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. Shortjaw kokopu ( Galaxias postvectisClarke) distribution, habitat selection and seasonal activity in the northeastern Tararua Ranges A thesis presented in partial fulfillment of the requirements for the degree of a Masters of Science in Ecology At Massey University, Palmerston North New Zealand. Scott Bowie 2002 Errata Page 3 (and other occurrences) : replace "Boubee" with "Boubee" . Page 4, line 5 replace " particular" with "particularly" Page 4 : footnote 2 is on page 5 Page 5, line 6: remove hyphen Page 5, para 3, line 6: replace "predating" with "preying on" Page 14, para 2, line 3: replace "fluorscine" with "fluorescein" Page 16, para 3, line 8: replace "preformed" with "performed" Page 18, Figure 2 add to caption "Pie segments are proportional to total number of fish of each species." Page 19 : Replace P values = 0. 00 l with < 0 . 00 l. Page 20, para 2: replace "pfankuch" with "Pfankuch". Page 40, Table 3. add to caption "Sample size, n = 67" Page 48, abstract last line: replace "sit" with " site" Page 57, line 4: replace "concordence" with "concordance" Page 63 , para 2: add "Reach length surveyed was 200 metres". Page 68, para 2: replace " principle" with "principal" Page 69, para 3: replace "elctrofished" with "electrofished" z_ Abstract Freshwater fish communities were surveyed at 59 sites in the Mangatainoka, Makakahi and Ruamahanga catchments of the northeastern Tararua Ranges during 2000/01. At each site, habitat characteristics were recorded and fish identified by spotlighting over a 100 m reach. Benthic invertebrate samples were also collected from 50 of these sites. Shortjaw kokopu (Galaxias postvectis Clarke) occurred at 16 sites, located in the Mangatainoka and Makakahi catchments only. Ninty-five shortjaw kokopu were caught in total, ranging from juveniles (:=:;90 mm) to adults (> 120 mm), with adults comprising approximately 75 % of the population. Six other fish species were also recorded. Koaro (G. brevipinnis Giinther), longfin eel (Anguilla dieffenbachii Gray), Cran's bully (Gobiomorphus basalis Gray), torrentfish (Cheimarrichthys fosteri Haast) and brown trout (Salmo trutta Linnaeus) all co-occurred with shortjaw kokopu; and a single banded kokopu ( G. fasciatus Gray) was found in the Ruamahanga catchment. Discriminant analysis found six habitat factors defined shortjaw kokopu presence. These were low percentages of debris jams, pasture and backwaters; high percentages of shrubs and riffles; and high conductivity. The invertebrate community also proved effective at predicting shortjaw kokopu presence. However, it appears that shortjaw kokopu are limited in distribution by recruitment rather than habitat. Different age classes of shortjaw kokopu were also found to use distinct microhabitats. Sand substrate, pool length, width at the top of the pool, velocity, gradient below the pool, and cobble in the habitat above the pool were found to discriminate between the age class microhabitats. At three sites in the Mangatainoka River, surveys were undertaken monthly, for 16 months. Number of shortjaw kokopu observed was greatly reduced at all three sites during winter and at a maximum in autumn. This showed that shortjaw kokopu exhibited reduced activity rather than seasonal movements within the catchment. Three methods for surveying fish communities were tested on shortjaw kokopu. Gee­ minnow traps failed to catch any shortjaw kokopu, but electrofishing and spotlighting both proved effective. While spotlighting caught more shortjaw kokopu at more sites, no significant difference in performance was found between the two methods. 11 Acknowledgements I am grateful to everybody who has influenced this project, especially to my supervisor, Dr Ian Henderson, who helped set the project up. While 3rd year limnology developed my interest in New Zealand's freshwater fish, it was Ian's inspiration for this project which made it happen. I am also incredibly grateful for his thorough editing skills and statistical know-how. To Sjaan Charteris who was incredible support when things went to custard. Her thorough editing, her driving ambition for me to finish and her ability to keep me on the straight and narrow, ensured that I gave my best, managed my time efficiently, but maintained my sanity in the process. For that, I am sincerely thankful. I would like to acknowledge the funding from the Department of Conservation and horizons.mw. The river-line run by horizons .mw proved incredibly beneficial, allowing trips to be planned efficiently. I am incredibly grateful to my parents, Robbie and Robyn, and sister Cookie, who ventured into the field when volunteers were in short supply. They endured much hardship to see me fulfil my aspirations, tolerating some rough sleeping conditions, some long-nights in semi-freezing water, some sever bruising, and some long days hiking into remote, but incredibly picturesque locations. To the Ecology Group, Massey University. Thankyou to all who ventured into the field, especially Kirsty Francis, Matt Wong, Cindy Jenkins, Debbie Kyngdon and Mark Hamer; and also to other volunteers, Callum Kay and Sarah Clarke, who were not always happy to be there, but were prepared to give up their time. Mike Joy and Russell Death had a lot of valuable advise; and Jens Jorgensen created some fine field equipment. Paul Barrett, Cathy Lake, Tracy Harris and Hayden Hewitt provided some excellent technical support; and Erica Reid, Barbara Just, Jodi Matenga and Diana Crow, all provided great administrative support. I am particularly grateful to the landowners in the Putara Valley, particularly Paul and Chrissy Ferry, Jenny and John Davidson and Jim and Bev Bryant for access to sampling sites. 111 Thank you also to the Manawatu Department of Conservation, and to Russell Death (Massey University), for the use of both gee-minnow traps and an electrofishing machine. Russell proved quite motivational, with his biting, yet thought-provoking talks. To Erica, Barb and Tracy for their willingness to play the scape-goat when I needed some light relief. To Sjaan for keeping me motivated, especially when the situation didn't always go my way. And to Matt for always indulging my need for a coffee break. Finally, a special thank you to Jim and Bev Bryant, and my father Robbie, for their help rescuing my car when it broke down, middle of the night, middle of nowhere. iv Table of Contents Abstract Acknowledgements Chapter 1 Chapter 2 Chapter 3 Chapter 4 Chapter 5 Synthesis Appendix 1 General Introduction Distribution and habitat selection of shortjaw kokopu ( Galaxias postvectis) in the northern Tararua Ranges Seasonal activity of shortjaw kokopu ( Galaxias postvectis) in the northern Tararua Ranges Evaluation of three methods for surveying shortjaw kokopu ( Ga/axias postvectis) in the northern Tararua Ranges Habitat selection of juvenile, sub-adult and adult shortjaw kokopu ( Galaxias postvectis) in the northern Tararua Ranges Finding a galaxiid nest in the northern Tararua Ranges ii iv 1 11 35 48 60 73 76 V General Introduction F.E. Clarke (1899) on shortjaw kokopu: _ "rrfie thin[ wesdana species I enfarge upon, more frequent(y inhabits the s{uggish ana muaay-6ottomea creek§, but is also f ouna in company 'UJith (iafa;rias kokopu in the gravd-bottomea ana some of the rocky creek§. In its proportions it somewhat appro?(jmates to the aescription off asciatus, though it grows much farger, but seUom beyona 10 in. in £ength. It is not as haray in the aquarium as q. kokopu, ana has genera{[y the same feeaing habits, e?(f,ept that it aoes not takg, a suiface-bait as we{{. Strange to say, it is se{aom, if ever, troub{ea with the ffesh-worms before mentioned. I have aistinguishea this one with the specific name · of postvectis, on account of its percu{iar ana constant markings." Chapter 1. General Introduction. New Zealand has 36 recognized species of native freshwater fish, with two more recently discovered, but yet to be formally classified (RM. Alli bone (DoC 1 : Wellington) pers. comm. November 2001). Seven other species of marine wanderers also frequent freshwater from time to time (McDowall 2000). Of these 36 species, seven are members of the bully family (Gobiomorphus spp., Eleotridae), three of the eel family (Anguilla spp., Anguillidae), two of the smelt family (Retropinnidae) and 20 of the family Galaxiidae. The remaining four species are lamprey (Geotria australis Gray), torrentfish (Cheimarrichthys fosteri), black flounder (Rhombosolea retiaria Hutton) and the now extinct grayling (Prototroctes oxyrhynchus Gunther). Many of New Zealand ' s freshwater fish require access to both marine and freshwater, commonly known as diadromy. There are three forms of diadromy: catadromy, living in freshwater but migrating to sea to spawn (e.g. eels); anadromy, living at sea but migrating into freshwater to spawn (e.g. lamprey); and amphidromy, migration between marine and freshwater but not related to spawning (e.g. torrentfish) (McDowall 1990). The galaxiidae family comprises five diadromous and 15 non-diadromous species. The diadromous species (whitebait) exhibit either catadromy, i.e. inanga (Galaxias maculatus (Jenyns)) or amphidromy, i.e. giant kokopu (G. argenteus (Gmelin)), banded kokopu (G.fasciatus), shortjaw kokopu (G. postvectis) and koaro (G. brevipinnis). Of the diadromous galaxiids, shortjaw kokopu are thought to be the rarest, listed as category A in the endangered species rankings (Molloy & Davis 1994). Sharing this endangered species rating are several New Zealand icons such as kiwi (Apteryx spp.), takahe (Porphyrio mantelli hochstetteri), black robin (Petroica traversi) and kakapo (Strigops habroptilus). Shortjaw kokopu are widely distributed throughout New Zealand, from Puysegur Point on the South Island's south coast, to Kaitaia in the north and Bay of Plenty in the east, but at any given location, they are generally found in very low numbers (approximately 1-3 fish per 100 m; McDowall 1990, McDowall et al. 1996). Several factors have been suggested that may explain the rarity of shortjaw kokopu. They may be confined to specific rnicrohabitats that are rare (i.e. particular stream and substrate size), their activity patterns may not complement most survey methods (i.e. they are hard to find) , they may be rare through over-harvesting of 1 Department of Conservation. 2 Chapter 1. General Introduction. juvenile whitebait, manmade barriers to migration may affect access to adult habitat, or competition/predation by introduced trout may decrease populations. A diadromous lifecycle, like that of shortjaw kokopu, is beneficial for many species as it allows them to distribute around New Zealand's coastline, colonizing many rivers. This can be impeded however, if the rivers contain barriers to whitebait migration. These barriers can be natural , i.e. waterfalls and dry reaches, or manmade, i.e. dams and weirs (McDowall 1984, 1990). A diadromous lifecycle may therefore place constraints on habitat selection. Habitat quality is an important determinant of shortjaw kokopu presence (Williams & Given 1981, McDowall 1984, Swales 1991, McDowall et al. 1996). Many studies suggest forest cover of the stream is an important component of shortjaw kokopu habitat (McDowall et al. 1977, Eldon 1983, Eldon 1984, Nicoll 1984, Main 1987, McDowall 1990, 1996, 1997, McDowall et al. 1996). However, other studies on West Coast populations of shortjaw kokopu have shown that shortjaw kokopu avoid forests dominated by beech (Nothofagus spp; McDowall et al. 1977, McDowall et al. 1996, McDowall 1997, 2000). Removing this forest cover is thought to be a major cause of declining native fish populations (McDowall 1984, 1990, Rowe et al. 1999), however the effects of exotic forest are less clear. Recent studies have found populations of banded kokopu in mature exotic forests (Hicks I 998, Rowe et al. 1999, Rowe 2000), although unmodified native forest is still thought to be preferred (Rowe et al. 1999). In the short term, these exotic forests act like nati ve forests, providing the overhead cover, humidity , and potential food supply required by banded kokopu. However, at a larger scale, exotic forests have several drawbacks. They have a limited lifespan (c. 25 years) before harvesting removes them, which in turn causes turbidity problems in the water (Rowe 2000). They also regulate the flow different to native forests, having higher flood peaks than native forest, but reducing water levels during dry periods (Hicks 1998). Banded kokopu , and galaxiids in general, are sensitive to turbidity (Boubee et al. 1997, Richardson et al. 1998, Rowe & Dean 1998, Richardson et al. 2001). Suspended sediments in the water restrict the migration of juvenile banded kokopu into these rivers (Boubee et al. 1997, Richardson et al. 1998, Richardson et al. 2001) and also restricting banded kokopu feeding (Richardson et al. 1998, Rowe & Dean 1998). Koaro were also found to avoid turbid habitats (Boubee et al. 1997, Richardson et al. 1998), but were 3 Chapter 1. General Introduction. better able to feed in these habitats (Richardson et al. 1998, Rowe & Dean 1998), which was attributed to their dispersal into glacial silt clouded rivers. Studies on the other diadromous galaxiids habitat, including shortjaw kokopu, are limited, but are suggested to be similar to the requirements of banded kokopu (Hicks 1998). Substrate type has also been identified as important for shortjaw kokopu, particular the presence of boulders and cobbles (McDowall 1990, 2000, McDowall et al. 1996). Fine sediments in the substrate may also be a problem for galaxiids through its impact on preferred prey (Main 1987, McDowall 1996, McDowall et al. 1996). However, more work is required on the habitat needs of shortjaw kokopu. Migratory access is a problem that faces all diadromous fish, including shortjaw kokopu (McDowall 1984, 1990, 1998). Based on analysis of NZFFD2 records from the South Islands, West Coast, McDowall (1998) suggested that most of New Zealand's diadromous fish are found at low altitudes and short distances inland. In contrast, non­ diadromous species tend to be further inland and at higher altitudes. While some diadromous fi sh, including shortjaw kokopu, are capable of significant inland migrations , McDowall (1998) found that most individuals colonised suitable habitat at downstream sites. He reasoned that, particularly for shortjaw kokopu, this was because of abundance of suitable habitat near the coast. Joy et al. (2000) found a similar altitude relationship in Taranaki ; however, shortjaw kokopu were found more commonly further inland. Jowett & Richardson (1995), Jowett et al. (1996) and Jowett et al. (1998) also found similar trends for shortjaw kokopu. There are many methods used to survey fish communities, including those with shortjaw kokopu present. Electrofishing machines are an effective non-lethal means of surveying and identifying fish in an entire stretch of river, catching many of the species in the community. However, R.F.G. Barrier ((DoC: Wellington) pers. comm. October 2001) has suggested that electrofishing is not a good estimator of some galaxiid communities because many species are either not caught or only in small proportions relative to their true abundance. In the case of galaxiids, including shortjaw kokopu, daytime refuge may mean hiding under rocks, so electrofishing will still stun them, but may not extracted them from between rocks. For some diadromous galaxiids, especially 4 Chapter 1. General Introduction. shortjaw kokopu, spotlight surveying is a suggested better method (R.F.G. Barrier (DoC: Wellington) pers. comm. October 2001). This method allows nocturnal fish communities to be surveyed during their active period. Other methods of fish community surveying are the use of traps and nets. However, these require fish to move around to encounter the traps . Highly territorial or site attached species may be under­ estimated by trapping. Shortjaw-kokopu are often found in the same or neighboring microhabitat between survey trips (e.g. Caskey 1999), so setting traps in one microhabitat may not catch the shortjaw kokopu from nearby microhabitats . However, more information is needed to determine the best method for shortjaw kokopu surveying. Most native fish , including shortjaw kokopu, become harder to find during winter (Cadwallader 1975, R.F.G. Barrier (DoC: Wellington) pers. comm. October 2001). This is a problem in all survey methods, but especially in spotlight surveys, which require fish to be active within their habitat. Some salmonids are known to become nocturnal in low water temperatures (:S5°C), regardless of the length of daylight. However, diadromous galaxiids in New Zealand are already nocturnal (McDowall 1990). A non­ diadromous galaxiid, Galaxias vulgaris Stokell, has been found to have peaks in activity relating to time since darkness fell. For the diadromous galaxiids, this may partially explain the perception that they are hard to find, surveys have been undertaken at the wrong times. Other observed patterns are of reduced activity during winter. However, for shortjaw kokopu, only seasonal growth rates have been studied (Caskey 1999), with annual activity pattern requiring investigation. Introduced trout have often been described as a limiting factor on native fish distribution, including shortjaw kokopu (McDowall 1984, McDowall et al. 1996). While not excluding adult shortjaw kokopu from specific habitat, McDowall et al. (1996) argues that trout hold the competitive advantage and do prey on juveniles. Brown trout are known to feed on migrating whitebait shoals (McDowall et al. 1996); and while there is no direct evidence of trout predating shortjaw kokopu whitebait in particular, Eldon (1983) surmised that in rivers which support large numbers of shortjaw kokopu whitebait (e.g. Buller River), trout predation on shortjaw kokopu is 2 New Zealand Freshwater Fish Database, maintained by the National Institute of Water and Atmosphere (NIWA ; McDowall & Richardson 1983). s Chapter 1. General Introduction. highly probable. Unfortunately, this problem is difficult to control, especially for diadromous species with a need for sea access (McDowall 1984), so further work is needed into the effects trout have on different size classes of shortjaw kokopu. 1.1. Study Area All study sites are located within a 100 km2 region of the northeastern Tararua Ranges, North Island, New Zealand at approximately 40° 40' S, 175° 30' E. Three main catchments drain the study area, the northern flowing Mangatainoka and Makakahi Rivers, and the southern flowing Ruamahanga River. The Mangatainoka and Makakahi Rivers are tributaries of the Manawatu River, which flows to the west coast of the North Island, while the Ruamahanga feeds into Cook Strait. The lower Manawatu River has a series of barrage dams at several sites upstream of the Manawatu Gorge (Anonymous 2001) while the Ruamahanga has a barrage control gate at Lake Wairarapa. All three rivers originate m the largely unmodified Tararua Forest Park. At lower altitudes, the canopy is dominated by red beech (Nothofagus fusca), intermixed with podocarp forest. At higher altitudes, red beech/podocarp forest is replaced by kamahi (Weinmannia racemosa) and leatherwood (Olearia colensoi) shrubs, in the Mangatainoka and Makakahi catchments, and by silver beech (N. menziesii) in the Ruamahanga (New Zealand Forest Service 1976). In the Tararua Ranges north of the main Mangatainoka catchment, all beech species are absent (Rogers & McGlone 1994). The Makakahi River also flows through an exotic tree plantation (Pinus radiata) at the park boundary . The documented fish community of the three study rivers in the northeastern Tararua Ranges has been relatively unknown until early 1999 when a large population of shortjaw kokopu was discovered in the headwaters of the Mangatainoka River (Anonymous 1999) (Table 1 ). However, the Mangahao River, one of the neighboring catchments of the Manawatu Rjver, has been heavily surveyed (Table 1). This is because of the desiltation process required for the power generation dam on the river (Boubee et al. 1995). 6 Chapter 1. General Introduction. Table 1. Fish species recorded from NZFFD records of the northern Tararua Ranges up to and including the discovery of a large population of shortjaw kokopu (Anonymous 1999). Common name Scientific name Lamprey Geotria australis Gray Longtin eel 1 Anguilla dieffenbachitj Gray Shortfin eel A. australis Richardson Common smelt Retropinna retropinna (Richardson) Shortjaw kokopu G. postvectis Clarke Banded kokopu G. fasciatus Gray Koaro G. brevipinnis Gunther Dwarf galaxies G. divergens Stokell Brown mudfish Neochanna apoda Gunther Common bully G. cotidianus McDowall Redfin bully Gobiomorphus huttoni (Ogilby) Upland bully G. breviceps Stokell Gran's bully G. basa/is Gray Torrentfish Cheimarrichthys fosteri Haast Brown Trout Sa/mo trutta Linnaeus Mangatainoka River ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ Makakahl River ✓ ✓ ✓ Ruamahanga River ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ Mangahao River ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ In this study, habitat features and invertebrate communities that characterise the presence of shortjaw kokopu in the northeastern Tararua Ranges are investigated. Habitat characteristics are examined in relation to the presence of three age classes, particularly juvenile shortjaw kokopu. The seasonal activity of shortjaw kokopu, the associated fish community and the best method for surveying shortjaw kokopu is also examined. This thesis is presented as four individual papers. This has resulted in some repetition in introductions and methods between chapters. Part of this work has also been partially presented in a report for the Department of Conservation (Bowie & Henderson 2002). 7 Chapter 1. General Introduction. 1 .2 References Anonymous 1999. Fish find gives scientists hope for rare species. Dominion, Wellington , New Zealand. 20-02-1999: 1. Anonymous 2001. Going with the flow on Mangatainoka flood control. New Zealand Fish and Game Magazine: Taranaki/Wellington Supplement, August: 6-7. Boubee, J. , Forsyth , D., Richardson, J. , Barrier, R. & James, M. 1995: 1995 Mangahao Dams desilting fish & invertebrate survey. NIWA, Hamilton, New Zealand. Survey ELE312-0l/1. Boubee, J. A. T., Dean, T. L., West, D. W. & Barrier, R. F. G. 1997 : Avoidance of suspended sediment by the juvenile migratory stage of six New Zealand native fish species. New Zealand Journal of Marine and Freshwater Research 31: 61- 69. Bowie, S. & Henderson, I. 2002: Shortjaw kokopu (Galaxias postvectis) in the northern Tararua Ranges: Distribution and habitat selection. Department of Conservation, Wellington, New Zealand. DOC Science Internal Series 30 Cadwallader, P. L. 1975: Spontaneous locomotory activity of Galaxias vulgaris Stokell (Pisces : Salmoniformes) . New Zealand Journal of Marine and Freshwater Research 9: 27-34. Caskey, D. 1999: Shortjawed kokopu (Galaxias postvectis) research and surveys June 1998- June 1999. Department of Conservation, Stratford, New Zealand. Clarke, F. E. 1899: Notes on New Zealand Galaxidae, more especially those of the Western Slopes: with descriptions of new spec ies, &c. Transactions and Proceedings of the New Zealand Institute 31: 78-91. Eldon, G. A. 1983: Enigma of the third kokopu. Freshwater Catch (NZ) 23: 17-18. Eldon, G. A. 1984: Carnivorous kokopu in captivity. Freshwater Catch (NZ) 25: 9. Hicks, B. 1998: Effects of forestry on fish. Fish and Game New Zealand 7: 38-41. Jowett, I. G., Hayes, J. W., Deans, N. & Eldon, G. A. 1998: Comparison of fish communities and abundance in unmodified streams of Kahurangi National Park with other areas of New Zealand. New Zealand Journal of Marine and Freshwater Research 32: 307-322. Jowett, I. G. & Richardson, J. 1995: Habitat preferences of common, riverine New Zealand native fishes and implication for flow management. New Zealand Journal of Marine and Freshwater Research 29: 13-23. 8 Chapter 1. General Introduction. Jowett, I. G., Richardson, J. & McDowall, R. M. 1996: Relative effects of in-stream habitat and land use on fish distribution and abundance in tributaries of the Grey River, New Zealand. New Zealand Journal of Marine and Freshwater Research 30: 463-475. Joy, M. K., Henderson, I. M. & Death, R. G . 2000: Diadromy and longitudinal patterns of upstream penetration of freshwater fish in Taranaki, New Zealand. New Zealand Journal of Marine and Freshwater Research 34: 531-534. Main , M . 1987: Feeding habits of the shortjawed kokopu. Freshwater Catch (NZ) 31: 13. McDowall, R. M. 1984: Designing reserves for freshwater fish 111 New Zealand. Journal of the Royal Society of New Zealand 14: 17-27. McDowall, R. M . 1990: New Zealand Freshwater Fishes: A natural hi story and guide, 2 Ed. Auckland, New Zealand. Heinemann Reed, 553 pp. McDowall, R. M. 1996: Volcanism and freshwater fish biogeography in the northeastern North Island of New Zealand. Journal of Biogeography 26: 139- 148. McDowall, R. M. 1997: Indigenous vegetation type and the distribution of the shortjawed kokopu, Galaxias postvectis (Teleostei: Galaxiidae), in New Zealand. New Zealand Journal of Zoology 24: 243-255. McDowall, R. M. 1998: Fighting the flow: dowstream-upstream linkages in the ecology of diadromous fish faunas in West Coast New Zealand Rivers. Freshwater Biology 40: 111-122. McDowall, R. M. 2000: The Reed Field Guide to New Zealand Freshwater Fish. Auckland, New Zealand. Reed Publishing (NZ) Ltd, 224 pp. McDowall, R. M., Eldon, G. A., Bonnett, M. L. & Sykes, J. R. E. 1996: Critical habitat for the conservation of shortjawed kokopu , Galaxias postvectis Clarke. Wellington, New Zealand. Department of Conservation, 80 pp. McDowall, R. M., Graynoth, E. & Eldon, G. A. 1977: The occurrence and distribution of fishes in streams draining the beech forests of the West Coast and Southland, South Island, New Zealand. Journal of the Royal Society of New Zealand 7: 405- 424. McDowall , R. M. & Richardson, J. 1983: The New Zealand freshwater fish survey: a guide to input and output. New Zealand Ministry of Agriculture and Fisheries , 15 pp. Molloy, J. & Davi s, A. 1994: Setting priorities for the conservation of New Zealand's threatened plants and animals, 2 Ed. Wellington, New Zealand. Department of Conservation, 64 pp. 9 Chapter 1. General Introduction. New Zealand Forest Service 1976: Tararua State Forest Park, Wellington, New Zealand. A.R. Shearer, Government Printer, 28 pp. Nicoll , G. 1984: Freshwater life surveyed in South Westland. Freshwater Catch (NZ) 24: 4-6. Richardson, J., Boubee, J. , Dean, T. , Hicks , M. , Rowe, D. & West, D. 1998: Effects of suspended solids on migratory native fish . Water and Atmosphere 6: 22-23. Richardson, J. , Rowe, D. K. & Smith, J. P. 2001 : Effects of turbidity on the migration of juvenile banded kokopu (Galaxias fasciatus) in a natural stream. New Zealand Journal of Marine and Freshwater Research 35: 191-196. Rogers, G. M. & McGlone, M. S. 1994: A history of Kaiparoro clearing and the limits of Nothofagus in the northern Tararua Ranges, New Zealand. New Zealand Journal of Botany 32: 463-482. Rowe, D. 2000: Can pine plantations restore whitebait fisheries? New Zealand Tree Grower 21: 34-35 . Rowe, D. K., Chisnall, B. L. , Dean, T. L. & Richardson , J. 1999: Effects of land use on native fish communities in east coast streams of the North Island of New Zealand. New Zealand Journal of Marine and Freshwater Research 33: 141- 151. Rowe, D. K. & Dean, T. L. 1998: Effects of turbidity on the feeding ability of the juvenile migrant stage of six New Zealand freshwater fish species. New Zealand Journal of Marine and Freshwater Research 32: 21-29. Swales, S. 1991: Threats and conservation of native fi sh. Freshwater Catch (NZ) 45: 19-21. Williams, G. R. & Given , D. R. 1981: The red data book of New Zealand: rare and endangered species of endemic terrestrial vertebrates and vascular plants, Wellington, New Zealand. Nature Conservation Council, 175 pp. 10 2. Abstract Distribution and habitat- selection kokopu ( Galaxias postvectis) in Tararua Ranges of shortjaw the northern Freshwater fish communities were surveyed at fifty sites m the Mangatainoka, Makakahi and Ruamahanga catchments in the northeastern Tararua Ranges. At each site habitat, invertebrate and fish communities were assessed. Shortjaw kokopu occurred in nine of the thirty-seven sites in the Mangatainoka and Makakahi catchments, but none were found in the Ruamahanga catchment. Shortjaw kokopu co-occurred with longfin eels, brown trout, Cran's bully and koaro. Shortjaw kokopu presence could be predicted by the invertebrate community composition and from habitat eharacteristics. Shortjaw kokopu were normally found in medium sized streams in native forest, with a high j (J density of riffles and high conductivity. Discriminant Function Analysis accurately predicted shortjaw kokopu occurrences in the three river catchments. It appears that shortjaw kokopu are limited in distribution by recruitment rather than habitat. Key Words: Shortjaw kokopu, Galaxias postvectis, distribution, habitat selection, Tararua Ranges, New Zealand. Chapter 2. Shortjaw kokopu habitat and distribution. 2.1 Introduction Shortjaw kokopu (Galaxias postvectis) occur in less than 2% of the NZFFD 1 records (McDowall et al. 1996a), yet their distribution ranges from Puysegur Point in the south , to Kaiatia in the north , along the length of the west coast, and across to the Bay of Plenty (McDowall 1990, 2000, McDowall et al. 1996a). Of the five species of diadromous galaxiidae, shortj aw kokopu are rarest and have been ass igned category A threatened species status (Molloy & Davis 1994 ). Although most shortjaw kokopu records in the NZFFD are of single, or a few individuals, there are some concentrated local populations, particularly on South Island' s West Coast, in Taranaki , and in the Bay of Plenty (McDowall 1990, 2000, McDowall et al. 1996a). Recent surveys also suggest Nelson/Marlborough (Studholme et al. 1999, Jack & Barrier 2000) , the Manawatu Ri ver catchment (Anonymous 1999) and the South Coast of the North Island (Rebergen & Joy 1999) have sizable populations of shortjaw kokopu. Published information on habitat requirements indicates the need for forest cover (Eldon 1983 , Main 1987, McDowall 1990, 1997), large substrate (Nicoll 1984, McDowall 1990, McDowall et al. 1996a), and pools (Eldon 1983, McDowall 1990, McDowall et al. 1996a). Data in the NZFFD suggest shortjaw kokopu are predominantly found at low altitudes (::: 125 rn ; McDowall 1998), and small distances inland (:S25 km; McDowall 1998), although exceptions are known (e.g. Caskey 1999, Studholme et al. 1999, Jack & Barrier 2000, Joy et al. 2000) . Juvenile shortj aw kokopu can climb obstacles such as small waterfalls almost as well as koaro (Galaxias brevipinnis) (McDowall 1990), so these are not the cause of the limited inland di stribution. McDowall et al. (1977) found that, on the West Coast of the South Island, shortjaw kokopu were generally present only in catchments where beech (Nothofagus spp. ) was absent or only a minor component of the forest, with similar results reported by Main (1989), McDowall et al. (1996a) and McDowall (1997). Main (1987) found that large populations of shortjaw kokopu occur in streams with a large component of bouldery substrate (>256 mm). The presence of shortjaw kokopu in the Mangatainoka River was first reported in February 1999 with 49 shortjaw kokopu observed by spotlight along a 500 m reach of a 1 New Zealand Freshwater Fish Database, maintained by the National Institute of Water and Atmosphere (NIWA; McDowall & Richardson 1983). 12 4th order stream (Anonymous 1999). Two days earlier, 8 shortjaw kokopu had been electrofished from the same reach (I.M. Henderson (Massey University: Palmerston North) pers. comm. February 2000). This new population was as far or further inland (c.180 km) than other records in the Manawatu catchment, Kahuterawa Stream (c.89 km; NZFFD, McDowall et al. 1996a), and the Mangahao River (c.176 km; NZFFD, Boubee et al. 1995). Koaro and brown trout (Salmo trutta) were also found at the same location, koaro occupying the same section of stream, and brown trout downstream of this section (c. 200 m; M.K. Joy (Massey University: Palmerston North) pers. comm. February 2000). In this survey I assess the distribution of shortjaw kokopu in the Mangatainoka River and adjacent catchments of the northeastern Tararua Ranges . I document the fish communities at these sites, assessing any associations between fish community and habitat characteristics, including invertebrate community composition. Finally, using three rivers, I build a model for predicting the occurrence of shortjaw kokopu based on habitat and invertebrate characteristics. 2.2 Methods The survey was carried out at 50 sites in the northern Tararua Ranges, 25 in the headwaters of the Mangatainoka River, 12 in the Makakahi River, and 13 in tributaries of the Ruamahanga River. The study area comprises the upper reaches of the three river catchments (Figure 1). The Mangatainoka and Makakahi Rivers converge, join the Manawatu River, and flow to the west coast. The Ruamahanga River flows south to its river mouth. Within the study area, the three catchments are generally forested, dominated by beech intermixed with podocarp forest (New Zealand Forest Service 1976). After leaving the study area, the Mangatainoka and Makakahi Rivers travel 180 km; and the Ruamahanga 150 km before reaching the sea. Sites were selected on their suitability for spotlight surveys, including ease of access and presence of large pools, while ensuring wide coverage of the catchments. Each site comprised a 100 m reach without major tributaries converging. 13 At each site, a range of catchment, habitat and chemical measures were recorded. Catchment variables were obtained from a 1:50000 topographic map (NZMS 260 S25: "Levin", 1995). These measures were stream order (Strahler 1952), altitude and gradient. The gradient was assessed from the spacing of contour lines within the 100 m reach and the altitude estimated at the mid point. Average width was estimated from five transects spread the length of the site. Average depth was recorded from five equally spaced points on each width transect; and velocity by timing the movement of fluorscine (BDH Laboratory Supplies: GPR™) dye along the length of the site. Substrate composition was recorded using the Wolman walk method (Wolman 1954) and the size categories in Table 1. Flow type, overhead cover, undercut banks, debris jams and riparian vegetation types were visually assessed on a percentage scale. The instream moss and periphyton cover were visually assessed on a IO-point scale (1 = least; 10 = most). Streambed stability was assessed using the bottom section of the pfankuch stability index (Pfankuch 1975, Death 1995). Conductivity (corrected to 25°C) and temperature were measured using an Orion (model 122) conductivity meter. Variation in conductivity due to rainfall and flow dilution was removed by using the residuals from a linear regression of conductivity and flow rates at a gauging station2 approx. 7 km downstream on the Mangatainoka River. Flow type was classified into backwater, pool, large pool, riffle, run and falls. Backwater was any area of still water connected to, but not influencing the mam channel during base flow. Pools were slow moving water less than 5 m long or lm deep, while large pools were greater than 5 m long or 1 m deep. Riffles were shallow swift broken water; runs were slow to moderately fast water with a calm or rippled surface; and falls were fast flowing water over a vertical drop. Riparian vegetation types were classified into the percentage of podocarp, beech, shrubs, exotic, pasture, tussock and bare rock in the riparian zone. Podocarp, beech and shrubs describe the native species in the riparian canopy vegetation; shrubs included ferns and toetoe. Exotic were any introduced tree species including pines and gorse. 2 horizons.mw river monitoring system at Larsens Road. 14 Pasture and tussock were the introduced and native grasses, respectively; and bare rock was the unvegetated bank side margins. ... ······· i 1 s l ···. f f f •' ····· ... ........ , N i < 0 3km Figure 1. The sites on three rivers of a distribution and habitat selection survey on shortjaw kokopu in the northern Tararua Ranges. A substrate size index was calculated from the percentage composition of size classes (Table 1) using: B * 8 + LC* 7 + C * 6 + LP* 5 + P * 4 + LG* 3 + G * 2 + SG * 1 + S & S * 0 Numbe1i-_ of 1:::.... Rocks Table 1. Substrate size classes. Boulders (B) Large Cobbles (LC} Cobbles (C) Large Pebbles (LP) Pebbles (P) Large Gravel (LG) Gravel (G) Small Gravel (SG) Coarse Sand (CS) Sand & Silt (S&S) > 256 mm 128-256 mm 64-128 mm 32-64 mm 16-32 mm 8-16 mm 4-8mm 2-4mm 1-2mm sl mm 15 A single sample of invertebrates was collected from a run or riffle within each site using a 500-µm mesh kick-net and disturbing the substrate for 30 s. Three locations within the habitat were disturbed for 10 s each. All invertebrate samples were preserved in a 10% formalin solution. Invertebrate samples were sorted and identified to a genus level using keys in Winterboum & Gregson (1989), and counted. If the invertebrates were not insect larvae, then they were grouped as Oligochaetae, Crustacea, Platyhelminthes, Potamopyrgus spp. or mites. Approximately 30 minutes after sunset, following completion of habitat assessment and invertebrate collection, a fish survey was caITied out, using two observers with 30-watt spotlights, powered by 12 volt, 7 amp hour batteries. Two upstream and two downstream traverses of the reach were made, taking c. 20-30 minutes per site. All fish were identified under the spotlight beam and most galaxiids were caught in hand nets and measured. The length of galaxiids unable to be captured was estimated to ±20 mm. At low altitude sites, a selection of eels and bullys were also captured to ensure species identification. Brown trout and longfin eels (Anguilla dieffenbachii) life stage was assessed by eye to be either juvenile or adult. Cran ' s bully (Gobiomorphus basalis) and koura (Paran ephrops planifrons) were counted but not measured. 2.2.1 Statistical analysis Invertebrate taxa occurring at less than five sites (10% of sites) were excluded from statistical analysis and invertebrate counts were log transformed. For the Mangatainoka invertebrate community analysis, the five taxa of chironomids were combined and taxa that did not occur in at least four of the Mangatainoka sites (16%) were ignored. The relationship between fish community structure, and invertebrate communities and habitat variables were analysed using Canonical Correspondence Analysis (CCA) with PCORD (Version 4.17; McCune & Mefford 1995). A Monte-carlo test with 1000 iterations was preformed to test the significance of the correspondence. Stepwise Discriminant Analysis (SDA) was run on subsets of the data, building a model to predict presence or absence of shortjaw kokopu. The initial analysis used only the Mangatainoka River sites, the second level used the Mangatainoka and Makakahi Rivers (Manawatu Rivers), and the third level used the Manawatu Rivers and the 16 Ruamahanga River (complete data set). Each set of variables selected by the SDA was run through a Discriminant Function Analysis (DFA) to determine the accuracy of the model , followed by a test on a new set of data. Both the SDA and DFA were analysed using SAS (2000). 2.3 Results Five native and one exotic species of fish were recorded. Native fish were found at 43 of the 50 sites (86%). Shortjaw kokopu occurred in only nine of the 50 sites surveyed (18%), in the Mangatainoka River (seven sites, 28 %) and Makakahi River (two sites, 16%; Table 2). Densities of shortjaw kokopu at the nine sites ranged from l to 7 fish per 100 m, with a total of 32 recorded. Of these, 26 were found in the Mangatainoka catchment, and six in the Makakahi catchment (Table 3). Koaro were found at five sites, four on the Mangatainoka River, and one on the Ruamahanga River. At the same site on the Ruamahanga River, the only banded kokopu found in this survey was also recorded. Brown trout and longfin eels were found in all surveyed catchments. Longfin eels were the most abundant and ubiquitous species , comprising 30-75 % of the fish recorded in each catchment and present in all but seven sites. These seven sites were high altitude (> 565 m, three sites), small 2nd order streams (three sites), or a heavily eroded, unstable stream (one site). Brown trout and Cran's bully were recorded mostly in low altitude sites without much forest cover. In the Ruamahanga catchment, the only brown trout recorded, occurred at one of the higher altitude sites not inhabited by Iongfin eel s. Cran ' s bully was found at nine sites, all of them low altitude, and all in the Mangatainoka and Ruamahanga Rivers . While not widely distributed, they often numerically dominated the sites where they did occur (Figure 2). Table 2. Occurrence of fish species by river catchment. River No. sites No. fish No. Mangatainoka Makakahi Ruamahanga 25 12 13 species 5 3 5 galaxiids 46 6 5 No. shortjaw kokopu 26 6 0 No. brown trout 13 4 1 17 Brown trout co-occurred with shortjaw kokopu at 3 si tes (Figure 3), all of which were at the lower range of altitudes containing shortjaw kokopu. A chi-squared test showed no sign ificant association (positive or negative) between brown trout presence and shortjaw kokopu presence (x2 = 0.7831, P = 0.3762). Koaro and banded kokopu were never found co-occurring with brown trout. Table 3. Species composition of all fish recorded from SO sites. Species Total No. fish No. occurrences Shortjaw kokopu ( Galaxias postvectis) Koaro ( Galaxias brevipinnis) Banded kokopu ( Galaxias fasciatus) Longtin eel (Anguilla dieffenbachia) Gran 's bully ( Gobiomorphus basal is) Brown trout ( Sa/mo trutta) = Shortjaw kokopu = Koaro - Banded kokopu = Longtin eel = Cran's bully = Brown trout C. 32 24 147 113 18 a. d. 9 5 43 9 12 b. Average number fish per occurrence 4 5 4 13 2 Figure 2. Distribution of the 6 fi sh species found in Mangatainoka, Makakahi and Ruamahanga River catchments. a. A ll rivers , b. Manawatu Rivers, c. Mangatainoka River, d. Makakahi Ri ver, and e. Ruamahanga River. 18 2.3.1.a Association of the fish community composition with habitat ' - Correlations between habitat and fish community was significant for both axes in all data sets. In the complete data set, axis l (P = 0.001) explains 21.4% of the variance and axis 2 (P = 0.002) explains 25.4%. In the Manawatu data set, axis I (P = 0.001) explains 34.2% and axis 2 (P = 0.001) explains 23.8%. In the Mangatainoka data set, axis 1 (P = 0.001) explains 47.3% and axis 2 (P = 0.001) explains 30.6%. Altitude is an important factor in all data sets, highly correlated with axis 1 or axis 2. Pasture, periphyton and run habitat are also important (Figure 4). Longfin eels show few associations with the habitat. The galaxiids show preference for more vegetative cover and higher altitudes, although banded kokopu only occur in the complete analysis due to the failure to find any in the Manawatu catchments. Cran's bully and brown trout show preference for open streams with run habitat. 45 40 35 ..c 30 U) u:: 0 25 ,._ Q) ..o 20 E :::::i z 15 10 c=::J Shortjaw kokopu ~ Koaro - Banded kokopu c=::J Longfin eel c=::J Cran's bully 111111111111111 Brown trout -NMV~~~oomo- NMV~~ ~romo-NMV~-NMV~~~oomo-N-NMV~~~romo-NM ooooooooo----------NNNNNNooooooooo---ooooooooo---- 0000000000000000000000000~~~~~~~~~~~~<<<<<<<<<<<<< ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ Sites Figure 3. Number of fish found at each survey site in a study of shortjaw kokopu distribution in the northern Tararua Ranges. Abbreviations are: MG01-MG25 = Mangatainoka River sites, MK0I-MK12 = Makakahi River sites, and RUA01-RUA13 = Ruamahanga River sites. 19 2.3.1.b Association of the fish community composition with invertebrates Correlations between invertebrate and fish community structure was significant for both axes in each data set. In the complete data set, axis 1 (P = 0.001) explains 27.8% of the variance and axis 2 (P = 0.003) explains 15.8%. In the Manawatu data set, axis 1 (P = 0.001) explains 35.8% and axis 2 (P = 0.001) explains 26.2%. In the Mangatainoka data set, axis 1 (P = 0.001) explains 42.2% and axis 2 (P = 0.001) explains 30.8%. Zelandobius spp. is correlated with axis 1 for all data sets (Figure 5). Aphrophila spp., Chironomidae and Oxyethira spp. are also strongly correlated with axis 1 or axis 2. Longtin eels show no distinct associations with any invertebrate community. The galaxiids all show a positive association with Zelandobius spp., and negative associations with Oxyethira spp. Cran's bully and brown trout are negatively associated with Zelandobius spp. 2.3.2 Predicting Shortjaw Kokopu presence 2.3.2.a Developing the Mangatainoka River model Habitat model High instream stability (i.e. low pfankuch score) was the strongest predictor of shortjaw kokopu presence. High percentage riffle and low percentage run were also selected by the stepwise analysis. Using these three variables, six sites were misclassified. Five sites were incorrectly predicted to have shortjaw kokopu present. The predicted probabilities of shortjaw kokopu occurrence in these cases were relatively low (P = 0.55, 0.56, 0.58, 0.61 and 0.74). One site was incorrectly predicted to not have shortjaw kokopu (P = 0.63). Invertebrate model Occurrence of Aphrophila was the best predictor for the presence of shortjaw kokopu. Other important variables were the presence of Zephlebia mayflies, Tanytarsus chironomids and Zelandobius stoneflies; and absence of Nesameletus mayflies, Polypedilum chironomids and Aoteapsyche caddisflies. Perfect prediction for shortjaw kokopu was achieved. 20 C\I UI ~ a. • cl 0 " .. • Shortjaw lmlmpu 0 • • Cran' bullys " " 0 ·-tl'Olf 0 " Run ~ Periph:oio ! o LDngfln eeb O o Altltude 6 Overhead cover " • a " " 0 Axis 1 •- • Banded kokopi b. • Cran's bully • .:Uwn trout " "" C\1 1 Pasfure"~Run Stream order o _!,g O " ~ • 00 : o • Altitude • 'il, • Longfln eel • 6 0 l! .o • 0 • " 0 • Shortjaw kokopu "" " Axis 1 c. • Koaro • Koaro " "" " " Cran'• buDy Pasture 6 C\11 " • 6 Longflneel _!,g ~ -Altitude • ~ arownL Periphyton ~thrubs • " • • • • • • Shortjaw kokopu Axis 1 Figure 4. Canonical Correspondence Analysis of fish communities and habitat variables in: a. complete data set; b. Manawatu data set; and c. Mangatainoka data set. D.. = Mangatainoka River, 0 = Makakahi River and • = Ruamahanga River. Shaded symbols represent shortjaw kokopu presence. Large squares are the fish community associations (•). Vectors show the highest four correlations between habitat variables and ordination axes. C\I Ill ~ a. • Bandedkokopu 0 •- • Shor1;1awkokopu 0 0 0 .. 0 6 Jl~anytarsus Zelandoblus O t.ngfln eel 0 o ~ 0 p,cyelhlra 6 6 600.t ~ 6 6 6 •-- 6 Chlronomldae • Cran'sbuDy Axis 1 b. • eran'sbully 6 6 C\I I Brown trout~ Aphroi;:,lla ,. o 111 • Oxyeft.1ra 0 ~ Chironomidae O '--\_oolongfin eel .. ~ Zeianlobius /j 6 6 .. 6 a a .......... 6 6 • Short;faw kokopu Axis 1 • Koaro c. • Cran's bully 6 6 C\I IBnlwn ~ Aphro~hlla 6 0 111 • Oxyei1ra0 ~ Chironomldae0 '--\..ool.ongfln eel ,..;-"W; Zelantoblus /j 6 6 .. 6 a a .......... 6 6 • Short;faw kokopu Axis 1 • Koaro Figure 5. Canonical Correspondence Analysis of fish and invertebrate communities in: a. complete data set; b. Manawatu data set; and c. Mangatainoka data set. .6.. = Mangatainoka River, 0 = Makakahi River and • = Ruamahanga River. Shaded symbols represent shortjaw kokopu presence. Large squares are the correlations between fish species and the ordination axes (•). Vectors show the highest four correlations between habitat variables and ordination axes. Chapter 2. Shortjaw kokopu habitat and distribution. Combined habitat and invertebrate model Eight variables were selected as indicators for the presence of shortjaw kokopu . Presence of Aphrophila was the best indicator, followed by high percentage beech forest, presence of Zephlebia, high percentage riffle habitat, absence of Peritheates midges, high pfankuch score, absence of Hydrobiosella caddisflies and high altitude. Perfect prediction for shortjaw kokopu presence was achieved. Testing the models on Makakahi River data A discriminant model based on habitat variables alone misclassified 50% of sites. Five cases were predicted of shortjaw kokopu occurrence (P = 0.60, 0.60, 0.73 , 0.89, 1.0) and one predicted absence (P = 0.6 1). Discriminant models based on invertebrate variables and the combined variables both misclassified two sites with shortjaw kokopu incorrectly to be absent (P = 1.0, 1.0). 2.3.2.b Developing the Manawatu Rivers model Habitat model Low pfankuch score (Figure 6) was the only predictor for the presence of shortjaw kokopu. Fourteen sites were incorrectly predicted, 13 predicted shortjaw kokopu to be present (P = 0.55-0.68) and one predicted shortjaw kokopu to be absent (P = 0.81). Invertebrate model Occurrence of Helicopsyche caddisflies was found to be the strongest predictor of shortjaw kokopu presence. The other important variables were the absence of Stenoperla stoneflies, Eriopterini crane flies and Polypedilum; and the presence of Tanytarsus. Eleven misclassifications occurred, all predicting shortjaw kokopu to be present (P = 0.97-1.0). Combined habitat and invertebrate model Fourteen variables predicted the presence of shortjaw kokopu. These were the absence of Eriopterini, Stenoperla, Nesameletus, Hydrobiosella caddisflies and Deleatidium; presence of Tanytarsus, Helicopsyche and Archichauliodes dobsonflies; low percentage debris jam, pasture and backwater; high percentage shrubs, riffles and high conductivity. High pfankuch was the first predictor chosen by the stepwise procedure, but was later removed. No sites were misclassified with these variables. 23 Chapter 2. Shortjaw kokopu habitat and distribution. 15 c 5 10 0 5 5 c :::, 10 0 0 15 20L----'----_._--~----' 10 20 30 40 50 Pfankuch index Shortjaw kokopu 1z1 Absent • Present Figure 6. The presence of shortjaw kokopu declines as instability (pfankuch score) rises in the Manawatu Rivers. Testing the models on Ruamahanga River data No site in the Ruamahanga River contained shortjaw kokopu, however, the habitat model misclassified 11 of 13 sites (P = 0.51-0.79). Twelve of 13 sites were misclassified using the invertebrate model (P = 1.00). However, perfect classification was achieved using the combined model. 2.3.2.c Developing the complete model Habitat model High conductivity, low percentage debris jams, pasture, podocarp, altitude and run; and high percentage riffles were identified as good predictors of shortjaw kokopu (Plate 1). One misclassification occurred predicting shortjaw kokopu to be present (P = 0.96). Invertebrate model The occurrence of Aphrophila and Coloburiscus and absence of Austroperla stoneflies were the best predictors of shortjaw kokopu presence. Eleven misclassifications occurred, seven predicting shortjaw kokopu to be present (P = 0.56-0.85) and four predicting absent (P = 0.53-0.78). 24 Chapter 2. Shortjaw kokopu habitat and distribution. Plate 1. Sites that shortjaw kokopu were present on the Mangatainoka (a, b, c, d, e & t) and Makakahi Rivers (g & h). 25 Chapter 2. Shortjaw kokopu habitat and distribution. Combined habitat and invertebrate model Ten variables predicted shortjaw kokopu presence. These were a high percentage of shrubs and beech ; high conductivity and stream order; low percentage run and undercut banks, small average stream widths; absence of Eriopterini and Austroperla; and presence of Coloburiscus. Absence of Aphrophila and low percentage of debris jams were the first predictors selected by the stepwise procedure, but were later removed. This model achieved perfect prediction. 2.4 Discussion Given the previous di scovery of a large population of shortjaw kokopu within the Mangatainoka catchment (Anonymous 1999), the sparse di stribution and low numbers of shortjaw kokopu found in other headwater reaches of the Mangatainoka catchment was unexpected. Many of the surveyed reaches had habitat characteristics matching those described in literature as suitable for shortjaw kokopu, such as large substrate (Nicoll 1984, McDowall 1990, McDowall et al. 1996a), forest cover (Eldon 1983, Main 1987, McDowall 1990, 1997), and pool habitat (Eldon 1983, McDowall 1990, McDowall et al. 1996a). Koaro were also expected to be more widespread in the survey . However, their appearance at the upper limits of shortjaw kokopu range may account for their apparent limited di spersal throughout the northern Tararua Ranges. The survey was designed for shortj aw kokopu, and while the larger main ri vers were also surveyed, smal ler streams at high altitude where koaro are more commonly found , were not always as easily accessible. Streams where koaro or banded kokopu occurred were often incorrectly predicted to contain shortjaw kokopu by di scriminant models. The appearance of banded kokopu in the survey, living in close proximity to koaro, was unexpected. Their lack of occurrence in the Mangatainoka or Makakahi catchments suggested the northeastern Tararua Ranges were beyond the limits of dispersal for thi s species. However, banded kokopu are regarded for their migratory abilities (McDowall 1990, 2000), almost as much as koaro. Banded kokopu and shortjaw kokopu do co­ occur on western slopes of the Tararua Ranges, however, in tributaries draining into the Manawatu River, such as the Kahuterawa Stream (NZFFD) . 26 Chapter 2. Shortjaw kokopu habitat and distribution. Longfin eels were the most widely distributed fish species, not surprisingly as they are known in virtually all accessible rivers in New Zealand (NZFFD, McDowall 1990, 2000). Longfin eels were found in most habitat types, excluding some high altitude sites and small 2nd order streams. Most of these streams often have accessibility problems, such as culverts, waterfalls or subterranean flows; but a brown trout were found in one of these high altitude streams. In the Manawatu Rivers, brown trout were restricted to the low altitude rivers, found in most of the streams with pastural vegetation , but also in the low altitude bush covered streams. The survey in February 1999 (Anonymous 1999) recorded a large brown trout less than 200 m below the convergence of the two 4th order streams where shortjaw kokopu were present (I.M. Henderson (Massey University: Palmerston North) pers. comm. February 2000). So while brown trout were expected in the main river, and other large branches, their occurrence in smaller streams was not expected. The single record of brown trout in the Ruamahanga catchment was unusual. Brown trout were observed in the Ruapae Stream, a tributary of the Ruamahanga River that was surveyed, at high altitude. It seems unusual that brown trout were not observed in larger tributaries downstream. Some of these streams were observed to have significant waterfalls bordering the main river, but others were much more accessible to migrating trout and found to contain Cran's bully which had a close association with trout in Mangatainoka streams. Cran ' s bullies were not as widely distributed as the other fish species, but were numerically as abundant as longfin eels in the Mangatainoka catchment. Cran ' s bully was restricted to low altitude large rivers. This was the only non-migratory native fish found . Shortjaw kokopu were found co-occurring with most other fish species, except banded kokopu. Longfin eels were found at all sites containing shortjaw kokopu, whereas koaro, brown trout and Cran's bullys were only found co-occurring with shortjaw kokopu in some sites. Furthermore, at no site did brown trout and Cran's bully's co­ occur with koaro or banded kokopu. 2.4.1 Fish Communities Canonical Correspondence Analysis showed distinct associations between fish communities and invertebrate communities or habitat variables. All the galaxiid species showed preferences for forested habitat at higher altitudes, a trait shared with the 27 Chapter 2. Shortjaw kokopu habitat and distribution. stonefly Zelandobius spp. Cran's bully and brown trout were associated with large open low altitude farmed streams, supporting cased caddisflies, Oxyethira and Aphrophila. Sites that contained shortjaw kokopu were quite distinct from other sites based on the Mangatainoka River analysis, but became less so as more rivers were included. This could mean that the Ruamahanga River with sites containing suitable habitat but lackjng shortjaw kokopu, is having an effect on the analysis. If this were the case, then we would not have expected to get perfect prediction from either the Manawatu River model or the complete river model. Generally, shortjaw kokopu were found at low to mid altitudes in the Manawatu catchments. Their downstream range overlaps with brown trout, and the upstream range, in some cases overlaps with koaro. While there was no overlap for koaro and brown trout, in some sites, the two ranges were less than 100 m apart. Longfin eels were present at all sites containing shortjaw kokopu, koaro, banded kokopu or Cran's bully, and at all but one site containing brown trout. 2.4.2 Shortjaw kokopu prediction models The shortjaw kokopu prediction models varied in effectiveness with up to 38% of sites misclassified with habitat models and 30% with invertebrate models. The habitat variables were chosen to allow rapid identification of site viability for shortjaw kokopu. While requiring numerical processing following collection, habitat variables were much less time- consuming than invertebrate sampling which needed sorting and identifying, but required less effort collecting samples in the field. A combined approach yielded the best results, but required more effort; however perfect prediction ensued. A true test of the models is to make predictions on sites not used to build the model. This was undertaken in two stages, using the Mangatainoka model to predict shortjaw kokopu occurrence in the Makakahi River, and using the Manawatu model to predict shortjaw kokopu occurrence in the Ruamahanga River. The Makakahi River had not been previously surveyed for fish communities but has the same downstream factors (common Manawatu stem) as the Mangatainoka River. Headwater tributaries of the Ruamahanga River have many site-specific habitat variables similar to sites in the Mangatainoka and Makakahi headwaters. However, the Ruamahanga River catchment does not share the same coastline, river mouth or lower river as the other two rivers. 28 Chapter 2. Shortjaw kokopu habitat and distribution. The failure of the Mangatainoka model to correctly predict shortjaw kokopu occurrence in the Makakahi River shows that the Mangatainoka is not indicative of shortjaw kokopu in the northeastern Tararua Ranges. By incorporating the Makakahi River into the model, the Manawatu model gains more predictive power, correctly classifying the Ruamahanga River sites. The model built from the complete data set also has perfect prediction but is untested due to a lack of a wider data set. However, both predictive models had several variables in common, particularly aspects of the habitat. Shortjaw kokopu presence is associated with high percentage shrubs, high conductivity and low percentage debris jams, although debris jams were later removed in the complete model. High proportion of shrubs in the riparian zone is associated with stable forested sites, matching described trends in the literature (Eldon 1983, Main 1987, McDowall 1990, 1997), whereas debris jams, also associated with forested catchment, seem to go against described trends (R.F.G. Barrier (DoC3 : Wellington) pers. comm. October 2001). Conductivity is an unexpected predictor, particularly as the three catchments are adjacent and share similar vegetation and geology (New Zealand Forest Service 1976), which are thought to regulate conductivity. Northern limjts of red beech in the Tararua Ranges occur in the Mangatainoka headwaters (Rogers & McGlone 1994), but red beech is dominant throughout the study area. Mudstone is present in the catchments of some of the lower Mangatainoka study sites (Stevens 1974); however, all sites containing shortjaw kokopu were in entirely greywacke catchments. Therefore, substrate and vegetation are not controlling conductivity. In this case, higher conductivity appears to be an indication of greater ground water influence rather than run off, which is a sign of greater flow stability. Eriopterini occurred in many 2nd order streams but also in large open canopied streams in the survey, contrasting the typical habitat of shortjaw kokopu. Previous shortjaw kokopu diet analysis (Main 1987, McDowall 1996, McDowall et al. 1996b) has found that aquatic diptera larvae make a minor component of shortjaw kokopu diet and thus this may be a reason why absence of Eriopterini predicted shortjaw kokopu presence, although other diptera larva, Aphrophila, were positively associated with shortjaw kokopu presence. 3 Department of Conservation. 29 Chapter 2. Shortjaw kokopu habitat and distribution. The failure to find shortjaw kokopu in the Ruamahanga catchment is difficult to explain. Although shortjaw kokopu are normally found in catchments with river mouths on the western coast of New Zealand (McDowall 1990), recent surveys have located several populations of shortjaw kokopu in streams both east and west of the Ruamahanga River mouth (Rebergen & Joy 1999, McDowall 2000). Therefore, river mouth access does not appear to be restricting shortjaw kokopu from accessing the Ruamahanga River. There are several point sources of pollution affecting the river, particularly urban settlements, the Waingawa freezing works and the International Timber Processors Ltd in Masterton , a floodgate controlling the water level in the lower Ruamahanga River, and a grade control weir at Te Ore Ore. However, these alone do not explain the lack of shortjaw kokopu in the Ruamahanga River. The occun-ence of two other diadromous galaxiid species in the Ruamahanga River of similar mjgrating ability (McDowall 1990) implies that other factors must be influencing shortjaw kokopu numbers. A pollution barrier to shortjaw kokopu migration in the Ruamahanga is a less likely explanation si nce banded kokopu, present in the upper Ruamahanga, are the most sensitive of the migratory galaxi id to turbidity, toxins, and suspended solids (Richardson 1997, West et al. 1997, Richardson et al. 1998, Rowe & Dean 1998, Jowett & Boustead 2000). Similarly, shortjaw kokopu are thought to be as good or better at pass ing waterfalls and manmade structures as banded kokopu (Jowett et al. 1998, Joy et al. 2000), so a physical baiTier in the lower Ruamahanga seems unlikel y. In contrast, banded kokopu were not found in the upper reaches of the Mangatainoka and Makakahi Rivers but they are known from lower altitude tributaries of the Manawatu River (pers. obs., M.K. Joy (Massey University: Palmerston North) pers. comm. February 2000, NZFFD, McDowall et al. 1996a). A reason for thi s may be a series of barrage dams between the Manawatu Gorge, and the Mangatainoka, and Makakahi River headwaters (Anonymous 2001). However, brown trout have been found to pass these barriers, and the appearance of juvenile (55-80 mm) galaxiids, both koaro and shortjaw kokopu (pers. obs.) shows that some migratory galaxiids are able to pass these structures . These barrage dams are less than six years old and given the expected life span of banded kokopu, at least nine years (McDowall 1990), if these structures were preventing migration of banded kokopu we would still expect to see some adults remaining in the upper catchment if it was suitable habitat. McDowall 30 Chapter 2. Shortjaw kokopu habitat and distribution. (1998) found there appeared to be a 'saturation point' for shortjaw kokopu migration. He proposed that the absence of shortjaw kokopu from many inland tributaries of West Coast rivers was due to a lack of recruiting juveniles and that only the closest habitat was being colonized. This theory could apply to shortjaw kokopu in the Ruamahanga River. If so, it would be expected to find shortjaw kokopu in tributaries closer to the sea, such as the Tauherenikau or Waiohine River. Shortjaw kokopu habitat in the northeastern Tararua Ranges is characterized by high stability, large substrate, 3rd-4th order streams, relatively low altitudes (340-430 m), with unmodified native forest canopy cover. Streams not having this combination of features are unlikely to have shortjaw kokopu present. Likewise, invertebrates, particularly the occurrence of Coloburiscus, Zelandobius and Tanytarsus, and the absence of the open stream invertebrates, such as Eriopterini, Aphrophila, Aoteapsyche, and Elmidae are useful predictors of shortjaw kokopu. Shortjaw kokopu were always found co-occurring with longfin eels, occasionally in the presence of brown trout, but more often in areas where brown trout are absent, and other galaxiid species, such as koaro occur. 31 Chapter 2. Shortjaw kokopu habitat and distribution. 2.5 References Anonymous 1999. Fish find gives scientists hope for rare species. Dominion, Wellington, New Zealand. 20-02-1999: 1. Boubee, J. , Forsyth , D., Richardson, J., Barrier, R. & James, M. 1995: 1995 Mangahao Dams Desilting Fish & Invertebrate Survey. NIWA, Hamilton, New Zealand. Survey ELE312-0l/l. Caskey, D. 1999: Shortjawed kokopu (Galaxias postvectis) research and surveys June 1998- June 1999. Department of Conservation, Stratford, New Zealand. Death, R. G. 1995: Spatial patterns in benthic invertebrate community structure: products of habitat stability or are they habitat specific? Freshwater Biology 33: 455-467 . Eldon, G. A. 1983: Enigma of the third kokopu. Freshwater Catch (NZ) 23: 17-18. Jack, D. & Barrier, R. 2000: Shortjawed Kokopu (Galaxias postvectis): Conservation status in Nel son/Marlborough - Year two, Interim Report 2000. Department of Conservation, Nelson, New Zealand. Jowett, I. & Boustead, N. 2000: Nowhere to hide: effects of high sediment loads on instream habitat for fish. Water and Atmosphere 8: 12-13 . Jowett, I. G. , Hayes, J. W., Deans, N. & Eldon, G. A. 1998: Comparison of fish communities and abundance in unmodified streams of Kahurangi National Park with other areas of New Zealand . New Zealand Journal of Marine and Freshwater Research 32: 307-322. Joy, M. K., Henderson, I. M. & Death, R. G. 2000: Diadromy and longitudinal patterns of upstream penetration of freshwater fish in Taranaki , New Zealand. New Zealand Journal of Marine and Freshwater Research 34: 531-534. Main , M. 1987: Feeding habits of the shortjawed kokopu. Freshwater Catch (NZ) 31: 13. McCune, B. & Mefford, M. J. 1995: PC-ORD. Multivariate Analysis of Ecological Data. 2, MjM Software Design Gleneden Beach, Oregon, USA. McDowall, R. M. 1990: New Zealand Freshwater Fishes: A natural history and guide, 2 Ed. Auckland, New Zealand. Heinemann Reed, 553 pp. McDowall , R. M. 1996: Supping from the surface and grazing the gravel: dietary habits of the shortj awed kokopu. Water and Atmosphere 4: 14. McDowall, R. M . 1997: Indigenous vegetation type and the distribution of the shortjawed kokopu , Galaxias postvectis (Teleostei: Galaxiidae) , in New Zealand. New Zealand Journal of Zoology 24: 243-255 . 32 Chapter 2. Shortjaw kokopu habitat and distribution. McDowall, R. M. 1998: Fighting the flow: downstream-upstream linkages in the ecology of diadromous fish faunas in West Coast New Zealand Rivers. Freshwater Biology 40: 111-122. McDowall, R. M. 2000: The Reed Field Guide to New Zealand Freshwater Fish. Auckland, New Zealand. Reed Publishing (NZ) Ltd, 224 pp. McDowall, R. M., Eldon, G. A., Bonnett, M. L. & Sykes, J. R. E. 1996a: Critical habitat for the conservation of shortjawed kokopu, Galaxias postvectis Clarke. Wellington. Department of Conservation, New Zealand. 80 pp. McDowall, R. M., Main, M. R., West, D. W. & Lyon, G. L. 1996b: Terrestrial and benthic foods in the diet of the shortjawed kokopu, Galaxias postvectis Clarke (Teleostei: Galaxiidae). New Zealand Journal of Marine and Freshwater Research 30: 257-269. McDowall, R. M. & Richardson, J. 1983: The New Zealand freshwater fish survey: a guide to input and output. New Zealand Ministry of Agriculture and Fisheries, 15 pp. Molloy, J. & Davis, A. 1994: Setting priorities for the conservation of New Zealand's threatened plants and animals, 2 Ed. Department of Conservation, Wellington, New Zealand 64 pp. New Zealand Forest Service 1976: Tararua State Forest Park, Wellington, New Zealand. A.R. Shearer, Government Printer, 28 pp. Nicoll, G. 1984: Freshwater life surveyed in South Westland. Freshwater Catch (NZ) 24: 4-6. Pfankuch, D. J. 1975: Stream reach inventory and channel stability evaluation. United States Department of Agriculture Forest Service, Region 1, Missoula Montana, USA. Rebergen, A. & Joy, M. 1999: Freshwater fish survey Aorangi Range, Wairarapa. Department of Conservation, Masterton, New Zealand. Richardson, J. 1997: Acute ammonia toxicity for eight New Zealand indigenous freshwater species. New Zealand Journal of Marine and Freshwater Research 31: 185-190. Richardson, J., Boubee, J., Dean, T., Hicks, M., Rowe, D. & West, D. 1998: Effects of suspended solids on migratory native fish. Water and Atmosphere 6: 22-23. Rogers, G. M. & McGlone, M. S. 1994: A history of Kaiparoro clearing and the limits of Nothofagus in the northern Tararua Ranges, New Zealand. New Zealand Journal of Botany 32: 463-482. 33 Chapter 2. Shortjaw kokopu habitat and distribution. Rowe, D. K. & Dean, T. L. 1998: Effects of turbidity on the feeding ability of the juvenile migrant stage of six New Zealand freshwater fish species. New Zealand Journal of Marine and Freshwater Research 32: 21-29. SAS 2000: SAS User's Guide: Statistics, Version 8, SAS Institute Inc Cary, North Carolina, USA. New Zealand Forest Service 1976: Tararua State Forest Park, Wellington, New Zealand. A.R. Shearer, Government Printer, 28 pp. Stevens, G. R. 1974: A tramper's geology of the Tararuas: A New Zealand geological survey handbook, Wellington, New Zealand. A.R. Shearer, Government Printer, 44pp. Strahler, A. N. 1952: Hypsometric (area-altitude) analysis of erosional topography. Bulletin of the Geological Society of America 63: 1117-1142. Studholme, B., Barrier, R. & Jack, D. 1999: Shortjawed Kokopu (Galaxias postvectis): Conservation status in Nelson/Marlborough - Year one, Interim Report 1999. Department of Conservation, Nelson, New Zealand. West, D. W., Boubee, J. A. T. & Barrier, R. F. G. 1997: Responses to pH of nine fishes and one shrimp native to New Zealand freshwaters. New Zealand Journal of Marine and Freshwater Research 31: 461-468. Winterbourn, M. J. & Gregson, K. L. D. 1989: Guide to the aquatic insects of New Zealand, 2 Ed. Auckland, New Zealand. Entomological society of New Zealand, 96pp. Wolman, M. G. 1954: A method of sampling coarse riverbed material. Transactions of the American Geophysical Union 35: 951-956. 34 3. Abstract Seasonal activity of shortjaw kokopu ( Galaxias postvectis) in the north_ern Tararua Ranges Freshwater fish communities were surveyed monthly at three sites over a 16 months period· in the Mangatainoka River and a tributary in the northeastern Tararua Ranges. Shortjaw kokopu (Galaxias postvectis) and longfin eels (Anguilla dieffenbachii) occurred in all sites, koaro (G. brevipinnis) in the two upper sites, brown trout (Salmo trutta) and Cran's bully (Gobiomorphus basalis) in the lower site. The number of shortjaw kokopu observed varied between months with greatly reduced numbers during winter and rising to a maximum in autumn. The variation in observed numbers is considered to be due to changes in activity rather than the seasonal movement of shortjaw kokopu within the catchment as the monthly pattern of numbers observed are the same at each site. Shortjaw kokopu ?id not show seasonal movement, instead having lower activity during colder months. Key Words: Shortjaw kokopu, Galaxias postvectis, seasonal activity, Tararua Ranges, ' New Zealand. Chapter 3. Seasonal activity of shortjaw kokopu . 3.1 Introduction Shortjaw kokopu are the rarest of the five species of the diadromous galaxiidae in New Zealand (McDowall 1996), occurring in 2% of NZFFD 1 records (McDowall et al. 1996a). Shortjaw kokopu are normally found in unmodified, predomjnantly podocarp, forest streams (Eldon 1983, Main 1987, McDowall 1990, 1997). They are most commonly found in pools and runs (Eldon 1983, McDowall 1990, McDowall et al. 1996a), often above large falls; and areas where large boulder and cobble substrate dominate (Nicoll 1984, McDowall 1990, McDowall 1996). Shortjaw kokopu are often considered a solitary species, found in small numbers, with only one or two individuals in any given reach . This has contributed to their category A classification (Molloy & Davis 1994) in the endangered species lists , along with their wide ranging (McDowall 1990, McDowall et al. 1996a, McDowall 2000) but sporadic distribution (McDowall 1996). Surveys of shortjaw kokopu populations are often conducted during warmer, summer periods, avoiding the inconveniences of winter (e.g. Caskey 1999, Studholme et al. 1999, Jack & Barrier 2000). However, these are generally qualitative surveys of occurrence at one time. Repeat surveys for most galaxiids, including shortjaw kokopu , are not often reported . Only Caskey (1999) has studied seasonality in shortjaw kokopu populations, focusing on growth rates over the year, particularly comparing winter and summer growth rates. Unlike some salmonid species that have been found to have either a diurnal or nocturnal activity pattern, dependent on both water temperature and photoperiodity (Fraser et al. 1995, Whalen et al. 1999), shortjaw kokopu are generally only nocturnally active (McDowall 1990). A seasonal pattern of nocturnal activity, with lowest activity in winter, has been documented in the Canterbury galaxias (Galaxias vulgaris) (Cadwallader 1975b) and shortjaw kokopu are expected to show a simi lar pattern (R.F.G Barrier (DoC2 : Wellington) pers. comm. October 2001). In this survey, I examine the effect that season has on observed numbers of shortjaw kokopu. By comparing seasonal patterns in three sites encompassing the extremes of 1 New Zealand Freshwater Fish Database, maintained by the National Institute of Water and Atmosphere (NTW A; McDowall & Richardson 1983) 36 Chapter 3. Seasonal activity of shortjaw kokopu. shortj aw kokopu range in the Mangatainoka catchment, it will be possible to distinguish changes in activity from seasonal migration within the catchment as causes of any variation. 3.2 Methods Surveys were carried out on three reaches of nver rn the headwaters of the Mangatainoka Ri ver, and one of its tributaries (Figure 1). The three sites were selected to encompass the distribution of shortjaw kokopu in the Mangatainoka River (Chapter 2) . All three sites had predominantly red beech canopy, although the lower site was where the river left the cover of the native bush. Sites differed in stream size, with the upper site including the confluence of two 3rd order stream (Table 1). The lower site was 1300 m below the middle site, and the upper site a further 300 m upper from the middle site . At all sites the general flow pattern of the river was of a pool-riffle-run sequence, although fa ll s were also present. Substrate was mainly large boulders (~256 mm) and cobbles (128-256 mm). Table 1. Characteristics of the three seasonal survey sites in the northeastern Tararua Ran es. Site Stream Altitude Gradient Number of pools and order (m) (m/200 m) runs Upper 3r &4 453 25 29 Middle 4th 401 13 19 Lower 5th 344 12 17 At each site a 200 m reach was searched for approximately 90 minutes allowing two upstream and two downstream traverses. In each sampling period, two observers with 30 watt spotlights, powered from 12 volt 7 amp hour batteries were used. All pools and runs were coded, so that the location of the fish could be recorded from month to month. Where possible, size was also recorded to identify individual fish. Size was treated as an 2 Department of Conservation . 37 Chapter 3. Seasonal activity of shortjaw kokopu. indication of age class (Studholme et al. 1999). During April and May, captured fish were also assessed for their reproductive status using the definitions in Charteris (2002). Surveys were carried out under low flow conditions to maintain consistent visibility in the water. From the fifth monthly survey onwards, water temperature was recorded at the conclusion of each survey. KEY [i]Tararua Forest Park D Native Forest - - ~3rd0rder streams - 2nd Order streams 0 Sites 0 2km I I I I I I I I I I N + Figure 1. Location of sites used in a seasonal survey of shortjaw kokopu activity in the northeastern Tararua Ranges. Shortjaw kokopu, koaro, longfin eel, brown trout, Cran's bully and torrentfish have all been recorded in the Mangatainoka River (Chapters 2 & 4). Any juvenile bullies were treated as Cran's bully because that is the only species positively identified from the Mangatainoka River headwaters (Chapter 2, NZFFD, M.K. Joy (Massey University: Palmerston North) pers. comm. February 2000). 1 '., 11_ :..- The data was grouped into seasons of three months a posterio1~~.llowing months to be sorted based on shortjaw kokopu activity. The first two months of survey, August and September 2000, were treated as a trial period, with the analysis of seasons based on subsequent seasonal data. The division of 14 monthly surveys into seasons meant spring and summer had four replicates, while autumn and winter only had three. 38 Chapter 3. Seasonal activity of shortjaw kokopu. 3.2.1 Statistical analysis A two-way analysis of variance (ANOV A) was used to compare shortjaw kokopu numbers among sites and seasons. Posthoc pairwise comparisons of the main effects were performed using Bonferroni tests. Fishers Least Squared Difference (LSD) tests were used to assess between season differences. All statistical analyses were carried out using (SAS 2000). 3.3 Results The largest population of shortjaw kokopu was found at the middle site, with 27 shortjaw kokopu being counted in a single sampling period (February 2001). Brown trout were not present at this site (Table 2), however, koaro were found in small numbers. The upper site had the least number of shortjaw kokopu with a maximum of 10 observed in one sampling period (December 2001). Brown trout were also absent from this site, but koaro were much more abundant with 16 koaro found during one sampling period (April 2001). The lower site had large numbers of shortjaw kokopu, with 19 being observed on two occasions (January 2001 & April 2001). Shortjaw kokopu commonly co-occurred with brown trout in the lower site with a maximum of six being observed in one sampling period (February 2001). However, koaro were absent from the lower site. Cran's bully were only present at the lower site, with a maximum of six counted during one sampling period (April 2001). Longfin eels were present at all sites. Although generally in low numbers, 15 were counted in one sampling period at the upper site (February 2001), a maximum of 10 at the middle site (February 2001), and 10 at the lower site (October 2001). Table 2. Fish species observed at the seasonal survey sites in the northeastern Tararua Ranges between August (2000) and November (2000). Site Shortjaw Kokopu Koaro Longfin eels Brown Trout Gran's Bully Upper ✓ ✓ ✓ Middle ✓ ✓ ✓ Lower ✓ ✓ ✓ ✓ 39 Chapter 3. Seasonal activity of shortjaw kokopu. 3.3.1 Seasonal activity As there was a marked decline in the number of shortjaw kokopu observed in May, this was designated as the start of winter. Seasonal changes in observed numbers followed a similar pattern in all five species of fish and at all three sites (Figure 2). Numbers were highest during summer and autumn but dropped during winter, following water temperature trends. Generally, the observed presence of the different fish species followed the water temperature changes, although time lags occurred between the changes in water temperature and the activity response of different fish species. Longfin eels appeared to most closely followed the temperature changes, with shortjaw kokopu and koaro both exhibiting slightly delayed responses to temperature change. During the October 2001 survey, three unidentified juvenile bully were found in the lower reach. 3.3.2 Shortjaw kokopu Seasonal variation in the number of shortjaw kokopu observed was similar at the three sites (Figure 3). The highest numbers of shortjaw kokopu were observed in autumn or summer, and the lowest in winter (Figure 4). Shortjaw kokopu were found in large congregations (~4 fish) in five pools and runs (Table 3) on 10 different survey trips (Table 4). No pool or run was observed to contain shortjaw kokopu in all 16 survey occasions, but most had shortjaw kokopu recorded on more than three occasions (Table 4). Table 3. Distribution of pool occupancy by shortjaw kokopu over 14 months in the Mangatainoka catchment. Minimum Maximum Mode 0 67 18 62 1 21 4 No. of shortjaw kokopu per pool 2 3 4 14 1 9 5 0 6 4 40 30 28 6 28 "-- 24 ., 3 22 e 20 f 18 ,! 16 'O a 14 j 12 "' Q 10 1i 8 E 6 :, z 4 2 30 28 6 28 ~24 .a 22 ; 20 ~18 ,! 16 'O fa 14 i 12 O 10 i : Z 4 2 a. A ()a (;)<:> <:::,() li:l() c:::,<:> C)" a" ()°' c:>" Q" a" a" c::," ()" ()°' a" 't-,§Y 'b0q" o(;-"~04."Q0°' •Jb-<::-", ~~,, ~v~~,, ").;:,<::-,, ')~,, 't-.:f>/ 'b"f~" o?Y ~o~" Date C. 0 .o: • ··o· · ·o·' ()() <::>o ()() ()a (:)a C)" Q°' t:1" <:>°' c" a" ()" a" a' o" a" 't-vO/ a:,q,<:V 0C-"~o4-"Q-<::-''<.~,, ~-t-" ~,,~.,,~,, ':Jv<:--" ')~,, 't-,~ff/t::Jl'q,, O(;-'~o4." Date Chapter 3. Seasonal activity of shortjaw kokopu. 30 28 628 "-- 24 ~ 22 I!! 20 f 18 ~ 16 .,, a 14 -fii 12 "' 0 10 1i 8 E 6 :, z 4 2 0 b. ·.ti .A t·: .-•0• •./J,•·· ••·e;·' V ,, .. 0 .0·· ··o:,·:o·· .. ~~~~~~~~~~~~~~~~ 't--l~' 'br,S~" &'" ~tJ-" <::Jbt>,, ").,,<::-"«~,,~~,, ~._,,~~,, )v<::-' ':i.~,, 'I-.:§¥ G:J0q" 0¢-,, ·~·f" • ........ 6, ....... ······O········ . ·V····· ·······•······· Date Key Shortjaw kokopu Koaro Longtin eel Brown trout Gran's bully Temperature Figure 2. Number of fish observed and the temperature recorded monthly over a 16 month period at a. upper site; b. middle site; and c. lower site. Table 4. Monthly distribution of pool occupancy by shortjaw kokopu in the Mangatainoka catchment. No. of shortjaw kokopu In pool Total No. () 1 2 3 4 5 6 seen August 2000 62 5 5 September 2000 63 4 4 October 2000 50 12 4 1 23 November 2000 54 6 6 1 21 December 2000 41 14 8 3 1 45 January 2001 44 13 6 1 3 47 February 2001 38 19 5 3 1 1 47 March 2001 43 15 6 1 1 1 40 April 2001 37 20 4 5 1 48 May 2001 56 10 1 12 June 2001 51 15 1 17 July 2001 52 13 1 1 22 August 2001 56 9 1 1 15 September 2001 51 9 5 1 1 26 October 2001 50 7 7 1 2 32 November 2001 49 7 6 3 1 1 37 41 Chapter 3. Seasonal activity of shortjaw kokopu. Season 30 ,------,=---.-,=---,------=:----t-----,----,----~----~-- Tr. Sp. Su. Au. ::, 25 a. j j 20 ~ :e-_g 15 Ill 0 ... ! 10 E ::, z 5 -e--- Upstream · · -o- · · Midstream --v- Downstream _GJ. • ····d Wi. Sp. Su. • ···P .· / -~ /'7 _:y/' I . ./ 'P-✓ / ti ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~/ ~/ 0-/ ~/ c,/ ~/ :Q/ ~/ ,_<.,/ ;:\/ ~/ ~/ ~/ ~/ c}/ ~/ ~ ~ 0 ~ ~ ~ ~ ~ ~ , ~ 1 ~ ~ 0 ~ Date Figure 3. Monthly changes in the number of shortjaw kokopu observed at three survey sites in the Mangatainoka catchment. Seasons are Tr. trial; Sp. spring; Su. summer; Au. autumn; and Wi. winter. 30 28 26 ::::, 24 §- 22 ~ ~ 20 ~ 18 t'15 0 ~ 14 -0 12 ... Q) 10 .0 E 8 ::::, z 6 4 2 0 --e-- Upper site · · -0 · · Middle site ~ Lower site Spring IJ Summer Autumn Winter Season Figure 4. Seasonal variation in mean number of shortjaw kokopu observed per survey at three sites in the Mangatainoka catchment. 42 Chapter 3. Seasonal activity of shortjaw kokopu. 3.3.3 Analysis of shortjaw kokopu activity Shortjaw kokopu numbers differed between sites (F2,30 = 30.7, P < 0.001). The upper site had lower numbers than the middle and lower sites (Bonferroni, P :s; 0.001, 0.010). However, there was no significant difference between middle and lower sites there was no significant difference (BonfeIToni, P = 0.118). Shortjaw kokopu numbers also differed between seasons (F3,30 = 14.59, P < 0.001), although between summer and autumn , and winter and spring there was no significant difference (Table 5). There was no significant interaction between season and site (F6,30 = 1.71 , P = 0.153). Table 5. Statistical significance between seasons, significant differences Fishers LSD test, (P :s; 0.05) are shown in bold. found by a Spring Summer Autumn Summer Autumn Winter 3.4 Discussion 0.059 0.006 0.403 0.264 0.012 0.001 There has been much research into the migration and diurnal activity patterns of freshwater fi sh species, particularly the anadromous salmonid species of the northern hemisphere. Temperature, photoperiodity and spate occurrence have often been identified as the principle determinants of salmonid migration and diurnal ac tivity (e.g. Whalen et al. 1999, Fraser et al. 1995 , Young 1998). Seasonal activity of galaxiidae species is less studied, with most work being carried out on non-diadromous species, particularly Galaxias vulgaris (e .g. Cadwallader 1975a, Cadwallader 1975b). All of the sites had reduced shortjaw kokopu activity during winter. Seasonal migration would have implied one of the sites had greater shortjaw kokopu activity during the winter and a significant site-season interaction would have been found. As this was not observed, seasonal migration seems unlikely . The same shortjaw kokopu were also found in the same pools when activity levels increased again. So, shortjaw kokopu must exhibit a limited activity pattern during winter. They may bury themselves in substrate, requiring less effort to hold position during flooding events, or remain in their daytime refuge, allowing a limited feeding period during part of the night to retain condition. As some shortjaw kokopu are found year round , it seems likely that they remain in daytime 43 Chapter 3. Seasonal activity of shortjaw kokopu. refuge; however, immediately following a flood, more shortjaw kokopu are active (pers. obs.). All of the species present in the Mangatainoka River and tributary, except Cran's bully, showed similar activity patterns over the 16 month studied. The reductions in activity all occurred around May. For the galaxiid species , this is the estimated spawning time (McDowall 1990), with spawning activity observed in Taranaki populations of shortjaw kokopu (S.C. Charteris (Massey University: Palmerston North) pers. comm. June 2001), which would explain the continued high activity levels through until May, beyond the period when temperature changes most rapidly. Longfin eels, however, are catadromous, spawning at sea (McDowall 1990), so do not need to maintain activity until a spawning period. Cran's bully showed similar activity throughout the year, although an increase occurred in early October when some juvenile bullies were found. During late autumn (April 2001), most large shortjaw kokopu were found in a ripe state, but by the start of winter (May 2001), most were spent. This is consistent with the paucity of adult shortjaw kokopu observed in May (2001), when only sub-adult fish were found. The larger fish may have been recovering from the energy expenditure associated with spawning. Koaro were only found above the range of brown trout, but did not become a dominant species in the fish community until shortjaw kokopu started to become fewer in number near the top of their range. Where shortjaw kokopu were less dominant, koaro utilized the pools and runs much more. Chadderton & Allibone (2000) have also described koaro as riffle dwellers while other fish species were present, but utilizing all habitat types when the other fish species reached their upstream distribution limits, which they called a "community controlled" effect. Brown trout were not observed at the lower site until January 2001. A cause of this may have been a large flood during October 2000. This flood affected the Mangatainoka headwaters, reconstructing many of the headwater tributaries by moving around boulders and large logs, replacing some pools with riffles. This occurred at all three survey sites. The lower site, however, had been worse hit, due to the much greater water 44 Chapter 3. Seasonal activity of shortjaw kokopu. flow, and this may have displaced brown trout from the fish community, requiring many months to return to pre-flood levels. Shortjaw kokopu were never active rn any pool or run for all 16 monthly surveys. However, eight pools and runs had shortjaw kokopu activity for more than 10 monthly surveys, and 15 for more than eight monthly surveys. The May (spawning) period may account for the greater shortjaw kokopu observance in some pools and runs, but not others (S.C. Charteris (Massey University: Palmerston north) pers. comm. June 2001). On this occasion, mostly sub-adult fish were observed in this survey, but not many of the large fish that were present in earlier months. Generally, these same fish 'reappeared' in the same pools in later months. However some did not, either moving into new, unrecorded habitats, or possibly dying. Of these, two were greater than 240 mm long and most were greater than 210 mm. Grouping the monthly data by seasons was useful to provide 'replication' for the statistical analysis but it may obscure some of the seasonal patterns. Only one night was surveyed each month, so even with ungrouped data, sudden changes in activity will be observed. This is particularly important in May. Although classified as winter due to the surveyed activity levels, May is the main month of spawning for shortjaw kokopu (S.C. Charteris (Massey University: Palmerston North) pers. comm. June 2001). The survey occurred late in the month, following the spates associated with spawning in the Mangatainoka catchment. At this time, the activity levels were low, but had the survey been carried out in early May, the activity levels may have been as high as in April. In conclusion, shortjaw kokopu have greatest activity during summer and autumn, and least during winter. While there was a difference between sites, this was consistent for the entire study, showing that shortjaw kokopu have a reduced activity period during winter, and don't show a seasonal movement pattern. The reduced activity in shortjaw kokopu appears to commence following spawning, particularly in the adult fish. 45 Chapter 3. Seasonal activity of shortjaw kokopu . 3.5 References Cadwallader, P. L. 1975a: Feeding habits of two fish species in relation to invertebrate drift in a New Zealand river. New Zealand Journal of Marine and Freshwater Research 9: 11 -26. Cadwallader, P. L. 1975b: Spontaneous locomotory act1v1ty of Galaxias vulgaris Stokell (Pisces: Salmoniformes). New Zealand Journal of Marine and Freshwater Research 9: 27-34. Caskey , D. 1999: Shortjawed kokopu (Galaxias postvectis) research and surveys June 1998- June 1999. Department of Conservation , Stratford, New Zealand. Chadderton, W. L. & Allibone, R. M . 2000: Habitat use and longitudinal di stribution patterns of nati ve fish from a near pristine Stewart Island, New Zealand, stream. New Zealand Journal of Marine and Freshwater Research 34: 487-499. Charteris, S. C. 2002: Spawning, egg development and recruitment of diadromous galaxiids in Taranaki New Zealand. Unpublished MSc Thesis, Massey University, Palmerston North, New Zealand. Eldon, G. A. 1983: Enigma of the third kokopu. Freshwater Catch (NZ) 23: 17-18. Fraser, N. H. C ., Heggenes, J ., Metcalfe, N. B. & Thorpe, J. E. 1995: Low summer temperatures cause juvenile Atlantic salmon to become nocturnal. Canadian Journal of Zoology 73: 446-451 . Jack, D. & Barrier, R. 2000: Shortjawed Kokopu (Galaxias postvecti s): Conservation status in Nelson/Marlborough - Year two, Interim Report 2000. Department of Conservation, Nelson, New Zealand . Main , M. 1987: Feeding habits of the shortj awed kokopu. Freshwater Catch (NZ) 31: 13. McDowall, R. M. 1990: New Zealand Freshwater Fi shes: A natural hi story and guide, 2 Ed. Auckland, New Zealand. Heinemann Reed, 553 pp. McDowall, R. M. 1996: Supping from the surface and grazing the gravel: dietary habits of the shortjawed kokopu. Water and Atmosphere 4: 14. McDowall , R. M. 1997: Indigenous vegetation type and the di stribution of the shortj awed kokopu , Galaxias postvectis (Teleostei : Galax iidae), in New Zealand. New Zealand Journal of Zoology 24: 243-255. McDowall , R. M . 2000: The Reed Field Guide to New Zealand Freshwater Fish. Auckland, New Zealand. Reed Publishing (NZ) Ltd, 224 pp. 46 Chapter 3. Seasonal activity of shortjaw kokopu. McDowall, R. M., Eldon, G. A., Bonnett, M. L. & Sykes, J. R. E. 1996a: Critical habitat for the conservation of shortjawed kokopu, Galaxias postvectis Clarke. Wellington, New Zealand. Department of Conservation, 80 pp. McDowall, R. M. & Richardson, J. 1983: The New Zealand freshwater fish survey: a guide to input and output. New Zealand Ministry of Agriculture and Fisheries, 15 pp. Molloy, J. & Davis, A. 1994: Setting priorities for the conservation of New Zealand's threatened plants and animals, 2 Ed. Wellington, New Zealand. Department of Conservation, 64 pp. Nicoll, G. 1984: Freshwater life surveyed in South Westland. Freshwater Catch (NZ) 24: 4-6. SAS 2000: SAS user's guide: statistics. Version 8, SAS Institute Inc Cary, North Carolina, USA. Studholme, B., Barrier, R. & Jack, D. 1999: Shortjawed kokopu (Galaxias postvectis): Conservation status in Nelson/Marlborough - Year one, Interim Report 1999. Department of Conservation, Nelson, New Zealand. Whalen, K. G., Parrish, D. L. & McCormick, S. D. 1999: Migration timing of Atlantic salmon smolts relative to environmental and physiological factors. Transactions of the American Fisheries Society 128: 289-301. Young, M. K. 1998: Absence of autumnal changes in habitat use and location of adult Colorado River cutthroat trout in a small stream. Transactions of the American Fisheries Society 127: 147-151. 47 4. Abstract Evaluation of three methods for surveying shortjaw kokopu ( Galaxias postvectis) in the northern Tararua Ranges Freshwater fi sh communities were surveyed at six sites in the Mangatainoka catchment in the northeastern Tararua ranges. At each site, spotlighting, electrofishing, and gee­ minnow trapping were tested for their effectiveness in locating shortjaw kokopu (Galaxias postvectis). Eleven shortjaw kokopu were caught by spotlighting, three by electrofishin·g, but none with gee-minnow traps. Also caught were brown trout (Salmo trutta), longfin eels (Anguilla dieffenbachii), Cran's bully (Gobiomorphus basalis), torrentfi sh (Cheimarrichthys fosteri) and koura (Paranephrops planifrons). There was Q no significant difference in the number of fish observed per sit by spotlighting and electrofishing. v Key Words: Shortjaw kokopu, Galaxias postvectis, electrofishing, spotlighting, gee­ minnow traps, Tararua Ranges, New Zealand. Chapter 4. Methods for surveying shortjaw kokopu. 4.1 Introduction Shortjaw kokopu (Galaxias postvectis) are considered rare, occurring in less than 2% of NZFFD1 records (McDowall et al. 1996b). Shortjaw kokopu are diadromous and found throughout most of New Zealand, yet have a Category A threatened species status (Molloy & Davis 1994 ). This is because of their sporadic distribution (mostly ~3 fish at any location; McDowall 1996) and their supposed habitat needs; streams with boulders and pools in podocarp dominated native forest (Main 1987, McDowall 1990, 2000, McDowall et al. 1996a) A possible factor contributing to the apparent rarity of shortjaw kokopu is that people may be looking in the wr