Copyright is owned by the Author of the thesis.  Permission is given for 
a copy to be downloaded by an individual for the purpose of research and 
private study only.  The thesis may not be reproduced elsewhere without 
the permission of the Author. 
 
The effect of poplar stand density 
on hill country pastures 
A thesis submitted in partial fulfilment 
of the requirements for the degree of 
Doctor of Philosophy (PhD) 
Massey University, Palmerston North 
New Zealand 
Andrew James Wall 
2006 
ii 
Abstract 
One-third of the North Island of New Zealand has been identified as requiring increased 
soil conservation if pastoral farming is to be sustainable. For over 50 years the planting of 
widely spaced poplar trees (Populus spp.) has been one of the main methods used to 
control soil erosion on hill pastures. Research has shown that these plantings have 
successfully decreased soil erosion but their impact on the productivity of pastoral farming 
has received little research attention. The research that has been undertaken has found 
poplars can suppress understorey pasture production by up to 40%, suggesting that farmers 
require more research information on the impact of planting conservation trees on the 
productivity of their farm if the use of conservation trees is to be more widely adopted on 
erosion prone land. 
The objective of this thesis was to provide comprehensive data on the relationship between 
the range of poplar densities used for soil conservation on the light and soil under poplars, 
and consequently the effect on understorey pastures. Three field sites on commercial sheep 
and beef hill farms, in regions with contrasting summer soil moisture availability, 
Manawatu (one site) and Central Hawke's Bay (two sites), were monitored for two years. 
Tree stocking rates ranged from 0 to 375 trees/ha. Measurements were based on units of 
four trees with most measurements either directly below the tree crowns or in the gaps 
between the trees, but more intensive transect measurements were also made. 
Photosynthetically active radiation (PAR) and the ratio of red to far red light (R:FR) were 
measured under the trees and in open pasture controls. Stand density indices used included 
all the commonly used measures of tree canopies, including digital photography, and stem 
diameter at breast height (DBH). PAR transmission was inversely related to all of the stand 
density indices with canopy closure based on digital photographs being the most robust of 
the indices used. PAR under the trees, relative to open pasture, was greater in the gaps than 
below tree crowns. Under a completely closed canopy, PAR transmission was reduced to 
15-20% and 50-55% of the open pasture in summer and winter, respectively. The R:FR 
under the trees, relative to open pasture, decreased markedly at high stand densities 
(allowing less than 40% PAR transmission) in summer, but was similar in winter. The 
change in PAR under the trees was shown to be a major factor limiting pasture growth, 
particularly directly below the tree crowns. For both summer and winter, canopy closure 
iii 
measured with a standard digital camera was strongly related to stand level PAR 
trans�sion (r2=0.88-0.97; P<O.OOOl) and was also a practical method of measuring 
canopy closure in the field. 
The soil measurements confirmed earlier research that soil pH increases under mature 
poplar trees. There was a 0.2 - 0.7 unit increase in soil pH in the upper 75 mm of soil over 
both contrasting regions. The soil fertility under the trees in terms of requirements for 
pasture growth was similar to that of the open pasture with calcium and potassium up to 
2.2 and 9.0 quick test units higher in the soil under the trees than in the open pasture, 
respectively. The direct cause of the increased concentration of some cations under the 
trees was the annual tree leaf litter. Overall, the soil fertility under the trees had the 
potential to produce similar pasture production to that of the open pasture with the added 
advantage of less acid conditions. 
Averaged over all sites the respective annual net herbage accumulation (ANHA) under 
poplar canopy closures of 25, 50 and 75 % was estimated from the equations developed to 
be 77, 60 and 48% of the open pasture. The greatest decrease was directly below the tree 
crowns where at canopy closures greater than 20% the ANHA was a relatively constant 
50% of open pasture. In the vertically projected gap between trees the ANHA decreased by 
6.6% relative to open pasture for each 10 % increase in canopy closure. At approximately 
80% canopy closure there was no difference between the ANHA directly below the trees 
and in the gap. Pasture net herbage accumulation (NHA) under the trees relative to open 
pasture was at its lowest in summer and autumn (36% of open pasture under a closed 
canopy), and at its greatest in early spring before tree canopy leafed out (72% of open 
pasture under a closed canopy). The botanical composition and feed value of the pasture 
under the trees was broadly similar to that of the open pasture. 
The greatest impact of the poplars on the pasture was decreased NHA due to shading. The 
decrease in NHA directly below mature unpruned poplars is substantial and would 
decrease farm profitability if the poplar stand density were high over a large area of the 
farm. The use of poplars for soil conservation is essential but these results show the 
importance of managing trees through pruning and thinning so that canopy closure is 
minimised. ANHA under the trees can be maintained at 75% of the open pasture if canopy 
closure is prevented from exceeding 30-40%. 
iv 
Acknowledgements 
My PhD could not have been competed without the assistance and guidance of so many 
people, but foremost my supervisors Associate Professor Peter Kemp and Dr Alec Mackay. 
I thank Associate Professor Peter Kemp for his great patience, guidance, and 
encouragement during my studies and for the numerous readings and editing of my PhD 
manuscript. To Dr Alec Mackay I thank you for your guidance, enthusiasm for the project, 
and willingness to answer many questions, especially during when I was setting up the 
field experiments. Associate Professor Peter Kemp and Dr Alec Mackay were instrumental 
and the catalyst in the development of concepts and understanding of tree-pasture systems 
enabling this study. 
Many thanks must go to all of the farmers involved in this project. Allister Clark and later 
Darrell Shellard (Kiwitea), John Dunderdale (Hautope), and Martin Meredith (Hautope) 
thank you for allowing me to use your properties for my fieldwork. 
I would like to acknowledge the support of all those people who helped me in so many 
ways during the course of my PhD thesis. To Yvonne Gray and the ladies in the Herbage 
Laboratory at AgResearch Grasslands, I thank you for helping me with the pasture 
composition dissections. Thanks to Ian Power for travelling all the way down from 
Ruakura to take light measurements at each of the farm sites. Thanks to Aurelio Guevara­
Escobar for teaching me how to analyse canopy digital images. Thanks also to Roger Levy 
for your tireless help with pasture cuts on some pretty rugged terrain and to Phil Budding 
for helping me with soil sampling and taking tree measurements. To all who helped me in 
the field, Naba Devkota, Tara Pande, Cameron McKinnon, Lee Matheson, Joanne Wall, 
Chris Wall, and many others, your assistance was greatly appreciated. 
For financial support during my studies, I would like to express my gratitude to 
AgResearch. Their financial contribution and the scholarship opportunity they provided 
enabled this study to be undertaken. 
My appreciation also goes out to my parents Suzanne and Chris Wall for their tremendous 
moral support and help over the years. 
Finally I wish to thank Leesa Roy my partner, for her love, support, understanding, and 
extreme patience over the many years of my field research and during the writing up 
process of this thesis. I will never forget. 
Contents 
1 Introduction and objectives 
2 The application of an experimental design to existing stands of poplar trees in 
New Zealand's North Island hill country 
3 The relationship between poplar stand density and PAR transmission 
4 The effect of poplar overstorey density on soil chemical properties 
5 The effect of poplar over storey density on understorey pasture production 
6 Conclusion 
Appendices 
v 
1 
10 
35 
126 
180 
289 
292 
vi 
List of tables 
Chapter 2 
Table 2.1 General stand and trial site characteristics. 18 
Table 2.2 Resource inventory for selecting experimental units at Kiwitea. 24 
Table 2.3 Resource inventory for selecting experimental units at Hautope 1. 25 
Table 2.4 Resource inventory for selecting experimental units at Hautope 2. 26 
Table 2.5 Regression parameters for P. x euramericana stand basal area 27 
(m2/ha) in relation to tree stocking rate (stemslha), at each 
commercial farm site, where Y = bX (constrained through origin). 
Chapter 3 
Table 3.1 
Table 3.2 
Table 3.3 
Table 3.4 
Table 3.5 
Table 3.6 
Dates when canopy closure was measured in the field. 
Overstorey (stand) density indices. 
Regression equations, and their respective coefficients, developed 
for overstorey (stand) density indices to estimate the percentage of 
above-canopy PAR (%DIFN) transmitted at the centre of a 
vertically projected canopy gap (Zone 3) between fully in leaf P. x 
euramericana trees. 
Regression parameters (standard errors in parentheses) for equations 
predicting estimated PAR transmission (Zone 3%DIFN) from stand 
green crown length (mIh), where Y=(a*b)/(b+x). 
Regression equations, and their respective coefficients, developed 
for overstorey (stand) density indices to estimate the percentage of 
above-canopy PAR (%DIFN) transmitted directly below (Zonel) 
the fully in leaf P. x euramericana crowns. 
Regression equations, and their respective coefficients, developed 
for over storey (stand) density indices to estimate the percentage of 
above-canopy PAR (%DIFN) transmitted at the centre of the 
vertically projected canopy gap (Zone 3) between the leafless P. x 
euramericana trees. 
43 
48 
52 
55 
58 
62 
Table 3.7 
Table 3.8 
Table 3.9 
Regression parameters (standard errors in parentheses) for equations 
predicting estimated Zone 3 %DIFN individually from CCNL, B� 
HPCC and CEV, where Y=a+h *x. 
Regression parameters (standard errors in parentheses) for equations 
predicting estimated PAR transmission (Zone3 %DIFN) from stand 
gr�en crown length (mIha), where Y=a+h *x. 
Regression equations, and their respective coefficients, developed 
for overstorey (stand) density indices to estimate the percentage of 
above-canopy PAR (%DIFN) transmitted directly below (Zone 1) 
the leafless P. x euramericana crowns. 
vii 
64 
65 
68 
Table 3.10 Regression models, and their respective coefficients, developed for 71 
selected stand indices to predict canopy closure (CC). 
Table 3.11 Relationships between the in-leaf canopy closure ratio (CCL) and 73 
estimated stand-level PAR transmission (weighted %DIFN). 
Table 3.12 Regression parameters (standard errors in parentheses) for equations 75 
predicting understorey R:FR, relative to open pasture (%), from 
estimated PAR transmission (%DIFN), where Y=a+h *In(x). 
Table 3.13 Light transmission through various closed stands of deciduous trees. 91 
Table 3.14 Maximum solar angle'!' above the horizon allowing direct-beam 93 
radiation to pass below the crown of an average sized tree to its base 
Chapter 4 
Table 4.1 
Table 4.2 
and also the shadow length produced by the same tree. 
Kiwitea soil chemical properties: directly below the poplar crowns 
(Zone 1), at the centre of the vertically projected canopy gap 
between the trees (Zone 3), and within the open pasture. 
Hautope 1 soil chemical and biochemical properties: directly below 
the polar crowns (Zone I), at the centre of the vertically projected 
canopy gap between the poplar trees (Zone 3), and within the open 
pasture. 
135 
141 
Chapter 5 
Table 5.1 
Table 5.2 
Table 5.3 
Table 5.4 
Table 5.5 
Table 5.6 
Table 5.7 
Table 5.8 
Regression models and their respective coefficients, developed for 
CCL to predicted winter understorey NHA, relative to open pasture 
production. 
Pasture composition at Kiwitea, expressed as a percentage of total 
sward biomass (%DM). 
Main high fertility responsive (HFR) grasses and legumes at 
Kiwitea, expressed as a percentage of total sward biomass (%DM). 
Pasture composition at Hautope 1, expressed as a percentage oftotal 
sward biomass (%DM). 
Main high fertility responsive grasses and legume species at 
Hautope 1, expressed as a percentage of total sward biomass 
(%DM). 
Pasture composition at Hautope 2, expressed as a percentage of total 
sward biomass (%DM). 
Nutritive value indices for the mixed-species pasture at Kiwitea: 
directly below the crowns of individual trees (Zone 1), at the centre 
of the vertically projected gap between trees (Zone 3), and in the 
open pasture. 
Nutritive value indices for mixed-species pasture at Hautope 1 : 
directly below the crowns of individual poplar trees (Zone 1), at the 
centre of the vertically projected gap between the poplar crowns 
(Zone 3), and in the open pasture. 
viii 
201 
216 
217 
222 
223 
226 
229 
235 
Table 5.9 Nutritive value indices for P. x euramericana leaves at Kiwitea. 242 
Table 5.10 Nutritive value indices for P. x euramericana leaves at Hautope 1. 243 
Table 5.11 Major mineral elements (p.g/g DM) in the mixed-species pasture at 245 
Kiwitea: directly below the crowns of individual poplar trees (Zone 
1), at the centre of the vertically projected gap between poplar 
crowns (Zone 3), and in the open pasture. 
ix 
Table 5. 12 Trace mineral elements (pg/g DM) in the mixed-species pasture at 246 
Kiwitea: directly below the crowns of individual poplar trees (Zone 
1), at the centre of the vertically projected gap between poplar 
crowns (Zone 3), and in the open pasture. 
Table 5. 13 Major mineral elements (pg/g DM) in the mixed-species pasture at 255 
Hautope 1: directly below the crowns of individual poplar trees 
(Zonel), at the centre of the vertically projected gap between poplar 
crowns (Zone 3), and in the open pasture. 
Table 5. 14 Trace mineral elements (p.g/g DM) in the mixed-species pasture at 256 
Hautope 1: directly below the crowns of individual poplar trees 
(Zone 1), at the centre of the vertically projected gap between poplar 
crowns (Zone 3), and in the open pasture. 
Table 5. 15 Major mineral elements (p.g/g DM) in Populus x euramericana 264 
leaves at Kiwitea. 
Table 5. 16 Trace mineral elements (p.g/g DM) in P. x euramericana leaves at 264 
Kiwitea. 
Table 5. 17 Major mineral elements (p.g/g DM) in Populus x euramericana 265 
leaves at Hautope 1. 
Table 5. 18 Trace mineral elements (�g/g DM) in P. x euramericana leaves at 265 
Hautope 1. 
Table 5. 19 New Zealand studies measuring annual net herbage accumulation 267 
(ANHA) under poplars. 
List of figures 
Chapter 2 
Figure 2.1 
Figure 2.2 
Chapter 3 
Figure 3.1 
Figure 3.2 
Figure 3.3 
Figure 3.4 
Figure 3.5 
Figure 3.6 
Schematic diagram of the pairing of tree and open pasture 
experimental units. 
The relationship between stand basal area and tree stocking rate at 
Kiwitea (-), Hautope 1 (�), and Hautope 2 (D). 
A schematic aerial VIew of four 'nuclei' trees (grey circles) 
defining the boundary ( dashed lines) of an experimental unit. 
Sample grid overlaid and centred on an experimental unit to 
estimate canopy closure (Knowles et al. 1999). 
Overstorey environments within the boundary of an experimental 
unit: Zone 1 - area within the expanded canopy gap, directly 
underneath the crown of the 'nucleus' tree located a the north­
eastern corner of an experimental unit; and Zone 3 - area at the 
centre of the vertically projected canopy gap (also refer to Figure 
3.1). 
The estimated percentage of above-canopy PAR (%DIFN) 
transmitted at the centre of a vertically projected canopy gap 
(Zone 3) between fully in leaf poplars, over a range of (a) canopy 
closure ratios (CCL) measured with a standard digital camera and 
(b) stand basal areas (BA). 
The estimated percentage of above-canopy PAR (%DIFN) 
transmitted at the centre of a vertically projected canopy gap 
(Zone 3) between fully in-leaf poplars, over a range of stand 
crown ellipsoidal volumes (CEV) at each farm site. 
The estimated percentage of above-canopy PAR (%DIFN) 
transmitted at the centre of a vertically projected canopy gap 
(Zone 3) between fully in-leaf poplars, over a range of stand green 
crown lengths (GCL) at each farm site. 
x 
16 
27 
41 
43 
47 
53 
54 
54 
Figure 3.7 
Figure 3.8 
Figure 3.9 
Figure 3.10 
Figure 3.11 
Figure 3.12 
Figure 3.13 
Figure 3.14 
Figure 3.15 
The estimated percentage of above-canopy PAR (%DIFN) 
transmitted directly below the fully in-leaf poplar crowns (Zone 
1), over a range of (a) stand canopy closure ratios (CCL) 
measured with a standard digital camera and (b) stand basal areas 
(BA). 
The estimated percentage of above-canopy PAR (%DIFN) 
transmitted at the centre of a vertically projected gap between 
leafless poplars (Zone 3), over a range of (a) canopy closure ratios 
(CCNL) measured with a standard digital camera and (b) stand 
basal areas (BA). 
The estimated percentage of above-canopy PAR (%DIFN) 
transmitted at the centre of a vertically projected canopy gap 
(Zone 3) between leafless poplars, over a range of stand green 
crown lengths (mIha). 
The estimated percentage of above-canopy PAR (%DIFN) 
transmitted directly below (Zone 1) the leafless poplar crowns, 
over a range of (a) stand canopy closure ratios (eeL) measured 
with a standard digital camera and (b) stand basal areas (BA). 
Mean estimated percentage of above-canopy PAR (%DIFN) 
transmitted directly underneath the individual poplar crowns 
(Zone 1) at each site. 
Relationships between overstorey (stand) density indices. 
Estimated stand-level PAR transmission (weighted %DIFN) 
across a range of in-leaf canopy closure ratios (CeL) in (a) 
summer and (b) winter. 
The relationship between the understorey R:FR, relative to the 
open pasture, and estimated PAR transmission (%DIFN) through 
the poplar canopy in (a) summer (in-leaf) and (b) winter (no-leaf). 
Schematic diagram illustrating the smaller angle of sky view seen 
from Zone 1 and its general skew away from the zenith compared 
to Zone 3. 
xi 
59 
63 
65 
69 
70 
72 
74 
76 
82 
Figure 3 . 16  
Figure 3. 1 7  
Figure 3. 1 8  
Figure 3. 1 9  
Figure 3.20 
Figure 3.21  
Chapter 4 
Figure 4. 1 
Figure 4.2 
Figure 4.3 
Figure 4.4 
Relationships between in-leaf canopy closure (CCL) and basal 
area (BA). 
Comparison of relationships between no-leaf canopy closure 
(CCNL) and in-leaf canopy closure (CCL). 
Comparison of the developed relationship between R;FR and 
estimated PAR transmission with data from other sources. 
Estimated pasture dry matter production (0) at the centre of a 
vertically projected canopy gap (Zone 3) and (�) directly below 
the poplar crowns (Zone 1 ), over a range of stand canopy closure 
ratios in (a) summer and (b) winter. 
Estimated tilleringlbranching (0) at the centre of a vertically 
projected canopy gap (Zone 3) and (�) directly below the poplar 
crowns (Zone 1 ), over a range of stand canopy closure ratios in (a) 
summer and (b) winter. 
Estimated (a) dry matter production and (b) tillerslbranches per 
plant for (0) perennial ryegrass/white clover and (.) cocksfootl 
lotus pastures at the centre of a vertically projected canopy gap 
(Zone 3) and for (�) perennial ryegrass/white clover and (�) 
cocksfootllotus pastures directly below the poplar crowns (Zone 
1 ), over a range of stand canopy closure ratios (CCL) in summer. 
Unbalanced nested factorial-treatment structure. 
Kiwitea 0-75mm soil chemical properties linearly related to poplar 
canopy closure (eCL). 
Kiwitea soil chemical properties linearly related to poplar canopy 
closure (CCL). 
Hautope 1 0-75 mm soil chemical properties linearly related to 
poplar canopy closure (eCL). 
XII 
89 
89 
96 
100 
103 
1 07 
132 
136 
137  
1 42 
Figure 4.5 
Chapter 5 
Figure 5. 1 
Figure 5.2 
Figure 5.3 
Figure 5.4 
Figure 5.5 
Figure 5.6 
Figure 5 .7 
Figure 5.8 
Figure 5.9 
Hautope 1 soil chemical properties linearly related to poplar 
canopy closure (CCL). 
Intensified sampling zones used for a limited number of 
experimental units at Kiwitea: Zones 1 & 5 - directly below the 
individual tree crowns on the northeast and southwest corners and 
experimental unit, respectively; Zone 3 - at the centre of the 
vertically projected canopy gap between the trees; and Zones 2 & 
4 - transitional area midway between the centre of the vertically 
projected canopy gap and its respective northeast and southwest 
edge. 
Annual net herbage accumulation (ANHA) over a range of in-leaf 
canopy closure ratios. 
Annual net herbage accumulation (ANHA) stratified across the 
inter-tree space at Kiwitea ( 1998-2000). 
Summer net herbage accumulation (NHA) over a range of in-leaf 
canopy closure ratios. 
Autumn net herbage accumulation (NHA) over a range of in-leaf 
canopy closure ratios. 
Winter net herbage accumulation (NHA) over a range of in-leaf 
canopy closure ratios. 
Spring net herbage accumulation (NHA) over a range of in-leaf 
canopy closure ratios. 
Net herbage accumulation (NHA), relative to the open pasture, 
across the inter-tree space at Kiwitea (1998-00) in: (a) summer, 
(b) autumn (c) winter, and (d) spring. 
Residual herbage mass in each of the three mam overstorey 
environments at (a) Kiwitea, (b) Hautope 1 ,  and (c) Hautope 2. 
xiii 
143 
1 87 
196 
197 
199 
200 
202 
203 
205 
208 
Figure 5.10 
Figure 5.11 
Figure 5.12 
Figure 5.13 
Figure 5.14 
Figure 5.15 
Figure 5.16 
Figure 5.17 
Figure 5.18 
Post-trimming Kiwitea residual herbage mass in (a) summer and 
(b) autumn over a range of poplar canopy closure ratios (eeL). 
Post-trimming Hautope 1 & 2 residual herbage mass in (a) autumn 
and (b) winter over a range of poplar canopy closure ratios (eeL). 
Residual herbage mass (RHM) stratified across the inter-tree 
space at Kiwitea in (a) late summer and (b) late autumn (1999-
2000). 
Relationship between spring pasture composition and poplar 
canopy closure (CeL) at Kiwitea. 
Relationship between the summer pasture composition and poplar 
canopy closure at Kiwitea. 
Relationship between HFR grasses and poplar canopy closure 
(eCL) at Hautope 1 in spring: (�) directly below the poplar 
crowns (Zone 1), and (0) at the centre of the vertically projected 
canopy gap between the trees (Zone 3). 
Relationship between dead pasture material and poplar canopy 
closure (eeL) at Hautope 1 in summer: (A) directly below the 
poplar crowns (Zone 1), and (0) at the centre of the vertically 
projected canopy gap between the trees (Zone 3). 
Spring relationships between pasture nutritive (feed) value indices 
and poplar canopy closure (eeL) at Kiwitea: (�) directly below 
the crowns of individual trees (Zone 1), and (0) at the centre of 
the vertically projected canopy gap between trees (Zone 3). 
Summer relationships between pasture nutritive (feed) value 
indices and poplar canopy closure (eeL) at Kiwitea: (�) directly 
below the crowns of individual trees (Zone 1), and (0) at the 
centre of the vertically projected canopy gap between trees (Zone 
3). 
xiv 
209 
210 
212 
218 
219 
224 
224 
231 
232 
Figure 5.19 
Figure 5.20 
Figure 5.21 
Figure 5.22 
Figure 5.23 
Figure 5.24 
Figure 5.25 
Spring relationships between pasture feed value indices and poplar 
canopy closure (eeL) at Hautope 1: (A) directly below the 
crowns of individual trees (Zone 1) and (0) at the centre of the 
vertically projected canopy gap between trees (Zone 3). 
Summer relationships between pasture feed value indices and 
poplar canopy closure (eeL) at Hautope 1: (A) directly 
underneath the crowns of individual trees (Zone 1), and (0) at the 
centre of the vertically projected canopy gap between trees (Zone 
3). 
Summer relationship between the dietary cation-anion difference 
(DeAD) and poplar canopy closure (eeL) at Hautope 1: (A) 
directly below the crowns of individual trees (Zone 1), and (0) at 
the centre of the vertically projected canopy gap between trees 
(Zone 3). 
Spring relationships between major pasture minerals and poplar 
canopy closure (eeL) at Kiwitea: (A) directly below the crowns 
of individual trees (Zone 1), and (0) at the centre of the vertically 
projected canopy gap between trees (Zone 3). 
Summer relationships between major pasture minerals and poplar 
canopy closure (eeL) at Kiwitea: (A) directly below the crowns 
of individual trees (Zone 1), and (0) at the centre of the vertically 
projected canopy gap between trees (Zone 3). 
Spring relationships between pasture trace minerals and poplar 
canopy closure (eeL) at Kiwitea: (A) directly below the crowns 
of individual trees (Zone 1), and (0) at the centre of the vertically 
projected canopy gap between trees (Zone 3). 
xv 
237 
238 
239 
248 
249 
250 
Summer relationship between strontium (Sr) and poplar canopy 251 
closure (eeL) at Kiwitea: (A) directly below the crowns of 
individual trees (Zone 1), and (0) at the centre of the vertically 
projected canopy gap between trees (Zone 3). 
List of appendices 
Chapter 4 
4.1 
4.2 
4.3 
Chapter 5 
5.1 
5.2 
Experimental unit selections for soil samples taken at 75-150mm 
soil depth. 
Annual Ca returned in the pasture and tree biomass at Kiwitea. 
Mass-balance sulfur cycling model. 
Regression parameters (standard errors in parentheses) for 
equations predicting ANHA from in-leaf stand canopy closure 
(CeL). 
Annual net herbage accumulation (ANHA) over a range of in-leaf 
canopy closures. 
xvii 
292 
293 
294 
295 
296 
Chapter 1 1 
1 Introduction and objectives 
1.1 Introduction 
The inherent geology and climate of the North Island hill country of New Zealand combine 
to make a deeply dissected landscape, which is prone to high rates of natural soil erosion 
(Molloy 1988; Eyles & Newsome 1992; Anonymous 1996). Adding to the land's 
instability has been the removal of forest and subsequent development of extensive 
pastoral farming (Trustrum et al. 1984; Blaschke et al. 1992; Miller et af. 1996). Overall, 
an estimated 3.7 million hectares or 33% of the North Island requires significant soil 
conservation measures in order for it to be able to physically sustain pastoral based land 
uses (Eyles & Newsome 1992). 
The problems of North Island hill country erosion have long been recognised, however, 
voluntary and Government subsidised soil conservation programmes have achieved only 
localised reductions in soil erosion (Glasby 1991; Anonymous 1996; Miller et al. 1996). 
Surveys between 1989 and 1992 indicated that about 80% of North Island hill country 
farmland needed soil erosion control measures and that 55% had not received adequate 
treatment (Clough & Hicks 1992). The Ministry for the Environment has identified that hill 
country soil erosion is a key issue that needs to be addressed within the next 10 to 15 years 
due to the potential size of the problem, the negative and often irreversible nature of soil 
erosions impacts, and its economic and social significance. As a key principle it also places 
the primary responsibility for achieving sustainable land management with individual 
farmers (Anonymous 1996). 
Also identified as an emerging issue, is the perception of New Zealand's extensive 
agricultural production systems by aflluent and well-informed consumers from key 
international markets (e.g. Europe, North America, and Asia) (William 1995; Mathews 
1996; West 1997). Several agencies monitoring world market trends have identified that 
New Zealand's environmental and animal welfare standards are a matter of increasing 
strategic significance to our agricultural export industries. The competitive advantage 
gained from New Zealand's quality image is seen as an essential element required for the 
optimum positioning of exports in premium markets and enables any threats to trade, in the 
post-GATT era, to be minimised (BayveI1993; West 1997). 
Chapter 1 2 
Soil conservation on North Island hill country farms has centred on the planting of exotic 
tree species, which counter the loss of soil strength resulting from pastoralism (Van 
Kraayenoord & Hathaway 1986; Hicks 1995; Miller et al. 1996). Where trees are 
appropriately planted and maintained for the type of erosion present they can reduce soil 
erosion by 50 to 80% (Hicks 1992, 1995). This lowers the magnitude of soil erosion close 
to levels found on stable hill country, even during large storm events (Miller et al. 1996). 
Trees are predominantly used because they are relatively inexpensive in relation to the 
large areas that require control measures (Van Kraayenoord 1976). The main tree species 
used for erosion control on pastoral hill land include poplar (Populus spp.), willow (Salix 
spp.) and radiata pine (Pinus radiata D. Don) (Van Kraayenoord & Hathaway 1986; 
Thompson & Luckman 1993; Wilkinson 1999). 
Since the early 1950s poplars have been extensively planted for soil erosion control on 
North Island hill country (Van Kraayenoord 1993). However, it has become evident that 
over the same period, most research on tree-pasture systems has been directed at the more 
commercial timber species - radiata pine (Hodgson 1997; Wall et al. 1997). Agroforestry 
research based on radiata pine shows that, over time, trees can significantly alter the 
microclimate, soil properties, and the water balance of farm systems (Fahey & Rowe 1992; 
Hawke & Wedderburn 1994; Maclaren 1996). These changes, along with trees competitive 
dominance, usually leads to reduced understorey pasture production in temperate zones 
(percival & Hawke 1985; Kellas et al. 1995; Scholes & Archer 1997). Predictably, the 
negative impact of trees on pasture production increases when they are planted at high 
stocking rates and also as the trees mature and their crowns develop in size (Hawke & 
Percival 1992). 
Animal production in New Zealand's hill country is based on the extensive grazing of open 
pastures. Therefore, any impact that trees have on pasture production and the micro climate 
will likely influence animal production. The shelter and shade provided by trees can reduce 
the exposure of farm animals to climatic extremes and in doing so provide animal welfare 
benefits (Gregory 1995; Mathews 1996). However, when farm animals are grazed 
continuously under evergreen trees (e.g. Pinus radiata and Eucalyptus) their productivity 
generally declines, relative to farm animals grazed on stable open pastures (Percival et al. 
1988; Knowles 1991; Bird et al. 1995). 
Chapter 1 3 
Many of the factors limiting pasture and animal production under evergreen trees may also 
be present under poplars. Nevertheless, there are a number of distinct contrasts between the 
tree species and their normal management requirements that may have different effects on 
pastoral hill farms. For example: poplars are deciduous species; depending on hybrid, they 
can maintain a narrow and less dense crown when planted at low stocking rates; their 
foliage is nutritious and reasonably palatable to browsing livestock; soil conservation 
plantings generally establish trees at low planting-to-final-tree-stocking ratios; and the use 
of individually protected pole planting stock allows trees to be planted into erosion 
susceptible areas of farms without having to exclude smaller livestock classes, such as 
sheep (Wall et al. 1997). 
Poplars have been shown to suppress understorey pastures by 20 to 40%, relative to stable 
open pastures (Gilchrist et al. 1993; Miller et al. 1996; Guevara-Escobar et al. 1997; 
Douglas et al. 2001). However, these studies are based on a small number of tree stocking 
rates and tree ages. The interactions between trees and their surrounding environment 
varies both spatially and temporally, with the extent and intensity of the relationships 
depending on the particular farm site, planting configuration, tree age, and silvicultural 
management. This dynamic nature of tree-pasture systems, along with the lack of 
homogeneity often found between different stands of trees makes inferences from the 
above studies of limited use for strategic farm planning. 
Over the last fifty years poplars have been extensively planted for soil erosion control on 
North Island hill country farms (Van Kraayenoord 1993; Haslett et al. 1995), and today 
there still remains great potential for further plantings (Clough & Hicks 1992); tree-pasture 
systems continue to be the most practical technology for sustaining pastoral farming on 
erosion prone hills. However, the lack of data to support the integration and management 
of such systems, along with the difficulty in universally applying current information, is a 
major hindrance to farmers when they consider soil conservation trees as a sustainable land 
management option. 
Chapter 1 4 
1 .2 Objectives 
As a consequence of the above, the general goal of this thesis was to provide farmers with 
comprehensive, applicable, and user-friendly information on the effects of widely-spaced 
poplars on the productivity of pastoral hill farms. Thereby, assisting farmers with more 
optimal integration and management of poplars as a soil conservation option and enabling 
fanners to make complex sustainable land-use decisions based on sound and robust 
information. 
Specific objectives for this thesis were to: 
1. Set up a treatment structure that would allow the dynamic nature of tree-pasture 
systems to be taken into account. 
2. Determine how different densities of poplars alter the quantity and quality of light 
reaching the understorey and to predict the likely consequences of the changes on 
pasture production. 
3. Determine, based on the tree's effect on light, which stand density index or indices 
is/are best for characterising poplar stands. 
4. Determine how different densities of poplar alter the soil resource status of bill 
farms, thereby providing farmers with an indication of the effects of poplars on the 
long-term viability of pastoral hill farm systems. 
5. Develop a predictive relationship between poplar stand characteristics and seasonal 
pasture production that can be used by farmers to gauge the impact of poplars on 
farm productivity. 
6. Determine the forage value of poplar leaves as a potential fodder source for 
browsing livestock. 
Chapter 1 5 
1.3 Thesis outline 
This thesis is divided into six chapters. Chapter 1 briefly discusses the main use of poplar 
trees on New Zealand hill country farms and reasons why research into their impacts on 
pastures is needed. More detailed discussion of previously published literature related to 
tree-pasture systems is given in each relevant Chapter and a review has been presented in 
the Proceedings of the New Zealand Grassland Association (Wall et al. 1997). Chapter 2 
descnbes the commercial fann sites and main treatment structure used throughout this 
study. Chapter 3 investigates how poplars affect both the quantity and quality of light 
reaching the understorey and the likely consequences on pasture production. Different 
stand density indices are compared for their simplicity, precision, and robustness as 
predictors of light transmission. The best index was then used in following chapters to 
characterise/quantify the density of each poplar stand. Chapter 4 examines how changes in 
poplar stand density affects topsoil fertility in relation to pasture production directly below 
and in-between the trees. Chapter 5 builds on information gathered from the previous two 
chapters and integrates it with actual pasture production and pasture quality measurements. 
The forage value of poplar leaves is also investigated. Chapter six summarises the main 
conclusions from each previous chapter. 
Chapter 1 6 
1.4 References 
Anonymous. (1996) Sustainable land management: A strategy for New Zealand. Ministry 
for the Environment, Wellington, New Zealand. 
Bayvel, A. C. D. (1993) Animal Welfare - a threat or an opportunity for research, farming 
and trade. Proceedings of the New Zealand Society of Animal Production 53: 223-225. 
Bird, P.R.; Kellas, J.D.; Kearney, G.A.; Cumming, K.N. (1995) Animal production under a 
series of Pinus radiata-pasture agroforestry systems in south-west Victoria, Australia. 
Australian Journal of Agricultural Research 46(6): 1299-310. 
Blaschke, P.M.; Trust� N.A; DeRose, R.C. (1992) Ecosystem process and sustainable 
land use in New Zealand steep lands. Agriculture, Ecosystems and Environment 41: 153-
178. 
Clough, P.; Hicks, D. (1992) Soil conservation and the Resource Management Act. 
New Zealand Institute of Economic Research, Wellington, New Zealand. 
Douglas, G. B.; Wa1croft, AS.; Wills, B.J.; Hurst, S.E.; Foote, AG.; Trainor, K.D.; Fung, 
L.E. (2001) Resident pasture growth and the micro-climate beneath young, wide-spaced 
poplars in New Zealand. Proceedings of the New Zealand Grassland Association 63: 131-
138. 
Eyles, G.O.; Newsome, P.F. (1992) A soil conservation approach to sustainable land use. 
The Proceedings of the International Conference on Sustainable Land Management, 
Napier, New Zealand, 17-23 November 1991 (ed Henriques, P.R.), pp 216-220. 
Fahey, B.D.; Rowe, L.K. (1992) Land-use impacts. Waters of New Zealand (ed Mosley, 
M.P.), pp 265-284. New Zealand Hydrological Society. 
Chapter 1 7 
Gilchrist, A.N.; deZ Hall, J.R.; Foote, A.G.; Bulloch, B.T. ( 1993) Pasture growth around 
broad-leaved trees planted for grassland stability. Proceedings of the XVII International 
Grassland Congress: 2062-2063. 
Glasby, G.P. ( 1991 )  A review of the concept of sustainable management as applied to 
New Zealand. Journal of the Royal Society of New Zealand 21 (2): 6 1 -81 .  
Gregory, N.G. ( 1 995) The role o f  shelterbelts in  protecting livestock: A reVIew. 
New Zealand Journal of Agricultural Research 38: 423-50. 
Guevara-Escobar, A.; Kemp, P.D.; Hodgson, 1.; Mackay, A.D.; Edwards, W.R.N. (1997) 
Case study of a mature Populus deltoides-pasture system in a hill environment. 
Proceedings o/the New Zealand Grassland Association 59: 179-185. 
Haslett, T.; Gilchrist, K.; Britton, B. ( 1995) Poplar for timber and other uses? What 's 
New in Forest Research 238. New Zealand Forest Research Institute Ltd, Rotorua, 
New Zealand. 
Hawke, M.F.; Percival, N.S. (1992) Sheep growth rates under Pinus radiata. Proceedings 
of the New Zealand Society of Animal Production 52: 229-231 .  
Hawke, M.F.; Wedderburn, M.E. ( 1 994) Microclimate changes under Pinus radiata 
agroforestry regimes in New Zealand. Agricultural and Forest Meteorology 71 :  1 33-145. 
Hicks, D. (1992) Impact of soil conservation on storm-damaged hill grazing lands in 
New Zealand. Australian Journal of Soil and Water Conservation 5(1): 35-40. 
Hicks, D.L. (1 995) Control of soil erosion on farmland. A summary of erosion 's impact on 
New Zealand agriculture, and farm management practices which counteract it. MAF 
Policy Technical Paper 95/4. MAF Policy, Wellington, New Zealand. 
Hodgson, J. ( 1997) Sustainability in pastoral systems: New Zealand experience. Fifth 
British Grassland SocietyResearch Conference: 1 - 16. 
Chapter 1 8 
Kellas, J.D.; Bird, P.R.; Cumming, K.N.; Kearney, G.A.; Ashton, A.K ( 1995) Pasture 
production under a series of Pinus radiata-pasture agroforestry systems in South-west 
Victoria, Australia. Australian Journal of Agricultural Research 46(6): 1285- 1297. 
Knowles, RL. ( 1991)  New Zealand experience with silvopastoral systems: A review. 
Forest Ecology and Management 45: 251 -67. 
Maclaren, J.P. (1996). Environmental effects of planted forests in New Zealand. Forest 
Research Institute Bulletin No. 198. New Zealand Forest Research Institute, Rotorua, 
New Zealand. 
Mathews, L.R (1 996) Animal welfare and sustainability of production under extensive 
conditions: a non-EU perspective. Applied Animal Behaviour Science 49: 41 -46. 
Miller, D.E.K; Gilchrist, A.N.; Hicks, D.L. ( 1996) The role of broad-leaved trees in slope 
stabilisation in New Zealand pastoral farming. Mountains of East Asia and the Pacific (eds 
Ralston, M.M.; Hughey, K.F.D.; O'Connor, KF.), pp 96-1 04. New Zealand Centre for 
Mountain Studies, Lincoln University, Canterbury, New Zealand. 
Molloy, L. ( 1993) Soils in the New Zealand landscape: the living mantle. New Zealand 
Society of Soil Science, Lincoln University, Canterbury, New Zealand. 
Percival, N.S.; Hawke, M.F. ( 1985) Agroforestry development and research ID 
New Zealand. New Zealand Agricultural Science 19(3): 86-92. 
Percival, N.S.; Hawke, M.F.; Jagusch, KT.; Korte, C.J.; Gillingham, A.G. (1988) Review 
of factors affecting animal performance in pine agroforestry. Agroforestry Symposium 
Proceedings (ed Mac1aren, P.), pp 165-1 74. Bulletin No. 139. New Zealand Ministry of 
Forestry and Forest Research Institute. 
Scholes, RJ.; Archer, S.R (1997) Tree-grass interactions in savannas. Annual Review of 
Ecological Systems 28: 517-544. 
Chapter 1 9 
Thompson, RC.; Luckman, P.G. (1993) Performance of biological erosion control m 
New Zealand soft rock terrain. Agroforestry Systems 2 1 :  19 1 -2 1 1 .  
Trustrum, N.A; Thomas, V.J. ; Lambert, M.G. (1 984) Soil slip erosion as a constraint to 
hill country pasture production. Proceedings of the New Zealand Grassland Association 
45: 66-76. 
Van Kraayenoord, C.W.S. ( 1 976) Plant materials for erOSIon control. New Zealand 
Agricultural Science 10( 1): 29-33. 
Van Kraayenoord, C.W.S.; Hathaway, RL. (1 986). Plant materials handbook for soil 
conservation volume I Principles and practices. Water and Soil Miscellaneous Publication 
No. 93 . National Water and Soil Conservation Authority, Wellington, New Zealand. 
Van Kraayenoord, C.W.S. ( 1 993) Poplar growing in New Zealand - past to present. A 
potential for growth: proceedings of a meeting to reactivate the New Zealand national 
poplar commission, 24 February 1993, Aokautere, New Zealand (ed Bulloch, B.), pp 9-15 .  
Manaaki Whenua - Landcare Research, New ZeaIand. 
Wall, AJ.; Mackay, AD.; Kemp, P.D.;  Gillingham, AG.; Edwards; W.RN. ( 1 997) The 
impact of widely spaced soil conservation trees on hill pastoral systems. Proceedings of the 
New Zealand Grassland Association 59: 1 71 - 1 77. 
West, A ( 1997) A challenge to create a consumer focus. Agricultural Science 1 0(4): 25-
27. 
Wilkinson, AG. (1 999) Poplars and willows for soil erosion control in New Zealand. 
Biomass and Bioenergy 16: 263-274. 
Williams, M. ( 1 995) Green consumerism and market trends. Landlink: 7 
Chapter 2 1 0  
2 The application of an experimental design to 
existing stands of poplar trees in New Zealand's 
North Island hi l l  country. 
CONTENTS 
2.1 Introduction 1 1  
2.2 Methodology 13 
2.2. 1 Selection of poplar stands in the Manawatu and Hawke's Bay regions 13 
2.2.2 Elimination process for simultaneously creating a gradient treatment 
structure and selecting experimental units 14 
2.2.3 An index for describing the overstorey density of an experimental unit 1 6  
2.2.4 Data analysis 17 
2.3 Results & Discussion 18 
2.3.1 General description of the selected commercial hill farms 18 
2.3.2 General description of selected experimental units 20 
2.3.3 Over storey densities 22 
2.4 Conclusion 
2.5 References 
28 
29 
Chapter 2 1 1  
2.1 I ntroduction 
Researchers have used a number of different approaches and experimental designs to 
investigate the impacts of trees on pastoral systems. Many of these studies are based on 
either systematic tree-spacing designs (Cameron et al. 1 989; Eastham & Rose 1990; Bird 
et al. 1 994), or more classical randomised designs (Anderson & Batini 1 983; Percival & 
Knowles 1 988; Pollock et al. 1 994; Kellas et al. 1 995). Systematic designs have an 
advantage in being able to accommodate a far greater number of replicated stand density 
treatments within a much smaller area than classical randomised designs (Nelder 1 962; 
Bleasdale 1 967; Cameron et al. 1 989; Bird et al. 1 994). However, as a trade-off, they 
require very homogenous environmental conditions, and also the statistical methods 
applicable to them are more restricted (Nelder 1 962; Cameron et al. 1989; Bird et al. 
1 994). 
Irrespective of the type of experimental design, where trees are planted specifically for 
research purposes a long time frame is required to adequately capture the temporal 
variation found within the mixed tree-pasture system (Knowles et al. 1 999). For more 
short-term experiments, an alternative is to modify the structure of older existing stands of 
trees through selective thinning and/or pruning (Anderson & Batini 1 983 ; Sibbald et al. 
1 99 1 ;  Devkota et al. 1 998). However, this approach has problems in relation to the typical 
size and structure of poplar soil conservation plantings in the New Zealand landscape. 
Large areas (or stands) of evenly spaced-apart trees are required. This contrasts with the 
relatively small pockets of unevenly spaced-apart poplar trees normally seen scattered over 
the hillsides of New Zealand (Wilkinson 1 988; van Kraayenoord 1993). In addition, the 
time and resources needed for modifying stands of large-sized trees on hill-to-steepland 
terrain would likely be substantial (Kellas et al. 1 995). 
One way to avoid the limitations of the above approaches is through the application of an 
experimental design to unmodified stands of mature trees, along with the use of their 
resident understorey vegetation (pyke & Zamora 1 982; Scanlan & Burrows 1990; Crowe 
1 993; Gilchrist et al. 1 993; Guevara-Escobar et al. 1 997). Nevertheless, the predetermined 
arrangement of these trees can impede the control of  extraneous sources of environmental 
variation. This in turn may confound the interpretation of any observed treatment (stand 
Chapter 2 1 2  
density) effects (Remmenga 1 98 1). Many of the environmental factors governing both the 
composition and production of hill pastures in New Zealand can vary substantially across a 
single paddock, as reviewed by Harris ( 1994) and White ( 1 994). 
The short-term nature of the research project, along with the unavailability of large, well­
established poplar stands, severely limited the number of approaches available for this 
thesis. Given the above limitations, the initial objective was to create a broad gradient of 
poplar overstorey densities based on existing unmodified stands of mature trees, which 
would form the main treatment structure for this thesis. Careful management of extraneous 
sources of environmental variation, not pertinent to the main treatment structure, would 
also be required as part of the experimental design. 
Chapter 2 1 3  
2.2 Methodology 
2.2.1 Selection of poplar stands in the Manawatu and Hawke's Bay regions 
A number of farmer interest groups assisted in finding commercial farm sites with 
potentially suitable stands of mature poplar trees. In January 1 998, the 
ManawatulRangitikei branch of 'Federated Farmers of New Zealand Inc.' helped to find 
eight prospective sites within the Manawatu region. Each of these sites was subsequently 
visited and a short reconnaissance taken to assess their general suitability for carrying out 
field research. The search for prospective sites within the Hawke's Bay region did not start 
in earnest until November 1 998, after an appropriate experimental design had been 
developed and successfully applied in Manawatu. Between December 1 998 and January 
1 999 fifteen sites were assessed in this region with help from the 'Hawke's Bay Regional 
Council' and the lower North Island 'New Zealand Landcare Trust' . 
The suitability of a site for the planned research depended on several factors. At the 
broadest scale, both the site's geography and livestock farming system had to be 
representative of where poplar trees are normally planted for soil erosion control purposes. 
Trees of similar age, and planted at a wide range of tree stocking rates, including adjacent 
open pastures areas, was also essential. At a more detailed scale, to increase the precision 
of the research, there needed to be an opportunity for minimising the extraneous sources of 
environmental variation found between the different overstorey environments. Hence, the 
more suitable sites had their stands of poplar trees and adjacent open pasture areas situated 
on a similar aspect and hillslope, and were subjected to comparable grazing management 
and fertiliser histories. All of the overstorey environments needed to be well managed and 
used regularly as part of the farm's overall grazing system. 
The logistical aspects of each potential farm site were also considered in relation to the 
resources available and the practicalities involved with running fieldwork over an extended 
period of time. Enthusiastic farmers were sought who would take a pro active approach 
towards the research programme and allow continuous access to the farm sites for a period 
of one to two years. The distance travelled to the sites each day could not be excessive and 
on-farm access by four-wheel-drive vehicle had to be available all year around. 
Chapter 2 1 4  
2.2.2 Elimination process for simultaneously creating a gradient treatment 
structure and selecting experimental units 
When applying experimental designs to unmodified stands of trees, which are not 
originally planted for research purposes, the treatments are inherently linked to the 
experimental units (i.e. they are not actually applied to the experimental units but instead 
come with them) (Urquhart 1 98 1). Thus, the process for simultaneously creating a 
structured set of overstorey (stand) density treatments and selecting appropriate 
experimental units (micro sites) on each commercial farm involved three stages: initial 
identification and basic characterisation of all potential experimental units; grouping of 
these units into strata with similar overstorey densities; and lastly, the selection of 
experimental units within each stratum. 
2.2.2.1 Stage 1 
In June 1 998, a field reconnaissance was made to identify and mark all of the potential 
experimental units with and without trees at the commercial farm selected for the 
Manawatu region. Experimental units with trees consisted of a nucleus of four trees, 
arranged approximately in a square grid-pattern that, in turn, were surrounded by buffer 
trees of similar age and spacing (thus minimising any possible border or edge effects). 
Experimental units without trees (controls) were adjacent open pasture areas located 
(>50m) away from the influence of tree shade, drift from leaf fall, and laterally extending 
tree roots. 
The position of the open pasture controls away from any interference by the trees was 
particularly critical as this could have severely distorted any comparisons made between 
the overstorey environments, thus introducing a major source of error (Ong 1 996). Under 
such circumstances the overall productivity of experimental units with trees will usually be 
overestimated relative to that found in the open (Ong et al. 1 996). 
Chapter 2 1 5  
2.2.2.2 Stage 2 
A resource inventory of each potential experimental unit was taken as part of the initial 
field reconnaissance. This inventory described the general physical characteristics of the 
units, including their aspect, macro-slope, position on the hillside, homogeneity of the 
ground surface, and any irregularities. Each experimental unit with trees was also tape 
surveyed, as described by Studman ( 1990). Using the stems of the four nuclei trees as 
corner reference points for the tape survey, the size, shape (based on 1 :  1 00 scale 
drawings), and representative tree stocking rate were determined and this information was 
then collectively used to group all of the experimental units into strata with similar 
dimensions. 
Experimental units with irregular (non-square) shapes were discarded from the selection 
process and directly adjacent units of the same size were combined together as a single unit 
(Urquhart 198 1 ). An approximate square shape was sought to maintain a similar spatial 
pattern of tree influence within the boundary of the four nuclei trees. The representative 
tree-stocking rate (on a per hectare basis) of each experimental unit was also calculated 
assuming the trees were planted on a perfectly square grid pattern. Tree-stocking rate 
(stemslha) was calculated using the formula: 
Stemslha = 1 0,000 m2/experimental unit area (m2). 
The systematic groupmg of experimental units served two mam functions: firstly, it 
provided a base for selecting a wide range of representative overstorey densities and 
secondly, it enabled a reasonably even spread to be chosen in order to obtain well defined 
relationships between the independent (overstorey density) and dependent variables 
studied (Steel & Torrie 1 980; Myers 1 990). 
2.2.2.3 Stage 3 
The general physical characteristics, included in the resource inventory, helped to 
determine the final selection of experimental units. Where possible the experimental units 
within each stratum were randomly selected. However, in order to reduce the effects of 
recognised causes of variation not pertinent to the required comparisons, the experimental 
units were also situated, as much as possible, on areas with a similar aspect, macro-slope, 
Chapter 2 1 6  
vertical position on the hillside, and away from any irregularities, such as stock camps, 
large amounts of tree debris, or areas with poor localised drainage (Figure 2. 1 ). This was a 
form of local error control, where the influence of known experimental heterogeneity was 
minimised between the experimental units (Steel & Torrie 1 980). The open pasture 
(control) experimental units served as benchmarks (or standards) in the final selection 
process. 
Paired open 
control for unit 1 
Paired open 
control for unit 2 
Figure 2.1 Schematic diagram of the pairing of tree and open pasture experimental units. 
In February 1 999, a similar selection process was used at two neighbouring commercial 
farms within the Hawke's Bay region. However, as an additional restraint on the selection 
process, some overlap in overstorey densities between all of the farm sites in both regions 
was sought. 
2.2.3 An index for describing the overstorey density of an experimental unit 
The overstorey density of each selected experimental unit was initially described in terms 
of their representative stand basal area. Usually, this index is strongly related to total stand 
biomass (Hoffinann & Usoltsev 2002; Johansson 2002), and is commonly used in forestry, 
and to lesser extent agroforestry, for quantitatively describing the density of a stand 
(Carron 1 968; Percival & Knowles 1 988; Scanlan 199 1 ;  Reed & Mroz 1 997). The basal 
area of an experimental unit was calculated from circumference (C) measurements taken at 
1 .4 m above ground level (upslope) on the stems of the four nuclei trees. Assuming a 
circular cross-sectional area for each stem, the mean diameter (D) and basal area (BA) at 
Chapter 2 1 7  
breast height over bark per stem were derived for the selected experimental units using the 
basic geometric formula: 
D = C/1t (units: mlstem) BA = 1tD2 / 4 (units: m2/stem) 
The representative stand basal area per hectare (m2/ha) for each experimental unit was 
calculated as the product of the representative tree stocking rate per hectare (refer section 
2.2.2.2) and the mean basal area per tree stem. 
2.2.4 Data analysis 
Simple linear regression analyses, using the REG procedure of SAS® (version 8.02 for 
Windows®, SAS Institute Inc. 1 999), were performed to determine the relationship 
between stand basal area and tree-stocking rate for each of the selected commercial farm 
sites. The SAS® NOINT option forced all of the regressions through the origin (SAS 
Institute Inc. 1 991) .  
In general, the small size of  the data sets hindered the analysis of  the residuals from the 
regression models. Nevertheless, for the commercial farm site selected in the Manawatu 
region one potential outlier was detected from the residual plots and more formal test 
statistics (R-student statistic, a = 0.05). Influence diagnostics (Cook's D, HAT DIAG, 
DFFITS) indicated that this suspect point exerted an appreciable influence on the slope of 
the regression, thus reducing the performance/precision of the fitted model (Myers 1 990). 
On a closer re-examination of the data, the suspect point was found to be the only 
experimental unit situated on a south-facing aspect, as opposed to the rest, which were on a 
north-facing aspect. Given that these two aspects may have different effects on the growth 
and development of the trees, the single experimental unit on the southern aspect was 
removed as an outlier. 
The simple linear regressions developed for each commercial farm site were compared by 
analysis of co variance (ANCOVA) using the GLM procedure of SAS® (Littell et al. 1 99 1 ;  
Kleinbaum et al. 1 998). 
Chapter 2 
2.3 Results & Discussion 
2.3.1 General description of the selected commercial hill farms 
1 8  
Selected for this research project were three commercial sheep and beef farms, located 
within the Southern Hill Country of the North Island. A general description of each farm is 
given in Table 2. 1 .  The first farm chosen (Kiwitea) was situated within the Kiwitea district 
of Manawatu, whereas, the second and third farms (Hautope 1 & 2) were two adjacent 
properties at Hautope in Hawke's Bay. 
Table 2.1 General stand and trial site characteristics 
Kiwitea Hautope 1 Hautope 2 
�. 
Poplar species Populus x euramericana P. x euramericana P. x euramericana 
Age (years) > 30 > 25 > 25 
E-< rJ) Position on hillside Upper-lower slopes Upper-lower slopes Mid-lower slope 
Location (Region) Manawatu Hawke's Bay Hawke's Bay 
Latitude Longitude 40.08°S 1 75.47°E 39.58°S 1 76.43°E 39.58°S 1 76.44°E 
Hill - steepland Hill Hill 
Macro-topography 
(20-36� ( 1 5-27� (23-28� � 
- Aspect North South-east North-west rJ) 
Altitude a.s.l (m) 3201 1 60 220 
Annual raUnfall (nun) 1 0622 8683 868 
Summer-moist Summer-dry Summer-dry 
lTopographic map reference: 260-T23 470 1 53 (Kiwitea), 260-V22 279327 (Hautope 1), and 260-V22 
295322 (Hautope 2) (Terralink 1 996; Terralink 1997); 2Suckling (1975); 3Rainfall observations from closest 
meteorological station at Patangata (NZMS 1 986). Abbreviation: a.s. l, above sea level. 
2.3.1.1 Livestock farming systems 
All three farms had sheep and beef cattle production systems that were fairly typical for 
their respective west and east coast regions of the North Island (MAF 1 99 1 ). Romney 
breeding ewe flocks, generally rearing most replacement ewes, and finisrnng the majority 
of lambs in prime condition, were common sheep policy features. The cattle policies were 
more varied. Nevertheless, they included either a traditional beef cow breeding herd or a 
more flexible dry cattle system, incorporating a mixture of finishing and store strategies. 
Overall, the Kiwitea farm carried a 60:40 mixture of sheep and cattle at 1 0.5 stock units 
Chapter 2 1 9  
per hectare (SU/ha) (Clark per. comm. 1 998). In comparison, both of the Hautope farms 
ran a slightly higher 64:36 ratio of sheep to cattle, at a lower stocking rate of around 9.2 
SU/ha (Meredith per. comm. 200 1 ;  Dunderdale per. comm. 2001) .  
2.3.1.2 Soils 
The soils at Kiwitea comprised of Raumai hill soils on the easier slopes and hilltops, and 
related Pohangina steepland soils for most areas that were steep and broken (Suckling 
1 975; Rijkse 1 977). These sandy loam soils are classed as intergrades between yellow-grey 
earths and yellow-brown earths; and are formed from unconsolidated (loose) marine sands, 
sandstones, and siltstones, interbedded with compacted greywacke gravel bands and 
purniceous bands (Rijkse 1 977). The underlying parent material is unstable, especially on 
the steeper slopes, and easily erodes in the form of slow healing slips and gullies (Suckling 
1 950, 1 975; Rijkse 1977). 
The soils at Hautope consisted of Haturna-Raukawa soils (Pohlen 197 1 ). These soils are 
classed as yellow-grey earths, derived from sandy mudstone, calcareous mudstone, and 
bentonitic and sandy mudstone (Pohlen 197 1). Associated with these soils is dense subsoil 
that impedes deep drainage and in many places forms a hard pan. Under highly saturated 
soil conditions this subsoil boundary often leads to slip and slump erosion, which usually 
heals readily, and also tunnel-gully erosion (pohlen 1 97 1 ;  McCaskill 1 973). 
2.3 . 1 .3 Climate 
The climates of Manawatu and Hawke's Bay are dominated by the main axial mountain 
ranges that stretch across the North Island in a north-east to south-west direction and the 
southern hemisphere's  mid-latitude westerly airstreams. These mountain ranges act as a 
barrier to weather systems moving eastwards and consequently, Hawke's Bay on their 
leeward side is significantly drier and sunnier than Manawatu, which is fully exposed to 
the predominant rain-bearing westerlies (Burgess 1 983 ; Zwartz 1997). As a result, the 
mean annual rainfall recorded at the nearest meteorological station to Kiwitea was slightly 
higher (22%) than for Hautope (Table 2. 1 )  and was also more evenly distributed 
throughout the year (data not shown). In contrast, with the westerly airstreams being 
particularly common in late spring and summer, Hautope experiences extremely variable 
rainfall during this period and often summer evapotranspiration rates are high enough to 
Chapter 2 20 
cause seasonal drought conditions especially on the sunny northerly aspects (Pohlen 1 97 1 ;  
Thompson 1987).  
2.3.1 .4 Poplar stands 
The stands of poplars selected at Kiwitea were a mixture of untended Populus x 
euramericana (Dode) Guinier black hybrid clones (Table 2. 1). The hybrids included a 
small number of 'Flevo' and more predominantly the Italian bred selections 145/5 1 ,  12 14  
and 1488 (Fung per. co mm. 2001).  All of the trees were planted before the 1970s, except 
for the small number of 'Flevo' trees (Clark per. co mm. 1998). Adequate soil moisture 
over the poplar-growing season, along with fertile and free-draining soils, had enabled 
good tree establishment across the entire hillsides. 
The older Italian bred hybrids also made up the poplar stands selected at the Hautope farm 
sites (Fung per. comm. 2001).  Hautope 1 contained I78 's and 12 14's strategically planted 
in stands to prevent mass movement erosion, with additional paired-plantings also running 
down the hillsides in main drainage lines. At Hautope 2, only the more drought resistant 
1214 was present, which was planted further downhill because of the extreme summer-dry 
nature of the particular northwest-facing hillside. None of the stands had received any 
silvicultural management (pruning or thinning) and all were planted prior to, or around the 
time of, the 1973 outbreak of Melampsora rust diseases in New Zealand (McGregor per. 
comm. 200 1 ;  Faulknor per. comm. 2001). 
2.3.2 General description of selected experimental units 
2.3.2.1 Kiwitea 
A total of four large open pasture areas and twelve experimental units with trees were 
selected at this site (Table 2.2) .  Replication of treatments was limited to the open pasture 
areas and two tree stocking rates at approximately 30 and 60 stemslha (Table 2.2) .  
As a former hill country research farm (Te Awa; 1945-1 969) the paddocks were 
intensively subdivided in relation to aspect (Suckling 1 954, 1975). Most experimental units 
(with and without trees) were located in four adjoining paddocks on a north-facing hillside, 
with an additional pair situated in two neighbouring paddocks over the hilltop on a 
southerly aspect (Table 2.2). 
Chapter 2 21  
The macro-topography varied between hill and steepland (Table 2.2). However, 
experimental units with higher tree stocking rates (especially >60 stemslha) were more 
associated with steepland, which also tended to be more heterogeneous and eroded 
(Table 2.2). This association was likely a consequence of the normal soil conservation 
practice of planting high tree stocking rates where erosion is severe and then widening the 
spacing as the trees extended out into more stable areas rv an Kraayenoord & Hathaway 
1 986) .  
All of  the experimental units were grazed similarly, with breeding ewes set stocked 
between lambing and weaning (August - December), and various mobs of sheep and cattle 
rotated on and off the paddocks, depending on the pasture covers, throughout the rest of 
the year (Clark per. cornm. 1 998). In late summer, cattle were often tactically used to 
remove any rank pasture cover (Clark per. cornm. 1 998). 
The paddocks were topdressed with superphosphate at a rate of 220-240 kg/ha/yr (average 
rate over the last 50 years) and only a single dressing of lime had been applied at 1 
tonnelha (Suckling 1975; Clark per. cornm. 1 998). 
2.3.2.2 Hautope 1 
Two large open pasture areas and ten unreplicated experimental units with trees were 
selected at this site (Table 2.3). The open and tree overstorey environments were located 
towards the upper hill slopes in two adjacent paddocks on a south-easterly aspect. The 
ground surface was generally smooth, but showed signs of previous soil erosion for the 
experimental units with trees (Table 2.3). 
Both of the adjacent paddocks were normally set stocked with breeding ewes just prior to 
lambing in mid August and remained so until weaning in NovemberlDecember. Rising 
two-year-old steers and heifers were normally added to the paddocks during November to 
help control peak spring pasture growth, and they remained set stocked with weaned lambs 
until late summer. Through autumn and winter all stock classes were rotated around the 
farm initially to 'clean up' rank summer pastures and then to maintain adequate pasture 
covers. 
Chapter 2 22 
Alternating between years, either sulfur superphosphate or single superphosphate was 
applied to the paddocks at a rate of300 kg!ha/yr (Meredith per. comm. 2001 ). 
2.3.2.3 Hautope 2 
Two large open pasture areas and four experimental units with trees, located within a 
single northwest-facing paddock, were selected at this site (Table 2.4). The experimental 
units were all positioned towards the base of the hillside, and their ground surface was 
generally terraced with little visual signs of soil erosion, except for at the highest tree­
stocking rate (Table 2.4). Shallow gullies separated the selected experimental units. 
Compared to Hautope 1 ,  lambing began a month earlier and accordingly breeding ewes 
were set stocked in late June. After weaning in NovemberlDecember the paddock was mob 
stocked with ewes for the rest of the year. Rising two-year-old heifers and steers were also 
finished on the farm; nevertheless they seldom grazed the selected paddock. 
The farm was topdressed with Dicalcic superphosphate at 200 kg!ha/yr (Dunderdale per. 
comm. 200 1 ). 
2.3.3 Overstorey densities 
The experimental units selected at each farm site provided a broad gradient of stand basal 
areas ranging between 0-43 m2/ha (Figure 2.2). The elimination process (refer to section 
2.2.2) allowed the increments in stand basal area at each site to be evenly spaced apart 
within a range typical of soil conservation and agroforestry plantings in New Zealand 
(Thompson & Luckman 1 993 ; Wilkinson 1 996; Knowles et al. 1 998; McElwee & 
Knowles 2000). Overall, both the range and spread of stand basal areas (used as an index 
of overstorey density) should aid in determining the appropriate form (linear or curvilinear) 
of the relationships to be studied. Considerable overlap in stand basal area was also present 
between the three commercial farm sites (Figure 2.2), which would help in comparisons 
between the two regions. 
The rate of change in stand basal area per unit increase in tree-stocking rate was 
significantly different between the three commercial farm sites (P<0.00 1 ;  Table 2.5). This 
Chapter 2 23 
most likely reflected differences in tree age, poplar hybrid, and the environmental site 
attributes for growing trees (Carron 1 968; Phillip 1 994). It also emphasises the 
unsuitability of tree stocking rate as an index of overstorey density where the relative size 
of the trees is not constant (Carron 1 968; Phillip 1 994). 
Table 2.2 Resource inventory for selecting experimental units at Kiwitea 
Experimental Paddock Tree-stocking Aspect Macro-slope Vertical Terrain Homogeneity Irregularities Selection 
lUlit rate (degrees) position on of ground 
(stemslha) hillside 
40 6 0 10  3 3 3 ... 
41  4 0 350 2 1 1 1 ... 
42 2 0 5 2 2 1 & 3 2 ... 
43 10  0 205 2 1 1 & 3 2 .. 
39 2 1 0  350 2 1 1 2 Poor drainage {man� rushes2 .. 
12  4 22 20 2 3 4 3 Tree debris 
36 3 25 5 3 2 1 & 3  2 ... 
25 3 27 10 2 2 1 2 E.U. shape 
30 3 29 350 2 2 1 2 ... 
23 3 32 345 3 4 1 & 3  3 ... 
1 4 34 15  2 1 1 2 Incomplete tree buffer 
8 3 38 335 2 1 1 & 3 1 Many impinging buffer trees 
38 3 39 350 2 1 1 & 3 1 ... 
1 0  4 4 1  50 3 2 4 3 
13  4 45 50 2 1 1 & 3  1 Stock camp 
14 4 47 50 2 2 1 & 3  1 
28 3 47 340 2 2 1 2 ... 
2 4 48 50 3 1 4 4 
9 3 49 355 3 2 4 3 IncOtnElete tree buffer 
29 3 5 1  0 2 1 1 2 .. 
1 1  4 56 340 2 2 4 2 Poor drainage, tree debris 
16  4 58 50 3 1 4 4 
33 3 60 3 1 0  3 1 1 4 Incomplete tree buffer 
26 3 62 0 2 2 1 & 3 1 Incomplete tree buffer 
7 3 63 355 2 1 1 2 ... 
3 4 64 350 3 4 4 3 Poor drainage {few rushes) ... 
1 5  3 73 340 3 1 3 4 
5 3 76 340 3 2 3 4 
24 3 77 5 3 2 4 2 Poor drainage {few rushes) ... 
27 3 83 350 3 2 3 4 
6 3 88 0 3 2 4 3 ... 
35 3 1 05 340 3 1 4 4 
4 5 1 19 1 50 3 1 4 4 Poor drain�e {few rushes) ... 
Rank: Macro-slope, < 12° ( 1), 12-28° (2), and > 28° (3); Vertical position on hillside, upper slope (I)  to lower slope (5); Terrain, smooth ground surface without ( 1 )  and with (2) 
signs of previous soil erosion, to a terraced ground surface without (3) and with (4) signs of previous soil erosion; Ground surface, homogeneous ( 1 )  to heterogeneous (5). 
I\.) � 
Table 2.3 Resource inventory for selecting experimental units at Hautope 1 
Experimental Paddock Tree Aspect Macro-slope Vertical Terrain Homogeneity Irregularities Selection 
unit stocking rate (degrees) position on of ground 
(stems/ha) hillside 
1 3 0 180 2 1 3 I • 
2 3 0 1 70 2 2 1 2 • 
1 02 2 21  240 2 4 1 2 
196 2 25 355 2 4 1 & 3 3 Incomplete tree buffer 
3 4 29 175 2 1 2 & 4  4 • 
72 3 53 1 30 2 3 4 3 
100 2 58 130 1 5 1 1 Poor drainage (many rushes) 
93 2 61 165 2 4 1 2 Sparse tree buffer 
4 4 63 145 2 2 2 2 Thistles • 
103 2 7 1  165 1 5 1 2 Poor drainage (many rushes) 
5 4 73 160 2 2 2 3 • 
56 2 74 165 1 5 1 1 Thistles 
6 4 74 1 10 2 1 4 2 • 
99 2 76 30 2 5 1 2 Incomplete tree buffer 
46 2 109 2 1 5  1 5 1 1 Incomplete tree buffer 
7 4 1 13 150 2 2 2 3 • 
1 77 2 125 155 1 5 1 1 Few thistles 
69 1 1 30 1 55 1 5 1 1 IncomElete tree buffer, thistles 
8 4 1 53 1 50 2 1 1 1 Few thistles • 
1 12 2 174 l i S 1 5 1 1 Incomplete tree buffer 
9 4 172 140 2 2 2 2 • 
175 2 193 135  1 5 1 1 Incomplete tree buffer 
70 4 209 120 2 3 2 & 4  3 Bare ground patches 
192 2 229 120 1 5 1 2 Many thistles 
10 4 222 125 2 2 2 3 • 
1 1  4 232 145 2 2 2 2 • 
82 4 267 1 25 2 4 2 1 Incomplete tree buffer 
1 89 4 337 135 2 2 2 & 4  2 
12 4 374 150 2 2 2 2 • 
Rank: Macro-slope, < 12° ( I), 12-28° (2), and > 28° (3); Vertical position on hillside, upper slope (1) to lower slope (5); Terrain, smooth ground surface without ( 1 )  and with (2) 
signs of previous soil erosion, to a terraced ground surface without (3) and with (4) signs of previous soil erosion; Ground surface, homogeneous (1)  to heterogeneous (5). After 
selection, the chosen experimental units were re-numbered 1-12. 
N c.n 
Table 2.4 Resource inventory for selecting experimental units at Hautope 2 
Experimental Paddock Tree stocking Aspect Macro slope Vertical Terrain Homogeneity Irregularities Selection 
unit rate (degrees) position on of ground 
(stems/ha) hillside 
1 1 0 330 2 3 3 1 '" 
2 1 0 275 2 4 1 & 3  1 '" 
3 1 30 3 10  2 4 3 3 '" 
1 79 2 47 90 2 3 1 3 
1 3 1  2 73 20 2 I 1 & 3  3 Incomplete tree buffer 
79 I 79 345 3 4 1 & 3 3 Incomplete tree buffer 
34 1 94 345 3 5 1 3 Very dense tree buffer 
146 2 98 20 2 5 1 & 3 4 Incom�lete tree buffer 
87 2 109 20 2 1 1 & 3 4 
103 2 1 1 6 20 3 5 1 2 Incomplete tree buffer 
4 1 130 340 2 4 1 & 3  3 '" 
93 2 1 52 90 2 2 1 & 3 4 Incomplete tree buffer 
5 1 1 7 1  345 2 4 1 & 3  3 • 
3 1  2 1 78 90 2 5 2 3 
1 86 1 1 8 1  345 2 5 2 3 
1 94 2 225 90 2 4 1 & 3 4 
100 2 240 90 3 5 1 & 3  4 
6 1 291 345 3 4 4 4 '" 
Rank: Macro-slope, < 1 2° ( 1 ), 12-28° (2), and > 28° (3); Vertical position on hillside, upper slope ( 1 )  to lower slope (5); Terrain, smooth ground surface without ( 1 )  and with (2) signs of previous 
soil erosion, to a terraced ground surface without (3) and with (4) signs of previous soil erosion;  Ground surface, homogeneous ( I )  to heterogeneous (5). After selection, the chosen experimental 
units were re-numbered 1 -6. 
N 
0) 
Chapter 2 27 
50 
40 
'(;' 
� 
"'S � 30 (1j 
� 
� 
� 20 .c 
] 
rJJ 
1 0  
0 
0 1 00 200 300 400 
Tree stocking rate (stemslha) 
Figure 2.2 The relationship between stand basal area and tree stocking rate at Kiwitea (.), Hautope 1 ( .. ), 
and Hautope 2 (D). The open circle (0) represents the south-facing experimental unit at Kiwitea, which was 
identified as an outlier in the regression analysis. 
Table 2.5 Regression parameters for P. x euramericana stand basal area (m2/ha) in relation to tree stocking 
rate (stemslha), at each commercial farm site, where Y = bX (constrained through origin). 
Site 
Kiwitea 
Hautope 1 
Hautope 2 
b (SE) 
0.465 (0.014) 
0 . 1 1 8  (0.009) 
0.043 (0.005) 
0.99 
0.95 
0.95 
RMSE 
2.54 1 
4.963 
1 .782 
n 
1 1  
10  
4 
All simple linear regressions were significant (p:S 0.003). Abbreviations: b, slope; SE, standard error; r2, 
adjusted coefficient of determination; RMSE, standard error of prediction; n, number of experimental units 
with trees. 
Chapter 2 28 
2.4 Conclusion 
The short time frame and the unavailability of extensive poplar stands, limited the number 
of approaches available for investigating the relationships between the overstorey density 
of a soil conservation poplar stand and its understorey abiotic and biotic components. 
Nevertheless, the careful selection of relatively small stands of poplar trees, along with the 
application of a stratified random sampling process, produced an even gradient of 
overstorey densities (measured as stand basal area) typical of soil conservation and 
agroforestry plantings in the Southern Hill Country of the North Island, New Zealand. The 
selection of farm sites in the Manawatu and Hawke's Bay regions provided contrasting 
summer-moist and summer-dry climates. The selection of stands from these two regions 
can be considered as broad types of replication with the purpose to increase or broaden the 
scope of inference for the study. Stage three of the elimination process helped to control 
extraneous sources of environmental variation found between the experimental units at 
each site. However, inherent in such soil conservation plantings is that higher tree stocking 
rates, and thus overstorey densities, tend to be more associated with previously eroded 
ground. The gradient of overstorey densities developed in this chapter forms the roam 
treatment structure for the experiments in Chapters three, four, and five. 
Chapter 2 29 
2.5 References 
Anderson, G.W. ; Batini, F.E. ( 1 983) Pasture, sheep and timber production from agro­
forestry systems with subterranean clover sown under 1 5-year-old Pinus radiata by a 
method simulating aerial seeding. Australian Journal of Experimental Agriculture & 
Animal Husbandry 23(121) :  123- 1 30. 
NZMS ( 1986) Rainfall observations for 1986: Stations in New Zealand, outlying Islands. 
New Zealand Meteorological Society, Wellington, New Zealand. 
Bird, P.R. ; Cayley, lW.D. ; Kearney, G.A. ; Aldridge, E.K. ( 1 994) Pasture growth among 
various tree species grown at various spacing. AFG Conference, pp. 297 - 300. 
Bleasdale, lK.A. ( 1967) Systematic designs for spacmg experiments. Experimental 
Agriculture 3 :  73 - 85. 
Burgess, S.M. (1 983) The climate and weather of Manawatu and Horowhenua. New 
Zealand Meterological Service, Wellington, New Zealand. 
Cameron, D.M.; Rance, S.J.; Jones, R.M. ; Charles-Edwards, D .A. ; Barnes, A. ( 1 989) 
Project STAG: an experimental study in agroforestry. Australian Journal of Agricultural 
Research 40(3):  699-714. 
Carron, L.T. ( 1 968) An outline of forest mensuration with special reference to Australia. 
Australian National University Press, Canberra, Australia. 
Crowe, S.R. ( 1 993) The response of Lolium perenne and Holcus lanatus to shading in 
relation to a silvopastoral agroforestry system. PhD thesis, Queen's University of Belfast, 
Ireland. 
Devkota, N.R.; Kemp, P.D.; Valentine, I . ;  Hodgson, J. ( 1 998) Performance of perennial 
ryegrass and cocksfoot cultivars under tree shade. Proceedings of the Agronomy Society of 
New Zealand 28: 129- 1 35 .  
Chapter 2 30 
Eastham, J. ; Rose, C.W. ( 1 990) Tree pasture interactions at a range of tree densities in an 
agroforestry experiment. 1. Rooting patterns. Australian Journal of Agricultural Research 
41(4): 683-695. 
Gilchrist, A.N.; deZ Hall, J.R; Foote, A.G.; Bulloch, B.T. ( 1 993) Pasture growth around 
broad-leaved trees planted for grassland stability. Proceedings of the XVII International 
Grassland Congress: 2062-2063. 
Guevara-Escobar, A.; Kemp, P.D.;  Hodgson, J. ; Mackay, A.D.; Edwards, W.RN. ( 1 997) 
Case study of a mature Populus deltoides-pasture system in a hill environment. 
Proceedings of the New Zealand Grassland Association 59: 179- 1 85.  
Harris, W. (1 994) Pasture as an ecosystem. Pastures: their ecology and management (ed 
Langer, R. H. M.), pp. 75 - 1 3 1 .  Oxford University Press, Auckland, New Zealand. 
Hoffinann, C.W.;  Usoltsev, V.A. (2002) Tree-crown biomass estimation in forest species 
of the Ural and ofKazakhstan. Forest Ecology and Management 158 :  59-69. 
Johansson, T. (2002) Increment and biomass in 26- to 9 1-year-old European aspen and 
some practical implications. Biomass and Bioenergy 23 : 245-255 .  
Kellas, lD.; Bird, P.R; Cumming, K.N.; Kearney, G.A. ; Ashton, A.K. (1 995) Pasture 
production under a series of Pinus radiata-pasture agroforestry systems in South-West 
Victoria, Australia. Australian Journal of Agricultural Research 46(6): 1285-1297. 
Kleinbaum, D.G. ; Kupper, L.L. ;  Muller, K.E. ;  Nizam, A. ( 1998) Applied regression 
analysis and other multivariable methods. Third edition. Duxbury Press. Brooks/Cole 
Publishing Company, California, USA. 
Knowles, R.L.; Horvath, G.c. ; Carter, M.A.; Hawke, M.F. ( 1999) Developing a canopy 
closure model to predict overstorey/understorey relationships in Pinus radiata 
silvopastoral systems. Agroforestry Systems 43(1) :  1 09-1 19. 
Chapter 2 3 1  
Littell, RC.; Freund, RJ.; Spector, P.c. ( 1991)  SAfiID System for linear models. Third 
edition. SAS Institute Inc., North Carolina, USA. 
MAF (199 1). Farm monitoring report: north central region .  Ministry of Agriculture and 
Fisheries, Palmerston North, New Zealand. 
McCaskill, L. W. ( 1973) Hold this land: a history of soil conservation in New Zealand. 
A.H. & A. W. Reed, Wellington, New Zealand. 
McElwee, H.F.; Knowles, RL. (2000) Estimating canopy closure and understorey pasture 
production in New Zealand-grown poplar plantations. New Zealand Journal of Forestry 
Science 30(3): 422-435. 
Myers, R.H. (1 990) Classical and modern regression with applications. Second edition. 
PWS - Kent Publishing Company, Boston, USA. 
Nelder, J.A. (1 962) New kinds of systematic designs for spacing experiments. Biometrics 
1 8 : 283 - 307. 
Ong, C.K. ( 1996) A framework for quantifying the various effects of tree-crop interactions. 
Tree-crop interactions: a physiological approach (eds Ong, C.K. & Huxley, P.), pp. 1 - 23. 
ICRAF and CAB International. University Press, Cambridge, UK. 
Ong, C.K.; Black, CR.; Marshall, F.M.; Corlett, J.E. (1996) Principles of resource capture 
and utilisation of light and water. Tree-crop interactions: a physiological approach (eds 
Ong, C.K. & Huxley, P.), pp. 73 - 1 58. ICRAF and CAB International. University Press, 
Cambridge, UK. 
Percival, N.S.; Knowles, R.L. (1 988) Relationship between radiata pine and understorey 
pasture production. Agroforestry symposium proceedings (ed Maclaren, lP.), pp. 1 52-1 64. 
Forest Research Institute and New Zealand Ministry of Forestry, Rotorua, New Zealand. 
Phillip, M.S. ( 1 994) Measuring trees and forests. Second edition. CAB International. 
University Press, Cambridge, UK. 
Chapter 2 32 
Pohlen, I.l. ( 1 97 1 )  Soils of Hawke's Bay region New Zealand. New Zealand Department of 
Scientific and Industrial Research, Wellington, New Zealand. 
Pollock, K.M.; Lucas, R.l.; Mead, D.l; Thomson, S.E. ( 1 994) Forage-pasture production 
in the first three years of an agroforestry experiment. Proceedings of the New Zealand 
Grassland Association 56: 1 79- 1 85 .  
Pyke, D.A. ;  Zamora, B.A. ( 1 982) Relationships between overstory structure and 
understory production in the Grand FirlMyrtle Boxwood habitat type of  Northcentral 
Idaho. Journal of Range Management 35(6): 769 - 773 . 
Reed, D.D.; Mroz, G.D. ( 1 997) Resource assessment inforested landscapes. John Wiley & 
Sons, Inc., New York, USA. 
Remrnenga, E.E. ( 198 1 )  Statistical approaches to studies involving annual crops. 
HortScience 1 6(5): 63 1 - 633 .  
Rijkse, W.C. ( 1 977) Soils of Pohangina County North Island, New Zealand. New Zealand 
Department of Scientific and Industrial Research, Wellington, New Zealand. 
SAS ( 1991 )  SA� System for Regression. Second edition. SAS Institute Inc., Cary, North 
Carolina, USA. 
SAS ( 1 999) Statistics version 8. 02 for Windows®. SAS Institute Inc., Cary, North Carolina, 
USA. 
Scanlan, J.c. ; Burrows, W.H. ( 1990) Woody overstorey impacts on herbaceous 
understorey in Eucalyptus spp. communities in central Queensland. Australian Journal of 
Ecology 1 5 :  1 9 1  - 197 .  
Scanlan, lC. ( 1991)  Woody overstorey and herbaceous understorey biomass in Acacia 
harpophylla (brigalow) woodlands. Australian Journal of Ecology 16: 521 -529. 
Chapter 2 33 
Sibbald, A.R.; Griffiths, J.H.; Elston, D.A. ( 1991)  The effects of the presence of widely 
spaced conifers on under-storey herbage production in the U.K. Forest Ecology & 
Management 45 : 7 1  - 77. 
Steel, R.G.D.;  Torrie, lH. ( 1 980) Principles and procedures of statistics, a biometrical 
approach. McGraw Hill Kogakusha, Ltd, Tokyo, Japan. 
Studman, C.l ( 1990) Agricultural and horticultural engineering: principles, models, 
systems and techniques. Butterworths of New Zealand Ltd, Wellington, New Zealand. 
Suckling, F.E.T. ( 1950) Manawatu Catchment Board Te Awa soil conservation 
experimental area: Report on operations from the initiation of the trials in 1945 to end of 
January 1949. Manawatu Catchment Board and Soil Conservation and Rivers Control 
Council, Wellington, New Zealand. 
Suckling, F.E.T. ( 1 954) Pasture management trials on unploughable hill country at 
Te Awa: 1 .  Establishment of experimental area and results for 1 949-5 1 .  New Zealand 
Journal of Science and Technology 36( 1 ) :  237 - 273 . 
Suckling, F.E.T. ( 1975) Pasture management trials on unploughable hill country at 
Te Awa. 3 .  Results for 1 959-69. New Zealand Journal of Experimental Agriculture 3(4): 
35 1 -436. 
Terralink ( 1 996) Kimbolton 260 - T23. Land Information New Zealand, Wellington, 
New Zealand. 
Terralink ( 1 997) Waipawa 260 - V22. Land Information New Zealand, Wellington, 
New Zealand. 
Thompson, C.S. ( 1 987) The climate and weather of Hawke's Bay. New Zealand 
Meterological Service, Wellington, New Zealand. 
Thompson, R.C.; Luckman, P .G. ( 1 993) Performance of biological erosion control in New 
Zealand soft rock hill terrain. Agroforestry Systems 2 1(2): 1 9 1 -2 1 1 .  
Chapter 2 34 
Urquhart, N.S. ( 1 98 1 )  The anatomy ofa study. HortScience 1 6(5): 621-627. 
van Kraayenoord, C.W.S. ( 1 993) Poplar growing in New Zealand - Past to present. A 
potential for growth: proceedings of a meeting to reactivate the New Zealand National 
Poplar Commission (ed Bullock, B.), pp. 9 - 1 5. Manaaki Whenua - Landcare Research 
New Zealand Ltd., Canterbury, New Zealand. 
Van Kraayenoord, C.W.S. ;  Hathaway, R.L. ( 1 986) Plant materials handbook for soil 
conservation volume 1. Principles and practices. Water and soil Miscellaneous Publication 
No. 93 . National Water and Soil Conservation Authority, Wellington, New Zealand. 
White, J.G.H. ( 1 994) Hill and high country pasture. Pastures: their ecology and 
management (eds Langer, R.H.M.), pp. 299 - 336. Oxford University Press, Auckland, 
New Zealand. 
Wilkinson, A. ( 1 996) The poplar scene - New Zealand and overseas. New Zealand Tree 
Grower 1 7( 1 ): 28-30. 
Wilkinson, A.G. (1 988) National report on activities related to poplar and willow 
cultivation period: 1985 - 1988. New Zealand National Poplar Commission, Palmerston 
North, New Zealand. 
Zwartz, D. ( 1 997) New Zealand Of icial Yearbook 1997. GP Publications, Wellington, 
New Zealand. 
Chapter 3 35 
3 The relationship between poplar stand density 
and PAR transmission 
CONTENTS 
3.1 Introduction 
3.2 Methodology 
3 .2 . 1 Treatments 
3 .2.2 Overstorey (stand) density indices 
3 .2.3 Light measurements 
3.3 Data analysis 
3.4 Results 
3 .4. 1 %DIFN within the vertically projected gap between fully in-leaf poplars 
3 .4.2 %DIFN directly underneath a fully in-leaf poplar crown 
3 .4.3 %DIFN within the vertically projected gap between leatless poplars 
3 .4.4 %DIFN directly underneath a leafless poplar crown 
3 .4.5 Selective relationships between stand density indices 
3 .4.6 Weighted (stand level) %DIFN over a range of canopy closures 
3 .4.7 The relationship between R:FR and %DIFN 
3.5 Discussion 
3 .5 . 1  The relationship between stand density indices and Zone 3 %DIFN 
for summer and winter 
3 .5 .2 The relationship between stand density indices and Zone 1 %DIFN 
for summer and winter 
3 .5 .3 Practical considerations 
3 .5 .4 Relationships between the best stand density indices for estimating 
PAR transmission (%DIFN) 
3.5 .5 Stand level %DIFN across a range of canopy closure ratios in summer 
and winter. 
3 .5.6 Potential error incurred through using %DIFN measurements 
3 .5 .7 The relationship between R:FR and %DIFN 
3.5 .8  The ecological significance of changes in the understorey light 
environment for pasture plants. 
3.6 Conclusion 
3.7 References 
36 
39 
39  
39  
45 
48 
50 
50 
56  
60 
66 
7 1  
73 
75 
77 
77 
82 
85 
85 
90 
9 1  
94 
98  
108 
110  
Chapter 3 36 
3.1 Introduction 
A central objective of most silvopastoral systems is to increase, or at least sustain over the 
long term, the total productive capacity of land through either more efficient utilisation 
and/or greater conservation of available resources (Wojtkowski 1998). This is normally 
attempted by mixing trees and understorey pastures that collectively enhance the stability of 
the environment, while on an interspecies basis impose different demands on the system's 
available resources (Sinclair et al. 2000). However, scope for the latter niche separation in 
mixed tree-pasture systems is somewhat limited, as both components usually require similar 
resources (energy, water, mineral nutrients) for growth and development (Ong 1996; Ong 
et al. 1996). 
Photosynthetic ally active radiation (PAR) is one resource where there is usually a strong 
overlap in requirements from both the overstorey and understorey plant species (Hanan & 
Begue 1995; Gendron et al. 1998; Lieff'ers et al. 1999). PAR is commonly defined as short­
wave radiation in the visible part of the spectrum, ranging between 380 nm to 710 nm in 
wavelength (Larcher 1980); and a critical determinant of its availability for photosynthesis 
at different levels within a mixed tree-pasture association is determined by the three­
dimensional structure and optical properties of the overstorey canopy (Grace et al. 1987; 
Sibbald & Griffiths 1992; Lieff'ers et al. 1999). 
Short-wave solar radiation reaching understorey pasture consists of two forms: direct-beam 
radiation, emitted relatively unimpeded from the sun to the understorey; and diffuse 
radiation, circuitously transmitted from all hemispherical directions, owing to scattering in 
the atmosphere andlor downward reflectance from and transmittance through leaves in the 
overstorey canopy (Federer & Tanner 1966; Reifsnyder et al. 1971; Hutchison & Matt 
1976, 1977; Endler 1993). Both forms of short-wave solar radiation passing directly 
through canopy gaps, without impinging on plant parts, are rich in PAR (Larcher 1980; 
Barnes et al. 1997). This contrasts with the reflected and through-leaf transmitted radiation, 
which is often depleted in photosynthetically active wavelengths as a result of selective 
absorption by foliage (Hughes et al. 1985; Wilson & Ludlow 1991; Barnes et al. 1997). In 
particular, ultraviolet, blue, and red wavelengths are highly absorbed by foliage compared 
Chapter 3 37 
with green, and especially far-red wavelengths (Messier & Bellefleur 1 988; Wilson & 
Ludlow 1 99 1 ;  Messier & Puttonen 1 995; Barnes et al. 1 997). 
A variety of external and internal factors govern the incidence ofP AR transmitted through a 
canopy (Hutchison & Matt 1 977; Lieffers et al. 1999). The incidence of short-wave solar 
radiation reaching the top of a canopy varies regularly with diurnal and seasonal changes in 
earth-sun geometry (seasonal variations increasing at higher latitudes), and irregularly with 
changing atmospheric conditions (Hutchison & Matt 1 977b; Canham 1 988; Sturman & 
Tapper 1 996; Comeau et al. 1 998; Lieffers et al. 1999). The subsequent transmission ofthis 
incident radiation to different parts of the understorey depends on its source (diffuse or 
direct-beam radiation) and the structure and optical properties of the overstorey canopy 
(Hutchison & Matt 1977b; Barnes et al. 1 997; Gendron et al. 1 998). 
Typical plantings of soil conservation trees in New Zealand's North Island Hill Country 
encompass a wide range of overstorey densities. Plantings range from almost single 
independent trees, with very large inter-tree gaps, to closed-canopy forest (Van 
Kraayenoord & Hathaway 1 986; Thompson & Luckman 1 993 ; Wall et al. 1 997; Wilkinson 
1 999). As a result of the trees spatial arrangement complex gradients of PAR transmission 
within and around adjacent canopy openings can be formed, depending on the size and 
orientation of the gap, dimensions of the surrounding trees, sun angle, sky conditions, and 
topography of the ground surface (Canham 1988; Canham et al. 1 990; Sibbald & Griffiths 
1 992; Runkle et al. 1 995; Lieffers et al. 1 999). 
Several methods are available for directly measuring both direct-beam and diffuse PAR 
transmission through scattered tree canopies (WUnsche et al. 1 995; Gendron et al. 1998; 
Lieffers et al. 1 999). To account for the high spatial and temporal variation, caused by the 
different forms of short-wave radiation, these methods usually involve either integrated 
arrays of quantum sensors placed simultaneously underneath and outside/above the canopy 
(Hutchison & Matt 1977a,b; Sibbald & Griffiths 1 992), or alternatively fewer sensors 
moved to various sampling positions with repeated readings taken over time (Chen et al. 
1 997; Hassika & Berbigier 1 998). 
Chapter 3 38 
However, intensive sampling over an extended period, with a large amount of specialised 
and often expensive equipment, limits the total number of experimental units (micro sites) 
that can be accurately sampled at any one time (Rich et al. 1 993 ; Wfulsche et at. 1 995 ;  
Comeau et al. 1 998; Gendron et  al. 1 998; Lieffers et  al. 1 999; Machado & Reich 1999; 
Englund et al. 2000). Thus, alternative instantaneous sampling techniques have been used as 
a way to indirectly estimate the average fraction of incident PAR transmitted or intercepted 
by a canopy (Anderson 1964a; Canham 1 988 ;  Chen et al. 1 99 1 ;  Hanan & Begue 1 995;  
Parent & Messier 1 996; Gendron et al. 1 998;  Lieffers et al. 1 999). Simulation models based 
on individual tree or stand architecture provide another alternative (Comeau et al. 1 998; 
Gendron et at. 1 998). 
While farmers have no control over the incidence of above-canopy PAR reaching soil 
conservation stands, they can control its transmission to the under storey pastures by 
manipulating the density of the overstorey tree canopy (e.g. by thinning, pruning, & 
pollarding). Given the importance of PAR transmission, as one of the key ecological 
processes driving understorey pasture production within temperate silvopastoral systems, 
the objective of this chapter is to define the relationship between the overstorey density of 
a poplar stand and a simplified estimate of incident PAR reaching the understorey. 
A number of different stand parameters are commonly used as overstorey density indices in 
both agroforestry and forestry (Carron 1 968; Schreuder et at. 1 993; Phillip 1994; Reed & 
Mroz 1 997). Thus, in addition to the main objective, a selection of commonly used 
overstorey density indices will also be evaluated for their simplicity, power, and robustness 
as predictors of estimated PAR transmission. 
Chapter 3 39 
3.2 Methodology 
3.2.1 Treatments 
The broad range of overstorey (stand) densities previously developed in Chapter 2 formed 
the main treatment structure for this study. An additional 1 8  experimental units from within 
a systematic-spacing 'Nelder' experimental design (Nelder 1 962), planted at AgResearch's 
Ballantrae Hill Country Research Station, were also sampled. Measurements were taken 
from 2 spokes of the fan shaped design. Each spoke contained 9 experimental units, ranging 
in size from 50 to 750 stemslha. This extra site further improved the overall spread of the 
overstorey densities, especially towards the lower density range. The Ballantrae site 
(Lat. 40. 1 8°S Long. 175 .50�) was planted with 'Tasman' poplars (P. x euramericana), on 
a gently sloping northeasterly facing hillside, at 1 3  5 m above sea level. When sampled the 
trees were five years old and their stems had been pruned to 3.65m ± 0.05m (mean ± SEM, 
n=72). 
3.2.2 Overstorey (stand) density indices 
In Chapter 2, the overstorey density! above each experimental unit was initially described in 
terms of stand basal area (m2/ha) (refer to Section 2.2.3). This is one of the main stand 
density indices traditionally used in forest resource inventories, as it is usually strongly 
related to stand volume or biomass (Carron 1 968; Phillip 1 994). However, the above 
relationship does not remain constant under the intensive silvicultural management usually 
associated with silvopastoral systems. For example, the pruning of lower tree branches can 
significantly reduce the density of a stand's overstorey, while at the same time not affecting 
its overall basal area (percival & Knowles 1 988). Thus, several alternative stand parameters 
were measured, which potentially could give a more integrated measure of both tree 
stocking rate (stemslha) and individual crown size. Measurements were taken from the four 
nuclei trees of each experimental unit (with trees) and were scaled up to a per unit area 
basis (or representative stand level) via their respective tree stocking rates (refer to Section 
2.2.2.2). 
I Overstorey density: the amount of tree material in a stand per unit area (Carron 1968). 
Chapter 3 40 
3.2.2.1 Stand parameters based on the height of the tree crowns 
The vertical height from ground level to the top (total height) and base (crown height) of 
the live tree crowns was measured either directly with a telescopic measuring pole (Senshin 
Industry Co.) or indirectly with a hypsometer (Suunto Precision Instruments). Ignoring 
epicormic side shoots, the base of a crown was the point where foliage occupied at least 
three of the four quadrants around the stem (Canham et al. 1 999). The green crown length 
(GCL) per stem, otherwise known as crown depth, was calculated as the difference between 
the total height and crown height (Carron 1 968; Phillip 1 994; Ellis & Hayes 1 997; Reed & 
Mroz 1 997): 
GeL/stem = total height - crown height (units: mlstem). 
3.2.2.2 Stand parameters based on the horizontal crown diameter 
The mean diameter of each tree crown was calculated from two horizontal tape 
measurements, which were taken at perpendicular angles to each other (the first direction 
determined at random). Two people were required for this measurement - with one person 
standing away from the tree, aligning a plumb-line on to its outermost edge, while the other 
marked the vertically projected point onto the ground (Carron 1 968; Schreuder et al. 1 993). 
This measurement allowed the horizontal cross-sectional area of the tree crowns (m2/stem) 
to be calculated using the same geometrical formula and assumptions used for determining 
stem basal area (refer to section 2.2.3 . 1 ). However, for this calculation, the maximum 
horizontal crown diameter was limited to within the mean distance between the trees for a 
particular experimental unit, based on the assumption that adjacent tree crowns do not 
overlap (Figure 3 . 1 ). 
When scaled up to a per unit area basis the above stand parameter gives a simple estimate of 
the canopy closure ratio on a horizontal plane (Schreuder et al. 1 993). Based on the 
assumptions of non-overlapping tree crowns with equal circular cross-sectional areas, and 
the trees planted on a square grid pattern, the theoretical maximum horizontally projected 
canopy closure ratio (HPCC) for the scaled up stands is 78.5% (Assmann 1970; Phillip 
1 994). 
Chapter 3 
Maximum horizontal 
crovvn diameter 
41 
Figure 3.1 A schematic aerial view of four 'nuclei' trees (grey circles) defining the boundary (dashed lines) 
of an experimental unit. The total area of an experimental unit represents the expanded canopy gap and the 
non-shaded area within the centre of the experimental unit represents the vertically projected canopy gap 
(Runkle 1982; Runkle et al. 1995). 
3.2.2.3 Stand parameter based on the volume of the tree crowns 
The ellipsoidal volume of each tree crown (CEV) was calculated from the measurements of 
crown diameter and green crown length using the formula: 
(units: m3/stem) 
where R = horizontally projected green crown radius per stem 
H = half the vertical green crown length (depth) per stem 
A solid vertical ellipse with perfect symmetry was assumed to represent the shape of the 
individual tree crowns (Wilkinson 1995; Karlik & Winer 1 999; Stadt & Leiffers 2000). 
Chapter 3 42 
3.2.2.4 Stand parameter based on colour digital images of the overstorey canopy 
A series of colour digital images were taken to estimate the degree of canopy closure (CC) 
above each experimental unit. All of the images were captured with a Sony MVC-FD7 
digital camera, set to view vertically upwards with the aid of a spirit-bubble-level, and 
mounted at 0.7 m above the ground (measured from the top of the lens) on a height 
adjustable tripod (Becker et al. 1 989; Bunnell & Vales 1 990; Englund et al. 2000). At the 
time of use, the spatial resolution (VGA 640 x 480 pixe1s2) of the Sony camera's image was 
not the highest available, compared to other digital cameras, for a sharper definition of the 
boundary between foreground-canopy and background-sky (Jahne 1 997; Wagner 1 998; 
Frazer et al. 200 1 ). However, its system of JPEG3 ( 1 :4 compression) image recording 
directly onto a low-cost 3.5-inch floppy disk effectively provided the camera with an 
unlimited memory capacity, enabling large numbers of images to be recorded relatively 
quickly at the remote farm locations (aided by carrying an extra rechargeable battery). All of 
the images were captured with the camera's standard lens (producing rectangular or non­
equidistant images), and were taken preferably under uniformly overcast sky conditions 
(e.g. Stratus clouds) to maximise image contrast and to minimise interference by direct 
sunlight (Roxburgh & Kelly 1 995; Wiinsche et al. 1 995; Englund et al. 2000; Frazer et al. 
2001 ). 
Field sampling 
The spacing between the four 'nuclei' trees, used to define an experimental unit, created an 
overstorey canopy that was naturally more open towards the plot centre (Figure 3 . 1 ). 
Therefore, estimating the average canopy closure by random sampling was inappropriate, as 
there was a chance the majority of sampling points could have been concentrated under a 
dense or open part of the overstorey canopy (Knowles et al. 1 999). Instead, to provide 
complete coverage of the canopy above each experimental unit, sampling points were 
restricted using a systematic star-shaped design developed by Knowles et al. ( 1 999t 
2 A pixel or picture element is the smallest individual unit of a digital image. Thus, image matrices with 
greater numbers of pixels have higher image resolution (McKennan 1 995). 
3 JPEG (Joint Photographic Experts Group) is a standardised compression technique used to reduce the size 
of a digital image (Frazer et al. 2001 ). 
4 Messier et al. ( 1998) also used a similar systematic sampling design for light measurements in boreal 
forests. 
Chapter 3 43 
Seventeen non-overlapping digital canopy images were taken for each experimental unit: at 
the centre of the plot, and at distances of 6 m and 12  m, radiating outwards along the eight 
major compass directions from the central point (Figure 3 .2). 
s 
E __ ---tt--__ -----1 __ -__ W 
12 m 
N 
Figure 3.2 Sample grid overlaid and centred on an experimental unit to estimate canopy closure 
(Knowles et al. 1999). 
To account for the deciduous nature of poplars, all experimental units were sampled when 
the trees were fully with (summer) and without (winter) leaves (Table 3 . 1 ). No 
measurements were made through spring and autumn because of the changing phenology of 
the overstorey canopy and the associated difficulty in determining a single point in time 
representative of these two seasons. 
Table 3.1 Dates when canopy closure was measured in the field. 
Site 
Kiwitea 
Hautope 1 
Hautope 2 
Ballantrae 
Summer 
2510111999 
13102/2000 
13/02/200 1 
2/03/2001 
Date 
Digital image processing and analysis 
Winter 
27/0811 999 
1 3/09/2000 
1 3/09/2000 
810912000 
The colour digital images obtained in the field were analysed using the methods and 
mainstream software packages described by Guevara-Escobar ( 1999). Essentially, digital 
image processing and analysis followed three stages: firstly, identifying and separating the 
Chapter 3 44 
overstorey canopy and background sky areas of an image through their respective colours 
(segmentation process); secondly, reducing these two colour-defined areas to purely black 
and white, respectively; and thirdly, counting the total number of pixels within each 
simplified category to give an estimate of canopy closure (i.e. the proportion of sky 
obscured by the poplar overstorey). Overall, the above method shared many similarities with 
the methodology used by Knowles et al. ( 1 999), McElwee & Knowles (2000), and Englund 
et al. (2000). 
The sky and canopy pixels of an image were differentiated using a colour filter created for 
each experimental unit in Corel Photo-Paintlil (version 9.0 for Windowslil, Corel Corp. 
1 999). An individual filter consisted of a palette of representative sky colours manually 
selected directly from an image via the software's colour masking function. A threshold 
value was set for the selected colours so that each would include 20 of their closest 
neighbours within the available colour spectrum. This reduced the total number of 
distinctive colours, and thus time, needed for identifying pixels belonging to the sky portion 
of an image. The variable-zoom function in Corel Photo-Paintlil aided in the investigation of 
finer image detail, but similar to the results of Englund et af. (2000), solely focusing on a 
small portion of the image often caused inaccuracy in separating other parts of an image. 
A semitransparent red overlay was added (tagged) to all of the pixels associated with the 
selected colours to visually check the precision of the segmentation process. Overall, 3 to 
1 0  colours were usually selected for separating the sky and canopy pixels of an image; 
depending on the degree of contrast between the two elements and also their respective 
uniformity of colour. Similar to Englund et af. (2000) and Frazer et al. (2001 ), images 
captured on completely overcast days were preferred, as this reduced the negative effects of 
colour halos and blurring of canopy edges. Once the filter (colour mask) was completed, it 
was then saved as a standard template for the remaining images of an experimental unit. 
Nevertheless, the precision of the standard filter was visually rechecked for each image and 
the original set of base colours were modified where necessary. Potential bias or non­
random variation, caused by human error, was minimised by having one-person process and 
analyse all of the digital images (Canham et al. 1990; Gendron et al. 1998). 
The individually tailored filters (colour masks) protected image pixels that fell within the 
colour ranges specified, but could also be inverted so protecting either the sky or overstorey 
Chapter 3 45 
canopy areas of an image. This enabled the entire originally unprotected canopy area to be 
converted to black, the mask inverted, and the remaining sky area converted to white. These 
simplified black and white binary images were saved as new files, and the black canopy 
pixels were identified and counted in SigmaScan@ Pro (version 4.0 for Windows@, SPSS 
Inc. 1 998). The counted number of black canopy pixels divided by the total number of 
pixels within the image matrix (640 x 480 pixels) represented the canopy closure ratio (CC). 
Overall, the estimated canopy closure ratio (CC) for an entire experimental unit was based 
on the mean of the seventeen images obtained in the field (Knowles et al. 1 999; McElwee & 
Knowles 2000). 
3.2.3 Light measurements 
3.2.3.1 Photosynthetically active radiation 
The percentage of above-canopy photosynthetically active radiation (PAR) transmitted 
through the tree canopy was estimated from diffuse non-intercepted radiation5 (DIFN) 
measurements taken with paired LAI-2000 Plant Canopy Analysers (LI-COR Inc., Lincoln 
NE, USA). These sensors measure short-wave radiation of less than 490 nm, 
simultaneously, from five integrated concentric annuli of the upward hemisphere, covering a 
1 500 field-of-view (Welles & Norman 1 991 ; Hanan & Begue 1 995; Lieffers et al. 1 999). 
Under a wide range of different canopy structures (or light gradients), several researchers 
have shown that instantaneous readings of understorey DIFN (Iu), relative to simultaneous 
above-canopy readings (10), can be used as a reliable estimate of PAR transmission (Hanan 
& Begue 1 995; Comeau et al. 1 998; Gendron et al. 1 998; Lieffers et al. 1 999; Machado & 
Reich 1 999): 
Estimated PAR transmission (%DIFN) = (Iu / 10) x 1 00 (units: %) 
This measurement is conceptually similar to the diffuse site factor obtained from 
hemispherical (fisheye) canopy photographs/images and also paired quantum sensor 
readings, taken simultaneously from above and below a canopy on completely overcast days 
(Messier & Puttonen 1 995; Parent & Messier 1 996; Gendron et al. 1 998; Messier et al. 
5 Diffuse non-intercepted radiation: diffuse short-wave radiation that passes directly through gaps in the 
canopy and is not scattered by or transmitted through the canopy biomass. 
Chapter 3 46 
1 998; Machado & Reich 1 999). However, an important difference is that the latter method 
accounts for scattered diffuse (transmitted and down-reflected) radiation from the 
overstorey, whereas the former two methods treat all overstorey elements (e.g. trunks, 
branches, leaves) effectively as opaque (Becker et al. 1 989; Rich et al. 1 993; Hanan & 
Begue 1 995; Roxburgh & Kelly 1 995; Gendron et al. 1 998;  Lieffers et al. 1 999). 
%DIFN measurements were taken using two synchronised and cross-calibrated LAI-2000 
Plant Canopy Analyser units; one was moved around the various sampling positions of 
interest, while the other was positioned permanently in a nearby open paddock, as an above­
canopy reference. Both LAI-2000 units had their short-wave radiation sensor fields of view 
(azimuth) restricted to 1 80 degrees, and were orientated to the general aspect of the farm 
site (refer to Table 2. 1). The view restriction prevented the LAI-2000 sensors from 'seeing' 
the operator or direct sunlight (Welles & Norman 1 991 ; Hanan & Begue 1995 ; Comeau et 
al. 1 998; Gendron et al. 1 998). At each sampling point 8 %DIFN measurements were taken 
with the portable LAI-2000 unit levelled horizontally at approximately one metre above the 
ground (Welles & Norman 1 99 1 ). All %DIFN measurements were obtained within a month 
of taking the digital canopy images (refer to Table 3 . 1 ), except for the summer %DIFN 
measurements at Ballantrae, which were obtained in the same month but in the year prior 
(this exception was due to the unavailability of equipment). 
3.2.3.2 Red to Far-red ratio 
The red to far-red ratio (R:FR) was measured using a single-point sensor (Skye Instruments 
LTD., Llandrindod Wells, Powys, U.K.) to determine any changes in the quality of light 
reaching the understorey. Sampling was carried out at the same time and place as the 
%DIFN measurements. 3 R:FR readings were taken from each sampling position. No R:FR 
measurements were taken at Kiwitea in the winter period. Results are presented as the R:FR 
relative to a reference open pasture (above-canopy) reading, averaged from measurements 
taken at the beginning and end of sampling at each site. 
3.2.3.3 Sampling positions within an experimental unit 
As mentioned in Section 3.2.2.4, the canopy of the poplar stands was often not continuous 
and usually contained a patchwork of tree crowns and inter-tree gaps. For this experiment, 
Chapter 3 47 
instead of studying a limited number of overstorey density treatments in detail, it was opted 
to study a much larger range less comprehensively. The sum of the interactions in mixed 
tree-pasture systems is likely to be minimised mid-way between the trees and, conversely, 
the most intense within close proximity to the trees (Runkle et al. 1 995; Ong et at. 1 996). 
Therefore, based on this premise, two extreme overstorey environments (or zones of tree 
influence) were chosen for sampling in each experimental unit. The first overstorey 
environment (Zone 1 )  was defined as the area directly below the vertical projection of the 
tree crown on the northeastern corner of an experimental unit (Figure 3 .3). This zone was 
positioned on the shaded side of the tree and received direct-beam radiation later in the day 
(compared to the opposite tree on the north-western corner of an experimental unit), when 
leaf and air temperatures, along with atmospheric humidity were likely to be less conducive 
for photosynthesis and plant growth (Wayne & Bazzaz 1 993). The second overstorey 
environment (Zone 3) was defined as the vertically projected gap (vpG; Ban et at. 1 998) 
between the crowns of the 4 nuclei trees (Figure 3 .3).  All light measurements were taken at 
the centre of each overstorey environment (zone of tree influence). 
Figure 3.3 Overstorey environments within the boundary of an experimental unit: Zone 1 - area within the 
expanded canopy gap, directly underneath the crown of the 'nucleus' tree located at the north-eastern corner 
of an experimental unit; and Zone 3 - area at the centre of the vertically projected canopy gap (also refer to 
Figure 3. 1) .  
Chapter 3 48 
3.3 Data analysis 
Linear and non-linear regression analyses, using the REG and NLIN procedures of SAS® 
(version 8.02 for Windows®, SAS Institute, Inc. 1 999), were performed to detennine 
relationships between the overstorey (stand) density indices (Table 3 .2), as single 
independent (regressor) variables, and estimated PAR transmission (%DIFN) , as the 
dependent (response) variable. Data from the 4 sites were pooled together, and separate 
regressions carried out for the 2 zones of tree influence (Zones 1 & 3) and 2 seasons 
(summer & winter). CurveExpert° (version 1 .37 for Wmdows®, Hyams 200 1 ), scatter plots, 
and applicable published ecological studies, were used to detennine potential regression 
models for the data. 
Table 3.2 Overstorey (stand) density indices 
Abbreviation 
DBH 
BA 
GCL 
HPCD 
HPCC 
CEV 
CCL 
CCNL 
Units 
mlha 
m2/ha 
mlha 
mlha 
% 
'000 m3/ha 
% 
% 
Definition 
Sum of stem diameters over bark at breast height ( l .4m) per hectare 
Sum of stem basal areas at breast height ( l .4m) per hectare 
Sum of vertically projected crown lengths per hectare 
Sum of horizontally projected crown diameters per hectare 
Horizontally projected canopy closure ratio 
Sum of stem crown ellipsoidal volumes per hectare 
Canopy closure ratio when trees are fully in leaf 
Canopy closure ratio when trees are completely without leaves (no leaf) 
For each overstorey density index, the simplest model with the smallest standard error of 
prediction (RMSE) , a high adjusted coefficient of detennination6, and no trend in the 
residuals was selected (Vales & Bunnell 1 988; Gendron et al. 1 998). The adequacy (quality 
of fit) of each regression model was checked through inspection of scatter, residual, and 
normal probability plots, along with more formal test statistics. Some of the independent 
variables were either natural logarithm or square root transformed to help meet the 
underlying model assumptions (as indicated in the appropriate tables). Systematic trends in 
6 Approximated r for the non-linear models: I -SSE/CSS, where SSE is the error sum of squares of the full 
model and CSS is the corrected total sum of squares. Adjusted r = 1 - ( l  - r) * [en - 1 )  I (n - m - I )], 
where n is the total number of observations and m is the number of model parameters. 
Chapter 3 49 
the residuaIs for DBH, HPCD, and GCL indicated additional independent or concomitant 
variables were needed for their regression models. 
Analysis of residuals (R-student statistic, Cl = 0.05) detected an outlier for regression models 
using CCNL, BA, HPCC, and CEV as single regressors of estimated PAR transmission 
(%DIFN) at the centre of a vertically projected canopy gap (Zone 3) when the trees were 
completely without leaves (winter). Based on influence diagnostics (Cook's D, HAT DIAG, 
DFFITS, and DFBETAS) this outlier exerted an appreciable influence on one or more of 
the regression coefficients and overall reduced the precision of the fitted models 
(Myers 1 990). Nevertheless, after rechecking, the suspect observation was not removed 
from the data as it was deemed a valid observation. 
Relationships between CCL and CCNL, BA, HPCC, and CEV were also investigated using 
regression analyses. Based on the model developed for CCL vs. HPCC a simplified stand­
level estimate of PAR transmission (weighted %DIFN) for each experimental unit was 
calculated using the equation: 
Weighted %DIFN = (Zone 1 %DIFN x Zone 1 area7) + (Zone 3 %DIFN x Zone 3 area) 
This simplified stand-level estimate of PAR transmission was regressed against the in-leaf 
canopy closure ratio (CCL) for both summer and winter seasons. 
The relationship between the R:FR and estimated PAR transmission (%DIFN) was 
investigated using a logarithmic function (Y = a + b*1nX). The linear forms of the 
regressions developed for each season (summer & winter) were compared by analysis of 
covariance (ANCOVA) using the GLM procedure of SAS@ (Littell et al. 1 991 ; Kleinbaum 
et a!' 1 998). 
Where appropriate F-tests were used to determine whether the coefficients of simple linear 
regressions were significantly (a = 0.05) different from specified values. 
7 Calculated as the proportion (fraction) of an experimental unit that the zone covered on a horizontal plane 
(this was derived from the relationship between HPCC and CCL). 
Chapter 3 50 
3.4 Results 
3.4.1 %DIFN within the vertically projected gap between fully in-leaf poplars 
Estimated PAR transmission at the centre of a vertically projected canopy gap (Zone 3 
%DIFN) was inversely related to all of the tested stand density indices (Table 3.3) .  
However, there were marked differences in the general form, complexity, and precision of 
the above relationships (Table 3 .3). 
CCL measured with a standard digital camera provided the simplest relationship with 
Zone 3 %DIFN (Table 3 .3) .  Out of the different stand density indices tested, the assigned 
simple linear model for CCL had one of the lowest standard errors of prediction 
(RMSE=5.39) and explained 97% of the variation in Zone 3 %DIFN (Table 3 .3).  
In addition, the slope of the relationship was not significantly different from -1 (P=0.0958) 
and the intercept was not significantly different from open pasture %DIFN (P=0.5226) 
(Table 3 .3;  Figure 3 .4a). Overall, Zone 3 %DIFN decreased at a constant rate from open 
pasture levels to 1 7% under the highest measured CCL of 82% (Figure 3 .4a). 
The more complex exponential decay, natural logarithm transformed quadratic, and 
quadratic regression models, with varying degrees of concave curvature, fitted using 
BA, HPCC, and CEV as single independent variables, respectively, also showed a very 
strong relationship with Zone 3 %DIFN (Table 3 .3). These relationships each accounted for 
over 90% of the variation in Zone 3 %DIFN and their standard errors of prediction 
(RMSE:S8.77) were similar to that of CCL (Table 3 .3). However, there were marked site 
differences for the relationship between CEV and Zone 3 %DIFN (Figure 3.5). As CEV 
increased, Zone 3 %DIFN decreased at a much faster rate under the young/small trees at 
Ballantrae, compared to at the other three sites (P<0.05). In addition, over a common range 
of CEV sampled in the field (i.e. 30,000-80,000 m3/ha) , Zone 3 %DIFN was consistently 
around 20% higher at Kiwitea than at Hautope 1 and 2 (Figure 3 .5). Overall, Zone 3 
%DIFN decreased rapidly up to a BA and CEV of 30 m2/ha (Figure 3Ab) and 1 20,000 
m3/ha (Figure 3 .5), respectively, thereafter remaining relatively constant at around 20%. 
In contrast, Zone 3 %DIFN decreased to 26% at the maximum theoretical HPCC of78.5% 
(Table 3.3) .  
Chapter 3 51 
The remaining square-root transformed simple linear models, using DBH, HPCD, and GCL, 
explained less than 65% of the variation in Zone 3 %DIFN and their standard errors of 
prediction were at least twice the size of the regression models fitted for the other stand 
density indices (RMS�1 8.52) (Table 3 .3). On an individual site basis, with trees of similar 
size and form, these three stand density indices were strongly related to Zone 3 %DIFN 
(Table 3 .4; Figure 3 .6). However, when pooled together, the scatter plots of the data 
(and also the model residuals) clearly indicated that some other unaccounted difference 
between the sites was also influencing PAR transmission (Figure 3 .6). In particular, there 
were marked differences in the relationship for young/small trees at Ballantrae compared to 
mature/large trees at Kiwitea and Hautope 1 (Figure 3 .6). 
Table 3.3 Regression equations, and their respective coefficients, developed for overstorey (stand) density indices to estimate the percentage of above-canopy PAR (%DIFN) 
transmitted at the centre of a vertically projected canopy gap (Zone 3) between fully in leaf P. x euramericana trees. 
Index Equation Coefficients 
a b c r2 RMSE 
DBHt y= a + b*sqrt(x) 1 00.8 1 -7.88 0.62 1 8.52 
(4.57) (0. 85) 
BA y= a + b*exp(c*x) 1 8.58 78.25 - 1 0.32 x 1 0-2 0.95 6.72 
(3. 07) (3. 09) (1. 1 7  xl (f2) 
GCLt y= a + b*sqrt(x) 98.23 - 1 . 1 3  0.57 1 9.62 
(4. 72) (0. 13) 
HPCDt y= a + b*sqrt(x) 1 00.29 - 1 .72 0.54 20.3 1 
(5. 19) (0.22) 
HPCC Y= a + b*In(x+ 1) + c*In(x+ 1 )2 94. 10  9.82 -5.80 0.96 6. 1 7  
(1. 86) (2.03) (0.44) 
CEV Y= a + b*x + C*X2 9 1 .59 -0.97 0.34 x 1 0-2 0.9 1 8.77 
(1. 65) (0.07) (0. 05 x 1 (f2) 
CCL Y= a +  b*x 95.48 -0.96 0.97 5.39 
(1.06) (0. 02) 
Standard errors of regression coefficients are in parentheses. All fitted models were highly significant (P<0.000 1 ). Pooled number of observations for the equations = 53. 
Pooled mean estimated PAR transmission (%DIFN) in the open = 94.77% ± 1 .42% (mean ± standard error, n= 1 0). Abbreviations: DBH, sum of stem diameters over bark at 
breast height ( l .4m) per hectare (m/ha); BA, sum of stem basal areas at breast height per hectare (m2/ha); GCL, sum of vertically projected green crown lengths per hectare 
(mlha); HPCD, sum of the horizontally projected crown diameters per hectare (mlha); HPCC, horizontally projected canopy closure ratio (%); CEV, sum of the crown 
ellipsoidal volumes per hectare ('000 m3/ha); CCL, canopy closure ratio (%) measured with a standard digital camera when the trees were fully in-leaf; sqrt, square root; In, 
natural logarithm; exp, base of natural logarithm; r2, adjusted coefficient of determination (approximated for non-linear models); RMSE (root mean square error), standard 
error of prediction. t Systematic trends in the residuals for DBH, GCL, and HPCD indicated that at least another parameter or concomitant variable was required for the 
regression models. 
Chapter 3 
,-.. 
� Q 
'#. '-' 
s:: 0 'c;; 
.§ <Il s:: '" ... -
� � 
� 
1;j 
. 5 -<Il � 
,-... 
� Q 
'#. '-' s:: 0 'c;; .� E <Il 
a ... -
� � 
� 
1;j 
. 5 -<Il � 
120 
100 ........ 0 ........ 00'0, , 
80 
, 
60 
40 
20 
0 
0 20 
120 
100 
80 
60 
40 
20 
........ ........ , ........ 
r::t-- , ........ � 
·0 " 
, , 
......... , ...... 
, • 
' �  
'�  . ...... 
40 60 80 
Canopy closure with leaves (%) 
------
• • 
• 
-----
• 
• -------
Ca) 
100 
Cb) 
o +---------.--------.---------,--------.-------�
o 10  20 30 40 50 
Stand basal area (m2/ha) 
53 
Figure 3.4 The estimated percentage of above-canopy PAR (%DIFN) transmitted at the centre of a vertically 
projected canopy gap (Zone 3) between fully in leaf poplars, over a range of (a) canopy closure ratios (CCL) 
measured with a standard digital camera and (b) stand basal areas (BA). Solid line - mean predicted response. 
Dashed lines - upper and lower 95% prediction limits. Symbols: (e) Kiwitea, (�) Hautope 1 ,  (0) Hautope 2, 
and (0) Ballantrae. 
Chapter 3 
1 20 
,.-.. 
Z 1 00 I � '"" ...... � a::. 
� '-' 80 I: • . 5: en en ·s cP en 60 1a .. -
� ... 0 p., 40 "0 Q) 1;$ 
. 8 -
20 en � 
0 
0 50 
• • • 
... 
A 
0 A 
...... 'A # 
100 
CEV ('000 m3Jha) 
• • 
• • 
150 
54 
• 
200 
Figure 3.5 The estimated percentage of above-canopy PAR (%DIFN) transmitted at the centre of a vertically 
projected canopy gap (Zone 3) between fully in-leaf poplars, over a range of stand crown ellipsoidal volumes 
(CEV) at each farm site. Symbols: (e) Kiwitea, (A.)  Hautope 1 ,  (0) Hautope 2, and (0) Ballantrae. 
1 20 
,.-.. 
� 1 00 ...... Q '#. '-' 
c 80 0 ·m ti'l 
·s ti'l 
60 1a .t:: 
� p., 40 "0 Q) 1;$ 
.8 -
ti'l 20 � 
0 
0 
� ...... ...... � o ...... O -� 
..... . 
• • 
2000 4000 
Site 
Kiwitea 
Hautope 1 
Hautope 2 
Ballantrae 
A A 
-- . . . . . . . 
6000 
Stand green crown length Cm/ha) 
GCL range 
(m/stem) 
20 - 3 5  
1 6  - 23 
8 - 1 3  
3 - 5  
8000 
Figure 3.6 The estimated percentage of above-canopy PAR (%DIFN) transmitted at the centre of a vertically 
projected canopy gap (Zone 3) between fully in-leaf poplars, over a range of stand green crown lengths 
(GCL) at each farm site. The sum of the stem diameters over bark at breast height (DBH; l Am) and the sum 
of the horizontally projected crown diameters per hectare (HPCD) also showed very similar relationships 
with Zone 3 %DIFN. Symbols: (-e-) Kiwitea, C· · ·  .. . . . ) Hautope 1 ,  (-·0-·) Hautope 2, and (-0-) Ballantrae. 
Chapter 3 55 
Table 3.4 Regression parameters (standard errors in parentheses) for equations predicting estimated PAR 
transmission (Zone 3 %DIFN) from stand green crown length (mlha), where Y=(a*b)/(b+x). 
Site a b r RMSE n P 
Kiwitea 9 1 .24 (4.99) 7 19.83 ( 120.48) 0.90 8.89 14 <0.0001 
Hautope 1 96.88 (6.3 1 )  922.93 ( 169.85) 0.90 9.06 12 <0.0001 
Hautope 2 9 1 .66 (5.41 )  1 89 1 .93 (480.95) 0.89 8.2 1 6 <0.0028 
Ballantrae 98.28 (0.85) 8092.99 (709.08) 0.90 2.57 21 <0.000 1 
Out of the range of potential functions tested the non-linear hyperbolic decay function was ranked the best 
for all of the sites. Abbreviations: r, approximated adjusted coefficient of determination; RMSE (Root mean 
square error), standard error of prediction; n, total number of observations. 
Chapter 3 
3.4.2 %DIFN d irectly underneath a fully in-leaf poplar crown 
56 
When going from open pasture to low poplar stand densities estimated PAR transmission 
directly below the fully in-leaf tree crowns (Zone 1 %DIFN) decreased at a much faster rate 
than within the vertically projected gap between the trees (Zone 3) (Figure 3 .4 vs. 3 .7). 
For each of the sites, Zone 1 %DIFN became relatively constant from a low stand density 
onwards, irrespective of the actual index used (Figure 3.7a,b). However, the average level 
of PAR transmitted below the trees varied significantly between the four sites. On average, 
Zone 1 %DIFN at Kiwitea, Hautope 1 ,  Hautope 2, and Ballantrae was 1 1% ± 2% (mean ± 
SEM), 26% ± 2%, 28% ± 4%, 76% ± 2%, respectively. The difference in Zone 1 %DIFN 
between Hautope 1 and 2 was not significant (P=0.6098). 
Out of the different stand density indices tested, CCL, BA, HPCC, and CEV accounted for 
at least 90% of the variation in Zone 1 %DIFN (Table 3 .5). However, the above site 
differences, along with the pooled datasets generally being unbalanced, reduced confidence 
in the appropriateness of the assigned regression models. In particular, the datasets for these 
four independent variables lacked data points for young/small and oldllarge trees at high and 
very low stand densities, respectively (Figure 3 .7a,b). 
Similarly to Zone 3, for the pooled data both BA and CEV had clear thresholds above 
which Zone 1 %DIFN became relatively constant. However, these thresholds were about 
half that for Zone 3, with %DIFN becoming constant at around 1 7% beyond roughly 
1 0  m2/ha (Figure 3 .7b) and 50,000 m3/ha (data not shown), respectively, for the two stand 
density indices. In contrast, Zone 1 %DIFN deceased at a diminishing rate for CCL 
(Figure 3.7a) and HPCC (data not shown) over their entire ranges, reaching approximately 
1 5 - 1 8% at the maximum canopy cover measured in the field (Table 3 .5). Nonetheless, for 
all of the above stand density indices it was clearly evident that at least another variable was 
required to account for site differences (as shown for CCL and BA in Figures 3.7a & b). 
The datasets for DBH, GCL, and HPeD were generally more balanced - having greater 
integration of data points from each site along the x-axis (data not shown). Even so, these 
three independent variables each accounted for only 70-75% of the variation in Zone 1 
%DIFN and their regression models RMSE were larger than for the other investigated stand 
Chapter 3 57 
density indices (Table 3.5). Similarly to Zone 3, the lower quality of fit and precision of the 
assigned regression models for the pooled data was likely caused by more pronounced 
relationship differences at each site for these three independent variables (data not shown). 
Table 3.5 Regression equations, and their respective coefficients, developed for overstorey (stand) density indices to estimate the percentage of above-canopy PAR (%DIFN) 
transmitted directly below (Zone 1 )  the fully in-leaf P. x euramericana crowns. 
Index Equation 
a 
DBHt Y = a + b*In(x+l )  93. 1 8  
(5.21) 
BA Y = a + b*exp(c*x) 16.83 
(2.22) 
GCLt Y = a/(1 + b*x) 94.33 
(5. 78) 
HPCDt Y = a/Cl  + b*x) 94.48 
(5.58) 
HPCC Y = II(a + b*x) 1 .06 X 1 0-2 
(0.03 x 1 (/2) 
CEV Y = a + (b*x)/(c+x) 93.84 
(2.27) 
CCL Y = a + b*In(x+ 1 )  94.57 
(3. 10) 
b 
- 1 8.37 
(1. 70) 
77.63 
(3.48) 
0.23 x 10-2 
(0. 05 x 1 (/2) 
0.42 x 10-2 
(0. 08 x 1 (/2) 
0.08 X 1 0.2 
(0.01 X 1{/2) 
-80.86 
(2. 91) 
- 1 8.07 
(0.95) 
Coefficient 
c 
-0.36 
(0. 06) 
5.62 
(1. 32) 
� RMSE 
0.75 1 7.61  
0.93 9.47 
0.72 1 8.52 
0.74 17.88 
0.92 9.99 
0.95 7.57 
0.90 1 0.95 
Standard errors for regression coefficients are in parentheses. All fitted models were highly significant (P<O.OOOI). Pooled number of observations for the models = 40. 
Pooled mean estimated PAR transmission (%DIFN) in the open = 94.74% ± 1 .4 1% (mean ± standard error, n=IO). Abbreviations: DBH, sum of stem diameters over bark at 
breast height ( l .4m) per hectare (mlha); BA, sum of stem basal areas at breast height per hectare (m2Iha); GCL, sum of vertically projected green crown lengths per hectare 
(mlha); HPCD, sum of the horizontally projected crown diameters per hectare (m/ha); HPCC, horizontally projected canopy closure ratio (%); CEV, sum of the crown 
ellipsoidal volumes per hectare ('000 m3/ha); CCL, canopy closure ratio (%) measured with a standard digital camera when the trees were fully in-leaf; In, natural logarithm; 
exp, base of natural logarithm; r2, adjusted coefficient of determination (approximated for the non-linear models); RMSE (root mean square error), standard error of 
prediction. t Systematic trends in the residuals for DBH, GCL, and HPCD were detected, indicating that at least another parameter or concomitant variable was required for 
the regression models. 
Chapter 3 59 
120 
\ (a) 
r-.. 1 00 \ � \ Cl \ <J. 80 '<> '-" 
c "-
.9 � ........ '" \ . § 60 
........ --'" \ ------;j \ - - -... - - -... 40 \ � o A 
- - -
" Po. "-
� ... "3 20 • ........ • ........ � ........ ....... .5 - ... ... .......... ----_e_. • • • '" � 0 - - - - - --- -
-20 
0 20 40 60 80 100 
Canopy closure with leaves (%) 
120 
(b) 
r-.. 100 
� .... Cl 
<J. 80 '-" 
c 0 ·Cii .!!.l 60 E '" 
;j ... ... 40 
� Po. � A "3 20 A A A A • � • • . 5 • ... , . • • • Cl) 
� 0 --- --------- --------
-20 
0 1 0  20 30 40 50 
Stand basal area (m%a) 
Figure 3.7 The estimated percentage of above-canopy PAR (%DIFN) transmitted directly below the fully in­
leaf poplar crowns (Zone 1 ), over a range of (a) stand canopy closure ratios (eeL) measured with a standard 
digital camera and (b) stand basal areas (BA). Symbols: (e) Kiwitea, ( .. ) Hautope 1 ,  (D) Hautope 2 ,  and (0) 
Ballantrae. Solid line - mean predicted response. Dashed lines - upper and lower 95% prediction limits. The 
data point clearly outside the lower 95% prediction limit for eCL vs Zone 1 %DIFN was identified as an 
outlier (R-student statistic, a = 0.05), but was not removed from the natural logarithm regression analysis. 
Note: the clumping or segregation of data points for the different sites (along the Y-axis) indicated that 
another unaccounted factor was also influencing Zone 1 %DIFN. 
Chapter 3 60 
3.4.3 %DIFN within the vertically projected gap between leafless poplars 
After leaf fall, Zone 3 %DIFN remained inversely related to each of the stand density 
indices, when all four of the sites were pooled together (Table 3 .6). However, in 
comparison to summer with the trees fully in leaf, Zone 3 %DIFN increased across the 
entire range of stand densities - more so towards the mid to higher end of the stand density 
range (Tables 3.3 & 3.6). For example, at a stand basal area of 5 m2lha and 40 m2/ha 
%DIFN was predicted to increase by 1 5% and 38%, respectively, once the trees had lost all 
of their leaves (Figure 3 .4b vs. 3 .8b). 
On average, the regression models fitted using CCNL, BA, HPCC, and CEV had a RMSE 
44% greater than their summer counterparts when the trees were fully in leaf (P=0.0193 ;  
Tables 3.3 & 3.6). Out of these four indices, the simple linear model based on CCNL 
produced the highest adjusted coefficient of determination (r2=0.82) and smallest standard 
error of prediction (RMSE=8.24; Table 3 .6). BA, CEV, and HPCC, on the other hand, each 
accounted for approximately 70% of the variation in Zone 3 %DIFN (Table 3 .6). Although 
the regression models fitted with DBH, GCL, and HPCD as single independent variables 
had a smaller RMSE than their equivalent models in summer, they still had the weakest 
relationship with Zone 3 %DIFN among the different stand density indices investigated 
(r2�0.43 ; Tables 3.3 & 3.6). Again, the unbalanced nature of the datasets for CCNL, BA, 
HPCC, and CEV (refer to Section 3.4.2) reduced confidence that the shape ofthe assigned 
curves was an accurate representation of the 'true' relationship. 
Residual scatter plots along with formal test statistics (e.g. Levene Median Test) of the 
above regression models indicated that each, except for CCNL and BA, contained 
heterogeneous variance (data not shown). This looked to be because of marked site 
differences in the relationships between Zone 3 %DIFN and the various stand density 
indices. All of the relationships at both Kiwitea and Hautope 2 were generally negative and 
linear. Whereas, at Hautope 1 and Ballantrae, once amongst the stands, there was no clear 
trend between Zone 3 %DIFN and any of the tested stand density indices. Nevertheless, 
Zone 3 %DIFN did vary significantly between these two sites (P<0.0001 ). On average, the 
younger/smaller trees at Ballantrae allowed 96% ± 1 % (mean ± SEM) of incident PAR 
Chapter 3 61 
to be transmitted to the understorey, while the olderllarger trees at Hautope 1 allowed 
54% ± 2% to be transmitted. 
Based only on Kiwitea and Hautope 2 in winter, the relationships for BA, HPCC, and CEV 
versus Zone 3 %DIFN changed from having some degree of negative concave curvature 
when the trees were fully in leaf in summer (Table 3 .3)  to a negative straight -line (linear) 
relationship when the trees were completely without leaf in winter (Table 3 .7). In contrast, 
the relationship between CC and Zone 3 %DIFN was linear for both periods. However, the 
slope of the relationships did not remain constant, with the rate of decrease in Zone 3 
%DIFN associated with increasing CC being 26% less in winter, when the trees were 
completely without leaf (Table 3 .3  vs. 3 .7). Overall, BA, HPCC, and CEV accounted for 
over 80% of the variation in Zone 3 %DIFN, while CCNL accounted for a slightly lesser 
amount at around 70% (Table 3 .7). 
The datasets for DBH, GCL, and HPCD were generally more balanced than for the other 
stand density indices (Figure 3 . 8  vs. 3 .9). Similarly in summer, the relationships for these 
three independent variables clearly depended on the site, as shown for GCL in Table 3 .8  and 
Figure 3 .9. Site differences in the slope of the relationships were particularly marked 
between Ballantrae, Hautope 2, and Kiwitea (Table 3 .8; Figure 3 .9). At each site, the 
individual relationships between these three independent variables and Zone 3 %DIFN also 
looked to change from being curvilinear in summer to linear in winter, as again shown for 
GCL (Figure 3.6 vs. 3 .9). 
Table 3.6 Regression equations, and their respective coefficients, developed for overstorey (stand) density indices to estimate the percentage of above-canopy 
PAR (%DIFN) transmitted at the centre ofthe vertically projected canopy gap (Zone 3) between the leafless P. x euramericana trees. 
Index Equation 
a 
DBH Y = a +  b*x 95.06 
(2. 75) 
BA Y = a + b*ln(x+ 1)  1 0 1 .64 
(2.39) 
GCL Y =  a +  b*x 93 .91 
(2.68) 
HPCD Y = a + b*x 94.96 
(3. 16) 
HPCC Y = a + h*x 98 . 1 3  
(1.97) 
CEV Y = a + b*x + c*x2 97. 1 1  
(1.94) 
CCNLt y2 = a + b*x 9379 . 17  
(215. 05) 
b 
-0.42 
(0.07) 
- 1 1 .74 
(1. 1 7) 
-8.79 x 10.3 
(1. 44 x 10-3) 
-2.09 x 10-2 
(0. 41 x ](f.2) 
-0.48 
(0. 04) 
-0.55 
(0. 08) 
- 137.79 
(8.85) 
Coefficients 
c 
0.20 X 10-2 
(0. 05 x 1 (f2) 
r RMSE 
0.43 1 3 .90 
0_66 1 0_69 
0.4 1 14.09 
0.33 1 5 .04 
0.70 10.04 
0.70 10 . 14  
0.82 8.24 
tZone 3 %DIFN was power transformed to improve the underlying assumptions of the linear model; nevertheless, the RMSE is given in natural units 
(Myers 1990). Standard errors for regression coefficients are in parentheses. All fitted models were highly significant (P<0.0001). Pooled number of observations 
for the equations = 52. Pooled mean %DIFN in the open = 97.6 % ± 1 .3% (mean ± standard error, n=9). Abbreviations: DBH, sum of stem diameters over bark at 
breast height ( l .4m) per hectare (m/ha); BA, sum of stem basal areas at breast height per hectare (m2/ha); GCL, sum of vertically projected green crown lengths 
per hectare (m/ha); HPCD, sum of the horizontally projected crown diameters per hectare (m/ha); HPCC, horizontally projected canopy closure ratio (%); CEV, 
sum of crown ellipsoidal volumes per hectare ('000 m3/ha); CCNL, canopy closure ratio (%) measured with a standard digital camera when the trees were 
completely without leaves; In, natural logarithm; r2, adjusted coefficient of determination; RMSE (root mean square error), standard error of prediction. 
Chapter 3 63 
1 4000 
(a) 
""' 12000 .. 
� , , 
0 1 0000 , 
'$. � '-' 
c 
8000 
, 
.9 , '" r::. , '" 'El , '" 6000 , , � , ., ... , ... , , 
� 4000 , . • , • p.. , • "0 , • Q) 
1ii 2000 .. , , . . 5 ... , (/) "'-� 0 , 
-2000 
0 20 40 60 80 1 00 
Canopy closure without leaves (%) 
140 
(b) 
Z 120 
f:= Cl '$. 1 00 '-' 
c:: ;-----.9 # (/) ------(/) 80 ---'S - --
(/) • 
� • • ... ... 60 .. 
� .. • • .. 
p.. • 
"0 • -
- .--------Q) 40 ... ---'" 
.5 ... (/) 
� 20 
0 
0 10 20 30 40 50 
Stand basal area (m2/ha) 
Figure 3.8 The estimated percentage of above-canopy PAR (%DIFN) transmitted at the centre of a vertically 
projected gap between leafless poplars (Zone 3), over a range of (a) canopy closure ratios (CCNL) measured 
with a standard digital camera and (b) stand basal areas (BA). Note: %DIFN was squared to normalise the 
data in (a). Symbols: (e) Kiwitea, ( .. ) Hautope 1 ,  (0) Hautope 2, and (0) Ballantrae. Solid line - mean 
predicted response. Dashed lines - upper and lower 95% prediction limits. The data point clearly outside the 
lower 95% prediction limit for BA was identified as an outlier (R-student statistic, a = 0.05), but was not 
removed. For BA, the clumping or segregation of ( " )  Hautope I data points away from the other sites (along 
the Y-axis) indicates that another unaccounted factor was also influencing Zone 3 %DIFN. 
Chapter 3 64 
Table 3.7 Regression parameters (standard errors in parentheses) for equations predicting estimated Zone 3 
%DIFN individually from CCNL, BA, HPCC and CEV, where Y= a + b*x. 
Index a b r RMSE 
CCNL 97. 1 0  (2.75) -0.7 1  (0. 10) 0.73 7.05 
BA 94.58 (1 .49) -0.94 (0.08) 0.89 4.48 
HPCC 98.26 (2.34) -0.37 (0.04) 0.81 5.92 
CEV 96.07 ( 1 .87) -0.20 (0.02) 0.85 5 .23 
Dataset for the above simple linear relationships includes only Kiwitea and Hautope 2 sites. All models were 
highly significant (P<O.OOO I ). Pooled number of observations for the equations = 19. Pooled mean PAR 
transmission (%DIFN) in the open = 95.61 ± 2.75 (mean ± stand error, n=4). Abbreviations: r, adjusted 
coefficient of determination; RMSE (Root mean square error), standard error of prediction. 
Chapter 3 65 
1 20.-----------------------------------------� 
,....., 
S 1 00 � �� 000 � O  0 Cl 4 0>0 � 0 '$- o , 0 '-' 
80 I: 0 • 0 ·Cil , 
.§ A • • Cl) 
60 A I: ea '" A A ... - • 
� A • A.c 40 A � 
.� 
tl 20 � 
0 
0 2000 4000 6000 8000 
Stand green crown length (m/ha) 
Figure 3.9 The estimated percentage of above-canopy PAR (%DIFN) transmitted at the centre of a vertically 
projected canopy gap (Zone 3) between leafless poplars, over a range of stand green crown lengths (m/ha). 
Symbols: (e) Kiwitea, (.)  Hautope 1 ,  (0) Hautope 2, and (0) Ballantrae. Stem diameter over bark at breast 
height (DBH; l .4m) and horizontally projected crown diameter (HPCD) also showed similar relationships 
with Zone 3 %DIFN. 
Table 3.8 Regression parameters (standard errors in parentheses) for equations predicting estimated PAR 
transmission (Zone 3 %DIFN) from stand green crown length (mlha), where Y= a + b*x. 
Site a b � RMSE n P 
Kiwitea 92.2 (2.9) ·0.0 1 3 4  (0.0021 )  0.78 6.3 1 3  <0.000 1 
Hautope 1 83.4 (7.5) -0.0 1 3 9  (0.0041 )  0.56 1 3.7 9 0.0 1 1 8  
Hautope 2 97.8 (1 .9) -0.0065 (0.001 1)  0.87 3 .4  6 0.004 1 
Ballantrae NS NS 2 1  0.8490 
Out of the range of potential functions tested the simple linear function was ranked the best for all four sites. 
Abbreviations: �, adjusted coefficient of determination; RMSE (Root mean square error), standard error of 
prediction; n, total number of observations. 
Chapter 3 
3.4.4 %DIFN directly underneath a leafless poplar crown 
66 
The general trend of the winter relationships between Zone 1 %DIFN and the different 
stand density indices was similar to in summer. All of the fitted regression models for the 
pooled data maintained strong negative concave curvature (Table 3.9). The decrease in 
Zone 1 %DIFN was particularly large when going from open pasture to low stand densities, 
and thereafter %DIFN remained relatively constant (Figures 3 . l 0a,b). As in summer, with 
increasing stand density %DIFN decreased at a much faster rate in Zone 1 compared to 
Zone 3 .  The only exception was at Hautope 1 ,  where %DIFN in both zones of tree 
influence was not significantly different, irrespective of the stand density (P=0.5415 ;  
Figure 3.8 vs. 3 . 1 0). 
One of the main differences between the two periods was a significant increase in the 
average level of estimated PAR transmission (%DIFN) in winter, when the trees were 
completely without leaves (P<O.OOOI) .  On average, Zone 1 %DIFN increased by 3 1  % ± 5% 
(mean ± SED) between summer and winter (excluding open pasture values). However, on 
an individual site basis, the increase in %DIFN relative to in summer was greater under the 
more mature/larger trees at Kiwitea, followed by those at Hautope 1 and 2, and then the 
younger/smaller trees at Ballantrae (Figure 3 . 1 1 ). As a result, Zone 1 %DIFN at Kiwitea, 
Hautope 1 ,  Hautope 2, and Ballantrae in winter averaged 54% ± 2% (mean ± SEM), 
5 1  % ± 3%, 68% ± 3%, and 83% ± 4%, respectively; the difference between Kiwitea and 
Hautope 1 was not significant (P=0.3787). In winter, the greater increase in Zone 1 %DIFN 
at Kiwitea compared to Ballantrae also meant that there was less variation amongst the sites 
than in summer (Figure 3 . 1 1) .  
Overall, the regression models based on CCNL, BA, HPCC, and CEV explained about 80% 
to 85% of the variation in Zone 1 %DIFN, while DBH, GCL, and HPCD explained around 
70% (Table 3.9). On average, the proportion of variation accounted for did not change 
significantly between the two periods (P=0.2228; Table 3 .5 & 3.9). 
For the pooled data, both CCNL and HPCC had clear thresholds (10% and 45%, 
respectively) above which Zone 1 %DIFN became constant at 54% (Figure 3 . 1 0a; Table 
3 .9). This contrasted with the relationships fitted for summer, which decreased at a 
Chapter 3 67 
diminishing rate over the entire ranges of CCL and HPCC (Figure 3 . 7a; Table 3 .5). Site 
differences and also the unbalanced nature of the pooled dataset may have caused the 
inconsistency in the form of the relationships between periods, as previously discussed in 
Section 3 .4.2. 
Both BA and CEV had similar threshold values to in summer, above which Zone 1 %DIFN 
became relatively constant (Figure 3. 1 0b; Table 3 .9). However, the same factors influencing 
the choice of regression model for CCNL and HPCC were also present for BA and CEV 
(Figure 3 . 1 0b). 
Again, similar to in summer, the datasets for DBH, GCL, and HPCD were more balanced 
(along the x-axis) than for the other stand density indices and showed more pronounced site 
differences in their relationship with Zone 1 %DIFN (data not shown). The site differences 
likely reduced the above regression models quality of fit and precision in comparison to the 
other stand density indices. 
Based on the maximum canopy closure ratio for in summer (82% CCL) and winter (52% 
CCNL), Zone 1 %DIFN for the pooled data was predicted to increase from 1 5% ± 5% 
(estimate ± standard error) to 52% ± 4% (Figures 3 .7a & 3 .  l Oa). These values, and the 
corresponding vertical shift in Zone 1 %DIFN between the two periods, closely resembled 
Zone 3 at the same stand density (Figures 3 .4a & 3 .8a). This indicates that the stand was 
effectively closed at this density, and thus, the canopy gap fractions 'seen' from Zones 1 
and 3 were similar. 
Table 3.9 Regression equations, and their respective coefficients, developed for overstorey (stand) density indices to estimate the percentage of above-canopy PAR (%DIFN) 
transmitted directly below (Zone 1) the leafless P. x euramericana crowns. 
Index Equation 
a 
DBH Y = a + b*In(x+ 1) 96.34 
(3. 33) 
BA Y = a + b*sqrt(x) + c*x 97.52 
(2.30) 
GCL Y = l/(a + b*sqrt(x» 1 .04 x 1 0-2 
(0. 04 x 1 (f2) 
HPCD Y = lI(a + b*sqrt(x» 1 .03 x 1 0-2 
(0. 04 x 1 (f2) 
HPCC Y = a + b*exp(c*x) 53 .87 
(1.91) 
CEV Y = a + b*sqrt(x) + c*x 95.92 
(2. 09) 
CCNL Y = a + b/(x+ 1 )  53 .63 
(1. 82) 
b 
-10.45 
(1.07) 
- 1 9.93 
(1. 94) 
0.02 X 1 0-2 
(0.00 x 1 (f2) 
0.03 X 10-2 
(0. 00 x 1 (f2) 
43 .04 
(3. 08) 
-8.95 
(0.95) 
4 1 .98 
(3.28) 
Coefficients 
c 
2. 13  
(0. 33) 
-0.08 
(0.02) 
0.45 
(0. 08) 
r2 RMSE 
0.71 10.73 
0.86 7.60 
0.68 1 1 .35 
0.69 1 1 .07 
0.84 8. 12  
0.87 7.24 
0.81 8.71 
Standard errors for regression coefficients are in parentheses. All fitted models were highly significant (P<O.OOOl). Total pooled number of observations for the models = 39. 
Pooled mean estimated PAR transmission (%DIFN) in the open = 97.6 % ± 1 .3% (mean ± standard error, n=9). Abbreviation: DBH, sum of stem diameters over bark at breast 
height ( l .4m) per hectare (m/ha); BA, sum of stem basal areas at breast height per hectare (m2/ha); GCL, sum of vertically projected green crown lengths per hectare (m/ha); 
HPCD, sum of the horizontally projected crown diameters per hectare (m/ha); HPCC, horizontally projected canopy closure ratio (%); CEV, sum of the individual crown 
ellipsoidal volumes per hectare ('000 m3/ha); CCNL, canopy closure ratio (%) measured with a standard digital camera when the trees were completely without leaves; sqrt, 
square root; In, natural logarithm; r2, adjusted coefficient of determination (approximated for non-linear models); RMSE (root mean square error), standard error ofprediction. 
(» CO 
Chapter 3 
120 .-------------------------------------------� 
(a) 
� 1 00 
Cl 
'<t. '-' 
s:: 
o :� rJl 
� -
"­
40 ..... 
-----_� 
_ ___.A._ __  
20 
o +---------.---------.---------.---------.-------� 
o 20 40 60 80 1 00 
Canopy closure without leaves (%) 
120 �--------------------------------------------� 
60 
40 
20 
� 
\. "-
o "-
\ 0  
\ 
\.. 
"­-...... 
.......... '--- ---
------
----
---
• 
• 
• • 
.'-... -- --
­
� ----
-�-- -
--
(b) 
o +---------.---------.---------.---------.-------� 
o 1 0  20 30 40 50 
Stand basal area (m2/ha) 
69 
Figure 3.10 The estimated percentage of above-canopy PAR (%DIFN) transmitted directly below (Zone 1 ) 
the leafless poplar crowns, over a range of (a) stand canopy closure ratios (CCNL) measured with a standard 
digital camera and (b) stand basal areas (BA). Symbols: (e) Kiwitea, (.6. )  Hautope 1 ,  (0) Hautope 2, and 
(0) Ballantrae. Solid line - mean predicted response. Dashed lines - upper and lower 95% prediction limits. 
Chapter 3 70 
100 
_ Summer 
..-- Winter � 80 
0 
'$. ........-
s:: 
.9 
60 '" 
.§ '" 
a ... -
� 40 � 
'"0 Cl) -'" 
.5 20 -'" � 
0 --'-----
Kiwitea Hautope 1 Hautope 2 BalIantrae 
Site 
Figure 3.1 1  Mean estimated percentage of above-canopy PAR (%DIFN) transmitted directly underneath the 
individual poplar crowns (Zone 1 )  at each site. Vertical bars represent the standard error of the mean. 
Chapter 3 
3.4.5 Selective relationships between stand density indices 
71 
When pooled across all four sites, the winter canopy closure ratio (CCNL) increased by 
6% for every 1 0% increase in the summer canopy closure ratio (CCL) (Table 3 . 1 0; Figure 
3 . 12a). However, based on the scatter of the data points, differences in the relationship 
between these two variables were particularly marked between the younger trees at 
Ballantrae and the older, more fully developed, trees at Kiwitea and Hautope 1 and 2 
(Figure 3 . 1 2a). A comparison of slopes from simple linear fimctions for each site using 
ANCOV A confirmed that Ballantrae was significantly different from the other three sites 
(P<0.01 ). As a result, a further two simple linear functions are presented in Table 3 . 10  
representing the different groups - separated as young (Ballantrae) and mature (Kiwitea & 
Hautope) stands. 
HPCC, BA, and CEV were all strongly related to CCL (Table 3 . 10; Figures 3 . 12b,c,d). 
Initially, when going from open pasture to low BA and CEV the increase in CCL was rapid, 
with CCL becoming relatively constant at around 70-80% beyond a BA of approximately 
20 m2/ha (Figures 3 . 12c,d). 
Table 3.10 Regression models, and their respective coefficients, developed for selected stand indices to 
predict canopy closure (CC). 
Index Equation 
a 
BA Y=a*(1 -exp{-b*x» 78.69 (3.25) 
CEV Y=a*xb 9. 1 4  ( 1 .62) 
HPCC Y=a*(1 -exp{ -b*x» 1 36.28 (28.82) 
CCLt Y= a + b*x -3.79 ( 1 .34) 
CCL t categorised into stand age groups: 
Young Y= a + b*x 0.08* (0.06) 
Mature Y= a + b*x -4.3 1 * (4.30) 
Coefficients 
b 
0. 1 05 (0.012) 
0.428 (0.039) 
0.0 1 0  (0.003) 
0.600 (0.030) 
· 0.061 (0.006) 
0.6 1 5  (0.067) 
r 
0.94 
0.90 
0.95 
0.92 
0.87 
0.77 
RMSE 
7.44 
9.49 
6.48 
5.37 
0. 1 8  
6. 1 8  
tIndex used to estimate the without leaf canopy closure ratio (CCNL). 'Coefficients (intercepts) were not 
significant (P>0.05). Standard errors are presented in parentheses. All fitted models were highly significant 
(P<O.OOOI). Abbreviations: r, adjusted coefficient of determination; RMSE, standard error of prediction. 
Total number of observations for the models based on all four sites, young, and mature categories were 43, 
1 8, and 26, respectively. Analysis of the linear model residuals for CCL-young and -mature indicated that 
transformations were required to meet the underlying assumption of homogenous variance. However, these 
models were left untransformed for ease of comparison. 
70 
60 
/ 
50 �
/ 
/ -& . ---- / ..  <f? / 40 /' 
'-' 
� 
/ 
/ / 
30 
/' /' / ./ 
U / /' /' 0' / • /' 
20 /' / / 
--/' / / /' /' /' 1 0  /' /' /' 
/' (a) / 
0 
0 20 40 60 80 1 00 
CCL (%) 
100 
-
-
-
-
-
-
-
---
-
-
-
..... 
-
• 80 " '" 
;' .. .  
.. 
.. 
/" . 
• 
---- 60 / ---------<f? I --
-
-
'-' R 
--...... 
-
.....1 / 
.
. 
'" U . "," u 40 0/ / / 
/ / I /0 I 20 / '/ 
/ 
/. 
0 
(c) 
0 10 20 30 40 50 
BA (m2/ha) 
1 00 
80 
,...... 60 ::R e....-
.....1 U 
U 40 
20 
0 
1 20 
l OO 
80 
----
::R e....-
.....1 60 U 
u 
40 
20 
0 
;' 
0 
.... 
..........
. 
..... "'
''' 
. 
'" 
'" 
--
'" 
-- 0 
--
--
..... ... ' 
-- • 
..... 
..... 
. '" 
--
'" 
p" 
--
'" 
'" 
",6 --
'" 
� <> '" ;' 
/ 
<> ,,
'" 
<> " 
<> ",' 
20 
'" 
--
" 
/''' .. 0 
..... 
'" 
...... .... 
.. 
. 
--
"",,,,,// 
'" 
--
--
--
40 60 80 
HPCC (%) 
-
-
-'  
-
-
-
-
--
--
-
-
_ ..... 
. --
-
-
-
-
-
--
-
--
/ 0 . _-
..... .... 
/ ...... .... 
"' ..... 
..... 
'" 
" 
;' 
/ • 
0 50 100 1 50 
CEV ('000 m3/ha) 
(b) 
1 00 
(d) 
200 
Figure 3.12 Relationships between overstorey (stand) density indices. Abbreviations: CCL, canopy closure ratio when trees are fully in leaf; 
CCNL, canopy closure ratio when trees are completely without leaf (no leaf); HPCC, horizontally projected canopy closure ratio; BA, stem basal 
areas at breast height ( l .4m) per hectare; and CEV, crown ellipsoidal volumes per hectare. Symbols: (e) Kiwitea, (.A)  Hautope 1, (0) Hautope 2, 
and (0) Ballantrae. Solid line - mean predicted response. Dashed lines - upper and lower 95% prediction limits. Datum clearly outside the lower 
95% prediction limit for CEV vs. CCL was identified as an outlier (R-student test, a=0.05) (Myers 1 990), but was not removed from the 
regression. 
Chapter 3 
3.4.6 Weighted (stand level) %DIFN over a range of canopy closures 
73 
In summer, the relationship between eeL and stand-level PAR transmission (weighted 
%DIFN) was curvilinear, with the rate of change in weighted %DIFN decreasing at higher 
eeLs (Table 3 . 1 1 ; Figure 3 . 13a). For example, at a eeL of 40% weighted %DIFN was 
reduced by 48% compared to the open pasture, whereas going from a CCL of 40% to 80% 
(near the maximum eeL measured in the field) it was only reduced by a further 33%. 
Overall, the regression model accounted for a very high proportion of variation in weighted 
%DIFN (r2=0.97). 
Conversely, in winter the rate of change in weighted %DIFN increased at higher eCLs 
(Table 3 . 1 1 ;  Figure 3 . 1 3b). At a high eCL of 80% weighted %DIFN was predicted to be 
over 3-fold greater under the leafless trees in winter than under the fully foliated trees in 
summer (Figures 3 . 13a,b). 
Table 3.1 1  Relationships between the in-leaf canopy closure ratio (CeL) and estimated stand-level PAR 
transmission (weighted %DIFN). 
Season Equation Coefficients 
a b c r2 RMSE n 
Summer Y =a - b*x + C*X2 94.94 ( 1 . 5 1 )  1 .42 (0. 13) 0.005 (0.002) 0.97 5.68 40 
Wintert y2 =a - b*x 9367 (260) 88.38 (5. 1 7) 0.88 7.56 39 
t%DIFN was squared to normalise the distribution of the regression residuals; nevertheless, the RMSE for 
the winter season is given in natural units (Myers 1 990). Standard errors of the coefficients are given in 
parentheses. Both regressions were highly significant (P<O.OOO I ). Abbreviations: �, adjusted coefficient of 
determination; RMSE, standard error of prediction, n, total number of observations. 
Chapter 3 74 
120 
(a) 
� 100 " " " Cl <>-. � � "-,,-'-' 80 c "� "-.9 "-ell , 
.§ "- , "- ........ 
� 60 "- , "- ........ ... ,0 0--.. ... 
........ ........ 
� 
........ "- , , . " � 40 ........ "0 , 
,. 
....... 0 ....... Q) ....... 
.� ......... ..,.... ....... . 1 ........ ....... _ , ... 
20 , • • Cl> ....... � ....... • , 
....... ....... . • ....... -.... -.... 
0 
0 20 40 60 80 100 
Canopy closure with leaves (%) 
14000 
(b) 
,-., 12000 N � , Cl , , � 1 0000 , '-' <> , , c , 0 , ·Cii 
8000 
, Cl> 'Q... ·8 , ell , a , , ....... ... 
6000 ....... 
, ... , 
� 
, , , , , ....... � , ....... -...... -...... "0 4000 , • • Q) , 
� -...... 
8 -...... , ·z , ell 2000 , � • , -...... ... -...... 
, 
, 
0 , 
0 20 40 60 80 1 00 
Canopy closure with leaves (%) 
Figure 3.13 Estimated stand-level PAR transmission (weighted %DIFN) across a range of in-leaf canopy 
closure ratios (CCL) in (a) summer and (b) winter (%DIFN squared to normalise data). Solid line - mean 
predicted response. Dashed lines - upper and lower 95% prediction limits. Symbols: (.) Kiwitea, (�)  
Hautope 1 ,  (D) Hautope 2, and (0) Ballantrae. 
Chapter 3 
3.4.7 The relationship between R:FR and %DIFN 
75 
In summer, the red to far-red ratio (R:FR) , relative to that received in the open pasture, 
progressively declined with reduced levels of estimated PAR transmission (%DIFN) under 
the fully in-leaf poplar stands (Table 3 . 1 2; Figure 3 . 14a). The decrease was especially 
pronounced below a %DIFN level of approximately 40% (Figure 3 . 14a). In contrast, the 
R:FR underneath the leafless poplar stands in winter, relative to the open pasture ratio, 
decreased less markedly (P<O.OOOI )  under reduced %DIFN and the rate of change was 
more constant over the entire range of measured transmission levels. There was also less 
variation around the fitted relationship in winter than summer (Table 3 . 12; Figure 3. 1 4b). 
The regression model fitted for summer showed signs of systematic variation in the 
residuals around the highest %DIFN levels, with the model tending to underestimate R:FR, 
relative to the open pasture ratio (Figure 3 . 14a). 
Table 3.12 Regression parameters (standard errors in parentheses) for equations predicting under storey 
R:FR, relative to open pasture (%), from estimated PAR transmission (%DIFN), where Y = a + b*ln(x). 
Season 
Summer 
Winter 
a 
1 1 .53 (2.76) 
62.48 (3.72) 
b 
1 8.50 (0.73) 
8.30 (0.86) 
0.89 
0.62 
RMSE 
5.42 
1 .93 
n 
83 
5 8  
Models were highly significant (P<O.OOO I). Abbreviations: r, adjusted coefficient o f  determination; RMSE, 
standard error of prediction; n, total number of observations. 
Chapter 3 76 
120 
G) 100 01) <:<S -= G) 0 '" r-.. 
8. �  80 <:<S G) 
� 3 
� �  60 � 0.. 
� 5 » 0.. G) 0 "' c..,. � 0 40 
'" G) "0 s:: � 20 
Ca) 
0 
0 20 40 60 80 100 120 
Estimated PAR transmission (%DIFN) 
120 
G) 100 01) <:<S -= G) 0 .... ,.-., 
8. �  80 
<:<S G) 
� 3 
� �  � 0.. 60 • •  s:: � G) » 0.. G) 0 .... c..,. � 0 40 
.... G) "0 s:: � 20 
Cb) 
0 
0 20 40 60 80 1 00 120 
Estimated PAR transmission (%DIFN) 
Figure 3.14 The relationship between the understorey R:FR, relative to the open pasture, and estimated PAR 
transmission (%DIFN) through the poplar canopy in (a) summer (in-leaf) and (b) winter (no-Ieat). Symbols: 
(e) Kiwitea, ( .. ) Hautope 1 ,  (0) Hautope 2, and (0) Ballantrae. Solid line - mean predicted response. Dashed 
lines - upper and lower 95% prediction limits. The two data points clearly outside the 95% prediction limits 
for the summer (in-leat) relationship were identified as outliers (R-student test, a=0.05) (Myers 1 990), but 
were not removed from the simple logarithm regression. No R:FR measurements were taken at the Kiwitea 
site for the winter period. 
Chapter 3 77 
3.5 Discussion 
3.5.1 The relationship between stand density indices and Zone 3 %DIFN for 
summer and winter 
3.5.1 .1  General relationship 
In both summer and winter PAR transmission at the centre of a vertically projected gap 
between the tree crowns (Zone 3 %DIFN) was inversely related to all of the tested stand 
density indices (Tables 3.3 & 3 .6). However, after leaf fall in autumn Zone 3 %DIFN 
increased especially in the mid to high poplar stand density range, which in turn reduced the 
slope of the relationships. As a result, at higher stand densities a greater proportion of 
annual PAR received on the ground would occur during leafless periods of the year 
(i.e. autumn through until spring). Using the depth of a canopy to vary its density, Wang & 
Baldocchi (1 989) found the same trend under a mixed deciduous oak-hickory stand 
(Quercus and Carya spp.). Anderson (1964b) also noted that the more heavily shaded a 
deciduous forest site was in summer the greater the percentage of the total year's  radiation 
budget was received in spring. 
3.5. 1 .2 Rankings of stand density indices for estimating Zone 3 %DIFN 
Out of the stand density indices tested, CCLINL measured with a standard digital camera 
had the simplest relationship with Zone 3 %DIFN for both summer and winter periods. 
CCLINL consistently had one of the highest adjusted coefficients of determination (r2) and 
one of the lowest standard errors of prediction (RMSE) out of the indices tested. 
In particular, the regression statistics for CCL suggested that the simple linear model fitted 
was a very good first approximation of the actual relationship, and that any gain in precision 
by including other independent (explanatory) variables was likely to be negligible. The -1 : 1 
slope of the regression line, along with the intercept not being significantly different from 
open pasture values, indicated a directly proportional inverse relationship between estimated 
Zone 3 %DIFN and CCL. Thus, overall, ceL appears to be a very good surrogate of Zone 
3 %DIFN across a wide range of poplar stand densities. This result agrees with Buckley et 
al. ( 1999), who found similar relationships for stands dominated with either deciduous red 
oak (Quercus rubra L.) or evergreen red pine (Pinus resinosa Ait.). Nevertheless, there 
Chapter 3 78 
were notable differences in the methodologies used by the two studies. More specifically, 
in the study by Buckley et al. ( 1999) light measurements were based on instantaneous 
quantum-sensor readings taken from a number of different locations for each stand density, 
and CCL was determined with a concave spherical densiometer. 
The relationship between CCNL and Zone 3 %DIFN was more complicated, with the 
unbalanced nature of the pooled dataset making it difficult to ascertain its general 'true' 
form. For the pooled data, when going from open pasture to a low CCNL, Zone 3 %DIFN 
decreased initially at a much slower rate in winter than summer. However, as CCNL 
continued to increase so too did the rate of decline in Zone 3 %DIFN. This contrasted with 
the constant (linear) rate of decline found for in summer. Factors contributing to the change 
in the relationship may have been the less specific absorption of PAR wavelengths by the 
leafless canopy, and also a greater proportion of diffuse radiation making up global (direct + 
diffuse) radiation penetrating to the stand floor. 
Site differences in the relationship between CCNL and Zone 3 %DIFN caused further 
complications (Figure 3 . 8a). At Ballantrae the small nature of the trees (i.e. height, trunk: 
and branch size) resulted in stands having only a very low CCNL and thus they caused little 
obstruction even when they were closely planted. However, the lack of relationship amongst 
the trees at Hautope 1 was less easy to explain. Potentially, the boundary trees projecting 
outwards from the four nuclei trees of each experimental unit at this site may have been 
more closely and similarly spaced apart. As a consequence, the incoming radiation could 
have been blocked similarly by th.e mass of tree trunks collectively visible low on the 
horizon, while, the trees nearest the point of interest caused little obstruction as they only 
appeared as narrow tapering columns (Anderson 1 964b). 
BA, HPCC, and CEV all had very strong relationships with Zone 3 %DIFN in summer 
(Table 3 .3) and moderately strong relationships in winter (Table 3 .6). In general, these 
relationships were either more complicated or their strengths weaker than for CCLINL 
(Tables 3 .3 & 3.6). It is likely that the same factors influencing the form of the winter 
relationship for CCNL also caused the relationships for BA, HPCC, and CEV in winter to 
straighten. Wang & Baldocchi (1989) showed a similar effect to the above when using the 
Chapter 3 79 
leaf area index (LA!) and woody biomass area index (WBI) of a stand to predict PAR 
transmission for foliated and non-foliated periods, respectively. 
Several researchers for a variety of different tree species have found strong inverse 
relationships between BA and PAR transmission (Perry et al. 1969; Vales & Bunnell 1988; 
Buckley et al. 1999). The BA and LA! of a stand are highly correlated (Buckley et al. 
1999). Thus, normally the form of the relationship shares similarities with Beer's law 
(Monsi & Saeki 1953, cited Vales & Bunnell 1988) and contains negative exponential 
curvature (as shown in Figure 3.4b). However, one notable exception between the assigned 
regression model for the summer period and Beer's law was that in the former there was no 
light extinction point. Instead, Zone 3 %DIFN approached a lower asymptote of around 
20% (Table 3.3; Figure 3.4b). 
Canham et al. (1999) noted that since most ofthe light that reaches a stand floor in summer 
penetrates through regions of the sky near the zenith, shading would be determined more by 
the horizontal cross-sectional shape/area of the crown rather than its vertical cross section. 
This may have been one of the primary reasons why HPCC was strongly related to Zone 3 
%DIFN especially over the fully in-leaf summer period (Table 3.3). Kuuluvainen & 
Pukkala (1987) also showed, using a theoretical model, that differences in the vertical 
shape of a tree had little effect on the area of shade it produced when the sun was 
positioned at a high elevation above the horizon. However, Sibbald et al. (1994) found that 
within stands of Sitka spruce (Picea sitchensis [Bong.] Carr.) at Scottish latitudes the 
horizontal component of incident light was more important for understorey pasture 
production than the vertical component. This would seem logical, as a high proportion of 
annual pasture production would occur in spring with the sun at a mid-elevation above the 
horizon. If the vertical component of the canopy was especially important in determining 
Zone 3 %DIFN then CEV should have provided a stronger relationship, as it takes into 
account both the horizontal and vertical components of the tree crowns. However, for the 
in-leaf summer period CEV had a higher RMSE and lower adjusted r2 than HPCC (Table 
3.3); whereas, in the leafless winter period the strength of the relationships for these two 
independent variables were similar (Table 3.6). This indicates that in the present study the 
horizontal component of the canopy more strongly influenced Zone 3 %DIFN. 
Chapter 3 80 
Variation in foliage density per unit of crown volume for the different aged trees (Wang et 
al. 1 995) at each site could have caused the marked site differences in the relationship 
between CEV and Zone 3 %DIFN for in summer (Figure 3 .5) .  Hybrid phenology or 
environmental stresses (e.g. soil moisture) may also have affected the density of foliage 
present on the individual tree crowns (Yates & McKennan 1 989; Messier et al. 1 998; 
Liefi'ers et al. 1 999). 
When pooled across all four sites DBH, GCL, and HPCD individually were poor to 
moderate predictors of Zone 3 %DIFN for both summer and winter periods (Tables 3 .3 & 
3 .6). This contrasts with findings by Vales & Bunnell (1 988) and Comeau et af. ( 1998) who 
found that DBH was strongly related to light transmission. These previous studies each 
related light transmission to DBH through the use of a single exponential decay type curve. 
Although the work by Vales & Bunnell ( 1988) was based on a broad range of stand 
structures, both previous studies were limited to small- to mid-sized trees stocked at a 
comparatively high rate per hectare. Conversely, in the present study there were large 
differences in the average size of the trees at each site, especially Ballantrae versus Kiwitea 
and Hautope 1 (data not shown). 
On an individual site basis, under trees of similar size and form, GCL, DBH, and HPCD 
were all strongly related to Zone 3 %DIFN, as shown for GCL. The general shape (i.e. 
negative concave curvature) of the relationships were also similar to those presented by 
Vales & Bunnell ( 1988) and Comeau et al. ( 1 998). Therefore, to improve the assigned 
regression models quality of fit for the pooled data and to generally make them more robust, 
another independent variable was likely required to account for the average size/form 
differences of the trees at each site. Similarly, in investigating the relationship between GCL 
and understorey pasture production for Pin us radiata D.Don stands, Percival et al. (1 984) 
and Percival & Knowles (1988) found that the mean GCL per tree was a useful covariate 
for combining data taken from a number of sites. This covariate became increasingly 
important as the range of tree ages increased (percival et al. 1984). 
Comeau et al. ( 1998) stated that further testing was required to determine if their developed 
relationship between DBH (sensu 'Lorimer's competition index') and PAR transmission for 
paper birch (Betula papyrijera Marsh.) could be generalised to other sites. The results from 
Chapter 3 81  
this study indicate that the above relationship would be in error if applied t o  stands 
containing trees of vastly different average size/form for a given DBHIha. 
Miller (1 959, cited Vales & Bunnell 1 988) proposed the hypothesis that canopy closure 
alone would not accurately predict light transmission because it does not take into account 
crown depth, which also influences light attenuation. However, out of the stand density 
indices that were tested, CCLINL measurements centred on Zone 3 were the least affected 
by site differences likely related to tree size. Pyke & Zamora (1 982) and Vales & Bunnell 
( 1 988) noted that any estimate of canopy closure made using an instrument with at least a 
moderate angle of view will include some crown depth and possibly serve as a better 
integrator of the three dimensional nature ofthe intercepting surface. 
Chapter 3 82 
3.5.2 The relationship between stand density indices and Zone 1 %DIFN for 
summer and winter 
3.5.2.1 General relationship 
In both summer and winter, as the stand density increased from open pasture to low stand 
densities there was a more marked decline in PAR transmission (%DIFN) directly below the 
poplar crowns (Zone 1 )  than in the vertically projected gap between the crowns (Zone 3). 
A number of factors likely contributed to the initially more rapid decline in Zone 1 
compared to Zone 3 %DIFN. In Zone 1 ,  the tree crown directly overhead, along with the 
surrounding trees, would obscure a larger portion of the total sky hemisphere. The tree 
crown directly above Zone 1 in particular would block out the disproportionately brighter 
region of the sky, which is generally located towards the zenith (Anderson 1 964a, 1 966). 
Furthermore, in comparison to Zone 3, a greater proportion of PAR transmitted from in­
between the tree crowns to Zone 1 would strike the ground at angles of incidence other 
than perpendicular to the plane of irradiation, this resulting in the energy received being 
spread over a larger area, as formalised in Lambert's cosine law (Barnes et al. 1 997). 
N � • S 
Zone 1 Zone 3 
Figure 3.15 Schematic diagram illustrating the smaller angle of sky view seen from Zone 1 and its general 
skew away from the zenith compared to Zone 3 .  
Chapter 3 83 
Figure 3.15 illustrates the smaller angle of sky view seen from Zone 1 and its general skew 
away from the zenith. From this figure it is apparent that at low stocking rates, smaller trees 
with either a greater crown height or less width will increase the angle of sky view seen 
from Zone 1; the latter two attributes in particular increasing the sky view towards the 
zenith. A greater crown height relative to total tree height would also reduce crown depth 
(GeL/stem) and thus the overall path length that light must pass through the canopy. 
Shorter path lengths usually increase the probability of light passing through the canopy 
without being intercepted (Hanan & Begue 1995). Differences in such tree attributes may 
have caused the significant differences in Zone 1 %DIFN found between the sites in both 
summer and winter. The aspect and slope of a site would also have influenced the region of 
sky seen from Zone 1 (Takenaka 1988) and the angle of incidence at which light strikes the 
ground (Bames et al. 1997). 
Variation in the above tree and site attributes would likely affect each of the relationships 
developed for Zone 1. For example, shorter, thinner, and higher pruned trees would reduce 
the initial rate of decrease in %DIFN when going from open pasture to low tree stocking 
rates. Differences in average crown depth may also affect the %DIFN as the stand becomes 
closed. Therefore, in order to make the developed regression models for Zone 1 more 
robust and precise, additional independent variables related to the above attributes will 
likely be required. 
Even in their current form, the predicted response for, Zone 1 %DIFN compared well with 
light measurements made by Guevara-Escobar et al. (1997) for a mature Populus x 
euramericana stand. Under a stand very similar to at Kiwitea and Hautope 1 ,  with an 
average BA of 14  m2Jha, PAR transmission was 20% directly below large untended trees. 
At the same stand density the predicted response was 17% PAR transmission, when using 
BA as the independent (predictor) variable (Figure 3.7b). 
On the other hand, there was considerable variation between field measurements taken by 
Douglas et al. (2001) and the predicted response for Zone 1 %DIFN at an equivalent stand 
density. Based on the analysis of hemispherical photographs, Douglas et al. (2001) 
estimated that solar irradiance under medium size Populus maximowiczii x P. nigra trees 
was 62% and 91 % of adjacent open-pasture plots in summer and winter, respectively. The 
Chapter 3 84 
BA for this study ranged between 2-8 m2/ha. In comparison, the predicted transmission of 
PAR at the lower end of this BA range for the present study was 55% and 73%, 
respectively (Figures 3.7b & 3 . l Ob). Differences in phenology between the poplar hybrids 
and the aforementioned tree and site attributes may have caused the above discrepancies. In 
addition, Douglas et al. (2001)  measured total short-wave radiation, which has a higher 
level of transmission compared to PAR, since it includes a greater range of wavebands not 
selectively absorbed by the tree leaves (e.g. near-infrared radiation) (Wang & Baldocchi 
1 989). 
Similarly to Zone 3, between summer and winter the relative increase in Zone 1 %DIFN 
was greater under high compared to low stand densities (Figures 3 .7a,b vs. 3 . l 0a,b). As a 
result, the variation in Zone 1 %DIFN across the range of stand densities was comparatively 
less in winter. Anderson ( 1964b) also found that, while the total daily radiation received in 
the understorey of three deciduous forest sites was very different in summer, there was little 
difference once the trees lost their leaves in winter. Two of the forest sites had an almost 
continuous tree canopy, while the other contained a large canopy gap of approximately 20m 
in diameter (Anderson 1 964b). The greater uniformity amongst the three forest sites in 
winter was attributed to the incoming radiation being blocked similarly by the mass of tree 
trunks collectively visible low on the horizon, while, the trees nearest the point of interest 
caused little obstruction as they only appeared as narrow tapering columns (Anderson 
1 964b). 
3.5.2.2 Rankings of stand density indices for estimating Zone 1 %DIFN 
For the pooled data of both summer and winter periods the regression models fitted using 
CCLINL, BA, HPCC, and CEV as single independent variables accounted for a greater 
proportion of the variation in Zone 1 %DIFN and had smaller RMSE than those using 
DBH, GCL, or HPCD (Tables 3 . 5  & 3 .9). However, site differences in the relationships 
were evident for all of the tested stand density indices, as shown for CCLINL and BA in 
Figures 3 .7a,b and 3 . l 0a,b. The differences were particularly marked for DBH, GCL, and 
HPCD as their datasets were generally more balanced. The imbalanced nature of the pooled 
datasets for CCLINL, BA, HPCC, and CEV may have contributed to the inconsistency in 
the form of relationships between the two periods. Generally, as a result of the above 
Chapter 3 85 
factors no single index was clearly better than any of the others in estimating Zone 1 
%DIFN. Further sampling under young/small and old/large trees at high and very low stand 
densities, respectively, along with the investigation of covariates to account for site 
differences, is required to strengthen the developed relationships. 
3.5.3 Practical considerations 
The stand density indices based on physical tree measurements had a number of practical 
advantages over the digital camera technique. Out of the stand density indices, DBH and 
BA followed by HPCD and HPCC were the easiest to measure and required minimal 
equipment. In comparison, the measurement of GCL and CEV generally took longer, 
especially under large trees, and required more specialised equipment in the form of a 
measuring pole and hypsometer. All of the above stand density indices can be sampled 
throughout the year under almost any weather conditions (Comeau et al. 1 998). 
Using a tape measure, tripod, and digital camera, the collection of canopy images was also 
relatively quick and easy. However, uniformly overcast conditions were needed to optimise 
contrast between the canopy and sky. This limits the usefulness of the technique in remote 
locations, where there may be only a narrow 'window' of opportunity for gathering data. 
Furthermore, the digital image analysis can take a considerable amount of time (Englund et 
al. 2000; Brandeis et al. 2001 ), especially if the images were not taken under optimum 
conditions. 
3.5.4 Relationships between the best stand density indices for estimating 
PAR transmission (%DIFN) 
3.5.4.1 Relationship between canopy closure with leaves (CCl) and basal area (BA) 
Many researchers have found a strong relationship between the CCL and BA of a stand, for 
a variety of deciduous and conifer species, and also using a variety of techniques for 
measuring canopy cover (e.g. spherical densiometers, canopy digital images, etc.). Similar 
to the relationship developed for the present study (Table 3 . 1 0; Figure 3 . 12c), most take the 
form of a growth model with an exponential rise to a maximum (upper asymptote) (Mitchell 
& Popovich 1 996; Buckley et al. 1 999; Knowles et al. 1 999; McElwee & Knowles 2000). 
Chapter 3 86 
In Figure 3 . 1 6, the data and the predicted response from the present study are compared 
with functions developed by McElwee & Knowles (2000) and Buckley et al. ( 1999) for 
deciduous poplar and oak stands, respectively. The study by McElwee & Knowles (2000) 
encompassed 40 different poplar stands (mainly P. deltoides x P. nigra) from the Bay of 
Plenty and East Coast regions of the North Island, New Zealand; while, the study by 
Buckley et al. ( 1999) was based on even-aged oak stands (dominated by Q. rubra L.) 
located in northern Lower Michigan, USA. Both of the functions developed in these studies 
were within the 95% prediction limits of the present study's growth model (data not 
shown). However, in comparison to McElwee & Knowles (2000), the present study's 
growth model tended to predict a higher CCL for a given BA, especially between 1 5m2/ha 
to 25m2/ha (Figure 3 . 1 6). The similarity between the function developed by Buckley et al. 
( 1 999) and the present study was surprising given the innate differences in crown phenology 
between the two tree species. 
There are a number of potential reasons why there was systematic variation between the 
growth model of McElwee & Knowles (2000) and the present study. The crown-height-to­
tree-height ratio (i.e. height ratios) of the respective stands of trees from the two studies 
may have been markedly different. Generally, stands with greater height ratios for a given 
BA, have a lower CCL (Knowles et al. 1999; McElwee & Knowles 2000). For example, 
through the pruning of lower branches; Devkota (2000) increased the height ratio of an 
even-aged and -stocked Italian grey alder (Alnus chordata) stand from 0.24 to 0.7 1 ,  causing 
CCL to decrease from 89% to 4 1  % (Devkota 2000; Devkota et al. 2001 ). For stands of P. 
radiata, Knowles et al. ( 1 999) included the height ratio as an independent variable to take 
into account the effects of green crown pruning and found it was a significant factor. 
McElwee & Knowles (2000) also investigated adding the height ratio into their empirical 
model, but it was not a statistically significant addition. An insufficient range of height ratios 
may have inhibited the reflection of this effect (McElwee & Knowles 2000). 
8 Height ratio is defined as the vertical height from ground level to the base of the live tree crown (crown 
height) divided by the vertical height from ground level to the top of the tree (total height) (Knowles et al. 
1 999; McElwee & Knowles 2000). 
Chapter 3 87 
The local site conditions may have been important, with environmental stresses, such as a 
soil moisture deficit, reducing normal leaf area display (Yates & McKennan 1 989). Given 
that similar poplar species and hybrids were sampled in the two studies it is unlikely that 
variation in individual crown phenology was, at least partially, responsible for differences 
between the functions describing the relationship between BA and CCL - although this 
could be a factor when applying the developed functions to newer hybrids. 
The different techniques and apparatus used to sample CCL could also have contributed to 
the variation found between the two studies. Bunnell & Vales ( 1 990) reported that from a 
common point in a forest understorey, techniques/apparatus that measure a wider angle of 
view of the upward sky-hemisphere gave a higher estimate of canopy closure (sensu crown 
completeness), with less variation around the mean. In general, the difference was greater 
under low CCLs (Bunnell & Vales 1 990). Similarly, underexposing canopy photographs 
causes an increase in the estimate ofCCL (Vales & Bunnell 1 988; Chen et al. 1 99 1 ) . 
Frazer et al. (2001)  found that differences in the resolution of digital-images of dense 
overstorey canopies have a small effect on the estimate of CCL ( <5% difference). In 
comparison, the subjectivity involved in techniques used to visually separate canopy and 
sky pixels of digital-images could be a greater source of error (Machado & Reich 1 999; 
Englund et al. 2000). Several researchers have repeated the processing and analysis of each 
image several times to reduce this potential error (McKennan 1995; Gendron et al. 1 998; 
Englund et al. 2000; Frazer et al. 2001) .  
Overall, the above factors indicate that any future model development of the relationship 
between eCL and BA should include the average height ratio of individual trees as an 
independent variable to make the function more robust in regards to common silvicultural 
management practices. In addition, the methodology for taking digital canopy-images (e.g. 
lens field of view & image exposure) and image analysis (e.g. repeated analyses) needs to be 
standardised to reduce the error incurred when estimating eCLINL. 
Chapter 3 88 
3.5.4.2 Relationship between canopy closure with (CCl) and without leaves (CCNl) 
The clear site differences between BaIlantrae and the other three sites for the relationship 
between CCL and CCNL indicated that at least some other unaccounted factor was 
significantly influencing the relationship (Figure 3 . 1 2a; Table 3 . 10). It is uncertain what 
factor or factors may have influenced this relationship, but obvious differences were the 
trees at BaIlantrae were less than one-fifth the age and had a height ratio two to five times 
greater than at the other sites (Table 2. 1 ). A young tree would have considerably less 
woody biomass area index (WBI) associated with its main stem and branches (Scanlan 
1 99 1 ;  Wang et al. 1995); while, a greater height ratio would reduce the WBI associated 
with branches and reduce the depth of the canopy. Based on a simple linear function forced 
through the origin, McElwee & Knowles (2000) estimated that CCNL was 3 1  % of CCL. In 
comparison, when the simple linear functions for the young (BaIlantrae) and mature 
(Kiwitea & Hautope) stands were also forced through the origin CCNL was 7% and 55% 
of CCL, respectively (Figure 3 . 1 7). It is difficult to speculate whether the differences in 
slope between these functions were due to the above aforementioned factors, as McElwee 
& Knowles (2000) did not disclose the specific age or height ratio of the stands used for 
developing their simple linear relationship. 
Chapter 3 
.-.. 
'$. '-' 
Vl 
V 
;> 
� 
..c -
. � 
v ... 
;:3 Vl 0 
U 
:>-0-0 C c:.s 
u 
1 00 ,--------------------------------------------, 
80 
60 
•• 
40 
• 
20 
... 
". .;' 
". 
. . . .
. . . • 
• • 
--
--
".--
-
.-- ---- ..... -­-
o �--------�---------r--------�--------_.--------� 
o 1 0  20 30 40 50 
Stand basal area (m2/ha) 
89 
Figure 3. 16 Relationships between in-leaf canopy closure (CCL) and basal area (BA). Symbols: (-e-) data 
points and predicted response from the present study, ( .. ) data point for a single poplar stand (Guevara­
Escobar 1 999), predicted response for a range of ( ) poplar (McElwee & Knowles 2000) and C .. · ) red oak 
stands (Buckley et af. 1 999). 
60 
• 
.-.. 
'$. 50 '-' 
Vl 
V 
;> 
c:.s 40 � 
5 0 ..c 
.� 30 
11) ... 
;:3 Vl 0 
U 20 :>-0-0 C 
'" 
U 1 0  
0 
0 20 40 60 80 1 00 
Canopy closure with leaves (%) 
Figure 3. 17  Comparison of relationships between no-leaf canopy closure (CCNL) and in-leaf canopy closure 
(CCL). Symbols: (-e-) mature (K iwitea & Hautope) and C ·0 . .  ) young stands (Bal lantrae), and (­
predicted response for a range of poplar stands from McElwee & Knowles (2000). 
Chapter 3 90 
3.5.5 Stand level %DIFN across a range of canopy closure ratios in summer 
and winter. 
Similar to the summer relationship derived in the present study, Vales & Bunnell ( 1 988), 
Barnes et al. ( 1997), and Buckley et al. ( 1999) also reported strong inverse relationships 
between CCL and PAR transmission. The relationships given by Vales & Bunnell ( 1 988) 
and Barnes et al. (1 997) for conifer tree species contained negative curvature, with the 
slope decreasing towards higher CCLs. This contrasted with Buckley et al. ( 1 999) who 
found a negative 1 :  1 linear relationship for both deciduous and conifer species. The 
sampling methods used (Anderson 1 964b; Vales & Bunnell 1 988; Englund et al. 2000) and 
structure of the stands (Barnes et al. 1997) could have caused the differences in the form of 
the above relationships. 
Parallel with Zone 3 %DIFN in winter, it is likely that less specific absorption of PAR 
wavelengths by the woody biomass and greater downward reflection of incident light at low 
CCNLs caused the relationship for weighted %DIFN to change shape from being concave 
to convex (Figure 3 . l3a,b). Overall, both the summer and winter relationships for weighted 
%DIFN compared well with single density measurements previously taken under the closed 
canopies of various deciduous tree species (Table 3. 1 3) .  Nevertheless, further sampling 
under young/small and oldllarge trees at high and very low stand densities, respectively, and 
especially covering the range between 1 0-40% CCL, would clearly strengthen the 
developed relationships for weighted %DIFN. 
Chapter 3 91  
Table 3.13 Light transmission through various closed stands of deciduous trees. 
Source Genus BA Total Height Light transmission (%) 
(m2/ha) (m) Summer Winter 
Hughes et al. (1985) Quercus 47 1 8  10  55 
Messier et al. ( 1998) Populus 33 20 10 
Betula 2 1  1 7  14  
Chen et al. ( 1997) Populus 26 2 1  26 55 
Carlson & Groat (1997)t Populus 36 1 9  1 8  
Mourelle et al. (200 1)  Populus 1 7  15  
Reifsnyder et  al. ( 1971/72)t Mixed hardwood 1 8  9 
Wang & Baldocchi ( 1989) Quercus & Carya 2 1  60 
tLight transmission was measured using pyranometers, which measure a wider range of short-wave 
radiation wavelengths compared to quantum sensors. Given that leaves preferentially absorb PAR over near-
infrared radiation wavelengths the above transmission values would likely be an overestimate of PAR 
transmission. In the study by Reifsnyder et al. ( 1971/72) maple were the dominant species, along with a 
considerable admixture of oak and ash. 
3.5.6 Potential error incurred through using %DIFN measurements 
A number of researchers have voiced concerns about using instantaneous measurements, 
such as %DIFN, to characterise seasonal PAR transmission levels (Anderson 1 964a,b; Gay 
et al. 1 97 1 ; Hutchison & Matt 1 977b; Canham 1 988; Canham et al. 1990; Stadt et al. 1 997; 
Tang et al. 1 999). The measurement of only diffuse short-wave radiation expressed as a 
percentage of open conditions (e.g. %DIFN) can mask important features of spatial and 
temporal variation in global (diffuse + direct) short-wave radiation below a stand of trees 
(Anderson 1 964b; Gay et al. 1 971). 
Due to scattering in the atmosphere diffuse radiation emanates from the sky hemisphere 
rather similarly from all compass directions (azimuths) with its brightness distribution 
increasing from a minimum at the horizon to a maximum at the zenith (Hutchison & Matt 
1 976; Stadt et al. 1997). As a result, within an inter-tree gap the general pattern of diffuse 
radiation tends to be quite uniform/symmetrical with the greatest level at the centre, roughly 
decreasing at a similar rate in all directions towards the gap edge and beyond (Takenaka 
1988; Canham et al. 1 990; Runkle et al. 1 995). 
Chapter 3 92 
In contrast, outside of the tropics large diurnal and seasonal changes in the apparent 
position of the sun (solar disk) causes direct-beam radiation to have a much more varied 
asymmetrical distribution. For example, at New Zealand's temperate latitude the sun never 
actually passes directly overhead, but instead during the course of a day appears to cross 
from east to west in the northern part of the sky (Sturman & Tapper 1996). As a 
consequence of the sun's angle with the ground considerably more direct-beam radiation is 
often transmitted to understorey areas adjacent to the sun (north9) facing edges of either 
widely spaced trees or large canopy gaps compared to opposite understorey areas adjacent 
to shade (south) facing edges. Yet both have a similar sky view and thus receive roughly 
equal amounts of diffuse radiation (Sibbald & Griffiths 1 992; Runkle et al. 1 995; Stadt et 
al. 1 997; Douglas et al. 2001). Therefore, by measuring only diffuse radiation spatially 
amongst stands with large inter-tree gaps global (diffuse plus direct) radiation transmission 
will likely be under- and overestimated in some areas (Lieffers et al. 1 999). 
The extent of the above skewed north-south gradient varies with the sun's elevation above 
the horizon, but also depends on the inter-tree gap configuration relative to tree size/form, 
cloud cover, and the topography of the site (Canham 1 988; Takenaka 1 988; Sibbald & 
Griffiths 1 992; Canham et al. 1 990; Lieffers et al. 1 999). Even though the latitude of the 
four sampled sites was generally similar, and thus so was the solar track across the sky, the 
latter factors varied considerably. At Kiwitea and Hautope 2 the steep north-facing hillsides 
that the stands of trees were situated on would likely have allowed direct-beam radiation to 
penetrate inter-tree gaps at an angle much closer to perpendicular to the ground than at 
Ballantrae and Hautope 1 reducing the skewed distribution (especially around the summer 
solstice). Table 3 . 14 shows that during the growing season direct-beam radiation must 
generally pass through gaps in the tree crowns to reach their base at Kiwitea and 
Hautope 2, while the shadow length produced by the same trees is much smaller than for 
the other two sites. 
9 In Northern Hemisphere latitudes outside of the tropics (>23.5"N) it is the opposite way around with the 
sun facing edges southward (poulson & Platt 1 989; Canham et al. 1990; Runkle et al. 1995). 
Chapter 3 93 
Table 3.14 Maximum solar angle'l' above the horizon allowing direct-beam radiation to pass below the 
crown of an average sized tree to its base and also the shadow length produced by the same tree. 
Season Time Solar elevation Shadow length as a proportion of crown radius 
(degrees) Kiwitea Hautope 1 Hautope 2 Ballantrae 
Autumn/spring equinox mom.lafter.t 22.5 9.9 1 3 .0 1 0.7 1 7.5 
noon 50.0 2.7 6.9 2.9 6 . 1  
Summer solstice mom.lafter. 37.4 5.4 7. 1 5 .8  9.5 
noon 73 .4 1 .2 1 .9 1 .3 2.2 
Winter solstice morn.lafter. 5 .5  42.6 56.0 46.2 75.2 
noon 26.6 4.5 42.0 4.9 14.5 
Crown radius (m) 7.3 4.6 3.0 1 .0 
'I' Max. angle (degrees) 2 1 .7 49.6 36.2 74.5 
tMorning and afternoon times were taken as 4 hours either side of solar noon. Formulas for determining the 
solar elevation and shadow lengths were obtained from Takenaka ( 1988), Sturman & Tapper ( 1996), and 
Ban et al. ( 1 998). Latitude was assumed to be 400S for all sites. Average crown height/radius ratio at 
Kiwitea, Hautope 1 ,  Hautope 2, and Ballantrae was 0.4, 1 .2, 0.7, and 3 .6, respectively. Ground slope for 
morning/afternoon was assumed to be 0°, while at noon slopes were 28�, 200S, 27�, and O� for Kiwitea, 
Hautope 1 ,  Hautope 2, and Ballantrae, respectively. Dates for autumn/spring (vernal) equinox, summer and 
winter solstices were: 2 1  March / 22 September, 2 1  December, and 2 1  June, respectively (Sturman & 
Tapper 1 996). It should also be noted that radiation hitting the ground at angles other than perpendicular to 
the plane of irradiance is distributed over a greater area, causing the amount of energy received per unit 
area to be less (Lambert's cosine law). Thus, at low angles of incidence, the intensity of direct-beam 
radiation will be considerable less than at higher angles (Barnes et al. 1 997). 
Nevertheless, several other researchers have shown that single point-in-time measures of 
solely diffuse radiation can provide a very good estimate of mean daily or seasonal global 
PAR transmission (Washitani & Tang 199 1 ;  Hanan & Begue 1 995 ; Messier & Puttonen 
1 995; Parent & Messier 1 996; Comeau et al. 1 998; Gendron et al. 1 998; Machado & Reich 
1 999). These studies spanned a wide range of stand types, including both homogenous and 
heterogeneous canopies, and measurements were taken from a number of different 
micro sites. In fact, Comeau et al. ( 1998), Gendron et al. ( 1 998), and Machado & Reich 
( 1 999) reported that for estimating global PAR transmission diffuse radiation methods 
including %DIFN were as good as hemispherical (fisheye) photography, which does take 
into account changes in solar elevation. 
Chapter 3 
3.5.7 The relationship between R:FR and %DIFN 
3.5.7.1 Under a fully in-leaf poplar canopy in summer 
94 
As shown in Figure 3 . 14a, the red to far-red ratio (R:FR) decreased from open pasture 
values at an increasing rate as the transmission of PAR (%DIFN) through the poplar canopy 
declined. This relationship shared similarities with the idealised relationships postulated by 
Lieffers et al. ( 1 999) for deciduous and coniferous dominated stands. Both of the above 
empirical and idealised relationships showed marked decreases in the R:FR below a PAR 
transmission level of about 40%. Within the vertically projected gap between the poplar 
crowns (Zone 3), estimated PAR transmission (%DIFN) was lower than 40% at a canopy 
closure >58% (Figure 3 .4a); whereas, directly below the crowns (Zone 1 )  of the mature 
unpruned trees at Hautope and especially Kiwitea this critical point was surpassed over the 
entire range of canopy closures (Figure 3 .7a). Conversely, at Ballantrae the %DIFN 
underneath the crowns (Zone 1) of the young pruned trees did not fall below 65% 
(Figure 3 .7a); and as a result would have lessened the magnitude of variation in R:FR across 
the tree-gap continuum. 
In the idealised relationships of Lieffers et al. ( 1 999) the R:FR reached a maximum at 
around 40% light transmission and thereafter remained relatively constant. This contrasted 
with the logarithmic relationship developed in the present study, which did not reach a 
maximum until 1 00% PAR transmission (Figure 3 . 1 4a). Similarly, Ritchie ( 1 997) measured 
a logarithmic increase in the R:FR with decreasing plant density based on a greenhouse 
experiment using Douglas fir (Pseudotsuga menziesii (Mirb.) Franco.) seedlings, which 
were systematically spaced apart in a miniature ''Nelder'' (Nelder 1 962) design. 
A number of previous studies, which measured the light environment under clear sky 
conditions and allowed data points to be converted to values relative to the open pasture, 
fitted the developed relationship well (Figure 3 . 1 8) .  The good fit of Ross et al. ( 1986) and 
Messier et al. ( 1 989) data from coniferous stands was inconsistent to that of  other 
researchers who have found differences in the spectral absorbance characteristics between 
broadleaf and conifer species (Coombe 1 957; Federer & Tanner 1966; Vezina & Boulter 
1966; Tasker & Smith 1 977; Morgan & Smith 1 98 1). In general the light spectrum under 
conifers tends to be more uniform, with a greater intensity of blue light (455-500 nm) and a 
Chapter 3 95 
lower intensity of infrared (>700 nm) light on clear days compared to under broadleaf 
species (Federer & Tanner 1 966; Morgan & Smith 1 98 1 ;  Smith 1 982). However, in more 
recent studies by Ross et al. ( 1 986) and Endler ( 1993) little difference in the attenuation of 
various light spectra, including the R:FR, were found between these two broad groups. 
Some of the variation observed among the studies in Figure 3 . 1 8  could have been the 
consequence of different sampling/measurement techniques (Smith 1 982; Morgan et af. 
1 985). In particular, atmospheric conditions (e.g. the degree of cloud cover) at the time of 
measurement can have a marked effect on the understorey R:FR (Morgan et al. 1 985; 
Messier et af. 1 989; Messier & Puttonen 1 995; Lieffers et al. 1 999; Reitmayer et af. 2002). 
Overall, confirmation of the appropriateness of the assigned logarithmic relationship using 
previously published research was hindered by their lack of data points above 40% PAR 
transmission (Figure 3 . 1 8). 
Chapter 3 96 
1 20 
� 1 00 OIl 0: 
c:: � 
� """' 
& �  80 0: e ell 0: :s 
. �  e: 0. 60 
a: c � ;>-. 0.  � 0 ... ...... £ 0 40 <rl ... � " c 
::> 20 
0 
o 20 40 60 80 1 00 1 20 
Estimated PAR transmission (%) 
Figure 3.1 8  Comparison of the developed relationship between R:FR and estimated PAR transmission with 
data from other sources. Symbol: (e) present study, (0) Morgan & Smith ( 1 98 1 ), (0) Ross et at. ( 1 986), ( ) 
Messier & Bellefleur ( 1 98 8), (_) Messier et at. ( 1 989), ( . )  Devkota (2000). All of the R:FR measurements 
were taken under predom i nantly clear sky conditions, except for the present study, where m easurements were 
taken under a range of atmospheric conditions. The sky conditions in the study by Morgan & Smith ( 1 98 1 )  
were not speci fied. Solid l ine - mean predicted response from the present study; dashed lines - upper and 
lower 95% prediction l im its. 
Chapter 3 97 
Under forest stands dominated by deciduous trees, Tasker & Smith (1 977), Pons ( 1 983), 
Hughes et al. ( 1 985), and Ross et al. (1986) all measured marked decreases in the average 
R:FR with leaf emergence in spring. Thereafter, the R:FR remained relatively stable over 
the main growing season until leaf senescence began in autumn. Nevertheless, similar to 
PAR transmission, over the course of a single day there may be considerable variation in the 
R:FR at any point on the ground, especially under open stands during clear-sky conditions 
(Ross et al. 1 986; Turnbull & Yates 1 993; Messier & Puttonen 1 995; Reitmayer et al. 
2002). As discussed in Section 3 .5.5, this temporal variation is caused by interactions 
between the changing sun-angle relative to the ground, stand biomass distnbution, and 
cloud cover (Endler 1993; Messier & Puttonen 1 995; Reitmayer et al. 2002). Single point­
in-time measurements, such as those used in the present study, cannot take into account 
such variation (Anderson 1 964a,b). As a result, this may limit the applicability of the 
developed relationships in this study to certain understorey plant responses where the time 
scale over which the variation in the R:FR occurs is important (e.g. phytochrome-mediated 
responses). Conversely, the light environment below a closed canopy tends to be much 
more uniform over time (Ross et al. 1 986; Turnbull & Yates 1 993). 
3.5.7.2 Under a leafless poplar canopy in winter. 
In contrast to summer (Figure 3 . 1 4a), as the percentage of PAR transmitted through the 
leafless poplar canopy in winter decreased, there was correspondingly a much smaller and 
more constant rate of decrease in the R:FR (Figures 3 . 1 4b). With all of Hautope l 's data 
points clumped at lower estimated PAR transmission (%DIFN) levels away from the other 
sites (Figure 3 . 1 4b), it is difficult to distinguish whether the pooled trend is significant or 
simply an artefact of variation between the sites. Regardless, the scatter of data points in 
Figure 3 . 1 4b show that without spectrally selective absorbing leaves the poplar overstorey 
has relatively little effect on the understorey R:FR, irrespective of stand density. The 
decrease in the impact of the overstorey may actually start to occur prior to leaf fall through 
chlorophyll degradation of the senescencing leaves (Ross et al. 1 986). Hughes et al. ( 1 985) 
found that under a deciduous oak (Q. robur L.) canopy, transmitted light was spectrally 
neutral (unaltered) after leaf fall in autumn and remained so up until bud break in the 
following spring. During the non-foliated period the R:FR remained close to unity under the 
oak canopy (Hughes et al. 1 985). 
Chapter 3 98 
3.5.8 The ecological sign ificance of changes in the understorey light 
environment for pasture plants. 
Decreases in PAR and the R:FR ratio under a poplar stand can have a number of effects on 
understorey pastures. With PAR wavelengths being directly involved in photosynthesis, a 
decrease in transmission through the tree canopy normally leads to a reduction in pasture 
carbohydrate and net dry matter (DM) production compared to in open pastures 
(Seo et al. 1989; Wong & Still 1996; Devkota et al. 1997; Sanderson et al. 1997). 
However, the reduction is not always in proportion to the decrease in PAR as plants can 
make compensatory changes (adaptations) in their morphology and physiology, which help 
to increase light interception and enhance the efficiency of carbon use in both 
photosynthesis and respiration (Ludlow et al. 1974; Corre 1983a; Assmann 1992; 
Sanderson et al. 1997; Wilson 1997; Healey et al. 1998). Nevertheless, below some critical 
threshold, even the above adaptations cannot compensate for the overall lower level of 
incident PAR reaching the plant, this resulting in decreased growth and development 
(Ludlow et al. 1974; Corre 1983a; Stuefer & Huber 1998). Conversely, several researchers 
have found that a change in the R:FR ratio of incident PAR does not directly influence the 
total aboveground DM production of individual pasture plants (Casal et al. 1985; Deregibus 
et al. 1985; Heraut-Bron et al. 1999; Devkota 2000). 
Combining the developed PAR transmission curves from Sections 3.4 with pasture 
production data given by Devkota et al. (1997) provides an initial estimate of the likely 
shape of the understorey pasture response to changes in the density of the poplar overstorey 
(Figure 3.19a,b). It is important to note that the pasture production data of Devkota et al. 
(1997) come from a glasshouse experiment, which used a range of spectrally neutral10 shade 
cloth densities (each providing a constant level of shade) to vary the intensity of PAR, and 
that soil nutrients and water were non-limiting for plant growth. Based on Figure 3.19a, 
pasture DM production in Zone 3 would decrease at a constant rate (i.e. linearly) with 
increasing CCL. In contrast, pasture production decreases initially at a much faster rate in 
Zone 1 than Zone 3, but thereafter falls at a diminishing rate. As discussed in Section 
3.5.2.1 for PAR transmission directly below the poplar crowns (Zone 1 DIFN%), reducing 
the crown depth and width of the individual trees (via pruning and species selection, 
Chapter 3 99 
respectively) would likely reduce the degree of negative curvature of the pasture production 
response. However, as the canopy becomes more uniformly closed both zones of tree 
influence approach similar rates of production regardless of the individual crown 
dimensions. The extinction point for pasture production in Figure 3 . 1 9a was at 85% CCL, 
which is the same as visual estimates made in the field by McElwee & Knowles (2000), and 
considerably higher than the 67% CCL estimated for P. radiata agroforestry (Knowles et 
al. 1 999; McElwee & Knowles 2000). 
The general shape of the pasture production response estimated for the two main zones of 
tree influence for the winter period was comparable to summer (Figures 3 . 1 9a,b). However, 
given that the maximum CCNL did not block more than 50% of the sky-hemisphere, 
pasture dry matter production was not estimated to fall below a similar value (Figure 
3 . 1 9b). 
10 Transmission of PAR through the shade cloth did not alter the R:FR from that measured in full sunlight. 
Chapter 3 1 00 
1 .4 
(a) 
1 .2 
---
01) '-' 
1 .0 .... 
§ 
0.. 
... IU 0.8 c.. 
... IU .... 
� 0.6 
C 
-0 .... 
0 0.4 0 
..c 
en 
0.2 
0.0 
0 20 40 60 80 1 00 
Canopy closure with leaves (%) 
1 .4 
(b) 
1 .2 
---
01) '-' 1 .0 
] 
c.. 
... IU 0.8 c.. 
... IU .... .... tU 
El 0.6 
C 
-0 .... 
0 0.4 0 
..c 
en 
0.2 
0.0 
0 20 40 60 80 1 00 
Canopy closure without leaves (%) 
Figure 3.19 Estimated pasture dry matter production (0) at the centre of a vertically projected canopy gap 
(Zone 3) and (�) directly below the poplar crowns (Zone 1 ), over a range of stand canopy closure ratios in 
(a) summer and (b) winter. The linear and non-linear relationships relating PAR transmission in the two main 
zones of tree influence (i.e. Zones I & 3) to stand canopy closure are given in Section 3 .4; whereas, the linear 
relationships between pasture production and PAR transmission are given in Devkota et al. ( 1997). For the 
latter, spectrally neutral shade cloth was used to vary the PAR transmission levels under glasshouse 
conditions; soil nutrients and water were non-limiting (Devkota et al. 1997). The botanical species 
composition for the above relationships was assumed to be 70% perennial ryegrass (Lolium perenne) and 
30% white clover (Trifolium repens). 
Chapter 3 1 01 
In general, plant morphology is more affected by changes in the intensity ofP AR rather than 
the R:FR ratio (Lee et al. 1 996; Lieffers et al. 1 999; Devkota 2000). Typical responses to 
reduced levels of PAR in the range of tree-shade include: an increase in leaf area ratiol l  
(Ludlow et al. 1 974; Corre 1 983a,b; Kephart et al. 1 992) and specific leaf area12 (Corre 
1 983a,b; Samarakoon et al. 1 990; Devkota et al. 1 997; Meziane & Shipley 1 999); leaf and 
petiole elongation in grasses (Taylor et al. 1 968; Wong & Wilson 1 980; Sanderson et al. 
1 997) and legumes (Solangaarachchi & Harper 1 987; Stuefer & Huber 1998); and a 
decrease in tillerlbranch appearance (Ludlow et al. 1 974; Wong & Stiir 1 996; Devkota et 
al. 1 997; Gautier et al. 1 999) and root biomass (Corre 1 983a; Eriksen & Whitney 1 9 8 1 ;  
Samarakoon et al. 1990; Wilson 1996). Overall, these changes generally produce more 
open (less dense) and etiolated understorey pastures, with higher shoot-to-root ratios than 
in the open (Wong & Wilson 1 980; Eriksen & Whitney 1 98 1 ). A reduction in the R:FR can 
also cause similar morphological responses (Casal et al. 1 985, 1 987b; Solangaarachchi & 
Harper 1 987; Thompson 1 993; Sanderson et al. 1 997). However, the responses tend to be 
more related to changes that enable plants to physically avoid impending shade (or 
alternatively, "forage for light") at the expense of leaf area development (Smith 1 982; Corre 
1 983b; Schmitt & Wulff 1 993 ; Heraut-Bron et al. 1 999). Regardless of the R:FR ratio, 
under very low PAR intensities, compensatory changes in plant morphology will be 
inhibited through restricted carbon nutrition (Morgan & Smith 1 98 1 ;  Casal et al. 1 986; 
Messier et al. 1 989; Stuefer & Huber 1 998; Heraut-Bron et al. 1999). 
Again, combining the developed PAR transmission curves with pasture production data 
given by Devkota et al. ( 1997) provides an initial estimate of the likely shape of the 
understorey pasture response in tillering/branching to changes in overstorey tree density 
(Figure 3 .20a,b). Within the vertically projected canopy gap (Zone 3) the relationship 
between tillerlbranch production and CCL was more convex in shape compared to the linear 
response shown for DM production (Figure 3 . 1 9a,b vs. 3 .20a,b). Directly below the poplar 
crowns (Zone 1), on the other hand, had a similar shaped response for both attributes 
(Figure 3 . 1 9a,b vs. 3 .20a,b). According to the simplified model developed, tillerlbranch 
I I  Leaf area ratio (LAR) is  the amount of leaf area displayed per unit of total plant biomass (m2/kg plant 
DM). 
12 Specific leaf area (SLA) is the leaf area per unit of leafweight (cm2/g DM). 
Chapter 3 1 02 
production in the perennial ryegrass/white clover sward would cease at around 85% CCL in 
summer (Figure 3 .20a). This contrasted with only a 30% reduction from open pasture levels 
at maximum canopy closure (52% CCNL) under the leafless poplars in winter (Figure 
3 .20b). 
Chapter 3 1 03 
30 
Ca) 
CIl Q) 25 -= u 
la .l5 ...... CIl .... 20 � la � -a  .... ... o Q) ... 0. 1 5  Q) "'O .0 Q) e u :::I :::I s:: "'O 
"'0 8 1 0  Q) 0. 
� 
§ u 5 � 
0 
0 20 40 60 80 1 00 
Canopy closure with leaves (%) 
30 
Cb) 
CIl Q) 25 ..c: u 
la .. .0 ...... 
� c:  � �  20 "' 0.  .... ... o Q) ... 0. 1 5  Q) "'O .0 Q) § g s:: "'O 
"'0 0 ... Q) 0. 10  1;; 
"3 
§ u 
5 u < 
0 
0 20 40 60 80 100 
Canopy closure without leaves (%) 
Figure 3.20 Estimated tilleringlbranching (0) at the centre of a vertically projected canopy gap (Zone 3) and 
(�) directly below the poplar crowns (Zone 1 ), over a range of stand canopy closure ratios in (a) summer 
and (b) winter. The linear and non-linear relationships relating PAR transmission in the two main zones of 
tree influence (i.e. Zones 1 & 3) to stand canopy closure are given in Section 4.4; whereas, the relationships 
between tillerslbranches and PAR transmission were derived from data given in Devkota et al. ( 1997). For 
the latter, spectrally neutral shade cloth was used to vary the PAR transmission levels under glasshouse 
conditions; soil nutrients and water were non-limiting (Devkota et al. 1 997). The botanical species 
composition for the above relationships was assumed to be 70% perennial ryegrass (LoZium perenne) and 
30% white clover (Trifolium repens). 
Chapter 3 1 04 
Solangaarachchi & Harper (1987), Gautier et al. (1999), and Devkota (2000) have all 
reported that a concomitant decrease in the R:FR enhances the effect of reduced PAR on 
tilleringlbranching. In the glasshouse experiment of Devkota et al. (1997) the R:FR was 
held constant at around that measured in open pastures, irrespective of the actual shade 
level. Therefore, in the developed relationships for summer, when the trees were fully in-leaf 
(Figure 3 .20a), the rate of decline in tillering/branching would probably be underestimated 
over certain parts of the curves. A small reduction in the R:FR of incident light has been 
shown to cause marked changes in the morphology of some plant species (Smith 1 982; 
Casal et al. 1986; Teuber & Laidlaw 1996). For example, Casal et al. ( 1987a) measured a 
strong decrease in the tillering of Lolium multiflorum resulting from a small reduction in the 
R:FR below values typical of full sunlight. 
There are a number of other factors that can also influence the response of understorey 
pastures to decreases in the level of PAR and R:FR ratio. The extent of changes in sward 
structure and total dry matter production under tree-shade will depend on the botanical 
species composition of the pasture (Figure 3 .21a,b). This is because there is considerable 
variation in shade tolerance and potential yield amongst and within different pasture species 
(Solangaarachchi & Harper 1987; Skalova & Krahulec 1 992; Devkota et al. 1 997, 1998; 
Naumburg et af. 2001). Shade tolerant (,shade') plants usually have lower growth rates in 
full sunlight, but in comparison to their shade intolerant ('sun') counterparts can maintain 
dry matter (DM) production closer to their maximum potential yields under reduced levels 
of PAR (Figure 3 .21a,b) (Boardman 1977; Bjorkman 1 981 ; Corn� 1 983b; Devkota et af. 
1 997; Reich et al. 1998; Naumburg et af. 2001) .  In relation to changes in plant morphology, 
shade intolerant species are normally more sensitive (or show greater plasticity) to small-to­
moderate decreases in the intensity of PAR and R:FR below open habitat values (Corre 
1 983a,b; Devkota et al. 1 997; Wilson 1997; Devkota 2000); typical responses include 
allocating a greater proportion of assimilates to organs that help move the plant's 
photosynthetic area into more favourable light conditions (Smith 1982; Corre 1983b). In 
contrast, shade tolerant species are less plastic, except when under very heavy shade 
(Devkota et al. 1997), and respond mainly to decreases in the intensity of PAR (Smith 
1 982; Corre 1983b; Caldwell 1 987) by allocating a greater proportion of assimilates into the 
development or maintenance of leaf area (Wong et al. 1985; Naumburg et al. 2001) .  
Chapter 3 1 05 
Nevertheless, there are exceptions to the above generalisations (Corre 1983b; Naumburg et 
al. 2001) .  
The effect of shade on understorey pasture growth can also depend on the stage of 
phenological development of the plants. Based on a glasshouse experiment using five grass 
species native to Arizona Pinus ponderosa forests, Naumburg et al. (2001 )  showed that a 
reduction in net assimilation rate (NAR) with increasing shade was greater for grasses when 
they were shifting to reproductive growth, compared to at flowering. In addition, shading 
slowed flower development for the naturally late flowering species by bringing forward the 
onset of reproductive growth with a similar or later date of actual flower emergence 
(Naumburg et al. 2001).  
Simply describing PAR and the R:FR relative to open pasture values can be misleading, as it 
is the daily integral of these factors (in absolute values of moles photons m·2) that 
determines plant growth and morphology (Wilson & Ludlow 199 1 ;  Wilson 1 997; McKenzie 
et al. 1 999). For example, in tropical and temperate climates the intensity of PAR in 
summer usually greatly exceeds the photosynthetic capacity of open pasture plants. Thus, 
even under moderate levels of shade the intensity of PAR may still actually exceed, or at 
least be near, the understorey plant's maximum photosynthetic capacity, causing little effect 
on growth and development. However, in winter the intensity of PAR in the open is 
normally much lower than in summer (often less than half the average daily total), which 
means that a similar level of shade in this season would likely have a much more 
pronounced impact on the understorey plants (Wilson 1 997). Furthermore, relative values 
based on single point-in-time measurements disguise important temporal variation in the 
understorey light environment (Anderson 1 964a,b), including sunfleck activity, which can 
have a significant impact on plant growth and development (Chazdon 1 988; Chazdon & 
Pearcy 1 99 1 ;  Washitani & Tang 1 99 1 ;  Wayne & Bazzaz 1993; Healey et al. 1 998; Tang et 
al. 1 999). Studies using shade cloth to vary the intensity of PAR also ignore the potential 
effects of sunflecks (Tumbull & Yates 1 993 ; Lee et al. 1 996). 
Soil fertility, water availability, and temperature also influence the response of pastures to 
changes in the level of light (Alberda 1 965; Eriksen & Whitney 1 98 1 ;  Corre 1 983c; 
Sanderson et al. 1 997; Meziane & Sbipley 1 999). Under soil nutrient, water, or low 
Chapter 3 1 06 
temperature stress, a plant's photosynthetic response curve shows full saturation (or reaches 
its maximum) at a lower level of light compared to when these abiotic factors are non­
limiting for plant growth and development (Alberda 1965; Wilson 1997). As a result, a 
moderate decrease in the level of PAR from open pasture values may have little impact on 
plant growth and development when other abiotic factors are more limiting. This was shown 
in an experiment by Blaclanan & Templeman ( 1938, cited Eriksen & Whitney 1 98 1 ), where 
under low soil nitrogen (N) levels the dry matter production of Agrostis capillaris and 
Festuca rubra was unaffected by up to 39% and 56% shade, respectively. In situations 
where the tree canopy improves the above abiotic factors, it is even possible to increase 
understorey photosynthesis over open conditions in full sunlight (Wong & Wtlson 1980; 
Wild et al. 1 993 ; Wilson 1996, 1 997). 
Chapter 3 1 07 
1 .4 
(a) 
1 .2 
,-.... 
bO '-' 
1 .0 
§ 
is.. .... 4) 0.8 c.. .... 4) -
\ -
� 0.6 \ C \ "0 
- \ 0 0.4 0 � ..s::: 
00 ........ 
� 
0.2 ........ ... 
- -� 
- -.....:-
0.0 
0 20 40 60 80 100 
30 
(b) 
- 25 c 
� 
c.. - ... .... 4) --....... c.. 20 '" ........ 4) ........... ..s::: Cl ............ C «I ........ .... 
J:) 1 5  
"-
'fil .... , � :::: ---
..... "-0 1 0  .... "-4) J:) 
... 
§ 
Z 5 "'�\..". 
11 
0 
0 20 40 60 80 100 
Canopy closure with leaves (%) 
Figure 3.21 Estimated (a) dry matter production and (b) tillerslbranches per plant for (0) perennial 
ryegrass/white clover and (.) cocksfootllotus pastures at the centre of a vertically projected canopy gap 
(Zone 3 )  and for (t.) perennial ryegrass/white clover and ( .. ) cocksfootllotus pastures directly below the 
poplar crowns (Zone 1 ), over a range ofCCLs in summer. Cocksfoot and lotus are considered shade tolerant 
species (Devkota 2000). The botanical species composition for the above relationships was assumed to be a 
mix of 70% grass and 30% legume. The linear and non-linear relationships relating PAR transmission in the 
two main zones of tree influence (i.e. Zones 1 & 3) to stand canopy closure are given in Section 3 .4; whereas, 
data relating pasture production variables to PAR transmission were obtained from Devkota et al. ( 1997). 
For the latter, spectrally neutral shade cloth was used to vary the PAR transmission levels under glasshouse 
conditions; soil nutrients and water were non-limiting (Devkota et al. 1997). 
Chapter 3 1 08 
3.6 Conclusion 
Estimated PAR transmission (%DIFN) was inversely related to all of the stand density 
indices tested. However, there were marked differences in the general form, complexity, and 
precision of the above relationships depending on the location (zone) within the stand, 
season, and stand density index used. When initially going from the open pasture to low 
poplar stand densities %DIFN decreased at a faster rate in Zone 1 than Zone 3. After the 
initial rapid decrease in Zone 1 %DIFN it then became relatively constant, with the level 
depending on the average size of the trees. Under a completely closed canopy, %DIFN in 
both zones of tree influence converged at around 1 5-20% and 50-55% in summer (in-leaf) 
and winter (no leaf), respectively. The relative increase in %DIFN between summer and 
winter was greater at higher stand densities and under larger, more mature, trees. This 
reduced the slope of the inverse relationships associated with Zone 3 %DIFN in winter, and 
also reduced the magnitude of decline in Zone 1 %DIFN. 
In summer, the R:FR decreased from open pasture values at an increasing rate as the 
transmission of PAR (%DIFN) through the fully in-leaf poplar canopy declined. 
The decrease was particularly marked below a %DIFN of 40%. In contrast, without 
spectrally selective leaves, in winter the poplar canopy had little effect on the understorey 
R:FR. 
Across multiple sites, CCLINL, BA, HPCC, and CEV were more strongly related to 
%DIFN than DBH, HPCD, or GCL. In particular, there was a directly proportional inverse 
relationship between CCL and Zone 3 %DIFN in summer. Unaccounted for site differences 
in the relationships for Zone 1 %DIFN were evident to varying degrees for all of the stand 
density indices investigated. One of the main factors not accounted for was likely the 
difference in average tree-crown size between sites. This factor also probably contnbuted to 
the lower quality of fit and precision of the regression models fitted for Zone 3 %DIFN 
when using DBH, HPCD, and GCL as single independent variables. Site variation in the 
average foliage density of the trees may have confounded the summer relationship between 
CEV and Zone 3 %DIFN. 
Chapter 3 1 09 
Further investigation of factors causing the above site differences is clearly required to 
improve the strength and robustness of the developed regression models. The average 
height ratio of the trees is likely an important factor that warrants investigation. 
Increasing the number of samples taken under young/small and old/large trees at high and 
very low stand densities, respectively, would also help to strengthen the regression models 
for both %DIFN and the R:FR ratio. Overall, the results show that the developed regression 
models should not be applied/extrapolated to stands with vastly different crown size or 
form. 
In general, canopy closure (CCL/NL) based on digital images taken with at least a moderate 
angle of view was able to account for variations in silvicultural management (e.g. pruning), 
crown shape, and crown density/porosity more than any of the other stand density indices 
investigated. However, there is a need for greater standardisation of the techniques used for 
capturing and analysing digital canopy images. 
BA, HPCC, and CEV were all strongly related to CCL. Similarly, CCNL and CCL were 
also strongly related. However, for the latter relationship there were marked differences 
between sites with young/pruned and mature/un pruned trees. 
Combining data from the present study with a previously conducted shade trial indicates 
that understorey pasture production would follow a similar pattern to changes in %DIFN. 
In summer, understorey pasture DM production was estimated to cease at 85% CCL, while 
not falling below 50% of open pasture in winter at the same overstorey (stand) density. 
Chapter 3 1 1 0 
3.7 References 
Alberda, T. ( 1 965) The influence of temperature, light intensity and nitrate concentration 
on dry-matter production and chemical composition of Lolium perenne L. Netherlands 
Journal of Agricultural Science 13(4): 335-360. 
Anderson, M.C. ( 1964a) Studies of the woodland light climate. I .  The photographic 
computation of light conditions. Journal of Ecology 52: 27-4 1 .  
Anderson, M.C. ( 1 964b) Studies of the woodland light climate. n.  Seasonal variation in the 
light climate. Journal of Ecology 52: 643-663 . 
Anderson, M. C. ( 1966) Stand structure and light penetration. n. A theoretical analysis. 
Journal of Applied Ecology 3 :  4 1 -54. 
Assmann, E. ( 1 970) The principles offorest yield study: studies in the organic production, 
structure, increment and yield of forest stands (ed Davis, P.W.). Pergamon Press Ltd, 
Sydney, Australia. 
Assmann, S.M. ( 1992) Effects of light quantity and quality during development on the 
morphology and stomatal physiology ofCommelina communis. Oecologia 92: 1 88- 1 95. 
Ban, Y.;  Xu, H.; Bergeron, Y.;  Kneeshaw, D.D. ( 1 998) Gap regeneration of shade­
intolerant Larix gmelini in old-growth boreal forests of northeastern China. Journal of 
Vegetation Science 9: 529-536. 
Bames, RV.; Zak, n.R.; Denton, S.R.; Spurr, S .H. ( 1997) Forest Ecology. Fourth edition. 
John Wiley & Sons Inc. New York, USA. 
Becker, P.; Erhart, n.w.; Smith A.P. (1 989) Analysis of forest light environments Part I. 
Computerized estimation of solar radiation from hemispherical canopy photographs. 
Agricultural and Forest Meteorology 44: 21 7-232. 
Chapter 3 1 1 1  
Bjorkman, O. ( 1981)  Responses to different quantum flux densities. Physiological plant 
ecology. Vol. 1. Responses to the physical environment (eds Lange, O.L.; Noble, P.S. ;  
Osmond, c.B.; Ziegler, H.), pp 57-107. Springer-Verlag, Berlin, Germany. 
Boardman, N.K. ( 1977) Comparative photosynthesis of sun and shade adapted plants. 
Annual Review of Plant Physiology 28: 355-377. 
Brandeis, T.J. ; Newton, M.; Cole, E. (2001 )  A comparison of overstorey density measures 
for describing understorey conifer growth. Forest Ecology and Management 1 52: 149- 157. 
Buckley, D.S.; Isebrands, J.G. ; Sharik, T.L. (1 999) Practical field methods of estimating 
canopy cover, PAR and LAI in Michigan oak: and pine stands. Northern Journal of Applied 
Forestry 16(1) : 25-32. 
Bunnell, F.L.; Vales, D.J. ( 1990) Comparison of methods for estimating forest overstorey 
cover: differences among techniques. Canadian Journal of Forest Research 20: 1 01 - 1 07. 
Caldwell, M.M. (1 987) Plant architecture and resource competition. Ecological Studies 
(eds Schulze, E.D.; Zwolfer, H.), pp 164- 1 79. Springer-Verlag, Berlin, Gennany. 
Canham, C.D. (1 988) An index for understorey light levels in and around canopy gaps. 
Ecology 69(5): 1634- 1 638. 
Canham, C.D.; Coates, K.D. ;  Bartemucci, P. ;  Quaglia, S. (1 999) Measurement and 
modelling of spatially explicit variation in light transmission through interior cedar­
hemlock forests of British Columbia. Canadian Journal of Forest Research 29: 1 775-
1 783. 
Canham, C.D.; Denslow, J.S.; Platt, W.J. ; Runkle, J.R.; Spies, T.A. ; White, P .S .  ( 1 990) 
Light regimes beneath closed canopies and tree-fall gaps in temperate and tropical forests. 
Canadian Journal of Forest Research 20: 620-63 1 .  
Carlson, D.W.; Groot, A. (1997) Microclimate of clear-cut, forest interior, and small 
openings in trembling aspen forest. Agricultural and Forest Meteorology 87: 3 1 3-329. 
Chapter 3 1 1 2 
Carron, L.T. (1 968) An outline offorest mensuration with special reference to Australia. 
Australian National University Press, Canberra, Australia. 
Casal, J.J.; Deregibus, V.A. ; Sanchez, RA. (1 985) Variations in tiller dynamics and 
morphology in Lolium mUltiflorum Lam. vegetative and reproductive plants as affected by 
differences in RedlFar-Red irradiation. Annals of Botany 56: 553-559. 
Casal, J.J.; Sanchez, RA.; Deregibus, V.A. ( 1986) The effect of plant density on tillering: 
the involvement of RlFR ratio and the proportion of radiation intercepted per plant. 
Journal of Experimental Botany 26(4): 365-371 .  
Casal, J.J.; Sanchez, R.A.; Deregibus, V.A. (1987a) Tillering responses of Lolium 
multiflorum plants to changes of Red/Far-Red ratio typical of sparse canopies. Journal of 
Experimental Botany 38( 194): 1 432- 1439. 
Casal, J.J.; Sanchez, RA.;  Deregibus, V.A. ( 1987b) The effect of light quality on shoot 
extension growth in three species of grasses. Annals of Botany 59: 1 -7. 
Chazdon, RL. ( 1 988) Sunflecks and their importance to forest understorey plants. 
Advances in Ecological Research 1 8: 1 -63 . 
Chazdon, RL. ; Pearcy, RW. ( 1991)  The importance of sunflecks for forest understory 
plants. BioScience 41 ( 1 1) :  760-766. 
Chen, J.M.; Black, T.A.; Adams. RS. ( 1991)  Evaluation of hemispherical photography for 
determining plant area index and geometry of a forest stand. Agricultural and Forest 
Meteorology 56: 1 29-143.  
Chen, J.M.; Blanken, P.D.; Black, T.A.; Guilbeault, M. ;  Chen, S .  ( 1997) Radiation regime 
and canopy architecture in a boreal aspen forest. Agricultural and Forest Meteorology 86: 
1 07-125. 
Chapter 3 1 1 3 
Comeau, P.G.; Gendron, F.; Letchford, T. (1 998) A comparison of several methods for 
estimating light under a paper birch mixedwood stand. Canadian Journal of Forest 
Research 28:  1 843-1850. 
Coombe, D.E. ( 1 957) The spectral composition of shade light in woodlands. Journal of 
Ecology 45: 823-830. 
Corre, W.I ( 1 983a) Growth and morphogenesis of sun and shade plants 1. The influence of 
light intensity. Acta Botanica Neerlandica 32(1/2): 49-62. 
Corre, W.J. ( 1983b) Growth and morphogenesis of sun and shade plants IT. The influence 
of light quality. Acta Botanica Neerlandica 32(3): 1 85-202. 
Corre, W.I (1 983c) Growth and morphogenesis of sun and shade plants III. The combined 
effects of light intensity and nutrient supply. Acta Botanica Neerlandica 32(4): 277-294. 
Deregibus, V.A.; Sanchez, RA.; Casal, J.J. ; Trlica, M.I ( 1985) Tillering responses to 
enrichment of red light beneath the canopy in a humid natural grassland. Journal of 
Applied Ecology 22: 199-206. 
Devkota, N.R (2000) Growth of pasture species in the shade in relation to alder silvo­
pastoral systems. PhD thesis, Massey University, New Zealand. 
Devkota, N.R;  Kemp, P.D.; Hodgson, J. (1 997) Screening pasture species for shade 
tolerance. Proceedings Agronomy Society of New Zealand 27: 1 19- 1 28. 
Devkota, N.R.; Kemp, P.D.; Valentine, 1.; Hodgson, I (1 998) Performance of perennial 
ryegrass and cocksfoot cultivars under tree shade. Proceedings Agronomy Society of New 
Zealand 28: 1 29-1 3  5. 
Devkota, N.R; Wall, A.J.; Kemp, P.D.; Hodgson, J. (2001)  Relationship between canopy 
closure and pasture production in deciduous tree based temperate silvopastoral systems. 
Proceedings of the XIX International Grassland Congress: 652-653 .  
Chapter 3 1 14 
Douglas, G.B. ;  Walcroft, AS.;  Wills, B.J.; Hurst, S.E.; Foote, AG.; Trainor, K.D. ;  Fung, 
L.E. (2001 )  Resident pasture growth and the micro-environment beneath young, widely­
spaced poplars in New Zealand. Proceedings of the New Zealand Grassland Association 
63 : 1 3 1 - 1 38.  
Ellis, le.; Hayes, lD. ( 1 997) Field guide for sample plots in New Zealand forests. FRI 
Bulletin No. 1 86 (eds Gadgill, R; Griffith, J.). New Zealand Forest Research Institute Ltd, 
Rotorua, New Zealand. 
Endler, lA. ( 1993) The color of light ID forests and its implications. Ecological 
Monographs 63( 1 ): 1 -27. 
Englund, S.R; O'Brien, ll; Clark, D.B. (2000) Evaluation of digital and film 
hemispherical photography and spherical densiometry for measuring forest light 
environments. Canadian Journal of Forest Research 30: 1999-2005. 
Eriksen, F.I.; Whitney, A.S.  ( 198 1) Effects of light intensity on growth of some tropical 
forage species. I. Interaction of light intensity and nitrogen fertilization on six forage 
grasses. Agronomy Journal 73 : 427-433 .  
Federer, e.A; Tanner, C.B. ( 1966) Spectral distribution of  light in the forest. Ecology 
47(4): 555-560. 
Frazer, G.W.; Fournier, RA; Trofymow, J.A; Hall, R.J. (2001)  A comparison of digital 
and :film fisheye photography for analysis of forest canopy structure and gap light 
transmission. Agricultural and Forest Meteorology 1 09: 249-263 . 
Gautier, H.; Varlet-Grancher, C.; Hazard, L. (1 999) Tillering responses to the light 
environment and to defoliation in populations of perennial ryegrass (Lolium perenne L.) 
selected for contrasting leaf length. Annals of Botany 83 : 423-429. 
Gay, L.W.; Knoerr, K.R.; Braaten, M.O. ( 1 971)  Solar radiation variability on the floor ofa 
pine plantation. Agricultural Meteorology 8 :  39-50. 
Chapter 3 1 1 5 
Gendron, P. ; Messier, c.;  Comeau, P.G. ( 1 998) Comparison of various methods for 
estimating the mean growing season percent photosynthetic photon flux density in forests. 
Agricultural and Forest Meteorology 92: 55-70. 
Grace, J.C.; Jarvis, P.G.; Norman, lM. ( 1 987) Modelling the interception of solar radiant 
energy in intensively managed stands. New Zealand Journal of Forestry Science 1 7(2/3): 
193-209 
Guevara-Escobar, A; Kemp, P.D.; Hodgson, J. ; Mackay, A.D.;  Edwards, W.R.N. ( 1 997) 
Case study of a mature Populus deltoMes-pasture system in a hill environment. 
Proceedings of the New Zealand Grassland Association 59: 1 79- 1 85 .  
Guevara-Escobar, A ( 1 999) Aspects of a poplar-pasture system related to pasture 
production in New Zealand. PhD thesis, Massey University, New Zealand. 
Hanan, N.P.; Begue, A ( 1 995) A method to estimate instantaneous and daily intercepted 
photosynthetically active radiation using a hemispherical sensor. Agricultural and Forest 
Meteorology 74: 1 55-168 .  
Hassika, P.;  Berbigier, P .  ( 1998) Annual cycle of  photosynthetically active radiation in 
maritime pine forest. Agricultural and Forest Meteorology 90: 1 57-17 l .  
Healey, K.D.;  Rickert, K.G.; Hammer, G.L.;  Bange, M.P. (1 998) Radiation use efficiency 
increases when the diffuse component of incident radiation is enhanced under shade. 
Australian Journal of Agricultural Research 49: 665-672. 
Heraut-Bron, V.; Robin, c.; Varlet-Grancher, C.; Afif, D.; Guckert, A. ( 1 999) Light 
quality (red:far-red ratio): does it affect photosynthetic activity, net CO2 assimilation, and 
morphology of young white clover leaves? Canadian Journal of Botany 77: 1425- 143 l .  
Hughes, J.E.; Morgan, D.C.; Black, C.R. ( 1 985) Transmission properties of an oak canopy 
in relation to photoperception. Plant, Cell and Environment 8: 509-5 16. 
Chapter 3 1 1 6 
Hutchison, B.A; Matt, D.R ( 1976) Beam enrichment of diffuse radiation in a deciduous 
forest. Agricultural Meteorology 17 :  93-1 1 0. 
Hutchison, B.A; Matt, D.R ( 1977a) The annual cycle of solar radiation in a deciduous 
forest. Agricultural Meteorology 1 8 :  255-265. 
Hutchison, B.A; Matt, D.R ( 1 977b) The distribution of solar radiation within a deciduous 
forest. Ecological Monographs 47: 1 85-207. 
Jabne, B. (1 997) Practical handbook on image processing for scientific applications. CRC 
Press, New York, USA 
Karlik, J.F.; Winer, AM. ( 1 999) Comparison of calculated and measured leaf masses of 
urban trees. Ecological Applications 9(4): 1 1 68- 1 1 76. 
Kephart, K.D.; Buxton, D.R; Taylor, S.E. (1 992) Growth of C3 and C4 perennial grasses 
under reduced irradiance. Crop Science 32: 1033-1038. 
Kleinbaum, D.G.; Kupper, L.L.; Muller, K.E.; Nizam, A. ( 1 998) Applied regression 
analysis and other multivariable methods. Third edition. Duxbury Press. Brooks/Cole 
Publishing Company, California, USA. 
Knowles, RL.; Horvath, G.C.; Carter, M.A.; Hawke, M.F. ( 1 999) Developing a canopy 
closure model to predict overstorey/understorey relationships in Pin us radiata 
silvopastoral systems. Agroforestry Systems 43( 1 ): 1 09- 1 1 9  
Kuuluvainen, T.; Pukkala, T .  ( 1987) Effect of  crown shape and tree distribution on the 
spatial distribution of shade. Agricultural and Forest Meteorology 40: 21 5-23 1 .  
Larcher, W. ( 1980) PhYSiological plant ecology. Second edition. Springer Verlag, Berlin, 
Gennany. 
Chapter 3 1 1 7 
Lee, D.W.; Baskaran, K.; Mansor, M.; Mohamad, H.; Yap, S .K. ( 1 996) Irradiance and 
spectral quality affect Asian tropical rain forest tree seedling development. Ecology 77(2): 
568-580. 
Lieffers, V.I.; Messier, c.; Stadt, K.I.; Gendron, F.; Comeau, P.G. ( 1999) Predicting and 
managing light in the understorey of boreal forests. Canadian Journal of Forest Research 
29: 796-8 1 1 . 
Littell, RC.; Freund, RJ. ; Spector, P.C. ( 1991 )  SA� System for linear models. Third 
edition. SAS Institute Inc., North Carolina, USA. 
Ludlow, M.M.; Wilson, G.L.; Heslehurst, M.R ( 1 974) Studies on the productivity of 
tropical pasture plants. V. Effect of shading on growth, photosynthesis and respiration in 
two grasses and two legumes. Australian Journal of Agricultural Research 25 : 425-433.  
Machado, J. ; Reich, P.B. ( 1 999) Evaluation of several measures of canopy openness as 
predictors of photosynthetic photon flux density in deeply shaded conifer-dominated forest 
understorey. Canadian Journal of Forest Research 29: 1 438-1444. 
McElwee, H.F. ; Knowles, RL. (2000) Estimating canopy closure and understorey pasture 
production in New Zealand-grown poplar plantations. New Zealand Journal of Forestry 
Science 30(3): 422-435.  
McKennan, G.T. (1 995) The use of digital images for data collection and analysis of light 
attenuation by tree canopies. Arboricultural Journal 19 :  267-274. 
McKenzie, B.A. ; Kemp, P.D.; Moot, D.l; Matthew, C.; Lucas, RI. ( 1999) Environmental 
effects on plant growth and development. New Zealand Pasture and Crop Science 
(eds White, J. ; Hodgson, I.), pp 29-44. Oxford University Press, Auckland, New Zealand. 
Messier, c.;  Bellefleur, P.  ( 1988) Light quantity and quality on the forest floor of pioneer 
and climax stages in a birch-beech-sugar maple stand. Canadian Journal of Forest 
Research 1 8: 6 1 5-622. 
Chapter 3 1 1 8 
Messier, C.; Honer, T.W. ;  Kimmins, lP. ( 1 989) Photosynthetic photon flux density, red: 
far-red ratio, and minimum light requirement for survival of Gaultheria shallon in western 
red cedar-western hemlock stands in coastal British Columbia. Canadian Journal of 
Forest Research 19: 1470- 1477. 
Messier, C.; Parent, S.; Bergeron, Y. (1 998) Effects of overstory and understory vegetation 
on the understory light environment in mixed boreal forests. Journal of Vegetation Science 
9:  5 1 1 -520. 
Messier, c.; Puttonen, P. ( 1 995) Spatial and temporal variation in the light environment of 
developing Scots pine stands: the basis for a quick and efficient method of characterizing 
light. Canadian Journal of Forest Research 25: 343-354. 
Meziane, D.; Shipley, B. ( 1 999) Interacting determinants of specific leaf area in 22 
herbaceous species: effects of irradiance and nutrient availability. Plant, Cell and 
Environment 22: 447-459. 
Mitchell, J.E.; Popovich, S.l (1997) Effectiveness of basal area for estimating canopy 
cover ofponderosa pine. Forest Ecology and Management 95 : 45-5 1 .  
Morgan, D.C.; Smith H. ( 198 1 )  Non-photosynthetic , responses to light quality. 
Encyclopaedia of Plant Physiology 12A: 1 09- 134. 
Morgan, D.C.; Warrington I.J.; Rook, D.A. (1 985) Some observations on the spectral 
distribution characteristics of short-wave radiation within Pin us radiata canopies. Plant, 
Cell and Environment 8(3): 20 1 -206. 
Mourelle, C.; Kellman, M.; Kwon, L .  (2001 )  Light occlusion at forest edges: an analysis of 
tree architectural characteristics. Forest Ecology and Management 1 54: 1 79- 192 .  
Myers, R.H. ( 1990) Classical and modern regression with applications. Second edition. 
PWS-Kent Publishing Company, Boston, USA. 
Chapter 3 1 1 9 
Naumburg, E. ; DeWald, L.E.; Kolb, T.E. (2001 )  Shade responses of five grasses native to 
southwestern V.S. Pinus ponderosa forests. Canadian Journal of Botany 79: 1 001 - 1 009. 
Nelder, J.A. (1 962) New kinds of systematic designs for spacing experiments. Biometrics 
1 8: 283-307. 
Ong, C.K. ( 1 996) A framework for quantifying the various effects oftree-crop interactions. 
Tree-crop interactions - a physiological approach (eds Ong, C.K.; Huxley, P.), pp 1 -23. 
ICRAF and CAB International. University Press, Cambridge UK. 
Ong, C.K.; Black, c.R.; Marshall, F. M.; Corlett, lE. ( 1 996) Principles of resource capture 
and utilisation of light and water. Tree-crop interactions - a physiological approach (eds 
Ong, C.K.; Huxley, P.), pp 1 30-1 50. ICRAF and CAB International. University Press, 
Cambridge UK. 
Parent, S. ;  Messier, C. ( 1 996) A simple and efficient method to estimate micro site light 
availability under a forest canopy. Canadian Journal of Forest Research 26: 1 5 1 - 1 54. 
Percival, N.S.; Bond, DJ. ; Hawke, M.F. ; Cranshaw, L.l Andrew, B.L.; Knowles, R.L. 
( 1984). Effects of radiata pine on pasture yields, botanical composition, weed populations, 
and production of a range of grasses. Proceedings of a technical workshop on agroforestry, 
pp 13-22. Ministry of Agriculture and Fisheries, Wellington, New Zealand. 
Percival, N.S.; Knowles, R.L. (1 988) Relationship between radiata pine and understorey 
pasture production. Agroforestry symposium proceedings (ed Madaren, lP.), pp 1 52- 163 .  
Forest Research Institute and New Zealand Ministry of  Forestry, Rotorua, New Zealand. 
Perry, T.O.; Sellers, H.E. ; Blanchard, C.O. ( 1969) Estimation of photosynthetically active 
radiation under a forest canopy with chlorophyll extracts and from basal area 
measurements. Ecology 50(1):  39-44. 
Phillip, M.S. (1 994) Measuring trees and forests. Second edition. CAB International. 
University Press, Cambridge, UK. 
Chapter 3 120 
Pons, T.L. ( 1983) Significance of inhibition of seed germination under the leaf canopy in 
ash coppice. Plant, Cell, and Environment 6: 385-392. 
Poulson, T.L.; Platt, W.J. ( 1 989) Gap light regimes influence canopy tree diversity. 
Ecology 70(3) :  553-555. 
Pyke, D.A.; Zamora, B .A. ( 198 1 )  Relationships between overstorey structure and 
understorey production in the Grand FirlMyrtle Boxwood habitat type of Northcentral 
Idaho. Journal of Range Management 35(6): 769-773. 
Reed, D.D.;  Mroz, G.D. ( 1997) Resource assessment inforested landscapes. John Wlley & 
Sons, Inc., USA. 
Reich, P.B . ;  Tjoelker, M.G.; WaIters, M.B.; Vanderklein, D.W.; Buschena, C. (1998) 
Close association of RGR, leaf and root morphology, seed mass and shade tolerance in 
seedlings of nine boreal tree species grown in high and low light. Functional Ecology 12 :  
327-338.  
Reifsnyder, W.E.;  Furnival, G.M.; Horowitz, lL. ( 1 971/1972) Spatial and temporal 
distribution of solar radiation beneath forest canopies. Agricultural Meteorology 9: 2 1 -37. 
Reitmayer, H.; Wemer, H.; Fabian, P .  (2002) A novel system for spectral analysis o f  solar 
radiation within a mixed beech-spruce stand. Plant Biology 4: 228-233 . 
Rich, P.M.; Clark, D.B.; Clark, D.A.; Oberbauer, S.F. ( 1993) Long-term study of solar 
radiation regimes in a tropical wet forest using quantum sensors and hemispherical 
photography. Agricultural and Forest Meteorology 65: 1 07- 127. 
Ritchie, G.A. ( 1997) Evidence for red:far red signalling and photomorphogenic growth 
response in Douglas-fir (Pseudotsuga menziesii) seedlings. Tree PhYSiology 17 :  1 6 1 - 1 68. 
Ross, M.S. ;  Flanagan, L.B.; La Roi, G.H. (1986) Seasonal and successional changes in 
light quality and quantity in the understorey of boreal forest ecosystems. Canadian Journal 
of Botany 64: 2792-2799. 
Chapter 3 1 21 
Roxburgh, J.R. ; Kelly, D. ( 1 995) Uses and limitations of hemispherical photography for 
estimating forest light environments. New Zealand Journal of Ecology 1 9(2): 2 1 3 -2 1 7. 
Runkle, IR. ( 1 982) Patterns of disturbance in some old-growth mesic forests of Eastern 
North America. Ecology 63(5): 1 533- 1546. 
Runkle, IR. ; Stewart, G.H.; Veblen, T.T. ( 1 995) Sapling diameter growth in gaps for two 
Nothofagus species in New Zealand. Ecology 76(7): 2 107-21 1 7. 
Samarakoon, S .P; Wilso� lR.; She1to� H.M. (1 990) Growth, morphology and nutritive 
quality of shaded Stenotaphrum secundatum, Axonopus compressus and Pennisetum 
clandestinum. Journal of Agricultural Science, Cambridge 1 14: 1 6 1 - 1 69. 
Sanderson, M.A.; Stair, D.W.; Hussey, M.A. (1 997) Physiological and morphological 
responses of perennial forages to stress. Advances in Agronomy 59: 1 72-224. 
Scanlan, J.C. ( 1 991)  Woody overstorey and herbaceous understorey biomass in Acacia 
harpophyl/a (brigalow) woodlands. Australian Journal of Ecology 16: 521 -529. 
Schmitt, J. ; Wulff, R.D. ( 1 993) Light spectral quality, photochrome and plant competition. 
Trends in Ecology and Evolution 2: 47-5 1 .  
Schreuder, H.T.; Gregoire, T.G.; Wood, G.B. ( 1993) Sampling methods for multiresource 
forest inventory. John Wiley & Sons, Inc. ,  USA. 
Seo, S.; Han, y.e. ; Park, M.S . ( 1 989) Grassland improvement and N fertilization 
management of woodland in Korea. Proceedings of the XVI International Grassland 
Congress: 1487-1 488. 
Sibbald, A.R.; Griffiths, lH. ( 1992) The effects of conifer canopies at wide spacing on 
some under-storey meteorological parameters. Agroforestry Forum 3 :  8- 15. 
Chapter 3 1 22 
Sibbald, A.R.; Griffiths, IH.; Elston, D.A. ( 1 994) Herbage yield in agroforestry systems as 
a function of easily measured attributes of the tree canopy. Forest Ecology and 
Management 65: 1 95-200. 
Sinclair, F. L. ;  Eason, W.R.; Hooker, J.E. (2000). Understanding and management of 
interactions. Agroforestry in the UK Bulletin 1 22, (eds Hislop, M.; Claridge, J.), pp 1 7-28. 
Forestry Commission, Edinburgh, UK. 
SkaIova, H.; Krahulec, F. ( 1992) The response of three Festuca rubra clones to changes in 
light quality and plant density. Functional Ecology 6:  282-290. 
Smith, H. ( 1 982) Light quality, photoperception, and plant strategy. Annual Review of 
Plant Physiology 33:  48 1 -5 1 8. 
Solangaarachchi, S .M.; Harper, IL. ( 1 987) The effect of canopy filtered light on the 
growth of white clover Trifolium repens. Oecologia 72: 372-376. 
Stadt, KI; Landhausser, S.M.; Stewart, J.D. ( 1 997) Comment - The effects of direct-beam 
light on overcast-day estimates of light availability. Canadian Journal of Forest Research 
27: 272-274. 
Stadt, KI; Lieffers, V.I (2000) MIXLIGHT: a flexible light transmission model for 
mixed-species forest stands. Agricultural and Forest Meteorology 1 02: 235-252. 
Stuefer, IF.; Huber, H. (1 998) Differential effects of light quantity and spectral light 
quality on growth, morphology and development of two stoloniferous Potentilla species. 
Oecologia 1 1 7:  1 -8. 
Sturman, A.P.; Tapper, N.J. ( 1996) The weather and climate of Australia and New Zealand 
(ed Gray, K). Oxford University Press, Australia. 
Takenaka, A. ( 1 988) An analysis of solar beam penetration through circular gaps in 
canopies of uniform thickness. Agricultural and Forest Meteorology 42: 307-320. 
Chapter 3 1 23 
Tang, Y.; Kachi, N.; Furukawa, A. ; Awang, M.B. ( 1999) Heterogeneity of light 
availability and its effects on simulated carbon gain of tree leaves in a small gap and the 
understorey in a tropical rain forest. Biotropica 3 1(2): 268-278. 
Tasker, R: Smith, H. (1 977) The function of phytochrome in the natural environment - V. 
Seasonal changes in radiant energy quality in woodlands. Photochemistry and 
Photobiology 26: 487-491 .  
Taylor, T.H.; Cooper, J.P.;  Treharne, K.J. ( 1968) Growth response of orchardgrass 
(Dactylis glomerata L.) to different light and temperature environments. I .  Leaf 
development and senescence. Crop Science 8 :  437-440. 
Teuber, N.; Laidlaw, A.S .  ( 1 996) Influence of irradiance on branch growth of white clover 
stolons in rejected areas within grazed swards. Grass and Forage Science 5 1 :  73-80. 
Thompson, L. ( 1 993) The influence of the radiation environment around the node on 
morphogenesis and growth of white clover (Trifolium repens). Grass and Forage Science 
48: 271-278. 
Thompson, RC.; Luckman, P.G. (1 993) Performance of biological erosion control in 
New Zealand soft rock hill terrain. Agroforestry Systems 21 (2): 19 1 -2 1 1 .  
Turnbull, M.H.;  Yates, D.l ( 1 993) Seasonal variation in the red/far-red ratio and photon 
flux density in an Australian sub-tropical rainforest. Agricultural and Forest Meteorology 
64: 1 1 1 - 1 27. 
Vales, D.l; Bunnell, F.L. (1 988) Relationships between transmission of solar radiation and 
coniferous forest stand characteristics. Agricultural and Forest Meteorology 43 : 201 -223 . 
Van Kraayenoord, C.W.S.;  Hathaway, RL. ( 1986) Plant materials handbook for soil 
conservation. Volume 1:  Principles and practices. Water and soil Miscellaneous 
Publication No. 93. National Water and Soil Conservation Authority, Wellington, New 
Zealand. 
Chapter 3 1 24 
Vezina, P.E.; Boulter, D.W.K. ( 1 966) The spectral composition of near ultraviolet and 
visible radiation beneath forest canopies. Canadian Journal of Botany 44: 1267-1284. 
Wagner, S. ( 1998) Calibration of grey values of hemispherical photographs for image 
analysis. Agricultural and Forest Meteorology 90: 1 03-1 1 7. 
Wall, AJ.; Mackay, A.D.;  Kemp, P.D.; Gillingham, AG.; Edwards, W.R.N. ( 1 997) The 
impact of widely spaced soil conservation trees on hill pastoral systems. Proceedings of the 
New Zealand Grassland Association 59: 1 7 1 - 1 77. 
Wang, H.; Baldocchi, D.D. ( 1 989) A numerical model for simulating the radiation regime 
within a deciduous forest canopy. Agricultural and Forest Meteorology 46: 3 1 3-337. 
Wang, J.R.; Zhong, AL.; Comeau, P.; Tsze, M.; Kimmins, IP. ( 1995) Aboveground 
biomass and nutrient accumulation in an age sequence of aspen (Populus tremuloides) 
stands in the Boreal White and Black Spruce Zone, British Columbia. Forest Ecology and 
Management 78: 1 27-1 3 8. 
Washitani, 1 . ;  Tang, Y. ( 199 1 )  Microsite variation in light availability and seedling growth 
of Quercus serrata in a temperate pine forest. Ecological Research 6: 305-3 1 6. 
Wayne, P.M.; Bazzaz, F.A (1 993) Morning vs. afternoon sun patches in experimental 
forest gaps: consequences of temporal incongruency of resources to birch regeneration. 
Oecologia 94: 235-243 . 
Welles, IM.; Norman, J.M. ( 1991) Instrument for indirect measurement of canopy 
architecture. Agronomy Journal 83:  8 1 8-825. 
Wild, D.W.M. ; Wilson, IR. ; Stili, W.W.; Shelton, H.M. ( 1993) Shading increases yield of 
nitrogen-limited tropical grasses. Proceedings of the XVII International Grassland 
Congress: 2060-2062. 
Wilkinson, A.G. (1999) Poplars and willows for soil erosion control in New Zealand. 
Biomass and Bioenergy 1 6: 263-274. 
Wilkinson, D.M. ( 1995) Modelling tree crowns as geometric solids. Arboricultural Journal 
1 9: 387-393. 
Chapter 3 1 25 
Wilson, J.R. ( 1 996) Shade-stimulated growth and nitrogen uptake by pasture grasses in a 
subtropical environment. Australian Journal of Agricultural Research 47: 1 075-1093. 
Wilson, lR. ( 1997) Adaptive responses of grasses to shade: relevance to turfgrasses for 
low light environments. International Turfgrass Society Research Journal 8 :  575-59 1 .  
Wilson, lR.; Ludlow, M.M. ( 199 1 )  The environment and potential growth of herbage 
under plantations. Forages for plantation crops. Proceedings of a workshop, Sanur Beach, 
Bali, Indonesia, 27-29 June 1990 (eds Shelton, H.M. ; Stili, W.W.), ppl0-24. Australian 
Centre for International Agricultural Research, Canberra, Australia. 
Wojtkowski, P. (1 998) The theory and practice of agroforestry design. A comprehensive 
study of the theories, concepts and conventions that underlie the successful use of 
agroforestry. Science Publishers, Inc., Enfield, USA 
Wong, C.C.; Rahim, H.; Mohd. Sharudin, M.A ( 1985) Shade tolerance potential of some 
tropical forages for integration with plantations. 1 .  Grasses. MARDI Research Bulletin 1 3 :  
225-247. 
Wong, C.C. ; Stili, W.W. ( 1 996) Persistence of tropical forage grasses ill shaded 
environments. Journal of Agricultural Science, Cambridge 126: 1 5 1 - 1 59.  
Wong, C.c. ; Wilson, J.R. ( 1 980) Effects of shading on the growth and nitrogen content of 
Green Panic and Siratro in pure and mixed swards defoliated at two frequencies. 
Australian Journal of Agricultural Research 3 1 :  269-285 .  
W"iinsche, J.N.; Lakso, AN.; Robinson, T.L. (1 995) Comparison of four methods for 
estimating total light interception by apple trees of varying forms. HortScience 30(2) :  272-
276. 
Yates, D.; McKennan, G. ( 1 989) Solar architecture and light attenuation by trees: conflict 
or compromise? Arboricultural Journal 1 3 :  7- 16. 

Chapter 4 1 26 
4 The effect of poplar overstorey density on soil 
chemical properties 
CONTENTS 
4.1 Introduction 
4.2 Methodology 
4.2. 1 
4.2.2 
4.2.3 
General site characteristics and treatment structure 
Field sampling 
Chemical and biochemical analysis techniques 
4.3 Data analyses 
4.4 Results 
4.4. 1 
4.4.2 
4.4.3 
4.4.4 
4.5 
4.5 . 1  
4.5.2 
4.5.3 
Effect of overstorey environment on soil properties within two separate 
soil strata at Kiwitea 
Effect of poplar overstorey density on soil properties within two 
separate soil strata at Kiwitea 
Effect of overstorey environment on soil properties within two separate 
soil strata at Hautope 1 
Effect of poplar overstorey density on soil properties within two 
separate soil strata at Hautope 1 
Discussion 
Soil pH 
The effect of overstorey environment on major plant-available 
cations 
The effect of overstorey environment on major plant-available 
127 
128 
1 28 
1 29 
1 30 
131 
132 
132  
133  
138  
139 
144 
144 
1 53 
anions 1 57  
4.5.4 Differences in soil organic carbon (SOC) between the overstorey 
environments 1 59 
4.5.5 The ecological significance of the soil nutrient status at each site 
in relation to pasture production 1 6 1  
4.6 Conclusion 163 
4.7 References 165 
Chapter 4 1 27 
4.1 Introduction 
Chapter 3 showed how a decrease in PAR transmission directly below poplar trees and 
with increasing stand density was likely to have a negative effect on understorey pasture 
production. Also briefly discussed in Section 3.5 .8  was that other limiting factors, such as 
soil moisture and nutrient stress, can have an impact on a pasture plant's response to 
changes in PAR. Trees can intercept and reduce significant amounts of effective rainfall 
reaching the soil surface (Guevara-Escobar 1 999; Guevara-Escobar et al. 2000; Douglas et 
al. 200 1 ;  McIvor et al. 2003). Nevertheless, several recent studies have shown that the soil 
moisture content in upper soil layers does not vary widely between tree and open 
pasture/grassland environments (Wilson & Kleb 1 996; Guevara-Escobar et al. 1 997; 
Guevara-Escobar 1 999; Power et al. 1 999, 2003 ; Herman et al. 2003; James et al. 2003; 
McIvor et al. 2003). In contrast, Guevara-Escobar et al. (2002) found the pH and levels of 
exchangeable base cations (Ca, Mg, & K) in the topsoil were higher directly under poplar 
trees than in adjacent open pasture and concluded that the soil from the former 
environment was superior in supporting pasture production, when no other factors were 
limiting. 
However, Guevara-Escobar et al. (2002) study was limited to a summer-moist climate 
(annual rainfall 1 l 00mm-1300mm), covered a very small range of stand densities, and 
spatially, comparisons were restricted to between directly below trees (analogous to Zone 
1 )  and in adjacent open pasture. Trees affect soils mainly through their litter, root activity, 
and changes to the micro climate (Van Goor 1 985). As such, variation in the regional 
climate, stand density, and proximity to trees in a stand will all likely strongly influence 
the impact of trees on soil properties (Fuller & Anderson 1 993; Nwaigbo et al. 1 997; 
Perrott et al. 1 999). Therefore, the objective of this chapter was to determine the spatial 
variation in the effect of poplar trees on soil properties, over a range of stand densities, and 
to evaluate whether any changes will likely affect understorey pasture production. To 
increase the scope of the study, measurements were taken from two climatically 
contrasting regions (summer-wet & -dry) where poplar trees are often planted for soil 
conservation work. 
Chapter 4 
4.2 Methodology 
4.2.1 General site characteristics and treatment structure 
1 28 
The broad range of overstorey (stand) densities previously developed in Chapter 2 for 
Kiwitea and Hautope 1 formed the main treatment structure of this study. These two sites, 
as opposed to Hautope 2, were selected because their larger/more fully developed trees 
would likely have had a greater impact on the nutrient balance of the mixed tree-pasture 
system, and also their poplar stands contained a wider range of densities. Both of these 
factors should help to clarify any trends in soil properties related to changes in stand 
density (Alban 1 982; Myers 1 990; Guevara-Escobar 1 999). The Kiwitea and Hautope 1 
sites represent summer-wet and summer-dry climates, respectively (refer to section 
2.3 . 1 .3). 
The poplars at each farm site had mainly been planted to control or prevent mass 
movement and fluvial soil erosion (refer to Section 2.3). As a consequence, tree spacing 
was not constant, instead being closer in areas with severe soil erosion and more widely 
spaced on relatively stable parts of the hill slopes. This typical planting pattern used for 
soil conservation trees (Wall et al. 1 997; Wilkinson 1 999) caused higher stand density 
experimental units to be situated on areas that showed greater visual signs of previous soil 
disturbance (refer to Section 2.3). 
Grasses dominated the pastures at both sites (refer to Section 5.4.4). At Kiwitea, grasses 
comprised over 50% of the total sward biomass, while at Hautope 1 this value exceeded 
60% and was often around 80%. Directly below the poplar crowns (Zone 1 )  at Kiwitea the 
proportion of high fertility responsive grasses (HFG') tended to be greater than in either 
the vertically projected canopy gap (Zone 3) or open pasture. Out of the different HFG, 
Lolium perenne and Poa spp. dominated the open pasture and Zone 1 ,  respectively, 
whereas, these two species were found in similar abundance in Zone 3.  All three main 
overstorey environments had a legume content around 8% of the total sward biomass 
(refer to Section 5.4.4. 1 ). At Hautope 1 ,  both Zones 1 and 3 contained a greater proportion 
HFG than in the adjacent open pasture. Lolium perenne and Poa spp. were the main high 
fertility responsive species in Zones 1 and 3 ,  whereas, Lolium perenne and Holcus lanatus 
I HFG included: Lolium perenne, Poa spp., Holcus lanatus, and Dactylis glomerata. 
Chapter 4 1 29 
dominated the open pasture. Less than 5% legwne was present in Zones 1 and 3 .  In 
comparison, the open pasture had around 1 0% legwne in spring, but this decreased 
markedly to below 5% in late summer (refer to Section 5.4.4.3). 
4.2.2 Field sampling 
Soil samples were taken at Hautope 1 and Kiwitea on 5 September and 17 October 2000, 
respectively. A 25 mm-diameter soil corer was used to collect samples from 0-75 mm and 
75- 150 mm soil depths. The upper 0-75 mm soil stratum was sampled for all of the 
experimental units (micro sites) that were selected in Chapter 2, except for the paired units 
located on a south facing aspect at Kiwitea. These two experimental units were excluded to 
limit comparisons at this site to a single northerly aspect (refer to Section 2.3,  Table 2.2). 
The lower 75- 1 50 mm stratum was sampled from a more limited range of experimental 
units at each of the sites. This range consisted of an open pasture controL and experimental 
units under low, medium, and high overstorey densities. Experimental units were stratified 
into overstorey density classes based on the proportion of diffuse non-intercepted radiation 
(%DIFN) received at the centre of their vertically projected canopy gap (Zone 3). Within 
each overstorey density class, two experimental units were randomly selected for sampling 
(Appendix 4. 1) .  
Amongst the stands of poplar the two main overstorey environments (or zones of tree 
influence) previously described in Section 3.2.3.3 were sampled separately. From within 
these overstorey environments, and also within the open pasture, 1 0  soil cores were 
randomly collected and bulked together for each soil stratum. Sampling near patches with 
visual signs of dung or urine spots was avoided (Cornforth 1 980) and the soil cores were 
only taken from a mediwn hill-slope ( 13-25 degrees) category (Saggar et al. 1 990; Lopez 
2000). 
The bulked soil samples were analysed for: pHw(H20); basic plant-available cations -
calciwn (Ca), magnesiwn (Mg), potassiwn (K), and sodium (Na); plant-available anions -
phosphate (P) and sulfate-sulfur (S04-S); anion storage capacity (ASC); soil organic 
carbon (SOC) and related soil organic matter (SOM). 
Chapter 4 1 30 
4.2.3 Chemical and biochemical analysis techniques 
Prior to analysis, the bulked soil samples were air dried at 33-35° for 1 8  hours, crushed, 
and then ground to pass through 2 mm-hole stainless steel sieves (Cornforth 1 980). 
To measure soil pH, 1 2  ml of a soil sample was suspended in 25 ml of distilled water, 
vigorously stirred, and then left to equilibrate overnight (1 6 hours) at 20°C. After briefly 
re-stirring the 1 :2 . 1  v/v water slurry, the soil pH was read using a combined glass-calomel 
electrode and direct reading digital pH meter (Cornforth 1 980; Blakemore et at. 1987). 
AgResearch Quick-test2 procedures were used to measure plant-available K, Ca., Mg, and 
Na, as outlined by Cornforth ( 1 980) and Lee et al. (1991).  The method involved shaking 
4.4 m1 of soil with a 20 m1 solution of IM ammonium acetate (NH40Ac) at pH 7 for 2 
minutes; after which, from the filtered extract, the displaced K, Ca, and Na were measured 
by flame emission spectrophotometry, while Mg was measured by atomic absorption 
spectrophotometry. 
Plant-available P was determined based on the method of Olsen et at. (1954). P was 
extracted from 4 ml of soil with an 80 ml solution of 0.5M sodium bicarbonate (NaHC03) 
at pH 8 .5  for 30  minutes. The concentration of P (units: J.!g/ml) in the filtered extract was 
measured through an Autoanalyser system, using a modified method of Murphy & Riley 
( 1 962) and Watanabe & Olsen ( 1965). 
Immediately plant-available S04-S was determined by extraction of 4 g of soil with a 
20 m1 solution of 0.02M potassium phosphate (K2HP04) at pH 4.0 for 30  minutes. The 
concentration of S04-S (units: ppm) in the filtered extract was measured by high 
performance ion chromatography (HPIC) (Watkinson & Kear 1 994). 
The ASC3 of the soil was determined by the method of Saunders ( 1965). This procedure 
involved mixing 5 g of soil with 25 ml of potassium dihydrogen phosphate buffer for 1 6  
hours. After this time lapsed, the remaining phosphate in the filtered solution was 
measured by spectrophotometry (Saunders 1 965; Cornforth 1980). 
2 Formally known as Ministry of Agriculture and Fisheries (MAY) Quicktest procedures. 
3 Formally known as Phosphate retention (Cornforth 1 980). 
Chapter 4 1 31 
SOC was measured by combusting solid soil samples and determining by infrared the 
amount of carbon dioxide (C02) produced. SOC content was converted to SOM by 
multiplying by a factor of 1 .724 (Foth 1978; Nelson & Sommers 1 982; Park et al. 1 994). 
4.3 Data analyses 
For each site, a split-plot analysis of variance (ANOV A) was conducted, using the general 
linear model (GLM) procedure of SAS@ (version 8.02 for Windows@, SAS Institute, Inc. 
1 999), to test the effects of overstorey environment (main plots), soil depth (subplots), and 
their interactions on the soil properties. In the GLM, a 'replicate nested within overstorey 
environment' parameter was treated as a random-effect, while all other parameters were 
considered fixed (Hedderley per. comm. 2002). The 3 x 2 factorial design was unbalanced, 
with treatment combinations having 2- 1 1  replicates (Figure 4. 1 ). As a result, Type 3 sums 
of squares were used instead of Type 1 ,  due to their more conservative nature. Diagnostic 
options provided in SAS@ were used to check the underlying ANOV A assumptions. These 
included studentised residual and normal probability plots, along with more formal test 
statistics (SAS 1 990). Several soil properties required transformation to meet either the 
assumption of normality or homogeneity of variance. These soil properties are clearly 
identified in the appropriate results tables. One data point was identified as an outlier and 
removed for the analysis of plant-available Mg at Kiwitea. Separation of treatment means 
was by the Tukey-Kramer multiple comparison method (PDIFF ADJUST=TUKEy) for 
unequal sample sizes (SAS 1990). 
Simple linear regression analyses, using the GLM procedure of SAS@, were performed to 
examine the relationship between the soil properties and the in-leaf poplar canopy cover 
ratio (CCL). Separate regression analyses were carried out for the two main overstorey 
environments amongst the poplar stands (Zones 1 & 3) and also for the two soil strata 
(0-75mm & 75-1 50mm). Differences between the regression equations were tested by 
analysis of covariance (ANCOV A), using the GLM procedure of SAS@ (Littell et al. 1 99 1 ;  
Kleinbaum et al. 1998). Where the regression equations for the two main overstorey 
environments amongst the poplar stands (Zones 1 & 3) coincided (i.e. were not 
significantly different in intercept or slope) the data were combined into a single function 
representing the entire understorey environment. The quality of fit of the simple linear 
Chapter 4 1 32 
regressions was checked through inspection of scatter, residual, and nonnal probability 
plots, along with more formal test statistics (SAS 1 990). 
Soil samples collected 
Environment: 
Soil depth (mm): 
Replicates: 
Kiwitea 
Hautope 1 
Open pasture 
0-75 
3 
2 
75- 1 5 0  
2 
2 
I 
0-75 
1 1  
1 0  
Amongst the stands of poplar 
Zone 1 
75-1 50 
6 
6 
I 
0-75 
1 1  
1 0  
Zone 3 
75-1 50 
6 
6 
Figure 4.1 Unbalanced nested factorial-treatment structure. Abbreviations: Zone 1, area directly below the 
poplar crown in the north-eastern corner of an experimental unit with trees; Zone 3, area within the vertically 
projected canopy gap between the four nuclei trees defining an experimental unit (refer to Section 3.2.3.3, 
Figure 3 .3); open pasture, adjacent open pasture area (control) away from the influence of the poplar trees. 
4.4 Results 
4.4.1 Effect of overstorey environment on soil properties within two 
separate soil strata at Kiwitea 
When averaged over both 0-75 mm and 75- 1 50 mm soil strata, soil pHw and plant­
available S04-S levels varied significantly between the three main overstorey 
environments (Table 4. 1) .  Also, there was weak evidence of an overstorey environment 
effect on the levels of K, Ca, and SOC (Table 4. 1). Soil pHw and S04-S levels were 
0.4 units and 1 -2 ppm, respectively, higher amongst the poplar trees (Zones 1 & 3) than in 
the adjacent open pasture. The concentrations of K tended to be greater in the vertically 
projected canopy gap between the trees (Zone 3), followed by directly below the poplar 
crowns (Zone 1 ), and then open pasture. Similarly, plant-available Ca was marginally 
higher in Zone 3 than in the open pasture. However, the Ca concentration in Zone 1 was 
not significantly different from either of the other two main over storey environments. The 
soil in the open pasture contained 1 %  more SOC than amongst the poplar stands. Plant­
available P, Mg, and Na in the soil, along with ASC, did not vary significantly between the 
Chapter 4 1 33 
three main overstorey environments (Table 4. 1). All of the soil chemical properties, except 
for ASC, decreased significantly with increasing soil depth (Table 4. 1 ). 
4.4.2 Effect of poplar overstorey density on soil properties within two 
separate soil strata at Kiwitea 
In the upper 0-75 mm of soil pHw generally increased under greater CCL (Figure 4.2a). 
The simple linear relationship, which included both main zones of tree influence (Zones 1 
& 3), was strong (r2=0.66; P<O.OOO I )  with little variation around the fitted function 
(CV=3%). The rate of increase in soil pHw amongst the poplar trees was 0. 1 units for 
every 1 0% increase in CCL (Figure 4.2a). In the lower 75- 1 50 mm of soil pHw was not 
significantly related to CCL. 
The concentrations of Ca, Mg, and K in the upper 0-75 mm of soil also increased under 
greater CCL (Figure 4.2b,c,d). Similar relationships occurred in both Zones 1 and 3 for Ca 
and Mg (Figure 4.2b,c), while for K the relationship was restricted to Zone 3 (Figure 4.2d). 
Overall, Ca, Mg, and K increased at a rate of 0.3, 1 .7, and 1 .2 quick-test units, respectively, 
for every 10% increase CCL (Figure 4.2b,c,d). However, the strength of the above 
relationships was generally weak (r2�0.41 )  and there was considerable variation around the 
fitted functions (CV=15-46%). In the lower 75- 1 50 mm of soil, Ca, Mg, and K were not 
significantly related to CCL. 
In contrast, the ASC in the upper 0-75 mm of soil decreased under greater CCL 
(Figure 4.3a). The relationship did not differ significantly between the two main overstorey 
environments amongst the poplar trees (Zones 1 & 3); and in Zone 3 a similar response 
also occurred further down the soil profile within the 75-150  mm stratum (Figure 4.3c). 
The strength of the linear relationship for 0-75 mm soil depth was weak (r2=0.25;  
P=O.007), increasing to a moderate level (r2=0.5 1 ;  P=0.03) further down the soil profile 
(75- 1 50 mm). 
The SOC content also decreased under greater CCL (Figure 4.3b,d). The relationship was 
limited to Zone 1 for the upper 0-75 mm of soil (Figure 4.3b), while it occurred in both 
zones further down the soil profile (75- 1 50 mm) (Figure 4.3d). SOC decreased at a rate of 
Chapter 4 1 34 
0. 1 7% for every 1 0% increase eeL, and the relationship did not vary significantly 
(P>O.05) between the two soil strata (Figures 4.3b,d). 
The concentrations of Na, P, and 804-S in the soil did not change significantly (P>O.05) 
with increasing eeL. 
Table 4.1 Kiwitea soil chemical propertiesl : directly below the poplar crowns (Zone 1), at the centre of the vertically projected canopy gap between the trees (Zone 3), and 
within the open pasture (Open pasture). 
Soil depth Environment pHw 
0-75 mm Open pasture 5.5 
Zone I 6. 1 
Zone 3 6.2 
Grand mean 5.9 
75- 150  mm Open pasture 5 .6 
Zone I 5.9 
Zone 3 5.9 
Grand mean 5.8 
Overall 0- 1 50 mm Open pasture 5.6 
Zone 1 6.0 
Zone 3 6.0 
Grand mean 6.0 
Analysis of variance 
Environment (A) P<0.05 
Soil depth (B) P<O. 1 
Interaction (A *B) NS 
SD 0. 1 
P 
(Ilg/mi) 
(lnY) 
17.8  (2.74) 
1 8.7  (2.79) 
1 9.8  (2.88) 
19. 1 (2.80) 
14.0 (2.22) 
7.8 (1 .98) 
1 0.5 (2.26) 
9.9 (2. 1 5) 
1 6.3  (2.48) 
1 4.9 (2.38) 
16.5 (2.57) 
15 .8  (2.58) 
NS 
P<O.OO I 
NS 
(0.22) 
K 
(QT) 
(InY) 
6.2 ( 1 .78) 
10.8 (2.3 1) 
1 3 . 1  (2.44) 
1 1 .3 (2. 18) 
4.3 ( 1 .26) 
7.0 (1 .89) 
8 .7  (2.09) 
7 .3  (1 .75) 
5.4 (1 .52) 
9.5 (2. 1 0) 
1 1 .5 (2.26) 
9.8 (2. 14) 
P<O. I 
P<O.OO I 
NS 
(0. 14) 
S04-S 
(ppm) 
(lnY) 
3.2 (1 .03) 
5.5 (1 .67) 
4.5 ( 1 .43) 
4.8 (1 .38) 
2.0 (0.38) 
3.2 (1 . 1 1) 
2.5 (0.90) 
2.7 (0.80) 
2.7 (0.70) 
4.6 (1 .39) 
3.8 ( 1 . 1 7) 
4.0 (1 .28) 
P<0.05 
P<O.OO I 
NS 
(0.24) 
Ca 
(QT) 
(lnY) 
5.5  ( 1 .70) 
6.8 (1 .90) 
7.7 (2.03) 
7. 1 ( 1 .88) 
4.8 ( 1 .59) 
5.7 ( 1 .71)  
6.3 (1 .82) 
5.8 ( 1 .7 1 )  
5.2 ( 1 .65) 
6.4 (1 .81)  
7.2 (1 .93) 
6.6 (1 .86) 
P<O.1  
P<0.05 
NS 
(0. 14) 
Mg 
(QT) 
26.5 
36.7 
36.0 
33 . 1  
24.0 
28.7 
27.2 
26.6 
25.2 
32.7 
3 1 .6 
32.2 
NS 
P<O.OO I 
NS 
3 .3 
Na 
(QT) 
( 1 tY) 
5 . 7  (0. 1 8) 
6.0 (0. 1 7) 
6.4 (0. 1 6) 
6. 1 (0. 1 7) 
4.5 (0.23) 
5.8 (0. 19) 
4.8 (0.22) 
5.2 (0.21 )  
5.2 (0.20) 
5.9 (0. 1 8) 
5.8 (0. 1 9) 
5 .8  (0. 18) 
NS 
P<0.05 
NS 
(0.03) 
ASC 
(%) 
(lnY) 
20.5 (3.02) 
14.5 (2.62) 
1 5 . 1  (2.70) 
15 .5  (2.78) 
22.3 (3.05) 
1 9.8  (2.87) 
1 7.3 (2.89) 
1 9. 1  (2.94) 
2 1 .2 (3.03) 
16.4 (2.74) 
1 5.9 (2.80) 
16.8 (2.78) 
NS 
P<O.O I  
NS 
(0.09) 
SOC 
(%) 
4.7 
3 .8 
3 .9 
4 . 1  
3.4 
2.5 
2.5 
2.8 
4. 1 
3 . 1  
3.2 
3.4 
P<O. 1 
P<O.OO I 
NS 
0.2 
SOM 
(%) 
8.2 
6.5 
6.8 
7.2 
6.0 
4.3 
4.3 
4.9 
7. 1 
5.4 
5.5 
5.9 
P<O. 1 
P<O.OOI 
NS 
0.3 
lAveraged across a range of different poplar stand densities. Abbreviations: QT, AgResearch Quick test units; InY, natural-logarithm transformed; SD, standard deviation; 
NS, non significant (P>O. l ); P, Phosphate; K, potassium; S04-S, sulfate-sulfur; Ca, calcium; Mg, magnesium; Na, sodium; ASC, anion storage capacity; SOC, soil organic 
carbon; and SOM, soil organic matter. 
� 
W 
01 
0-75 mm 
6.8 ,.----------------, 
Y = 5.56 (0.08) + x·O.O I (0.001) 
6.6 r2=0.66, RMSE=0.20, P<O.OOOI 
6.4 
6.2 
6.0 
5.8 
5.6 
o 
to • 
(a) 
5.4 +-
-----,-------,------,--------,-----4 
o 20 40 60 80 100 
60 .-------------------------------, 
Y = 26.49 (2.40) + x·0. 1 7  (0.04) 
li' ;=0.41 ,  RMSE=5.58, P=0.0004 
'2 50 
(c) 
• 
;::l 
t; 
41 ] 40 • 
� § 30 } 20 • • 
1 0 +------,-------,------,--------.r---� 
o 20 40 60 80 100 
Canopy closure ratio with leaves (%) 
0-75 mm 
1 2 ,.------------------------------, 
Y = 5.62 (0.50) + x·0.027 (0.008) 
,-... 1 1  r
2=0.29, RMSE=1 . 1 6, P=0.003 '" '§ 10  
9 
8 
7 
6 
5 
o 
• • 
o 
• 0 
o 0 
(b) 
4 +-----�----�------r_----�----� 
o 20 40 60 80 lOO 
25 ,.------------------------------, 
Y = 5 .97 (2.53) + x·0. 1 2  (0.04) 
';i;' r2=0.33, RMSE=5. 1 8, P=0.02 
.t: 20 
§ 
t; � 1 5  
'3 
.z § ID 
.� 
cl: 5 o 
o 0 
o 
(d) 
0 +------,------,.------,--------,-----1 
o 20 40 60 80 lOO 
Canopy closure ratio with leaves (%) 
Figure 4.2 Kiwitea 0-75 mm soil chemical properties linearly related to poplar canopy closure (CCL). Symbols: (e) directly below the poplar 
crowns (Zone 1 )  and (0) at the centre of the vertically projected canopy gap between the trees (Zone 3). Standard errors for the simple linear 
regression coefficients are given in parentheses. Abbreviations: r2, adjusted coefficient of determination; RMSE, standard error of prediction. 
30 
---. 
� 25 
. � <) � 20 
<) d) tlO Cd ... 1 5  .9 
'" 
I'l 0 
� 10 
5 
6 
5 ---'#. 
'-" 
I'l 4 � <) <) 
.� 3 
g 
2 
0-75 mm 
• 
0 
Y = 19.94 ( 1 .65) - x·0.08 (0.03) 
r2=0.25, RMSE=3.85, P=0.007 
0 20 40 60 
• 
Y = 4.85 (0.27) - x·0.0 1 8  (0.005) 
r2=O.5 1 ,  RMSE=0.55, P=0.003 
0 20 40 60 
• 0 o .  • • • 
80 
• 
• 
80 
Canopy closure ratio with leaves (%) 
75-150 mm 
30 
(a) (c) 
� 25 e...-
.� <) � 20 
<) 
� 1 5  � 0 0 1iS 
I'l 0 
� 10  Y = 22. 17  ( 1 .44) - x·0.07 (0.03) 
5 
r2=0.5 1 ,  RMSE=2. 14, P=0.03 
100 0 20 40 60 80 1 00 
6 
(b) I Y::  3.46 (0.23) � x·0.0 16  (0.004) (d) 5 r2=O.59, RMSE-0.36, P=0.0008 ---� e...-
I'l 0 4 � <) 
'§ 3 
g 
2 
• 
1 
100 0 20 40 60 80 1 00 
Canopy closure ratio with leaves (%) 
Figure 4.3 Kiwitea soil chemical properties linearly related to poplar canopy closure (CCL). Symbols: (e) directly below the poplar crowns 
(Zone 1) and (0) at the centre of the vertically projected canopy gap between the trees (Zone 3). Standard errors for the simple linear function 
coefficients are given in parentheses. Abbreviations: r2, adjusted coefficient of determination; RMSE, standard error of prediction. � 
W 
.......,j 
Chapter 4 1 38 
4.4.3 Effect of overstorey environment on soil properties within two 
separate soil strata at Hautope 1 
In the upper 0-75 mm of soil pHw was 0.2-0.3 units higher amongst the poplar trees 
(Zones 1 & 3) than in open pasture (Table 4.2). However, the reverse occurred in the lower 
75- 1 50 mm of soil as a result of the open pasture pHw increasing by 0.3 units, while pHw 
decreased in Zone 1 and did not change significantly in Zone 3 (Table 4.2). At this lower 
soil depth the difference in pHw between Zone 1 and open pasture was only marginal 
(P=0.07). 
The concentration of K in the upper 0-75 mm of soil was highest in Zone 1 ,  followed by 
Zone 3, and then the open pasture (Table 4.2). With increasing soil depth, K decreased by 
3-5 quick-test units in Zones 1 and 3, while not changing significantly in the open pasture 
(Table 4.2). The decrease in Zone 3 K made it comparable with the open pasture, whereas 
in Zone 1 the concentration remained significantly higher than in the other two overstorey 
environments (Table 4.2). 
For Mg there was a weak interaction (P=0.07) between overstorey environment and soil 
depth (Table 4.2). Mg in the upper 0-75 mm of soil tended to be 1 3 . 1- 1 4.2 quick-test units 
greater in both Zones 1 and 3 compared to in the open pasture (Table 4.2). However, 
further down in the 75-1 50 mm soil stratum, Mg did not vary significantly between the 
three main overstorey environments (Table 4.2). 
Ca in the upper 0-75 mm of soil did not differ significantly among the three main 
overstorey environments (Table 4.2). Nevertheless, in Zones 1 and 3 the concentration 
tended (P=0.06) to decrease by 0 .8  quick-test units with increasing soil depth, whereas, in 
the open pasture it did not change significantly. As a result, in the 75- 1 50 mm of soil, 
Zone 1 Ca tended (P=O.06) to be slightly lower than in the open pasture (Table 4.2). 
Again, little difference was found with Na. When averaged across both soil depths, the 
highest concentrations ofNa were in Zone 1 ,  followed by Zone 3, and then open pasture 
(Table 4.2). 
Chapter 4 1 39 
For both soil depths Zone 1 ,  followed by Zone 3, tended (P=0.08) to have the greatest 
concentration of P (Table 4.2). In all three main overstorey environments P decreased with 
soil depth (Table 4.2). 
In the upper 0-75 mm of the open pasture soil S04-S was around 1 1  ppm greater (P=0.06) 
than amongst the poplar trees (Table 4.2). However, in the lower 75- 1 50 mm of soil the 
reverse occurred as a result of a significant (p<0.00 1 )  decrease in the level of S04-S in the 
open pasture. 
The Ase increased significantly with soil depth when averaged across all three main 
overstorey environments (Table 4.2). Also, there tended (P=0.05) to be an interaction 
between the overstorey environment and soil depth (Table 4.2). In the upper 0-75 mm soil 
stratum, the ASe in Zone 3 was marginally higher than in Zone 1 or the open pasture. 
Whereas, further down in the 75- 1 50 mm soil stratum, Zones 1 and 3 both had an Ase 
around 6% greater than in the open pasture (Table 4.2). 
When averaged over both soil depths, Zone 1 had 0.5-0.6% more soe than in Zone 3 or 
the open pasture (Table 4.2). The SOC in all three main overstorey environments 
decreased significantly with soil depth (Table 4.2). 
4.4.4 Effect of poplar overstorey density on soil properties within two 
separate soil strata at Hautope 1 
In the upper 0-75 mm of soil � Na, and Mg increased under greater eCL (Figure 4.4). 
The simple linear relationship for K did not differ significantly (P>0.05) between the two 
main overstorey environments amongst the poplars (Figure 4.4a). In contrast, for Na and 
Mg the relationships only occurred in Zone 1 and Zone 3, respectively (Figure 4.4b,c). The 
rate of increase for these basic cations was 0.9, 0.5, and 2.3 quick-test units, respectively, 
for every 1 0% increase CeL. However, the strength of the above relationships was weak 
(r2::;0.35), with a large amount of variation around the fitted functions. Furthermore, data 
points were concentrated around the two extremes in CCL, which reduced confidence in 
the fitted functions. 
Chapter 4 1 40 
Soil pHw, ca, P, S04-S, ASC, SOC in the upper soil stratum (0-75 mm) did not change 
significantly (P>0.05) over the limited range of CCLs. 
In the lower 75-1 50 mm of soil, P and Na, along with ASC, increased under greater CCL 
(Figure 4.5b,c,d). This contrasted with soil pHw, which at the same depth decreased with 
increasing CCL (Figure 4.5a). P and Na increased by 0.6 J1g1ml and 0.4 quick-test units, 
respectively, for every 1 0% increase in CCL (Figure 4.5c,d). The slope of the simple linear 
relationship for Na was not significantly different (P>0.05) from the same relationship 
found in the upper 0-75 mm of soil (Figures 4.4b & 4.5d). Conversely, soil pHw decreased 
by 0.05 units for every 1 0% increase in CCL (Figure 4.5a).  Overall, the strength of the 
above relationships ranged from weak to strong (r2=O.34-0.64). However, the relationships 
were based on a very limited range of CCLs and the soil properties, except for soil pHw, 
exhibited disproportionately greater variation towards higher CCLs. 
Table 4.2 Hautope 1 soil chemical and biochemical properties' :  directly below the poplar crowns (Zone 1 ), at the centre of the vertically projected canopy gap between the 
poplar trees (Zone 3), and within the open pasture (Open pasture). 
Soil depth Environment pHw P K S04-S Ca Mg Na ASC SOC SOM 
(Ilg/ml) (QT) (ppm) (QT) (QT) (QT) (%) (%) (%) 
(lnY) (lnY) (lnY) 
0-75 mm Open pasture 5.5 1 5.0 (2.62) 7.5 24 .0 (3.03) 8.5 39.0 7.0 ( 1 .94) 1 6.0 4.3 7.4 
Zone 1 5 .8 20.7 (2.96) 16.5 1 3 .5 (2. 5 1 )  8.6 52. 1 10.3 (2.32) 16.9 4.9 8.5 
Zone 3 5.7 16.2 (2.76) 1 1 .6 12.4 (2.50) 8.8 53.2 9.0 (2. 1 8) 20.6 4.2 7.3 
Grand mean 5.6 18. 1  (2.78) 1 1 .9 14.0 (2.68) 8.6 48. 1 9.4 (2. 1 5) 17.8 4.5 7.8 
75-150  mm Open 5.8 3 .5 ( 1 .24) 6.0 4.5 ( 1 .44) 9.0 49.0 6.0 ( 1 .79) 1 6.0 2.3 3.9 
Zone 1 5.6 7.3 ( 1 .93) 1 1 .  7 1 1 .5 (2.38) 7.8 49. 1 9.8 (2.35) 22. 1 2.8 4.8 
Zone 3 5.6 5 .8 ( 1 .75) 8.6 10.7 (2.39) 8.0 5 1 .7 9.3 (2.29) 22.8 2.4 4.2 
Grand mean 5.7 6. 1 ( 1 .64) 8.8 10. 1 (2.07) 8.2 49.9 9. 1 (2. 14) 20.3 2.6 4.4 
Overall 0- 1 50 mm Open 5.6 9.3 ( 1 .93) 6.7 14.3 (2.24) 8.8 44.0 6.5 ( 1 .86) 16.0 3.3 5.7 
Zone 1 5.7 14.0 (2.44) 14. 1 12.5 (2.44) 8.2 50.6 10. 1 (2.33) 1 9.5 3 .8 6.7 
Zone 3 5.6 1 1 .0 (2.26) 10. 1  1 1 .5 (2.44) 8.4 52.4 9.2 (2.24) 2 1 .7 3.3 5.7 
Grand mean 5.6 13 .5 �2.42) 1 1 .7 12.5 �2.4 1 )  8.4 5 1 . 1  9.3 (2.2 1)  1 9.8 3.8 6.5 
Analysis of variance 
Environment (A) NS P<O. l P<O.O l NS NS NS P<O.OO I P<O. 1 P<O.Ol P<O.OO l 
Soil depth (B) NS P<O.OO I P<O.OO l P<O.OO l P<O. 1 NS NS P<O.Ol P<O.OOI P<O.OO l 
Interaction (A *B) P<O.OO I NS P<0.05 P<O.OO I P=0.05 P<O. l P<O. l P=0.05 NS NS 
SD 0. 1 (0. 17) 1 .0 (0.20) 0.4 4.4 (0.08) 1 . 8  0.3 0.5 
'Averaged across the range of poplar stand densities. Abbreviations: QT, AgResearch Quick test units; SEM, standard error ofthe mean; NS, non significant; P, phosphate; 
K, potassium; S04-S, sulphate-sulphur; Ca, calcium; Mg, magnesium; Na, sodium; ASC, anion storage capacity; SOC, soil organic carbon; and SOM, soil organic matter. 
.....Iio. 
� .....Iio. 
()"75 mm 
25 
'Cii' • 
• 
.';:: 20 § 
1;; 
B 
ii 1 5  
'3 
C" '-' 
.� 1 0  
ell 
� 
'0 
� 5 Y = 8.36 (2.57) + x·0.09 (0.04) 
r2=<l. 14, RMSE=3 .97, P=0.05 
A 
0 20 40 60 
65 
� 60 j 
'! 55 � 
B 
ii 50 
'g. 
'-' 45 
.� 
i 40 
::E 35 i Y = 38.2 (5.6) + x·0.23 (0.09) 
30 1
r2=O·33, RM�E=8.3, P=0.�3 � 
0 20 40 60 
Canopy closure ratio with leaves (%) 
. .. 
• 
• 
. � 
(a) 
80 
A 
� 
� � 
� 
(c) 
80 
0-75 mm 
1 3  
1 2  r2=0.35, RMSE=1 .59, P=O.03 
r � 7.22 ( 1 .07) + x·O.O, (0.02) 
'Cii' .� 1 1  .. 
1;; 10  ¥ 
� 9 '3 
C" '-' 8 
§ 
� 7 
6 
5 
0 20 40 
• 
60 
Canopy closure ratio with leaves (%) 
• •  
.. 
(b) 
80 
Figure 4.4 Hautope 1 0-75 mm soil chemical properties linearly related to poplar canopy closure (CCL). Symbols: ("' ) directly below the poplar 
crowns (Zone 1) and (�) at the centre of the vertically projected canopy gap between the trees (Zone 3). Standard errors for the simple linear 
function coefficients are given in parentheses. Abbreviations: r2, adjusted coefficient of determination; RMSE, standard error of prediction. � � I\.) 
75-150 mm 
6 0L 5.8 +--
a 
5 6 1 � " " 
5.4 " 
,,-.. 
El 
� '-' 
� 
-a ell 0 
� 
5.2 i 
Y = 5.8 1 (0.06) - x·0.005 (0.001) 
1r2=0.64, RMSE=0.09, P=0.0004 5.0 i i 
0 20 40 
12  
Y = 3 . 19 ( 1 . 16) + x·0.06 (0.02) 
r2=0.34, RMSE=l .77, P=0.02 
10  
8 
6 
4 
2 I 
0 20 40 
60 
60 
Canopy closure ratio with leaves (%) 
" 
"" 
" 
(al l 
I 
80 
(c) 
I 
80 
75-150 mm 
35 r: 1 5.54 (2.05� + x·0. 12 (0.04) (b) 
,,-.. r2=O.45, RMSE-3. 12, P=0.006 
� 30 " 
.e 
g J " � 25 "" u 
� 
S 20 ti 
.§ 
� 15  
10  
0 20 40 60 80 
12 
1 1  J (d) " " " ,,-.. 
.� 
§ 1O � " 
ti 
£ 9 i � 
·3 0- 8 '-' 
§ 
:e 7 
c55 
6 � Y = 6.76 (0.8 1) + x·0.04 (0.0 1) 
5 I r2=O.4, RMSE=1 .23, P=O.009 i i 
0 20 40 60 80 
Canopy closure ratio with leaves (%) 
Figure 4.5 Hautope 1 soil chemical properties linearly related to poplar canopy closure (CCL). Symbols: (A)  directly below the poplar crowns 
(Zone 1 )  and (�) at the centre of the vertically projected canopy gap between the trees (Zone 3). Standard errors for the simple linear function 
coefficients are given in parentheses. Abbreviations: r2, adjusted coefficient of determination; RMSE, standard error of prediction. � � 
W 
Chapter 4 1 44 
4.5 Discussion 
4.5.1 Soil pH 
In the upper 0-75 mm of soil at both Kiwitea and Hautope l one of the main impacts that 
poplar trees had was a 0.2-0.7 unit increase in soil pHw compared to the adjacent open 
pasture (Tables 4. 1 & 4.2). At Kiwitea, the elevation in soil pHw was greater under higher 
CCL (Figure 4.2a). However, no such effect was found at Hautope 1 .  Based on the scatter 
plot of the data, the poor range of CCL sampled (skewed between 45-75% CeL) and the 
high amount of variation in soil pHw under the trees possibly contributed to the lack o f  any 
significant relationship at this site (data not shown). The greater concentration of organic 
matter in the soil under the trees at Hautope 1 than at Kiwitea (t-test, P=0.0006) could have 
increased its pH buffering capacity (Helyar 1 987; Curtin & Rostad 1 997; Giddens et al. 
1 997; Alfredsson et al. 1 998). A similar effect would also be achieved through higher clay 
content in the soil (During 1 984; McBride 1 994; Wong et al. 2000). 
Using a similar "paired-site" approach to this study, Guevara-Escobar et al. (2002) 
measured the soil pHw at 1 0  different sites in the same Manawatu-Wanganui region as 
Kiwitea. In the upper 0-75 mm of soil, pHw directly under poplar trees (> 5 years old) was 
0.5- 1 .2 units higher than in open pasture. In general, the soil pHw was around 5 . 5  units in 
the open pasture and 6. 1 units directly under the trees. This corresponded very closely to 
the soil pHw levels measured under the same respective overstorey environments at 
Kiwitea (Table 4. 1 ). However, the increase in soil pHw under the trees was more marked 
than at Hautope 1 (Table 4.2). 
Spatially, Guevara-Escobar et al. (2002) measurements were restricted to directly below 
the trees (analogous to Zone 1 )  and in the open pasture. The more intensive stratified 
sampling used in the present study indicates that poplar trees have a similar effect on 
soil pHw both directly under and in the gaps between their crowns (Tables 4. 1 & 4.2; 
Figures 4.2a & 4.5a). 
The higher soil pHw under poplars is the opposite of that normally found under the main 
exotic tree species planted in New Zealand - Pinus radiata D. Don. In genera4 P. radiata 
has been shown to acidify the soils of unimproved (Alfredsson et al. 1998) and improved 
Chapter 4 1 45 
(Hawke & O'Connor 1 993 ; Giddens et al. 1 997; Parfitt et al. 1 997; Cossens & Hawke 
2000) pasture/grassland. Based on the Tikitere Agroforestry Research Area near Rotorua, 
Hawke & O'Connor (1 993) reported that greater soil acidification occurred under older 
stands of P. radiata planted at a higher stocking rate; both attributes are associated with 
greater CCL. 
Soils under legume-based pasture also generally become more acidic with time (Ridley et 
al. 1 990; Bolan et al. 1 99 1 ;  Haynes & Williams 1 993). If unchecked this can cause pasture 
production losses (especially pH <5 .5) through reduced nutrient availability and phytotoxic 
effects of elevated AI and Mn levels in the soil solution (Edmeades 1 986; Dodd et al. 
1 992; McLaren & Cameron 1996; Wheeler & O'Connor 1 998). Adding lime to pastures is 
widely used in New Zealand to control/ameliorate soil acidification (Sinclair 1 995; 
Wheeler 1 997). However, on hill pastures that do not allow ground-based applications, 
liming is generally considered uneconomic owing to high aerial-application costs (Sinclair 
1 995 ; MacLaren 1 996; de Klein et al. 1997). For example, on moderately buffered 
New Zealand sedimentary soils 6- 10  tonneslha of good quality limestone (approx. 80% 
CaC03) is usually required to raise the soil pH by 1 unit (Morton et al. 1 994; Sinclair 
1 995). Therefore, the integration of widely spaced poplar trees across hill pastures could 
potentially be used as an alternative bioremediation measure for controlling soil 
acidification. 
4.5.1 . 1  Potential processes causing a greater decrease in open pasture soil pH, 
compared to amongst the poplar stands 
Many different processes affect the balance of hydrogen ions CHl and hydroxyl ions 
(OIr) in soil solution, which in turn determines pH (Van Breemen et al. 1 983; Binkley & 
Richter 1 987; Helyar & Porter 1 989; Sinclair 1 995). In temperate legume-based pastures 
generally the most important processes are associated with the carbon (C) and nitrogen (N) 
cycles (Helyar & Porter 1 989; Ridley et al. 1 990b; Bolan et al. 1 99 1 ;  Tang et al. 1 999; 
Bolan et al. 2003). 
Carbon (C) cycle 
Additional W ions are added to soil solutions from the C cycle via: the dissociation of 
carbonic acid (H2C03), formed originally from carbon dioxide (C02) respired by plant 
roots and heterotrophic soil organisms; and from the synthesis and dissociation of organic 
Chapter 4 1 46 
acids (Van Breemen et al. 1 983;  Binkley & Richter 1987; Bolan et al. 1 99 1 ;  de Klein et al. 
1 997). 
Dissociation of carbonic acid (H2C03) 
The dissociation ofH2C03 releases both It and bicarbonate (HC03) into the soil solution 
(Binkley & Richter 1 987; de Klein et al. 1 997). Where rainfall exceeds evapotranspiration, 
these HC03- anions are leached from the topsoil in association with companion cations 
(e.g. exchangeable Ca2+, Mg2+, K+, Na l. This leaves a pool of additional W, which either 
replace some of the removed cations on soil colloid exchange complexes (along with Al3l, 
or are used in the weathering (e.g. hydrolysis) of minerals (Binkley & Richter 1 987; 
McBride 1 994; Fisher & Binkley 2000). 
de Klein et al. ( 1 997) noted some studies have shown that the extra pool of It created 
through the dissociation of H2C03 is relatively small compared to other acidifYing 
processes. Nevertheless, Parfitt et al. ( 1 997), Chen et al. (2000), and Saggar et al. (2001)  
all measured significantly higher concentrations of CO2 in soils under open pastures 
compared to under adjacent conifer (Pinus spp.) stands. In the study by Parfitt et al. 
( 1997), the resulting elevated concentration of RC03 - in the open pasture soil was the main 
inorganic anion regulating the leaching of base cations.  
Parfitt et al. (1 997) attributed elevated CO2 and HC03- concentrations in open pasture soil 
to higher rates of root and microbial respiration and also to a lower rate of gas diffusion. 
Similarly to Kiwitea (Table 4. 1 ), the open pasture had a greater amount of SO M than under 
the trees, and this, along with a lower soil C:N ratio, was postulated to favour enhanced 
microbial respiration. In contrast, at Rautope 1 the SOM content directly below the trees 
(Zone 1 )  was higher than in the open pasture (Table 4.2). Nevertheless, the biochemical 
composition of this SOM may still have inhibited its rate of turnover. Kochy & Wilson 
( 1 997) reported that the aboveground litter from a mixed-grass prairie decomposed faster 
than senesced leaves shed from an adjacent Populus tremuloides (aspen) forest. 
Differences in the decomposition rate, and thus presumably CO2 production, was mainly 
related to the biochemical nature of the litter, rather than to plant-induced changes to the 
micro climate, although both had a significant effect. Overall, it was postulated that the 
decomposition rate of the senesced poplar leaves was limited by a lower N concentration, 
in comparison to the prairie litter (Kochy & Wilson 1 997). 
Chapter 4 1 47 
Another possible cause of lower CO2 concentrations in the soil atmosphere under poplar 
trees is the presence of substances that restrict the growth of heterotrophic decomposer 
communities. Populus species contain condensed tannins (proanthocyanidins) (Driebe & 
Whitman 2000; Kemp et al. 2001).  These high molecular weight secondary compounds 
form resilient complexes with proteins, making them less available as substrates for 
decomposer soil organisms, and can inhIbit the activity of many different enzymes 
associated with the breakdown of organic matter (Benoit et al. 1 968; Benoit & Starkey 
1 968; Lewis & Starkey 1 968; Schimel et at. 1 996; Fierer et al. 2001 ;  Schofield et at. 
200 1 ). Also, some tannins are toxic to certain groups of soil organisms (Baldwin et al. 
1 983; Schimel et al. 1 996; Fierer et al. 2001). In contrast, most hill pasture plants either do 
not contain or have only trace amounts of condensed tannins (Jackson et al. 1 996; Wang et 
al. 1 996). 
Warmer soil temperatures, as a catalyst of enhanced plant root and microbial respiration, 
and slower gas diffusion through wetter soils in autumn and spring, could also increase the 
HC03- concentration in open pasture (Kowalenko et al. 1 978; Taylor & Parkinson 1 988; 
Parfitt et al. 1 997). The soil under poplar trees is normally 0.5-5.0° cooler than in adjacent 
open pasture (Crowe 1 993; Guevara-Escobar et at. 1 997) and can also be drier at certain 
key times of the year, such as late autumn through until early spring (Guevara-Escobar 
et al. 1997; Douglas et al. 200 1).  
Synthesis and dissociation of organic acids 
If cations and anions are not taken up in equilibrium by plant roots then either W or OIr 
ions are released back into the soil solution to maintain charge neutrality within the soil­
plant system (Binkley & Richter 1 987; Sinc1air 1 995 ; Tang et al. 2003). These W and OH­
ions are generated inside the plants mainly through the dissociation of organic acids and 
decarboxylation of organic acid anions, respectively (Bolan et al. 1 99 1 ). Therefore, the 
absorption of a greater ratio of inorganic cations-to-anions by plants causes two main 
effects: the excretion of W into the soil solution, which lowers the soil solution pH; and 
the formation of large quantities of charge-balancing organic acid anions inside the plants 
(Sinc1air 1 995; de Klein et al. 1 997; Tang et al. 1 999, 2003). However, the complete 
recycling and mineralisation of this plant material back into the same soil releases the 
inorganic ions taken up, and oxidation of the synthesized organic acid anions, along with 
the ammonification of the organic nitrogen compounds, consumes the previously excreted 
Chapter 4 1 48 
W resulting in no net change in soil solution pH (Binkley & Richter 1 987; Sinclair 1 995; 
de Klein et al. 1 997; Fisher & Binkley 2000). 
Many temperate pasture legumes in particular take up an excess of cations over anions 
(Braschkat & Randall 2004), thus releasing W back into the soil solution to maintain 
charge neutrality (Bolan et al. 199 1 ;  Haynes & Williams 1993; Sinclair 1995; de Klein et 
al. 1 997). If grazing animals consume this legume material and then permanently remove it 
from the system (via animal products or the transfer of dung and urine to camp areas) then 
the cycle described above is uncoupled and the W previously released cannot be 
neutralised, this causing soil acidification (Bolan et al. 1 99 1 ;  Sinclair 1 995 ; Fisher & 
Binkley 2000). In hill pastures, grazing animals usually ingest herbage from over a wide 
area of the paddock and return a large proportion of this material, in the form of dung and 
urine, back to relatively small camp areas (Gillingham & During 1973; Gillingham et al. 
1 980). At both Kiwitea and Hautope 1 ,  this effect (i.e. net transference of alkalinity) was 
likely enhanced in the open environment because of the greater pasture/legume production 
and animal carrying capacity than amongst the poplar trees (refer to Section 5 .4. 1) .  
Nitrogen (N)  cycle 
In legume-based pastures where new N inputs are solely added through symbiotic N2 
fixation, a permanent change in soil pH occurs via the N cycle only when nitrate (N03-) is 
leached from the system (Binkley & Richter 1 987; de Klein et al. 1997). Generally N03- is 
weakly held in the soil and, if not absorbed by plants or soil microbes, is prone to being 
leached with companion cations in drainage water (Bolan et al. 1 991) .  This prevents W 
produced during the nitrification process from being neutralised and leads to a permanent 
increase in soil acidity (Helyar 1976; Bolan et al. 1 99 1 ;  Sinc1air 1995; de Klein et al. 1 997; 
Fisher & Binkley 2000). Assuming the pastures within all three main overstorey 
environments at Kiwitea and Hautope 1 were equally utilised by grazing animals, then the 
greater pasture/legume production in the open, compared to amongst the poplar stands 
(refer to Section 5.4. 1 ), would result in a larger quantity of N being returned to the soil in 
dung and urine patches (Stee1e et al. 1 984; Heng et al. 1 991 ;  Scholefie1d et al. 1 993). The 
high N concentration aggregated in dung and especially urine patches (30- 100 g N/m2) 
usually far exceeds the immediate requirements of pasture plants and soil microbes (Ball & 
Ryden 1 984; Haynes & Williams 1 99 1 ,  1 993). Thus, in the open pasture there would be 
Chapter 4 1 49 
greater potential for excess N to be transformed into N03- and leached as a salt than under 
the trees (Zones 1 & 3). 
N03- leaching normally occurs in late autumn and winter (Holland & During 1 977; Field 
et al. 1 985; Sherwood & Fanning 1 989; Scholefield et al. 1 993; Ridley et al. 2001). Often 
N03- levels build up during summer when plant sinks are inactive due to water stress and 
the wetting up of soil in early autumn can also cause a flush in N mineralisation and 
nitrification (Helyar 1976; Heng et al. 1 99 1 ;  Scholefield et a1. 1 993 ; Herman et al. 2003). 
However, the large quantity of poplar litter and aqueous leachates released in autumn 
(Taylor et al. 1 989; Guevara-Escobar 1 999) could reduce the size of the labile N03- pool. 
Thibault et al. ( 1 982) found that dormant bud and foliar leachates from balsam poplar 
(Populus balsamifera L.) specifically inhibited nitrification. Lodhi & Killingbeck (1 980) 
and Baldwin et al. (1983) reported similar effects in ponderosa pine (Pinus ponderosa 
Dougl.) and balsam :fir (Abies balsamea [L.] Mill) stands, respectively, and attributed the 
inhibition to tannin and phenolic compounds from the trees. These C-rich secondary 
compounds can also decrease soil N03- levels by stimulating N immobilisation, while at 
the same time reducing N mineralisation (Azhar et al. 1 986; Fox et al. 1 990; Palm & 
Sanchez 1 99 1 ;  Schimel et al. 1 996; Fierer et al. 2001). Lower soil water content under the 
trees, compared to in the open pasture through autumn to spring would also reduce N03-
leaching (refer to C cycle). 
An additional pathway: Alkalisation of incident precipitation by the poplar canopy and litter 
The flux of W reaching the mineral soil in precipitation may be greater under open pasture 
compared to the combined effects of poplar trees and understorey pasture. Throughfall and 
stemflow under broadleaved trees is usually less acidic than incident rainfall (Parker 
1 983; Pastor & Bockheim 1 984; Potter et al. 199 1 ;  McBride 1 994; Roule et al. 1999; Neal 
2002). Also, a significant proportion of free It ions can be neutralised in reactions with 
forest floor organic horizons (Richter et al. 1 983 ; Richter 1 986; Pohlman & McColl 1988; 
Shibata et al. 1 995). Cation exchange reactions on foliage, bark, and litter surfaces appear 
to be the primary mechanism for the neutralisation of this acidity (potter et al. 1 99 1 ;  
Cappellato et al. 1993 ; Shibata et al. 1 995). However, if this was a major pathway for soil 
pHw differences, then Zone 1 would be expected to have a higher pHw than Zone 3 .  This 
was clearly not the case, as the soil pHw in Zones 1 and 3 were not significantly different 
at either of the sites (Tables 4. 1 & 4.2; Figure 4.2a). The annual fluxes of W deposited, 
Chapter 4 1 50 
especially in unpolluted regions, may also be relatively small compared with native soil 
acidity (Richter 1 986). Furthermore, low CEC soils such as at Kiwitea are typically acidic 
and as a consequence are less susceptible to base cation displacement and leaching by 
atmospheric acids (Richter 1986). 
4.5.1 .2 Potential processes causing an increase in  soil pH amongst the poplar 
stands, compared to in the open pasture 
Before the stands of poplar were planted at Kiwitea, Suckling ( 1975) had measured a soil 
pH of 5 .4 to 5 .5  units in the same paddocks used for the present study. Based on this 
historical data, soil pHw in the open pasture has not appreciably decreased over time, but 
instead has increased amongst the poplar trees. Furthermore, the significant relationships 
between soil pHw and CCL at the two farm sites indicate that the trees have a direct effect 
on soil pHw (Figure 4.2a & 4.5a). 
Guevara-Escobar et al. (2002) found that simply mixing poplar leaf litter into soil from an 
open pasture increased its pHw. Noble et al. ( 1 996) reported a similar effect for several 
other different tree species and also showed that the change in soil pH was dependent on 
the amount of litter added. In the latter study, the leaf litter contained an excess of 
inorganic cations over anions, with the charge balance maintained by synthesised organic 
acid anions. From this it was reasoned that during the decomposition of leaf litter, 
oxidation (decarboxylation) of organic acid anions would consume W from the soil, 
causing an increase in soil solution pH. Other alkalising-mechanisms proposed are that 
organic acid anions and functional groups are protonated (Hoyt & Turner 1 975; Wong 
et al. 1 998; Marx et al. 2002) or exchange with the terminal OIr ions of soil minerals 
(Hue et al. 1 986; Hue & Amien 1 989). Out of the major cations in leaf litter, the level of 
Ca was highly correlated with ash alkalinity, which in turn was used as an estimate 
(index) of the organic acid anion content (Noble et al. 1 996; Noble & Randall 1 999). In 
summer, the Ca concentration in the poplar leaves at Kiwitea and Hautope 1 was 4- and 6-
times greater than in the open pasture herbage, respectively (refer to Sections 5 .4.7 & 
5.4.8). 
Several studies have shown that poplar leaves contain a greater ratio of major inorganic 
cations-to-anions (Lousier & Parkinson 1 976; Pastor & Bockheim 1 984; Noble et al. 
1 996; Singh 1 998; Berthelot et al. 2000), with very little Ca or Mg, relative to other 
Chapter 4 1 51 
ions, being retranslocated prior to leaf fall (Baker & Blackmon 1 977; McColl 1 980; 
Pastor & Bockheim 1 984; Singh 1 998). Thus, at both Kiwitea and Hautope 1 ,  the oxidation 
of synthesised organic acid anions, during the decomposition of fallen poplar litter, could 
have been a major pathway for increasing the soil pHw (Haynes & Mokolobate 200 1). The 
effect depending on the amount of poplar leaf added to the soil, which in turn is related to 
CCL. Alternatively, the organic acid anions may have leached from the poplar foliage as a 
salt with mobile inorganic cations such as K and been carried into the soil via throughfall 
and stemflow (parker 1 983). 
However, as previously discussed for the C cycle in Section 4.5. 1 . 1 ,  an excess uptake of 
inorganic cations over anions into plant roots must be matched by the excretion of W back 
into the soil to maintain charge neutrality (Binkley & Richter 1 987; Bolan et al. 1 99 1 ) .  
This counter-balancing any alkalinity created from the oxidation o f  the organic acid 
anions. Therefore, in order for the above process to make the topsoil more alkaline an 
equal amount of acidity has to be imparted somewhere else within the soil profile (Noble 
et al. 1 996; Tang et al. 1 999; Marschner & Noble 2000). In other words, alkalinity is not 
independently synthesised through this mechanism, but is simply transferred (Haynes & 
Mokolabate 200 1 ). This could explain why in the 75- 1 5 0  mm of soil at Hautope 1 pHw 
was lower amongst the poplar trees, compared to the open pasture, while the situation was 
reversed in the top 75 mm of soil (Table 4.2). Greater cycling of nutrients through denser 
stands of poplar would have a compounding effect, and this occurred in the lower, less 
well buffered, 75-1 50 mm of soil at Hautope 1 (Figure 4.5a). In contrast, pHw levels within 
both soil strata at Kiwitea were higher amongst the poplar trees than in the open pasture. 
These site differences may have been caused by the dense subsoil (fragipan) at Hautope 1 
(refer to Section 2.3. 1 .2) forcing the poplar trees to absorb nutrients from shallower soil 
horizons, whereas, the more permeable soils at Kiwitea would have allowed deeper and 
more extensive root development. 
Uncoupling of the C cycle through the accumulation of SOM in Zone 1 at Hautope 1 
(Table 4.2) could also enhance soil acidification, if the excess inorganic cations over 
anions in the accumulated SOM were originally sourced in situ (Richter 1 986; Bolan et al. 
1 99 1 ;  de Klein et al. 1 997; Giddens et al. 1 997; Parfitt et al. 1 997; Tang et al. 1 999; 
Marschner & Noble 2000). 
Chapter 4 1 52 
Another pathway in which organic matter, rich in organic acid amons and inorganic 
cations, could have been added to the topsoil amongst the poplar trees was through the 
camping of grazing animals. Sibbald & Agnew ( 1 996) found that widely spaced trees acted 
as a focal point where sheep congregated, rested, defecated, and urinated to a greater extent 
than in the inter-canopy gap areas. If this dung and urine came from ingested open pasture 
herbage, which in turn was rich in organic acid anions and inorganic cations, then its 
decomposition and mineralisation in the soil under the trees would cause the net 
consumption ofW (or release ofOIT) (Wong & Swift 2003). This is because the counter­
balancing acidifying processes originally associated with the synthesis of this herbage, 
before it was ingested, occurred in the open pasture and not under the trees (refer to 
Section 4.5. 1 . 1 ). Animal dung containing a high concentration of Ca and Mg relative to 
inorganic anions is normally balanced with carbonate (Cot), giving it a pH in the range 
of 7.0-8.0 (Haynes & Williams 1993). Several researchers have measured significant 
increases in soil pH below dung patches (During et al. 1 973 ; Haynes & Williams 1 99 1 ;  
Whalen et al. 2000). Nguyen & Goh ( 1 992) found consistently higher soil pH levels in 
animal camp areas compared with non-camp areas in permanent pastures that had been 
grazed and fertilised for more than 1 5  years. 
Out of the two farm sites investigated, the potential for transferring alkalinity from the 
open pasture to amongst the trees in the form of dung deposits was greater at Hautope 1 .  
This was because the paddocks at this site had a greater mix of tree and large open pasture 
areas, whereas, at Kiwitea the poplar stands more homogenously covered the paddocks. 
The K concentration in soils has been used as an indicator of nutrient transfer by grazing 
animals (Hawke & Tombleson 1 993; Hawke & Gillingham 1996; Gillingham & Hawke 
1 997). At Hautope 1 ,  soil K was greater amongst the poplar stands, especially directly 
below the poplar crowns (Zone 1 ), compared to in the open pasture (Table 4.2). This 
indicated that at this site animal camping under the trees could have been a significant 
factor. Conversely, amongst the poplar stands at Kiwitea the concentration of soil K tended 
(P<0. 1 )  to be greater at the centre of the inter-canopy gap (Zone 3) in comparison to the 
open pasture (Table 4. 1 ). 
The translocation of exchangeable AI from upper to lower soil strata could also have 
contributed to the changes in soil pH at each site (Smith et al. 1995). AI contributes to soil 
acidity through the way in which it undergoes hydrolysis (depending on the pH of the soil 
Chapter 4 1 53 
solution) to produce W (Binkley & Richter 1 987; McBride 1 994). Many orgamc 
compounds (e.g. tannins, phenolics, other soluble organic acids/anions) released into the 
soil from tree leachates, root exudates, litter and its decomposition products can 
temporarily complex AI (Malcolm & McCracken 1968; Pohlman & McColl 1 988). In 
soluble form, these organic-AI complexes can be leached down the soil profile (Bloomfield 
1 954; DeLong & Schnitzer 1 955), where they are subsequently broken down by oxidation 
and microbial attack, or alternatively they become saturated with metals and precipitate out 
of solution (Davies 1 971 ;  David et al. 1 995; Fox 1995). However, at both sites the overall 
contribution of exchangeable AI to soil acidity was likely relatively small due to the low 
amount of sesquioxides present (as indicated by the low ASC) and relatively high soil pH 
(Noble et al. 1 996; Wong et al. 1 998, 2000). 
4.5.2 The effect of overstorey environment on major plant-available cations 
Overall, the concentrations of major plant available cations (K, Ca, Mg, & Na) in both 
Zones 1 and 3 were either similar to or greater than in the open pasture (Table 4. 1 & 4.2). 
Most cations were also positively, but weakly, related to CCL (Figure 4.2 & 4.4). 
However, the relative differences in cation concentration amongst the three main 
overstorey environments, and with increasing CCL, varied markedly between the two farm 
sites. For example, at Kiwitea a positive linear relationship between CCL and Mg in the 
upper 0-75mm of soil was found in both Zones 1 and 3, while for K it was restricted to 
Zone 3 (Figure 4.2c,d), whereas, the exact opposite occurred at Hautope 1 (Figure 4.4a,c). 
Guevara-Escobar et al. (2002) also measured similar or greater concentrations of major 
cations in the topsoil directly below young (5 year old) and mature (>25 year old) poplar 
trees when compared with adjacent open pastures. However, the present study shows that 
the trees influence is not solely restricted to within the crown domain, but can extend well 
into the inter-canopy gap area. 
Hawke & O'Connor (1 993) and Perrott et al. (1 999) found the concentration of 
exchangeable cations (K, Ca, Mg, & Na) in the top 0-75mm of soil decreased under older 
stands and greater tree stocking rates of P. radiata. Nevertheless, for New Zealand paired­
site studies the relative difference . between tree and adjacent open pasture/grassland areas 
has been variable in magnitude and direction (Maclaren 1 996; Giddens et al. 1 997; Parfitt 
et a. 1 997; Alfredsson et al. 1 998). 
Chapter 4 1 54 
Potassium (K) 
At both farm sites, plant available K in the upper 0-75mm of soil was higher amongst the 
poplar trees (Zones 1 & 3) than in the open pasture, which supports the findings of 
Guevara-Escobar et al. (2002) under mature poplar trees, but not young trees, in the same 
region as Kiwitea. High rates of nutrient return in litterfall, throughfall, and stemflow, 
along with nutrient transfer by grazing animals and reduced leaching all likely contnbuted 
to the elevated levels of K in the topsoil amongst the trees (Adams & Boyle 1 979; Pastor & 
Bockheim 1 984; Sibbald & Agnew 1 996; Parfitt et al. 1 997). In forest ecosystems K is 
largely cycled in throughfall rather than litterfall (Carlisle et a1. 1 966; Parker 1 983; Swank 
1 986, cited Blair 1 988). Nevertheless, Guevara-Escobar ( 1 999) found that adding senesced 
poplar leaves to an open pasture soil increased the level of exchangeable K 4-fold after two 
months of incubation. Little K, Mg, or Ca is stored in poplar tissue over winter (Baker & 
Blackmon 1 977). The higher soil pH amongst the poplar trees would also enhance K 
adsorption through its positive effect on the soil's cation exchange capacity (CEC; Haynes 
& Williams 1 993). 
In comparing the two sites for levels of K there was no consistent pattern between the two 
overstorey environments amongst the trees. As noted above, at Hautope 1 plant available K 
in the upper 0-75mm of soil was significantly greater in Zone 1 than Zone 3 ,  which was the 
reverse of what was found at Kiwitea. In Hautope 1 's more extreme summer-dry climate 
grazing animals seeking shade directly below the poplar crowns (Zone 1 )  may have 
transferred K in dung and urine above that required by the trees and understorey pasture or 
lost through soil leaching. Furthermore, the higher SOM content in Zone 1 (Table 4.2) 
would have increased the soil's CEC and thus its ability to retain K (Williams 1 980). 
In contrast, at Kiwitea under a milder and less heterogeneous micro climate created by the 
higher proportion of tree cover, the K returned to the soil in dung and urine may have been 
more evenly distributed over the entire paddock. A higher concentration of K in Zone 3 
could then have resulted from an overall lower demand from the pasture in comparison to 
the combined demand from the trees and understorey pasture. Alternatively, grazing 
animals may have actually preferred the limited canopy-gap areas and thus returned a 
greater amount of dung and urine to this overstorey environment. 
Chapter 4 1 55 
Sodium (Na) 
At Hautope 1 ,  the higher concentration of plant available Na in Zone 1 ,  followed by 
Zone 3, and then the open pasture was probably influenced by the same factors affecting 
the relative concentrations of K. Additional Na from outside of the tree-pasture system 
would also have been added to Zone 1 by the large surface area of the trees catching sea 
salt (parker 1983 ; Parfitt et al. 1997). Dry deposition of Na in sea salt would also have 
occurred at Kiwitea. However, it is unclear why the concentrations of plant available Na at 
this site did not vary significantly between the three main overstorey environments, or 
change with increasing CCL. The elevated concentration of other major cations (e.g. Ca, 
Mg, & K) in the soil amongst the trees may have displaced or prevented Na from being 
adsorbed onto the more limited cation exchange sites (Fisher & Binkley 2000), with the Na 
in soil solution leaching from the free draining soil under high rainfall (refer to Section 
2.3). 
Calcium (Ca) 
At Kiwitea several factors likely caused the increase in plant available Ca in the upper 
0-75mm of soil under greater CCL (Figure 4.2b). In comparison to the open pasture, more 
Ca would have been added to the soil surface through the Ca rich poplar leaf litter. As 
previously discussed in Section 4.5 . 1.2, the Ca concentration in poplar leaves at Kiwitea in 
summer was 4-times greater than in the open pasture herbage, with little re-translocation or 
leaching of this cation likely occurring from tree leaves before leaf fall (parker 1983 ; 
Pastor & Bockheim 1984; Potter et al. 1991). Assuming this site had a net primary 
production (NPP) capacity of 13 t DMlhalyr, with poplar leaves making up 3 . 1  t DMlha/yr 
in the tree-pasture system (Guevara-Escobar 1999), it was estimated that annually twice as 
much Ca was added to the soil under the trees relative to the open pasture (Appendix 4.2). 
However, for enrichment to occur, the Ca mineralised from leaf litter needed to originally 
have come from some other soil stratum (Alban 1982; Parfitt et al. 1997). Under increasing 
CCL, tree leaf litter makes up a larger proportion of the total organic matter cycled in the 
soil, while less Ca would be lost from the system through nutrient leaching or the grazing 
animals. The higher soil pH amongst the trees (Table 4. 1 ;  Figure 4.2a) would also increase 
the soil's ability to retain Ca, but this may have been negated to some extent by the lower 
SOM content (Table 4. 1;  Figure 4 .3b). The absorption of soil Ca by trees is usually high 
(Fisher & Binkley 2000). This could explain why the concentration of Ca in Zone 3 was 
Chapter 4 1 56 
only significantly greater than the open pasture, instead of both Zones 1 and 3 when 
averaged over the entire range ofCCLs (Table 4 . 1 ). 
At Hautope 1 ,  the similar level of plant available Ca in the upper 0-75 mm of soil across 
the three main overstorey environments may have been caused by the less marked 
differences in soil pH (Table 4.2), more restricted rooting zone caused by the dense 
subsoil, and overall lower leaching potential of the site (refer to Section 2.3). The restricted 
rooting zone would reduce the ability of the trees to absorb Ca from lower soil horizons 
and transfer it to the soil surface, while the lower leaching potential of the site would 
reduce the amount of Ca lost especially from the open pasture. In addition, considerable 
amounts of Ca may have been immobilised in the build up of SOM directly below the trees 
(Zone 1 ;  Table 4.2). Within the 75- 1 50 mm soil stratum at Hautope 1 ,  the lower levels of 
plant available Ca amongst the poplar trees (Zones 1 & 3) than in the open pasture was 
likely related to the lower soil pH, along with the trees being forced to take up Ca from this 
soil depth because of the dense subsoil (refer to Section 2.3. 1 .2). 
Magnesium (Mg) 
Similarly to Ca at Kiwitea, the increase in plant available Mg in the upper 0-75mm of soil 
under higher CCL at both farm sites (Figures 4.2c & 4.4c) was likely related to 
considerable Mg inputs from poplar leaf litter (parfitt et al. 1 997; Davis 1 998; Guevara­
Escobar 1 999; Sharma et al. 2001). Adams & Boyle (1 979) found that adding Populus 
grandidentata Michx. leaf leachate to the topsoil of a sandy Michigan spodosol enriched 
the latter with Mg (and also K). In marine climates, throughfall and stemflow transferred to 
the soil directly below trees (Zone 1 )  can also be enriched in Mg through the deposition 
of sea salt (parker 1983; Giddens et al. 1997; Parfitt et al. 1 997). This latter process would 
help counter the loss of Mg from the soil through tree uptake. Alternatively, at Hautope 1 
there may have been considerable transfer in dung (Hawke & Tombleson 1 993) and 
greater weathering of Mg-rich clays (Saunders 1 978) in the more restricted root zone 
(Alban 1 982). 
Chapter 4 1 57 
4.5.3 The effect of overstorey environment on major plant-avai lable anions 
Phosphate (P) 
The similar Olsen-P values among the three main overstorey environments at Kiwitea was 
somewhat surprising, given that the soil amongst the trees contained less SOM (Table 4. 1 ), 
which was also likely to be more recalcitrant (refer to Section 4.5 . 1 . 1 )  than in the open 
pasture. A lower SOM concentration would reduce the total size of the organic P pool, 
while the greater refractory nature of the SOM would reduce the rate at which the organic 
P is mineralised. Amongst the trees, the higher soil pHw (Table 4. 1 )  and also the release of 
appreciable quantities of organic acid anions (refer to Section 4.5 . 1 .2) could have increased 
the Olsen-P by reducing the ASC of the soil (Figure 4.3a,c); with the organic acid anions 
also directly competing for anion exchange sites (Saunders 1965; Roberts et al. 1996; 
Perrott et al. 1999). Furthermore, organic acid anions ( e.g. oxalate) can release inorganic P 
into the soil solution through the dissolution of oxide surfaces (Fox & Comeford 1 992; 
Fox 1995). Prior to leaf fall, a high proportion of phosphorus is retranslocated out of poplar 
leaves back into perennial tissue (Baker & Blackmon 1 977; Pastor & Bockheim 1984). 
As a result, it is unlikely that poplar leaf litter is a major factor determining soil Olsen-P 
levels. 
Based on soil samples taken in spring under young (5 year old) and mature (> 25 year old) 
poplar trees, Guevara-Escobar et al. (2002) also did not find any significant difference in 
Olsen-P between tree and adjacent open pasture environments. In contrast, most New 
Zealand studies investigating the effect of afforestation with pines have found that trees 
increase plant available P in the soil (Davis & Lang 1 99 1 ;  Hawke & O'Connor 1 993; 
Condron et a1. 1996; Chen et al. 2000; Cossen & Hawke 2000; Chen et al. 2003). This has 
been partly attributed to the enhanced mineralisation of organic P previously built up under 
pasture/grassland (Condron et al. 1996; Chen et al. 2000). Competition for adsorption sites 
by organic acid anions produced by the trees, desorption or dissolution of organic species, 
and the breakdown of soil aggregates exposing previously protected SOM may also be 
involved (Perrott et al. 1999). 
All of the previously discussed mechanisms operating under poplar trees at Kiwitea would 
also likely be present at Hautope 1 .  However, at this site there was greater SOM in Zone 1 
(Table 4.2), increasing the organic P pool available for mineralisation, and also there was 
Chapter 4 1 58 
greater probability of nutrient transfer occurring by the grazing animals (refer to Section 
4 .5 . 1 .2). This may have resulted in the elevated Olsen-P level in Zone 1 compared to 
Zone 3 and the open pasture (Table 4.2). Giddens et al. ( 1 997) also attributed greater 
Olsen-P levels under P. radiata stands to animal transfer at 'paired sites' where grazing 
animals had free access to both environments. 
In the lower 75-1 50mm of soil at Hautope 1 ,  the increase in Olsen-P with greater CCL was 
probably due to the lower soil pHw (Table 4.2; Figure 4.5a), increasing the positive charge 
on soil colloids, and a greater concentration of sesquioxides (Section 4.5 . 1 .2). Both 
mechanisms would lead to a higher ASC, which at this soil depth was also positively 
related to CCL (Figure 4.5b). 
5ulfate (504-5) 
Given that >90% of sulfur (S) in most topsoils is tied up in organic matter (McLaren & 
Cameron 1 996; Fisher & Binkley 2000), it was surprising that at Kiwitea the immediately 
available S04-S levels in Zones 1 and 3 were marginally, but significantly, higher than in 
the adjacent open pasture (Table 4. 1 ). At both farm sites in summer, the concentration of S 
in the tree leaves was around 2-fold greater than in the open pasture herbage (refer to 
Sections 5 .4.7 & 5.4.8). Therefore, the addition of poplar leaf litter to the topsoil in both 
Zones 1 and 3 could have been a major pathway for S04-S enrichment. Based on a 
glasshouse soil incubation experiment, Guevara-Escobar ( 1 999) reported that adding 
senesced poplar leaves (at a rate typical of a dense plantation) to an open pasture soil 
caused a 5-fold increase in immediately available S04-S after two months of incubation. In 
contrast, at Hautope 1 the concentration of immediately available S04-S in the upper 0-75 
mm of soil was markedly higher in the open pasture compared to amongst the poplar trees 
(Table 4.2). This may have been caused by an uneven rate of S04-S added to the two 
systems through recent fertiliser applications (refer to Section 2.3.2.2). 
Parfitt et al. ( 1997) also attributed a higher concentration of S04-S under P. radiata than in 
open pasture to the trees absorbing this nutrient from lower soil horizons and redistributing 
it in leaf litter onto the soil surface. In addition, a significant quantity of S from sea salt 
was also trapped by the tree crowns and transferred to the soil in throughfall (Parfitt et al. 
1997). Soil S04-S enrichment via the latter process would also be expected in Zone 1 at 
Chapter 4 1 59 
Kiwitea and Hautope 1 because of each of the sites relatively close proximity to the 
coastline. 
The lower net herbage accumulation (NHA) amongst the trees at both fann sites (refer to 
Section 5.4. 1)  would reduce S04-S taken up by pasture and lost through grazing animals 
(via product and excreta) (Saggar et al. 1 990). Furthermore, similar mechanisms described 
for the N cycle in Section 4.5. 1 . 1 ,  such as greater net immobilisation and less leaching at 
key times of the year may also play a role in reducing the amount of S04-8 lost from the 
tree-pasture system. 
4.5.4 Differences in soil organic carbon (SCC) between the overstorey 
environments 
The effect of poplar trees on the SOC concentration was also not consistent across both 
farm sites. For example, when averaged over the range of overstorey densities (CCLs) 
investigated at Kiwitea, there was weak evidence (P<0. 1)  that the open pasture had around 
1 % more SOC than amongst the poplar trees (Table 4. 1 ). A weak-to-moderate negative 
linear relationship between CCL and SOC also occurred especially for Zone 1 (Figure 
4.3b,d). This contrasted with Hautope 1 ,  where Zone 1 contained approximately 0.5% 
more SOC than the other two main overstorey environments (Table 4.2) and CCL did not 
significantly affect the SOC concentration. 
Several researchers have measured an increase in SOC (or associated SOM) when poplar 
trees are planted into previously cultivated (arable) agricultural land (Singh et al. 1 989; 
Hansen 1 993; Makeschin 1 994; Park et af. 1 994; Thevathasan & Gordon 1 997; Marquez et 
al. 1 999; Jha et af. 2000; Kaur et al. 2000; Tolbert et al. 2000; Saviozzi et al. 2001). 
Afforestation of permanent pasture/grassland is more variable, with the concentration of 
SOC usually being similar to or lower than in open pasture (Crowe 1 993; Fuller & 
Anderson 1 993; Saviozzi et al. 2001 ; Guevara-Escobar et al. 2002). Similar New Zealand 
results have also been found for afforestation of pasture/grassland with P. radiata and 
other conifer species (Giddens et al. 1 997; Parfitt et al. 1 997; Alfredsson et al. 1 998; 
Perrott et al. 1999; Chen et al. 2000, 2003;  Halliday et al. 2003). Davis & Condron (2002) 
concluded from an extensive review of these predominantly paired-site studies that, on 
average, conifers initially reduce the organic carbon concentration of pastoral soils; 
Chapter 4 1 60 
however, after about 20 years there is generally little difference between the two 
ecosystems. 
The organic carbon, or associated orgaruc matter, content of a soil is governed 
simultaneously by the amount of organic matter added and lost from the soil (Foth 1 978; 
Van Cleve & Powers 1 995; Post & Kwon 2000). Therefore, at the Kiwitea farm site there 
are two possible ways for the SOC (and associated SOM) content to be greater in the open 
pasture than amongst the poplar trees: either there has been a comparatively greater 
amount of organic matter added to the open pasture soil, relative to its decomposition; or a 
greater amount of organic matter has been lost amongst the poplars, relative to its addition. 
Without baseline data taken prior to the trees being planted it is difficult to establish which 
of these mechanisms was mainly responsible for the observed difference, especially given 
that it is uncertain whether the soils of the improved hill pasture had attained a 'steady­
state' in their carbon content. 
Using the annual net herbage accumulation of pasture legume from Section 5.4. 1 as a 
proxy for N2-fixation (Lambert 1 987; Guevara-Escobar 1 999), a larger amount of organic 
matter would likely have been cycled in the open pasture through overall greater net 
primary production (NPP) compared to amongst the poplar trees (Walker 1 956; Paustian et 
al. 1 992; Cole et al. 1995; Murata et al. 1 995). Given that the open pasture and poplar 
stands at Kiwitea were split into separate paddocks, little of the extra N or organic matter 
produced in the open pasture could have been transferred, via animal dung and urine, to 
among the poplar trees. 
Counteracting the higher biomass-C inputs in the open pasture is the likelihood that poplar 
litter is more recalcitrant (refer to Section 4.4.2 . 1 ) .  However, the increase in soil pH and 
potentially the greater release of organic acid anions (refer to Section 4.4.2) would reduce 
bonding between organic matter and inorganic colloids (e.g. clays), this increasing the 
formers exposure to microbial and biochemical decomposition processes (Curtin et al. 
1 998; Perrott et al. 1999). Alternatively, the difference in SOC content may simply be the 
effect of soil erosion of the site prior to the planting of poplars, with the soil still in the 
process of slowly accumulating organic carbon back to previous non-eroded levels. 
Davis & Condron (2002) postulated that large differences in SOC for many of the paired-
Chapter 4 1 61 
site studies they reviewed could also have been caused by initial differences in soil carbon 
stocks rather than effects of afforestation per se. 
Similarly, based on annual legume net herbage accumulation (refer to Section 5.4. 1 )  as a 
proxy for Nrfixation at Hautope 1 ,  amongst the poplar stands (Zones 1 & 3) would be 
expected to be more N limited than in the open pasture. However, unlike at Kiwitea, the 
poplar stands were intermixed with large open pasture areas in the same paddock, allowing 
greater opportunity for the transfer of N and organic matter to amongst the trees in the 
form of animal urine and dung (During et al. 1 973; Paustian et al. 1 992; Haynes & 
Williams 1 993 ; Haynes & Naidu 1 998; Metherell 2003). Also, in comparison to the open 
pasture, the trees ability to access soil water at depth in the summer-dry environment, 
along with higher levels of essential macro-nutrients, such as P and K, could have led to a 
greater NPP and thus litter-C inputs. 
4.5.5 The ecological significance of the soil nutrient status at each site in 
relation to pasture production 
The slightly-to-moderately acid soil pHw and the medium-to-high levels of plant-available 
P and K in the upper 0-75 mm of soil at Kiwitea (Table 4. 1)  were all close to the biological 
optimum4 for maximum pasture production on New Zealand sheep and beef farms, 
irrespective of the overstorey environment (Edmeades et a/. 1 984; Morton et al. 1 994). 
In contrast, the level of immediately available SO/--S at the time of sampling in spring 
was low for all three main overstorey environments and potentially could have limited 
pasture production (Morton et al. 1 994). 
Nevertheless, SO/--S deficiency throughout the rest of the year in the well-developed 
pasture was unlikely because of substantial inputs through: the addition of superphosphate 
fertiliser (refer to Section 2.3.2. 1 ), mineralisation of SOM, and to a lesser extent, from 
rainfall (wet) and dry deposition (Saunders et a/. 1 98 1 ;  Cornforth et al. 1 983; Ledgard & 
Upsdell 1 99 1 ). This was confirmed using Sinclair & Saunders ( 1984) mass-balance model, 
which accounts for such gains to the S cycle (Nguyen et al. 1 993). Based on this model, it 
was estimated that the application of superphosphate at a rate of 220 kglha/yr should have 
4 Soil test values that, on average for New Zealand sedimentary soils, are associated with 97% maximum 
pasture production (Morton et al. 1 994). 
Chapter 4 1 62 
provided the soil at Kiwitea with an extra 5.5 kg of S/ha above maintenance requirements 
(Appendix 4. 1 ). Some additional SO/--S may also have been absorbed by pasture plants 
from further down the soil profile (Gregg et al. 1 977), but was probably only a small 
amount given the very low levels of immediately available SO/--S measured within the 
75- 150 mm soil stratum (Guevara-Escobar 1999). 
In the upper 0-75 mm of soil at Hautope 1 ,  the moderately acid soil pHw and medium-to­
high levels of plant-available P, K, and SO/--S were close to the optimum for maximum 
pasture production on New Zealand sheep and beef farms (Edmeades et al. 1984; Morton 
et al. 1994). 
The higher pHw in the upper 0-75 mm of soil amongst the trees than in the open pasture at 
Kiwitea and Hautope 1 should have enhanced legume N2-fixation by increasing rhizobial 
nodulation (McKenzie et al. 1999) and molybdenum (Mo) availability (Sherrell & 
Metherell 1985; Wheeler & O'Connor 1998). However, as previously discussed for the C 
and N cycles in Section 4.5. 1 . 1, the large pulse of C-rich poplar residues (including 
secondary organic compounds such as tannins) added to the soil in autumn could 
potentially reduce the size of the inorganic N, S, and P pools available for the understorey 
pasture plants. Furthermore, in the following spring, mineralisation of this stored N could 
reduce the competitive advantage of legumes over grasses in the generally N deficient 
sites, resulting in depressed understorey legume growth and N2-fixation from late-spring 
through until autumn (Luscombe & Fletcher 1982). 
For New Zealand sedimentary soils the level of plant-available Ca and Na in the upper 
0-75 mm of soil at Kiwitea was generally medium-to-Iow and low, respectively, whereas 
the level of Mg was high (Cornforth 1980; Blakemore et al. 1987). Hautope 1 also 
contained high levels of plant-available Mg, but had medium, as opposed to medium-to­
low, levels of Ca and Na (Cornforth 1 980; Blakemore et al. 1 987). However, while these 
major soil cations play significant roles in the general chemistry, biology, and physical 
properties of soils, and are also important in relation to animal nutrition and health, the 
varying levels found in New Zealand soils seldom directly limit pasture production 
(During 1984; McLaren & Cameron 1996). 
Chapter 4 1 63 
4.6 Conclusion 
At Kiwitea and Hautope 1 the soil pHw and concentration of essential major cations (Ca, 
Mg, � & Na) and anions (P & 804-8) in the top 0-75 mm of soil amongst the poplar trees 
(Zone 1 & 3) were similar to, or greater than, in the adjacent open pasture. Nevertheless, 
based on the standard basic soil tests taken in spring, the soil fertility of both tree and open 
pasture environments was close to the biological optimum required for maximum pasture 
production on New Zealand sheep and beef farms. The only exception was 804-8 at 
Kiwitea, which at the time of measurement was low in all three main overstorey 
environments. Therefore, it appears that poplar trees in summer-wet and summer-dry 
regions of the North Island, New Zealand, do not limit understorey pasture production 
through their effects on soil chemical properties. 
80il pHw in the upper 0-75mm of soil was 0.2-0.7 units higher amongst the poplar trees 
than in adjacent open pasture. The increase in soil pHw depended on CCL, but was also 
strongly affected by the pHw buffer capacity of the soil. The influence of the trees on soil 
pHw extended across the inter-tree gap (Zones 3) and was not restricted only to directly 
below the tree crowns (Zone 1 ). Chemical reactions involved with the addition of poplar 
organic material to the soil (e.g. ligand exchange, oxidation, andlor protonation of 
synthesised organic acid anions, & release of inorganic cations) were probably the main 
causes of the increase in soil pHw. Lower HC03- and N03- leaching may also have reduced 
the rate of natural acidification in the soil amongst the trees (Zones 1 & 3). Conversely, 
greater transference of alkalinity away from the open pasture, in the fonn of animal 
products and excreta, may have occurred through thiS environment's overall higher 
grazing-animal carrying capacity. Comparison between the Kiwitea and Hautope 1 sites 
indicates that impact of poplars on soil pHw is affected by the structure of the soil and 
poplars can even cause acidification (via excess cation/anion uptake) in the rhizosphere 
just above impenneable soil horizons such as fragipans. 
In general, the concentration of major plant available cations in the top 0-75mm of soil 
increased under higher CCL, although there was considerable variation between sites and 
overstorey environments amongst the trees. Enrichment was likely caused by the 
redistribution of cations from lower soil horizons through poplar leaf litter and leachates, 
an increase in the CEC of the soil associated with higher soil pHw, and lower soil water 
Chapter 4 164 
leaching. External sources Na and Mg would also be added through the trees catching sea­
salts. Variation in relationships between the two sites was probably caused by differences 
in SOM content, leaching potential, nutrient transfer by grazing animals, and soil structure 
(affecting tree rooting depth). 
Amongst the trees (Zones 1 & 3), the release of appreciable quantities of organic acid 
anions into the soil (via poplar leaf litter and leachates) and the higher soil pHw likely 
increased Olsen-P values. Site differences were then determined by the variation in SOM 
content between the three main overstorey environments. The level of S04-S in the soil 
amongst the trees was probably enhanced directly from the tree litter itself and also from 
sea-salt deposition, while less S04-S would have been lost through the grazing animals and 
soil water leaching. Similarly to cations, differences in animal grazing behaviour between 
the sites may have been a confounding factor for both Olsen-P and S04-S. 
The poplar trees had contrasting effects on the levels of SOC at Kiwitea and Hautope 1 .  
Without baseline data taken prior to tree planting, it is difficult to speculate what the 
probable causes were for the site differences. Variation in N availability may have been a 
significant factor through its affect on NPP and thus C inputs. 
Amongst the trees (Zones 1 & 3), the addition of large quantities poplar litter and leachate 
to the soil (especially in autumn), reduced leaching via less soil water drainage, and a 
lower animal carrying capacity could potentially close/tighten the N, P, and S cycles. 
Further study of temporal changes (fluxes) in these major nutrients is needed, as changes in 
their concentrations could affect seasonal understorey pasture production, reduce fertiliser 
input costs, and also lower pollution of the wider environment. Some processes involved, 
such as the timing and quantity of leaf fall, could also be manipulated to improve 
synchronization between the supply of these soil nutrients and demand by understorey 
pasture plants. 
Chapter 4 1 65 
4.7 References 
Alban. D.H. ( 1 982) Effects of nutrient accumulation by aspen, spruce, and pine on soil 
properties. Soil Science America Journal 46: 853-86 1 .  
Alfredsson, H. ; Condron, L.M.; Clarholm, M; Davis, M.R ( 1 998) Changes in soil acidity 
and organic matter following the establishment of conifers on fonner grassland in 
New Zealand. Forest Ecology and Management 1 12: 245-252. 
Azhar, E.S.; Vandenabeele, 1; Verstraete, W. (1 986) Nitrification and organic nitrogen 
formation in soils. Plant and Soil 94: 383-399. 
Baker, lB.; Blackmon, B.G. (1 977) Biomass and nutrient accumulation in a cottonwood 
plantation - the first growing season. Soil Science Society of America Journal 41 :  632-636. 
Baldwin, I.T.; Olson, RK.; Reiners, W.A. ( 1983) Protein binding phenolics and the 
inhibition of nitrification in subalpine balsam fir soils. Soil Biology and Biochemistry 
1 5(4): 41 9-423. 
Ball, P.R; Ryden, lC. ( 1984) Nitrogen relationships in intensively managed temperate 
grasslands. Plant and Soil 76: 23-33. 
Benoit, RE.; Starkey, RL. (1968) Enzyme inactivation as a factor in the inhibition of 
decomposition of  organic matter by tannins. Soil Science 1 05(4): 203-208. 
Benoit, RE.; Starkey, RL.; Basaraba, J. ( 1 968) Effect of purified plant tannin on 
decomposition of some organic compounds and plant materials. Soil Science 1 05(3) :  1 53-
1 58. 
Berger, T.W.; Kollensperger, G.; Wimmer, R (2004) Plant-soil feedback in spruce (Picea 
abies) and mixed spruce-beech (Fagus sylvatica) stands as indicated by dendrochemistry. 
Plant and Soil 264(1 12): 69-83. 
Chapter 4 1 66 
Berthelot, A.; Ranger, 1.; Gelhaye, D. (2000) Nutrient uptake and immobilisation in a 
short-rotation coppice stand of hybrid poplars in north-west France. Forest Ecology and 
Management 128 :  1 67-179. 
Binkley, D.; Richter, D. ( 1 987) Nutrient Cycles and WBudgets of Forest Ecosystems. 
Advances in Ecological Research 16: 2-5 1 .  
Blakemore, L.C.; Searle, P.L.; Daly. B.K. ( 1 987) Methods for chemical analysis of soils. 
NZ Soil Bureau Scientific Report 80. Department of Scientific and Industrial Research, 
Lower Hutt, New Zealand. 
Bloomfield, C. ( 1954) A study of podzolization. V. The mobilization of iron and 
aluminium by aspen and ash leaves. Journal of Soil Science 5(1):  50-56. 
Bolan, N.S.; Hedley, M.I. ; White, RE. ( 1991)  Processes of soil acidification during 
nitrogen cycling with emphasis on legume based pastures. Plant and Soil 1 34 :  53-63. 
Bolan, N.S.; Adriano, D.C.; Curtin, D. (2003) Soil acidification and liming interactions 
with nutrient and heavy metal transformations and bioavailability. Advances in Agronomy 
78: 21 5-272. 
Braschkat, 1.1.; Randall, P.J. (2004) Excess cation concentrations in shoots and roots 
of pasture species of importance in south-eastern Australia. Australian Journal of 
Experimental Agriculture 44(9): 883-892. 
Cappellato, R; Peters, N.B.; Ragsdale, H.L. ( 1993) Acidic atmospheric deposition and 
canopy interactions of adjacent deciduous and coniferous forests in the Georgia Piedmont. 
Canadian Journal of Forest Research 23 (6): 1 1 14- 1 124. 
Chen, C.R; Condron, L.M.; Davis, M.R; Sherlock, RR (2000) Effects of afforestation on 
phosphorus dynamics and biological properties in a New Zealand grassland soil. Plant and 
Soil 220: 1 5 1 - 1 63 .  
Chapter 4 1 67 
Chen, C.R; Condron, L.M.; Davis, M.R;  Sherlock, RR (2003) Seasonal changes in soil 
phosphorus and associated microbial properties under adjacent grassland and forest in 
New Zealand. Forest Ecology and Management 1 77(1/3): 539-557. 
Cole, D.W.; Compton, J.E. ;  Homann, P.S.; Edmonds, R.L. ;  Van Miegroet, H. ( 1995) 
Comparison of carbon accumulation in Douglas Fir and Red Alder forests. Carbon Forms 
and Functions in Forest Soils (eds McFee, W.W. ; Kelly, lM.), pp 527-546. Soil Science 
Society of America, Inc., Wisconsin, USA. 
Condron, L.M.; Davis, M.R; Newman, R.H.; Cornforth, I .S.  (1 996) Influence of conifers 
on the forms of phosphorus in selected New Zealand grassland soils. Biology and Fertility 
of Soils 21 : 37-42. 
Cornforth, I.S. ( 1980) Soils and fertilisers: Soil analysis - interpretation. Aglink FPP 556. 
Ministry of Agriculture and Fisheries, Wellington, New Zealand. 
Cornforth, I .S.;  Smith, G.S.; Wheeler, D.M. ( 1983) Soil testing and plant analysis for 
sulphur. Proceedings of the technical workshop on sulphur in New Zealand agriculture 
(eds Gregg, P.E.H; Syers, J.K.), pp 49-60. Occasional Report No. 5 .  Department of Soil 
Science, Massey University, Palmerston North, New Zealand. 
Cossens, G.G.; Hawke, M.F. (2000) Agroforestry in Eastern Otago: results from two long­
term experiments. Proceedings of the New Zealand Grassland Association 62: 93-98. 
Crowe, S.R (1993) The response of Lolium perenne and Holcus lanatus to shading in 
relation to a silvopastoral agroforestry system. PhD thesis, Queen's University of Belfast. 
Curtin, D. ; Rostad, H.P.W. (1 997) Cation exchange and buffer potential of Saskatchewan 
soils estimated from texture, organic matter and pH. Canadian Journal of Soil Science 
77(4): 621 -626. 
Curtin, D.; Campbell, c.A.; Jalil, A. (1998) Effects of acidity on mineralisation: pH­
dependence of organic matter mineralisation in weakly acidic soils. Soil Biology and 
Biochemistry 30(1 ): 57-64. 
Chapter 4 1 68 
David, M.B.; Vance, G.F. ; Krzyszowska, AJ. (1 995) Carbon controls on Spodosol 
nitrogen, sulfur, and phosphorus cycling. Carbon Forms and Functions in Forest Soils (eds 
McFee, W.W.; KelIy, 1.M.), pp 329-354. Soil Science Society of America, Inc., Wisconsin, 
USA 
Davies, RI. ( 1 971 )  Relation of polyphenols to decomposition of organic matter and to 
pedogenetic processes. Soil Science 1 1 1 (1) :  80-85. 
Davis, M.R; Lang, M.H. ( 1991 )  Increased nutrient availability in topsoils under conifers in 
the South Island high country. New Zealand Journal of Forestry Science 2 1 : 165-1 79. 
Davis, M.R; Condron, L.M. (2002). Impact of grassland afforestation on soil carbon in 
New Zealand: a review of paired-site studies. Australian Journal of Soil Research 40: 675-
690. 
de Klein, C.A; Monaghan, RM.; Sinclair, AG. ( 1 997). Soil acidification: a provisional 
model for New Zealand pastoral systems. New Zealand Journal of Agricultural Research 
40: 541 -557. 
DeLong, W.A.; Schnitzer, M. ( 1955) Investigations on the mobilization and transport of 
iron in forested soils: I. The capacities of leaf extracts and leachates to react with iron. Soil 
Science Society of America Journal 1 9: 360-363 . 
Dodd, M.B.; Orr, S.1. ; Jackson, B.L.J. ( 1 992) Soil aluminium, phosphate, and moisture 
interactions affecting the growth of four white clover (Trifolium repens L.) lines. New 
Zealand Journal of Agricultural Research 35 :  4 1 1 -422. 
Douglas, G.B.; Walcroft, AS.; Wills, B.J.; Hurst, S.E.; Foote, AG.; Trainor, K.D.; Fung, 
L.E. (2001 )  Resident pasture growth and the micro-environment beneath young, widely­
spaced poplars in New Zealand. Proceedings of the New Zealand Grassland Association 
63 : 1 3 1 - 1 38. 
Driebe, E.M. ; Whitham, T.G. (2000) Cottonwood hybridisation affects tannin and nitrogen 
content of leaf litter and alters decomposition. Oecologia 123 :  99- 1 07. 
Chapter 4 1 69 
During, C. ( 1 984) Fertilisers and soils in New Zealand farming. Third edition. 
Government Printer, Wellington, New Zealand. 
During, C.; Weeda, W.C.; Dorogaeff, F.D. ( 1973) Some effects of cattle dung on soil 
properties, pasture production, and nutrient uptake. IT. Influence of dung and fertilisers on 
sulphate sorption, pH, cation-exchange capacity and the potassium, magnesium, calcium, 
and nitrogen economy. New Zealand Journal of Agricultural Research 1 6: 1 3 1 - 1 38.  
Edmeades, D.C. (1 986) Soils and Fertilisers. Lime: effects on soils, pasture, animals. 
Aglink FPP 567. Ministry of Agriculture and Fisheries, Wellington, New Zealand. 
Edmeades, D.E. ; Pringle, R.M.; Shannon, P.W.; Mansell, G.P. ( 1 984) Effects of lime on 
pasture production on soils in the North Island of New Zealand 4. Predicting lime 
responses. New Zealand Journal of Agricultural Research 27: 371 -382. 
Field, T.R.O.; Ball, P.R.; Theobald, P.W. ( 1985) Leaching of nitrate from sheep-grazed 
pastures. Proceedings of the New Zealand Grassland Association 46: 209-2 14. 
Field, T.R.O.; Theobald, P.W.; Ball, P.R.; Clothier, B.E. (1 985) Leaching losses of nitrate 
from cattle urine applied to a lysimeter. Agronomy Society of New Zealand 1 5 : 137- 14 1 .  
Fierer, N.; SchimeL lP.; Cates, R.G.; Zou J. (2001) Influence of balsam poplar tannin 
fractions on carbon and nitrogen dynamics in Alaskan taiga floodplain soils. Soil Biology 
and Biochemistry 33 :  1 827- 1 839. 
Fisher, R.F.; Binkley, D. (2000) Ecology and management of forest soils. Third edition. 
John Wiley & Sons, Inc., New York, USA. 
Foth, H.D. (1 978) Fundamentals of Soil Science. Sixth edition. John Wiley & Sons, Inc. 
Fox, R.H.; Myers, R.J.K.; Vallis, I. ( 1990) The nitrogen mineralization rate of legume 
residues in soil as influenced by their polyphenoL lignin, and nitrogen contents. Plant and 
Soil 129: 25 1 -259. 
Chapter 4 1 70 
Fox, T.R (1995) The influence of low-molecular-weight organic acids on properties and 
processes in forest soils. Carbon Forms and Functions in Forest Soils (eds McFee, W.W.; 
Kelly, J.M.), pp 43-62. Soil Science Society of America, Inc., Wisconsin, USA. 
Fox, T.R; Comerford, N.B.; McFee, W.W. (1 990) Phosphorus and aluminium release 
from a spodic horizon mediated by organic acids. Soil Science Society of America Journal 
54: 1 763- 1 767. 
Fox, T.R; Comerford, N.B. (1 992) Influence of oxalate loading on phosphorus and 
aluminium solubility in spodosols. Soil Science Society of America Journal 56: 290-294. 
Fuller, L.G. ; Anderson, D.W. ( 1993) Changes in soil properties following forest invasion 
of Black soils of the Aspen Parkland. Canadian Journal of Soil Science 73: 6 1 3-627. 
Giddens, .M.; Parfitt, RL.; Percival, H.J. ( 1 997) Comparison of some soil properties under 
Pinus radiata and improved pasture. New Zealand Journal of Agricultural Research 40: 
409-416. 
Gillingharn, A.G. (1 980) Phosphorus uptake and return in grazed, steep hill pastures. I. 
Pasture production and dung and litter accumulation. New Zealand Journal of Agricultural 
Research 23 : 3 1 3-32 1 .  
Gillingham, A.G.; During, C .  (1973) Pasture production and transfer of fertility within a 
long established hill pasture. New Zealand Journal of Experimental Agriculture 1 :  227-
232. 
Gillingharn, A.G.; Hawke, M.F. ( 1997) The effects of shelterbelts on adjacent pastures and 
soils in a temperate climate. Proceedings of the XVIII International Grassland Congress 6: 
5-6. 
Gregg, P.E.H.; Goh, K.M.; Brash, D.W. ( 1 977) Isotopic studies on the uptake of sulphur 
by pasture plants. II. Uptake from various soil depths at several field sites. New Zealand 
Journal of Agricultural Research 20: 229-233. 
Chapter 4 1 71 
Guevara-Escobar, A. ( 1 999) Aspects of a poplar-pasture system related to pasture 
production in New Zealand. PhD thesis, Massey University, New Zealand. 
Guevara-Escobar, A.; Edwards, W.RN.; Morton, RH. ; Kemp, P.D. ; Mackay, AD. (2000) 
Tree water use and rainfall partitioning in a mature poplar-pasture system. Tree Physiology 
20: 97- 1 06. 
Guevara-Escobar, A; Kemp, P.D.; Hodgson, J.; Mackay, AD.; Edwards, W.RN. ( 1 997) 
Case study of a mature Populus deltoides-pasture system in a hill environment. 
Proceedings of the New Zealand Grassland Association 59: 1 79- 1 85. 
Guevara-Escobar, A; Kemp, P.D.; Mackay, AD.; Hodgson, J. (2002) Soil properties of a 
widely spaced, planted poplar (Populus deltoides)-pasture system in a hill environment. 
Australian Journal of Soil Research 40: 873-886. 
Halliday, J.C. ;  Tate, K.R ;  McMurtrie, RE.; Scott, N.A (2003) Mechanisms for changes in 
soil carbon storage with pasture to Pin us radiata land-use change. Global Change Biology 
9(9): 1294- 1 308. 
Hansen, E. ( 1 993) Soil carbon sequestration beneath hybrid poplar plantations in the North 
Central United States. Biomass and Bioenergy 5 :  43 1 -436. 
Hawke, M.F. ; Gillingham, AG. ( 1 996) Nutrient transfer by livestock adjacent to managed 
and unmanaged shelterbelts. New Zealand Tree Grower 1 6(1) :  3 5-37. 
Hawke, M.F.; O'Connor, M.B. ( 1 993) Soil pH and nutrient levels at Tikitere Agroforestry 
Research Area. New Zealand Journal of Forestry Science 23(1 ): 40-48. 
Hawke, M.F.; Tombleson, ID. ( 1 993) Production and interaction of pastures and 
shelterbelts in the central North Island. Proceedings of the New Zealand Grassland 
Association 5 5 :  1 93- 1 97.  
Haynes, RI; Williams, P.H. ( 1 993) Nutrient cycling and soil fertility in the grazed pasture 
ecosystem. Advances in Agronomy 49: 1 1 9- 1 99. 
Chapter 4 1 72 
Haynes, RJ. ; Naidu, R ( 1 998) Influence of lime, fertiliser and manure applications on soil 
organic matter content and soil physical conditions: A review. Nutrient Cycling in 
Agroecosystems 5 1 (2): 1 23-1 37. 
Haynes, RJ.; Mokolobate, M.S. (2001 )  Amelioration of AI toxicity and P deficiency in 
acid soils by additions of organic residues: a critical review of the phenomenon and the 
mechanisms involved. Nutrient Cycling in Agroecosystems 59: 47-63. 
Haynes, RJ.; Williams, P.H. ( 1991)  Effect of grazing animals on the nutrient status of 
pasture soils. Soil and plant testingfor nutrient deficiencies and toxicities (eds White, RE.; 
Currie, L.D.), pp 224-236. Occasional Report No. 5. Fertiliser and Lime Research Centre, 
Massey University, Palmerston North, New Zealand. 
Helyar, KR ( 1976) Nitrogen cycling and soil acidification. The Journal of the Australian 
Institute of Agricultural Science 42(4): 2 17-221 .  
Helyar, KR (1987) Acid soils update, Research information. NSW Agriculture and 
Fisheries, Australia. 
Helyar, KR; Porter, W.M. (1 989) Soil acidification, its measurement and the processes 
involved. Soil acidity and plant growth (ed Robson, A.D.), pp 61- 10 1 .  Academic Press, 
Sydney, Australia. 
Heng, L.K; White, RE.; Bolan, N.S . ;  Scotter, D.R ( 1991)  Leaching losses of major 
nutrients from a mole-drained soil under pasture. New Zealand Journal of Agricultural 
Research 34: 325-334. 
Herman, D.J. ; Halverson, L.J.; Firestone, M.K. (2003) Nitrogen dynamics in an annual 
grassland: oak canopy, climate, microbial population effects. Ecological Applications 
1 3(3): 593-604. 
Holland, P.T. ; During, C. (1 977) Movement of nitrate-N and transformations of urea-N 
under field conditions. New Zealand Journal of Agricultural Research 20: 479-488. 
Chapter 4 1 73 
Houle, D.; Ouimet, R; Paguin, R ;  Laflamme, J.G. (1 999) Interactions of atmospheric 
deposition with a mixed hardwood and a coniferous forest canopy at the Lake Clair 
Watershed (Duchesnay, Quebec). Canadian Journal of Forest Research 29(12) :  1 944-
1 957. 
Hoyt, P.B.; Turner, RC. (1 975) Effects of organic materials added to very acid soils on ' 
pH, aluminium, exchangeable:NI-4, and crop yields. Soil Science 1 19(3): 227-237. 
Hue, N.V.; Amien, I. (1 989) Aluminium detoxification with green manures. 
Communications in Soil Science and Plant Analysis 20: 1499- 1 5 1 1 .  
Hue, N.V.; Craddock, G.R.; Adams, F. ( 1 986) Effect of organic acids on aluminium 
toxicity in subsoils. Soil Science Society of America Journal 50: 28-34. 
Jackson, F,S. ;  Barry, T.N. (1 996) The extractable and bound condensed tannin content of 
leaves from tropical tree, shrub and forage legumes. Journal Science Food Agriculture 7 1 :  
1 03- 1 1 0. 
James, S.E.; Partel, M.; Wilson, S.D.; Peltzer, D.A (2003) Temporal heterogeneity of soil 
moisture in grassland and forest. Journal of Ecology 91 (2): 234-239. 
Jha, M.N.; Gupta, M.K. ; Dimri, B.M. (2000) Effect of agroforestry practices on soil 
properties. Annals of Forestry 8(2): 225-228. 
Kaur, B.; Gupta, S.R.; Singh, G. (2000) Soil carbon, microbial activity and nitrogen 
availability in agroforestry systems on moderately alkaline soils in northern India. Applied 
Soil Ecology 1 5(3): 283-294. 
Kemp, P.D.; Mackay, AD.; Matheson, L.A.; Timmins, M.E. (2001)  The forage value of 
poplars and willows. Proceedings of the New Zealand Grassland Association 63 : 1 15- 1 19. 
Kleinbaum, D.G. ; Kupper, L.L. ; Muller, K.E.; Nizam, A ( 1998) Applied regression 
analysis and other multivariable methods. Third edition. Duxbury Press. Brooks/Cole 
Publishing Company, California, USA 
Chapter 4 1 74 
Kochy, M.; Wilson, S.D. (1 997) Litter decomposition and nitrogen dynamics in aspen 
forest and mixed-grass prairie. Ecology 78(3): 732-739. 
Kowalenko, C.G.; Ivarson, K C . ;  Cameron, D.R (1978) Effect of moisture content, 
temperature and nitrogen fertilization on carbon dioxide evolution from field soils. Soil 
Biology and Biochemistry 10:  41 7-423 .  
Lambert, M.G. ( 1 987) Nitrogen fixation during improvement of North Island hill country 
pastures. New Zealand Journal of Experimental Agriculture 15 :  267-270. 
Ledgard, S.F.; Upsdell, M.P. ( 1991 )  Sulphur inputs from rainfall throughout New Zealand. 
New Zealand Journal of Agricultural Research 34: 105-1 1 1 . 
Lee, R ;  Cornforth, L.S. ;  Edmeades, D.C.; Watkinson, l.H. ( 199 1 )  A comparative study of 
MAF (Ruakura) and DSIR Land Resources soil analysis results. New Zealand Journal of 
Agricultural Research 34: 227-233.  
Lewis, lA. ; Starkey, RL. ( 1 968) Vegetable tannins, their decomposition and effects on 
decomposition of some organic compounds. Soil Science 1 06(4): 241-247. 
Littell, RC.; Freund, RJ.; Spector, P.c. ( 199 1 )  SASID System for linear models. Third 
edition. SAS Institute Inc., North Carolina, USA. 
Lodhi, M.A.K; Killingbeck, KT. (1 980) Allelopathic inhibition of nitrification and 
nitrifying bacteria in a ponderosa pine (Pinus Ponderosa Dougl.) community. American 
Journal of Botany 67(1 0): 1 423-1429. 
Lopez, l.F. (2000) Ecology of pastoral communities in a heterogeneous environment. PhD 
thesis, Massey University, New Zealand. 
Lousier, J.D.; Parkinson, D. ( 1 976) Litter decomposition in a cool temperate deciduous 
forest. Canadian Journal of Botany 54: 4 19-436. 
Chapter 4 1 75 
Luscombe, P.C. ; Fletcher, RH. ( 1 982) Nitrogen fertiliser on grazed hill pastures. 
Proceedings of the New Zealand Grassland Association 43 : 1 7 1 - 1 8 1 .  
Madaren, IP ( 1996) Environmental effects of planted forests in New Zealand. FRl bulletin 
1 98. Forest Research Institute, Rotorua, New Zealand. 
Makeschin, F. ( 1 994) Effects of energy forestry on soils. Biomass and Bioenergy 6 : 63-79. 
Malcolm, RL. ; McCracken, RI ( 1968) Canopy drip: a source of mobile soil organic 
matter for mobilization of iron and aluminium. Soil Science Society of America Journal 32: 
834-838.  
Marquez, C.O.; Cambardella, C.A.; Isenhart, T.M.; Schultz, R.C. (1999) Assessing soil 
quality in a riparian buffer by testing organic matter fractions in central Iowa, USA. 
Agroforestry Systems 44(2/3) :  1 33- 1 40. 
Marschner, B . ;  Noble, A.D. (2000) Chemical and biological processes leading to the 
neutralisation of acidity in soil incubated with litter materials. Soil Biology and 
Biochemistry 32(6): 805-8 13 .  
Marx, M. ;  Marschner, B.; Nelson, P.  (2002) Short-term effects of  incubated legume and 
grass materials on soil acidity and C and N mineralisation in a soil of north-east Australia. 
Australian Journal of Soil Research 40(7): 1 23 1 - 124 1 .  
McBride, M.B. ( 1994) Environmental chemistry of soils. Oxford University Press, New 
York, USA. 
McColl, J.G. ( 1 980) Seasonal nutrient variation in trembling aspen. Plant and Soil 54: 323-
328. 
McIvor, 1.,  Hurst, S.; Fung, L.; Van Den Dijssel, C.; Green, S.; Douglas, G.; Foote, L. 
(2003) Silvopastoral management of 'Veronese' poplars on an erosion-prone hillslope. 
Environmental management using soil-plant systems (eds Currie, L.D.; Stewart, R.B. ;  
Chapter 4 1 76 
Anderson, C.W.N.), pp 177-1 86. Occasional Report No. 16. Fertiliser and Lime Research 
Centre, Massey University, Palmerston North, New Zealand. 
McKenzie, B.A.; Kemp, P.D.;  Moot, D.I; Matthew, C.; Lucas, RI (1999) Environmental 
effects on plant growth and development. New Zealand Pasture and Crop Science 
(eds White, J.; Hodgson, 1), pp 29-44. Oxford University Press, Auckland, New Zealand. 
McLaren, RG.; Cameron, K.C. (1 996) Soil science: Sustainable production and 
environmental protection. Revised edition. Oxford University Press, Auckland, New 
Zealand. 
Metherell, A.K. (2003) Management effects on soil carbon storage in New Zealand 
pastures. Proceedings of the New Zealand Grassland Association 65: 259-264. 
Morton, J. ; Roberts, A.H.C.; Edmeades, D.C. (1 994) Fertiliser use on sheep and beef 
farms: the principles and practice of soil fertility and fertiliser use on sheep and beef 
farms. New Zealand Fertiliser Manufacturers Research Association and New Zealand 
Pastoral Agriculture Research Institute Ltd. 
Murphy, J.; Riley, IP. (1962) A modified single solution method for the determination of 
phosphate in natural waters. Analytica Chimica Acta 27: 3 1-36. 
Murata, T.; Nguyen, M.L.; Goh, K.M. ( 1 995) The effects of long-term superphosphate 
application on soil organic matter content and composition from an intensively managed 
New Zealand pasture. European Journal of Soil Science 46: 257-264. 
Myers, RH. ( 1 990) Classical and modern regression with applications. Second edition. 
PWS-Kent Publishing Company, Boston, USA. 
Nea� C. (2002) Interception and attenuation of atmospheric pollution in a lowland ash 
forested site, Old Pond Close, Northamptonshire, UK. Science of the Total Environment 
282/283 : 99- 1 1 9. 
Chapter 4 1 77 
Nelson, D.W.;  Sommers, L.E. ( 1 982) Organic carbon. Methods of soil analysis, Part 2 (ed 
Page, AL.). Second edition. 
Nguyen, M.L. ; Gob, K.M. ( 1993) An overview of the New Zealand sulphur cycling model 
for recommending sulphur requirements in grazed pastures. New Zealand Journal of 
Agricultural Research 36: 475-49 1 .  
Noble, A.D.;  Zenneck, 1. ;  Randall, P.J. (1 996) Leaf litter ash alkalinity and neutralisation 
of soil acidity. Plant and Soil 1 79: 93-302. 
Noble, A.D.; Randall, P.l ( 1999) Alkalinity effects of different tree litters incubated in an 
acid soil ofN.S.W., Australia. Agroforestry Systems 46(2): 147-160.  
Nwaigbo, L.C.; Miller, H.G.; Sibbald, AR; Hudson, G. (1 997) Soil nutrient redistribution 
about the tree in a silvopastoral system. Proceedings of the XVIII International Grassland 
Congress 6: 1 3- 14. 
Olsen, S.R; Cole, C.V. ;  Watanabe, F.S.;  Dean, L.A ( 1954) Estimation of available 
phosphorus in soils by extraction with sodium bicarbonate. United States Department of 
Agriculture Circular No. 939. U.S. Government Printing Office, Washington, USA 
Palm, C.A; Sanchez, P.A ( 1 991)  Nitrogen release from the leaves of some tropical 
legumes as affected by their lignin and polyphenolic contents. Soil Biology and 
Biochemistry 23(1) :  83-88. 
Parfitt, RL.; Percival, H.J.; Dahlgren, R.A; Hill, L.F. ( 1 997) Soil and solution chemistry 
under pasture and radiate pine in New Zealand. Plant and Soil 191 : 279-290. 
Park, l; Newman, S.M.; Cousins, S.H. (1 994) The effects of poplar (P. trichocarpa x 
deltoides) on soil biological properties in a silvoarable system. Agroforestry Systems 25:  
1 1 1 - 1 18 .  
Parker, G.G. ( 1 983) Throughfall and stemflow in the forest nutrient cycle. Advances in 
Ecological Research 1 3 :  58- 120. 
Chapter 4 1 78 
Pastor, J.; Bockheim, J.G. (1 984) Distribution and cycling of nutrients in an aspen-mixed­
hardwood-spodosol ecosystem in Northern WISCOnsin. Ecological Society of America 
65(2): 229-252. 
Paustian, K. ; Parton, W.J.; Persson, J. (1 993) Modeling soil organic matter in organic­
amended and nitrogen-fertilized long-term plots. Soil Science Society of America Journal 
56: 476-488. 
Perrott, K.W.; Ghani, A.; Q'Connor, M.B.; Waller, J.E. ( 1 999) Tree stocking effects on 
soil chemical and microbial properties at the Tikitere Agroforestry Research Area. New 
Zealand Journal of Forestry Science 29(1) :  1 1 6-130. 
Pohhnan, A.A.; McColl, J.G. ( 1 988) Soluble organics from forest litter and their role in 
metal dissolution. Soil Science Society of America Journal 52: 265-271 .  
Post, W.M.; Kwon, K.C. (2000) Soil carbon sequestration and land-use change: processes 
and potential. Global Change Biology 6: 3 1 7-327. 
Potter, C.S.; Ragsdale, H.L.; Swank, W.T. ( 1991 )  Atmospheric deposition and foliar 
leaching in a regenerating Southern Appalachian forest canopy. Journal of Ecology 79: 97-
1 1 5.  
Power, LL. ;  Dodd, M.B.; Thorrold, B.S. ( 1999) A comparison of pasture and soil moisture 
under Acacia melanoxylon and Eucalyptus nitens. Proceedings of the New Zealand 
Grassland Association 6 1 :  203-207. 
Power, LL.; Thorrold, B.S.; Balks, M.R. (2003) Soil properties and nitrogen availability in 
silvopastoral plantings of Acacia melanoxylon in North Island, New Zealand. Agroforestry 
Systems 57: 225-237. 
Richter, D.D. ( 1 986) Sources of Acidity in some Forested Udults. Soil Science Society of 
America Journal 50: 1 584 - 1 589. 
Chapter 4 1 79 
Richter, D.D.; Johnson, D.W. ;  Todd, D.E. ( 1 983) Atmospheric sulfur deposition, 
neutralisation, and ion leaching in deciduous forest ecosystems. Journal of Environmental 
Quality 1 2 :  263-270. 
Ridley, A.M.; Slattery, W.J. ; Helyar, KR. ; Cowling, A ( 1 990a) The importance of the 
carbon cycle to acidification of a grazed annual pasture. Australian Journal of 
Experimental Agriculture 30: 529-537. 
Ridley, A.M.; Slattery, W.J. ; Helyar, KR; Cowling, A ( 1 990b) Acidification under 
grazed annual and perennial grass based pastures. Australian Journal of Experimental 
Agriculture 30: 539-544. 
Ridley, A.M.; White, RE.; Helyar, KR.; Morrison, G.R; Heng, L.K; Fisher, R (200 1 )  
Nitrate leaching loss under annual and perennial pastures with and without lime on a 
duplex (texture contrast) soil in humid southeastern Australia. European Journal of Soil 
Science 52(2): 237-252. 
Roberts, AH.C.; Morton, J.D.; O'Connor, M.B.; Edmeades, D.C. (1 996) Building a solid 
foundation for pasture production in Northland: P, K, S and lime requirements. 
Proceedings of the New Zealand Grassland Association 57: 1 1 9- 1 25 .  
Saggar, S . ;  Hedley, C.B. ;  Salt, G.J. (200 1 )  Soil microbial biomass, metabolic quotient, and 
carbon and nitrogen mineralisation in 25-year-old Pinus radiata agroforestry regimes. 
Australian Journal of Soil Research 39: 49 1 -504. 
Saggar, S. ;  Mackay, AD.; Hedley, M.I; Lambert, M.G. ; Clark, D.A ( 1 990) A nutrient­
transfer model to explain the fate of phosphorus and sulphur in a grazed hill-country 
pasture. Agriculture, Ecosystems and Environment 30: 295-3 1 5 . 
SAS ( 1 990) SAS/STA� user 's guide, Version 6, Volume 1. Forth edition. SAS Institute 
Inc., North Carolina, USA 
Chapter 4 1 80 
Saunders, W.M.H. ( 1965) Phosphate retention by New Zealand soils and its relationship to 
free sesquioxides, organic matter, and other soil properties. New Zealand Journal 
Agriculture 8 :  30-57. 
Saunders, W.M.H. (1978) Soils and fertilisers: clay mineralogy and soil fertility. Aglink 
FPP 554. Ministry of Agriculture and Fisheries, Wellington, New Zealand. 
Saunders, W.M.H.; Cooper, D.M. ; Sinclair, AG. (1981)  Soils and fertilisers: Sulphur. 
Aglink FPP 649. Ministry of Agriculture and Fisheries, Wellington, New Zealand. 
Saviozzi, A; Levi-Minzi, R; Cardelli, R;  Riffaldi, R (2001 )  A comparison of soil quality 
in adjacent cultivated, forest and native grassland soils. Plant and Soil 233: 25 1 -259. 
Schimel, l.P.;  Van Cleve, K; Cates, RG.; Clausen, T.P.; Reichardt, P.B. ( 1 996) Effects of 
balsam poplar (Populus balsamifera) tannins and low molecular weight phenolics on 
microbial activity in taiga floodplain soil: implications for changes in N cycling during 
succession. Canadian Journal of Botany 74: 84-90. 
Scholefield, D.; Tyson, KC.; Garwood, E.A.; Armstrong, AC.; Hawkins, l. ; Stone, AC. 
( 1 993) Nitrate leaching from grazed grassland lysimeters: effects of fertilizer input, field 
drainage, age of sward and patterns of weather. Journal of Soil Science 44: 601 -6 13 .  
Schofield, P . ;  Mbugua, D.M.;  Pell, AN. (2001 )  Analysis of condensed tannins: a review. 
Animal Feed Science and Technology 9 1 :  2 1 -40. 
Sherrell, C. G.; Metherell, AK ( 1985) Soils and fertilisers: Molybdenum deficiency in 
pastures. Aglink FPP 63 1 .  Ministry of Agriculture and Fisheries, Wellington, New 
Zealand. 
Sherwood, M.; Fanning, A ( 1 989) Management systems to reduce the impact of nitrates 
(ed Germon, l.C.), pp32-42. Elsevier, London, UK. 
Chapter 4 1 81 
Sbibata, H.; Satoh, F. ; Tanaka, Y.; Sakuma, T. ( 1995) The role of organic horizons and 
canopy to modifY the chemistry of acidic deposition in some forest ecosystems. Water, Air, 
and Soil Pollution 85(3): 1 1 1 9- 1 124. 
Sibbald, AR. ;  Agnew, R. ( 1996) Silvopastoral National Network Experiment. Annual 
report 1 995. Agroforestry Forum 7(1) :  7- 10. 
Sinclair, AG. ( 1982) Soils and Fertilisers: Sulphur - sulphate leaching index. Aglink FPP 
652. Ministry of Agriculture and Fisheries, Wellington, New Zealand. 
Sinclair, AG. ( 1995) Modelling soil acidification in pastoral farming. Fertiliser 
requirements of grazed pasture and field crops: Macro- and micro-nutrients (eds Currie, 
L.D.; Loganathan, P.), pp 198-205. Occasional Report No. 8. Fertiliser and Lime Research 
Centre, Massey University, Palmerston North, New Zealand. 
Sinclair, AG.; Saunders, W.M.H. ( 1984) Sulphur. Fertiliser and lime recommendations 
for pastures and crops in New Zealand (eds Cornforth, 1.S.; Sinclair, A.G.), pp 1 5- 1 7. 
Second edition. Ministry of Agriculture and Fisheries, Wellington, New Zealand. 
Singh, B. ( 1998) Biomass production and nutrient dynamics in three clones of Populus 
deltoides planted on Indogangetic plains. Plant and Soil 203 : 15-26. 
Singh, K; Chauhan, H.S.; Rajput, D.K; Singh, D.V. ( 1 989) Report ofa 60 month study on 
litter production, changes in soil chemical properties and productivity under poplar 
(P. deltoides) and eucalyptus (E. hybrid) interplanted with aromatic grasses. Agroforestry 
Systems 9: 37-45. 
Smith, C.l; Goh, KM.; Bond, W.J.;  Freney, lR. (1 995) Effects of organic and inorganic 
calcium compounds on soil-solution pH and aluminium concentrations. European Journal 
of Soil Science 46: 53-63 . 
Steeie, KW.; Judd, M.J. ; Shannon, P.W. ( 1984) Leaching of nitrate and other nutrients 
from a grazed pasture. New Zealand Journal of Agricultural Research 27: 5-1 1 .  
Chapter 4 1 82 
Suckling, F.E.T. ( 1 975) Pasture management trials on unploughable hill country at 
Te Awa. ill. Results for 1 959-69. New Zealand Journal of Experimental Agriculture 3 :  
3 5 1 -436. 
Tang, C.; Sparling, G.P.;  McLay, C.D.A.; Raphael, C. ( 1 999) Effect of short-term legume 
residue decomposition on soil acidity. Australian Journal of Soil Research 37: 561 -673 . 
Tang, C.;  Rengel, Z. (2003) Role of plant cation/anion uptake ratio in soil acidification. 
Handbook of soil acidity (ed Rengel, Z.), pp 57-8 1 .  Marcel Dekker Inc., New York, USA 
Taylor, B.R; Parkinson, D. ( 1 988) Aspen and pine leaf litter decomposition in laboratory 
microcosms. IT. Interactions of temperature and moisture level. Canadian Journal of 
Botany 66: 1 966- 1 973. 
Thevathasan, N.V.;  Gorgon, A.M. ( 1 997) Poplar leaf biomass distribution and nitrogen 
dynamics in a poplar-barley intercropped system in southern Ontario, Canada. 
Agroforestry Systems 3 7: 79-90. 
Thibault, I.R ;  Fortin, I.A.; Smimoff, W.A ( 1 9 82) In vitro allelopathic inhibition of 
nitrification by balsam poplar and balsam fir. American Journal of Botany 69(5): 676-679. 
Tolbert, V.R; Thomton, F.C.; Ioslin, I.D.; Bock, B.R; Bandaranayake, W.; Houston, 
AB.; Tyler, D .D.;  Mays, D.A; Green, T.H.; Pettry, D.E. (2000) Increasing below-ground 
carbon sequestration with conversion of agricultural lands to production of bio-energy 
crops. New Zealand Journal of Forestry Science 30: 1 38 - 1 49. 
Van Breemen, N.; Mulder, 1; Driscoll, C.T. (1 983) Acidification and alkalinization of 
soils. Plant and Soil 75: 283-308. 
Van Cleve, K.; Powers, RF. ( 1 995) Soil carbon, soil formation, and ecosystem 
development. Carbon Forms and Functions in Forest Soils (eds McFee, W.W.; Kelly, 
I.M.), pp 1 55-200. Soil Science Society of America, Inc., Wisconsin, USA 
Chapter 4 1 83 
Van Goor, c.P. ( 1985) The impact of tree species on soil productivity. Netherlands 
Journal of Agricultural Science 33 :  1 33-140. 
Wall, A.J.; Mackay, A.D.; Kemp, P.D.; Gillingham, A.G.; Edwards, W.RN. ( 1 997) The 
impact of widely spaced soil conservation trees on hill pastoral systems. Proceedings of the 
New Zealand Grassland Association 59: 1 71 - 177. 
Wang, Y.; Douglas, G.B.; Waghorn, G.C. ; Barry, T.N.; Foote, A.G. ; Purchas, RW. (1 996) 
Effect of condensed tannins upon the performance of lambs grazing Lotus corniculatus and 
Lucerne (Medicago sativa). Journal of Agricultural Science, Cambridge 126:  87-98. 
Watanabe, F.A. ; Olsen, S.R (1 965) Soil Science Society of America Proceedings 29: 677 
Walker, T.W. (1 956) The nitrogen cycle in grassland soils. Journal Science Food 
Agriculture 7: 66-72. 
Watkinson, IH.; Kear, M.I ( 1 994) High performance ion chromatography measurement 
of sulphate in 20mM phosr'hate extracts of soil. Communications in Soil Science and Plant 
Analysis 25(7&8): 10 15- 1033.  
Whalen, IK.; Chang, C. ;  Clayton, G.W.; Carefoot, IP. (2000) Cattle manure amendments 
can increase the pH of acid soils. Soil Science Society of America Journal 64: 962-966. 
Wheeler, D.M. (1 997) Temporal changes in some soil chemical properties at four depths 
following the surface application of lime. New Zealand Journal of Agricultural Research 
40: 309-3 1 6. 
Wheeler, D.M. ; O'Connor, M.B. ( 1 998) Why do pastures respond to lime? Proceedings of 
the New Zealand Grassland Association 60: 57-6 1 .  
Wilkinson, A.G. (1 999) Poplars and willows for soil erosion control in New Zealand. 
Biomass and Bioenergy 1 6: 263-274. 
Chapter 4 1 84 
Williams, C.R. ( 1 980) Soil acidification under clover pasture. Australian Journal of 
Agriculture and Animal Husbandry 20: 5 6 1 -567. 
Wilson, S.D.; Kleb, H.R. (1 996) The influence of prairie and forest vegetation on soil 
moisture and available nitrogen American Midland Naturalist 1 36(2): 222-2 3 1 .  
Wong, M.T.F; Nortc1i:ff, S.;  Swift, R.S. ( 1 998) Method for determining the acid 
ameliorating capacity of plant residue compost, urban waste compost, farmyard manure, 
and peat applied to tropical soils. Communications in Soil Science and Plant Analysis 29: 
2927-2937. 
Wong, M.T.F.; Gibbs, P. ;  NortclifI, S . ;  Swift, R.S. (2000) Measurement of the acid 
neutralizing capacity of agroforestry tree prunings added to tropical soils. Journal of 
Agricultural Science, Cambridge 1 34: 269-276. 
Wong, M.T.F.; Swift, R.S. (2003) Role of organic matter in alleviating soil acidity. 
Handbook of soil acidity (ed Rengel, Z.), pp 337-358.  Marcel Dekker Inc., New York, 
USA. 
Chapter 5 
5 The effect of poplar overstorey density on 
understorey pasture production 
CONTENTS 
5.1 Introduction 
5.2 Methodology 
5.2. 1 
5 .2.2 
Treatments and sampling positions 
Measurements 
5.3 Data analyses 
5 . 3 . 1  
5.3.2 
5 .3 .3 
5.3 .4 
Net herbage accumulation and residual herbage mass 
Pasture composition, nutritive value, and mineral content 
Poplar leaves nutritive value and mineral content 
A check of underlying model assumptions 
5.4 Results 
5.4. 1 
5.4.2 
5.4.3 
5.4.4 
5.4.5 
5 .4.6 
5.4.7 
5.4.8 
Annual net herbage accumulation 
Seasonal net herbage accumulation 
Residual herbage mass below 2 5  mm trimming height 
Pasture botanical species composition 
Pasture nutritive values (proximate analysis) 
Poplar leaf nutritive values 
Pasture mineral content 
Poplar leaf mineral contents 
5.5 Discussion 
5.5 . 1  
5.5.2 
5.5.3 
5.5 .4 
5 .5 .5 
5.6 
5.7 
Annual net herbage accumulation (ANHA) 
Seasonal net herbage accumulation (NHA) 
Residual herbage mass (RHM) below 25 mm trimming height 
Botanical composition 
Pasture and poplar leaf feed values 
Conclusion 
References 
1 85 
186 
187 
1 87 
1 88 
192 
1 92 
1 92 
1 93 
1 93 
194 
1 94 
1 98 
206 
2 1 3  
227 
240 
244 
26 1 
266 
266 
269 
271 
274 
278 
279 
280 
Chapter 5 1 86 
5.1 I ntroduction 
The effect of poplar trees on annual pasture dry matter production, relative to adjacent 
open pasture, has been shown to range from a negligible decrease under young isolated 
trees (Gilchrist et al. 1993) to a 40-50% decrease under mature trees with high inter-crown 
interference (Crowe 1993; Guevara-Escobar et al. 1 997). Similarly, in stands with low 
inter-crown interference poplars have little impact on the nutritive value of understorey 
pastures (Douglas et al. 2005). However, below mature stands with high inter-crown 
interference Guevara-Escobar et al. (1 997) found the understorey pasture had a slightly 
lower nutritive value for grazing animals compared to an adjacent open pasture. 
In New Zealand, pasture production measurements around poplars have been restricted to 
within close proximity of individual tree crowns (Gilchrist et al. 1 993 ; Guevara-Escobar et 
al. 1 997; Douglas et al. 2001 ,  2005). Although a tree's influence on understorey pasture 
decreases with greater distance from the base of its stem (Clements et al. 1 988; Gilchrist et 
al. 1 993) this relationship also depends on the size and proximity of the surrounding trees 
(Sibbald et al. 1 99 1). Lack of information about how pasture production within the inter­
tree gap area is affected over a range of stand densities is inhibiting the development of a 
predictive stand level model, which could be used by farmers and resource managers to 
gauge the impact of poplars on farm pastoral productivity. This information would also 
help to develop management strategies that minimise any negative effects that these trees 
might have when integrated into hill pasture systems. 
In Chapter 3 canopy closure (CC) measured with a standard digital camera was shown to 
be a very good index of stand density and accounted for both the size and proximity of 
trees within a stand. As such, this index is used in this chapter to characterise the density of 
the poplar stands and is used to relate how changes in stand density affects understorey 
pasture productivity both directly below and in the gap between trees. 
Chapter 5 
5.2 Methodology 
5.2.1 Treatments and sampling positions 
1 87 
Similarly to Chapters 3 and 4, the broad range of overstorey (stand) densities previously 
developed in Chapter 2 formed the main treatment structure of this study. At all three farm 
sites, sampling amongst the poplar stands was restricted to within two main overstorey 
environments (or zones of tree influence): Zone 1 ,  area directly below the vertical 
projection of the tree crown on the north-eastern corner of an experimental unit; and 
Zone 3, area at the centre of the vertically projected gap between the crowns of the four 
nuclei trees (Figure 5. 1) .  To evaluate how representative the above sampling method was 
for an entire experimental unit, sampling was intensified at Kiwitea for two stand densities 
(50% & 70% CCL, each replicated twice). The total number of sampling zones was 
increased to five, spaced out along a linear transect running diagonally between two of the 
'nuclei' trees defining an experimental unit (Figure 5 . 1) .  
N 
Figure 5.1 Intensified sampling zones used for a limited number of experimental units at Kiwitea: 
Zones 1 & 5 - directly below the individual tree crowns on the northeast and southwest corners of an 
experimental unit, respectively; Zone 3 - at the centre of the vertically projected canopy gap between the 
trees; and Zones 2 & 4 - transitional area midway between the centre of the vertically projected canopy 
gap and its respective northeast and southwest edge. 
Chapter 5 1 88 
5.2.2 Measurements 
5.2.2.1 Net herbage accumulation 
Pasture net herbage accumulation (NHA) was measured on a dry weight basis using the 
standard double-trim technique (Radcliffe 1 974; Lucas & Thompson 1 994). Measurements 
were taken over two years at Kiwitea (October 1 998 - September 2000), and for one year 
at Hautope 1 and 2 (May 1 999 - April 2000). Under 0.7m2 grazing exclusion cages 
pastures were trimmed with a portable shearing hand-piece to a uniform height of 25mm 
above ground. After a month, all re-growth above the pre-trimmed height was harvested 
from inside a 0. lm2 metal quadrat and taken to the laboratory for processing. Following 
each harvest, the grazing exclusion cages were moved to a new location that was raked free 
of tree debris and trimmed. Each exclusion cage did not return to the same area for at least 
three months. Placement of cages was restricted to a medium hill-slope ( 1 3 -25°) category 
(Saggar et al. 1 990; L6pez 2000). Harvests were missed in some winter and late 
summer/autumn months because of slow pasture growth. Overall, 6-7 harvests were taken 
each year. 
In the laboratory, all tree material (leaves, catikins, small twigs, bark, etc.) was discarded 
from the herbage samples by hand sorting. The samples were then washed to remove any 
soil contamination and dried in a forced-draught oven at 800e for 24 hours. Dried samples 
were weighed on electronic scales to 2 decimal places (d.p.). The NHA per unit area per 
day was calculated by dividing the dried herbage collected from each quadrat by the total 
number of days separating the initial trim and harvest cuts. 
NHA data from individual cuts was pooled to provide seasonal and annual totals, which 
are presented relative to the open pasture environment. Where a NHA cut interval 
overlapped across two seasons, the measurement was assigned to the latter season, because 
this is when the vast majority of herbage accumulated (Radcliffe 1 974). Seasons were 
related to the phenological cycle of the trees: spring - bud burst to full leaf; summer - full 
leaf; autumn - leaf fall; and winter - no leaf At all three farm sites, bud burst began in 
September and full leaf display was not attained until late October. Summer lasted from 
November until February, when premature leaf fall occurred, likely because of a 
combination of soil moisture stress and leaf-rust disease (McElwee & Knowles 2000). 
Chapter 5 1 89 
At Kiwitea and Hautope 2 the trees were completely devoid of leaves by June. In contrast, 
leaf fall finished one month earlier in May at Hautope 1 .  
NHA was scaled up to a stand-level (weighted) estimate using the general equation: 
Weighted NHA(Zones 1 & 3) = (Zone 1 NHA x Zone 1 area) + (Zone 3 NHA x Zone 3 area) 
The two area variables for the above equation were calculated as the proportion of an 
experimental unit that each zone covered on a horizontal plane. This, in turn, was derived 
from a simple linear relationship between HPCC and CCL (refer to Section 3.2.2). 
At Kiwitea, NHA for the more intensively stratified experimental units was scaled up to a 
stand-level estimate using the equation: 
Weighted NHA (Zones 1-5) = (Zone 1 NHA x � Zone 1 area) + (Zone 2 NHA x Y3 Zone 3 
area) + (Zone 3 NHA x Y3 Zone 3 area) + (Zone 4 NHA x Y3 Zone 3 area) + (Zone 5 NHA 
x � Zone 1 area) 
5.2.2.2 Residual herbage mass below 25mm trimming height 
At the end of each season, an area similar to where each grazing exclusion cage was sited 
was trimmed to 25mm above ground level, with the cut herbage above this height removed 
and discarded. Immediately afterwards, a second cut to ground level was made inside a 
0. l m2 quadrat, with all of the remaining herbage harvested and taken to the laboratory for 
processing (refer to Section 5 .2.2. 1 ). Residual herbage mass (RHM) was calculated in 
kg DMlha for each season. Quadrat cuts were taken at Kiwitea on 22/1 0/99, 1 1103/00, 
1 7/05/99, and 1 0- 1 1 /08/99 for spring, summer, autumn, and winter, respectively. Whereas, 
at Hautope 1 and 2, residual herbage cuts were made on the 1 -211 0/99, 1 2- 1 3/03/00, 
1 4/05/99, and 1 7- 1 8/08/99 for the same respective seasons. The summer season cut for all 
three farm sites was delayed for two months because of very low pasture cover in February 
and March 2000. 
5.2.2.3 Pasture composition by weight 
Coinciding with the timing of NHA measurements, additional 0. l m2 quadrat samples for 
pasture composition analysis were cut from under the grazing exclusion cages in both late 
spring and summer. Field-sampling dates at Kiwitea over the two-year period were 29-
30/10/98 and 22110/99 for spring, and 07/01/99 and 4-5/03/00 for summer. In comparison, 
Chapter 5 1 90 
Hautope 1 and 2 were sampled on the 1 -2/1 0/99 and 12-1 3/03/00 for spring and summer, 
respectively. All of the experimental units (micro sites) selected in Chapter 2 were sampled, 
except for the paired units located on a south facing aspect at Kiwitea. These experimental 
units were excluded to limit comparisons at this site to a single northerly aspect (refer to 
Section 2.3, Table 2.2). Amongst the trees, sampling was restricted to Zones 1 and 3 
(Figure 5 . 1 ). 
In the laboratory, each fresh sample was initially spread out over a clean bench and any 
particularly large pieces of plant material (e.g. Plantago lanceolata leaves) were tom into 
smaller pieces so that all plant material was of a fairly similar size. The quantity of herbage 
was reduced down to a manageable size for hand sorting by repeatedly mixing and then 
discarding half of the sample. Sub-samples prepared for sorting were dissected into high 
fertility responsive (HFR) grasses, low fertility tolerant (LFT) grasses, legume species, 
other species, and dead pasture material. HFR grasses consisted of Lolium perenne, 
Dactylis glomerata, Holcus lanatus, and Poa spp. LFT grasses included species such as 
Agrostis tenuis, Cynosurus cristatus, and Anthoxanthum odoratum. Tree material was 
excluded from the analyses. Dissected herbage was separately dried in an oven at 60°C for 
48 hrs and weighed using electronic scales to 3 d.p. The dry weight of each category was 
expressed as a percentage of the total herbage dry weight in the dissected sub-sample 
(excluding tree material). 
5.2.2.4 Pasture nutritive value 
At Kiwitea and Hautope 1, a third 0. lm2 quadrat for pasture nutritive value analysis was 
cut from under the grazing exclusion cages at the same time as the pasture botanical 
composition samples. However, sampling was restricted to a limited range of stand 
densities at each of these sites. This range consisted of an open pasture control, and 
experimental units under low and medium-to-high overstorey densities (Appendix 5 . 1 ). 
Each overstorey-density class was replicated twice and contained two sub-units (e.g. Zones 
1 and 3 amongst the trees). 
In the laboratory, each sample was thoroughly mixed and a sub-sample taken to measure 
dry matter content. The latter was calculated from measurements of the sub-samples fresh 
(wet) and dry weights (measured to 2 d.p.). On the same day of collection in the field, 
about 200g of mixed herbage from each main sample was washed and placed separately in 
Chapter 5 1 91 
an oven to be dried at 60°c for 48hrs. The dried herbage was ground in a 'Cyclotec Mill' to 
pass through a 1 mm diameter sieve and then stored in airtight containers until subjected to 
chemical analysis. Feed value indices were measured by near infrared reflectance 
spectroscopy (NIRS; Corson et al. 1 999) at the ICP Facility, Grasslands Research Centre, 
Palmerston North. Samples were analysed for crude protein (CP), lipid, soluble 
carbohydrates (Sol CHO), neutral detergent fibre (NDF), acid detergent fibre (ADF) , in 
vitro organic matter digestibility (OMD), and metabolisable energy (ME). 
5.2.2.5 Pasture mineral content 
Two strips of pasture, each approximately 0.5m long, were cut with a portable shearing 
handpiece (75mm wide) from around the same locations and at the same time as the 
nutritive value cuts. However, the samples were taken from outside of the grazmg 
exclusion cages to prevent possible contamination by the cages galvanised coating 
(Guevara-Escobar 1 999; Mackay et al. 1 995). The methodology for processing the samples 
in the laboratory was the same as that used for nutritive value analysis, except the dry 
matter content of the fresh herbage was not measured. The mineral concentration in the 
samples was measured by Plasma Emission Spectrometry, at the ICP Facility, Grasslands 
Research Centre, Palmerston North. 
5.2.2.6 Nutritive value and mineral content of poplar leaves 
At Kiwitea and Hautope 1 ,  poplar leaves (lamina + petioles) were plucked from lower tree 
branches in spring, summer, and autumn for nutritive value and mineral content analysis. 
Samples were taken from one nuclei tree associated with each of the experimental units 
used for determining pasture nutritive value and mineral contents (i.e. 4 trees were sampled 
per site per season). Leaves collected in autumn were senescing and had started to turn 
yellow. Field-sampling dates at Kiwitea were 22/1 0/99, 07/0 1 100, and 04/04/00, for spring, 
summer, and autumn, respectively. In comparison, Hautope 1 was sampled on the 
1 711 0/99, 05/0 1 /00, and 29/03/00 for the same respective seasons. The methodology for 
processing the samples in the laboratory was the same as that used for nutritive value 
analysis. The nutritive value of the poplar leaves was measured by Near Infrared 
Reflectance Spectroscopy (NIRS), while the mineral content was measured by Plasma 
Emission Spectrometry. 
Chapter 5 
5.3 Data analyses 
5.3.1  Net herbage accumulation and residual herbage mass 
1 92 
The relationship between NHA and CCL was examined by regression analysis, using the 
REG and NLIN procedures of SAS@ (version 8.02 for Windows@, SAS Institute, Inc. 
1 999). For each site, the above relationship was separately studied for Zone I ,  Zone 3, and 
at the stand level (weighted across an experimental unit). Regression model selection was 
by the 'forward method', where initially the most appropriate model is assumed to be a 
straight-line (Kleinbaum et al. 1 998). The adequacy (quality of fit) of each regression 
model was checked via inspection of scatter, residual, and normal probability plots, along 
with more formal test statistics (SAS 1 990). CurveExpert@ (version 1 .3 7  for Windows@, 
Hyams 2001 )  and applicable published ecological studies were also used to determine 
potential regression models for the data. For each relationship, the simplest model with the 
smallest standard error of prediction (RMSE) and no trend in the residual plots was 
selected. Slopes of the relationships for different sites, seasons, or years (Kiwitea), were 
compared using t-tests (comparisons only made between the same regression models). 
A split plot analysis of variance (ANOV A) was conducted for the more intensively 
sampled experimental units at Kiwitea, using the GLM procedure of SAS@, to test the 
effects of CCL (main plots), understorey location (subplots), and their interaction on NHA. 
Where appropriate, individual treatments, including an open pasture control, were 
compared using 95% Cl. 
Similar statistical analyses were carried out for residual herbage mass. 
5.3.2 Pasture composition, nutritive value, and mineral content 
For each site, a split-plot-in-time analysis of variance (ANOV A) was conducted, using the 
GLM procedure of SAS@, to test the effects of the overstorey environment (main plots), 
season (subplot), and their interactions on the various pasture characteristics. In the general 
linear model, a 'replicate nested within over storey environment' parameter was treated as a 
random-effect, while all other parameters were considered fixed (Hedderley per. comm. 
2002). 
Chapter 5 1 93 
The 3 x 2 factorial design for the botanical species composition data was unbalanced, with 
treatment combinations having 4- 1 0  replicates. As a result, Type 3 sums of squares were 
used instead of Type 1 ,  due to their more conservative nature. The data sets were balanced 
for nutritive value and mineral concentration analyses. 
Separate regression analyses were carried out for the two main overstorey environments 
amongst the poplar stands (Zones 1 & 3) and also for the two seasons (spring and 
summer). Differences between regression equations were tested by analysis of co variance 
(ANCQV A), using the GLM procedure of SAS® (Kleinbaum et al. 1 998; Littell et al. 
1 99 1 ). Where regression equations for the two main overstorey environments amongst the 
poplar stands (Zones 1 & 3) coincided (i.e. were not significantly different in intercept or 
slope) the data were combined into a single function representing the entire understorey 
environment. 
5.3.3 Poplar leaves nutritive value and mineral content 
Differences in poplar leaf nutritive value and mineral content between spring, summer, and 
autumn were analysed by multivariate analysis of variance (MANOVA), using the 
ANOVA procedure of SAS® (Littell et al. 1 991) .  
5.3.4 A check of underlying model assumptions 
Diagnostic options provided in SAS® were used to check the underlying regression and 
ANOVA assumptions. These included studentised residual and normal probability plots, 
along with more formal test statistics (SAS 1 990). Several data sets required 
transformation to meet either the assumption of normality or homogeneity of variance. 
Pasture characteristics that required transformation are clearly identified in the appropriate 
figures and tables. 
Chapter 5 
5.4 Results 
5.4.1 Annual net herbage accumulation 
5.4.1 .1  Directly underneath the poplar crowns 
1 94 
Annual net herbage accumulation (ANHA) in the open pasture was 1 3 .2 ± 0.9, 1 3 .4 ± 0.9, 
1 0.9 ± 0.2, and 1 3 .5 ± 1 .7 t DMlha (mean ± SEM) at Kiwitea in 1 998-99, 1 999-00, and at 
Hautope 1 and 2, respectively. At all of the sites, ANHA was lower directly below the 
poplar crowns (Zone 1 )  than in the open pasture (Figure 5.2a). In particular, the rate of 
decline in ANHA was very high when initially going from the open environment to a low 
poplar canopy cover (eeL). The lack of data points between 0-40% eeL made it difficult 
to identify the exact form of the relationship and as a result, a number of potential 
regression models, ranging from a gentle exponential to a sharp hyperbolic decay curve, 
fitted the data reasonably well. Out of the potential regression models tested, a simple 
power curve was selected because of its slightly better fit in relation to the RMSE and 
distribution of residuals. Overall, Zone 1 ANHA averaged across all three sites became 
relatively constant at around 50% of open pasture beyond approximately 20% eeL 
(Figure 5 .2a). Again, the lack of data points over the lower eeL range inhibited a more 
accurate estimation of this threshold. 
The trees effect on Zone 1 ANHA varied significantly between sites (Appendix 5. 1 . 1  & 
5 .2a). At Hautope 1 ,  the rate of decline in ANHA with increasing eeL was less than at 
Kiwitea, and on average ANHA was reduced to 65% ± 1 % of the open pasture, compared 
to 47%, 3 8%, and 44% ± 3% for Kiwi98-99, Kiwi99-00, and Hautope 2, respectively. 
Although the rate of decline in ANHA, relative to the open pasture, did not vary 
significantly between the two measured years at Kiwitea (Appendix 5 . 1 . 1  & 5.2a), ANHA 
directly below poplar crowns was on average marginally higher in 1 998-99 than 1 999-00 
(paired t-test; P<O.OOO I).  The high level of error associated with the small sample size 
(n=8) and high variance, especially for open pasture measurements, in the relationship for 
Hautope 2 reduced its statistical power for making comparisons with the other sites 
(Appendix 5 . 1 . 1  & 5.2a). 
Cha�er 5 1 95 
5.4.1 .2 Within the vertically projected canopy gap between trees 
ANHA in the vertically projected gap between the trees (Zone 3) decreased :from open 
pasture values at a constant rate with increasing eeL (Figure 5.2b). Across all sites, the 
rate of decline was 6.6% ± 0.5% of open pasture production for every 1 0% increase in 
eeL (Figure 5 . 2b). Between the sites, Zone 3 ANHA at Hautope 1 and Kiwi98-99 
appeared to decrease at a lesser rate than at Hautope 2 and Kiwi99-00. However, only the 
difference in the slope of the relationship for Hautope 1 and Kiwi99-00 was statistically 
significant (Appendix 5. 1 .2 & 5.2b). Zone 1 and 3 ANHA converged at around 80% eeL, 
indicating that at this level the poplar overstorey had become relatively homogeneous and 
closed (Figures 5.2a,b). 
5.4.1 .3 Weighted across the inter-tree space (stand level) 
Stand level (weighted) ANHA decreased at a slightly diminishing rate with increasing 
eeL (Figure 5.2c). Pooled across all sites, ANHA under a eeL of 25%, 50% and 75% was 
estimated to be 77%, 60%, and 48% of the open pasture, respectively. Overall, the simple 
exponential decay regression model accounted for a high proportion of the variation in 
ANHA (r2=0.77). However, the degree of negative curvature varied significantly between 
sites. Hautope 1 exhibited the least negative curvature, followed by Kiwi98-99, and then 
Kiwi99-00 and Hautope 2 (Appendix 5 . 1 .3 & 5 .2c). The high standard error associated 
with the Hautope 2 relationship resulted in it not being significantly different from any of 
the other sites (Appendix 5. 1 .3). Based on the individual site relationships predicted 
ANHA at 75% eeL varied anywhere between 35% and 65% of the open pasture 
(Appendix 5 . 1 .3 & 5.2c). 
1 40 140 
'""' 0 Y = 99.64 (3.06) . (x+1 r0. l  8 (0.01 )  j 1 2 0  � 0 (b) i 1 20 0 2 0 i r =0.76, RMSE=13.8, P<O.OOO l i � 0.. 1 00 0.. 100 
&3 r:: 0.. � 0 80 80 
C � ... ... ...  
� 
60 ... -" � 60 0 ... 0 
40 <: 40 � .  M V V r:: 20 r:: 20 • 0 0 Y = 99.33 (2.74) - 0.66 (0.05) • X N N 
0 
(a) 
0 
r2=0.73, RMSE=13.2 1 ,  P<O.OOO l 
0 20 40 60 80 1 00 0 20 40 60 80 1 00 
Canopy closure with leaves (%) 
'""' 1 40 
Y = 99.27 (2.76) • e-O.OI (0.001 ) · X 1 1 20 0 0 0.. i r2=0.77, RMSE=12.96, P<O.OOO I 
8. 1 00 
0 .... 0 80 � .......... ... � � 
� 
60 
40 1 • -0 • • •  V • • ... 
tb 20 � 
0 
(c) 
0 20 40 60 80 100 
Canopy closure with leaves (%) 
Figure 5.2 Annual net herbage accwnulation (ANHA) over a range of in-leaf canopy closure ratios. Standard errors for regression coefficients are 
given in parentheses. Symbols: (0) Kiwitea98-99 (e) Kiwitea99-00, (A)  Hautope 1 ,  (0) Hautope 2. Solid line - mean predicted response. 
Abbreviations: r2, adjusted coefficient of detennination; RMSE, standard error of prediction. 
� 
CO 0> 
Chapter 5 1 97 
5.4.1 .4 Intensively stratified across the inter-tree space at Kiwitea 
Across the inter-tree space, ANHA was significantly affected by the interaction between 
the density of the poplar stand, defined by its eeL, and the understorey location (zone of 
tree influence) (p=0.033; Figure 5.3). Directly below the poplar crowns, on either the 
shade-facing (Zone 1)  or sun-facing (Zone 5) side of the trees, ANHA was similar at 
around 42% of open pasture, irrespective of eeL. At a stand density of 70% eeL, ANHA 
within the gap centre (Zone 3) and adjacent intermediate/transitional areas (Zones 2 & 4) 
also did not differ significantly from Zones 1 and 5 (P>0.05). In contrast, at 50% ceL, 
ANHA in Zones 2, 3, and 4 increased to around 62% of open pasture production (p<0.05). 
The difference between Zone 4 ANHA at 50% eeL and Zone 3 ANHA at 70% eeL was 
not statistically significant (p>0.05; Figure 5.3). 
In general, there was good comparability between stand level (weighted) estimates of 
ANHA made using either two or five zones of tree influence. Weighted ANHA based on 
Zones 1 & 3 was 55 ± 4.2% and 4 1 . 1  ± 2.9% of the open pasture under a eCL of 50% and 
70%, respectively. This compared with 52.7 ± 4.2% and 4 1 . 1  ± 3 .7% of open pasture 
production, respectively, estimated using the five zones of tree influence (Zones 1 -5). 
Thus, while there is little difference between the methods in estimating ANHA at high 
stand densities, the more simplified method marginally overestimates herbage production 
« 5%) at medium stand densities (Paired t-test; P=0.0035). 
120 
., 100 
t---f---!---�---� 
... 
B 
'" 80 oS 0.. 
Q Q 0.. 60 0 
<f?-
'-' 
� 40 
� 20 
0 
2 3 4 5 
Zone of tree influence 
Figure 5.3 Annual net herbage accumulation (ANHA) stratified across the inter-tree space at Kiwitea 
( 1998-2000). Vertical bars represent the standard error of the mean (SEM). Symbols: (T)  open pasture (0) 
50% eeL (.) 70% eeL. Sample size: 20 experimental units with trees and 6 experimental units without 
trees (open pasture). 
Chapter 5 
5.4.2 Seasonal net herbage accumulation 
5.4.2.1 Summer and autumn 
1 98 
In summer (Figure 5 .4) and autumn (Figure 5 .5) the shape of the relationships for Zone 1 ,  
3 and stand level (weighted) NHA was generally similar to those for ANHA (Figure 5 .2). 
Overall, the rate of decline in NHA associated with increasing eeL did not vary 
significantly between the two consecutive seasons for each environment (P>0.05). 
Averaged over both seasons, Zone 1 NHA became relatively constant at around 36% of 
open pasture production by at least 20% eeL (Figures 5 .4a & 5.5a). Nevertheless, there 
were significant differences in the trees effect on Zone 1 NHA between sites/years. 
In summer, Zone 1 NHA at Kiwi99/00 was reduced to 27 ± 3% of open pasture 
production, compared with 45 ± 4%, 38  ± 2%, and 47 ± 5% for Kiwi98/99, Hautope 1 and 
Hautope 2, respectively. In autumn, Zone 1 NHA at Kiwi99/00 was reduced further to 1 8  ± 
3% of open pasture production, while at Kiwi98/99, Hautope 1 and Hautope 2 under storey 
herbage accumulation was 44 ± 4%, 42 ± 4%, and 30 ± 5%, respectively. Zone 1 NHA in 
the latter three sites/years did not vary significantly between the two consecutive seasons 
(P>0.05). However, unlike in summer, the difference between Kiwi99/00 and Hautope 2 in 
autumn was not significant (P>0.05). 
In contrast to Zone 1, NHA in Zone 3 and at the stand level (weighted) decreased more 
slowly with increasing eeL (Figures 5.4b,c & 5 .5b,c). For example, at 20% eeL Zone 3 
and weighted NHA was 79% and 76% of the open pasture, respectively, when pooled 
across all sites and the two consecutive seasons. In both summer and autumn, Zone 3 NHA 
decreased at a significantly faster rate at Kiwitea in 1 999-00 than 1998-99 (P<0.05). 
Whereas, at the stand level (weighted), over the same two seasons, the rate of decline in 
NHA at Kiwi99/00 was also significantly greater than Kiwi98/99 and Hautope 1 (P<0.05). 
Similarly to ANHA, the high error associated with the relationships determined for 
Hautope 2 would likely have reduced the statistical power for maldng comparisons with 
other sites (data not shown). 
Overall, the strength of the summer relationships was strong (Figure 5 .4) . However, the 
strength of these relationships decreased to a more moderately strong level in autumn, 
mainly due to a disproportionately large amount of variation in the open pasture (Figure 
5 .5). 
1 60 1 60 0 Y = 99.25 (3 .43) - 15 . 1 8  (1 .04) · In(x+ l)  0 sqrt(Y) = 9.95 (0.22) - 0.047 (0.004) · X ,...... I 140 j � 140 2 r2=0.69, RMSE=17.47, P<O.OOO I r =0.79, RMSE= 1 5.6 1 ,  P<O.OOOI a 1 20 � !l 120 � 0 8. 100 I:i 8. 100 0 0 .... .... 0 80 0 � � 80 
� 60 � 60 40 .... c t') 40 Q) • Q) I:i 20 I:i � • • 0 20 Ca) N Cb) 
0 0 
0 20 40 60 80 1 00 0 20 40 60 80 1 00 
Canopy closure with leaves (%) 
1 60 ,...... 0 sqrt(Y) = 9.86 (0. 1 9) - 0.053 (0.004) • X 1 140 2 r =0.78, RMSE=1 5.65, P<O.OOOI 
Po 120 � 
8. e 0 � 1 00 • 
0 
� 80 '-' 
� 60 
"0 40 i 
.d) 20 
� Cc) 
0 
0 20 40 60 80 1 00 
Canopy closure with leaves (%) 
Figure 5.4 Summer net herbage accumulation (NHA) over a range of in-leaf canopy closure ratios. Standard errors for regression coefficients are 
given in parentheses. Symbols: (0) Kiwitea98-99 (e) Kiwi99-00, (0) Southern aspect Kiwi98-99 C+) Southern aspect Kiwi99-00, C. )  Hautope 1 ,  
(0) Hautope 2 .  Solid line - mean predicted response. Abbreviations: r2, adjusted coefficient of detennination; RMSE, standard error of prediction. � (0 
(0 
160 
,....., 
i 1 40 
1 
t 
0.. 120 e 
1'1 � 100 i 
'Cl 
'*- SO '-' 
� 60 40 
� 
� 20 
0 
0 
160 
,....., [J 
1 1 40 t 
� 120 e 
IU i • §' 100 .... 0 
::R e.... SO 
� 60 
] 40 
-a 
2: ] 
.d) 
� 
0 
Y = 99.S5 (5.23) - 16.35 ( 1 .59) · In(x+ 1 )  
r2=0.65, RMSE=23.S1 ,  P<O.OOOI 
• 
20 40 
�� 0 0 o 
o •• • • • • • 
60 SO 
(a) 
1 00 
sqrt(Y) = 9.90 (0.3 1 )  - 0.06 1 (0.006) • X 
r2=O.65, RMSE=23.59, P<O.OOOI 
-: . .  • • • (c) 
20 40 60 80 100 
Canopy closure with leaves (%) 
160 0 
,....., sqrt(Y) = 9.S9 (0.34) - 0.057 (0.006) • X 
1 1 40 t r2=0.5S, RMSE=25.9S, P<O.OOOI e 0.. 120 
[ 1 00 i e 
0 
0 .... 0 SO '*-
'-' 
� 60 40 ...., 
� 20 1 0 � . �  N (b) 0 • 
0 20 40 60 SO 100 
Canopy closure with leaves (%) 
Figure 5.5 Autumn net herbage accumulation (NHA) over a range of in-leaf canopy closure ratios. Standard errors for regression coefficients are 
given in parentheses. Symbols: (0) Kiwitea98-99 (e) Kiwi99-00, (0) Southern aspect Kiwi98-99 (+) Southern aspect Kiwi99-00, (A)  Hautope 1,  
(0) Hautope 2. Solid line - mean predicted response. Abbreviations: r2, adjusted coefficient of determination; RMSE, standard error of prediction. N o o 
Chapter 5 201 
5.4.2.2 Winter and spring 
At Kiwitea and Hautope 2 in winter the relationship between NHA and eeL for Zones 1 
and 3, and also at the stand level (weighted), did not change significantly from the 
preceding two seasons (P>0.05; Table 5 . 1 ;  Figure 5.6). In comparison, at Hautope 1 NHA 
particularly under higher eeLs recovered towards open pasture levels. As a result, Zone 1 
and 3 NHA for this site decreased at only a marginal rate with increasing eeL (Table 5 . 1 ;  
Figure 5.6). 
Table 5.1 Regression models, and their respective coefficients, developed for CCL to predicted winter 
understorey NHA. relative to open pasture production. 
Zone Site Equation Coefficients 
a b r RMSE p 
1 &3 Y=a-b*ln(x+ 1 )  99.22 (3.50) 14.88 ( 1 .09) 0.8 1 14.30 <0.000 1 
2 Y=a-b*x 1 00.6 1 (4.72) 0.2 1 (0.08) 0.29 9.68 0.0277 
3 1 &3 sqrt(Y)=a-b*x 9.86 (0.22) 0.048 (0.004) 0.74 1 6.00 <0.000 1 
3 2 Y=a-b*x 99.99 (6.74) 0.28 (0. 1 2) 0.24 1 3 .81 0.04 1 8  
Weighted 1 &3 sqrt(Y)=a-b*x 9.83 (0.2 1 )  0.052 (0.004) 0.79 1 5. 1 2  <0.000 1 
Weighted 2 Y=a-b*x 1 00.4 1 (3.97) 0.24 (0.07) 0.43 8 . 1 4  0.0063 
Standard errors for coefficients are presented in parentheses. Abbreviations: Y, understorey NHA relative to 
open pasture production (unit: %); X, in-leaf canopy cover (unit: %); Site I, Kiwitea; Site 2, Hautope 1 ;  
Site 3, Hautope 2 ;  r, adjusted coefficient of determination; RMSE, standard error of prediction; 
Total number of observations for regression models based on Sites 1 &3 and Site 2 were 44 and 1 4, 
respectively. At Site 2 (Hautope 1 )  the slope of the relationship for Zone 1 and 3 were not significantly 
different (P>0.05). 
At Hautope 1 ill spring, NHA did not change significantly with increasing eeL, 
irrespective of environment amongst the trees (Figure 5 .7). On average, Zone 1 NHA at 
this site was 1 8  ± 6% higher than in the open pasture, while Zone 3 NHA did not differ 
significantly from either Zone 1 or the open pasture environment. Zone 1 NHA also 
remained relatively constant across the range of eeLs at Kiwitea and Hautope 2, but 
averaged only 59% of open pasture production (Figure 5.7a). There was also significant 
yearly variation, with Zone 1 NHA at Kiwitea in 1998-99 and 1999-00 averaging 5 1  ± 3% 
and 68 ± 3% of the open pasture, respectively. Similarly to Hautope 1 in winter and spring 
(Figure 5.6 & 5.7), NHA at Kiwitea and Hautope 2 in spring recovered to a greater extent 
at higher eeLs. Thus, the overall magnitude of decline in NHA amongst the trees was less 
than for other seasons. 
1 60 
,-.. e 
� 1 40 rd 120 0 0.. 
= 
8- 1 00 0 
"0 80 '#. '-' 
� 60 40 
011 = 
20 � 
0 
0 
1 60 ,-.. 
� 140 • 0.. 1 20 0 
[ 0 1 00 ...... 0 
'#. 80 '-' 
� 60 "'" 40 011 
� 20 
� 
0 
0 
A 
---_.&_�-
A A --
41  A 
0 0 
20 40 60 
C)-__ .. .4 ------ .. o A - -� 
A 
0 
20 40 60 
Canopy closure with leaves (%) 
160 
,-.. • 
� 140 1 20 0 0.. 
= 
8- 100 A AA 0 A A ...... ---0 80 ---'#. A AA '-' A 
� 60 
eA 0 
40 ,.., 011 • = 20 0 
(a) N (b) 
0 
80 1 00 0 20 40 60 80 1 00 
Canopy closure with leaves (%) 
(c) 
80 1 00 
Figure 5.6 Winter net herbage accumulation (NHA) over a range of in-leaf canopy closure ratios. Symbols: (0) Kiwitea98-99 (e) Kiwi99-00, 
(0) Southern aspect Kiwi98-99 (+) Southern aspect Kiwi99-00, (A)  Hautope 1 ,  (D) Hautope 2. Solid line - mean predicted response for Kiwitea 
and Hautope 2. Dashed line - mean predicted response for Hautope 1 .  
1 60 160 (b) ,-., (a) ,-., 
i 140 ...... e 140 0 ... � 0 120 0 � 120 0 • 0.. � • 0.. 8. § • 1 00 • 0.. 100 0 0 ..... • ..... 0 80 0 80 <J. • � --- i .. c:. .  0 . 0  0 � 60 '\ � 60 • 0 • 8 &> 0 • 0 0 0 40 0 is> G) M 40 0 • i Cl) Cl) c c 20 Y = 99.84 (4.52) - 0.35 (0.09) ... X � 20 Y = 94.30 (5.65) - 0.28 (0. 1 1 )  ... X 0 
r2=0. 1 ,  RMSE=27.3 1 ,  P=0.0 1 32 
N 
r2=0.22, RMSE=2 1 .84, P=O.OOOI 
0 0 
0 20 40 60 80 1 00 0 20 40 60 80 100 
Canopy closure with leaves (%) 
160 ,-., (c) 
i 1 40 0 .\ 0.. 120 0 fl i · 0.. 1 00 ... 0 ..... • • 0 
� e..,., 80 
� 60 0 0 0 C "1:l 40 � � 20 Y = 98.54 (4. 8 1 )  - 0.33 (0.09) '" X r2=0. 1 7, RMSE=23 .25, P=0.0008 
0 
0 20 40 60 80 100 
Canopy closure with leaves (%) 
Figure 5.7 Spring net herbage accumulation (NHA) over a range of in-leaf canopy closure ratios. Standard errors for regression coefficients are 
given in parentheses. Symbols: (0) Kiwitea98-99 (e) Kiwi99-00, (0) Southern aspect Kiwi98-99 (+) Southern aspect Kiwi99-00, (A) Hautope 1 ,  
(0) Hautope 2 .  Solid line - mean predicted response. Abbreviations: r2, adjusted coefficient of determination; RMSE, standard error of prediction. 
N 
o eN 
Chapter 5 204 
5.4.2.3 Intensively stratified across the inter-tree space at Kiwitea 
In summer, NHA varied significantly across the space between the trees (p=0.0006). 
Averaged over both 50% and 70% eeL, NHA directly below the trees on the shade-facing 
(Zone 1 )  and sun-facing (Zone 5) sides of the canopy gap was reduced to 30% of the open 
pasture (Figure 5 .8a). In contrast, NHA within the centre of the gap (Zone 3) and 
transitional areas (Zones 2 and 4) was reduced to 50% of open pasture production 
(Figure 5 .8a). Overall, when averaged across the entire space between the trees, NHA 
tended (p=0.0696) to be 14% higher (relative to the open pasture) under 50% than 70% 
eeL (data not shown). 
Depending on the density (eeL) of the poplar stand, understorey NHA in autumn varied 
significantly with location between the trees (p=0.01 1) .  At 50% eeL, NHA in Zones 2, 3, 
and 4 was not significantly different from the open pasture (P>0.05; Figure 5 .8b). 
However, directly below the trees on both sides of the canopy gap (Zones 1 and 5) NHA 
was reduced to 35% of open pasture production. NHA did not vary significantly across the 
entire inter-tree space (Zones 1 -5) under 70% eeL and was maintained at a level similar to 
directly below the trees at 50% eeL (Figure 5.8b). The difference in NHA between Zone 4 
at 50% eeL and Zone 3 at 70% eeL was not statistically significant (P>0.05 ; Figure 
5 .8b). 
In winter and spring there were no significant differences in NHA across the inter-tree 
space (Zones 1 -5) or between stand densities (50 and 70% eeL). On average, NHA 
amongst the poplar trees was 46% of open pasture production in winter (Figure 5 .8c). This 
increased to 70% in spring (Figure 5.8d). 
For each season, there was good comparability between estimates of stand level (weighted) 
NHA using either two or five zones of tree influence. Relative to estimates based on all 
five zones of tree influence, alternative estimates using just Zones 1 and 3 marginally 
overestimated weighted NHA in summer by 9% (paired t-test; P=0.0457). Similarly, 
autumn NHA at the medium stand density of 50% eeL was slightly overestimated by 10% 
when weighted using Zones 1 and 3 (Paired t-test; P=0.043 1 ). Otherwise, there were no 
significant differences between these two methods. 
120 120 
"..... 1 00 
I 80 p. 
!---t ---f ---1---I "..... 1 00 
I 80 p. 
1---t --1---1---I 
5 p. 60 0 � 60 ..... ..... 0 0 
-;.R � 40 -;.R � 40 
� 20 � 20 
0 
(a) (b) 
0 
2 3 4 5 2 3 4 5 
120 120 
"..... 1 00 
I 80 
"..... 100 
i 80 p. 
f-- - t - - -f 
-
--1-- -I f-
-
-1---� ---!-- -I 
8. 60 0 
c 
8- 60 0 .... ..... 0 
-;.R 40 � 
� 20 
0 
-;.R 40 � 
� 20 
....---- I I ..... 
(c) (d) 
0 0 
2 3 4 5 2 3 4 5 
Zone of tree influence Zone of tree influence 
Figure 5.8 Net herbage accumulation (NHA), relative to the open pasture, across the inter-tree space at Kiwitea ( 1998-00) in: (a) summer, 
(b) autumn (c) winter, and (d) spring. Vertical bars represent the standard error of the mean (SEM). Symbols: (T)  open pasture, (0) 50% eeL, 
(e) 70% eeL, and (_) 50% and 70% eeL averaged together. N o c.n 
Chapter 5 
5.4.3 Residual herbage mass below 25 mm trimming height 
5.4.3.1 Effect of overstorey environment on Kiwitea residual herbage mass 
206 
In late spring, the residual herbage mass (RHM) of the open pasture was 784 ± 74 kg 
DMlha (Figure 5 .9a). In comparison, Zone 1 RHM was 30% lower than in the open pasture 
(p=0.0224), whereas Zone 3 was not significantly different from either of the other two 
main overstorey environments (Figure 5 .9a). By late summer, the RHM in the open pasture 
had increased to 1 572 . 1 89 kg DMIha (Figure 5.9a). This amount of herbage was 2.4-
(p=0.004) and 1 .9-fold (p=0.009) greater than in Zone 1 and 3, respectively (Figure 5 .9a). 
The relative differences between the three main overstorey environments in late autumn 
were similar to those measured in summer. Nevertheless, the overall amount ofRHM was 
markedly reduced, with the open pasture having around 700 kg DMlha (Figure 5 .9a). 
In late winter, the RHM ranged between 300-500 kg DMlha, and was not significantly 
different between the three overstorey environments (Figure 5.9a). 
5.4.3.2 Effect of poplar overstorey density on Kiwitea residual herbage mass 
In late spring, the RHM in Zone 1 and 3 was relatively constant at approximately 550  kg 
DMIha and 700 kg DMlha, respectively, over the entire range of poplar canopy closures 
(data not shown). Similarly, in late summer the RHM in Zone 1 also remained relatively 
constant at around 700 kg DMlha (data not shown). This contrasted with the REM in 
Zone 3,  which decreased from open pasture levels at a rate of 120 ± 23 kg DMlha for every 
1 0% increase in eeL (Figure 5 . 1 Oa). In autumn, the RHM in both Zones 1 and 3 decreased 
at a lesser (p<0.05) rate of 46 ± 6 kg DMIha for every 1 0% increase eeL (Figure 5 . 1Ob). 
The strength of both the summer and autumn linear relationships was strong (r2=0.62-0.66; 
P<0.000 1 ). However, a more accurate determination of the general shape of the 
relationships was hindered by the lack of data points between 0-40% eeL (Figures 
5 . 1 Oa,b). In late winter, the RHM in Zones 1 and 3 was relatively constant at 
approximately 350 kg DMlha, irrespective of eeL (data not shown). 
Chapter 5 207 
5.4.3.3 Effect of overstorey environment on Hautope 1 & 2 residual herbage mass 
Over the entire year at Hautope 1 ,  the RHM in the open pasture was at least twice as high 
as amongst the trees (Zones 1 & 3) (P<0.05; Figure 5.9b). At the same time, the RHM was 
not significantly different in Zones 1 and 3 (P>0.05; Figure 5.9b). The open pasture RHM 
reached a maximum of 1700 kg DMlha in late spring and a minimum of 850 kg DMIha in 
late autumn/winter (Figure 5.9b). 
At Hautope 2 in spring, there was little difference in the RHM amongst the three main 
overstorey environments (P>0.05; Figure 5.9c). In summer, the RHM amongst the trees 
(Zones 1 & 3) was 38% lower than in the open pasture (P<0.05; Figure 5.9c). This 
increased further to between 55%-60% in late autunm and winter in Zone 1 (P<0.05), 
while the RHM in Zone 3 was not significantly different from either of the other two main 
overstorey environments (Figure 5.9c). 
5.4.3.4 Effect of poplar overstorey density on Hautope 1 & 2 residual herbage mass 
In late spring and summer the RHM amongst the trees (Zones 1 & 3) remained constant at 
around 500-600 kg DMlha, irrespective of the eeL (data not shown). Whereas, in late 
autumn the RHM decreased from open pasture levels at a rate of 82 ± 7 kg DMlha for 
every 10% increase in eeL (Figure 5 . 1  l a). Similarly to Kiwitea, the determination of the 
precise form of the relationship was hindered by the lack of data points between 0-40% 
eeL (Figures 5 . 1 0a,b & 5. 1 1 a). Analysis of the residual errors, for the individual zones of 
tree influence, indicated that the relationship between Zone 3 RHM and eeL might have 
convex curvature; while, there was weak evidence for some form of concave curvature 
associated with Zone 1 (Figure 5. l 1 a). Nevertheless, the poor and unbalanced spread of the 
data points, especially towards the lower range of poplar canopy closures, inhibited the use 
of such relationships with confidence. A similar relationship to autumn also occurred in 
winter (P>0.05; Figure 5 . 1 1b). Variation in RHM around the fitted function in winter was 
greater than in autumn (Figure 5 . 1 1 a,b). 
I 2000 (a) 
1500 � 
Vl 
� 1 000 v 
� .D t> ..cl 500 
] :9 � 0 
� 2000 
� 
� 1500 Vl 
� 1 000 v 
� -e v ..cl 500 � 
.g 
� 0 
- Open pasture 
� Zone 1 
_ Zone 3 
Spring Summer Autumn Winter 
(c) 
Spring Summer Autumn Winter 
� 2000 .------------------, 
� 
M 
� 
� 
I 
� 
� 
1 500 
1000 
500 
o -'---
Spring Summer Autumn Winter 
Figure 5.9 Residual herbage mass in each of the three main over storey environments at (a) Kiwitea, (b) Hautope 1 ,  and (c) Hautope 2. Vertical 
bars represent the standard error of the mean (SEM). Abbreviations: Zone 1 ,  directly below the crowns of individual trees; and Zone 3, at the 
centre of the vertically projected canopy gap between trees. N o (X) 
Chapter 5 
2500 
.--. 
� 2000 
� '" 1 500 � 
� 1 000 1: 
� 
� 500 '" 
� 
0 
0 
1 200 
� 1 000 
0 
� 800 '" 
� 600 
� 1: 
� 400 
] 
:2 '" 200 � 
0 
0 
20 
20 
40 
40 
Y = 1 566 ( 1 19) - x*1 1 .9 (2.3) 
r2=0.62, RMSE=320, P<O.OOOl 
o 
o 
(a) 
60 80 100 
Y = 666 (37) - x*4.6 (0.6) 
r2=0.66, RMSE=106, P<O.OOO I 
(b) 
60 80 1 00 
Canopy closure with leaves (%) 
209 
Figure 5.10 Post-trimming Kiwitea residual herbage mass in (a) swnmer and (b) autumn over a range of 
poplar canopy closure ratios (CCL): (e) directly below the individual tree crowns (Zone 1 ), and (0) at the 
centre of the vertically projected canopy gap between trees (Zone 3). Standard errors for regression 
coefficients are given in parentheses. Abbreviations: r, adjusted coefficient of determination; RMSE, 
standard error of prediction. 
Chapter 5 
1200 
� 1000 
0 
� 800 
<Il 
� 600 � <.:I -e 
] 400 
] :g <Il 200 � 
0 
1600 
� 
1400 
1200 0 
Cl) 
C 1000 <Il 
� 800 � <.:I -e 600 ] 
] 400 :g <Il 
� 200 
0 
0 20 
0 20 
40 
40 
Y = 875 . 1  (39.6) - x*8.2 (0.7) 
�=0.78, RMSE=122, P<O.OOOI 
(a) 
60 80 1 00 
Y = 1 024.5 (66.7) - x*9.7 ( 1 .2) 
�=0.63, RMSE=205.6, P<O.OOOI 
(b) 
60 80 1 00 
Canopy closure with leaves (%) 
21 0 
Figure 5.1 1  Post-trimming Hautope 1 & 2 residual herbage mass in (a) autumn and (b) winter over a range of 
poplar canopy closure ratios (CCL). Directly below the individual tree crowns (Zone 1) at (A)  Hautope 1 and 
(_) Hautope 2, and at the centre of the vertically projected canopy gap between trees (Zone 3) at (6) Hautope 
1 and (D) Hautope 2. Standard errors for regression coefficients are given in parentheses. Abbreviations: r, 
adjusted coefficient of determination; RMSE, standard error of prediction. 
Chapter 5 21 1 
5.4.3.5 Spatial variation in residual herbage mass between the trees at Kiwitea 
In late summer, depending on the density (eeL) of the poplar stand, the RHM varied 
significantly with location between the trees (P=0.045; Figure 5 . 1 2a). Under 50% eeL, 
RHM in Zone 3 was 55% greater than in Zone 1 (Figure 5 . 12a). The RHM in Zone 3 also 
tended (p:SO. l )  to be at least 49% greater than in all of the zones of tree influence at 70% 
eeL, except for Zone 4. However, these differences were only significant at P:sD.05 for 
Zone 2 and 5 (Figure 5 . 1 2a). In general, the RHM in Zone 5 at 70% eeL, was 
significantly lower than all other zones of tree influence, except for Zone 2 at the same 
eeL, and also Zone 1 at 50% eeL (Figure 5 . 1 2a). Overall, the summer RHM in the open 
pasture was significantly greater than in all of the stratified zones of tree influence, except 
for Zone 3 at 50% eeL (Figure 5 . 12a). On average, the difference in RHM was 2.3-fold 
greater in the open pasture. 
The RHM was not significantly different between any of the zones of tree influence or 
stand densities in late autumn (P>0.05; Figure 5 . 12b). However, under both 50% and 70% 
eeL the RHM in Zone 1 was approximately 55% less than in the open pasture (Figure 
5 . 1 2b). In late winter and spring, the post trimming RHM did not vary significantly across 
the inter-tree space or between the different stand densities, including the open pasture 
(P>0.05 ; data not shown). 
Chapter 5 
2000�------------------------------------� 
1 1 600 
� � 1 200 
Cl) 
� -e 800 
] 
] 
� 400 
� 
�--t--1---�--f 
�/ - -- - -f ·-- ·-- - J: 
,/ 
Ca) 
o �-----r-----'------.------r----_.r---� 
1 2 3 4 5 
1 000.-------------------------------------�
800 
600 
� -e 400 
] 
] 
� 200 
� 
(b) 
f---f---f---f---f  
o �-----r-----,------.------r----_.----� 
2 3 4 5 
Zone of tree influence 
21 2 
Figure 5.12 Residual herbage mass (RHM) stratified across the inter-tree space at Kiwitea in (a) late summer 
and (b) late autumn ( 1999-2000). Vertical bars represent the standard error ofthe mean (SEM). Symbols: (T)  
open pasture, (0) 50% eeL, and (.) 70% eeL. 
Chapter 5 
5.4.4 Pasture botanical species composition 
21 3 
5.4.4.1 Effect of overstorey environment and season on Kiwitea pasture 
composition 
General pasture composition 
Grasses dominated the pasture botanical composition, accounting for over 70% and 50% of 
total sward biomass in spring and summer, respectively (Table 5 .2). Overall, Zone 1 
tended to contain 1 0- 18% more high fertility responsive (HFR) grasses than either Zone 3 
or the open pasture (P=0.05; Table 5 .2). In contrast, no significant differences between the 
three main overstorey environments were evident for low fertility tolerant (LFT) grasses, 
legumes, other species (flat weeds, thistles, etc.), or dead pasture material (Table 5 .2). 
Between spring and summer, the proportion of HFR grasses, when averaged across all 
three main overstorey environments, decreased by 20% in the sward, with a compensatory 
4% and 1 8% increase in other species and dead pasture material, respectively. 
Over the same period of time, the total legume content decreased from 1 0% to 6% 
(Table 5 .2). 
Main high fertility responsive grasses and legumes 
The most abundant HFR grass in the open pasture was Lolium perenne, with a small 
amount of Poa (Table 5.3) and trace amounts of Dactylis glomerata and Holcus lanatus 
also present « 0.5%DM). Averaged over both seasons, the open pasture contained 1 1 - 14% 
more Lolium perenne than in Zones 1 and 3 (P<0.0001 ). However,' the difference between 
these three main overstorey environments tended to fall in summer due to a reduction of 
Lolium perenne in the open pasture (P=0.0646; Table 5.3).  
Poa and Lolium perenne were the dominant HFR grasses in Zones 1 and 3 (Table 5.3).  
In spring, Zone 1 ,  followed by Zone 3, and then the open pasture, had the greatest Poa 
content (Table 5.3). Between spring and summer, the proportion of Poa in Zone 3 was 
reduced to a level comparable with the open pasture. Similarly, the proportion in Zone 1 
also decreased, but remained significantly greater than in the other two main overstorey 
environments (Table 5.3). 
Chapter 5 2 14  
When averaged over both seasons, Zone 1 contained 4-5% more Holcus lanatus than 
Zone 3 or the open pasture (Table 5.3). This contrasted with Dactylis glomerata, which did 
not vary significantly between the three main overstorey environments (Table 5 .3). 
The sward content of both these species did not change between spring and summer 
(Table 5 .3). 
The main legumes in the swards were Trifolium repens and to a lesser extent various Lotus 
species (Table 5 .3). The proportion of these legumes varied little between the three main 
overstorey environments or two consecutive seasons. Overall, the sward biomass contained 
around 8% legume (Table 5.2). 
5.4.4.2 The effect of poplar overstorey density on Kiwitea pasture composition 
General pasture composition 
In spring, the proportion of high fertility responsive grasses in Zone 1 showed a strong 
(r2=0.59, P=0.0002) positive linear relationship with increasing CeL (Figure 5 .  1 3a). 
Conversely, the proportion of low fertility tolerant grasses within the same zone of tree 
influence decreased at a similar absolute rate (P>0.05), although the strength of this 
relationship was much weaker (r2=0. 3 7, P=0.006) (Figure 5. 1 3b). The rate of change for 
both relationships was approximately 2-4% of the total sward biomass for every 1 0% 
increase in CeL (Figure 5 . 1 3a,b). 
No clear trends were evident in summer for any of the general pasture composition 
categories. The only exception was a weak (r2=0.25, P=0.02) negative linear relationship 
between the dead pasture material content in Zone 3 and increasing eCL (Figure 5 . 1 4a). 
Main high fertility responsive grasses 
Lolium perenne and Poa showed clear trends with increasing CeL (Figure 5 .  1 3c,d). 
In both Zones 1 and 3, the proportion of Lolium perenne in the sward decreased at a 
constant rate of 2.6% ± 0.4% (prediction ± SE) for every 1 0% increase in eCL (Figure 
5 . 1 3c). This contrasted with Poa in Zone 1 ,  which increased at a higher (P:S0.05) rate of 
5.3% ± 0.6% for every 1 0% increase in CeL (Figure 5 . 1 3d) .  Thus, the positive linear 
relationship between the content of HFR grasses in Zone 1 and CCL (Figure 5 . 1 3 a) was, at 
Chapter 5 21 5 
least in part, due to a more than compensatory increase in Poa o ver the decrease in Lolium 
perenne. 
Overall, the Lolium perenne content of the open pasture had disproportionately greater 
variability around the fitted function, compared to the two main zones of tree influence 
amongst trees (Zones 1 & 3) (Figure 5 . 1 3c). This contrasted with the Poa spp. content, 
which was more variable in Zones 1 and 3 (Figure 5. 1 3d). The intercept for the latter 
simple straight-line function was not significantly different from zero (P=O. 1 336; Figure 
5. 1 3d). 
In summer, the proportion of Lolium perenne in Zone 1 also showed a clear negative trend 
with increasing CCL (Figure 5 . 1 4b). However, both the intercept and slope of this 
simple linear relationship were significantly (P:::;O.05) lower than in spring (Figures 5 . 1 3c 
& 5 . 1 4b). 
Chapter 5 21 6 
Table 5.2 Pasture composition at Kiwitea1, expressed as a percentage of total sward biomass (%DM). 
Season Environment HFR LFT Legume Other Dead 
grasses grasses species matter 
(1nx) ("x) 
Spring'98&'99 Open pasture 37 37  (3.6) 9 1 5  3 (3.5) 
Zone 1 60 23 (3.0) 8 6 3 (2.3) 
Zone 3 47 25 (3.0) 1 3  1 3  2 (3.2) 
Grand mean 50 27 (3. 1) 1 0  1 1  3 (2.9) 
Summer'98&'99 Open pasture 24 29 (3.3) 8 1 7  22 (4.0) 
Zone 1 35  27  (3.2) 5 1 1  22 (3.0) 
Zone 3 29 28 (3.2) 6 1 8  19  (3.8) 
Grand mean 30  28 (3.2) 6 1 5  21  (3.5) 
Overall Open pasture 30  33 (3.4) 9 1 6  12 (3.7) 
Zone 1 48 25 (3. 1 )  7 9 12  (2.7) 
Zone 3 38  27  (3. 1 )  10 15  1 1  (3.5) 
Grand mean 40 27 (3.2) 8 13  12 (3.2) 
Analysis of variance 
Environment (A) P=0.05 NS NS NS NS 
Season (B) P<O.OOI NS P<0.05 P<0.05 P<O.OO I 
Interaction (A *B) NS NS NS NS NS 
SD 10 (0.4) 5 7 (0.4) 
) Averaged across the entire range of poplar stand densities over a two year period (1998-2000). Transformed 
values are given in parentheses. Abbreviations: Zone 1, directly below the crowns of individual poplar 
trees; Zone 3, at the centre of the vertically projected gap between poplar crowns; Inx, natural-logarithm 
transformed; -vx, square-root transformed; SD, standard deviation; NS, not significant (1)2:0. 1); DM, dry 
matter; HFR grasses, high fertility responsive grasses (Lolium perenne, Dactylis glomerata, Holcus lanatus, 
and Poa spp.); LIT grasses, low fertility tolerant grasses (Agrostis capillaris, Cynosurus cristatus, and 
Anthoxanthum odoratum). 
Chapter 5 21 7 
Table 5.3 Main high fertility responsive (HFR) grasses and legumes at Kiwiteai, expressed as a percentage 
of total sward biomass (%OM). 
Season Environment Lolium Dactylis Holcus Poa Trifolium Lotus 
perenne glomerata lanatus spp. repens spp. 
(In[x+l ]) CYx) ("Jx) 
Spring'98&'99 Open pasture 32 o (0.03) o (O.3S) 4 (2.04) 8 1 
Zone 1 12  3 (0.92) 6 (2. 12) 38 (S.99) 6 2 
Zone 3 16  3 (0.7S) 1 (0.78) 26 (4.72) 10 3 
Grand mean 1 8  3 (0.66) 3 ( 1 .2 1)  26 (4.64) 8 2 
Summer'98&'99 Open pasture 23 o (0.00) o (0.22) 1 (0.S8) 6 2 
Zone 1 14 4 ( 1 .00) 4 ( 1 .S6) 12 (3.09) 5 1 
Zone 3 1 5  6 ( 1 .02) 1 (0.70) 6 ( 1 .S6) 6 0 
Grand mean 17 4 (0.79) 2 (0.93) 7 ( 1 .9S) 6 1 
Overall Open pasture 27 o (0.01)  o (0.28) 3 ( 1 . 3 1 )  7 2 
Zone 1 1 3  4 (0.96) 5 (1 .84) 25 (4.S4) 5 
Zone 3 16  5 (0.88) I (0.74) 1 6  (3. 1 4) 8 2 
Grand mean 1 7  3 (0.73) 3 (1 .07) 1 7  (3.30) 7 1 
Analysis of variance 
Environment (A) P<O.OO I NS P<O.O I P<O.OI NS NS 
Season (B) NS NS NS P<O.OOI NS NS 
Interaction (A *B) P<O . I  NS NS P<0.05 NS NS 
SO 7 (0.39) (0.63) (0.92) 5 3 
iAveraged across a range of poplar stand densities over a two-year period ( 1998-2000). Transformed values 
are given in parentheses. Abbreviations: Zone 1, directly below the crowns of individual poplar trees; Zone 3 ,  
at the centre of the vertically projected gap between poplar crowns; In[x+ I ] ,  natural-logarithm transformed; 
..Jx, square-root transformed; SO, standard deviation; NS, not significant (�O. I ); DM, dry matter. 
1 00 60 
(a) Y = 30.24 (2.44) - x·0.26 (0.04) 
80 � 50 r
2=O.57, RMSE=7.08, P<O.OOOl 
� ... ... � 40 � 60 '-' '-" <U 
� s:: s:: 30 � � 40 <U I:)., 
� c: 20 :::s '--20 c Y = 36.72 (4. 1 1 )  + x·0.38 (0.08) 0...1 10  
2 0 (c) r =0.59, RMSE=l l .O, P<0.0002 • 
0 0 • 
0 20 40 60 80 1 00 0 20 40 60 80 1 00 
1 00 80 
Y = 37.08 (3.71)  - x·0.23 (0.07) (b) Y = 5.01 (3. 16) + x·0.53 (0.06) 
2 ?=O.82, RMSE=8.49, P<O.OOOl 80 r =0.37, RMSE=9.97, P<O.006 
� � 60 ... ... � 60 � '-' � '-' '" ci. 40 
� 40 ... 0. '" o::l 
5 c 20 � 20 
... ... (d) ... 
0 0 
0 20 40 60 80 1 00 0 20 40 60 80 1 00 
Canopy closure with leaves (%) Canopy closure with leaves (%) 
Figure 5.13 Relationship between spring pasture composition and poplar canopy closure (CCL) at Kiwitea. Symbols: (A)  directly below the 
poplar crowns (Zone 1 )  and (0) at the centre of the vertically projected canopy gap between the trees (Zone 3). Standard errors for simple linear 
regression coefficients are given in parentheses. Abbreviations: r2, adjusted coefficient of determination; RMSE, standard error of prediction; 
HFR grasses, high fertility responsive grasses, LFT grasses, low ferti lity tolerant grasses. 
Chapter 5 21 9 
Figure 5.14 Relationship between the summer pasture composition and poplar canopy closure at Kiwitea. 
Symbols: (A)  directly below the poplar crowns (Zone 1 )  and (0) at the centre of the vertically projected 
canopy gap between the trees (Zone 3). Standard errors for simple linear regression coefficients are given in 
parentheses. Abbreviations: r, adjusted coefficient of determination; RMSE, standard error of prediction. 
Chapter 5 220 
5.4.4.3 Effect of overstorey environment and season on Hautope 1 pasture 
composition 
General pasture composition 
Grasses made up at least 60%, and usually exceeded 80%, of the total sward biomass 
(Table 5 .4). Averaged over the two consecutive seasons, Zones 1 and 3 contained around 
27% more HFR grasses compared to the open pasture (Table 5 .4). Conversely, the 
proportion of LFT grasses in the pastures did not differ significantly amongst the three 
main overstorey environments, but did decrease by about 1 0% between spring and summer 
(Table 5 .4). Both Zones 1 and 3 contained very little legume in their swards (::;1 %). This 
contrasted with the open pasture, which in spring had 1 2% legume (Table 5 .4). Between 
spring and summer, the legume content in the open pasture decreased to 2%, but still 
remained significantly higher than amongst the trees (Zones 1 & 3) (Table 5 .4). The main 
' other species' in the pastures were flat-weeds, such as Plantago lanceolata, Taraxacum 
officinale, and Cerastium glomeratum. Collectively, these species accounted for less than 
6% of the total sward biomass, and were generally in similar abundance amongst the three 
main overstorey environments (Table 5 .4). Dead pasture material also made up only a 
small proportion « 1 0%) of the pasture in spring, irrespective of overstorey environment 
(Table 5 .4). However, between spring and summer, this general category increased by 
30% in the open pasture, and by approximately 7% amongst the trees (Zone 1 & 3) 
(Table 5 .4). 
Main high fertility responsive grasses and legumes 
In the open pasture the main HFR grass was Lolium perenne, followed by Holcus lanatus 
(Table 5 . 5). Amongst the trees (Zone 1 & 3) both Lolium perenne and Poa were co­
dominant (Table 5.5). All three main overstorey environments in spring had about 22% 
Lolium perenne. During the ensuing summer, the proportion of Lolium perenne did not 
change significantly in the open pasture, but did tend to increase by 1 4% and 9% in 
Zones 1 and 3, respectively (P=0.0894; Table 5 .5). Averaged over the two consecutive 
seasons, the open pasture contained 6-7% more Holcus lanatus than in Zones 1 & 3 
(Table 5 . 5). In contrast, the understorey pastures in Zones 1 and 3 had around 23% more 
Poa, with only a trace amount « O . 5%DM) of this species evident in the open pasture 
environment (Table 5.5). Between spring and summer the Poa content decreased (Table 
Chapter 5 221 
5.5). Dactylis glomerata made up only a small proportion of total sward biomass in all 
three main overstorey environments « 5%DM; Table 5.5). 
The two main legume species in the open pasture were Trifolium repens and T dubium; the 
latter was not evident in summer (data not shown). In comparison, over both consecutive 
seasons, only a trace amount (�l %DM) of Trifolium repens was present in Zones 1 and 3 
(Table 5.5). 
5.4.4.4 Effect of poplar overstorey density on Hautope 1 pasture composition 
General pasture composition 
In both spring and summer, the proportion ofHFR grasses amongst the trees (Zones 1 & 3) 
increased exponentially with increasing eeL (Figure 5. 1 5). However, this relationship was 
based on a very limited range of eeLs and accounted for only 40% of the variation shown 
by this general category (Figure 5. 1 5). 
In summer, the proportion of dead pasture material in the sward was negatively related to 
eeL (Figure 5 . 16). Amongst the trees (Zones 1 & 3), the dead material content of the 
sward decreased at a constant rate of 3% ± 0.5% for every 1 0% increase in eeL 
(Figure 5 . 16) .  The strength of this relationship was moderate (r=o.56; P<O.OOOI ). 
Chapter 5 222 
Table 5.4 Pasture composition at Hautope 1 1, expressed as a percentage of total sward biomass (%OM). 
Season Environment HFR LFf Legume Other Dead 
grasses grasses species matter 
C..Jx) ( l/[x+lD (lnx) 
Spring' 99 Open pasture 30 52 12 (3.35) 2 (0.40) 4 (1 .40) 
Zone 1 52 40 o (0.37) o (0.78) 7 (1 . 83) 
Zone 3 55 35 1 (0.57) 5 (0.75) 5 (1 .52) 
Grand mean 50 40 2 (0.95) 2 (0.7 1)  6 ( 1 .58) 
Summer'99 Open pasture 25 36 2 ( 1 .22) 2 (0.5 1 )  34 (3.48) 
Zone 1 60 26 o (0.34) o (0.86) 13 (2.43) 
Zone 3 56 29 o (0.30) 1 (0.72) 14 (2.54) 
Grand mean 52 29 1 (0.47) 1 (0.74) 17 (2.82) 
Overall Open pasture 28 44 7 (2.29) 2 (0.46) 19 (2.44) 
Zone 1 56 33 o (0.36) o (0.82) 1 0  (2. 1 3) 
Zone 3 55 32 1 (0.44) 3 (0.73) 10 (2.03) 
Grand mean 5 1  34 2 (0.71 )  2 (0.72) 1 1  (2. 1 4) 
Analysis of variance 
Environment (A) P<O.O I  NS P<O.OOI P<O. 1 NS 
Season (B) NS P<O.OI  P<O.O I NS P<O.OO I 
Interaction (A *B) NS NS P<O.O I  NS P<O.O I 
SO 1 2  1 3  (0.69) (0.22) (0.46) 
IAveraged across the entire range of poplar stand densities over a one-year period ( 1 999-2000). Transformed 
values are given in parentheses. Abbreviations: Zone 1 ,  directly below the crowns of individual poplar trees; 
Zone 3, at the centre of the vertically projected gap between poplar crowns; lnx, natural-logarithm 
transformed; -'/x, square-root transformed; lI[x+l ], inverse transformed; SO, standard deviation; NS, not 
significant (�0. 1 ); OM, dry matter; HFR grasses, high fertility responsive grasses (Lolium perenne, Dactylis 
glomerata, Holcus lanatus, and Poa spp.); LIT, low fertility tolerant grasses (Agrostis capillaris, Cynosurus 
cristatus, and Anthoxanthum odoratum). 
Chapter 5 223 
Table 5.5 Main high fertility responsive grasses and legume species at Hautope 1 1, expressed as a percentage 
of total sward biomass (%DM). 
Season Environment Lolium Dactylis Holcus Poa Trifolium 
perenne glomerata lanatus spp. repens 
( lI[x+ l D  (xl13) 
Spring' 99 Open pasture 22 1 8 (0.22) o (0.23) 4 
Zone 1 20 3 1 (0.83) 29 (2.97) 0 
Zone 3 24 2 o (0.86) 28 (2.92) 
Grand mean 22 2 2 (0.74) 24 (2.04) 1 
Summer'99 Open pasture 1 7  7 (0. 1 9) o (0. 1 0) 2 
Zone 1 34 4 3 (0.58) 19 (2.47) 0 
Zone 3 33 2 2 (0.79) 1 8  (2.34) 0 
Grand mean 3 1  3 3 (0.60) 15  ( l .64) 1 
Overall Open pasture 1 9  1 8 (0.2 1 )  o (0. 1 6) 3 
Zone 1 27 4 2 (0.70) 24 (2.72) 0 
Zone 3 29 2 1 (0.83) 23 (2.63) 
Grand mean 26 3 3 (0.67) 20 (2.26) 
Analysis of variance 
Environment (A) NS NS P<O.O l P<O.OO I P<O.OO I 
Season (B) P<O. l NS NS P<0.05 NS 
Interaction (A *B) P<O. 1 NS NS NS NS 
SD 9 5 (0.27) (0.52) 
1Averaged across the entire range of poplar stand densities over a one-year period (1999-2000). Transformed 
values are given in parentheses. Abbreviations: Zone 1, directly below the crowns of individual poplar trees; 
Zone 3 ,  at the centre of the vertically projected gap between poplar crowns; ( l I[x+l ]), inverse transformed; 
x1l3, cubic-root transformed; SD, standard deviation; NS, not significant (1'20. 1); DM, dry matter. 
Chapter 5 224 
1 00 
2 Y = 28.7 (5.4)*exp
O.OO9 (O.OO3)"x 
Cl �=0.4, RMSE= 1 2.2, P=0.OOO5 
'$. 80 '-' ·0 ell Q) Q ell 
� • 60 Q) 
.� ell C 0 c.. 40 ell Q) 
... 
. .q 
'€ 
20 � 
� 
::c 
0 
0 20 40 60 80 
Canopy closure with leaves (%) 
Figure 5.15 Relationship between HFR grasses and poplar canopy closure (CCL) at Hautope 1 in spring and 
summer: (.&) directly below the poplar crowns (Zone 1 ), and (0) at the centre of the vertically projected 
canopy gap between the trees (Zone 3 ). Standard errors for regression coefficients are given in parentheses. 
Abbreviations: r, adjusted coefficient of determination (approximated for the non-linear function; refer to 
Section 4.3); RMSE, standard error of prediction. 
50 
r--
40 ::is Cl '$. 
'-' 
ca 30 ·c Q) 
g 
Q) !3 20 -ell «l c.. 
"0 «l Q) 10 Cl 
0 
0 20 
Y = 33.7 (3 .3) - x*0.3 1 (0.05) 
r2=0.56, RMSE=7.0, P<O.OOO l 
40 
o 0 
o 
60 
� o 
Canopy closure with leaves (%) 
80 
Figure 5.16 Relationship between dead pasture material and poplar canopy closure (eCL) at Hautope 1 in 
summer: (.&)  directly below the poplar crowns (Zone 1 ), and (0) at the centre of the vertically projected 
canopy gap between the trees (Zone 3). Standard errors for regression coefficients are given in parentheses. 
Abbreviations: r2, adjusted coefficient of determination; RMSE, standard error of prediction. 
Chapter 5 225 
5.4.4.5 Effect of overstorey environment and season on Hautope 2 pasture 
composition 
General pasture composition 
Overall, LFT grasses made up over 75% of the total sward biomass and thus, dominated 
the pastures at this site (Table 5 .6). In spring, Zone 1 contained a significantly higher 
proportion of LFT grasses compared to the open pasture. Nevertheless, in summer no 
significant differences were evident amongst the three main overstorey environments 
(Table 5 .6). HFR grasses, legumes, and other species collectively made up less than 1 5% 
of the total sward biomass; and their proportions in the pasture varied little amongst the 
three main overstorey environments or over the two consecutive seasons (Table 5.6). 
The legume species included Trifolium subterraneum, T. repens and T. dubium, while the 
HFR grasses consisted almost entirely of Lolium perenne, except for trace amounts of 
Holcus lanatus and Poa « 1  %DM; data not shown). All three main overstorey 
environments in spring had a similar proportion of dead pasture material (Table 5 .6). 
Between spring and summer the dead pasture material content in Zones 1 and 3 increased 
by 9-1 5%, while not changing significantly in the open pasture. As a result, in summer 
Zone 1 had a significantly higher dead pasture material content than in the open pasture 
(Table 5 .6). 
No significant relationships between the pasture composition categories and CCL were 
detected at this site. 
Chapter 5 226 
Table 5.6 Pasture composition at Hautope 21, expressed as a percentage of total sward biomass (%DM). 
Season Environment HFR LFT Legume Other Dead 
grasses grasses species matter 
(In[x+ 1]) 
Spring' 99 Open pasture 6 ( 1 .33) 72 5 0 1 7  
Zone 1 4 ( 1 .30) 87 0 0 9 
Zone 3 2 (0.58) 84 0 0 1 4  
Grand mean 4 ( 1 .07) 8 1  2 0 1 3  
Summer'99 Open pasture 4 ( 1 . l7) 80 1 0 1 5  
Zone 1 7 ( 1 .84) 68 0 1 24 
Zone 3 4 ( 1 . l7) 73 0 1 23 
Grand mean 5 ( 1 .39) 74 0 0 2 1  
Overall  Open pasture 5 ( 1 .25) 76 3 0 1 6  
Zone 1 5 ( 1 .57) 78 0 0 1 7  
Zone 3 3 (0.87) 78 0 0 1 8  
Grand mean 4 ( 1 .23) 77 1 0 1 7  
Analysis of variance 
Environment (A) NS NS NS NS NS 
Season (B) NS P<0.05 NS NS P<O.OI 
Interaction (A *B) NS P<O.05 NS NS P<0.05 
SD (0.84) 7 3 0.5 5 
I Averaged across the entire range of poplar stand densities over a one-year period ( 1999-2000). Natural 
logarithm transformed (1n[ x+ I ]) values are given in parentheses. Abbreviations: Zone I ,  directly below the 
crowns of individual poplar trees; Zone 3 ,  at the centre of the vertically projected gap between poplar 
crown.s; DM, dry matter; HFR grasses, high fertility responsive grasses (Lolium perenne, Dactylis giomerata, 
Hoicus ianatus, and Poa spp.); LIT grasses, low fertility tolerant grasses; SD, standard deviation; NS, not 
significant W�O. l ). 
Chapter 5 
5.4.5 Pasture nutritive values (proximate analysis) 
227 
5.4.5.1 Effect of overstorey environment and season on Kiwitea pasture nutritive 
values 
Dry matter 
In spring, the dry matter (DM) content of the freshly cut (wet) pasture amongst the trees 
(Zones 1 & 3) was on average 5 .2% lower than in the open pasture (Table 5.7). Between 
spring and summer, there was little change in the DM content of the open pasture, 
whereas, in Zones 1 and 3 it increased to a level comparable with the open pasture 
environment (Table 5.7). 
Organic matter digestibility and related metabolisable energy 
In spring, the organic matter digestibility (OMD) of the pasture in the three mam 
overstorey environments was similar (Table 5 .7). However, between spring and summer all 
three decreased, with a greater degree of reduction in the open pasture. Thus, in summer 
the OMD of the understorey pasture amongst the trees (Zones 1 & 3) was on average 5 .9% 
greater than in the open pasture (Table 5.7). The differences in the metabolisable energy 
(ME) content of the pastures mirrored those for OMD (Table 5 .7). 
Crude protein 
Crude protein (CP) in the pastures decreased between spring and summer, irrespective of 
the overstorey environment (Table 5.7). Overall, Zones 1 and 3 tended to contain 1 .6-2. 1%  
more CP than the open pasture (P=O.08 1 5; Table 5.7). 
Neutral detergent fibre 
In general, the variation in neutral detergent fibre (NDF) across the three main overstorey 
environments and two consecutive seasons was not large (�6% DM; Table 5 .7). However, 
the concentration of NDF in the pasture tended to depend on an interaction between the 
overstorey environment and season (P=O.0747; Table 5.7) .  In spring, Zone 1 had 4.6% 
more NDF than in the open pasture; whereas, the concentration in Zone 3 was not 
significantly different from either of the other two main overstorey environments 
(Table 5 .7). Between spring and summer, the NDF in Zones 1 and 3 did not change 
significantly, whereas in the open pasture it increased to a level comparable with the 
other two overstorey environments (Zones 1 & 3) (Table 5.7). Unlike in spring, the 
concentration of NDF in Zone 3 in summer was marginally (but significantly) lower than 
in Zone 1 (Table 5.7). 
Chapter 5 228 
Acid detergent fibre 
Overall, Zone 1 had 1 .6% and 2.4% more acid detergent fibre (ADF) than in Zone 3 and 
the open pasture, respectively (Table 5 .7). Averaged across all three main overstorey 
environments, the concentration of ADF in the pasture also slightly increased over the two 
consecutive seasons (Table 5.7) .  
Soluble carbohydrate 
Overall, the concentration of soluble carbohydrate (Sol CHO) was greatest in the open 
pasture, followed by Zone 3, and then Zone 1 (Table 5.7). Nevertheless, there tended to be 
an interaction between the overstorey environment and season (P=0.0848). In spring, the 
open pasture contained 2.4%-3.8% more Sol CHO than amongst the trees (Zones 1 & 3); 
whereas, at the same time, Zone 3 tended to have a greater concentration of Sol CHO than 
Zone 1 (P=0.0533). In summer, the open pasture continued to have a greater Sol CHO 
level than in Zone 1 .  However, the concentration in Zone 3 was not significantly different 
from either ofthe other two main overstorey environments (Table 5 .7). 
Lipid 
Averaged across the three main overstorey environments, the amount of lipid in the pasture 
decreased by 0.6% between spring and summer (Table 5.7). There also tended to be an 
overstorey environment by season interaction (P=O.0828). In spring, Zones 1 and 3 had 
0.4% more lipid (on average) than in the open environment. However, between spring and 
summer the lipid content in all three main overstorey environments decreased to a similar 
level (Table 5 .7). 
Ash 
Overall, the ash content in Zone 3 was marginally greater than in the open pasture; 
whereas, in Zone 1 the ash content was not significantly different from either of the other 
two main overstorey environments (Table 5.7). 
Dietary anion-cation difference 
Averaged across the three mam overstorey environments, the dietary anion-cation 
difference (DACD) of the pasture decreased by 28% between spring and summer 
(Table 5.7). 
Table 5.7 Nutritive value indices for the mixed-species pasture' at Kiwitea: directly below the crowns of individual trees (Zone 1 ), at the centre of the vertically projected gap 
between trees (Zone 3), and in the open pasture. 
Season Environment DM CP Lipid Ash ADF NDF Sol CHO DCAD OMD ME 
(%) (% DM) (% DM) (% DM) (% DM) (% DM) (% DM) (meq/kg DM) (% DM) (MJ/kg DM) 
Spring'98&'99 Open pasture 2 1 . 8  20.0 4.6 12.0 26.6 45.5 6.7 537 74.4 1 0.7 
Zone 1 16.3 2 1 .9 5 . 1  1 1 .6 28.9 50. 1 2.9 530 72.9 1 0.5  
Zone 3 16.9 2 1 . 8  4.9 12.6 2S. 1 47.5 4.3 560 73 . 1  1 0.5 
Grand mean I S.4 2 1 .2 4.S 12 . 1  27.9 47.7 4.6 542 73.5 10.6 
Summer'99&'00 Open pasture 24. 1  1 5.8 4. 1 1 0.7 29.0 49.9 5 .9 356 60.S S .S 
Zone 1 25.4 17 . 1  4.3 I 1 .S 3 1 .6 50.9 4.2 405 66.4 9.5 
Zone 3 24.3 l S.2 4.3 12.0 29.2 47.2 5.3 400 67.0 9.6 
Grand mean 24.6 17.0 4.2 1 1 .5 29.9 49.3 5 . 1  3S7 64.7 9.3 
Overall Open pasture 23.0 1 7.9 4.4 1 1 .3 27.S 47.7 6.3 447 67.6 9.S 
Zone 1 20.S 1 9.5 4.7 1 1 .7 30.2 50.5 3.6 467 69.6 1 0.0 
Zone 3 20.6 20.0 4.6 12.3 2S.6 47.3 4.S 480 70.0 10.0 
Grand mean 2 1 .5 1 9. 1  4.5 l 1 .S 2S.9 4S.5 4.9 465 69. 1  9.9 
Analysis of variance 
Environment (A) NS P<O. l NS P<0.05 P<O.Ol  NS P<0.05 NS NS NS 
SEM (A) 0.6 0.5 0.05 0.3 0.3 0.6 0.3 1 1 .8  0.8 0.09 
Season (B) P<O.OO I P<O.OOl P<O.OOl NS P<O.OO l P<O. 1 NS P<O.OO l P<O.OO l P<O.OOI 
SEM (B) 0.5 0.4 0.05 0.2 0.2 0.5 0.2 9.6 0.6 O.OS 
Interaction (A *B) P<O.O l NS P<O. 1 NS NS P<O. 1 P<O. 1 NS P<0.05 P<O.Ol 
SEM (A*B) 0.9 0.7 O.OS 0.4 0.4 0.9 0.4 1 6.7 1 . 1  0. 13  
'Averaged across the entire range of poplar stand densities over a two-year period ( 1 99S-2000). Abbreviations: SEM, standard error ofthe mean; NS, not significant (1)2:0. 1 ); 
DM, dry matter; CP, crude protein; ADF, acid detergent fibre; NDF, neutral detergent fibre; Sol CHO, soluble carbohydrate; DCAD, dietary cation-anion difference; OMD, in 
vitro organic matter digestibility; ME, metabolisable energy. !\.) !\.) 
CD 
Chapter 5 230 
5.4.5.2 The effect of poplar overstorey density on Kiwitea pasture nutritive values 
In spring, the DM content of the freshly cut (wet) pasture decreased at a constant rate of 
0.7% ± 0. 1%  for every 1 0% increase in CCL (Figure 5 . l 7a). Similarly, the Sol CHO 
concentration of the pasture also decreased with increasing CCL (Figure 5 . 1 7b). However, 
based on the limited spread of data points, the fonn of the latter relationship was 
exponential as opposed to linear. For both nutritive (feed) value indices, their relationships 
with CCL were not significantly different (P>0.05) between the two main zones of tree 
influence (Zones 1 & 3) (Figure 5 . 1 7a, b). 
Conversely, the CP, lipid, NDF, and ADF contents of the pasture increased marginally 
with increasing CCL (Figure 5 .  1 7c,d,e,f). The relationships for NDF and ADF were 
significant only in Zone 1 (Figure 5 . 1 7  d,e), whereas, for CP the relationship was 
significant only in Zone 3 (Figure 5. 1 7c). Similar (P>0.05) relationships occurred for lipid 
in both of the main zones of tree influence (Zones 1 & 3) amongst the poplar stands 
(Figure 5. 1 7f). 
In summer, the rate of change in Zone 1 ADF was similar (P>0.05) to that observed in 
spring (Figures 5. 1 7  e & 5 . 1 8c). The OMD and related ME content of the pasture were also 
positively related to increasing CCL (Figure 5 . 1 8a, b); however, the maximum variation in 
these nutritive value indices was less than 8% DM and 1 MJ/kg DM, respectively, over the 
entire measured range of canopy covers (Figure 5 . 1 8a,b). 
24 24 30 
Y = 21 .5 (0.6) - x*0.07 (0.01) (c) (e) 
22 r2=O.81 ,  RMSE=1 .2, P<O.OOOl 0 29 22 
'0' 20 2 � 28 a::: � 20 '-' a::: � '-' '-' � 1 8  1J 27 
1 6  
1 8  
Y = 20.0 (0.3) + x*0.028 (0.007) 
26 Y = 26.8 (0.5) + x*0.03 (0.01) 
(a) r2=O.68, RMSE=O.68, P=O.007 r2=O.5 1 ,  RMSE=1 .0, P=O.03 
14  16  25 
0 20 40 60 80 100 0 20 40 60 80 100 0 20 40 60 80 100 
1 0  54 5.4 
Y = 6.6 (0.6)*exp·O.0095 (O.OO23)*x (d) (f) 
r2=0.61 ,  RMSE=1 .2, P=O.002 52 ... 5.2 ... 8 
2 2 
50 0 
� 6 2 5.0 '-' @ 48 @ 
a 
a::: � 4.8 '-' 
� 46 "0 4 's.. '0 44 � 4.6 r/J ... 0 2 Y = 45.7 ( 1 .2) + x*0.07 (0.03) 4.4 Y = 4.60 (0.08) + x*0.006 (0.001 )  42 
(b) r2=O.44, RMSE=2.5, P=O.04 r2=O.57, RMSE=O. l 6, P=O.003 
0 40 4.2 
0 20 40 60 80 100 0 20 40 60 80 100 0 20 40 60 80 100 
Canopy closure with leaves (%) Canopy closure with leaves (%) Canopy closure with leaves (%) 
Figure 5.17 Spring relationships between pasture nutritive (feed) value indices and poplar canopy closure (CCL) at Kiwitea: ( . )  directly below the crowns of individual trees (Zone 
I ), and (0) at the centre ofthe vertically projected canopy gap between trees (Zone 3). Standard errors for regression coefficients are given in parentheses. Abbreviations: r2, adjusted 
coefficient of determination; RMSE, standard error of prediction; DM, dry matter content; Sol CHO, soluble carbohydrates; CP, crude protein; NDF, neutral detergent fibre; ADF, 
acid detergent fibre. 
I\) 
W � 
75 34 
(a) (c) 
70 0 
'0' 0
'" 
a;:: 
� 65 
32 
� a;:: 30 ......, 
� � 
60 28 
Y = 60.8 ( 1 .0)*expO.0014 (O.OOO3)*x Y = 29. 1  (0.5) + x*0.04 (0.01 )  
?=0.66, RMSE=2.2, P=O.OO08 
55 
?=O.66, RMSE=I .O, P=O.009 
26 
0 20 40 60 80 1 00 0 20 40 60 80 1 00 
10.5 Canopy closure with leaves (%) 
(b) 
1 0.0 0 
� 0'"  9.5 
� 9.0 � 
8.5 Y = 8.8 (0. 1 )  + x*O.O l l (0.003) 
?=0.62, RMSE=O.3, P=O.OOl 
8.0 
0 20 40 60 80 100 
Canopy closure with leaves (%) 
Figure 5.18 Summer relationships between pasture nutritive (feed) value indices and poplar canopy closure (CCL) at Kiwitea: (A)  directly below the crowns of individual 
trees (Zone 1), and (0) at the centre of the vertically projected canopy gap between trees (Zone 3). Standard errors for regression coefficients are given in parentheses. 
Abbreviations: r2, adjusted coefficient of determination; RMSE, standard error of prediction; OMD, in vitro organic matter digestibility; ME, metabolisable energy; ADF, acid 
detergent fibre. 
I\) 
W 
I\) 
Chapter 5 233 
5.4.5.3 Effect of overstorey environment and season on Hautope 1 pasture nutritive 
values 
Dry matter 
Overall, the DM content of the freshly cut (wet) pasture amongst the trees (Zones 1 & 3) 
was on average 4.7% lower than in the open pasture (Table 5 .8). In spring, the DM content 
of the pastures was very high at above 80%, whereas, in summer this decreased markedly 
to around 20% (Table 5.8). 
Organic matter digestibility and related metabolisable energy 
In spring, the OMD of the pasture was similar for all three main overstorey environments 
(Table 5 .8). However, over the ensuing season, while the OMD of the open pasture 
decreased slightly by 8%, it increased (Zone 1 )  and tended to increase (p=0.0870; Zone 3) 
by on average 6% amongst the trees (Table 5 .8). 
The differences in ME mirrored those for OMD. Thus, the three mam overstorey 
environments were not significantly different in spring, but the understorey pasture 
(Zones 1 & 3) had a greater ME content compared to the open pasture in summer 
(Table 5 . 8). This divergence was caused mainly by a drop in the ME content of the open 
pasture between spring and summer, as opposed to an increase amongst the poplar trees 
(Zones 1 & 3) (Table 5 .8). 
Crude protein 
Overall, the understorey pasture amongst the trees (Zones 1 & 3) had 4% more ep than in 
the open pasture (Table 5 .8). When averaged across the three main overstorey 
environments, the ep in the pastures also increased by 4.8% over the two consecutive 
seasons (Table 5 . 8). 
Neutral detergent fibre 
In spring, the open pasture had on average 4.7% less NDF compared to the understorey 
pasture (Zones 1 & 3) (Table 5. 8). However, between spring and summer, the NDF content 
in the open pasture increased by 4%, whereas, it decreased in Zones 1 and 3 by on average 
6 . 1  %. Thus, the difference in NDF concentration, between the three main overstorey 
environments, was reversed in summer (Table 5 . 8). 
Chapter 5 234 
Acid detergent fibre 
In spring, the level of ADF in the pastures was similar for all three main overstorey 
environments (Table 5.8). Between spring and summer, the ADF content in the open 
pasture did not change significantly. This contrasted with amongst the trees (Zones 1 & 3), 
which decreased by on average 3 .6% (Table 5.8). 
Soluble carbohydrate 
Overall, the open pasture had 2.9% more Sol CRO than in Zones 1 and 3 (Table 5 . 8). 
When averaged across the three main overstorey environments, the concentration of Sol 
CRO in the pastures slightly decreased over the two consecutive seasons (P=O.05 1 4; Table 
5.8). 
Lipid 
Between spring and summer the concentration of lipid in the pastures increased by on 
average 1 .3% (Table 5 . 8). Nevertheless, an interaction tended to occur between the 
overstorey environment and season (P=O.0843). In spring, all three main overstorey 
environments contained approximately 3 .4% lipid. Over the ensuing summer season this 
concentration increased further, but was especially marked in Zones 1 and 3 .  Thus, in 
summer, the understorey pasture directly below the tree crowns (Zone 1 )  and at the centre 
of the inter-canopy gap (Zone 3) had, and tended (p=O.0586) to have, respectively, a 
greater concentration of lipid compared to the open pasture (Table 5 . 8). 
Ash 
In spring, the ash content of the pastures was not significantly different between the three 
main overstorey environments (Table 5.8). Generally, the amount of ash produced 
increased over the two consecutive seasons, but was again especially marked in Zones 1 
and 3 .  As a result, the understorey pastures (Zone 1 &3) contained on average 1 . 8% more 
ash than in the open pasture in summer (Table 5.8). 
Dietary anion-cation difference 
In spring, the DACD of the pastures in all three main overstorey environments was similar 
(Table 5 . 8). Between spring and summer, the DACD in the open pasture decreased by 
1 8%, while in Zone 1 it increased by 1 7%. No significant change occurred in Zone 3 .  
Thus, overall in summer, the DACD o f  the understorey pastures (Zone 1 & 3 )  was on 
average 3 1  % greater than in the open pasture (Table 5.8). 
Table 5.8 Nutritive value indices for mixed-species pasture I at Hautope I :  directly below the crowns of individual poplar trees (Zone 1), at the centre of the vertically 
projected gap between the poplar crowns (Zone 3), and in the open pasture. 
Season Environment DM CP Lipid Ash ADF NDF Sol CHO DCAD OMD ME 
(%) (% DM) (% DM) (% DM) (% DM) (% DM) (% DM) (meq/kg DM) (% DM) (MJ/kg DM) 
(lnx) 
Spring'99 Open pasture 87.2 12.2 3.5 10.2 32.2 54. 1 8.2 53 1 66.2 9.4 (2.24) 
Zone 1 84. 1  14.8 3 .5 10.5 33.5 58.6 4 . 1  496 63.7 9.0 (2.20) 
Zone 3 83.2 13 .2 3 .2 10.4 34.1  59.0 5 . 1  546 64.5 9.2 (2.21) 
Grand mean 84.9 1 3 .4 3.4 10.4 33.2 57.2 5 .8 524 64.8 9.2 (2.22) 
Summer'OO Open pasture 25.3 14.0 4.4 1 1 .3 33 . 1  58. 1 6.4 438 58.3 8.1 (2.09) 
Zone 1 19.2 20.6 5.0 1 3 . 1  29.7 5 1 .6 4.2 579 7 1 .3 9.7 (2.27) 
Zone 3 19.3 19.9 4.9 1 3 . 1  30.7 53.8 4.2 570 69. 1 9.4 (2.24) 
Grand mean 2 1 .3 1 8.2 4.7 12.5 3 1 .2 54.5 4.9 529 66.2 9. 1 (2.20) 
Overall Open pasture 56.2 1 3 . 1  3 .9 10.7 32.6 56. 1 7.3 484 62.3  8.8 (2. 1 6) 
Zone 1 5 1 .7 1 7.7 4.3 1 1 .8 3 1 .6 55. 1 4 . 1  537 67.5 9.4 (2.23) 
Zone 3 5 1 .3 16.6 4.0 1 1 .8 32.4 56.4 4.6 558 66.8  9.3 (2.23) 
Grand mean 53 . 1  1 5 .8 4. 1 1 1 .4 32.2 55.8 5 .3  527 65.5 9. 1 (2.21) 
Analysis of variance 
Environment (A) P<O.O I  P<0.05 NS P<O.OO l NS NS P<0.05 P::;0.05 P<O. l NS 
SEM (A) 1 .2  0.8 0. 1 0.2 0.5 0.8 0.3 1 8  1 .2 (0.02) 
Season (B) P<O.OO I P<O.OO I P<O.OOI P<O.OO I P<O.O I  P<0.05 P=0.05 NS NS NS 
SEM (B) 1 .0 0.6 0. 1 0. 1 0.4 0.6 0.3 1 5  1 .0 (0.01)  
Interaction (A *B) NS NS P<O. 1 P<0.05 P<0.05 P<O.O I NS P<O.05 P<O.O l  P<0.05 
SEM (A*B) 1 .7 1 . 1  0.2 0.3 0.7 1 . 1  0.5 25 1 .7 (0.03) 
IAveraged across the entire range of poplar stand densities over a one-year period (1999-2000). Transformed values are given in parentheses. Abbreviations: lnx, natural-
logarithm transformed; SEM, standard error of the mean; NS, not significant (�0. 1); DM, dry matter; CP, crude protein; ADF, acid detergent fibre; NDF, neutral detergent N 
fibre; Sol CHO, soluble carbohydrates; DCAD, dietary cation-anion difference; OMD, in vitro organic matter digestibility; ME, metabolisable energy. W 01 
Chapter 5 236 
5.4.5.4 Effect of poplar overstorey density on Hautope 1 pasture nutritive values 
In spring, the Sol CHO concentration in the pasture decreased at a constant rate of 0.6 % 
DM ± 0. 1% DM for every 10% increase in CCL (Figure S . 1 9a). Both of the main zones of 
tree influence (Zones 1 & 3) had similar (P>O.OS) negative linear relationships, and the 
overall strength of the combined relationship was strong (r2=0.76; P=O.OOO I) .  In Zone 1 ,  
the DCAD of the understorey pasture also decreased marginally with increasing CCL 
(Figure S . 1 ge). However, the strength of this negative linear relationship was moderate 
(r2= 0.44; P=0.04). 
Conversely, the CP, NDF, and ADF in the pasture were positively related to CCL (Figure 
5 . 19b,c,d). These nutritive value indices increased at 0.4 ± 0. 1% DM, 0.7 ± 0 . 1% DM, and 
0.2 ± 0. 1 % DM, respectively, for every 10% increase in CCL. However, the rate of change 
relative to the open pasture level was at least twice as large for CP compared to the other 
two indices (Figure 5 . 1 9b,c,d). In addition, the relationships for CP and ADF were 
restricted to Zone 1 (Figure S . 1 9b,d); whereas, similar (P>O.OS) relationships for NDF 
occurred in both Zones 1 and 3 (Figure 5 . 19c). No clear trends were evident for DM, 
OMD, ME, lipid, or ash in the spring pastures. 
In summer, all of the nutritive value indices, except for the DM, Sol CHO, and lipid 
contents, showed significant relationships with CCL (Figure 5.20). The OMD, ME, ash, 
and DCAD in the pasture increased at a constant rate of2. 1 ± 0.3% DM, 0.3 ± 0.04 MJ/kg 
DM, 0.27 ± 0.06% DM, and 24 ± 5 meq/kg DM, respectively, for every 1 0% increase in 
CCL (Figures 5.20a,b,f & 5 .2 1 ). Similarly, the concentration of CP also increased with 
increasing CCL (Figure 5.20c). However, based on the very limited spread of data points 
the form of this relationship was exponential as opposed to linear. In the open pasture the 
predicted CP content was 14% DM, increasing to 22% at maximum measured CCL 
(Figure 5 .20c). Conversely, the level of NDF and ADF in the pasture decreased at a 
constant rate of 1 .0 ± 0.2% DM and 0.6 ± 0 . 1  % DM, respectively, for every 1 0% increase 
in CCL (Figure 5 .20d,e). In general, for each of the above nutritive value indices, the 
relationships in both main zones of tree influence (Zones 1&3) were similar (P>0.05) 
(Figures 5.20 & 5.21).  
1 0  64 600 
(a) (c) (e) 
8 62 0 :i 550 :i :i 60 
0 
� 6 i 500 '-' @ @ � 58 4 � '-' U � ... ... '0 ... 56 tZl � 450 
2 
Y = 8. 1 (0.5) - x·0.06 (0.01)  54 Y = 54.4 (0.8) + x·0.07 (0.0 1 )  Y = 530 (10) - x·0.6 (0.2) 
r2=O.76, RMSE=l .O, P=O.OOO l r2=O.64, RMSE=1 .6, P=O.OOl r2=O.44, RMSE=20.0, P=O.04 
0 52 400 
0 20 40 60 80 0 20 40 60 80 0 20 40 60 80 
1 8  35 Canopy closure with leaves (%) 
(b) (d) 
1 6  
34 ... 
... 
:i :i 33 
� 1 4  � � '-' 
� � 32 
1 2  
Y = 12.3 (0.4) + x·0.04 (0.0 1)  
3 1  
Y = 32.2 (0.3) + x·0.02 1 (0.008) 
r2=O.74, RMSE=0.8, P=O.OO4 r2=O.45, RMSE=O.7, P=O.04 
1 0  30 
0 20 40 60 80 0 20 40 60 80 
Canopy closure with leaves (%) Canopy closure with leaves (%) 
Figure 5.19 Spring relationships between pasture feed value indices and poplar canopy closure (CCL) at Hautope 1 :  ( . )  directly below the crowns of individual trees (Zone 1 ), and 
(0) at the centre of the vertically projected canopy gap between trees (Zone 3). Standard errors for regression coefficients are given in parentheses. Abbreviations: r, adjusted 
coefficient of determination; RMSE, standard error of prediction; CP, crude protein; ADF, acid detergent fibre; NDF, neutral detergent fibre; Sol CHO, soluble carbohydrates; 
I\.) 
DCAD, dietary cation-anion difference. � 
SO 24 36 
(a) (c) � 6 (e) 
75 � �  22 34 
2 70 
20 
� 32 � @ I S  � 65 � � '-' 
� 
'-' 16  � fJ 30 60 14 � o 
55 Y = 57.7 ( 1 .3) + x*0.21 (0.03) Y = 13.7 ( 1 .0)"'expO.0067 (O.OO1 3)*x 
2S Y = 33.5 (0.7) - x"'0.06 (0.01 )  � 12  
r2=O.S6, RMSE=2.7, P<O.OOOI r2=O.71 ,  RMSE=2.2, P=0.0003 r2=O.58, RMSE=1 .5, P=O.002 
50 10  26 
0 20 40 60 SO 0 20 40 60 SO 0 20 40 60 80 
1 1  65 1 5  
(b) (d) (f) 
o � 14 � 
� 1 0  60 � � 13  
� 9 es, 55 � '-' � � 1 2  � 8 50 � o � 1 1  Y = S.O (0.2) + x*0.027 (0.004) Y = 5S.5 ( 1 . 1) - x*O. l O  (0.02) Y == 1 1 .4 (0.3) + x*0.027 (0.006) 
r2=O.S4, RMSE=O.4, P<O.OOOl r2=O.66, RMSE=2.2, P==O.OOOS r2=O.66, RMSE=O.6, P=O.OOOS 
7 45 10 
0 20 40 60 SO 0 20 40 60 SO 0 20 40 60 SO 
Canopy closure with leaves (%) Canopy closure with leaves (%) Canopy closure with leaves (%) 
Figure 5.20 Summer relationships between pasture feed value indices and poplar canopy closure (CCL) at Hautope 1 :  (�) directly underneath the crowns of individual trees 
(Zone 1 ), and (0) at the centre of the vertically projected canopy gap between trees (Zone 3). Standard errors for regression coefficients are in parentheses. Abbreviations: 
r2, adjusted coefficient of determination; RMSE, standard error of prediction; CP, crude protein; ADF, acid detergent fibre; NDF, neutral detergent fibre; OMD, in vitro organic 
I'J 
matter digestibility; ME, metabolisable energy. � 
Chapter 5 
700.-------------------------------------� 
:? 600 
Cl 
01) 
� � 500 
'-' 
� 400 
Y = 434 (27) + x*2.4 (0.5) 
1=0.63, RMSE=55, P=0.001 
300 +---------r--------,--------�--------� 
o 20 40 60 80 
Canopy closure with leaves (%) 
239 
Figure 5.21 Summer relationship between the dietary cation-anion difference (DCAD) and poplar canopy 
closure (CCL) at Hautope 1 :  ( .. ) directly below the crowns of individual trees (Zone 1 ), and (0) at the centre 
of the vertically projected canopy gap between trees (Zone 3). Standard errors for regression coefficients are 
given in parentheses. Abbreviations: r, adjusted coefficient of determination; RMSE, standard error of 
prediction. 
Chapter 5 
5.4.6 Poplar leaf nutritive values 
5.4.6.1 Seasonal effects at Kiwitea 
240 
The DM content of the fresh (wet) poplar leaves increased from spring through until 
autumn (Table 5.9). In spring the DM content was 26% ± 1% and this increased by 
4.2% and then a further 2.4% over the spring-summer and summer-autumn intervals, 
respectively (Table 5.9). Between spring and summer the concentrations of Sol CHO, CP, 
and lipid in the leaves did not change significantly. However, over the ensuing autumn 
season these three indices decreased by 5.2%, 7.5%, and 0.2% DM, respectively 
(Table 5.9). Similarly, while the OMD and ME content of the leaves decreased slightly 
between spring and summer, these two indices also decreased markedly over the ensuing 
autumn period (Table 5.9). Over the same period the ADF and NDF concentration 
increased by around 9% DM (Table 5.9). The ash content and DCAD of the leaves varied 
little over time (Table 5.9). 
In both spring and summer, the DM, Sol CHO, OMD and ME content of the poplar leaves 
were 6-8%, 6-7%, 5-12% DM, and 0.8- 1 .8 MJ/kg DM, respectively, higher than in the 
mixed-species pasture (P<0.001 ;  Tables 5.7 & 5 .9). Conversely, over the same two 
consecutive seasons, the ADF, NDF, and ash concentrations were 4-5%, 13- 14%, and 
around 2%, respectively lower in the poplar leaves, compared to the mixed-species pasture 
(P<O.OO l ;  Tables 5.7 & 5.9). There was little difference in lipid and the DCAD between 
the two types of plant material in spring (P>0.05). However, both of these feed value 
indices were significantly higher in the poplar leaves in summer (P<0.0 1 ;  Tables 5.7 & 
5.9). The concentration of CP in the two types of plant material was not significantly 
different in either spring or summer (P>0.05; Tables 5.7 & 5.9). 
5.4.6.2 Seasonal effects at Hautope 1 
The DM content of the fresh (wet) poplar leaves tended to increase from spring through 
until late autumn (Table 5 . 10). In spring, the leaves contained 30% ± 3% DM and this 
increased by 5% and then 16% over the summer and autumn seasons, respectively 
(Table 5. 1 0) .  Similarly, between spring and summer the Sol CHO content of the leaves 
increased marginally by 2% DM. However, this concentration then decreased by 8% DM 
during the ensuing summer/autumn period (Table 5. 1 0). Over the same latter period, the 
Chapter 5 241 
CP, OMD and ME content of the leaves also decreased by 5.8%, 1 7. 1%, and 2.4 MJ/kg 
DM, respectively, while the ADF and NDF concentration tended to increase by around 
8% DM (Table 5 . 10). In spring the DCAD was 531  meq/kg DM. This decreased by 1 2.8% 
and then 35 .8% in summer and autumn, respectively (Table 5. 1 0). The lipid and ash 
concentrations ofthe leaves did not vary significantly over time (Table 5 . 1 0). 
In spring, the DM content of the fresh (wet) poplar leaves was 35% of that in the mixed­
species pasture. However, in summer the reverse occurred, with the DM content of the 
latter being 61% of that in the poplar leaves (P<0.00 1 ;  Tables 5.8 & 5 . 1 0). Analogous to 
Kiwitea, the concentration of Sol CHO, OMD, and ME content of the poplar leaves were 
5-8%, 1 0- 12%, and 1 .4- 1 .8 MJ/kg DM, respectively, higher than in the mixed-species 
pasture, over both spring and summer (P<0.05; Tables 5.8 & 5 . 1 0). At the same time, 
the ADF and NDF concentrations in the poplar leaves were 4.4-6.9% and 2 1 .8-23%, 
respectively, lower than in the mixed-species pasture (P<0.05; Tables 5.8 & 5. 1 0). Also 
similar to Kiwitea in spring, the DCAD did not differ significantly between the two 
types of plant material (P>0.05). However, in summer the DCAD of the poplar leaves was 
marginally ( 12%) lower than the mixed-species pasture (P<0.05; Tables 5.8 & 5. 1 0). 
In spring, the CP and Lipid concentrations were 5.5% and 0.9% DM, respectively, higher 
in the poplar leaves than mixed-species pasture (P<O.Ol). However, in the following 
summer, the lipid concentration was 0.6% DM greater in the pasture (P<O.OO l ), while 
there was little difference in CP between the two types of plant material (Tables 5.8 & 
5. 1 0). 
Table 5.9 Nutritive value indices for P. x euramericana leaves at Kiwitea. 
Season DM CP Lipid Ash ADF NDF Sol CHO DCAD OMD ME 
(%) (%DM) (%DM) (%DM) (%DM) (%DM) (%DM) (meq/kg DM) (%DM) (MJ/kg DM) 
Spring 26.8 19.8 5.0 10.0 24.0 34.4 1 1 .3 562 78.8 1 1 .4 
Summer 3 1 .0 1 8.5 4.9 9.7 25.3 35.5 1 1 .2 5 1 9  76.6 1 1 . 1  
Autumn 33 .4 1 1 .0 4.7 10.3 34.2 44.0 6.0 448 63.4 9.2 
Grand mean 30.4 16.4 4.9 10.0 27.8 37.9 9.5 5 1 0  72.9 10.5 
Analysis of variance 
Significance P<0.05 P<O.OI P<0.05 NS P<O.O I  P<O.O I  P<O.05 NS P<O.OI  P<0.05 
SEM (n=12) 0.9 1 .3 0. 1 0. 1 1 .4 1 .3 0.8 1 8  2 . 1  0.3 
Averaged over a two-year period ( 1998-2000). Abbreviations: SEM, standard error of the mean; NS, not significant (p�0. 1 ); DM, dry matter; CP, crude protein; ADF, 
acid detergent fibre; NDF, neutral detergent fibre; Sol CHO, soluble carbohydrate; DCAD, dietary cation-anion difference; OMD, in vitro organic matter digestibility; 
ME, metabolisable energy. 
Table 5.10 Nutritive value indices for p, x euramericana leaves at Hautope 1 .  
Season DM CP Lipid Ash ADF NDF Sol CHO DCAD OMD ME 
(%) (%DM) (%DM) (%DM) (%DM) (%DM) (%DM) (meq/kg DM) (%DM) (MJ/kg DM) 
Spring 3 0. 1  1 8.9 4,3 1 0.0 26.3 3S.4 1 1 . 1  S 3 1  7S.3 1 0.6 
Summer 34.7 1 7. 1  4 . 1  1 1 .0 26.8 3 1 . 5  1 3 .2 463 78.3 1 0.9 
Autumn SO.9 1 1 .3 4. 1 1 0.0 34,S 39. 1 5,6 297 6 1 .2 8.S 
Grand mean 3 8.6 I S.7 4. 1 1 0.3 29.2 3S.3 1 0.0 430 7 1 .6 1 0.0 
Analysis of variance 
Significance P<O , 1  P<0.05 NS NS P<O. l P<O, l P<O. l P<0.05 P<0.05 P<O.OS 
SEM (n=12) 2.8 1 . 1  0. 1 0.4 1 .3 1 .2 1 .2 32 2.4 0.3 
Averaged over a one-year period 1 999-2000. Abbreviations: SEM, standard error of the mean; NS, not significant (P�O. l ); DM, dry matter; CP, crude protein; ADF, 
acid detergent fibre; NDF, neutral detergent fibre; Sol CHO, soluble carbohydrate; DeAD, dietary cation-anion difference; OMD, in vitro organic matter digestibility; 
ME, metabolisable energy. 
Chapter 5 244 
5.4.7 Pasture mineral content 
5.4.7.1 Effect of overstorey environment and season on Kiwitea pasture mineral 
concentrations 
Major mineral elements 
Overall, the concentrations of potassium (K) and magnesium (Mg) in the understorey 
pasture amongst the trees (Zones 1 & 3) were 3 1  % and 1 3%, respectively, greater than in 
the adjacent open pasture (Table 5 . 1 1) .  When averaged across all three main overstorey 
environments, these two major cations decreased significantly between spring and summer 
(Table 5 . 1 1) .  In spring, the concentration of phosphorus (P) in Zones 1 and 3 was 1 9% 
greater than in the open pasture. Nevertheless, all three main overstorey environments 
decreased to a similar level of O.3%DM over the ensuing season (Table 5 . 1 1). Between 
spring and summer the concentration of sulfur (S) in Zones 1 and 3 decreased by 20-22%, 
whereas, it did not change significantly in the open pasture. As a result, Zone 1 contained 
significantly less S than in the open environment (Table 5 . 1 1) .  For both calcium (Ca) and 
sodium (Na) there were no significant differences between overstorey environments or 
seasons (Table 5 . 1 1) .  
Trace mineral elements 
Averaged over the two consecutive seasons, the concentration of manganese (Mn) was 
50% greater in the open pasture than in Zones 1 and 3 (Table 5 . 12). On the other hand, 
strontium (Sr) tended (P=0.053 1)  to form a graded series, with Zone 3 having the greatest 
and the open pasture having the lowest concentration (Table 5 . 12). Both of these trace 
elements, along with iron (Fe) and aluminium (AI), accumulated in the pastures between 
spring and summer (Table 5 . 12). Conversely, copper (Cu) and nickel (Ni) decreased over 
the same period of tirne (Table 5 . 12). The levels of zinc (Zn), boron (B) and cobalt (Co) in 
the pastures were not significantly different between overstorey environments or seasons 
(Table 5 . 12) .  
Table 5.11 Major mineral elements (J.1g1g DM) in the mixed-species pasturel at Kiwitea: directly below the crowns of individual poplar trees (Zone 1 ), at the centre of the 
vertically projected gap between poplar crowns (Zone 3), and in the open pasture. 
Season Environment Ca K Mg Na P S 
(lnx) 
Spring'98&'99 Open pasture 4228 2 1 672 1 96 1  1 1 3 5  (7.02) 3 722 2679 
Zone 1 4 1 89 30449 2400 1 03 8  (6.93) 45 1 3  2898 
Zone 3 5049 29358 225 1 1 200 (7.08) 4338 296 1 
Grand mean 4488 27 1 59 2204 1 124 (7.0 1 )  4 1 9 1  2846 
Summer'98&'99 Open pasture 3953 1 8773 1 932 1 4 9 1  (7.26) 32 1 7  2572 
Zone 1 4783 22733 2 1 02 1 3 1 9  (7. 17) 3252 2246 
Zone 3 49 1 2  23 1 1 3 2080 1 1 93 (7.08) 3270 236 1  
Grand mean 4549 2 1 540 203 8 1 334 (7. 1 7) 3246 2393 
Overall Open pasture 4090 20223 1 946 1 3 1 3  (7. 14) 3470 2626 
Zone 1 4486 26591 225 1 1 1 78 (7.05) 3 882 2572 
Zone 3 4980 26235 2 1 66 1 1 97 (7.08) 3804 266 1 
Grand mean 45 1 9  24350 2 1 2 1  1 229 (7.09) 3 7 1 9  2620 
Analysis of variance 
Environment (A) NS P<0.05 P=0.05 NS NS NS 
SEM (A) 1 75 767 40 (0.07) 65 67 
Season (B) NS P<O.OOI P<O.O I NS P<O.OO l P<O.OOl 
SEM (B) 1 45 626 33 (0.06) 53 5 5  
Interaction ( A  *B) NS NS NS NS P<O.OI P<O.05 
SEM (A*B) 247 1 084 57 (0. 1 1  ) 92 95 
lAveraged across the entire range of poplar stand densities over a two-year period ( 1 998-2000). Transformed values are given in parentheses. Abbreviations: lnx, natural-
logarithm transformed; SEM, standard error of the mean; NS, not significant (�0. 1);  DM, dry matter; Ca, calcium; K, potassium; Mg, magnesium; Na, sodium; P, 
phosphorus; S, sulfur. 
N � 
C.11 
Table 5.12 Trace mineral elements (�glg DM) in the mixed-species pasture! at Kiwitea: directly below the crowns of individual poplar trees (Zone 1 ), at the centre of the 
vertically projected gap between poplar crowns (Zone 3), and in the open pasture. 
Season Environment Cu Fe Mu Zn AI B Co Ni Sr 
(lnx) (lux) (x·1x l0·3) 
Spring'98&'99 Open pasture 1 9.4 2 1 1 (5.30) 1 3 7 . 1  (4.90) 36.3 206.5 (5.25) 20.2 0.50 0.55 24.6 
Zone 1 1 7.9 1 82 (5.20) 62.7 (4. 13) 36. 1 236.3 (4.50) 22.5 0.55 0.70 24.9 
Zone 3 1 7.5 1 68 (5. 1 2) 84. 1  (4.42) 45.6 1 89.3 (5.39) 20.0 0.40 0.55 30.2 
Grand mean 1 8. 3  1 82 (5.21) 86. 1 (4.48) 39.3 2 1 1 .5 (5.05) 20.9 0.48 0.60 26.6 
Summer'99&'00 Open pasture 1 5.0 265 (5.57) 1 55.0 (5.03) 36.6 40 1 . 8  (2.63) 1 7.7 0.48 0.35 26.2 
Zone 1 1 4 . 1  327 (5.76) 1 20.9 (4.71 )  33.8 574.0 (2.0 1 )  1 9.0 0.73 0.43 32.2 
Zone 3 14.5 273 (5.59) 122.5 (4.79) 4 1 . 8  477.3 (2.30) 1 9. 8  0.63 0.53 3 3 . 1  
Grand mean 1 4.6 293 (5.64) 128.3 (4.84) 3 7.4 500.9 (2.32) 1 8.8 0.64 0.43 30.5 
Overall Open pasture 1 7.2 238 (5.43) 1 46. 1 (4.96) 36.5 304 . 1  (3.94) 1 8.9 0.49 0.45 25.4 
Zone 1 1 6.0 255 (5.48) 9 1 . 8  (4.42) 34.9 405 . 1  (3.26) 20.8 0.64 0.56 28.5 
Zone 3 1 6.0 220 (5.35) 103.3 (4.60) 43.7 333.3 (3.85) 1 9.9 0.5 1 0.54 3 1 .7 
Grand mean 1 6.4 238 (5.42) 1 07.2 (4.66) 3 8.4 356.2 (3.68) 1 9. 8  0.56 0.52 28.5 
Analysis of variance 
Environment (A) NS NS P<O . O I  NS NS NS NS NS P=0.05 
SEM (A) 0.3 (0. 1 0) (0.09) 3.2 (0.52) 1 .2 0.06 0.06 0.9 
Season (B) P<O.OO I P<O.O I P<O.O I NS P<O.OI NS NS P<0.05 P<O.O I 
SEM (B) 0.3 (0.08) (0.08) 2.6 (0.42) 1 .0 0.05 0.05 0.8 
Interaction (A *B) NS NS NS NS NS NS NS NS NS 
SEM (A*B) 0.5 (0. 1 4) (0. 1 3) 4.5 (0.73) 1 .7 0.08 0.08 1 .3 
JAveraged across the entire range of poplar stand densities over a two-year period ( 1 998-2000). Transformed values are given in parentheses. Abbreviations: lnx, natural-
logarithm transformed; x·1x I 0·3, inverse transformed; DM, dry matter; SEM, standard error of the mean; NS, not significant (�0. 1 ); Cu, copper; Fe, iron; Mu, manganese; 
Zn, zinc; Al, aluminium; B, boron; Co, cobalt; Ni, nickel; Sr, strontium. The values for AI, B, Co, Ni, and Sr are based on one year ( 1 999-2000). 
I\) � 0> 
Chapter 5 247 
5.4.7.2 The effect of poplar overstorey density on Kiwitea pasture mineral 
concentrations 
Major mineral elements 
In spring, the concentrations ofK, Mg, P, and S in the pasture showed moderate through to 
strong (r2=0.59-O.79) positive linear relationships with increasing eeL (Figure 5.22). 
Similar (P>0.05) linear relationships were evident in both Zone 1 and 3 for K, Mg, and P, 
whereas, for S the relationship solely occurred in Zone 3 (Figure 5 .22). The rate of change 
in K, Mg, P, and S was 1220 ± 190 f.!glg DM, 56 ± 1 2  f.!glg DM, 1 1 5 ± 28 f.!glg DM, and 
50 ± 12  f.!glg DM, respectively, for every 1 0% increase in eeL (Figure 5.22). No clear 
trends were apparent for either ea or Na in the spring pastures. 
In summer, linear relationships similar to those found in spring occurred for K and Mg 
(Figure 5 .23a,b). However, the intercept of the relationship for K was 1 9% lower than in 
spring (P=0.05), and for Mg the relationship was significant only in Zone 3 .  In addition to 
these two major cations, the concentration of ea also showed a moderate (r2==0.42), but 
positive, linear relationship with increasing eeL. This major cation accumulated at a rate 
of 144 ± 48 f.!glg DM for every 10% increase in eeL (Figure 5.23c). No clear trends were 
apparent for P, S, or Na in the summer pastures. 
Trace mineral elements 
Amongst the trees (Zone 1 & 3) in spring, the concentrations of eu and Mn decreased at a 
constant rate of 0.3 ± 0. 1 f.!glg DM and 9.0 ± 2.0 f.!glg DM, respectively, for every 10% 
increase in eeL (Figure 5.24). The strength of the relationship for eu was moderate 
(r2=0.56), whereas, for Mn it was strong (r2=0.65) (Figure 5 .24). 
Sr was the only trace element that showed a significant relationship with eeL in summer 
(Figure 5 .25). However, the strength of this relationship was a moderate fit (�=O.5 1). In 
both Zones 1 and 3 Sr increased at a constant rate of 1 .0 ± 0.3 f.!glg DM for every 10% 
increase in eeL (Figure 5.25). 
40000 SOOO 
(a) (c) 0 0 
3Soo0 0 4S00 
� 30000 
� 
� 
� 
4000 
..:; 2Soo0 ::I. '-' 
� p.. 
3S00 · 20000 Y = 3699 ( IS I )  + x* l 1 .5 (2.8) Y = 21963 ( 101S)  + x*122 (19) 
r2=O.79, RMSE=21S3, P<O.ooOI r2=O.S9, RMSE=320, P=O.002 
ISoo0 3000 
0 20 40 60 80 100 0 20 40 60 80 100 
2800 3400 
(b) (d) 
2600 3200 • 
� 
0 0 0 2400 
� 
3000 
� 
2200 
� ..:; 2> 2800 
� 
2000 r/) 
1 800 2600 Y = 26S9 (S8) + x*S.O ( 1 .2) Y = 1966 (68) + x*S.6 (l .2) 
r2=O.63, RMSE=I44, P=O.OOI r2=O.69, RMSE=1 19, P=O.006 
1600 2400 
0 20 40 60 80 1 00 0 20 40 60 80 100  
Canopy closure with leaves (%) Canopy closure with leaves (%) 
Figure 5.22 Spring relationships between major pasture minerals and poplar canopy closure (CCL) at Kiwitea: (j.)  directly below the crowns of 
individual trees (Zone 1 ), and (0) at the centre of the vertically projected canopy gap between trees (Zone 3). Standard errors for regression 
coefficients are given in parentheses. Abbreviations: r2, adjusted coefficient of determination; RMSE, standard error of prediction; K, potassium; 
Mg, magnesium; P, phosphorus; and S, sulfur. 
30000 .-----------------, 
25000 
� 
� 20000 
:!. '-' 
1 5000 
Y = 17796 ( l OS 1 )  + x·SS (20) 
r2=O.62, RMSE=2294, P=O.ool 
1 0000 +-----,-----.-----r----.---------l 
o 20 40 60 80 100 
2300 .,------------------,
o 
2200 
� 2100 
Y = 1910  (50) + x·3.0 (Ll) 
r2=O.49, RMSE=103, P=O.03 
o 
i 
'-' 2000 
� 
1900 
(b) 
l S00 +---,.-- ---.-----r---,----4 
o 20 40 60 SO 100 
Canopy closure with leaves (%) 
7000 
(c) 
6000 
� 5000 
� 
� 4000 
a 
3000 
2000 
0 20 
o o 
Y = 3935 (259) + x·14.4 (4.8) 
r2=O.42, RMSE=549, P=O.Ol 
40 60 80 100 
Canopy closure with leaves (%) 
Figure 5.23 Summer relationships between major pasture minerals and poplar canopy closure (CCL) at Kiwitea: ( . )  directly below the crowns of 
individual trees (Zone 1), and (0) at the centre of the vertically projected canopy gap between trees (Zone 3). Standard errors for regression 
coefficients are given in parentheses. Abbreviations: r2, adjusted coefficient of determination; RMSE, standard error of prediction; K, potassium; 
Mg, magnesium; and Ca, calcium. 
Chapter 5 250 
22 
(a) 
20 
2 
Cl 
� 1 8  '-' 
;:3 
U 
0 
1 6  A-
Y = 19.5 (0.4) - x*0.03 (0.0 1 )  
?=0.56, RMSE=O.85, P=O.003 
1 4  
0 20 40 60 80 1 00 
200 
(b) 
1 5 0  
,.-.. 
� 
bO 
1 00 � '-' 
� 2 0  
50 A 
Y = 1 33 .6 ( 1 0.6) - x*0.9 (0.2) 
?=0.65, RMSE=22.4, P=0.0009 
0 
0 20 40 60 80 1 00 
Canopy closure with leaves (%) 
Figure 5.24 Spring relationships between pasture trace minerals and poplar canopy closure (CCL) at 
Kiwitea: ( .. ) directly below the crowns of individual trees (Zone 1 ), and (0) at the centre of the vertically 
projected canopy gap between trees (Zone 3). Standard errors for regression coefficients are given in 
parentheses. Abbreviations: �, adjusted coefficient of determination; RMSE, standard error of prediction; Cu, 
copper; and Mn, manganese. 
Chapter 5 
40 .-----------------------------------------� 
35 
25 
Y = 26.3 (1 .5) + x*O. l (0.03) 
r2=O.51, RMSE=3.2, P=0.005 
20 +--------.------�r-------.-------�------� 
o 20 40 60 80 100 
Canopy closure with leaves (%) 
251 
Figure 5.25 Summer relationship between strontium (Sr) and poplar canopy closure (CCL) at Kiwitea: (A) 
directly below the crowns of individual trees (Zone 1 ), and (0) at the centre of the vertically projected 
canopy gap between trees (Zone 3). Standard errors for regression coefficients are given in parentheses. 
Abbreviations: r, adjusted coefficient of determination; RMSE, standard error of prediction. 
Chapter 5 252 
5.4.7.3 Effect of overstorey environment and season on Hautope 1 pasture mineral 
concentrations 
Major mineral elements 
Overall, the K and Mg concentrations in the understorey pastures amongst the trees 
(Zones 1 & 3) were 66% and 47% greater than in the open pasture, respectively (Table 
5 . 1 3). Across the three main overstorey environments, Mg increased by 1 6% between 
spring and summer (Table 5. 1 3). Similarly, over the same two consecutive seasons, K 
increased in Zone 3, but did not change significantly in Zone 1 or the open pasture 
(P=O.0492; Table 5 . l3). While the level of Ca in the pastures did not differ among the 
three main overstorey environments, this major mineral element also increased by 26% 
over spring and summer (Table 5. l 3). 
The understorey pasture in Zone 1 had the greatest concentration of Na, followed by 
Zone 3,  and then the open pasture environment (Table 5 . l 3). However, unlike the former 
major cations, across the three main overstorey environments, the level ofNa decreased by 
28% between spring and summer (Table 5 . 1 3). 
In spring, the average concentration of P in the understorey pasture amongst the trees 
(Zones 1 & 3) was 24% greater than in the open pasture (Table 5 . 1 3). This difference 
increased to 63% in summer, as the level in the understorey pasture (Zone 1 & 3) increased 
further, while it did not change significantly in the open pasture (Table 5 . l 3). Overall, the 
concentration of S also tended to be 1 6% greater in Zones 1 and 3 ,  compared to in the open 
pasture (P=O.08; Table 5. 1 3). 
Trace mineral elements 
In spring, the understorey pasture amongst the trees (Zones 1 & 3) contained 58% more Fe 
than in the open pasture (Table 5 . 14). At the same time, the concentration of AI in Zone 1 
was also 27% greater than in the open pasture, whereas, Zone 3 was not significantly 
different from either of the other two main overstorey environments (Table 5 . 14). 
Nevertheless, during the ensuing summer the respective concentrations of Fe and AI in all 
three main overstorey environments increased to a similar level (Table 5 . 14) .  
Chapter 5 253 
Overall, the concentrations of Zn and Sr in the understorey pasture amongst the trees 
(Zones 1 & 3) were 20% and 32%, greater than in the open pasture, respectively (Table 
5 . 14). These trace minerals, along with all of the others measured, except for B and Ni, 
accumulated in the pastures over spring and summer (Table 5 . 14). 
5.4.7.4 Effect of poplar overstorey density on Hautope 1 pasture mineral 
concentrations 
Major mineral elements 
In spring, the respective concentrations ofNa, K, Mg, and P in the pasture increased at a 
constant rate of 203 ± 44 �g/g DM, 1 1 60 ± 1 80 �g/g DM, 1 02 ± 16  �g/g DM, and 1 05 ± 
16  �g/g DM, respectively, for every 10% increase in eeL (Figure 5.26). The strengths 
of these positive linear relationships were strong (r2=0.65-O.79), and the relationships were 
not significantly different (P>0.05) between Zones 1 and 3 (Figure 5.26). No clear trends 
were apparent for either ea or S in the spring pastures. 
In summer, the concentrations ofNa, K, Mg, and P in the pasture showed strong to very­
strong (r2=0.71 -0.84) positive linear relationships with increasing eeL (Figure 5.27). For 
Na the rate of change was significantly (P<O.OS) different between the two main zones of 
tree influence, with the level increasing twofold in Zone 1 compared to Zone 3 (Figure 
S .27c). Nevertheless, both of these relationships (i.e. the intercepts and slopes) were not 
significantly different (P>0.05) from the single combined relationship that was observed in 
spring (Figures 5.26c & 5 .27c). Similarly, the rate of change in Mg was not significantly 
different (P>0.05) between the two consecutive seasons (Figures 5 .26b & 5 .27b). In 
contrast, the rate of change in K and P was 1 .8- and 2.4-fold, respectively, greater in 
summer compared to spring (p�0.05) (Figures 5 .26a,d & 5 .27a,e). 
Unlike in the spring pastures, ea and S showed significant, but moderate fit (r2=0.42-0.49), 
positive linear relationships with increasing eeL (Figure 5.27d,f). These two major 
minerals increased at a rate of 8 1  ± 29 �g/g DM and 98 ± 33 �g/g DM, respectively, for 
every 10% increase in eeL (Figure 5 .27d,f). The relationship for ea was restricted to 
Zone 1 (Figure 5 .27d). 
Chapter 5 254 
Trace mineral elements 
In spring, the concentrations of ell, Zn, Fe, and Sr increased at a constant rate of 0. 1 7  ± 
0.05 /lg/g DM, 1 .2 ± 0.3 /lg/g DM, 7.3 ± 1 .8 /lg/g DM, and 1 .0 ± 0.3 /lg/g DM, 
respectively, for every 1 0% increase in eeL (Figure 5 .28). For each of these trace minerals 
the relationships found in Zone 1 and 3 were similar (P>0.05). Overall, the strengths of fit 
of the combined (Zones 1 & 3) linear relationships were moderate to moderately strong 
(r2=0.53-0.63) (Figure 5.28). 
In summer, the rate of change in Sr was not significantly different (P>O.OS) from that 
observed in spring (Figures 5.28d & 5 .29b). Unlike in spring, the concentration of Mn in 
Zone 1 was significantly related to eeL. Mn decreased by 28.6 ± 8.0 /lg/g DM for every 
1 0% increase in eeL (Figure S.29a). 
Table 5.13 Major mineral elements (�g/g DM) in the mixed-species pasture I at Hautope 1 :  directly below the crowns of individual poplar trees (Zone 1 ), at the centre of the 
vertically projected gap between poplar crowns (Zone 3), and in the open pasture. 
Season Environment Ca K Mg Na P S 
(lnx) (lnx) C..Jx) C..Jx) 
Spring'99 Open pasture 343 7 1 5562 (9.65) 1351 (7.2 1 )  1225 (34.66) 2387 (48.83) 2294 
Zone 1 3585 232 1 0  ( 10.05) 1 99 8  (7.59) 2698 (5 1 . 87) 3 0 1 5  (54.88) 2453 
Zone 3 3367 20952 (9.94) 1 8 1 7  (7.50) 2 1 73 (46.23) 2900 (53.80) 23 7 1  
Grand mean 3463 1 9908 (9.88) 1 722 (7.43) 2032 (44.25) 2768 (52.50) 2373 
Summer'OO Open pasture 4064 1 3 8 8 1  (9.53) 1485 (7.30) 75 1 (27.32) 2483 (49.76) 2 1 32 
Zone 1 4524 27268 ( 10.20) 2279 (7.73) 2 1 90 (46.76) 4 1 56 (64.34) 2700 
Zone 3 4509 26464 ( 10. 1 5) 2255 (7.7 1)  1437 (37.87) 392 1  (62.37) 278 1 
Grand mean 4365 22538 (9.96) 2006 (7.58) 1459 (37.32) 3520 (58.83) 2537 
Overall Open pasture 3750 14722 (9.59) 1 4 1 8  (7.25) 988 (30.99) 2435 (49.30) 22 1 3  
Zone 1 4054 25239 ( 1 0. 1 3) 2 1 3 9  (7.66) 2444 (49.3 1) 3585 (59.61 )  2576 
Zone 3 393 8  23708 ( 10.04) 2036 (7.60) 1 805 (42.05) 34 1 1  (58.09) 2576 
Grand mean 3914 2 1 223 (9.92) 1 864 (7.50) 1 746 (40.79) 3 144 (55.67) 2455 
Analysis of variance 
Environment (A) NS P<O.OI P<O.O I  P<O. OO I P<O.O I P<O. 1 
SEM (A) 1 0 1  (0.04) (0.03) ( 1 .56) (0.98) 84 
Season (B) P<O.OOI NS P<O.O I P<O.O I P<O.OO I NS 
SEM (B) 82 (0.03) (0.02) (1 .27) (0.80) 68 
Interaction (A *B) NS P=0.05 NS NS P<0.05 P<O. 1 
SEM (A*B) 143 (0.06) (0.04) (2.2 1) ( 1 .39) 1 1 8 
lAveraged across the entire range of poplar stand densities over a one-year period (1 999-2000). Transformed values are given in parentheses. Abbreviations: Inx, natural-
logarithm transfonned; ...Jx, square-root transformed; SEM, standard error of the mean; NS, not significant (P;;:O. I);  DM, dry matter; Ca, calcium; K, potassium; Mg, 
magnesium; Na, sodium; P, phosphorus; and S, sulfur. 
N 01 01 
Table 5.14 Trace mineral elements (l1g/g DM) in the mixed-species pasturel at Hautope 1 :  directly below the crowns of individual poplar trees (Zone 1), at the centre of the 
vertically projected gap between poplar crowns (Zone 3), and in the open pasture. 
Season Environment Cu 
lnx 
Spring'99 Open pasture 1 1 .0  (2.39) 
Zone 1 1 1 .8 (2.47) 
Zone 3 12 . 1 (2.49) 
Grand mean 1 1 .7 (2.45) 
Summer' 00 Open pasture 14.8 (2.68) 
Zone 1 14.6 (2.67) 
Zone 3 1 5 .7  (2.75) 
Grand mean 1 5 . 1 (2.70) 
Seasons combined Open pasture 1 2.9 (2.54) 
Zone 1 1 3 .2 (2.57) 
Zone 3 13 .9 (2.62) 
Grand mean 13 .4 (2.58) 
Analysis of variance 
Environment (A) NS 
SEM (A) (0.03) 
Season (B) P<O.OOI 
SEM (B) (0.02) 
Interaction (A *B) NS 
SEM (A*B) (0.04) 
Fe 
x·lxlO-3 
6 8 . 5  ( 14 .81)  
1 1 6.4 (8.67) 
99.8 ( 10.90) 
1 00. 1 ( 1 1 .46) 
3 8 1 .0 (2.99) 
259.0 (4.20) 
276.8 (3.64) 
290.5 (3.6 1)  
224.7 (8.90) 
1 87.7 (6.43) 
1 88 .3 (7.27) 
1 95.3 (7.53) 
P<O. I 
(0.59) 
P<O.OOI 
(0.48) 
P<O.O I 
(0.83) 
Mn 
lnx 
248 . 0  (5.50) 
1 52 . 5  (4.93) 
1 8 1 .3 (5. 1 4) 
1 83 . 1  (5. 19) 
362.3 (5.87) 
1 93 . 0  (5.22) 
282.3 (5 .47) 
262.6 (5.52) 
305 . 1  (5.68) 
1 72 .8  (5.08) 
23 1 . 8 (5.3 1 )  
222.8 (5.35) 
NS 
(0.07) 
P<O.OI 
(0.06) 
NS 
(0. 1 0) 
Zn 
x-'xlO-3 
27.3 (36.85) 
34.9 (29. 1 1) 
32.7 (30.87) 
32.5 (32.28) 
92.8 ( 1 1 .70) 
1 20.2 (8.86) 
1 00.8 ( 10.38) 
1 07.0 ( 1 0.3 1)  
60.0 (24.28) 
77.6 ( 1 8.99) 
66.7 (20.62) 
69.7 (2 1 .29) 
P=O.05 
( 1 .08) 
P<O.OO I 
(0.89) 
NS 
(1 .53) 
AI B Co Ni Sr 
x-'xlO-3 x-'xlO-3 xl13 '-Ix 
1 1 3.8 (8.83) 16.8 (68.86) 0.43 (0.73) 0.95 (0.97) 1 7.0 
1 44.5  (7.00) 1 7. 5  (59.82) 0 .2 8  (0.65) 0.98 (0.97) 23.0 
128.0 (8. 17) 1 1 .8 (86.06) 0.30 (0.67) 1 . 1 3 ( 1 .06) 2 1 .6 
1 3 1 .8 (8.00) 1 5 . 1  (7 1 .58) 0 .32 (0.68) 1 .03 ( 1 .00) 2 1 .2 
4 12.0 (2.99) 14 .7  (72.53) 0.58 (0.82) 1 .20 ( 1 .09) 2 1 .0 
266.0 (4.30) 1 6.9 (65.98) 0 .50 (0.79) 1 . 1 3 ( 1 .05) 28. 1 
248.5 (4.07) 1 5 .4 (65.49) 0.48 (0.78) 1 .20 ( 1 .09) 27.5 
288.2 (3.79) 1 5.9 (68.00) 0.5 1 (0.80) 1 . 1 7  ( 1 .07) 26.4 
262 .9 (5.9 1 )  1 5 . 8  (70.69) 0.50 (0.78) 1 .08 ( 1 .03) 1 9.0 
205.3 (5.65) 1 7.2 (62.90) 0.39 (0.72) 1 .05 ( 1 .0 1 )  25.6 
1 88.3 (6. 1 2) 1 3 .6 (75.78) 0 .39 (0.72) 1 . 1 6  ( 1 .07) 24.5 
2 1 0.0 (5.89) 1 5.5 (69.79) 0.4 1 (0.74) 1 . 1 0 (1 .04) 23 .8 
NS NS NS NS P<0.05 
(0.36) (8.07) (0.01) (0.04) 0.70 
P<O.OOI NS P<O.OOI NS P<O.OOI 
(0.29) (6.59) (0.0 1 )  (0.04) 0.57 
P<0.05 NS NS NS NS 
(0.5 1)  ( 1 1 .42) (0.0 1 )  (0.06) 0.99 
1 Averaged across the entire range of poplar stand densities over a one-year period ( 1999-2000). Transformed values are given in parentheses. Abbreviations: Inx, natural-
logarithm transformed; x-'xlO-3, inverse transformed; xli3, cubic-root transformed; ...Jx, square-root transformed; DM, dry matter; SEM, standard error of the mean; NS, not 
significant (P;::O. 1 ); Cu, copper; Fe, iron; Mn, manganese; Zn, zinc; AI, aluminium; B, boron; Co, cobalt; Ni, nickel; Sr, strontium. 
!\.) 
CJ'I 0> 
30000.--------------------------. 
(a) 
25000 
� � 20000 
3 
� 
1 5000 
y = 1 5295 (910) + x'" 1 1  6 ( 1 8) 
r2=O.78, RMSE=1 886, P<O.ooOI 
1 0000 +-----..----,-----,..;...-----1 
o 20 40 60 80 
3000 ,.-----------------. 
2500 
Y = 1 3 1 6  (77) + x'" 1 0.2 ( 1 .6) 
r2=O.79, RMSE=160, P<O.OOOI 
(b) 
'i 2000 
'-' 
� 
1 500 
1 000 +-----..-----,----�--_4 
o 20 40 60 80 
Canopy closure with leaves (%) 
4000 
3000 
� 
� 2000 
3 
� 
1 000 
0 
0 
(C) 
20 
y =  1227 (21 7) +x"'20.3 (4.4) 
;=0.65, RMSE=449, P=O.0009 
40 60 80 
4000.-----------------------, 
3500 
� � 3000 
3 
p., 
2500 
Y = 2352 (80) + x'" 1 0.5 ( 1 .6) 
r2=O.79, RMSE=1 66, P<O.ooOI 
o 
(d) 
2oo0 �----�------�----_r----_4 
o 20 40 60 80 
Canopy closure with leaves (%) 
Figure 5.26 Spring relationships between major pasture minerals and poplar canopy closure (CCL) at Hautope 1 :  ( . )  directly below the crowns 
of individual trees (Zone 1 ), and (0) at the centre of the vertically projected canopy gap between trees (Zone 3). Standard errors for regression 
coefficients are given in parentheses. Abbreviations: r2, adjusted coefficient of determination; RMSE, standard error of prediction; Na, sodium; K, 
potassium; Mg, magnesium; and P, phosphorus. 
35000 4000 5000 
(a) Y Zone 3 = 777 (88) + x· I O. 7 (2.0) (C) (e) ... "' 0  
30000 0 "' 0  r2:00.79, RMSE=1 79, P=O.002 
3000 4000 
:E 25000 * * 0 .ag 20000 � 2000 .ag 3000 :::1. ,.:; '-' :::1. � '-' � 1 5000 Z � 
1 000 2000 
10000 Y = 13939 (2033) + x*21 7  (41 )  YZone t = 81 3  (155) + x*22. 1 (3.6) Y = 2499 (242) + x*25.7 (4.9) 
?=O.71 ,  RMSE=42 16, P=O.0004 r=0.84, RMSE=3 15, P=O.0009 1 000 
?=O.71 ,  RMSE=503, P=O.0004 
5000 0 
0 20 40 60 SO 0 20 40 60 SO 0 20 40 60 
3000 5500 3500 
(b) Y = 4052 ( 125) + x*S. l (2.9) (d) (t) 
?=O.49, RMSE=255, P=0.03 0 ... 
2500 0 0 5000 3000 ... 0 
� � � 
.ag 2000 � 4500 � 2500 :::1. :::1. '-' '-' ,.:; 
� l3 (/) 
1 500 4000 2000 
Y = 1486 (107) + x*1 3 . 1  (2.2) Y = 2 149 ( 16 1 )  + x*9.S (3.3) 
?=O.77, RMSE=222, P=O.OOO l ?=O.42, RMSE=335, P=O.Ol 
1000 3500 1 500 
0 20 40 60 80 0 20 40 60 SO 0 20 40 60 
Canopy closure with leaves (%) Canopy closure with leaves (%) Canopy closure with leaves (%) 
Figure 5.27 Swnmer relationships between the major pasture minerals and poplar canopy closure (CCL) at Hautope I :  (A)  directly below the crowns of individual trees (Zone I), 
and (0) at the centre of the vertically projected canopy gap between trees (Zone 3; 0). Standard errors for regression coefficients are given in parentheses. Abbreviations: r, adjusted 
coefficient of determination; RMSE, standard error of prediction; Na, sodiwn; K, potassiwn; Mg, magnesiwn; Ca, calcium; P, phosphorus; and S, sulfur. 
80 
SO 
N 
c.n CD 
1 5  
Y = 10.91 (0.24) + x·0.017  (0.005) (a) 
14  ?=O.53, RMSE=O.49, P=0.004 
� 1 3  0 
� ... 0 
3 12  � ;:l U 
1 1 ....-
10 1 
0 20 40 60 80 
40 
(b) ... ... 
35 
� 
� 30 
:::1. '-' 
rG 
25 
Y = 26.7 ( 1 .4) + x·0. 12  (0.03) 
?=O.63, RMSE=2.8, P=O.OOl 
20 
0 20 40 60 80 
Canopy closure with leaves (%) 
160 
140 ?=O.59, RMSE=18. 1 ,  P=O.002 
j y � 66.1 (8.7) + x·0.73 (0. 1 8) 
� 120 . 
Cl 
� loo :::1. '-' Q) 80 ... 
60 v 
0 
... ... 
(c) 
40 
0 20 40 60 80 
30 
I (d) ... 0 
25 
� 
� 20 :::1. '-' ... \/) 
1 5  
1 Y = 16.5 (1 .3) + x·0. 10  (0.03) ?=O.57, RMSE=2.6, P=O.003 
10  
0 20 40 60 80 
Canopy closure with leaves (%) 
Figure 5.28 Spring relationships between the trace pasture minerals and poplar canopy closure (CCL) at Hautope 1 :  ( . )  directly below the 
crowns of individual trees (Zone 1), and (0) at the centre of the vertically projected canopy gap between trees (Zone 3). Standard errors for 
regression coefficients are given in parentheses. Abbreviations: r2, adjusted coefficient of determination; RMSE, standard error of prediction; Cu, 
copper; Zn, zinc; Fe, iron; and Sr, strontium. 
N CJ1 CO 
Chapter 5 
500�------------------------------------� 
400 
,.-... 
::E 
Cl 
00 
300 Ob ::1. 
'-' 
� 
200 
100 
35 
30 
2 
Cl 
00 25 Ob ::1. 
'-' ... r/J 
20 
0 
(a) 
Y = 363 (35) - x*2.9 (0.8) 
x2=O.62, RMSE=70, P=0.01 ... 
20 40 60 
Canopy closure with leaves (%) 
(b) 
Y = 20.9 (0.9) + x*0. 12  (0.02) 
x2=0.77, RMSE=1 .9, P=0.0001 
80 
15 +---------�--------.---------.-------� 
o 20 40 60 80 
Canopy closure with leaves (%) 
260 
Figure 5.29 Summer relationships between the trace pasture minerals and poplar canopy closure (CCL) at 
Hautope 1 :  ( .. ) directly below the crowns of individual trees (Zone 1), and (0) at the centre of the vertically 
projected canopy gap between trees (Zone 3). Standard errors for regression coefficients are given in 
parentheses. Abbreviations: r2, adjusted coefficient of determination; RMSE, standard error of prediction; 
Mn, manganese; and Sr, strontium. 
Chapter 5 
5.4.8 Poplar leaf mineral contents 
5.4.8.1 Seasonal effects at Kiwitea 
Major mineral elements 
261 
Ca and Na accumulated in the poplar leaves from spring through until late autumn 
(Table 5 . 1 5). These two major minerals increased by 42% and 80%, respectively, between 
spring and summer, and then increased a further 30% and 22%, respectively, between 
summer and just prior to leaf fall in autumn (Table 5 . 1 5). In contrast, the concentration of 
P decreased by 28% and then 2 1% over the same two intervals (Table 5 . 1 5) .  No significant 
changes occurred in the levels of K, Mg, or S over the three consecutive seasons (Table 
5 . 1 5) .  
In spring, the respective concentrations of Ca, Mg, and S in the poplar leaves were 2.6-
(P<O.Ol ), 2. 1- (p<O.OO l ), and 1 .6-fold (p<0.05) greater than the .average concentration in 
the mixed-species pasture for all three main overstorey environments (Tables 5. 1 1  & 5 . 1 5). 
This contrasted with P and Na, which were 35% (p<0.001)  and 29% (PSO.05) lower in the 
poplar leaves compared to the mixed-species pastures (Tables 5 . 1 1  & 5 . 15) .  
In summer, the difference in Ca, Mg, and S increased further to 3 .6- (P<O.OOl ), 2.5-
(P<O.OO l )  and 2.0-fold (P<0.05), respectively (Tables 5. 1 1  & 5 . 1 5). On the other hand, the 
difference in P remained at around 40%, and for Na there was no significant difference 
(p>0.05) between the poplar leaves and mixed-species pasture (Tables 5. 1 1  & 5 . 1 5). Both 
types of plant material had similar (P>0.05) levels of K in spring and summer (Tables 5 . 1 1  
& 5 . 1 5). 
Trace mineral elements 
Mn, Co, and Sr accumulated in the poplar leaves from spring through until late autumn 
(Table 5 . 1 6). Similarly, Fe, Al, and B also increased between spring and summer, but 
thereafter did not change significantly (Table 5 . 1 6). The concentrations of Cu, Zn, and Ni 
did not vary significantly over the three consecutive seasons (Table 5 . 1 6) .  
In spring the respective concentrations of Zn, B, Co, and Sr in the poplar leaves were 2.6-
(p<0.01),  2.9- (p<0.01),  4.9- (p<0.05), and 3 .6-fold (P<0.00 1 )  greater than the average 
Chapter 5 262 
concentrations in the mixed-species pasture (Tables 5 . 1 2  & 5 . 16). Conversely, Cu, Fe, and 
AI were 1 .8- (p<O.Ol), 2.5- (p<O.OOl )  and 6.2-fold (p<O.OOl ), respectively, greater in the 
mixed-species pasture compared to the poplar leaves (Tables 5 . 12  & 5 . 16). 
In summer, the differences in B, Co, Sr, Fe, and AI, increased further to 5 .7- (p<O.Ol ), 6.9-
(p<O.05), 4.7- (p<O.OOl), 2.9- (p<O.OOl ), and 6.5-fold (p<O.OOl ), respectively. Conversely, 
Zn decreased slightly to 2.5-fold (p<O.Ol), and, unlike in spring, the concentration of Cu 
was similar between the two types of plant material. Over both seasons the poplar leaves 
and the mixed-species pasture had similar (p>O.05) concentrations of Mn and Ni (Tables 
5 . 1 2  & 5 . 16). 
5.4.8.2 Seasonal effects at Hautope 1 
Major mineral elements 
Ca accumulated in the poplar leaves from spring through until late autumn (Table 5 . 1 7) .  
In contrast, the concentration of K tended (P=O.0789) to decrease between spring and 
summer, and then significantly decreased between summer and autumn (Table 5 . 1 7) .  
No significant changes occurred for Mg, Na, P, or S over the three consecutive seasons 
(Table 5. 1 7). 
In spring, the respective concentrations of Ca, Mg, and S in the poplar leaves were 5.2-
(p<O.O l), 2.2- (p<O.OOl ), and 1 .6-fold (p<O.05) greater than the average concentration in 
the mixed-species pasture (Tables 5 . 1 3  & 5 . 1 7). Conversely, P was 29% (p<O.05) lower in 
the poplar leaves compared to the mixed-species pasture (Tables 5 . 13  & 5. 1 7) .  Overall, the 
relative differences in Mg, S, and P were very similar to those measured at the Kiwitea site 
(Section 5.4.8. 1 ). Both types of plant material had similar (p>O.05) levels of K and Na 
(Tables 5 . 1 3  & 5 . 17). 
In summer, the concentration of Ca in the poplar leaves was 5.8-fold (P<O.O l )  greater than 
in the mixed-species pasture, whereas the differences for Mg and S varied little from 
spring (Tables 5 . 1 3  & 5. 1 7). Unlike in spring, the mixed-species pasture contained 43% 
more K than the poplar leaves (P<O.05). Similarly, P was also 2.5-fold (P<O.OOl)  greater in 
the pasture compared to the poplar leaves (Tables 5 . 1 3  & 5 . 17) .  
Chapter 5 263 
Trace mineral elements 
Between spring and summer the concentrations of AI, Co, and Sr in the poplar leaves 
increased by 49%, 71%, and 48%, respectively, but thereafter did not change significantly 
(Table 5. 1 8). Similarly, Mn increased by 26% between spring and summer, but in autumn 
decreased to a level that was not significantly different from either spring or summer 
(Table 5 . 1 8) .  The concentration of Zn was similar in spring and summer. However, this 
trace mineral increased by 57% between summer and autumn (Table 5 . 1 8). Cu, Fe, B, and 
Ni varied little over the three consecutive seasons (Table 5 . 1 8). 
In spring, the average concentration of Fe and Al in the mixed-species pasture was 1 . 6-
(p<0.01 )  and 2.3-fold (p<O.OOI ), respectively, greater than in the poplar leaves (Tables 
5 . 1 4  & 5 . 1 8) .  In contrast, the respective concentrations of Zn, B, Co, and Sr in the poplar 
leaves were 4.9- (P<O.OO l ), 5 .3- (p<0 . 0 1 ), 8 .4- (p<O .OO I),  and 5.0-fold (p<O.OO I )  greater 
than in the mixed-species pasture (Tables 5 . 14 & 5. 1 8). Both types of plant material 
contained similar amounts ofCu, Mn, and Ni (p>0.05; Tables 5 . 1 4  & 5 . 1 8). 
In summer, the average concentrations of Fe, Al, Cu, and Ni in the mixed-species pasture 
were 6.0- (P<O.OO I ), 3 .7- (P<O.OO I ), 1 .4- (P<O.O I ), and 1 .7-fold (P<O.O I ), respectively, 
greater than in the poplar leaves (Tables 5 . 1 4  & 5. 1 8). Similarly to spring, the levels o f Zn, 
B, Co, and Sr in the poplar leaves were 1 .4- (P<0.05), 6 . 1 - (p<0.0 1 ), 9.2- (p<0.01), and 
6 . 0-fold (P<O . O I )  greater than in the mixed-species pasture (Tables 5 . 1 4  & 5 . 1 8). This 
contrasted with Mn, which was not significantly different between both types of plant 
material (p>0.05; Tables 5 . 1 4  & 5 . 1 8). 
Chapter 5 264 
Table 5.15 Major mineral elements (�g1g DM) in Populus x euramericana leaves at Kiwitea. 
Seasoo Ca K Mg Na P S 
Spring 1 1 5 74 240 1 0  4636 799 27 1 1  4664 
Summer 1 6392 22 1 52 503 1 1440 1 945 4709 
Autumn 2 1 3 5 5  22397 5735 1 755 1 53 6  4240 
Grand mean 1 644 1  22853 5 1 34 1 3 3 1  2064 4538 
Analysis of variance 
Significance P<O . O I  NS NS P<0.05 P<0.05 NS 
SEM (0=1 2) 1283 799 1 98 1 48 168 2 1 5  
Abbreviations: DM, dry matter; SEM, standard error of th e  mean; NS, not significant �O. I ); Ca, calcium; 
K, potassium; Mg, magnesium; Na, sodium; P, phosphorus; S, sulfur. 
Table 5.16 Trace mineral elements (�g1g DM) in P. x euramericana leaves at Kiwitea. 
Season Cu Fe Mo Zn AI B Co Ni Sr 
Spring 1 0 . 1  75. 1  96.6 1 03 .7 34.0 6 1 .3 2.4 0.6 94.7 
Summer l3.3 98.4 1 49.3 93 . 5  75. 1 1 07.3 4.2 0.3 1 42.3 
Autumn l 3 .0 88.7 1 9 1 .0 1 3 1 .6 75.5 89.3 5.3 0.4 1 67.8 
Grand mean 1 2 . 1  87.4 145.6 1 09.6 6 1 .5 86.0 4.0 0.4 l 34.9 
Analysis of variance 
Significance NS P<O.OS P<O.OS P<O.OI P<O.OS P<O. 1 P<O.OS NS P<O.OI 
SEM (n=1 2) 0.6 4.3 1 2.9 7.7 6.2 6.7 O.S 0.0 10. 1  
Abbreviations: DM, dry matter; SEM, standard error of the mean; NS, not significant �0. 1); Cu, copper; 
Fe, iron; Mo, manganese; Zn, zinc; AI, aluminium; B, boron; Co, cobalt; Ni, nickel; Sr, strontium. 
Chapter 5 265 
Table 5.17 Major mineral elements (Jlg/g DM) in Popu/us x euramericana leaves at Hautope 1 .  
Season Ca K Mg Na P S 
Spring 1 7845 20388 3722 1747 1 967 3747 
Summer 2523 1 1 57 1 8  4633 281 0  1 387 441 9  
Autumn 33768 1 29 1 7  4644 2834 1 397 3098 
Grand mean 256 1 5  1 6341 4333 2463 1 584 3755  
Analysis of variance 
Significance P<O . 1  P<O.O I NS NS NS NS 
SEM (n=12) 2378 1081  209 256 1 07 267 
Abbreviations: DM, dry matter; SEM, standard error of the mean; NS, not significant (�0. 1 ); Ca, calcium; 
K, potassium; Mg, magnesium; Na, sodium; P, phosphorus; S, sulfur. 
Table 5.18 Trace mineral elements (Jlg/g DM) in P. x euramericana leaves at Hautope 1 .  
Season Cu Fe Mn Zn AI B Co Ni Sr 
Spring 1 3 .5 58.8 1 83.3 1 56.5 55.7 82.0 2 .8 1 .0 1 03.0 
Summer 1 0.9 50.7 23 1 .0 147.3 83.0 96.2 4.8 0.7 1 52.3 
Autumn 1 1 .6  5 1 .6 226.3 230.8 80.4 1 12.3 7.4 0.7 200.0 
Grand mean 1 2.0 53.7 2 13.5 1 78.2 73.0 96.9 5 .0 0.8 1 5 1 .8 
Analysis of variance 
Significance NS NS P<0.05 P<0. 1 P<0.05 NS P<0.05 NS P<0.1 
SEM (n=12) 0.7 3 . 1  1 7.7 12.9 4.3 6.5 0.7 0. 1 14.5 
Abbreviations: DM, dry matter; SEM, standard error of the mean; NS, not significant (1'2:0. 1); Cu, copper; 
Fe, iron; Mn, manganese; Zn, zinc; AI, aluminium; B, boron; Co, cobalt; Ni, nickel; Sr, strontium. 
Chapter 5 
5.5 Discussion 
5.5.1 Annual net herbage accumulation (AN HA) 
5.5. 1 .1 At the stand level 
266 
Most previous silvopastoral studies in wet and dry climates have typically shown that 
understorey pasture dry matter (DM) production is negatively related to tree canopy 
closure. However, the slope and general shape of these relationships varies markedly 
between tree species (Jameson 1 967; Pyke & Zamora 1 982; Percival & Knowles 1988; 
Mitchell & Bartling 1991 ; Knowles et al. 1 999; Devkota et al. 200 1 ;  Power et al. 2001).  
For Pinus radiata, Knowles et al. (1 999) found a strong negative linear relationship with 
pasture DM production ceasing at around 70% canopy closure (pasture growth extinction 
point). Similarly, Mitchell & Bartling ( 1991 )  and Fernandez et al. (2002) also measured 
little pasture growth at canopy closures greater than 70% for Pinus ponderosa Doug. 
(Laws). In contrast to the study by Knowles et al. (1999), pasture DM production 
(weighted ANHA) under the poplars decreased at a diminishing rate over the range of 
eeLs measured (i.e. the shape of the relationship was concave in nature) and was still 50% 
of open pasture at 70% eeL (Figure 5.2c). Power et al. (1999) also showed that young 
Acacia melanoxylon trees (6-9 years old) were able to maintain a greater level of 
understorey pasture DM production than under P. radiata, especially towards higher tree 
canopy closures. Pasture DM production under 25%, 50%, and 75% A. melanoxylon 
canopy closure was 83%, 65% and 35% of the open pasture, respectively (taken directly 
from figure), which is very similar to that measured for the poplars in the present study 
(power et al. 2001).  However, the overstorey/understorey relationship for A. melanoxylon 
was convex in shape, which differs from the linear and concave relationships found for 
P. radiata and poplar, respectively. 
5.5.1 .2 Directly underneath the tree crown (Zone 1 )  
Out of the two main zones of tree influence amongst the poplars (Zones 1 & 3) the rate of 
decline in ANHA was particularly marked in Zone 1 when initially going from the open 
pasture to a low CCL (Figures 5 .2a,b). Beyond 20% eeL Zone 1 ANHA became relatively 
constant at around 50% of the open pasture (Figure 5.2a). The strong negative concave 
curvature of this relationship shared many similarities with the relationship between Zone 
Chapter 5 267 
1 estimated PAR transmission (%DIFN) and CCLINL (Figure 3 .7a & 3 .  l Oa). Estimates 
from the developed logarithm mode� using CCL as the independent variable, also 
compared well with previous single stand density measurements taken under untended 
poplars in New Zealand (Table 5 . 1 9). At sites with low, low-to-medium, and high inter­
crown interference Gilchrist et al. ( 1993), Douglas et al. (2001),  and Guevara-Escobar et 
al. ( 1997) measured 1 1% (not significant), 14-34%, and 40% reductions in understorey 
pasture DM production compared to the open pasture, respectively (Table 5 . 1 9). It is 
likely that differences in stand canopy closure, tree age (size), and hybrid phenology would 
have confounded comparisons between these studies. The trees in Gilchrist et al. ( 1 993) 
and Douglas et al. (2001 )  studies were approximately one-third to one-half the age of the 
trees in the present study. As discussed in Section 3 .5 .2. 1 at low tree stocking rates, smaller 
trees with either greater crown height or less width will increase the quantity of PAR 
transmitted to the understorey pasture directly below their crowns. The understorey pasture 
DM production in the study by Douglas et al. (2001)  was also probably greater than 
predicted (Table 5 . 19) because the sampling area for pasture measurements extended out 
beyond the vertical projection of the tree crown. 
Table 5.19 New Zealand studies measuring annual net herbage accumulation (ANHA) under poplars. 
Source Region Poplar hybrid BA eeL ANHA Predicted 
(m2/ha) (%) (%OP) Zone 1 ANHA 
Gilchrist et al. Gisbome Populus x NA NA 89 NA 
( 1 993) euramericana 
Douglas et al. Manawatu P. maximowiczii x 2-8 1 5-45 86 60-50 
(20 0 1 )'1' P. nigra 
Otago P. x euramericana 1 -3 8-20 66 67-58 
Guevara-Escobar Manawatu P. x euramericana 14 70 60 46 
et al. ( 1 997) 
'I' Includes measurements that were taken only on the shaded (south) side of the trees and up to 0.45 
tree height distances away from the base of the tree stems. Where not provided eeL was estimated from 
BAlha (Table 3 . 1 0). Predicted Zone 1 ANHA was calculated from the logarithm model in Figure 5 .2a. 
Abbreviations: BA, stand basal area per hectare; eeL, stand in-leaf canopy closure; NA, not available. 
Chapter 5 268 
5.5.1 .3 Within the vertically projected gap between trees (Zone 3) 
Within the vertically projected gap between the trees (Zone 3) ANHA decreased by 6.6% 
of open pasture production for each 1 0% increase in eeL. This linear relationship differed 
markedly from the strong negative concave curvature found directly below the trees (Zone 
1 )  with increasing eeL. However, it did share similarities with the linear relationship 
found between Zone 3 %DIFN and eeL (Figure 3.4a), although the slope of the 
relationship for ANHA was significantly (t-test; P<O.OOOl )  less than for %DIFN. Zone 1 
and 3 ANHA converged at around SO% eeL, which supports the finding in Section 3 .4.4 
that poplar stands are effectively closed beyond this eeL. 
McIvor et al. (2003) measured pasture DM production at the midpoint (analogous to Zone 
3) between young 6 year-old 'Veronese' poplar (Populus x euramericana) with an 
estimated eeL of 20% (estimated from a BA of 2.75m2/ha). Pasture DM production was 
1 0- 1 5% lower amongst the trees than in adjacent open pasture. This compared very well 
with the predicted 14% lower pasture DM production (ANHA) estimated from the 
developed function for Zone 3 ANHA (Figure 5 .2b). In scaling up pasture DM production 
measurements taken from directly under poplar trees to a stand level, Guevara-Escobar 
(1 999) assumed the gap was equal to open pasture. However, results from this study 
clearly show that pasture DM production (ANHA) within the gap between trees (Zone 3) 
varies with gap size or overall stand canopy closure. 
5.5.1 .4 Spatial variation around the trees at Kiwitea 
Across the inter-tree space any change in ANHA depended on the eeL (Figure 5 .3). In the 
United Kingdom. Sibbald et al. ( 1 991 )  reported a similar effect amongst 3-Sm tall Sitka 
spruce (Picea sitchensis [Bong.] earr.) trees planted on a square-grid pattern at 1 56, 27S, 
and 625 stems/ha. Amongst small (3-5m tall) unpruned trees, pasture DM production was 
lower near the corners of the inter-tree space, particularly on the shaded north-eastern I side 
of the trees. In contrast, there was less spatial variation in pasture DM production amongst 
taller (Srn) pruned trees at the same stocking rates, which would have had a more 
uniformly closed canopy 'seen' from any position in the understorey (refer to Section 
3 .5.2. 1 ). 
IThe forest site was located at latitude 5 5°42'N. In the Southern hemisphere the side of a tree shaded from 
direct solar radiation is on the south side of the tree. 
Chapter 5 269 
However, the greater reduction in pasture DM production on the shaded, relative to the 
sunny, side of the trees at low overall stand canopy closures in the study by Sibbald et al. 
( 1 99 1 )  differed from the results found in the present study. Instead, the distribution of 
ANHA between mature unpruned poplar trees was symmetrical in shape, irrespective of 
CCL (Figure 5.3). Douglas et al. (2001 )  also measured significantly lower pasture DM 
production within close proximity to the shaded side of small (1 1m  tall) relatively isolated 
poplar trees, compared to the opposite sunny side of the tr�es. However, around larger 
( 1 7.4m tall) trees at another site in the same study there was no significant difference 
between shaded and sunny sides of the trees. Similarly, Clements et al. ( 1 988) also did 
not find any significant difference in pasture DM production with orientation around 
widely spaced ash (Fraxinus excelsior) trees ( 1 5m tall) at Chiddingfold in the United 
Kingdom. Since the stand level (weighted) model developed in Section 5.4. 1 .3 assumes 
that ANHA in all directions directly below the trees is similar to the shaded side of the 
tree (Zone 1), this may lead to weighted ANHA being underestimated for stands with 
small, widely spaced trees (refer to area shaded red in Figure 5 .30) .  Nevertheless, the 
weighted model may have compensated for this by assuming that beyond the tree-crown 
periphery (e.g. in Zones 2 & 4) ANHA increases rapidly to the same level found at the 
centre of the vertically projected gap (Zone 3) (refer to area shaded grey in Figure 5.30) .  
This latter assumption for the simplified 'two zones of tree influence' model also likely 
contributed to the slight « 5%) overestimation of weighted ANHA at medium stand 
densities (e.g. 50% CCL) when compared to estimates based on five zones of tree 
influence (Section 5.4. 1 .4). 
5.5.2 Seasonal net herbage accumulation (NHA) 
5.5.2.1 Directly underneath (Zone 1) and within the gap (Zone 3) between tree 
crowns 
Differences in NHA amongst the three main overstorey environments (open pasture, Zone 
1 & 3) varied with season and largely reflected the presence or absence of leaves on the 
poplar trees. The greatest decrease in Zone 1 and 3 NHA with increasing CCL occurred 
when the trees were fully in-leaf in summer and also during leaf fall in autumn. Over these 
two consecutive seasons Zone 1 NHA was reduced to around 36% of the open pasture 
beyond approximately 20% CCL. In contrast, Zone 3 NHA decreased at a slower and more 
Chapter 5 270 
constant rate with increasing CCL, reaching a similar level to Zone I once the canopy had 
become closed at about 80% CeL (Figures 5 .4a,b & 5.5a,b) . 
...... Q) ... 3 '" '" Q. 
c: Q) Q. 50 0 
"#. '-' 
:2 
z 
-< 
2 3 4 5 
Zone of tree influence 
Underest imated D Overestimated 
Figure 5.30 Schematic diagram (not to scale) of annual net herbage accumulation (AN HA) across an inter­
tree space: (-) skewed distribution with less ANHA on the shaded (south ) side of the trees and (-) assumed 
distribution for the simplified 'two zones of tree influence' model. 
Previous research has also shown that the greatest reduction in pasture DM production 
under poplar trees usually occurs around summer (Crowe & McAdam 1 992b; Crowe 1 993 ; 
Gilchrist et al. 1 993; Guevara-Escobar et al. 1 997; Guevara-Escobar 1 999; Douglas et al. 
200 1 ,  2005). However, the timing and extent of pasture recovery towards open pasture 
levels from autumn through until spring varies between different studies. In I reland, 
understorey pasture NHA below mature (>30 year old) evenly spaced (278 sternslha) 
Populus serotina trees was similar to the open pasture at the end of the growing season in 
autumn (Crowe 1 993). Whereas, in New Zealand, pasture NHA under relatively isolated 
poplars d id not fully recover to open pasture levels until winter (Gilchrist et al. 1 993 ; 
Douglas et al. 200 1 ,  2005). 
Guevara-Escobar ( 1 999) reported more variable results based on a two-year study with 
mature (unpruned) poplar trees that had high inter-crown interference. In the first winter 
understorey pasture NHA recovered to 76% of the open pasture and was not significantly 
Chapter 5 271 
different from the latter environment, but in the second winter it remained suppressed at 
only 47% of the open pasture. Similarly, across the three farm sites in the present study 
there were inconsistencies in the extent and rate at which the understorey pasture recovered 
back towards open pasture levels after autumn. At Kiwitea and Hautope 2 there was no 
significant recovery until spring; whereas at Hautope 1 ,  understorey pasture NHA was at 
least 83% of the open pasture in winter and was slightly greater than the open pasture in 
sprmg. 
5.5.2.2 Spatial variation around the trees at Kiwitea 
The more intensively stratified measurements taken across the inter-tree space showed 
NHA directly below the mature unpruned trees on both their shaded (Zone 1)  and sunny 
(Zone 5) sides was similar, irrespective of CCL or season. In contrast, NHA within the 
vertically projected canopy gap (Zones 2, 3, & 4) varied significantly depending on both 
eCL and season (Figure 5 .8). Similarly to ANHA, in summer and autumn stand level 
NHA was marginally overestimated (:SI 0%) by assuming pasture DM production 
immediately beyond the tree crowns (Zones 2 & 4) equalled that at the centre of the 
vertically projected canopy gap (Zone 3). From Figure 5 .8a,b it is apparent that there is 
actually a more graduated increase in NHA with distance from the trees during these two 
seasons. In contrast, with there being little variation in NHA across the inter-tree space in 
winter and spring (Figure 5.8c,d), omitting the two intermediate/transitional zones had 
little impact on stand level estimates. 
5.5.3 Residual herbage mass (RHM) below 25 mm trimming height 
At Kiwitea the RHM in the open pasture and in the understorey amongst the trees (Zones 1 
& 3) had a similar seasonal pattern to that measured by Cossens ( 1984) for a P. radiata 
agroforestry trial in Otago, New Zealand. Maximum and minimum RHM in both studies 
occurred during summer and winter, respectively (Figure 5 .9a). In late spring, the RHM in 
the open pasture was above 700 kg DMIha (Figure 5.9a) indicating that the herbage mass 
was within the 700-2000 kg DMIha range for set-stocked pastures where enough residual 
leaf area should have remained so not to detrimentally affect NHA markedly (Bircham & 
Hodgson 1 983, 1984; Smetham 1990). However, Zone 1 RHM was around 550 kg DMlha. 
Hence, trimming the understorey pasture in Zone 1 down to 25mm could have caused a 
Chapter 5 272 
pronounced lag in the rate of pasture re-growth at this time through insufficient remaining 
green (photosynthetic) leaf tissue. Although the RHM in Zone 3 was not significantly 
different from Zone 1 or the open pasture, the actual mean level across the range of stand 
densities studied was around 700 kg DMIba (Figure 5.9a). In late spring neither Zone 1 or 
3 RHM was affected by CCL. 
Averaged over the range of stand densities studied, the greatest difference in RHM 
between the open pasture and amongst the trees (Zones 1 & 3) occurred in late summer, 
and in general the open pasture contained at least twice as much RHM than in Zone 1 and 
3 (Figure 5 .9a). However, the significant relationship between Zone 3 RHM and CCL 
indicated that the effect of the trees on RHM depended on their density (Figure 5 . 1 0a). 
Crowe & McAdam (1992a) reported the largest difference in sward biomass density (kg 
DMlha/cm) for pastures maintained at 50 mm surface height also occurred in summer, with 
the open pasture being significantly greater than under storey pasture below mature 
Populus serotina trees. 
Based on previous sward-dynamics research for Lolium perenne-dominant open pastures, 
the late summer RHM in all three main overstorey environments should have contained 
enough residual leaf area not to adversely affect NHA (Figure 5.9a). However, due to 
morphological changes in sward structure caused by tree-shade the residual leaf area in 
Zone 1 and 3 may actually have been much lower than predicted for the given level of 
herbage mass. Changes in plant morphology under tree-shade include increases in specific 
leaf area (SLA) via taller and thinner leaves and an increase in stem/petiole length 
(Solangaarachchi & Harper 1 987; Samarakoon et al. 1 990a; Sklilovli & Krahulec 1992; 
Balocchi & Phillips 1 997b; Sanderson et al. 1 997; Wilson 1 997). As a result, a higher 
proportion of total plant leaf area could have been positioned in upper sward horizons 
reducing the leaf area available in the lower horizons for re-growth after defoliation. 
Crowe (1993) reported shading Holcus lanatus and Lolium perenne pasture for more than 
4 weeks caused the leaf biomass and leaf area index (LA!) further down the sward to 
decrease with a corresponding increase in stem biomass. 
In late autumn, both Zones 1 and 3 had a very low RHM of 300-400 kg DMlha, when 
averaged over the range of stand densities studied, whereas the RHM in the open pasture 
was around 700 kg DMIha (Figure 5.9a). The low RHM in Zones 1 and 3 along with 
Chapter 5 273 
morphological adaptations brought about by tree-shade likely resulted in very little leaf 
area remaining under 25mm trimming height, and thus would have detrimentally affected 
the rate of pasture re-growth. The significant relationship between RRM and eeL 
indicated that the effect of the trees on RHM also depended on their density (Figure 5 . lOb). 
However, the slope of the relationship was lower than in summer, indicating the RHM was 
less affected by eeL in autumn. Two factors that may have lessened the effect of the trees 
on RHM could have been an increase in PAR transmission and a change in the quality of 
light towards levels found in the open pasture, especially at higher eCLs (refer to Sections 
3.5.2 and 3 .5.7). 
In late winter, the RHM in Zone 1 and 3 changed little from autumn, but was in the range 
(300-500 kg DMIha) typically found in open pastures for that particular time of the year 
(Nicol & Nicoll 1 987; Thompson & Poppi 1 990). Crowe & McAdam (l992a) also found 
the herbage biomass density below 50mm sward surface height in winter was similar for 
open and understorey pastures. However, unlike at Kiwitea, the herbage biomass density in 
the two environments converged in the preceding autumn season. Similarly to Kiwitea, 
Zone 1 and 3 RHM at Hautope 1 and 2 was particularly low in summer and autumn, 
compared to the open pasture, with the levels likely affecting the rate of pasture re-growth 
(Figure 5 .9b,c). 
Overall, the differences in RHM between the three main overstorey environments suggest 
that understorey pasture below poplar trees should not be grazed as hard as in the open 
during summer and autumn, in order to leave adequate residual leaf area so not to 
markedly reduce herbage re-growth and thus NHA. This could be achieved through shorter 
grazing times on the particular paddock or through the use of stock such as cattle, which 
cannot graze as closely to the ground as sheep. However, the grazing intensity would need 
to be intensive enough, at least periodically, so not to allow excessive herbage loss from 
tissue death and decay (leaf senescence). 
Chapter 5 274 
5.5.4 Botanical composition 
5.5.4.1 Kiwitea 
Averaged over the range of stand densities studied, the understorey pasture in Zone 1 
contained 1 0-20% more high fertility responsive (HFR) grasses than either Zone 3 or the 
open pasture (Table 5.2). This difference was mainly due to a greater content of Poa spp. 
and to a lesser extent Holcus lanatus in Zone 1 compared to Zone 3 and the open pasture 
(Table 5 .3). Previous research has shown that both these species often make up a high 
proportion of pastures under trees. Under poplars the content of Poa has been found to 
increase in stands with low (Guevara-Escobar 1 999) and high (Guevara-Escobar 1 999; 
Crowe & McAdam 1 992b) inter-crown interference. Crowe & McAdam ( 1 992b) also 
measured a greater content of Holcus lanatus, than in adjacent open pasture, under mature 
Populus serotina trees with high inter-crown interference. Similarly, the proportion of Poa 
spp. and Holcus lanatus has also been shown to increase in New Zealand agroforestry 
trials with Pin us radiata (percival et al. 1 984; West et al. 1 99 1 ). 
In the poplar understorey (Zones 1 & 3) the Poa spp. content likely increased through 
proliferation over autumn-spring into gaps created by tree-shade (Section 3 . 5 . 8) and the 
smothering effect of fallen poplar leaf litter (Bason 1 988; McAdam 1 996). Wells & Haggar 
( 1984) found the most favourable time of the year for colonisation of newly sown Lolium 
perenne swards by Poa annua was in autumn, rather than in spring. Poa spp. generally 
have good winter growth and tillering ability compared to many other temperate grasses, 
including Lolium perenne (Vartha 1 973a; Harris 1 990). Also, coming into autumn, 
prolonged tree-shade may have reduced below ground competition from other resident 
pasture species by inhibiting their root development (Devkota et al. 1 998). 
In autumn, the density of the understorey pasture decreased with increasing eCL, as 
indicated by Figure 3 .20a for tillering and Figure 5 . 1  Ob for RHM. Based on the above 
discussion, such sward conditions would be more conducive for the ingression of Poa spp. ,  
and may explain why i n  spring the proportion o f  Poa spp. i n  Zone 1 was positively related 
to CCL (Figure 5. 1 3d). After Poa annua sets seed in spring it then dies as part of the 
obligate cycle of an annual (Harris 1 990). This, along with the intolerance of Poa spp. to 
soil moisture stress (Vartha 1 973b; Garwood & Sinclair 1 979) could have caused the 
Chapter 5 275 
significant decrease in the content of Poa spp. in Zone 1 and 3 between spring and summer 
(Table 5.3). 
Several factors may have lead to the higher content of Holcus lanatus in Zone 1 than Zone 
3 or the open pasture (Table 5.3) .  Based on an artificial shade experiment, Devkota et al. 
( 1 997) found Holcus lanatus had greater DM production and tillering ability than most 
other common hill country pasture species screened under very low PAR transmission 
levels ( 14% of incident radiation). This high level of shade was similar to in Zone 1 at 
Kiwitea in summer (Section 3 .4.2). Devkota (2000) also found sheep had greater 
preference to graze away from heavy shade if given the choice. Laxer grazing in the 
shade under trees along with the strong mid-season (summer) growth of Holcus lanatus 
would increase its competitiveness against other common hill country pasture species 
(Watt & Haggar 1 980; Frame 1 991 ). The increase in Holcus lanatus under Pinus radiata 
has been attnbuted to lower grazing pressure imposed during the initial years of tree 
establishment (Gillingham 1 984; Percival et af. 1 984). A temporary 2-3 year period of 
retirement from grazing is also required for successfully establishing poplar trees if 
unprotected planting stock is used (Stace 1 996). 
Averaged over both seasons, the open pasture contained 1 1 - 14% more Lolium perenne 
than amongst the trees in Zones 1 and 3 (Table 5.3). In the understorey of mature poplar 
stands with high inter-crown interference both Crowe ( 1 993) and Guevara-Escobar ( 1999) 
measured a lower proportion of Lolium perenne than found in adjacent open pastures. 
However, in stands with low to medium inter-crown interference the proportion of Lolium 
perenne in the understorey has been found to be similar to (Douglas et al. 2005) or greater 
than (Guevara-Escobar 1 999) in the open pasture. This indicates that the proportion of 
Lolium perenne in under storey pastures depends on stand density, which was confirmed by 
the negative relationships between Lolium perenne and CCL in Figures 5 . 1 3c and 5 . 14b. 
Similarly, the proportion of Lolium perenne under Pinus radiata has also been reported 
to decrease with increasing tree stocking rate or age (Cossens 1984; Percival et al. 1 984; 
Hawke 1 991) .  
Up to medium-to-high levels of shade the perfonnance of Lolium perenne in terms of DM 
production is comparable to more shade tolerant grasses such as Holcus lanatus and 
Dactylis glomerata (Devkota et al. 1 997; 1 998). However, as shading increases, the rate of 
Chapter 5 276 
tiller production by Lolium perenne is suppressed to a greater extent than for shade tolerant 
species, this reducing the ability of Lolium perenne to proliferate vegetatively and 
withstand defoliation (Devkota et al. 1997; 1998). In New Zealand agroforestry trials with 
Pinus radiata, lower Lolium perenne contents under trees have also been partly attributed 
to reduced grazing pressure during tree establishment (Gillingham et al. 1 976; Cossens 
1 984; Knowles 1991 ). 
Between spring and summer the proportion of Lolium perenne in the open pasture 
decreased towards levels found in Zone 1 and 3 (Table 5 .3).  This may have been due to a 
greater rate of senescence in the open pasture than amongst the trees (Crowe & McAdam 
1992b; Crowe 1993) or due to delayed phenological development (e.g. inflorescence) in 
the tree understorey (Guevara-Escobar 1 999). 
The similar legume, other species (weeds), and dead matter content in each of the three 
main overstorey environments (Table 5.2) confirms previous findings by Guevara-Escobar 
( 1 999) based on trees of a comparable age. Douglas et al. (200 1 )  also did not find any 
significant differences in each of the above botanical composition categories when 
comparing the understorey pasture on the shaded side of intermediate aged (8- 1 5  years) 
poplar trees (similar to Zone 1 )  with adjacent open pasture. However, at a site in the same 
region as Kiwitea the legume and weed content was lower on the sunny side of the trees. 
Douglas et al. (2005) also reported significant seasonal variation in the legume content at 
the same site. In summer and autumn, the proportion of legume in the open pasture was 5-
7% higher than in swards beneath the trees, whereas in winter and spring there was little 
difference between environments. 
5.5.4.2 Hautope 1 and 2 
In contrast to Kiwitea, at Hautope 1 both Zone 1 and 3 contained a higher proportion of 
HFR grasses than in the open pasture, when averaged over the range of stand densities 
studied (Table 5.4). In spring this difference was mainly due to a greater content of Poa 
spp. in Zone 1 and 3 compared to in the open pasture, whereas in summer Zone 1 and 3 
contained a greater proportion of Lolium perenne (Table 5 .5). The same factors 
contributing to the increase in Poa amongst the trees at Kiwitea were also likely present at 
Hautope 1 .  However, it is less clear why the proportion of Lolium perenne in Zone 1 and 3 
Chapter 5 277 
tended to increase between spring and summer, while not changing significantly in the 
open pasture (Table 5 .5). Site differences in the effect of the trees on soil fertility, soil 
moisture, and animal grazing behaviour, relative to the open pasture, are possible 
contributing factors. The greater content of Holcus lanatus in the open pasture (Table 5 . 5) 
was opposite to Kiwitea and could indicate that either grazing pressure or soil N 
availability was lower in the open pasture, as both factors increase this species 
competitiveness against other grasses such as Lolium perenne (Watt & Haggar 1 980; 
Frame 1 991). 
The lower proportion of legume and weeds in Zone 1 and 3 was similar to that found by 
Douglas et al. (200 1) when comparing the understorey pasture on the sunny side of 
intermediate aged trees (8- 1 1 years) with adjacent open pasture. The lower dead matter 
content in Zone 1 and 3 in summer may have been caused by a lower rate of senescence 
(Crowe & McAdarn 1 992b; Crowe 1 993) or delayed phenological development (e.g. 
inflorescence) (Guevara-Escobar 1 999). 
In comparison to Kiwitea and Hautope 2, the higher content of winter active grasses (e.g. 
Lolium perenne and Poa spp.) and lower content of dead matter amongst the trees than in 
the open pasture at Hautope 1 (Table 5 .4 & 5 .5) could account, at least partly, for the 
greater recovery in understorey NHA measured at this site in winter and spring (Section 
5 .4.2.2). Other factors such as nutrient cycling may also be involved, especially nitrogen 
(N) availability in spring (Section 4.5. 5). Several studies have shown that low to 
moderately high levels of shade can actually improve pasture growth in naturally low N 
environments (Wong & Wilson 1 980; Wild et al. 1 993; Wilson 1 996, 1 997). The high 
spring NHA in Zone 1 and 3 despite a low legume content strongly indicates the 
understorey pasture was obtaining soil N from an alternative source. 
At Hautope 2 there was little variation in botanical composition between the three main 
overstorey environments (Table 5 .6). The most prominent difference was in summer Zone 
1 and 3 had a greater proportion of dead matter than in the open pasture (Table 5.6), which 
contrasted with both of the other sites (Table 5 .2 & 5.4). 
Chapter 5 
5.5.5 Pasture and poplar leaf feed values 
278 
The nutritive value of the spring pastures at Kiwitea and Hautope 1 for all three main 
overstorey environments (Table 5 .7  & 5.8) was high and would easily meet the 
requirements of grazing sheep and cattle (Hodgson & Brooks 1 999). In summer as the 
pastures matured their nutritive values declined. However, primary indicators of nutritive 
value, such as in vitro organic matter digestibility (OMD) and metabolisable energy (ME), 
remained higher in Zone 1 and 3 than in the open pasture (Table 5.7 & 5.8) and were 
positively related to CCL (Figure 5 . 1 8 & 5 .20). Averaged for both seasons over the range 
of stand densities studied, the crude protein (CP) content of the understorey pasture in 
Zone 1 and 3 was also higher than in the open pasture (Table 5 .7  & 5.8). At Hautope 1 in 
summer the higher nutritive value amongst the trees in Zone 1 and 3 was likely related to 
the lower dead matter content in the understorey pasture (Table 5 .4), which was also 
positively related to CCL (Figure 5 . 1 6). Soluble carbohydrate (Sol CHO) was the only 
nutritive value component in the understorey pasture that was lower than in the open 
pasture (Table 5 .7  & 5.8) and in spring was negatively related to CCL (Figure 5 . 1 7  & 
5 . 1 9).  Low carbohydrate reserves would detrimentally affect initial plant re-growth after 
defoliation, especially if only low residual leaf area remained (Hodgson 1990). In contrast 
to the mature poplar stands, Douglas et 01. (2005) found very little difference in the 
nutritive value of pastures under widely spaced intermediate aged trees (8- 1 1 years) and in 
the open environment. Douglas et al. (2005) attributed this to the similar botanical 
composition in the tree and open pasture environments. 
The spring nutritive value of fresh poplar leaves at Kiwitea and Hautope 1 was similar to 
pasture, while the summer comparisons showed poplar leaves to have markedly higher 
nutritive values (Table 5 .9  & 5. 1 0).  Thus, fresh poplar leaf material would serve very 
satisfactory as a supplementary feed during either the spring or summer months. Even 
though the senescing autumn leaves had lower nutritional values (Table 5 .9 & 5 . 10) than in 
spring and summer they still would have considerable potential as an emergency fodder 
source. However, both Guevara-Escobar (1 999) and Douglas et al. (2005) found poplar 
leaves quickly lose their nutritional value once they are on the ground for more than a few 
days. 
The mineral content in spring and summer pastures at Kiwitea (Table 5. 1 1  & 5 . 12) and 
Hautope 1 (Table 5 . 1 3  & 5 . 14) was sufficient to meet the dietary requirements of sheep 
Chapter 5 279 
and cattle (Grace 1 984). At both sites, the concentrations of major mineral elements in the 
pastures of each over storey environment mirrored differences previously identified for soil 
fertility (Section 4.4. 1 ). All of the minerals measured in poplar leaves at Kiwitea (Table 
5 . 1 5  & 5 . 1 6) and Hautope 1 (Table 5 . 1 7  & 5 . 1 8), except for phosphorous (P), were also 
above minimum requirements for sheep and cattle (Grace 1 984). As a sole feed, dietary P 
may be deficient especially as the high calcium (ea) concentration that was also present in 
the leaves would depress P absorption. 
5.6 Conclusion 
At the stand level, ANHA decreased at a diminishing rate with increasing eeL. However, 
the suppression of ANHA under the poplar stands was less than for other tree species such 
as Pinus radiata, especially towards higher levels of canopy closure. Below 30-40% eeL 
stand level (weighted) ANHA was greater than 75% of open pasture production. Amongst 
the trees, the rate of decline in ANHA was much greater in Zone 1 than Zone 3 when 
initially going from the open pasture to a low eeL. This decrease appeared to be strongly 
related to the level of PAR transmission in the respective environments. Seasonally, the 
greatest decrease in NHA amongst the trees occurred in summer and autumn, with the 
change also strongly related to eeL. However, there was considerable variation in the rate 
of recovery in winter and spring between sites, which was probably caused by site-specific 
differences in pasture botanical composition and in spring soil fertility. During summer 
and autumn low RHM amongst the trees indicated understorey pasture should not be 
grazed as hard as open pasture for these seasons. 
The most consistent difference in pasture botanical composition between the three main 
overstorey environments was a greater proportion of Poa spp. amongst the trees than in the 
open pasture, especially in spring. Other differences also occurred for Holcus lanatus and 
Lolium perenne. However, these differences were inconsistent across sites and were likely 
confounded by variation in PAR transmission (%DIFN) under the trees, soil fertility and 
moisture, and animal grazing behaviour. 
The nutritive value of the understorey pasture amongst the trees was as good and often 
slightly better than in the open pasture, particularly in summer. Fresh leaf prunings from 
the trees in summer had a forage value comparable to open pasture and could be used to 
Chapter 5 280 
partly offset any reduction in understorey pasture NHA. However, once the tree leaves 
reach the ground they would reduce the nutritive value of the total feed on offer if they are 
not quickly consumed within 2-3 days. The concentration of major and minor minerals in 
the understorey pasture was generally greater than in the open pasture and likely reflected 
the higher soil pH and fertility under the trees. 
Chapter 5 281 
5.7 References 
Balocchi, o.A.; Phillips, C.l.C. ( 1 997b) The morphology and development of Lotus 
uliginosus and Trifolium subterraneum under Pinus radiata canopy in southern Chile. 
Agroforestry Systems 37: 1 5 -26. 
Bircharn, 1.S.;  Hodgson, J. ( 1 983) The influence of sward condition on rates of herbage 
growth and senescence in mixed swards under continuous stocking management. Grass 
and Forage Science 38: 323-33 1 .  
Bircharn, J.S. ;  Hodgson, J. ( 1 984) The effects of change in herbage mass on rates of 
herbage growth and senescence in mixed swards. Grass and Forage Science 39: 1 1 1 - 1 1 5. 
Clements, RO.; Ryle, G.J.A. ; Jackson, c.A.; Hanks, C. ( 1 988) Pasture growth under wide­
spaced ash trees. British Grassland SOciety: Research meeting at the Welsh Agricultural 
College, Aberystwyth. Research Meeting No. 1 .  British Grassland Society, Berkshire, UK 
Corson, D.C.; Waghorn, G.C. ; Ulyatt, M.J.; Lee, 1. ( 1 999) NIRS: Forage analysis and 
livestock feeding. Proceedings of the New Zealand Grasslands Association 6 1 :  1 27- 1 32. 
Cossens, G.G. (1 984) Grazed pasture production under Pin us Radiata in Otago. 
Proceedings of a technical workshop on agroforestry, pp. 39-48. Ministry of Agriculture 
and Fisheries, Wellington, New Zealand. 
Crowe, S.R; McAdam, J.H. (1 992a) Some effects of grazing management on a mixed 
species sward poplar-sheep agroforestry system. Proceedings of the 14th General Meeting 
of the European Grassland Federation, Lahti, Finland, June 8-11, 1992. European 
Grassland Federation. 
Crowe, S.R; McAdam, lH. (1 992b) Sward dynamics in a mature poplar agroforestry 
system grazed by sheep. Aspects of Applied Biology 29: 4 13-4 1 8. 
Chapter 5 282 
Crowe, S.R. (1 993) The response of Lolium perenne and Holcus lanatus to shading in 
relation to a silvopastoral agroforestry system. PhD thesis, Queen's  University of Belfast, 
Ireland. 
Devkota, N.R.; Kemp, P.D. ; Hodgson, I. (1 997) Screening pasture species for shade 
tolerance. Proceedings Agronomy Society of New Zealand 27: 1 19- 1 28. 
Devkota, N.R.; Kemp, P.D.; Valentine, I . ;  Hodgson, I. ( 1998) Performance of perennial 
ryegrass and cocksfoot cultivars under tree shade. Proceedings Agronomy Society of New 
Zealand 28: 129-1 3  5 .  
Devkota, N.R. (2000) Growth of pasture species in the shade in relation to alder silvo­
pastoral systems. PhD thesis, Massey University, New Zealand. 
Devkota, N.R.; Wall, A.I.; Kemp, P.D.; Hodgson, I. (2001) Relationship between canopy 
closure and pasture production in deciduous tree based temperate silvopastoral systems. 
Proceedings of the XIX International Grassland Congress: 652-653 .  
Douglas, G.B.;  Walcroft, A.S.; Wills, B.J. ; Hurst, S.B.; Foote, A.G.; Trainor, K.D.; Fung, 
L.E. (2001 )  Resident pasture growth and the micro-environment beneath young, wide­
spaced poplars in New Zealand. Proceedings of the New Zealand Grasslands Association 
63 : 13 1- 138 .  
Douglas, G.B.;  Walcroft, A.S.;  Hurst, S.B.; Potter, J.F.; Foote, A.G.; Fung, L.E. ; Edwards, 
W.R.N. ;  van den Dijssel, C. (2005) Interactions between widely spaced young poplars 
(Populus spp.) and introduced pasture mixtures. Agroforestry Systems: Direct Online. 
Eason, W.R. ( 1 988) Effect of tree leaf litter on sward botanical composition and growth. 
British Grassland Society: Research meeting at the Welsh Agricultural College, 
Aberystwyth. Research Meeting No. 1 .  British Grassland Society, Berkshire, UK. 
Ferruindez, M.E.;  Gyenge, J.B. ;  Dalla Salda, G.; Schlichter, T.M. (2002) Silvopastoral 
systems in northwestern Patagonia I :  growth and photosynthesis of Stipa speciosa under 
different levels of Pinus ponderosa cover. Agroforestry Systems 55: 27-35. 
Chapter 5 283 
Frame, J. ( 199 1 )  Herbage production and quality of a range of secondary grass species at 
five rates of fertiliser nitrogen application. Grass and Forage Science 46: 1 39-1 5 1 .  
Garwood, E.A.; Sinclair, J. (1 979) Use of water by six grass species. 2. Root distribution of 
use of soil water. Journal of Agricultural Science 93(1): 25-35 .  
Gilchrist, A.N.; DeZHall, J.R.; Foote, A.G.; Bulloch, B.T. ( 1993) Pasture growth around 
broad-leaved trees planted for grassland stability. Proceedings of the XVII International 
Grassland Congress: 2062-2063 . 
Gillingham, A.G.; Klomp, B.K.; Peterson, S.E. ( 1 976) Stock and pasture management for 
establishment of radiata pine in farmland. Proceedings of the New Zealand Grasslands 
Association 37:  38-5 1 .  
Gillingham, A.G. (1 984) Agroforestry on steep hill country: Results to year 13 .  
Proceedings of a technical workshop on agroforestry, pp. 33-38. Ministry of Agriculture 
and Fisheries, Wellington, New Zealand. 
Grace, N.D. (1 984) The determination of mineral requirements of sheep and cattle. 
Proceedings of the New Zealand Society of Animal Production 44: 1 39- 141 . 
Guevara-Escobar, A.; Kemp, P.D.; Hodgson, J. ; Mackay, A.D.; Edwards, W.R.M. ( 1 997) 
Case study of a mature Populus deltoides-pasture system in a hill environment. 
Proceedings of the New Zealand Grasslands Association 59: 179- 1 85. 
Guevara-Escobar, A. (1 999) Aspects of a poplar-pasture system related to pasture 
production in New Zealand. PhD thesis, Massey University, New Zealand. 
Harris, W. (1 990) Pasture as an ecosystem Pastures: their ecology and management (ed 
Langer, R. H. M.), pp. 75 - 1 3 1 .  Oxford University Press, Auckland, New Zealand. 
Hawke, M.F. ( 1991)  Pasture production and animal performance under pine agroforestry in 
New Zealand. Forest Ecology and Management 45 : 1 09-1 1 8. 
Chapter 5 284 
Hodgson, J. (1 990) Grazing management: science into practice. Longman Group UK Ltd, 
Essex, England. 
Hodgson, J. ; Brookes, I.M. ( 1 999) Nutrition of grazing animals. New Zealand Pasture and 
Crop Science (eds White, 1. ; Hodgson, 1.), pp 1 17-1 32. Oxford University Press, 
Auckland, New Zealand. 
Jameson, D.H. ( 1 967) The relationship of tree overstorey and herbaceous understorey 
vegetation. Journal of Range Management 20: 247-249. 
Kleinbaum, D.G; Kupper, L.L.; Muller, K.E.; Nizam, A. (1 998) Applied regression 
analysis and other multivariable methods. Third edition. Duxbury Press. Brooks/Cole 
Publishing Company, California, USA. 
Knowles, RL. ( 1991)  New Zealand experience with silvopastoral systems: A review. 
Forest Ecology and Management 45: 25 1 -267. 
Knowles, RL.; Horvath, G.C.; Carter, M.A.; Hawke, M.F. ( 1 999) Developing a canopy 
closure model to predict overstorey/understorey relationships in Pinus radiata 
silvopastoral systems. Agroforestry Systems 43(1 ): 109- 1 19. 
Littell, RC.; Freund, R.J.; Spector, P.C. ( 1 991)  SAS® System for linear models. Third 
edition. SAS Institute Inc., North Carolina, USA. 
L6pez, I . .F. (2000) Ecology of pastoral communities in a heterogeneous environment. PhD 
thesis, Massey University, New Zealand. 
Lucas, RJ.; Thompson, K.F. (1994) Pasture assessment for livestock managers. Pastures 
their ecology and management (ed Langer, RH.M.), pp 241-262. Oxford University Press, 
Auckland, New Zealand. 
Mackay, A.D.;  Saggar, S . ;  Trolove, S.N.; Lambert, M.G. ( 1995) Use of an unsorted pasture 
sample in pasture testing for sulphur, phosphorus, and nitrogen. New Zealand Journal of 
Agricultural Research 38 :  483-493. 
Chapter 5 285 
McAdam, J.H. ( 1 996) Vegetation change and management in temperate agroforestry 
systems. Aspects of Applied Biology 44: 95-100. 
McElwee, H.F. ; Knowles, RL. (2000) Estimating canopy closure and understorey pasture 
production in New Zealand-grown poplar plantations. New Zealand Journal of Forestry 
Science 30(3): 422-435. 
McIvor, I. ;  Hurst, S.; Fung, L.; Van Den Dijssel, C.; Green, S.; Douglas, G.; Foote, L. 
(2003) Silvopastoral Management of "Veronese" poplars on an erosion-prone hillslope. 
Environmental management using soil-plant systems (eds Currie, L.D.; Stewart, RB.; 
Anderson, C.W.N.), pp177-1 86. Occasional Report No. 1 6. Fertiliser and Lime Research 
Centre, Massey University, Palmerston North, New Zealand. 
Mitchell, lE.; Bartling, P.N.S. ( 1991 )  Comparison of linear and nonlinear overstory­
understory models for ponderosa pine. Forest Ecology and Management 42: 195-204. 
Nicol, A.M.; Nicoll, G.B. ( 1 987) Pastures for beef cattle. Livestockfeeding on pasture (ed 
Nicol, A.M.), pp 1 1 9- 132. Occasional Publication No. 1 0. New Zealand Society of Animal 
Production, Ruakura Agricultural Research Centre, Hamilton, New Zealand. 
Percival, N.S.; Bond, DJ.; Hawke, M.F.; Cranshaw, L.J. Andrew, B.L.; Knowles, RL. 
( 1 984) Effects of radiata pine on pasture yields, botanical composition, weed populations, 
and production of a range of grasses. Proceedings of a technical workshop on agroforestry, 
pp 1 3-22. Ministry of Agriculture and Fisheries, Wellington, New Zealand. 
Percival, N.S.; Knowles, RL. (1 988) Relationship between radiata pine and understorey 
pasture production. Agroforestry symposium proceedings (ed Maclaren, J.P.), pp 1 52-1 63 .  
Forest Research Institute and New Zealand Ministry of Forestry, Rotorua, New Zealand. 
Power, I.L.; Dodd, M.B. ;  Thorrold, B.S. ( 1 999) A comparison of pasture and soil moisture 
under Acacia melanoxylon and Eucalyptus nitens. Proceedings of the New Zealand 
Grasslands Association 61 : 203-207. 
Chapter 5 286 
Power, LL. ; Dodd, M.B.; Thorrold, B.S. (2001 )  Deciduous or evergreen: Does it make a 
difference to understorey pasture yield and riparian zone management? Proceedings of the 
New Zealand Grasslands Association 63 : 1 1 5-1 1 9. 
Pyke, D.A.; Zamora, B.A (1982) Relationships between overstorey structure and 
understorey production in the Grand FirlMyrtle Boxwood habitat type of Northcentral 
Idaho. Journal of Range Management 35(6): 769-773. 
RadcIi:ffe, J.E. (1 974) Seasonal distribution of pasture production in New Zealand 1 .  
Methods of measurement. New Zealand Journal of Experimental Agriculture 2 :  337-340 
Saggar, S.; Mackay, A; Hedley, M.l; Lambert, M.G.; Clark, D.A (1990) A nutrient­
transfer model to explain the fate of phosphorus and sulphur in a grazed hill-country 
pasture. Agriculture Ecosystems and Environment 30: 295-3 15 .  
Samarakoon, S.P.; Shelton, H.M.; Wilson, J.R (1990a) Voluntary feed intake by sheep and 
digestibility of shaded Stenotaphrum secundatum and Pennisetum clandestinum herbage. 
Journal of Agricultural Science, Cambridge 1 14: 1 43-1 50. 
Sanderson, M.A; Stair, D.W.; Hussey, M.A (1997) Physiological and morphological 
responses of perennial forages to stress. Advances in Agronomy 59: 1 72-224. 
SAS (1990) SAS/STA� user 's guide, Version 6, Volume 1. Forth edition, SAS Institute 
Inc. North Carolina, USA. 
SAS (1999) Statistics version 8. 02 for Windows@. SAS Institute Inc., Cary, North Carolina, 
USA 
Skcilova, H. ; Krahulee, F. ( 1992) The response of three Festuca rubra clones to changes in 
light quality and plant density. Functional Ecology 6: 282-290. 
Chapter 5 287 
Sibbald, AR; Griffiths, I.H.; Elsto� D.A ( 1991)  The effects of the presence of widely 
spaced conifers on understorey herbage production in the U.K. Forest Ecology and 
Management 45: 71 -77. 
Smetham, M.L. (1 990) Pasture management. Pastures: their ecology and management (ed 
Langer, R H. M.), pp. 197-240. Oxford University Press, Auckland, New Zealand. 
Solangaarachchi, S.M.; Harper, IL. (1 987) The effect of canopy filtered light on the 
growth of white clover trifolium repens. Oecologia 72: 372-376. 
Stace, C. (1 996) Poplar establishment methods and costs. Tree Grower August: 1 5- 1 7. 
Thompso� K.F.; Poppi, D.P. (1990) Livestock production from pasture. Pastures their 
ecology and management (ed Langer, R.H.M.),  pp 263-283. Oxford University Press, 
Auckland, New Zealand. 
Vartha, E.W. (1 973a) Effects of shade on the growth of Poa trivialis and perennial 
ryegrass. New Zealand Journal of Agricultural Research 1 6: 38-42. 
Vartha, E. W. (1 973b) Effects of defoliation and nutrients on growth Poa trivialis L. with 
perennial ryegrass. New Zealand Journal of Agricultural Research 16: 43-48. 
Watt, T.A; Haggar, RI. ( 1980) The effect of defoliation upon yield, flowering and 
vegetative spread of Holcus lanatus growing with and without Lolium perenne. Grass and 
Forage Science 35:  227-234. 
Wells, G.I. ; Haggar, RI. ( 1 984) The ingress of Poa annua into perennial ryegrass swards. 
Grass and Forage Science 39: 297-303 . 
West, G.G.; Dean, M.G.; Percival, N.S. ( 1991 )  The productivity ofMaku Lotus as a forest 
understorey. Proceedings of the New Zealand Grassland Association 53 :  169-1 73 .  
Chapter 5 288 
Wild, D.W.M.; Wilson, J.R; Stili, W.W.; Shelton, H.M. ( 1 993) Shading increases yield of 
nitrogen-limited tropical grasses. Proceedings of the XVII International Grassland 
Congress: 2060-2062. 
Wilson, IR (1 996) Shade-stimulated growth and nitrogen uptake by pasture grasses in a 
subtropical environment. Australian Journal of Agricultural Research 47: 1 075-1093 . 
Wilson, J.R (1 997) Adaptive responses of grasses to shade; Relevance to turfgrasses for 
low light environments. International Turfgrass Society Research Journal 8 :  575-591 .  
Wong, C.C.; Wilson, J.R (1 980) Effects of shading on the growth and nitrogen content of 
Green Panic and Siratro in pure and mixed swards defoliated at two frequencies. 
Australian Journal of Agricultural Research 3 1 :  269-285. 
Chapter 6 289 
6 Conclusions 
The research presented in this thesis provides the following conclusions: 
When initially going from the open pasture to low poplar stand densities estimated PAR 
transmission (%DIFN) decreases at a much faster rate directly below the trees (Zone 1 )  
than in the inter-tree gap (Zone 3). Thereafter, Zone 1 %DIFN becomes relatively constant, 
with the level depending on the average size of the tree. At complete canopy closure 
%DIFN is 1 5-20% and 50-55% of the open pasture in summer and winter, respectively. In 
summer the poplar canopy also reduces the R:FR, especially when PAR transmission 
levels fall below 40% of the open. 
Based on previous pasture-shade trials the above reduction in %DIFN will have a major 
effect on pasture net herbage accumulation (NHA) and general sward structure. Again, the 
impact will initially be more marked in Zone 1 rather than Zone 3 .  However, as the density 
of the poplar stand increases the productivity and structure of the understorey pasture 
becomes spatially homogeneous. Under conditions where the level of PAR is the only 
limiting factor, understorey NHA should cease at around 85% CCL, while not falling 
below 50% of the open pasture in winter because of reduced shading. At low stand 
densities the use of smaller trees with either a greater crown height (through pruning) or 
less crown width improves Zone 1 %DIFN and thus reduces any negative impact that the 
trees initially have on understorey pastures. 
Canopy closure (CCLlCCNL), measured with a standard digital camera, is a very good 
predictor of %DIFN. In particular, there is a directly proportional inverse-relationship 
between Zone 3 %DIFN and CCL. However, unaccounted for site differences in the 
relationship for Zone 1 %DIFN was evident to varying degrees for all of the stand density 
indices investigated. One of the main factors unaccounted for is likely the difference in 
average tree-crown size between sites. Especially for DBH, HPCD, and GCL as single 
independent variables, the above unaccounted for factors markedly reduced their ability to 
accurately predict %DIFN. As such, these three stand density indices should be used with 
caution when applied across multiple sites. 
Chapter 6 290 
The soil pHw and concentration of essential major cations (ea, Mg, K, & Na) and anions 
(p & S04-S) in the upper 75 mm of soil under mature stands of poplar are as good as, and 
in some cases slightly better than, in the adjacent open pasture. Also, the influence of the 
trees on soil properties is not restricted to within the crown domain (Zone 1 ), but instead 
often extends out into the inter-tree gap (Zone 3). One of the main impacts that mature 
poplars have is an increase in soil pHw, although the extent of change depends on the soil 
texture and tree rooting depth. The cycling of excess cations over inorganic anions from 
lower soil horizons to the topsoil is a major pathway for many of the changes in soil 
chemical properties that occur under the trees. Overall, the basic soil tests taken indicate 
poplars do not adversely affect understorey pastures through their effects on the soil. 
However, the seasonal changes in soil N, P, and S04-S availability associated with leaf fall 
could have both positive and negative impacts and requires further investigation. 
Under increasing eeL the change in ANHA of the understorey pasture broadly follows a 
similar pattern to %DIFN. Beyond 20% eeL, Zone 1 ANHA becomes relatively constant 
at around 50% of the open pasture. In contrast, Zones 3 ANHA decreases from open 
pasture levels at a constant rate of 6.6% for every 10% increase in eeL. Given the 
decrease in ANHA appeared to be strongly related to the change in %DIFN, the use of 
short, narrow, or high pruned trees should reduce the suppression of ANHA in Zone 1 .  
Seasonally, the greatest decrease in NHA amongst the trees from open pasture levels 
occurs in summer and autumn. Thereafter, understorey pasture production recovers 
towards open pasture levels. However, there were site differences, with the trees at 
Hautope 1 having a less detrimental impact on NHA than at Kiwitea and Hautope 2 in 
winter and spring. The cause may have been site-specific differences in botanical 
composition and in spring soil fertility. Fresh leaf prunings from poplars in summer have a 
nutritive value similar to open pasture and could be used to partly offset any reduction in 
understorey pasture production during this season. 
The botanical composition and feed value of the understorey pasture is broadly similar to 
that of the open pasture. The most consistent difference in botanical composition was a 
greater content of Poa spp. in the understorey pasture. In contrast, the decrease in ANHA 
directly below mature unpruned trees (Zone 1) and amongst trees at high stand densities is 
substantial and will have a significant impact on farm profitability if situated over a large 
area of a farm. Therefore, ongoing management of poplar stand density is necessary to 
Chapter 6 291 
minimise the negative impact that trees have on pasture ANHA. To maintain ANHA levels 
at around 75% of the open pasture CCL should not exceed 30-40%. At Kiwitea and 
Hautope 1 ,  this is equivalent to less than 30 mature unpruned trees per hectare. 
Chapter 4 292 
Appendix 4. 1 Experimental un it selections for soil samples taken at 75-1 50mm soil 
depth . 
4.1 .1  Kiwitea 
Experimental unit Zone 3 %DIFN Overstorey density Selection 
class 
42 90 * 
40 90 Open 
41 86 * 
3 9  7 8  
30 56 * 
Low 
3 6  5 4  
23 5 1  * 
28 30 
38 30 * 
3 29 Medium 
29 26 * 
24 26 
7 IS  * 
6 1 3  
High * 
4.1.2 Hautope 1 
Experimental unit Zone 3 %DIFN Overstorey density Selection 
class 
2 1 00 * 
99 
Open * 1 
6 50 * 
3 43 Low * 
4 3 8  
8 34 
5 29 * 
Medium 
1 0  27 
7 26 * 
1 1  24 
9 2 1  High * 
12 20 * 
Chapter 4 
Appendix 4.2 Annual Ca returned in the pasture and tree biomass at Kiwitea. 
Assumptions: 
a. Net primary production = 1 3  t DM/ha/yr (refer to section 5.4. 1 ); 
b. Poplar leafbiomass = 3 . 1  t DM/ha/yr (Guevara-Escobar 1999); 
c. Ca concentration in autumn tree leaves = 2. 1 3% (refer to section 5 .4.8. 1 ); 
d. Ca concentration in understorey and open pasture = 0.48% and 0.41 %, 
respectively (refer to section 5 .4.7. 1) ;  and 
e. All organic material is returned to the soil 
Quantity of Ca returned to the soil in the poplar-pasture system. 
Poplar leaflitter: 3 100 kg DMlha/yr x 2. 1 3% Ca 
Understorey pasture: 9900 kg DMlha/yr x 0.48% Ca 
Total 
Quantity of Ca returned to the soil in the open pasture system. 
Open pasture 1 3000 kg DM!ha/yr x 0.41 % Ca 
Difference between systems. 
kg CaJha/yr 
= 66.0 
= 47.5 
= 113.5 
= 53.3 
= 60.2 
293 
Chapter 4 
Appendix 4.3 Mass-balance sulfur cycling model 
Based on formulas given by SincIair & Saunders (1 984) and Nguyen & Goh ( 1993). 
Assumptions 1 :  
a. Annual stocking rate2 = 1 2.65 stock units per hectare (SU/ha) 
b. T otaI sulfate lost in animal products (assuming steep hill) = l A  kg S/SU 
c. End of season leaching of sulfate = 6 kg S/ha 
d. Net sulfate innnobilisation = nil 
e. Atmospheric sulfate input = (Ledgard & Upsdell 1 991)  
£ Fraction of rainfall sulfate leached = 0.25 
g. Soil test (0-75 mm depth) = 4.8 Quick test units 
1 .  Pasture development index (PDI) 
PDI = years since pasture improvement commenced (to a maximum of20) 
x average stocking rate over that period 
x 0.005 for steep hill country. 
= 20 x 12.65 x 0.005 
= 1.27 
2. Sulfate leaching index (SLI) classification = 5 
3 .  Maintenance sulfate requirement (S) 
294 
S = loss in animal products (a x b) + end of season leaching (c) + immobilisation (d) 
- sulfate added from rainfall (e x I -f). 
= ( 12.65 x lA) + 6 + 0 - ( e x  (1  - 0.25) 
= 23 .71 kg S/ha - 1 .5 kg S/ha 
= 22.21 kg S/ha 
4. Modification of the maintenance requirement (MS) due to previous pasture development 
MS = maintenance requirement (S) x [2 - 0. 125 x soil test (g) - 0.5 x -vPDI] 
= 22.21 x [2 - 0. 125 x 4.8 - 0.5 x -V(1 .27)] 
= 18.66 kg S/ha 
5.  Sulfate added (SA) from annual topdressing with single superphosphate @ 1 1  % sulfate 
SA = 220 kg/ha x 0. 1 1  
= 24.2 kg S/ha 
6. Excess sulfate added in fertiliser above the modified maintenance requirements 
= 24.2 kg S/ha - 1 8.66 kg S/ha 
= 5.5 kg S/ha 
1 Many of the assumptions are directly related to the sulfate-leaching index (SLI) (Sinclair & Saunders 1 984). 
2 Average stocking rate between the maximum of 14.8 SU/ha achieved on the farm in 1 969 and the current 
rate of 1 0.5 SU/ha. The grazing system was classified as "set-stocking or extensive rotational grazing" given 
that the current stocking rate was less than 75% of the potential carrying capacity (Metherell & Morrison 
1 984, cited Nguyen & Goh 1 993). 
Chapter 5 295 
Appendix 5. 1 Regression parameters (standard errors in parentheses) for 
equations predicting ANHA from in-leaf stand canopy cover (CCl). 
5.1.1 Zone 1 ANHA (% open pasture), where Y = 1 00*(X + 1) b 
Site b r RMSE N P< 
Kiwitea98-99 -0. 1 896 (0.0209) a IjI 0.85 10.829 1 7  0.0001 
Kiwitea99-00 -0.2495 (0.024 1 )  a 0.90 1 0.388 1 7  0.0001 
Hautope 1 -0. 1 044 (0.0062) b 0.95 3 .686 1 4  0.000 1 
Hautope 2 -0.2 1 84 (0.0597) ab 0.72 1 8.039 8 om 
5.1.2 Zone 3 ANHA (% open pasture), where Y = 1 00 + b*X 
Site b r RMSE N P< 
Kiwitea98-99 -0.5723 (0.0868) a 0.73 12. 143 8  1 7  0.000 1 
Kiwitea99-00 -0. 8 1 22 (0.0859) ab 0.85 12.0198 1 7  0.0001 
Hautope 1 -0.5664 (0.0564) ac 0.88 6.43 1 1 1 4  0.0001 
Hautope 2 -0.9248 (0.2474) abc 0.65 1 7.4939 8 om 
5.1.3 Weighted ANHA (% open pasture), where Y = 1 00*ebox 
Site b r RMSE N P< 
Kiwitea98-99 -0.0099 (0.0012) a 0.84 10.72 1 5  1 7  0.000 1 
Kiwitea99-00 -0. 0 1 42 (0.00 15) b 0.89 10.843 1 1 7  0.000 1 
Hautope 1 -0.0067 (0.0004) c 0.95 3 .7757 1 4  0.000 1 
Hautope 2 -0.0 143 (0.0043) abc 0.67 1 7.65 1 1  8 0.0 1 
'l'Within column, regression parameters with the same letter were not significantly different (P>0.05). 
140 
� 1 20 � 0 0 i gj 
140 
� 120 � 0 (b) 0 i gj 
(a) 
p.. 1 00 
I:l <U p.. 
p.. 1 00 
[ 
80 0 � � 
� 
60 
40 
0 80 � '-' 
� 60 
« 40 
.... . ..... . .. .. . .... ... � ... ... . . ........ .  .a\ 
• o 
<"l 
<U 
I'l 20 � 
<U 
I:l 20 • 0 N 
0 0 
0 20 40 60 80 1 00 0 20 40 60 80 1 00 
Canopy closure with leaves (%) 
140 
I 1 20 � 0 (c) 0 i 
p.. 
I:l 1 00 <U p.. 0 I;j 
� 
� 80 
1 
� I!! 
� 60 
.� . .... � .... \\ 
"0 40 <U .... 
-a 20 .i) 
� 
0 
0 20 40 60 80 100 
Canopy closure with leaves (%) 
Appendix 5.2 Annual net herbage accumulation (ANHA) over a range of in-leaf canopy closures. Symbols: (0) Kiwitea98-99 (.) Kiwitea99-00, 
CA)  Hautope 1 ,  (0) Hautope 2. I\.) CO 0)