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. 
 



Growth Studies with Hemp (Cannabis saliva L.) 

A thesis presented in partial fulfillment of the requirements for the degree of 

Master of Science (Horticultural Science) 

At Massey University, Palmerston North, 

New Zealand 

Arthur Randall Gibson 

2007 



Growth studies with hemp (Cannabis sativa L). 

Abstract 

The effect of latitude, sowing date and density on yield, height and growing time were 

studied by sowing two varieties of hemp for seed at three sites in New Zealand. 

Predation and poor germjnation hampered results. Latitude effects on height could not 

be confirmed but Anka, unlike Finola, suffered a reduction in height with later sowing. 

Anka was the larger plant with higher seed yield, but the dwarfed Finola gave a higher 

harvest index. Thus seed yield for later sowing is reduced for cv Anka, unchanged for 

cv Fino la. Plant density data allowed estimated yields at 30, 60 and 90 plants. m-2 to be 

determjned. At 30 plants.m-2
, seed yield in both varieties was reduced with later sowing , 

and confirmed Anka with the greater gross yield against Finola having the higher 

harvest index. At 60 and 90 plt.m-2
, the results confirmed only that Anka is the bigger 

plant. Earlier sowing than the 1 st sowing in our trial (October) could be an advantage. 

Higher sowing rates than 40-50 plt.m-2 would be of little advantage for Anka, but rates 

higher than 90 plt.m-2 should be explored for Finola. The asymptotic yield maximum for 

Anka falls dramatically with later sowing, whereas Finola does not. This suggests 

double cropping might be explored, with an early sowing of Anka followed by Finola. 

Growing time data indicated the lower latitude site had longer growing times overall. 

Day length sensitivity (shorter growing time with later sowing) was confirmed for 

Anka, but Finola was unaffected. This suggests Finola is a day-length neutral plant. 

Leaf production in both varieties was higher at higher latitudes, and was reduced by 

later sowing date. 

A second experiment examined the nitrogen and phosphorus uptake of a fibre hemp 

cultivar (EILl) with respect to three sowing dates and two latitudes. Harvests were 

taken monthly to establish growth trends. From biomass figures, equations were 

established for leaf/stem relationships. Uptake values (on a dry matter basis) tended to 

stabilize as the plant matured at: N(stem) 3.04%, N(leaf) 4.5 %, P(stem) 0.26% and 

P(leaf) 0.45 %. High initial concentration of both N and P fell rapidly. For maximum 

uptake seed should be sown early (October or earlier) and plants grown as long as 

possible (5 months or more). 

I 



Acknowledgements 

I would like to thank my supervisor, Dr. M . A. Nichols, for his unstinting advice and 

encouragement. His knowledge and experience were invaluable, when apparently 

incomprehensible data drove me to the point of despair. Financial support from the 

Foundation for Research in Science and Technology was invaluable, simply because 

without that support, and that of my mentor, Ashley Berrysmith of NZ Freshcuts (land 

and personnel for trials), the whole project would have been impossible. The help and 

advice of Massey University staff, from both the Soils Laboratory (chemistry analysis), 

and Plant Growth Unit (laboratory services) were vital for preparation of data. Last, but 

of course not least , I must thank my fiancee Susan Lawrence for her support. As my 

most ardent critic, she provided advice when I needed it, stress when I didn't (or didn't 

think I did) and above all belief in myself when I had none. Without all these 

extraordinary people the project would not have begun, let alone been completed. I 

thank them all. 

II 



Table of Contents 

Abstract ........................................................................................................................ ! 

Acknowledgements .................................................................................................... II 

Table of Contents ...................................................................................................... III 

List of Figures ............................................................................................................ V 

List of Tables ............................................................................................................. VI 

1. Introduction ......................................................................................................... l 

2. Experiment 1: Growth Parameters of Industrial Hemp (Cannabis sativa L.) 
for Seed in New Zealand . ......... .... .... ...... ......... ........ .... ............... ....... ....... ... ........ 3 

2.1. Introduction ................................................................................................. 3 

2.2. Materials and Methods ......... ..... ..... ......... .... .................. .. .. ........ .. .. ........ ...... 3 
2.2.1. Sites and Cultivars .. .. ... .... ............... ........................ ..... .. ........ .. ... ... ...... .. 3 
2.2.2. Problems and Limitations ...................................................................... 4 

2.2.2.1. Seed quality ................... .. ...... ... .... ........ ... ............ ............ ......... .. ... 4 
2.2.2.2. Predation ........................................................................................ 4 
2.2.2.3. Watering in .................................................................................... 5 
2.2.2.4. Fertilisation .................................................................................... 5 
2.2.2.5. Wind damage ................................................................................. 5 

2.2.3. Analytical Methods ................................................................................ 6 
2.2.3.1. Height analysis .. ........................... ................ .... .. ... ........ ......... .... .... 6 
2.2.3.2. Time .................. .. ... ... ... ... .... ..... .. ............................ .... .. .......... ... .... 6 
2.2.3.3. Seed yield ...................................................................................... 6 
2.2.3.4. Plant density effects .............. .. ......... ..... ...... ............. ... ................... 7 
2.2.3.5. Growing time .. ................................................ ... .. .......... ................ 7 
2.2.3.6. Leaf weight .................................................................................... 8 

2.3. Results and Discussion ................................................................................ 8 
2.3.1. Height analysis .. ... .................................... ..... .. .......... .... ..... .. ...... .. ......... 8 

2.3.1.1. Site effect .. ... ..... ... .... .......... .. .................... ........... ...... .. ................... 8 
2.3.1.2. Variety effect: ...... .. ...... ............. ..... ........ .... ...................... ............ .. 8 
2.3.1.3. Sow-variety interaction .... ....... .......... ....... ........ ............... ... ..... ....... 9 

2.3.2. Seed yield analysis ............................................................................... 10 
2.3.3. Plant density ........................................................................................ 11 

2.3.3.1. Sowing .... ..................................................................................... 11 
2.3.3.2. Variety ......................................................................................... 12 
2.3.3.3. Slope (productivity with maximum competition) .......................... 12 

2.3.4. Plant density effects at selected calculated densities ............................. 12 
2.3.4.1. Density 30 plants.m·2 

............•.....•..•.•.•...........•.......•••......••..•........ 13 

2.3.4.1.1. Sowing date··················-····-··-··-··-·-······-···································· 13 

2.3.4.1.2. Variety effects ... ... ... .... ......... ... ......... .... ......... ... ...... ...... ..... ... _ .... 13 

2.3.4.1.3. Sow* Variety interaction····-······· ················· ··· ··-··--···-··············· 14 

2.3.4.2. Density 60 plants.m·2 
..•....•....•..•..•...••.............••......•........•.......•...• 14 

2.3.4.2.1. Variety: ........... .................. .. ......... ......... ..... ........ .. ................. .... 14 

2.3.4.3. Density 90 plants.m·2 
................................................................... 15 

III 



2.3.4.3.1. Variety ... ..... ..... .. ... ................... .......................... ....... ................ 15 

2.3.4.4. Summary of Plant Density Effects ...................................... .. ....... . 16 
2.3.5. Growing time analysis ........ ........ .. ...... ...... .. ........ .. .... ... ... ... ... .... .. ... ... ... 18 

2.3.5.1. Site effect ... .... ......... ...... ... ..... .. .... ................. ..... .. .. .. ................. .... 18 
2.3.5.2. Sowing date effect.. ...................................................................... 19 
2.3.5.3. Site * sowing interaction ........... ... ... ... ... ..... ... ..... ... ............. ..... ..... 19 

2.3.6. Leaf weight analysis ..... .. .................... .. ....... ... .... .. ....... ......... ..... .......... 20 
2.3.6.1. Site effect .... .. .............. .................................................... .......... ... 21 
2.3.6.2. Sowing date effect. ....................................................................... 21 
2.3.6.3. Site * sowing date interaction ..... ..... ..... ...... ... .. ... .. .... .... .. ........ ..... . 22 

2.4. Conclusions .... .... ....... ... ... ......... .... .... ........... .. .. ... ... .. ... ... ... .... .. ........ .... ........ 22 

3. Experiment 2: Growth and nutrient uptake of industrial hemp (Cannabis 
sativa L.) in nutrient rich effluent . .... ....... ....... .... ........... ... .. ... ..... .... ... ... ... .... .... 25 

3.1. Introduction ... .... .. ....... ...................... .... .. ... .......... ......................... ..... ........ 25 

3.2. Materials and Methods .......... .. ..... ............. .. ...................... ... ...... ........... .... 26 
3.2.1. Sites and Cultivar ......... .... ...... .............. ................................... ........... .. 26 
3.2.2. Problems and Limitations ..... ....... .. ... ...... ... .. ........... ... ....... ... ..... ... ... ..... 26 

3.2.2.1. Herbicide damage ....... .. ...... ........ ... ........ .. ...... ... ........... .... ... ... ... ... 26 
3.2.2.2. Weighting limitations ................................................................... 26 

3.2.3. Analytical Methods ................................................................... ..... ...... 27 
3.2.4. Allometric Relationships ..... ..... ... ... .... .... .. ... .. .... ... ...... .... ... .... ...... .... .. ... 27 
3.2.5. Nutrient uptake ................................... ... .. ...... .. ....... ............................. 27 

3.3. Results and discussion ... ... .. ... .. ........ .............................. .. ..... ..... ... ......... .... 28 
3.3.1. Allometric Relationships .. ..... ............ ..... ... ... ... ........ ... .... ...... .. .. ... ... ...... 28 
3.3.2. Nitrogen and phosphorus uptake ......... .......... ................ .. .... ......... ....... . 30 

3.3.2.1. Stem and leaf nitrogen uptake ... .. ... ....... ................................. ... ... 30 
3.3.2.2. Stem and leaf phosphorus uptake ........ .... ....... ................ ... ........... 30 
3.3.2.3. N and P uptake for stem .. ... ... ... ........ ... .. ... ...... .......... ..... ............... 31 
3.3.2.4. N and P uptake for leaf.. ............................................................... 32 

3.3.3. Harvest date effects ..... ... .. .... .. ................. ......... .. ....... .. .. ...... ....... .. ........ 32 

3.4. Conclusions ....... .. ...... ............. .... ..... ................... ... ... ... .... .. .......... ..... ........ .. 33 
3.4.1. Leaf/stem dry weight ........ ........... .. ........ ...... ........ ......... ... ... .. ........ ..... .. 33 
3.4.2. Nitrogen and phosphorus uptake ................................... ... ... ... ... .. ........ . 34 

4. General Conclusions ... ...... ................. .. ..... ........ ... .............. .......... ......... .. .......... 35 

5. Future Work ...... .... ..... .. ................................... .... .......... ........ .. .... .. .... ...... ..... ... .. 36 

References .... ....... .................................. ............. ...................... ............ ........... .... ...... 38 

Appendix A ... ...... ..... .. .... .... ...... .. ............ ..... .. ........... .... .................. ... .... .. .. ....... .... ..... . 41 

Appendix B ..... ..... ... .. ................. .......... .. ........ .......................... ....... ... .. ... ... .. ...... .. ... ... 41 

Appendix C ............ .. ...... .... .... ....... .... .... .. ............. ..... ......... ............ ........................ ... . 42 

Appendix D ... ....... ... ..... .. ........ .... .... ... .... .. .. ..... .. .. .. ..... ... .......... ..... ... .. ..... ..... .... .... ..... ... 43 

Appendix E ....... ..... ..... .... ... .... ... .... ... ... .... ........ ..... .... ... ......... ... ...... .. ......... ...... ....... ..... 43 

Appendix F . ........ ........... .... ...... ..... .... ......... ......... .. ........ ...... ... ... ... .... ........ ..... ... .......... 43 

Appendix G ... ........ .... ...... ...... .... .. .. ... ....... ........... ..... ....... ........ ... ... .. .... .... .. .... ... .......... 44 

IV 



List of Tables 

Table 1 Site Means of Overall Plant Height .... ...... .... .. ... ........ .... .. ... ..... .. .... .. .. .. ....... 8 

Table 2 Varietal Means of Overall Height ......... ... ...... ... ....... ...... ........ ....... ..... ... .... . 8 

Table 3 Overall intercept means ... ..... ... ... ... .. ... ....... ........... ... ... ....... ...................... 12 

Table 4 Effect of Variety on Above Ground Biomass .......... .... .. ... .... ....... .......... ..... 13 

Table 5 Variety Effects of Density at 60 Plants.m·2 
. . .... . . . .........••.. ... •. . ...•.• ..• •... . ...•• 14 

Table 6 Variety Effects at Density of 90 Plants.m-2 .... ........ ....... ......... .... ... .......... 15 

Table 7 Site Effect for Growing Time to Harvest ... ..................................... ..... .... .. 18 

Table 8 Leaf I Biomass Ratio Means for Kerikeri and Rangiriri . ........................ .. 21 

Table 9 Site v Sowing Date Interaction for Leaf/Biomass Ratio .. .................... ...... 22 

Table 10 Nutrient uptake (g.m· 2
) for 30 plant.m-2 

......... . ..... . ..... .•.•.....•..... . ......•....... . 33 

VJ 



1. Introduction 

Industrial hemp ( Cannabis saliva L) is a potentially valuable crop, which represents an 

emerging industry in New Zealand (McIntosh 1998, McPartland et al. 2004). The 

history of hemp is very old however, and is even mentioned by Herodotus, the "father 

of history" as a source of cloth in burial ceremonies (Herotodus 73-75). Rollins (1731 ), 

as part of a diatribe on the need for a productive working life, refers to the usefulness of 

hemp in clothing - in context apparently for some 4000 years. This is confirmed by Lu 

and Clarke (1995) who present a detailed account of the history and use of hemp in 

China from circa 4000BC. They report the cultivation and use of hemp in c lothing, 

paper, food and medicine. They give details of sowing dates, fertiliser requirements, 

plant densities and cultivation dating from some 200 years BC. Clearly hemp cultivation 

was widespread and well understood in pre-biblical time, at least in China. Captain 

James Cook ( 1779) refers to New Zealand flax in a comparison to hemp "There is one 

plant that serves the natives instead of hemp and flax , and which excels all that are 

applied to the same purposes in other countries." Thus hemp has seen many centuries of 

use, in Western society and ancient Greece, as well as in China, in clothing, rope, oil 

and medicine (although medicinal properties may refer primarily to high THC 

varieties). 

In more recent times, research effort has been somewhat sparse, if increasing. Work is 

reported in various aspects of fibre reinforcement in automotive construction (Gulati 

and Sain 2005), and ballistic reinforcement (Wambua et. al. 2007). 

Detailed work is also reported on the biochemical or nutritional characteristics of hemp 

oil and its constituents (Tang et. al., 2006, Calloway 2004, Calloway, 2006, Leizer et. 

al. 2000 and others), and this effort also appears to be increasing. 

On the subject of the growth of hemp plants, less work seems to have been reported, 

although again the effort is apparently increasing. Some comment is made of the genetic 

variability of hemp, emphasized by statements that the plant is dioecious, with enough 

variability to select monoecious varieties (Moliterni et. al., 2004 ). Moliterni noted that 

monoecious varieties will revert in time to dioecious, a situation which must be 

monitored. Plant density and self thinning effects in relation to fibre yield and quality 

has been reported (Werf et. al. 1995, Drew et. al. 1977). Werf found that self thinning in 

hemp is more pronounced than in other plants. One trial (Lehoczky, 2006) reports the 

l 



effect of hemp intrusio n into ma ize as a weed, and Bennett et. al. (2006) reported the 

e ffect of variety and harvest time on fibre quality. 

L ittle appears to have been repo rted on the nitrogen and phosphorus uptake potential of 

hemp, although Amaducci et. al. 2002 reported a hemp trial on plant density (30-250 

p lt.m-
2

) and N fertilization (100-200 kg.Ha-' ), finding no significant differe nces in yield 

between (fibre) varieties. 

This project was intended to examine the growth parameters of industrial he mp, in an 

atte mpt to derive the most appropriate growing conditions (planting date, latitude and 

dens it y) fo r the production o f hemp seed oil in New Zealand . Given the paucity of 

publicly ava il able in for matio n o n he mp growth for seed in this country, a preliminary 

trial was devised to identify areas for further, more detailed study. 

Conc urrently with our trial, however, results fro m an undergraduate experiment 

indicated that hemp may also be of bene fit as a bioremediation crop (G ibson and 

N icho ls in prep). A second experiment was undertaken to explore this possibility, by 

sowing the same cultivar on two sites. T his investigation therefore invo lves two distinct 

experime nts. 

Experiment I : three (arable) sites, two replicates, two varieties, fo ur sowing dates and 

three sowing dens ities. 

Experiment 2: two (efflue nt disposal) sites, one cultivar, three sowing dates, three 

sowing densities and monthly har vests. 

2 



2. Experiment 1: Growth Parameters of Industrial Hemp (Cannabis sativa L.) for 

Seed in New Zealand. 

2.1. Introduction 

As a commercially useful plant, hemp has been used for centuries for cloth and rope, 

and as food and medicine (Lu and Clarke, 1995). More recently, the beneficial 

effects of hemp seed oil have attracted more detailed attention (Callaway, 2004, 

Callaway et. al. 2005) and hemp is currently being cultivated for this purpose (Tang 

et. al. 2006, McIntosh 1998, McPartland et. al. 2004.). Little has been published on 

the growth parameters of the plant in New Zealand, and in particular there seems 

little known of the effects on seed harvest index of latitude, plant height, maturation 

time and yield, although Struik et. al. (2000) did report latitude effects in relation to 

fibre. A detailed model development project under controlled laboratory conditions 

is also reported (Lisson et. al. 2000), based on the APSIM-hemp protocol, which 

derives a full model encompassing all major growth factors, except latitude. Further, 

the cultivars used in their trial were different than available to us, and any 

relationship to the current trial could not readily be assessed. 

A growth analysis experiment (Gibson and Nichols in prep) has been conducted 

under hydroponic greenhouse (NFT) conditions, but these results may have little 

bearing on field parameters, and clearly contribute nothing to the effects of latitude 

or sowing densities. The current experiment was intended to address some of these 

issues, and define future areas by sowing at different latitudes, at different dates and 

sowing densities, with particular emphasis on oil, (i.e. seed) rather than fibre (stem) 

production. Latitudes covered in this New Zealand study ranged from 35° 14'S 

(Kerikeri) to 40° 22'S (Massey), being the broadest range logistically feasible, and 

available. 

2.2. Materials and Methods 

2.2.1. Sites and Cultivars 

The cultivars used were cv Anka and cv Finola at Massey University Tiritea 

campus (Fruit Crops Unit, longitude 175° 37'; latitude 40° 22'S); Rangiriri, 

South Waikato, longitude 175° 07'; latitude 37° 26 'S and Kerikeri, Bay of 

Islands, longitude 173° 56 '; latitude 35° 14'S. Anka is a Canadian bred, 

monoecious cultivar, certified for cultivation in Canada from 1999. It is medium 

3 



height, and can be harvested for either oil or fibre, under appropriate co nditions 

for eac h. Fino la is a cultivar de veloped and bred in Finland specifically as an 

early flowering, short stemmed, oil produc ing dioecious variety. It has been 

certified for cultivatio n in Canada since J 999. 

Seeds in our trial were direct drilled into prepared plots, as follows. Two 

replicates of J .25m w ide plots were sown, each containing three 2.0m sub-plots 

end-to-end, at three sowing densit ies for each c ultivar, with a 1 .Orn guard plot at 

each e nd. Consecutive sowings, one month apart, were sown alongside the 

previous sowing. Harvest samples o f l .Oni each were take n from the centre of 

eac h plot, such that each sample had guard rows surrounding it. Bird netting was 

used to cover the freshly sown area. 

Seed beds were no t ro utinely irrigated. This was to ensure natural growmg 

conditions, and to avoid unwanted soil/water/plant interactio ns due to the close 

prox imity of earlier sowings. Water was applied under excessively dry 

condit ions. 

Regular (monthly) site visits were scheduled for main te nance, sowmg and 

harvesting. 

2.2.2. Proble ms and Limitations 

Several problems were e ncountered during the growing phase o f the trial, which 

impacted on the extent and qualit y of the data collected. 

2.2.2.1. Seed quality 

Due to de lays in licensing, import and availability, sowing at the planned 

starting date (September) was not possible. Sowing therefore commenced in 

October 2005, using residual Anka and Fino la seed fro m an earlier trial. In 

fac t some seed was 4 years o ld at purchase (thus 5 years at sowing) which 

resulted in poor germination rates. We increased sowing rates throughout the 

experime nt as this problem became apparent. 

2.2.2.2. Predation 

B ird netting was not totally effecti ve, and other predators, inc luding (but no t 

limited to) slugs, springtail, possums, mice and pukeko a lso contributed to 

low plant densities. 

4 



2.2.2.3. Watering in 

Seeds were not routinely watered in, to minimize crossover interaction on 

the adjacent, previously sown plots. In some cases current soil moisture may 

not have been sufficient to ensure good soil/moisture contact with the 

germinating seed. 

2.2.2.4. Fertilisation 

The sites were either fertilized, or believed to be of adequate fertility. This 

may have been insufficient in some cases. Where poor germination was 

encountered, however, weeds grew well. Thus fertiliser _application should 

not have been the dominant problem at these sites. 

2.2.2.5. Wind damage 

A severe wind event at the Rangiriri site, some weeks after first sowing, 

affected some plants. The smaller (Fino la) plants were damaged to some 

degree by this, but Anka appeared unaffected. In some aspects of data 

reduction, this produced anomalous results, in which case Finola data was 

removed. In most cases, however, the damage did not appear to serious ly 

affect results, and data was therefore retained. 

The above factors, or interactions between them, resulted in losses - the first 

sowing at the Kerikeri site, and the first three sowings at Massey. Newer seed, 

availab le toward the end of the trial, was used for a late sowing at Massey in late 

January. Anka produced harvestable seed, but replicates were notably different. 

Finola did not produce a crop, possibly due to severe weed infestation. Data 

from this sowing was inspected in aJJ aspects of data reduction, but was 

excluded due to unacceptably high errors, and lack of equivalent data from other 

sites. 

Harvest was conducted as closely as possible to the "80% ripe" stage. The 

geographic spread of sites and problems of determining ripeness remotely made 

this difficult. This, and the Joss of ripe seed due to shattering and bird predation, 

resulted in some errors in the seed harvest data. 

From each plot at each site, the following data were recorded: 

• Average height at harvest (m), 

• Weight of above ground biomass (stem, foliage and seeds). 

5 



• Sowing and harvest dates. 

• Plant density (plant.m-2). 

2.2.3. Analytical Methods 

Data were analysed based on the premise that plants grow exponentially. 

Transforming data to natural logarithms will thus produce a straight line, from 

which a linear regression yields intercept and slope for interpretation. The 

regression also yields an estimate of error for statistical analysis. Similarly, an 

asymptotic limitation to growth such as competition for light, nutrient or other 

resource can be modeled by a reciprocal transform. Again, the resulting linear 

regress ion allows interpretation of the intercept and slope of the line. 

The regression data were analysed by analysis of variance (SAS Institute Inc., 

Cary, NC, USA version 9.1). Initially, for each analysis, a full factorial model 

including all main effects and interactions was prepared. The least significant F 

ratio from this analysis was then removed, and the process iterated until only 

factors at the 5% significance level remained. This model was then examined. 

2.2.3.1. Height analysis 

Height (m) was recorded and analysed. 

2.2.3.2. Time 

Time to harvest was recorded in days. Sowing time was recorded as months 

from first sowing (October). 

2.2.3.3. Seed yield 

Seed yield and above-ground biomass were converted to per plant , and 

transformed to natural logs. These data were plotted, displaying the seed to 

biomass ratio and highlighting anomalous data due to seed loss from bird 

strike and shattering. 

The following protocol was devised to accommodate this loss of seed. Linear 

regressions of (ln) seed weight v (In) biomass were calculated for each 

cultivar with an assumed slope of 1.0. Where the (exponentiated) seed 

weight was less than 25 % of the estimated seed weight (based on the 

regression equation) that data was discarded. Removal of this data altered the 

regression relationship, and the process was iterated until no further data 

6 



warranted removal. The residual dataset thus obtained was analysed, and the 

significance mode l examined. 

2.2.3.4. Plant density e ffects 

It was assumed that at high plant densities, internal shading would result in 

an asymptotic yie ld maximum. This could be modeled by a reciproca l 

transform according to the 'de nsity equation' 1/w = Ap + B (Drew and 

Flewelling, 1977), where p is plant density in plants.m-2, and w is the plant 

yie ld. Plant density was therefore regressed against reciprocal above ground 

biomass yie ld (per plant) from each site , sowing date and variety. 

Replicates were combined. Suspect data was scrutinized by (a) plotting data 

fro m each variety (i.e. combining data fro m a ll sowings and s ites) and (b) 

plotting data fo r each sowing and variety (i.e. combining site data) . These 

plots highlighted anomalous data due to errors in labe ling, late harvesting 

and other handling errors. Massey data was found to be unusable due to 

pauc it y and high variability, and was excluded from further analys is o f thi s 

parameter. 

Intercepts (yield with no competitio n) o f plant de nsit y regressio n coeffic ie nt s 

so deri ved were analysed by analysis o f variance. Analys is o f slo pe data 

(yie ld with maximum competition) was not possible due to the se lection 

procedure as discussed. The regression equations were however used to 

calculate expected productivity values at se lected plant de ns ities of 30, 60 

and 90 plants.m-2, based on the range of densities observed. T hese values 

(for reciprocal expected yie ld) were analysed, regressed and the resulting 

equations used to calculate a trend line at interpolated values (0 - 90) for 

each cultivar. The reciprocal slope fro m these equations is presented as the 

asymptotic maximum yield for each cultivar. 

2.2.3.5. Growing time 

Days to harvest was analysed w ith respect to month of planting, fro m 

October through January. So me unavo idable error was introduced due to 

slight variation in sowing date at several disparate sites. Due to logistic 

difficulties, harvesting could be two weeks fro m ideal ripeness. 

7 



2.2.3.6. Leaf weight 

Leaf weight was recorded as per plant, and compared to the total above 

ground biomass. Data was converted to natural logs, analysed by analysis of 

variance and examined. 

2.3. Results and Discussion 

2.3.1. Height analysis 

The significance model (appendix A) was derived. Only sowing data from 

Rangi.riri and Kerikeri were included. The (5 %) significant effects found were 

for Site, Variety and the sowing* variety interaction. 

2.3.1.1.Site effect 

The means (Table 1) show a significant site effect, with the more southerly 

site taller by 24.6%. In the absence of Massey data (the most southerly site) 

this effect cannot be reliably linked to latitude (a trend requires at least 1 

degree of freedom, or at least 3 points), and thus requires further study. The 

differences observed could be due to uncontrolled local variables such as soil 

type, drainage, moisture regime and climate, rather than latitude. 

Table 1 Site M eans of Overall Plant Height 

Site Means 

Site Mean (m) 

Kerikeri 0.65 

Rangiriri 0.81 

Note: Combined mean heights of anka and fin ola cultivars. 

2.3.1.2.Variety effect: 

Variety means (Table 2) clearly reflect the expected overall height difference 

in these two varieties. 

Table 2 Varietal Means of Overall Height 

Varietal Means 

Variety Mean (m) 

Anka 1.01 

Finola 0.45 

Note: Combined mean of Rangiriri and Kerikeri sites, and all sowings. 

8 



The height of Anka, reported here at 1.01 m, was observed on-site as often 

considerably taller than this. Reasons become clear in Figure l below, and 

are discussed in that context 

2.3.1.3. Sow-variety interaction 

The means are presented in Figure 1. 

1.5 

I 
1: , 
Cl 

"iii 
::c 

0.5 

• Anka = -0.1671x + 1.5596 

R' = 0.8914 

&,,....---.....,6,---- 6----..A 

Finola = -0.0235x + 0.5348 

R' = 0.552 

O+----~---~--~------~ 

Sowing Date (months) 

Figure 1 The effect of sowing date on var ieta l height. 

• anka 
ll finola 

- Linear (anka) 

- - Linear (finola) 

This figure shows that while Finola height is virtually unaffected by later 

sowing, Anka is markedly so. It could be speculated that Finola, unlike 

Anka, is a day length neutral flowering variety. This is unusual for hemp, 

possibly related to the extremely high latitudes (Finland) where Finola was 

developed. Further trials, with extended sowing dates (and possibly 

laboratory trials) would help to clarify this, but the possibility could exist for 

double cropping if Finola (or a similar) variety is involved in late sowing. 

The height data presented above elicits some information, some possibilities and 

some indications for further work. Since stem growth normally ceases at 

floration , it follows from Figure l that Anka is a day-length sensitive plant, but 

Finola may not be. Finola was developed in Finland (Callaway, 1996) 

specifically as an early flowering variety, and it is perhaps unsurprising that this 

is so. If true, the result highlights the natural genetic variability of hemp 

(Moliterni et.al. 2004), and offers encouragement to developers breeding 

varieties for specific growing conditions. 

Of interest is the relatively low average height of Anka (Table 2) compared to 

Fino la. The reason becomes clear with reference to Figure l; Anka can be grown 

9 



to some I.Sm with early sowing, but will attain Jess than I.Orn if sown late. For 

oil seed production this could be significant. Anecdotal accounts of harvest 

difficulties with taller plants highlight the need for a management tool to 

minimize these problems, and later sowing can clearly achieve this. However the 

impact of low height on seed yield limits this advantage, as discussed in section 

2.3.2 below. 

The height difference shown (Tab le 1) between si tes is also of interest. Studies 

have shown some biochemical responses to latitude, notably THC/BCD ratios 

(Leizer et. al. 2000), and it could be that height is also affected. It is unfortunate 

that Joss of data at our most southerly site precluded statistical confirmation of 

this (at least three sites are required to indicate a trend), but the possibility 

certainly warrants further study. If confirmed, a further management tool would 

be available indicating high latitude sites for fibre production, and low latitudes 

for oil seed production. Note that this result relates to height only, not 

specifically seed yield, which is discussed below. 

2.3.2. Seed yield analysis 

The graphed final dataset, selected as described above (section 2.2.2.3), is shown 

in Figure 2. 

... q. 1 

01 

£ 0 
.E 
01 
"iii ., 
s: 
"C 
a, ·2 
a, 

en 
.3 

• 

0 
Finola = 1.0501 x • 1.8702 • 

R2 = 0.5984 

• 
• 

• 

. .,,. 

0 .,, .,, 

0 .,,. 

OIi O / / 
/ 0 0 a 

o.,..... / a> 0 0 
.,, • 0 

• .. .,,. 
0 .,,. .,,. 

/ o 
.,,. 0 

.,, 0 0 
0 

0 

o / _,, • Anka = 1.0895x - 2.6826 

_,,. _,, 1° 0 R2 = 0.6925 
0 

0 

0 

.4 +-------------------~ 
-0.5 0 .5 1.5 2.5 3 .5 4.5 

Above Ground Biomass ln(g.plf1
) 

o anka 

• l inola 
- Linear (hnola) 

- - Linear (anka) 

Figure 2 Final dataset of ln(seed) v ln(total biomass) , with damaged data removed 

This shows the clear difference between varieties, and that the selected data (for 

both varieties) has a slope overall very close to 1.0, as defined by the transform 

and selection process. With the average varietal slope set to 1, the intercept for 

individual points, as (ln)seed yield/biomass (g) can be calculated as follows; 

JO 



(ln)intercept = ln(seed) - (ln)biomass. 

The intercept differences displayed between cultivars are as follows; 

Finola intercept= (exp) -1.8702 = 0.154 g(seed).g(biomass- 1
) 

Anka intercept= (exp) - 2.6826 = 0.054 g(seed).g(biomass- 1
) 

Thus finola has a higher harvest index, by a factor of almost 3. Anka is a larger 

plant however, and further analysis is required to confirm which cultivar is the 

higher producer on an area basis. This point is discussed below. 

2.3.3. Plant density 

From the basic density equation, 1/w = Ap + B, where p is plant density in 

plants. m-2
, regressions were obtained for each site, sowing and variety 

(replicates combined), yielding an intercept (reciprocal yield with no 

competition) for each variable . Slope data (reciprocal yield at asymptotic density 

limit) could not be directly analysed , as the slope had been arbitrarily set to 1.0 

during the selection process (section 2 .2.3.3). Analysis of variance (intercept) 

produced the following result s, including sow ing date and variety effects, as 

shown in appendix B. 

2.3.3.1.Sowing 

Massey data was again unu sable, and Rangiriri sowing one (October) data 

was excluded due to wind damage. Kerikeri sowing one data was also Jost. 

The results (Figure 3) therefore contain three points, which do however 

illustrate a clearly significant trend. 

'<11 

"' 

0.160 

E 0.,20 

0 
ii 
~ 

¥ 0.080 

"' :§ 
'C 
ai 
> 0040 
'C 

" Ji 

y = -0.0349x + 0.2154 
R' = 0.9916 

0.000-1--------~ --~---~--~ 

0 

Sowing Date (months) 

• Seed 
- unear (Seed) 

Figure 3 Effect of sowing date on expected seed yield at intercept (no competition) 

11 



This figure demonstrates the overall effect of later planting times on 

expected harvest index for the two cultivars (combined) under study, 

assuming no compet ition between plants. 

This confirms the importance of early sowing, as reported above in relation 

to height, and provides a gu ide to how effective such a strategy could be. 

2.3.3.2. Variety 

Overall means for Anka and Finola harvest index, are presented in Table 3, 

including data converted from reciprocal means. 

Table 3 Overall intercept nieans 

Variety Effect On Seed Yield (Intercept) 

Variety Mean (In) Mean (exp) 

Anka -2.6771 0.069 

Finola -1. 7786 0.169 

Note: Mean (In) extracted from data in Appendix B. 

This table shows the higher harvest index of Finola compared to Anka, with 

no competition. Finola harvest index is some 2.4 times that of Anka under 

these conditions. 

2.3.3.3. Slope (productivity with maximum competition) 

The est imat ion of maximum productivity from thi s data was not possible, as 

discussed above (section 2.3.3). Data variabilit y and seed losses required an 

arbitrary assumption of an overall slope of 1.0 to extract statistical ly 

significant effects. Although technically possible to examine internal 

differences due to site, sowing date and variety, re lating any such 

conclusions to real world conditions is not practical, and thus not reponed. In 

any future trial, with more contro lled and consistent data, such an analysis 

may well be appropriate. 

2.3.4. Plant density effects at se lected calculated densities 

Biomass productivity data estimates were prepared for dens ities 30, 60 and 90 

plants.m·2, and analysed separate ly to preserve data independence. 

12 



2.3.4.1.Density 30 plants.m-2 

Analysis gave significant effects for sow, variety and the sow * variety 

interaction (appendix C), as follows. 

2.3.4.1.1. Sowing date 

Results of the effect of sowing date on total above ground biomass are 

shown in Figure 4. 

500 -N 

'E 
d, 

- 400 
"O 
ai 
> 
1/) 
1/) >JO 
m 
E 
.Q 
al 
"O 200 

C: 
:::, 

e 
~ 100 

> 
0 
.0 
< 

• 

• 

biomass = -107 .4959x + 517.4422 

R' = 0.8665 

• 

• 

0+-------------------
0 

Sowing Date (months) 

• biomass 
- Linear (biomass) 

Figure 4 The effect of sowing date on total above ground biomass at 30 plants.m-2. 

This figure confirms a reduction in total above ground biomass yield 

with later sowings. It will be noted that the sowing * variety interaction 

is significant at this density level which, since both varieties perform 

differently, implies that this finding should be considered in conjunction 

with the interaction analysis below (section 2.3.4.1.3). 

2.3.4.1.2. Variety effects. 

Data from the two varieties are presented in Table 4. 

Table 4 Effect of Variety on Above Ground Biomass 

Effect of Variety on Yield 
(30 Plants.m·2

) 

Variety Mean 

Anka 

Finola 

321.2 

115.3 

Note Higher productivity of Anka at 30 plants.m-2 

This table shows that Anka has an expected biomass yield nearly three 

times that of Finola (ratio 2.8). This again confirms Anka as the larger 

13 



plant, and suggests that Finola could probably be sown at higher 

densities than Anka. 

2.3.4.1.3. Sow* Variety interaction 

Results of this interaction are shown in Figure 5. 

-N ·e 
~ 
"C 
a; 
> 
V) 

1000 

800 

Vl 600 
ea 
E 

.Q 
ID 
"C 400 

C 
:::, 

e 
Cl 

~ 
0 
~ 
< 

200 

• 

• 

Anka biomass= ·194.1540x + 864 .9394 

R' = 0.8053 

er - - - _o_ - - - ...Q 
Finola biomass = -20.8378x + 169.9449 - -

R2 = 0.7736 O+----~---~---~--~---~ 

Sowing Date (months) 

• anka 
0 finola 

- Linear (anka) 

- - Linear {finola 

Figure 5 Interaction effects of both varieties against sowing date at 30 plants.m·2
. 

This figure dramatically demonstrates the difference in effect of sowing 

date for the two varieties (at this density level). Anka productivity is 

markedly reduced by later sowing dates, whereas Finola reduces very 

little. Figure 5 also effectively resolves the issues raised in Figures 3 and 

4, highlighting the different response to sowing date by the two varieties. 

This could be significant if double cropping is contemplated, as late 

sowing of Anka is clearly unproductive compared to Finola. 

2.3.4.2.Density 60 plants.m·2 

Results are presented in appendix D, indicating the significant effect of 

variety alone. 

2.3.4.2.1. Variety: 

Means for this effect are shown in Table 5 

Table 5 Variety Effects of Density at 60 Plants.m·2
• 

Effect of Variety on 
Yield (60 Plants.m-2

) 

Variety Mean 

Anka 

Finola 

376 .7 

156.5 

14 



This table shows that the overall above ground biomass at 60 plants.m-2 

is in similar ratio between varieties to that at 30 plants .m-2 (ratio 2.4 ), but 

at increased level for both varieties. Thus at 60 plants.m-2, the asymptotic 

yield limit for biomass productivity has not been approached by either 

variety. 

2.3.4.3. Density 90 plants.m-2 

Results at this sowing density are presented in appendix E. 

2.3.4.3.1. Variety 

Means for the only significant effect (variety) are presented in Table 6. 

Table 6 Variety Effects at Density of 90 Plants.m-2 

Effect of Variety on Yield 
(90 Plants.m-2

) 

Variety 

Anka 

Finola 

Mean 

403.9 

180.5 

Again , the difference between varieties is in si milar ratio to that found at 

30 and 60 plants.m-2 (ratio 2.2), and is again numerically higher in both 

varieties than for 30 or 60 plants.m-2. Since progressively higher yields 

are recorded for 30 60 and 90 plants.m-2
, it follows that the asymptotic 

yield limit is not reached at these levels. Due to lower than expected 

plant densities in this trial, calculations at densities above 90 were not 

warranted, but should be explored in further trials. The derived 

equations, however, were combined with seed yield data to calculate 

trend lines for seed yield as shown in Figures 6 (Anka) and 7 (Finola). 

15 



2.3.4.4. Summary of Plant Density Effects 

"i 
E 
~ 
"O 

80 

60 

aj 40 

> 
l / --------------: ~~~~---· 

20 ,,..,,. , ... ' ............. ... . .... - · 

:/(-~:-~:_. -- ----- - --·- ·- ·- -· - --·-

sow 1 asymptote= 60.0 

sow 2 asymptote= 34.3 

sow 3 asymptote = 28.4 

sow 4 asymptote = 19.1 

0+------~--~--~--~-~--~ 
20 40 60 80 100 120 140 

Plant Density (plt.m.2) 

- ankasowl 
--- ankasow2 

· • · •· anka sow 3 

- ·- ·anka sow4 

Figure 6 Seed yield trend line for Anka, showing asymptotic maxima 

Figure 6 clearly shows an asymptotic trend for Anka, illustrating that earlier 

sowing dates provide considerably higher yields. The possibility that earlier 

sowing could be even more productive is clearly shown. It is also clear that 

sowing densities for Anka greater than 40 plants. m·2 give progressively less 

response with later sowing. Thus earlier sowing will attain higher seed yield 

with lower densities. 

40 

~ 30 
N 

·e 
~ 
-0 
cij 20 

;;: 
-0 
Q) 
Q) 
en 

10 
I 

I 
I 

I 

I 

I 
I 

/ 
/ 

/_: / 
, : / 

I . /. 

I:.:/ 
1!·, ., 

20 

,,--------

.' 
/ 

40 

,,' 

60 

---

BO 

- sow 2 asymptote= 44 .0 

sow 3 asymptote = 42.7 

sow 4 asymptote = 41 .4 

100 120 

Plant Density (plt.m-2
) 

140 

- - - finola sow 2 

· --·· lmolasow3 

- - - ·linola sow 4 

Figure 7 Seed yield trend line for Finola, showing asymptotic maxima 

Figure 7 illustrates that Finola also exhibits higher yields for earlier planting, 

with the possibility of higher yields for even earlier sowing dates, as does 

Anka. Only three sowings are presented for Finola, due to missing and 

damaged data. The trend lines indicate that Finola could benefit from higher 

plant densities than derivable from our data, possibly in excess of 100 plants 

16 



per square metre. Again, higher de nsities will be required with later sowing 

to maintain high yields. 

The asymptotic seed yield limits (exponentiated s lope from the derived 

equations) for both Anka and Finola are compared in Figure 8, showing 

trends for the two culti vars. 

80 

-N 

E 
~ 60 
QI .. 
0 .. a. 
E 
>, 40 

"' <( 

"Cl 
'a; 
> 
-g 20 

• 
Fino la = • 1 .3029x + 46.593 

A'= 0.9992 

• 

• anka 
o linola 

- linear (anka) 
- - linear (finola) 

QI 
en Anka = - 13.138x + 68.56 

A'= 0.8924 

0 

0 3 

Sowing Date (months) 

Figure 8 Asymptotic seed yields for vari eties v sowing date 

This figure clearly illustrates the effect of sowing date on seed yield 

performance of Anka and Fino la. Anka productivity clearly reduces with 

later sowing, whereas Fino la does not. Thus Anka requires an early sowing 

date to perform well, possibly earlier than achieved in our trial (October), but 

Fino la performance is effectively unaffected. It will be recalled that earl y 

sowing of Anka results in taller plants, and that this cou ld impact on 

harvesting problems; this would not be true of Fino la. 

"Cl 
'a; 
> 
.!:? 

1200 

0 600 
Q. 
>-
"' <( 

"' :(l 300 

E 
0 
ai 

• 
Anka = -192.09x + 1002.4 

R2 = 0.8925 

Finola = ·8.4612x + 302.36 

R2 = 0 .9993 

0-l-------~---~- --~--~ 
2 3 

Sowing Date (months) 

Figure 9 Effect of sowing date on biomass for Anka and Finola 

17 

• anka 
D hnola 

- Linea, (anka) 

- - L,nea, (hOOa) 



The figure shows the biomass yield reduction for Anka with delays in 

sowing date, where Finola is again essentially unaffected. When compared to 

Figure 8, it also shows that Finola harvest index is higher, when planted late. 

This confirms the possibility for double cropping with Anka sown early, 

followed by Finola. 

This possibility should be explored with further trials, extending the overall 

growing period (possibly required to allow adequate growing time for both 

varieties), and identifying any harvest problems due to height which may 

occur. 

2.3.5. Growing time analysis 

The model resulting from these data included site, sow, and the site * sow 

interaction, as presented in appendix F. These effects, significant at the 5% level, 

were based on omitting all Massey data, and the first sowing at Rangiriri. 

2.3.5.1.Site effect 

The means for site data are presented in Table 7 

Table 7 Site Effect for Growing Time to Harvest. 

Site Effect for 'Days To 
Harvest' 

Site 

Kerikeri 

Rangiriri 

Mean 

76 

59 

The growing time would appear to be significantly shorter at Rangiriri than 

at Kerikeri, which is considerably to the North. This result could indicate a 

latitude effect but, in the absence of the Massey data (the most southerly site) 

to establish a trend, this could not be verified . If such a latitude effect could 

be established, it would impinge on the double cropping concept mentioned 

above. It may, for example, be possible to double crop at higher latitudes, 

but not at lower. Conversely, the earlier sowing dates feasible in warmer 

climates (lower latitudes) could negate this. Further trials are required to 

clarify the extent and importance of these effects. 

18 



2.3.5.2. Sowing date effect 

The results are presented in Figure 9. A clear reduction in growing time with 

later planting is apparent, notwithstanding high variability. 

80 

en so 
> 
"' ~ 
Cl) 

E 
i= 40 

Cl 
C: 

'3: 
e 

(!) 20 

y = ·6.2772x + 86.452 
R2 = 0.4967 

• 

• growing time 
-Linear( rowi lffle) 

0+----~--------------

Sowing Date (months) 

Figure 10 Effect of sowing date on growing time to harvest 

Again, this result requires clarification. In the absence of clear varietal 

differences (discussed in section 2.3.5.3), either variety could be day-length 

sensitive. In which case early sowing of the first day length sensitive crop 

would not necessarily result in earlier harvest. Other indications, as outlined 

above, suggest that Anka, but not Finola, could be day-length sensitive. If so, 

early sowing of Anka would provide higher yield, but not necessarily earlier 

harvest. This requires a determination of the latest feasible sowing date, 

which is not clear from the data. 

2.3.5.3. Site * sowing interaction 

Means for this effect are presented in Figure 11. 

u, 
> 
"' "C 

100 

80 

~ 60 

E 
j:: 
Cl 
C 40 

-~ 
c'.; 

20 

_________ ,. 
D D 

y = 59.2778 

R2 =0.0000 

Sowing Date (months) 

• Kerikeri 

D Aanglrlri 

Figure 11 Sowing date versus growing time interaction effect between sites 

19 



Despite inherent errors, and paucity of data, this figure clearly shows the 

difference in growing time between the two varieties. It would appear that 

Finola can be sown as late as January and provide an adequate crop. 

Growing time therefore, depends on variety and sowing date, with the 

possibility of a latitude effect. Certainly, it has been shown that latitude 

affects some biochemical processes (Leizer et. al. 2000), and these results 

could indicate more fundamental effects. Clearly a more detailed 

investigation into day-length sensitivity and latitude effects is warranted, at 

least for the varieties studied here. 

2.3.6. Leaf weight analysis 

Leaf weight (natural log) was plotted against (natural log) total above ground 

biomass, as shown in Figure 12. 

-(11 

~ 1 

Leaf = 0.9001x · 0.5081 

R2 = 0.8981 

• 

Above ground Biomass ln(g.plf1
) 

• leaf wt 
- Linear (leaf WI ) 

Figure 12 Leaf weight per plant plotted against total above ground biomass 

Figure 12 shows a clear correlation between total leaf weight and above ground 

biomass, irrespective of variety, with a slope of 0.9001. Thus as the (natural log) 

total weight rises, the natural log leaf weight rises less, by a factor of 0.9. As 

weight is a function of height, it follows that as the plant matures, it 

allometrically changes from leaf to stem production. This could be important as 

a factor in quality fibre production or, assuming a difference in chemical uptake 

between leaf and stem (discussed in section 3.3.2), in bioremediation. 

Analysis of the ratio of (ln) leaf wt and (In) above ground biomass, with Massey 

data excluded, yielded results as shown in appendix G. Significant effects 

included site and sowing date, and the interactions site * sow, and site * variety. 

20 



With variety not significant as a main effect, the interaction of variety and site 

was not considered further. 

2.3.6.1. Site effect 

Means for this effect are shown in Table 8. 

Table 8 Leaf I Biomass Ratio Means for Kerikeri and Rangiriri. 

Leaf/Biomass Site Effect 

Site 

Kerikeri 

Rangiriri 

Mean 

0.49 

0.67 

Mean(exp) 

1.63 

1.96 

Mean differences for this effect, with Rangiriri being higher by 37% as 

natural logs, 20% as the exponent, could be due to latitude effects. As 

before, loss of data from the Massey site precludes a clear conclusion but, 

together with other results in this thesis, indicate the need for further detailed 

studies into latitude effects. 

2.3.6.2.Sowing date effect 

This effect is demonstrated in Figure 13. 

0.8 

.,g 06 

Ill ... 
1/) 
1/) 
Ill 
E o.• 
0 
iii 
--'i 
Q) 

...J 0.2 

Leaf ratio= -0.0765x + 0.7988 

R2 = 0.9875 

0+-----------------~ 

Sowing Date (months) 

Figure 13 Effect of sowing date on leaf/biomass ratio 

• leafratio 

- linear (leaf ratio) 

Figure 13 clearly shows an effect of planting date on the (In) leaf wt, (In) 

above ground biomass ratio, which decreases with later sowing date. This 

could be a significant result, given that allometric changes, and therefore 

growth distribution could affect stem and fibre quality. It will be noted, 

21 



however, that later sowing dates result in lower height maxima, lower seed 

yields and lower above ground biomass for Anka. 

This result, therefore, could reflect a plant response to changing day-length, 

vernalization, and the need to harvest more daylight during shorter day­

length regimes. It is clear, however, that random events such as rainfall and 

sunshine hours cannot explain the clear trend shown. Further, unmonitored 

variables such as soil type, drainage and fertility levels are also not 

responsible, as they did not change during the experiment. Thus latitude 

remains the likely candidate, by mechanisms not clear from the data. Future, 

more detailed investigations are required to explain these results, and other 

allometric inferences. 

2.3.6.3.Site * sowing date interaction 

Means for this effect are shown below in Table 9 

Table 9 Site v Sowing Date Interaction for Leaf/Biomass Ratio. 

Leaf/Biomass Site * Sowing Date 

Site Sow Mean 

Kerikeri 2 0.53 

Kerikeri 3 0.41 

Kerikeri 4 0.51 

Rangiriri 1 0.71 

Rangiriri 2 0.76 

Rangiriri 3 0.70 

Rangiriri 4 0.46 

This table illustrates the wide variation of (In) leaf wt v (In) above ground 

biomass compared over the two sites reported. No simple trend is apparent in 

this case, and the results may be due to confounding and uncontrolled 

factors. 

2.4. Conclusions 

Plants sown at the higher latitude (Rangiriri) grew 24.6% taller than at the lower 

latitude (Kerikeri). Absence of precise data from the southernmost site (Massey), 

and consequent loss of the third point required to establish a trend, precluded 

verification of a direct link to latitude however, and a more comprehensive trial is 

22 



required to exclude environmental effects. Shorter plants could be advantageous for 

harvesting of seed however, and confirmation of this finding could affect 

appropriate sites for cultivation, either for seed (high latitude) or stem (low latitude). 

Lower seed yield under these conditions could result, negating this advantage. 

The taller variety, Anka, was shown to grow nearly twice as tall as Finola, but more 

significantly, the height of Anka is considerably reduced by later sowing. Finola was 

not greatly affected. If final height is a consideration when planning cultivation, 

whether for stem (fibre) or seed ( oil) care should be taken in selection of site, variety 

and sowing date. 

Anka is the larger plant, but biomass and seed production was considerably reduced 

by later sowing. In contrast, Finola biomass and seed production was unaffected by 

delayed sowing. The possibility therefore exists for increasing overall productivity 

by double cropping, with Anka followed by Finola. Extended sowing dates could 

improve this possibility, and should be explored in further trials. 

Plant density analysis showed that, without competition, seed yields conformed to 

the directly measured findings, indicating that under the conditions of our trial, 

relatively little competition occurred. Analysis of above ground biomass at 30 

plants.m-2, showed yield reduced with later sowings by some 75 %. Thus a 

considerable penalty in biomass (and therefore seed yield) is attracted by delays in 

sowing date. As the smaller (albeit more productive) plant, Finola biomass 

productivity was virtually unaffected by later sowing, thus highlighting the dramatic 

effect of later sowing of Anka. If later sowing is envisaged, whether due to weather 

constraints or proposed double cropping, Finola should be encouraged compared to 

Anka. At 60 plants.m-2, and at 90 plants.m-2
, Anka produces considerably more 

biomass than Finola, and each variety produces more than at lower densities. In 

neither case is the asymptotic yield (yield with maximum competition) approached, 

although Anka seed yield shows relatively minor response to densities above 50 

plants.m-2. Densities above 90 plants.m-2 for Finola should be explored in further 

trials. 

Analysis of growing time to harvest showed that the lower latitude site (Kerikeri) 

required longer growing time than did the higher latitude (Rangiriri) site. Clear 

co1Telation with latitude could not be established, due to loss of data from the 

highest latitude (Massey) site. The difference was substantial however (76 days v 59 

23 



days) so the possibility should be explored in further trials. Growing time was also 

significantly affected by sowing date, to a similar degree. Again, early sowing data 

was unavailable, but sowing in November required 76 days to mature on average, 

sowing in January only 60. It was found that Anka was again greatly affected by 

later sowing, but Finola was not. Thus it would appear that Finola is not a day­

length sensitive plant, but Anka conforms to the norm for hemp. 

Comparing leaf weight (In) with above ground biomass (In) showed a correlation, with 

a slope of 0.9. Thus as the plant increases in size, relative leaf production falls, by a 

factor of 2.5. The ratio of ln(leaf weight) to ln(above ground biomass) was shown to be 

significantly affected by site and variety. The lower latitude site gave a higher ratio, and 

the ratio was lower with later sowing date. Any possible latitude effect could not be 

confirmed in the absence of precise data from the highest latitude site under study. 

24 



3. Experiment 2: Growth and nutrient uptake of industrial hemp (Cannabis 

sativa L.) in nutrient rich effluent. 

3.1. Introduction 

The potential of bioremcdiation as an effluent purification method has long been 

known. McCall and Gibson (1979) found that down-slope runoff through grass root 

mat assisted removal of nitrogen and phosphorus from drainage water. This is an on 

going subject of research, as shown by Sawabe et. al. (2006), who ide ntified 

matches between target pollutants and appropriate plants. In recent times, studies 

have been conducted on sewage e ffluent (nitrogen and phosphorus) purification 

(Yang Lei; et. al. 2001), comparing different wetland substrates using napier grass 

as a test plant. 

Little has been reported in the literature regarding the growing of industrial hemp 

for the removal of nitrogen and phosphorus from nutrient rich solutions, although 

hemp is well documented as a medium for removal of heavy metals (Tlustos et al. 

2006) and hydrocarbons (Liste, 2006). For several years, hemp has been used in a 

project to purify soils, including the radioactive contaminated soils near Chernobyl 

(Charkowski 1999; Khan 2005). 

In a recent study (Gibson & Nichols in prep), the authors have defined the growth 

and nutrient uptake characteristics of industrial hemp unde r NFf hydroponic 

conditions. The inference was that hemp may have va lue as a conunercially usefu l 

bioremediation crop, particu larly in the removal of N and P. Little has been repo11ed 

on N and P uptake in he mp, but a study has been reported on uptake by forage 

grasses in North Eastern U.S. Ketterings, et. al. (2006) reported rates for N, on 

average over several fertiliser application rates, of 0.0200 g(N).g(DM)" 1
, and for P, 

0.0029 g(P).g(DM)" 1
, and Fosu (2004) investigated nitrogen accumulation and 

release by sunn hemp, calopo mucuna and devil bean. 

This section of the thesis examines the growth and nutrient uptake characteristics of 

hemp under New Zealand field conditions. 

25 



3.2. Materials and Methods 

3.2.1. Sites and Cultivar 

Two New Zealand sites were used, one at Feilding (Lat 40° 14'S), utilizing 

tertiary treated sewage effluent as feedstock, and one at Ashburton (Lat 43° 

54'S), using effluent from a major meat processing plant. 

The same cultivar (EILl) was used at both sites. EILJ is a dioecious fibre type 

cultivar developed in Queensland by EcoFibre Industries Ltd 

Three sowings were made at each site. Each sowing at each site comprised three 

lOm by 2m plots, at three plant densities. The 3 sowings were direct drilled into 

unprepared soi l (Feilding) and into cultivated soil (Ashburton) at monthly 

intervals. Random l.Om2 destructive harvests were taken monthly from each 10 

x 2m plot, starting one month after sowing. These samples were weighed, leaf 

and stem weights recorded, sub samples dried at 80C for 48 hours and weighed. 

The sub samples were fine ground, and analysed for total nitrogen (Keldahl) and 

total phosphorus (Phosphomolybdate). Plant numbers for each plot and harvest 

were also recorded, thus all data could be analysed on per plant, and on per m2 

basis. 

3.2.2. Problems and Limitations 

3.2.2.1.Herbicide damage 

Several plots were inadvertently damaged by herbicidal spray during the 

Feilding trial. This limited the usefulness of data from this site for density 

analysis, and exacerbated the data weighting limitations inherent to the 

experimental design. 

3.2.2.2.Weighting limitations 

Successive harvests on successive sowings imply fewer harvest samples 

available from later sowings. Thus 5 samples were available from sowing 1, 

4 from sowing 2 and 3 from sowing 3. In fact losses due to herbicidal spray 

resulted in only 1 sample being available for harvest 5. This affected the 

statistical validity of averaged results. 

26 



3.2.3. Analyt ical Methods 

Statistical analyses were performed to the same protocol and with the same 

software as for experiment 1. As both exponential growth and asymptotic limits 

to growth were observed, and the effects are independent, both transforms 

(natural logarithm and reciprocal) were applied where applicable (section 3.2.4). 

Jn order to test for diffe rences between sites, sowing dates and harvests, sample 

data were converted to per plant and natural log ( In) transforms performed. 

However as interest in thi s project is in nutrie nt uptake on an area basis, all 

results were only finally considered on an area basis. 

Given that sowings were performed at d ifferent plant densities, results on an 

area basis would be value less without compensatio n for this variatio n. For this 

reason, and because some data was lost, results were 'standardized ' to 30 

plants. m-
2

. As dry matter yield is fundamentally asymptotic due to shad ing and 

self thinning, rec iprocals were taken of total (In) dry m atter per square metre, 

and plotted aga inst plant de nsity. The resulting linear regression formulae were 

used to estimate total dry matter yield at 30 plant s.m-2 • B y this means, va lues for 

each harvest, sowing and site could be regressed against harvest date. Due to 

variations and ano malies in sowing dates between sites, Ashburton sowing date 

data was adjusted by 1 mo nth to compe nsate, allowing all avail able data to be 

used in a single calculation. 

3.2.4. Allo metric Relationships 

Allo metric analysis between leaf and ste m was based o n the re lationship of (ln) 

leaf a nd (In) stem against total weight. The resulting regression equations were 

used to calculate leaf and stem weights for each ha rvest , sowing and site. 

3.2.5. Nutrient uptake 

Total (In) nitrogen and (In) phosphorus data, for both leaf and ste m, were 

regressed against (ln) leaf and (ln) ste m biomass data respective ly to deri ve an 

expected uptake nutrie nt value. From these regressio ns, and the bio mass data 

o utlined above, uptake values for both nitrogen and phosphorus were derived, on 

a per harvest basis, irrespecti ve of sowing date or site. 

27 



3.3. Results and discussion 

The following data were recorded for each site, plot and harvest: 

• Leaf dry weight (g) 

• Stem dry weight (g) 

• Plant numbers (count) 

• Total Nitrogen for leaf and stem(%) 

• Total Phosphorus for leaf and stem (%) 

3.3.1. A1lometric Relationships 

Leaf weights (natural log) were plotted against stem (natural log) weights, as 

shown in Figure 14, on an area (square metre) basis. 

8.0 

-')'E 6.o 

en 
c 
(/) 

::i 4.0 

E 
0 
iii 
1a 

Cl) 2.0 
...I 

• 

• 

y = 0.618546x + 1.826974 

R
2 

= 0.833569 

0.0 +---------~---------~ 
-2.0 0.0 2.0 4.0 6.0 8.0 

Stem Biomass ln(g.m-2
) 

• Series1 

- Linear (Series 1) 

Figure 14 Leaf versus stem (In) all ometric relationship, on an area basis 

This figure clearly shows the relationship between leaf and stem during plant 

growth, giving ri se to the predictive model: 

(ln)leaf = 0.6185 x (ln)stem + 1.8270 

Further, the slope (less than 1.0) illustrates the trend, throughout growth, of an 

allometric change from dominant leaf production to dominant stem production. 

This is more clearly demo nstrated when stem and leaf (ln) weights are plotted 

against total (In) weight, as shown in Figure 15 (per plant basis), and Figure 16 

( area basis) . 

28 



6 

- 4 

0.. 
en 
c -
.E 

2 

C'l 
·a; 
~ - 0 
10 
Cl> 

:::: 
E 
Cl> -2 en 

.,, 
·2 

Leaf= 0.834056x • 0.427748 

A
2 

= 0.904604 

• 
Stem= 1.180607x • 1.141588 

A
2 

= 0.965912 

Total weight Jn(g .plf1
) 

+ stem 

D leal 
- Linear (stem) 

- - Linea r (leaf) 

Figure 15 Stem and leaf (In) weights v total (In) plant weight 

The equivalent c hart, Figure 16 below, illustrates a similar trend on an area 

basis. 

8.0 

~ 

"! 
E 6 .0 

Ol 
c 
.E 4.0 
Cl 
"iii 
~ 
co 

Cl> 
2.0 

.J 

--E 
Cl> - 00 

(/) 

·20 
0.0 

Leaf = O. 797861 x + 0.350278 

R2 = 0.918133 _.r;j" 

--0 

....al1 19 
.... 'a 

2.0 

Srem = 1.215718x • 1.963915 
R2 = 0.978407 

4.0 

Total Weight ln(g.m.2) 

6.0 80 

Figure 16 Stem and leaf (In) weights v total (In) weight, on an area basis 

• s1em 
Oleaf 

The inherent concern with an area mode of presentation, internal data 

dependency, is evidenced by the artificia lly higher R2 values disp layed in Figure 

16 than in Figure 15. The potential problem is, however, negated by the similar 

relationship demonstrated in Figure 14. The Figure 16 results are more useful as 

predictive models. 

It wi ll be noted that these relationships are true across both sites, and all sowings 

and harvests. Thus, the pattern of growth of hemp is independent of these 

parameters, and can be estimated fro m the following equations: 

(log)leaf = 0.834056 x (log total weight) - 0.427748 

(log)stem = 1.180607 x (log total weight) - J.1 4 1588 

29 



3.3.2. Nitrogen and phosphorus uptake. 

Nitrogen and phosphorus uptake was calculated as discussed. 

3.3.2.1.Stem and leaf nitrogen uptake 

Both stem and leaf N uptake were compared with total N uptake (natural 

log), on an area basis ln(g. m-2
) as shown in Figure 17. 

':' 
E 
~ 2 
C 

Cl) 
.le 
ea 
- 0 Q. 
:::> 
z .... 
~ 2 

...I 
"C 
C 
ea 
E 
Cl) 

en 

Leaf= 0.957486x - 0.193791 

R2 = 0.975897 

Stem = 1.162649x - 1.686549 

R2 = 0.945368 

-•+----- -~---- -------- ----~ ., 
Total N Uptake ln(g.m-2

) 

• stem 
o leaf 

Figure 17 Relationship between stem and leafN, and total N (natural log) 

This relationship shows the overall higher level of N in leaves than in stem, and 

the differing slopes apparent as the plant grows (ages). While the regression 

lines do not cross in the range under study, as they do for the biomass data 

shown above, a similar trend is apparent in both. 

3.3.2.2.Stem and leaf phosphorus uptake 

Correspondingly a similar pattern emerges when the phosphorus data is plotted 

(Figure 18), but at a significantly lower level. 

-N 

·e 
Cl O 

c 
Cl) 

.le 
ea a. -2 

:::> 
ll. .... 
~ -4 

...I 
"C 
C 
ea -s 
E 
Cl) 

en 

Leaf= 0.9496x - 0.3834 

R2 = 0.9741 

Stem = 1.092944x - 1.320372 

R2 = 0.888580 

D .,. 

• 

-8 +------~----------~~----~ 
-6 -4 -2 

Total P Uptake ln(g.m-2) 

• s1em P 
O sleaf P 

Figure 18 Relationship between stem and leaf P, and total P (natural log) 

30 



It can be seen that P follows a similar relationship to that of N, and once 

again the lines converge, but do not cross in the range of our data. 

Comparison of Figures 17 (N) and 18 (P) show the differences in uptake 

levels for the two nutrients. 

Stem and leaf uptake rates for both nitrogen and phosphorus were calculated 

(on an area basis) in Figures 19 and 20. 

3.3.2.3.N and P uptake for stem 

-"' 
·~ 2 

Cl 
c 
-; 0 
~ 
(U 

a. 
::, ·2 

E 
Cl) 

en .4 

ll. 
"O 
C 
(U ·6 

z 

• 

y = 0.756148x · 3.492626 
R2 = 0.94921 4 

y = 0.812229, · 5.958469 

R2 = 0.91 1966 

-8 +----~------~-----~ 
·2 

Stem Weight ln(g.m·2) 

• Ni1rogen 

D Phosphorus 

- Linear (Nitrogen) 

- - linear (Phosphorus) 

Figure 19 Stem N and P uptake rates v stem weight , on an area basis 

For stem uptake, it is clear that N and P are taken up at similar rates for a 

given size of plant , as evidenced by the closely parallel regression trend 

lines. It will also be noted that the intercept for each element corresponds to 

the (natural log) uptake of that element for l g of stem dry matter (natural 

log O = 1.0, so the intercept represents this value). Thus 

Stem nitrogen uptake 

Stem phosphorus uptake 

= 0.0304g. dry matter stem 

= 0.0026g. dry matter stem 

31 



3.3.2.4.N and P uptake for leaf 

Uptake values for N and P are presented in Figure 20. 

-"' ·e 2 

~ 
.E 
41 

.!I<: 0 
ea 
Q. 
:::, 
c. 
"C -2 
C: 
ea 
z 
m -4 
41 
..J 

Nitrogen= 0.929891x - 3. 102046 

R2 = 0.982919 

• 
0 

~~~'ff"d 

_,,cr oB ,,, 
/ 

0 / 
q.. '-"o Phosphorus= 0.990068x - 5.757832 

';, / R2 = 0.903142 
/ 0 

~ 

-6 +------------------~ 
0 

Leaf Weight ln(g.m"2) 

Figure 20 Leaf N and P uptake v Leaf wt , on an area basis 

• Nitrogen 

O Phosphorus 
- Unear (Nitrogen) 

- - linear Phos horus 

Once again, the lines are closely parallel , indicating that hemp takes up N 

and P in leaves at virtually the same rate irrespective of plant size. Again , the 

intercepts indicate a model for N and P uptake, as discussed above. 

Leaf nitrogen uptake 

Leaf phosphorus uptake 

= 0.0450g. dry matter leaf 

= 0.0032g. dry matter leaf 

3.3.3. Harvest date effects 

The effect of harvest date (in months) on total N and P uptake is presented in 

Figure 21. 

-"' ·e 

20 

16 

~ 12 

41 
.!I<: 
ea 
Q. 
:::, 8 

c. 

z 

• 

• Nitrogen 

0 Phosphorus 

• ..Q----"'l:J ____ o ___ J:l. __ _ 

Harvest Date (months) 

Figure 21 Effect of harvest date on nutri ent uptake, across si tes and sowing dates 

This figure shows the expected increase in total uptake with later harvests and 

larger plants. A quadratic line has been added for clarity. Since both 

32 



nitrogen (%) and phosphorus (%) reduce for both stem and leaf over the growth 

cycle of the plant, Figure 21 indicates that the rapidly increasing total biomass 

more than compensates. It is thus clear that allowing the plants to reach maturity 

will optimize nutrient removal from the growth medium, to the asymptotic 

maximum illustrated and discussed above. 

Total nitrogen and phosphorus uptake estimates for each of five successive 

harvests were calculated as outlined above, with results shown in Table 10. It 

should be noted that these estimates were based on combined results of three 

successive sowings. Thus the harvest data contained progressively fewer data 

points, and the last (5 th
) only 1. The later harvests should therefore be considered 

indicative only, due to the inherent weighting constraints outlined in section 

3.2.2.2. 

Table 10 Nutrient uptake (g.m-2)for 30 plant.m·2 

Nutrient Uptake (g.m-2) for 30 
Plant.m·2 

Nitrogen Phosphorus 

Harvest Leaf Stem Leaf Stem 

1 

2 

3 

4 

5 

1.24 

5.05 

6.19 

9.27 

13.33 

0.28 

1.60 

2.06 

3.39 

5.32 

0.11 

0.48 

0.60 

0.92 

1.35 

0.03 

0.18 

0.24 

0.41 

0.66 

This table, subject to the constraints indicated, shows the steadily increasing 

uptake values for both N and P. Thus the rapidly increasing biomass in both leaf 

and stem more than compensate for the observed reduction in both N and P 

concentrations over the life of the plant. 

3.4. Conclusions 

3.4.1. Leaf/stem dry weight 

The dry weight allometric relationship of stem and leaves (on a per plant basis) 

in industrial hemp can be described by the formulae : 

Leaf weight = 0.834056(total weight) - 0.427748 

Stem weight= l.180607(total weight) - 1.141588 

33 



This is true, with high probability, irrespective of latitude ( 41 °S to 43 °S), sowing 

date (October to December) or harvest date (monthly from one month after 

sowing). The (natural log) leaf dry weights decrease with respect to (natural log) 

stem dry weight as total (natural log) above ground biomass increases. Although 

this thesis is concerned with hemp seed oil ( experiment 1) and nutrient uptake 

(experiment 2), this allometric relationship, in relation to density effects found in 

experiment 1, could be significant for hemp fibre production. 

When leaf and stem weights are considered on an area basis, the equivalent 

formulae are: 

(ln)leaf = 0.79786l(]n total weight)+ 0.350278 

(ln)stem = l.215718(]n total weight) - 1.963915 

These results explain a high level of total variability, irrespective of latitude, 

sowing date or harvest date. When considered in conjunction with density 

effects, as discussed in section 2, this could be relevant to fibre production. 

3.4.2. Nitrogen and phosphorus uptake. 

Leaf and stem uptake values for nitrogen and phosphorus were compared against 

total plant N and P respectively. These showed a similar trend to the weight 

figures reported above, but with less pronounced tendency to converge at higher 

plant weight. 

Stem nitrogen and phosphorus uptake values were compared to total stem 

weight, and equivalent results calculated for leaves. The results showed the 

following relationships: 

Stem N uptake= (exp) -3.492626 = 0.0304 g.m-2 D .M. stem 

Stem P uptake= (exp) -5.958469 = 0.0026 g.m-2 D.M. stem 

LeafN uptake= (exp)-3.102046 = 0.0450 g.m-2 D.M. leaf 

Leaf P = (exp) -5 .757832 = 0.0032 g.m·2 D.M leaf 

Plotting values for these equations as leaf/stem ratios against total weight gives a 

consistent pattern for the two elements, with nitrogen being consistently some 3-

5 times the uptake rate of phosphorus. The relationships show initial high uptake 

rates, settling to a stable, lower value as the plant matures. The values for leaf 

nitrogen and phosphorus compare favorably with the work of Ketterings et. al. 

34 



(2006) for grass of 0.0200 g(N).g(DM)-1 and 0.0029 g(P).g(DM)-1 respectively. 

Thus nitrogen uptake by hemp leaf would appear to be more than twice that for 

grass. Even for stem (not directly comparable to grass) uptake is some 50% 

higher. However the equivalent rates for phosphorus are similar to those for 

grass. Hemp would thus appear to be a favorable alternative to grass in terms of 

N and P uptake, while still producing a potentially valuable product as an output. 

Sites differences could well be due to differences between nutrient feedstock and 

local soil/climate effects, but it can be stated that optimum removal is achieved 

by sowing as early as possible (October or earlier), and harvesting as late as 

possible (5 months or longer). The high biomass accumulation in hemp more 

than compensates for the high concentration of both Nitrogen and Phosphorus in 

young plants. 

4. General Conclusions 

The two experiments in this project, while apparently similar, were in fact designed for 

complementary purposes. Experiment 1 focused on seed production, and the growth and 

density parameters involved; experiment 2 primarily explored nitrogen and phosphorus 

uptake from effluent irrigation. Experiment l , with two varieties, provided insight into 

factors of height, growing time, harvest index, and plant density effects, and gave 

indications of latitude effects on these factors. More importantly, the two varieties 

(cv Anka and cv Finola) gave information on the interactions between these effects, 

particularly varietal differences (sections 2.3, 3.3). This was possible, notwithstanding, 

problems due to poor seed quality, predation, weed infestation and wind damage, all of 

which contributed to low yield and density, and missing plots. 

Results were achieved despite these problems by statistical modeling, based on the 

premise that plants grow exponentially. Transforming data to natural logarithms will 

thus produce a straight line, from which a linear regression yields intercept and slope for 

interpretation, with an estimate of error for statistical analysis. Similarly, an asymptotic 

limitation to growth such as competition for light, nutrient or other resource can be 

modeled by a reciprocal transform. Again, the resulting linear regression allows 

interpretation of intercept and slope of the line. As these two effects are independent, 

both can be used as applicable, as was done to delineate allometric relationships (section 

3.3.1), where both exponential growth and internal shading occurred. 

35 



By these means it is concluded that the harvest index of Anka falls dramatically with 

later sowing, where the Finola harvest index does not. This different result between 

cultivars is reflected in biomass, growing time, seed yield and height, and indicates the 

strong possibility that Finola is a day-length neutral plant. Experiment 2, with only one 

cultivar (cv EILI), could not provide this level of insight, but chemical analysis showed 

that greater quantities of nitrogen and equivalent quantities of phosphorus, can be 

extracted compared to grass. It was shown that the exponentially increasing biomass 

yield more than compensates for the reduction in %N and %P throughout the growth 

cycle of the plant. Successive harvests throughout the plant growth cycle, provided 

insight into allometric changes in chemical composition, complementary to the single 

harvest data available from experiment 1. Conversely, plant density data from 

experiment 1 was used to delineate density effects, where data from experiment 2, with 

several plots damaged by herbicidal spray, could not. 

5. Future Work 

Both these experiments were intended as preliminary trials, to define parameters for 

more exhaustive future investigations, and to identify useful areas of further study. 

The main points identified include: 

• Latitude effects. Some evidence was obtained indicating latitude effects in plant 

height, growing time and harvest index. This was inconclusive, primarily due to loss 

of data at our most southerly site (experiment 1) and the more northerly site 

(experiment 2). Further trials are required, at more sites, to confirm any such effect. 

• Sowing date. Strong evidence was found indicating potential improvements in 

productivity and yield from an extended growing period, possibly leading to double 

cropping potential using different varieties. Effects of frost damage in spring , 

vernalization and floration inhibiting growth were not investigated however, and 

further trials are required to establish effective sowing date limits. 

• Plant density. Varietal differences and asymptotic limits were established, but it is 

clear that these limits were not approached at the densities achieved in our trial. 

Further trials at higher densities are required to refine yield limits. 

• Varieties. Very clear varietal differences in yield have been established with two 

varieties in experiment 1, particularly with respect to sowing date, and allometric 

36 



relationships with one variety in experiment 2. A trial with a wider range of varieties 

is required, to identify and quantify the extent of these points of difference. 

• Nutrient uptake. Our results confirm that considerable quantities of nitrogen and 

phosphorus can be removed from nutrient rich effluent. No information on varietal 

differences was gained however, and further trials with different varieties are clearly 

warranted. 

A tria l or series of trials can be envisaged involving a wide range of cultivars, latitudes, 

plant densities, sowing dates and effluent levels. Our data confirms that such trial(s) are 

warranted. This was one of our main objectives, with the expectation throughout our 

trial period of a reali st ic reso lution of New Zealand licensing regulations (at that time 

under consideration) for growing he mp. In fact the regulations, after a four year delay, 

were finally promulgated in July 2006 (Misuse of Drugs (Industrial Hemp) Regulations 

2006), and came into e ffect August that year. The main thrust of the new regulations is 

to severe ly inhibit any further trials in this area. The new licensing require ments invo lve 

$550 per site (plus GST) , an add itio nal $ 100 (f' the crop is to be grown for research or 

breeding purposes (my italics). A further clause invo lves additio nal costs for testing of 

each plot for THC levels, and a requirement that the crop be not sown w ithin 5km of a 

res ide ntial zone. Thus to find suit able sites, and conduct tria ls as indicated by our 

results, will require considerable additio nal funding, making any such research 

prohibitively expensive without private funding. 

37 



References 

Amaducci, S., Errani, M., Venturi, G. (2002) . Response of hemp to plant population and 

nitrogen fertilisation. Italian Journal of Agronomy 6 (2) : 103-111 

Bennett, S.J., Snell, R. , Wright, D. (2006) . Effect of variety, seed rate and time of 

cutting on fibre yield of dew-retted hemp. Industrial Crops and Products, 24 

(I): 79-86. 

Charkowski, E., (1999). Hemp "Eats" Chernobyl Waste, Offers Hope for Hanford 

retrieved on 15/01/2006 from 

http://hemp.net/news/990 l /06/hemp _ eats_chernobyl_ waste. html. 

Callaway, J.C. , Laakkonen, T .T. (1996). Cultivation of cannabis oil seed varieties in 

Finland. J Int Hemp Assoc 1996:3(1):32-34 

Callaway, J.C. , (2004). Hempseed as a nutritional resource: An overview. Euphytica 

140 (1-2): 65-72. 

Callaway, J., Schwab, U., Harvima, I., Halonen, P., Mykkanen, 0., Hyvonen., P., 

Jarvinen, T. (2005). Efficacy of Dietary Hempseed Oil in Patients with Atopic 

Dermatitis. Journal of Dermatological Treatment. 16:87-94. 

Cook, J., (1779) Narrative of the Voyages Round the World; Chapter II: Narrative of 

Captain Cook's First Voyage Round the World. 

Drew, T. J. , Flewelling, J. W. , (1977). Some Recent Japanese Theories of Yield-Density 

Relationships and their Application to Monterey Pine Plantations. Forest 

Science, Volume 23, Number 4, 1 pp. 517-534(18). 

Fosu, M., (2003) Nitrogen accumulation and release by sunn hemp, calopo, mucuna and 

devil bean in semi-arid Ghana. Agricultural and Food Science Journal of Ghana 

2: 141-153 

Gulati, D., Sain, M. (2006). Fungal-modification of natural fibers: A novel method of 

treating natural fibers for composite reinforcement. Journal of Polymers and the 

Environment 14 (4): 347-352. 

Herodotus: The History (Chap. 73-75); Univ. Chicago Press; David Greene, transl.; 

ISBN 0-226-32770-I. 

Ketterings, Q. M., Godwin, G., Kiker, T. F., Barney, P., Hunter, M., Cherney, J. H., 

Beer, S. (2006). Nitrogen, phosphorus, potassium, magnesium and calcium 

removal by brown midrib sorghum sudangrass in the Northeastern USA. Journal 

of Agronomy and Crop Science 192 (6): 408-416. 

38 



Khan, A.G. (2005). Role of soil microbes in the rhizospheres of plants growing on trace 

metal contaminated soils in phytoremediation. Journal of Trace Elements in 

Medicine and Biology 18 355-364 

Lehoczky, E., Nemeth, T., Reisinger, P. , Radimszky, L., Komives, T. (2006). Effect of 

weediness on the water content of the so il: A field study. Communications in 

Soil Science and Plant Analysis 37 ( 15-20): 2673-2678. 

Leizer, C., Ribnicky, D., Poulev, A, Dushenkov, S., Raskin, I. (2000). The 

Composition of Hemp Seed Oil and Its Potential as an Important Source of 

Nutrition. Journal of Nutraceuticals, Functional & Medical Foods Vol. 2(4) 

Lisson, S.N., Mendham, N.J., Carberry, P.S. (2000). Development of a hemp (Cannabis 

sativa L.) simulation model l. General introduction and the effect of temperature 

on the pre-emergent development of hemp. Australian Journal of Experimental 

Agriculture 40 (3): 405-411 

Lisson, S.N., Mendham, N.J., Carberry, P.S. (2000). Development of a hemp (Cannabis 

sativa L.) s imulation model 2. The flowering response of two hemp cu lti vars to 

photoperiod. Australian Journal of Experimental Agriculture 40 ( 3 ): 413-417 

Lisson, S.N., Mendham, N.J., Carberry, P.S. (2000). Development of a hemp (Cannabis 

sativa L.) simulation model 3. The effect of plant density on leaf appearance, 

expansion and senescence. Australian Journal of Experimental Agriculture 40 

(3): 419-423 

Lisson, S.N., Mendham, N.J., Carberry; P.S. (2000). Development of a hemp (Cannabis 

sativa L.) simulation model 4. Model description and validation. Australian 

Journal of Experimental Agriculture 40 (3): 425-432 

Liste, H.H., Prutz, I. (2006). Plant performance, dioxygenase-express ing rhizosphere 

bacteria, and biodegradation of weathered hydrocarbons in contaminated soil. 

Chemosphere 62 (9): 1411-1420 

Lu, X., Clarke, R.C. (1995). The cult ivation and use of hemp 

(Cannabis saliva L.) m ancient China. Journal of the international Hemp 

Association Vol. 2 No. 1. 

McIntosh, D. J., (1998). The 5 minute guide to industrial hemp in New Zealand. New 

Zealand Hemp Hemp association: pp62. 

McColl, R.H.S. , Gibson, AR., (1979). Downslope movement of nutrients m hill 

pasture, Taita, New Zealand. 2. Effects of season, sheep grazing, and fert iliser. 

New Zealand Journal of Agricultural Research 22 (]): 151-161. 

39 



McPartland, J. M., Cutler, S., McIntosh, D. J. (2004). Hemp production in Aotearoa. 

Journal of Industrial Hemp 9(1):105-115 

Misuse of Drugs (Industrial Hemp) Regulations 2006 and the Misuse of Drugs 

Amendment Regulations 2006. 

Moliterni, Y.M.C., Cattivelli, L., Ranalli, P., Mandolino, G. , (2004). The sexual 

differentiation of Cannabis sativa L.: A morphological and molecular study 

Euphytica. 140 (J-2): 95-106 

Rollin, C., (1731). Rollin's Ancient History: Egypt. Of The Egyptian Soldiers and War. 

Sawabe, A., Takeda, R., Komemushi, S. , (2006). Phytoremediation: searching for plants 

with high environmental purification capacity. Memoirs of the Faculty of 

Agriculture of Kinki University 39 : 1-8 

Struik, P.C. , Amaducci, S., Bullard, M.J ., Stutterheim, N.C., Venturi, G., Cromack, 

H.T.H. (2000). Agronomy of fibre hemp (Cannabis sativa L.) in Europe. 

Industrial Crops and Products 11 (2-3): 107-118 

Tang, C.H., Ten, Z. , Wang, X.S ., Yang, X.Q. (2006) . Physicochemical and functional 

properties of hemp (Cannabis saliva L.) protein isolate. Journal Of Agricultural 

And Food Chemistry 54 (23): 8945-8950. 

Tlustos, P., Szakova, J., Hruby, J., Hartman, I., Najmanova, J., Nedelnik, J., Pavlikova, 

D., Batysta, M. (2006). Removal of As, Cd, Pb, and Zn from contaminated soil 

by high biomass producing plants. Plant, Soil and Environment 52 (9) : 413-423 

Wambua, P ., Yangrimde, B. , Lomov, S., Yerpoest, I. (2007). The response of natural 

fibre composites to ballistic impact by fragment simulating projectiles. 

Composite Structures 77 (2): 232-240. 

Werf, H.M.G. van der, Wijlhuizen, M., Schutter, J.A.A. de. (1995) Plant density and 

self-thinning affect yield and quality of fibre hemp (Cannabis sativa L.). Field 

Crops Research 40 (3): 153-164. 

Yang, L., Chang, H.Ting., Huang, M.N.L. (2001). Nutrient removal in gravel- and soil­

based wetland microcosms with and without vegetation. Ecological Engineering 

18 (1): 91-105 

40 



Appendix A. 

Final 5% significance model for height at harvest, from Rangiriri and Kerikeri (sowings 
2, 3 and 4). 

Dependent Variable: height 

Source 

Mode l 
Error 

Corrected Total 

A-Square 
D.765637 

Source 

site 
variety 
sowing*variety 

Le vel of 

site 

Kerikeri 
Rangiriri 

Le vel of 

variety 

anka 
finola 

.... 

Level of Level 

DF 

6 

64 

70 

Rangi /Keri height 

height 

Sum of 
Squares 

6.41447698 
1.96348077 

8.37795775 

Mean Square F Value 

1 . 06907950 34.85 
0 . 03067939 

Coeff Var 
23.84665 

Root MSE 
0. 175155 

height Mean 
0.734507 

DF Mean Square F Value 

Pr > F 

< .0001 

Pr > F Type I SS 

0.46018791 
5.46667081 
0.48761826 

0.46018791 
5.46667081 
0.12190457 

15.00 0.0003 
1 

4 

N 

35 
36 

N 

36 
35 

of 

Rangi / Keri height 

178 .19 
3.97 

- - - - - - - - - - - -height- - - - - - - - - - -

Mean 

0.65285714 
0.81388889 

Std Dev 

0.32402407 
0.35246299 

- - - - - - - - - - - - height- - - - - - - - - - -

Mean 

1. 00694444 
0.45428571 

Std Dev \ 

0.25971214 
0.13249386 

- - - - - - - - - - - - height- - - - - - - - - - -

< .0001 
0.0061 

sowing variety N Mean Std Dev 

DEC2005 anka 12 1 . 07500000 0 .19598237 
DEC2005 finola 12 0.40416667 0.08106769 
JAN2006 anka 12 0.85000000 0.18090681 
JAN2006 finola 11 0.48636364 0 . 15666989 
NOV2005 anka 12 1 .09583333 0 . 32083579 
NOV2005 finola 12 0.47500000 0.14538351 

Appendix B. 

Model derived for reciprocal seed yield (intercept) across sites, sowings and varieties; 
Rangi /Keri seed (ln) yield 

Dependent Variable: intercpt intercpt 

Source 

Model 
Error 

Corrected Total 

A-Square 
0.379835 

Source 

sow 

DF 

3 

44 

47 

Sum of 
Squares 

13.44751580 
21.95604562 

35.40356142 

Coeff Var 
-31 .97603 

Root MSE 
0.706400 

DF Type I SS 

2 3.81596311 

41 

Mean Square F Value Pr > F 

4.48250527 8.98 < .0001 
0.49900104 

intercpt Mean 
-2.209155 

Mean Square F Value Pr > F 

1.90798155 3.82 0.0294 



variety 9.63155270 9.63155270 19.30 <.0001 

Rangi/Keri seed (ln) yield 

Level of - - - - - - - - - - -intercpt- - - - - - - - - -

sow N Mean Std Dev 

2 20 -1 . 93967701 0.79772658 
3 13 -2.16786360 0 .76243213 
4 15 -2.60424407 0.94570378 

Level of - - - - - - - - - - - intercpt- - - - - - - - - -

variety N Mean Std Dev 

anka 23 -2.67713262 0.80182449 
finola 25 -1.77861510 0.69486452 

Appendix C. 

Analysis of variance and means for plant density effects at 30 plants.m2
. 

Rangi / Keri density 30 

Dependent Variable: yield30 yield30 

Source 

Model 
Error 

Corrected Total 

R-Square 

0.930889 

Source 

sow 
variety 
sow*varie ty 

Level 

sow 

1 
2 
3 
4 

Level 

of 

of 

Sum of 
DF Squares Mean Square F Value Pr > F 

7 431391.3654 61627.3379 11 .55 0.0042 
6 32027.3824 5337.8971 

13 463418.7478 

Coeff Var 

33.46921 

Root MS E 

73. 06091 

yield30 Mean 

218.2929 

DF 

3 

1 

3 

Type I SS 

163129.1726 
148400.1373 
119862.0555 

Mean Square F Value Pr > F 

54376.3909 10.19 0.0091 
148400.1373 27 . 80 0.0019 
39954.0185 7.48 0.0188 

Rangi / Keri density 30 

- - - - - - - - - - -yield30- - - - - - - - - - -

N Mean Std Dev 

2 461 . 569267 461 .205799 
4 232 . 033669 131. 334485 
4 180.919042 91 .937951 
4 120.287881 59.089717 

- - - - - - - - - - -yield30- - - - - - - - - - -

variety N Mean Std Dev 

anka 7 321.249271 218.223995 
finola 7 115.336572 69.866944 

Level of Level of - - - - - - - - - - - y ie ld30 - - - - - - - - - - -

sow variety N Mean Std Dev 

anka 787.691015 
1 f inola 1 135.447519 
2 anka 2 319.073215 65 .149952 
2 f inola 2 144.994124 131.141889 
3 anka 2 246.877333 38.261891 
3 finola 2 114.960750 80.571728 
4 anka 2 164.576391 47.846197 
4 finola 2 75.999371 18.429239 

42 



Appendix D. 

Analysis of variance and means for plant density effects at 60 plants.m2
• 

Dependent Variable: yield60 

Rangi / Keri density 60 

yield60 

Source 

Model 
Error 

OF 
Sum of 

Squares Mean Square F Value Pr > F 

47.1548318 
100. 3381214 

47.1548318 5.64 0.0351 

Corrected Total 

A-Square 

0.319709 

12 

13 147.4929533 

Coeff var 

65.07221 

OF 

Root MSE 

2.891628 

8.3615101 

yield60 Mean 

4.443721 

F Value Pr > F Source 

variety 

Type I SS 

47.15483185 

Mean Square 

47.15483185 5.64 0.0351 

Appendix E. 

Level of 

variety 

anka 
finola 

Rangi / Keri density 60 

- - - - - - - - - - -yield60- - - - - - - - - - -

N 

7 

7 

Mean 

6.27898740 
2.60845486 

Std Dev 

3.84806644 
1 . 38398154 

Analysis of variance and means for plant density effects at 90 plants.m2
. 

Rangi / Keri / Massey density 90 

Dependent Variable: yield90 yield90 

Source 

Model 
Error 

Corrected Total 

A-Square 

0.310100 

Sum of 
OF Squares Mean Square F Value 

174733.3619 174733.3619 5.39 
12 388740.2491 32395.0208 

13 563473 . 6110 

Coeff Var 

61 . 59577 

Root MSE 

179.9862 

yield90 Mean 

292 .2054 

OF Mean Square F Value 

Pr > F 

0.0386 

Pr > F Source 

variety 

Type I SS 

174733.3619 174733.3619 5.39 0.0386 

Rangi / Keri /Massey density 90 

Appendix F. 

Level of 

variety 

anka 
finola 

N 

7 

7 

- - - - - - - - - - -yield90- - - - - - - - - - -

Mean 

403.923595 
180.487212 

Std Dev 

238.484289 
88.967890 

Analysis of variance and means for growth time (in days) to harvest 

Rangi/Keri Days to harvest no sowing 1 

Dependent Variable: growingdays 

Source OF 

growingdays 

Sum of 
Squares 

43 

Mean Square F Value Pr > F 



Model 5 10618 . 35382 2123.67076 60.90 <.0001 

Error 65 2266.57576 34 . 87040 

Cor r ected Total 70 12884.92958 

R-Squa re Coeff Var Root MSE growingdays Mean 

0.824091 8.714680 5.905116 67 . 76056 

Source OF Type I SS Mean Square F Value Pr> F 

site 5254 . 964498 5254.964498 150 . 70 <. 0001 
sow 2 3538.063119 1769 .031559 50 . 73 < , 0001 
site•sow 2 1825.326203 912 . 663102 26 . 17 <.0001 

Rangi / Keri Days to harvest no sowing 1 

Level of - -- - ----- growingd ays- - -- --- - -

site N Mean Std Dev 

Kerikeri 35 76.4857143 11.9148238 

Rangiriri 36 59.2777778 8.9494169 

Level of - --------growingdays--- ----- -

sow N Mean Std Dev 

2 24 70 . 2500000 15.1147783 
3 24 74.9166667 11 . 9379434 
4 23 57.6956522 5.4140791 

Leve l of Level of ----- -- --growingd ays- ----- ---

site sow N Mean Std Dev 

Kerikeri 2 12 85.0000000 0.0000000 
Kerikeri 3 12 83.0000000 0.0000000 
Kerikeri 4 11 60.0909091 7.0064905 
Rangiriri 2 12 55 . 5000000 1.7320508 
Rangiriri 3 12 66 .8333333 12.4669259 
Rangiriri 4 12 55 . 5000000 1. 7320508 

Appendix G. 

Leaf weight and total above grou nd biomass data analysis of variance (5%) model. 

Dependent Var iable: ratio 

Source 

Model 
Error 

Corrected Total 

R- Squa re 

0.495681 

Source 

s ite 
sow 
site•sow 
site•variety 

Rangi / Keri /Massey leaf ratio 

ratio 

Sum of 
OF Squares Mean Square F 

8 1.29817176 0 . 16227147 
57 1.32079277 0 . 02317180 

65 2.61896453 

Coeff Var Root MSE ratio Mean 

25.29028 0 .152223 0.601903 

OF Type I SS Mean Square F 

0.53548550 0 . 53548550 
3 0 . 26873919 0.08957973 
2 0.27965420 0.13982710 
2 0.21429286 0. 10714643 

Rangi / Keri /Massey leaf ratio 

44 

Val ue Pr > F 

7.00 < .0001 

Va lue Pr > F 

23 . 11 < .0001 
3.87 0.0138 
6.03 0.0042 
4.62 0.0138 



Level of - - - -- - - - - - - -ratio - - -- - - -- - -- -

site N Mean Std Dev 

Kerikeri 25 0.48655105 0. 18257610 
Rangiriri 41 0.67223908 0.17912721 

Level of - ----- ----- -ratio- -- ----- - -- -

sow N Mean Std Dev 

12 0.71362398 0 . 13255284 
2 19 0.65281308 0.19331291 
3 19 0.58107766 0.21510936 
4 16 0.48238542 0 .17968898 

Level of Level of ----- -------ratio---------- - -

site SOW N Mean Std Dev 

Kerikeri 2 9 0.53452132 0.22137676 
Ke r ikeri 3 8 0.41127830 0 . 12700497 
Kerikeri 4 8 0.50785726 0. 17943313 
Rangiriri 1 12 0 . 71362398 0.13255284 
Rangiriri 2 10 0.75927566 0.06775549 
Rangiriri 3 11 0.70456810 0.17932692 
Rangiriri 4 8 0.45691359 0.1 8843990 

Level of Level of ------------ratio----- -------

site variety N Mean Std Dev 

Kerikeri anka 15 0.55138252 0.16591230 
Kerikeri finola 10 0.38930386 0.16899158 
Rangiriri anka 20 0 . 63489008 0.1 8708716 
Rangiriri finola 21 0 . 70780955 0 .1 6792295 

45