Trap spacing and layout: experiments in stoat control in the Dart Valley,

Transcription

1 Trap spacing and layout: experiments in stoat control in te Dart Valley, SCIENCE FOR CONSERVATION 118 B.L. Lawrence and C.F.J. O Donnell Publised by Department of Conservation P.O. Box Wellington, New Zealand

2 Science for Conservation presents te results of investigations by DOC staff, and by contracted science providers outside te Department of Conservation. Publications in tis series are internally and externally peer reviewed. July 1999, Department of Conservation ISSN ISBN Tis publication originated from work done under Department of Conservation Investigation no. 1936, carried out by B. L. Lawrence and C. F. J. O Donnell, Science & Researc Unit, Department of Conservation, Private Bag, Cristcurc. It was approved for publication by te Manager, Science & Researc Unit, Science Tecnology and Information Services, Department of Conservation, Wellington. Cataloguing-in-Publication data Lawrence, B. L. Trap spacing and layout : experiments in stoat control in te Dart Valley, / B.L. Lawrence and C.F.J. O'Donnell. Wellington, N.Z. : Dept. of Conservation, v. ; 30 cm. (Science for conservation, ; 118.) Cataloguing-in-Publication data. Includes bibliograpical references. ISBN Mustela erminea--control--new Zealand. 2. Trapping--New Zealand. I. O'Donnell, C. F. J. (Colin F. J.), II. Title. III. Series: Science for conservation (Wellington, N.Z.) ; zbn

3 CONTENTS Abstract 5 1. Introduction 5 2. Study area 6 3. Metods Trapping Tunnels Tunnel Configuration 8 Trial 1 8 Trial Analysis 8 4. Results Trial Trial Distribution of catc Timing of captures 9 5. Discussion Recommendations Acknowledgments References 13

4 Abstract Control of stoat (Mustela erminea) populations in soutern beec (Notofagus spp.) forests is required to maintain te viability of treatened bird populations. Intensive stoat control using a grid of traps, wit trap spacing of 100 m (grid trapping), increases te breeding success of moua (yellowead: Mooua ocrocepala), but is expensive. An alternative is to set traps only around te perimeter of an area to be protected (perimeter trapping). We tested weter perimeter trapping could be as effective at reducing stoat numbers as is intensive grid trapping. During trials we obtained similar catc rates on perimeters as on intensive grids. Assuming tat stoat populations were similar in eac study area, we conclude tat perimeter trapping around a 100 a (1 km 2 ) block can be as effective as intensive grid trapping wit 100 m trap spacing. Trials were undertaken wile stoat densities were low, and te implications of tis are discussed. 1. Introduction In New Zealand beec forests (Notofagus spp.), irregular but severe predation by stoats (Mustela erminea) as been implicated in te decline of several indigenous bird species, in particular moua (yellowead: Mooua ocrocepala) (Elliott 1996a; O Donnell 1996; O Donnell et al. 1996). Development of metods to control stoats, and tus prevent furter decline in tese bird species, as occurred over te last two decades. Te present study is part of tis ongoing researc. King (1980, 1981), King & Edgar (1977) and King & McMillan (1982) reviewed and tested metods for controlling stoats. King (1984) concluded tat stoat control is probably only wortwile in specific situations, namely during te nesting season and only for te protection of a few endangered bird species tat are most adversely affected by stoat predation. O Donnell et al. (1996) evaluated te effectiveness of suc control for moua populations in te Eglinton Valley using mark IV Fenn traps to trap stoats. Nesting success in a 500 m 900 m area containing a grid of trapping tunnels at 100 m intervals was compared to tat in a similar non-treatment area 1 km away. Nesting success in te trapped area was 80% compared wit 36% success in te area not trapped. During te same period, lines rater tan grids of trapping tunnels were operated in te Dart, Routeburn and Hawdon Valleys. Te moua populations witin 600 m of tese lines collapsed by 60% in consecutive spring territory censuses (B. Lawrence, C. O Donnell unpublised data). From tis we concluded tat protection of moua by lines of traps was unlikely to be successful. In te Eglinton Valley, significantly more stoats were trapped on te perimeter tan in te interior of te intensive trapping grid (Dilks et al. 1996). If a perimeter of traps at 100 m spacings is as effective in increasing moua breeding success as a grid, te cost of a trapping operation could be reduced by over 50% bot in terms of traps and servicing time. During we tested weter perimeters of traps could be equally as effective at reducing stoat numbers as intensive grids in te Dart Valley.. 5

5 2. Study area Te Dart Valley study area contained two sites, Lake Sylvan and Millflat, wic were located on opposite sides of te Dart River at te ead of te Lake Wakatipu catcment, partly in te Mt. Aspiring National Park ( 'S, 'E, Figure 1). Te glacial valley is steep sided, wit a large braided river, its floor being 3 5 km wide at tis point. Te two study sites were 2.4 km apart at te nearest point. Figure 1. Location of study sites in te Dart Valley, Sout Island, New Zealand. New Zealand Nort Island 35 Dart River Kinloc River Rees Glenorcy 40 S Queenstown Sout Island 45 Lake Wakatipu E 180 Millflat Study Area Lake Sylvan r R i v e Glenorcy - Paradise Road Sylvan Study Area r t D a R o u t e B u r n Routeburn - Kinloc Road 0 1 km Diamond Lake 6

6 Bot sites are m a.s.l. and ave similar forest dominated by red beec (Notofagus fucsa) wit a canopy m tall. Tere are significant stands of silver beec (N. menziesii) and mountain beec (Notofagus solandri var. cliffortioides) were soils are tin over glaciated rock knobs. Te forest as been cut over in part, in bot locations. Te understorey is open, wit little plant species variation in areas of beec regeneration subsequent to logging. Te climate is wet, (2000 mm annual rainfall at Routeburn), but temperatures are moderated by te strong nort-west airflows of te Dart Valley (lowest canopy temperature during te winter of 1996 was 6.8 C). 3. Metods 3.1 TRAPPING TUNNELS Trapping tunnels were 600 mm long, made of roug sawn timber and ad a square cross section of mm (King et al. 1994). A wooden bar and two wire bars were placed orizontally over eac end to prevent non-target species entering te traps. Two mark IV Fenn traps were set in eac tunnel wit two ens eggs as bait between tem. Trial 1 A B Trial 2 C D Figure 2. Trapping tunnel configurations in trials 1 and 2 ( denote tunnels at 100 m centres). A, m perimeter of 28 tunnels Millflat 1992/33; Sylvan 1993/94. B, 500 m 900 m grid of 60 traps Sylvan 1992/93; Millflat 1993/94. C, m perimeter of 38 tunnels Millflat 1994/95. D, m perimeter and median of 46 tunnels Sylvan 1994/95.. 7

7 3.2 TUNNEL CONFIGURATION Te study consisted of two trials. Te first trial compared te efficacy of perimeter trapping at killing stoats wit tat of grid trapping. Te second trial was to test weter a perimeter enclosing 90 a was as effective at killing stoats as te perimeters enclosing 45 a used in te first trial (Figure 2). Trial 1 Te study sites were bot 500 m 900 m. In te 1992/93 season, te Lake Sylvan site was set out in a grid trapping format wit tunnels at 100 m intervals (Figure 2), 60 tunnels of traps in total. At Millflat, te site was laid out in a perimeter trapping format, wit a total of 28 tunnels of traps at 100 m intervals (Figure 2). In 1993/94 tis configuration was reversed: te grid at Millflat and te perimeter at Lake Sylvan. Trial 2 In 1994/95 bot sites were enlarged to measure 900 m 1000 m. Millflat was operated as a perimeter of 38 tunnels, wile Lake Sylvan was a perimeter wit a median line of 8 extra tunnels, 46 in total, effectively providing two 45 a perimeters for comparision (Figure 2). In all years te traps were set from te last week in November until te tird week in February, a period of 12 weeks eac season. Traps were cecked weekly. 3.3 ANALYSIS As low numbers of stoats were caugt eac year, te probability of occurrence of stoat kills was small, and a Poisson distribution was used to generate expected values for random location of kills. Fiser s exact test was used to test if kill rates on perimeters were different from internal kill rates. 4. Results 4.1 TRIAL 1 Te first year bot sites trapped caugt te same number of stoats (seven stoats eac). Te second year te perimeter site caugt 11 stoats, te grid site caugt 10. In te grid, te kill rate on te periperal line of tunnels was compared to tat of te interior positions eac year. In year 1, four of te seven captures were on te periperal line of tunnels (te difference periperal tunnels to all interior tunnels was not significant, P = 0.695: Table 1). In year 2, two of te ten captures were on te periperal line of tunnels. In bot years te central 12 traps caugt fewer stoats tan te outer two lines of traps, being zero to seven and two to eigt, respectively. Tis difference was not significant (P = ). 8

8 TABLE 1. THE NUMBER OF KILLS AT PERIPHERAL TUNNELS COMPARED TO NUMBERS OF KILLS AT INTERNAL TUNNELS IN TWO GRID TRAPPING OPERATIONS (TRIAL 1). CENTRE REFERS TO THE 12 TUNNELS AT THE CENTRE OF THE GRID NUMBER OF KILLS NUMBER OF TRAPS FISHER S EXACT TEST 1992/93 Sylvan perimeter 4 28 P = /93 Sylvan internal 3 32 NS 1993/94 Millflat perimeter 2 28 P = /94 Millflat Internal 8 32 NS Outside 2 rows, 1992/93 and 1993/94 Centre 12 traps, 1992/93 and 1993/ P = NS TABLE 2: Χ 2 TEST FOR GOODNESS OF FIT SHOWING THE CLOSE DISTRIBUTION OF KILLS WE WOULD EXPECT FROM A RANDOM POISSON DISTRIBUTION TO THE OBSERVED DISTRIBUTION. TUNNELS WITH 0 KILLS PER TUNNEL PER SEASON TUNNELS WITH 1 KILL PER TUNNEL PER SEASON TUNNELS WITH >1 KILL PER TUNNEL PER SEASON Expected Observed χ 2 = 0.08 χ 2 (0.05),2 = 5.99 P > TRIAL 2 Te perimeter only site caugt 11 stoats, and te perimeter plus median site caugt 12 stoats. No stoats were caugt on te median line. Distribution of catc Te expected probabilities of a kill at eac location (generated from a Poisson distribution) were compared to te observed kills. Kills did not differ significantly from te Poisson random distribution (Table 2). Timing of captures Toug te number of captures eac season were very similar for bot study sites, Sylvan and Millflat, te timing of captures varied markedly. In some years captures were spread trougout te season (Millflat 1992/93), wile in oter years most kills occurred over a sort period (Sylvan 1994/95). Wen te kills are summed for tree years and bot areas, te erratic annual pattern of captures for eac area sows a more general seasonal pattern. Te total kills for eac week of te season are sown in Figure 3.. 9

9 Figure 3. Total stoat kills for eac week of te season ( combined). Total kills / week Dec Jan Feb Week of te season In trial 1 we compared te average lengt of time to kill te pool of available stoats between grids and perimeters. Grids took 8.4 (SD = 2.2) weeks on average to kill te available stoats wile perimeters took 6.9 weeks (SD = 3.1). Tis difference was not significant (P = 0.12 two sample t-test assuming equal variance). 5. Discussion Equal numbers of stoats were caugt in eac study area eac year. Tis is despite varying numbers and densities of tunnels. Doubling te size of te perimeter caugt no more stoats. Nor did putting twice as many traps in te same area. Tere may well ave been no more trappable stoats available. It suggests tat te population in bot study areas is similar, but tat te level fluctuates from year to year. If we assume tis is true, it suggests tat density of traps at te levels used, from 38 per 100 a to 60 per 50 a, as little influence on total stoat kill. Tere was no significant difference in te average time it took to kill te available stoats in perimeters compared to grids. Te fact tat tere was not even a trend towards perimeters taking longer, makes us confident tat te less dense perimeter of traps removed stoats just as quickly as te denser grid. Tis leads to te conclusion tat perimeters of tis size are as effective at removing stoats as grids in terms of time. Te perimeter does not appear to stop stoat activity witin te area concerned. In te grids of trial 1, more stoats were caugt in te internal traps tan on te periperal traps. On tis scale te trap perimeter is not acting as a protective barrier to furter immigration once residential stoats are removed, because internal catces are spread in time in te same manner as periperal catces. Two possibilities could lead to te perimeter giving less protection. Te first possibility is tat a stoat den is establised in te centre of te perimeter, and te young may not come into contact wit te trap sites until substantial predation as occurred in te den vicinity. Tis is based on te premise tat young stoats stay witin a few undred metres of te den initially and te activity of te moter is concentrated tere as well. If te site as been trapped in te previous year, as was usually te case in tese trials, te likeliood of resident stoats and dens in te area is reduced. Millflat was trapped for te first time in 1992 and resulted in early catces. Tis was reinforced in te Dart Valley in 1995/96. A 2 3 week delay in poison egg 10

10 take occurred in te two sites trapped te previous year, compared wit eigt oter poison egg perimeters wic ad not been previously trapped (B. Lawrence and P. Dilks unpublised data). To ensure we test for effects of trapping on normal residential populations, future trials sould be in areas were stoats ave not been controlled te previous year. Te second possibility is tat in a period of ig stoat numbers, wen territories are significantly smaller (Murpy & Dowding 1995), greater stoat activity in smaller geograpic areas may mean a grid does drop te general stoat population more quickly tan a perimeter does. However, tis is regarded as unlikely as te reduction in average ome range areas in irruption years (160 ± 35 to 80 ± 7 a: Murpy & Dowding 1995) is similar in scale to te trial perimeters enclosing a. Stoats sould encounter traps at te same rate in eiter year. Te perimeter design was suggested by Dilks et al. (1996). Tey found tat river edges and perimeter traps running at rigt angles to te Eglinton Valley floor caugt more stoats tan forest interior traps. Given individual ranges of stoats in an irruption year of 50 to 105 a (n = 7: Murpy & Dowding 1995), te size of trapping perimeters trialled during te present study ensures a ig probability tat stoats witin te perimeter will encounter traps at some time. But ow muc larger could perimeters be wile ensuring tat bot resident and migrant stoats still encounter traps? More knowledge of ow stoats beave witin teir ome ranges and ow new stoats invading an area beave, is needed before an optimum teoretical perimeter size can be designed. Te following questions still need to be addressed. Wat proportion of stoats are likely to be resident or transient? How important is trap encounter rate in determining kill rate? How important is unger in determining trap rate? How do sedentary stoats patrol teir territories randomly or around a set pattern? Future researc could focus on establising te teoretically best pattern of traps to optimise encounter rate by resident and immigrant stoats. Te probability of stoats encountering traps at te edge of teir ome ranges is lower tan if traps are at te centre of te range. For example, Murpy & Dowding (1995) found tat 70% of stoat activity was occurring in core ranges of 20 a witin te ome range. Trap layout sould reflect tese varying probabilities. Timing of trapping is critical. Te stoat removal period, 6-8 weeks from early December, coincides wit te igest at risk period for moua, particularly were tey are nesting for a second time (O Donnell et al. 1996). Late spring trapping in irruption years as sown it is unlikely tat female stoats can be trapped ten (King 1994). Trapping in July/August would be wort trialling as one female stoat caugt ten may prevent up to ten juvenile stoats becoming active in early December (King & Moody 1982). Tese trials were undertaken in years of low stoat densities. Tey terefore need to be tested in a stoat irruption year: not only to test tat te kill pattern is te same, but also to test tat te perimeter is as effective at protecting moua as are grids. It is important tat future trials provide clear answers. Using te data from te Eglinton Valley in 1987/98 (Elliott 1996b) and 1990/91 (O Donnell et al. 1996) we can calculate te replications required to give us te power to detect te estimated differences. We need two replicates to establis a difference of five predation events in trapped to untrapped areas, and tree replicates to establis a difference of four predation events. Te 1990/91 results indicate te latter requirement.. 11

11 6. Recommendations To establis weter a perimeter of traps is sufficient to protect moua in a stoat irruption year, it is recommended: 1. Tat 1000 m 1000 m perimeters wit traps at 100 m centres wit a median line be establised during te next stoat irruption, wit a grid and non-treatment area to act as controls. 2. Tat productivity of moua at eac site be monitored closely enoug to establis te level of predation and nest success. 3. Tat tree replicates of te trial be carried out. 4. Tat future trials occur in areas were no stoat control occurred in te previous year. 5. Tat trials of te effectivenes of winter trapping on moua productivity be carried out. 7. Acknowledgments Tanks to Debbie Kennett wo did te field work , Kevin Mitcell and Ian Soutey wo did te field work in 1994/95, Peter Dilks, Mick Hutcins and Ricard Kennet for logistic support and encouragement, and Peter Webb for commenting on te draft. 12

12 8. References Dilks, P.J.; O Donnell, C.F.J.; Elliott, G.P.; Pillipson, S.M. 1996: Te effect of bait type, tunnel design, and trap position on stoat control operations for conservation management. New Zealand Journal of Zoology 23: Elliott, G.P. l996a: Moua and stoats: a population viability analysis. New Zealand Journal of Zoology 23: Elliott, G.P. 1996b: Productivity and mortality of moua (Mooua ocrocepala). New Zealand Journal of Zoology 23: King, C.M. 1980: Field experiments on te trapping of stoats (Mustela erminea). New Zealand Journal of Zoology 7: King, C.M. 1981: Te effects of two types of steel traps upon captured stoats (Mustela erminea). Journal of Zoology (London) 295: King, C.M. 1984: Immigrant killers. Introduced predators and te conservation of birds in New Zealand. Oxford University Press, Auckland. King, C.M. 1994: Monitoring and control of mustelids on conservation lands. Part 1: Planning and assessing an operation. Department of Conservation Tecnical Series 3. Wellington, New Zealand. King, C.M.; Edgar, R.L. 1977: Tecniques for trapping and tracking stoats (Mustela erminea): a review and a new system. New Zealand Journal of Zoology 4: King, C.M.; McMillan, C.D. 1982: Population structure and dispersal of peak year coorts of young stoats. New Zealand Journal of Ecology 5: King, C.M.; Moody, J.E. 1982: Te biology of te stoat (Mustela erminea) in National Parks in New Zealand: IV. Reproduction. New Zealand Journal of Zoology 9: King, C.M.; O Donnell, C.F.J.; Pillipson, S.M. 1994: Monitoring and control of mustelids on conservation land. Part 2: Field and worksop guide. Department of Conservation Tecnical Series 4. Wellington, New Zealand. Murpy, E.C.; Dowding, J.E. 1995: Ecology of te stoat in Notofagus forest: ome range, abitat use and diet at different stages of te beec mast cycle. New Zealand Journal of Ecology 19: O Donnell, C.F.J. 1996: Predators and te decline of New Zealand forest birds: an introduction to te ole nesting bird and predator programme. New Zealand Journal of Zoology 23: O Donnell, C.F.J.; Dilks, P.J.; Elliott, G.P. 1996: Control of a stoat (Mustela erminea) population irruption to enance moua (yellowead) (Mooua ocrocepala) breeding success in New Zealand. New Zealand Journal of Zoology 23: