CORE Metadata, citation and similar papers at core.ac.uk Provided by UC Research Repository Available online at www.science.canterbury.ac.nz/nzns

Mohoua ochrocephala abundance in the Catlins following aerial 1080 control

Jakob K. Katzenberger1,2* and James G. Ross1

1 Centre for Wildlife Management and Conservation, Lincoln University, PO Box 84, Lincoln 7647, Canterbury, New Zealand 2 Workgroup on Endangered , J.F. Blumenbach Institute of Zoology and Anthropology, Georg-August University of Göttingen, Bürgerstrasse 50, 37073 Göttingen, Germany * Corresponding author: [email protected]

(Received August 2016, revised and accepted December 2016)

Abstract

Pest control using aerially-distributed 1080 bait could threaten non-target native species either by primary or secondary poisoning. To understand the impact of aerial 1080 control on the abundance of mohua (Mohoua ochrocephala), a vulnerable endemic forest bird, we analysed bird counts from the Catlins State Forest Park recorded over the period 1998-2002. Statistical modelling showed that mohua occupancy varied during the study and actually increased after 1080 control in 1999, but not significantly so. Concurrently with high predator numbers in the area during a beech ( spp.) mast event, mohua abundance significantly declined in 2001 but then recovered in 2002. In conclusion, this study shows no negative effect for the nationally vulnerable mohua following a single aerial 1080 possum (Trichosurus vulpecula) control operation. In fact, with improvements in the experimental design and survey effort this study suggests future work could show positive effects of aerial possum control on populations, based on the increase in mohua occupancy observed directly after 1080 application.

Keywords: 1080, Maclennan Range, mohua, New Zealand, pest control, possum Trichosurus vulpecula, sodium fluoroacetate, .

New Zealand Natural Sciences (2017) 42:1-8 © New Zealand Natural Sciences 2 New Zealand Natural Sciences 42 (2017)

Introduction especially vulnerable to mammalian pred- ators, which are highly abundant at the Introduced mammalian pests such same time (Elliott, 1996). In particular, as stoats (Mustela erminea), ship rats stoats were thought to be a main reason (Rattus rattus) and possums (Trichosurus for the dramatic population declines in vulpecula) pose a significant threat to mohua (O’Donnell et al., 1996) when native forest in New Zealand by increasing in response to high rodent direct predation of eggs, chicks and numbers after beech (Nothofagus spp.) adults, thereby limiting populations of tree mass seeding (masting) events (Cho- a number of endangered species (Innes quenot, 2006). Nonetheless, ship rats et al., 2010). Possums also impact forest (Rattus rattus) can also predate heavily on structure by direct browsing and spread mohua after rapid population increases bovine TB (Mycobacterium bovis) to cattle connected to factors like mild winters, and farmed deer (Sadleir, 2000). Due beech masting or possibly high levels of to these impacts pest control is applied stoat control (Dilks et al., 2003, Innes et over large areas of native forest in New al., 2010). Zealand, benefitting populations of The aim of this study was to quantify native birds and plants (Baber et al., 2009, any impact from of a single aerial 1080 Nugent et al., 2010, O’Donnell & Hoare, control operation on mohua abundance 2012). On a landscape scale, pest control in the Catlins State Forest Park and is often carried out by aerial broadcasting thereby contribute towards a better un- of sodium fluoroacetate (1080) poisoned derstanding of the risks and benefits these bait (Eason et al., 2011) targeting control operations have for non-target possums and rodents, while predators native bird species. (especially stoats) die from secondary poisoning after scavenging on poisoned Materials and methods carcasses (Murphy et al., 1999). However, Study area aerial 1080 control is associated with a risk of poisoning non-target species like This study was conducted from forest birds, either directly by bait uptake October 1998 to November 2002 in or indirectly when transferred through the Maclennan Range of the Catlins the food-web (Veltman & Westbrooke, State Forest Park, located in the south 2011). eastern corner of the of The mohua (or yellowhead, Mohoua New Zealand. The Catlins State Forest ochrocephala) is a small, insectivorous, Park lies on the boundary between the forest , endemic to the South and Southland regions, between Island of New Zealand. The species has Balclutha and . Approximately had strong declines over the past decades 50 % of the area in the Maclennan ranges (Dilks et al., 2003, O’Donnell, 1996) and is southern beech forest habitat and was has disappeared even from extensive areas designated as priority area for mohua of relatively unmodified native forest, conservation (O’Donnell, 1993). thus this species is classified as “Nation- Bird monitoring ally Vulnerable” (Robertson et al., 2012). The breeding strategy of mohua, which is Two different methods were applied a hole-nester with a long incubation time to monitor mohua abundance in this in late summer, makes their populations study. Firstly, 1-km grid square search Katzenberger & Ross: Mohua abundance following aerial 1080 control 3 as described in O’Donnell & Dilks by DOC (Department of Conservation, (1986) was carried out with the following 2016a) modifications. At least 30 minutes was Statistical analysis spent in each square and deliberate searches for mystery birds were permitted, The presence/absence data from the grid to overcome differences in observer search was analysed with a Generalized experience and weather conditions. Also Linear Mixed Effects Model (GLMM) a bird ‘squeaker’ was used to stimulate with a binomial error distribution. As mohua calling. Additionally, at least some of the squares were revisited each one five-minute bird count was usually year, a random intercept for the square ID conducted in each grid square, following was used to control for non-independence the standard method described in Dawson of these observations. Significance of & Bull (1975). The monitoring work fixed effects was evaluated with a type was conducted annually over the four III Wald chi-square test. These analyses years 1998-2002 in October and early were carried out using R v. 3.1.2 (R November, being the time of greatest Core Team, 2016), the ‘lme4’ package mohua calling (Table 1). v. 1.1-7 (Bates et al., 2015) and the ‘car’ package (Fox & Weisberg, 2011). Where Aerial 1080 control significant differences existed, post-hoc Aerial 1080 control was conducted means testing was conducted using Tukey in the study area in winter 1999 with contrasts with the ‘multcomp’ package bait containing sodium fluoroacetate (Hothorn et al., 2008). (1080). The 1080 control operation used Differences in the mean numbers of Waimate RS5 cereal-pollard baits, which birds observed during 30-minute searches were dyed green and cinnamon lured. was analysed using ANOVA. The five- Bait was sowed at 3 kg/ha with a 1080 minute bird count data were analysed concentration of 0.15 % w/w and no using a Generalized Linear Model with a prefeeding of non-toxic bait was applied. Poisson error distribution. Where signifi- This control operation was primarily cant differences existed, post-hoc means targeting possums with funding provided testing was conducted using Fishers LSD by TB Free NZ to minimise the risk of test. These analyses were carried out in TB spreading to nearby cattle and deer GenStat version 15. herds. For further details on pest control see current standard operating procedures

Table 1. Mohua monitoring effort 1998-2002 in the Maclennan Range. Aerial 1080 control in the study area was conducted in winter 1999, between the first and second sampling period.

Survey date 1-km grid squares 1-km grid squares with No. of five min. bird surveyed mohua present counts 6/10 – 6/ 11/98 84 53 83

6/10 – 7/ 11/99 37 26 32

4/10 – 6/ 11/00 55 42 32

1/10 – 1/ 11/01 78 38 72

7/10 – 8/ 11/02 112 88 101 4 New Zealand Natural Sciences 42 (2017)

Results Discussion Mohua occupancy The results from this study show that The number of squares surveyed for possum control did not have a detectable mohua occupancy varied substantially impact on mohua in the Maclennan over the study years, with lower survey Range, since no significant differences effort in years 1999 and 2000 (Table were found comparing either occupancy 1). The estimated average proportion or abundance of mohua before and after of squares with mohua occupancy the aerial 1080 drop in winter 1999. In 2 showed strong annual variation (X 360 = fact, mohua occupancy increased from 13.93, P = 0.008; Fig. 1A) with a high 70 % to 90 % in the study area following in 1999, shortly after the aerial 1080 aerial 1080 broadcasting (Fig. 1A), drop; however, this increase was not suggesting a positive effect of the control statistically significant (Tukey post-hoc operation. However, this substantial test, P = 0.24). After a predation irruption increase was not statistically significant observed in the study area during late and this is most likely attributed to the 1999 (G. Loh, pers comm., DOC, 2006), lower number of surveys conducted post- average mohua occupancy declined 1080 control in 1999 (37, compared to sharply over the next two years to a 84 in 1998). Similarly, mohua abundance minimum in 2001 that was significantly also did not show a significant difference lower than years 1999 and 2002 (Tukey before and after possum control. Based post-hoc test, both P = 0.02). on the above results we conclude that the aerial 1080 control in 1999 did not Mohua abundance have any unwanted non-target effects for There was little variation in the numbers mohua in the months and years following of birds observed during searches of the control operation. This hypothesis occupied grid squares (F4,239 = 2.41, P is also supported by more research = 0.6), with a mean of five individuals conducted since this study. For example, found per square over all years. In an 11 year study conducted by O’Donnell contrast, the five-minute bird count & Hoare (2012) who monitored mohua data over all surveyed squares indicated status with 1080 control for rats and a small but non-significant decrease in possums in the Landsborough Valley the average mohua per five minute count also demonstrated no negative impact in 1999 compared to 1998 (Fig 1B). of multiple control operations. Thus the Further pairwise comparisons showed direct impact of aerial 1080 drops on that the average value of birds counted mohua populations is now viewed as most for 2001 was significantly lower than the likely either negligible or highly beneficial values for 1998 (Fisher post-hoc test , P (cf. Veltman et al., 2014). = 0.005) and 2002 (Fisher post-hoc test, Nevertheless, the ongoing bird moni- P = 0.007). The five-minute bird counts toring in our study showed a significant were generally conducted following a decline in both mohua occupancy and grid square search, therefore the annual abundance occurring two years after a variation in effort was similar to the these predator irruption was first observed surveys (Table 1). in summer 1999. This high predator abundance was most likely related to the heavy beech masting that occurred in Katzenberger & Ross: Mohua abundance following aerial 1080 control 5

Figure 1. (A) proportion of squares with mohua occupancy in October 1998-2002 in the Maclennan Range estimated with a binomial GLMM fitted on grid square data. Error bars show SEM while lowercase letters indicate significant differences from a post-hoc analysis of means by Tukey contrasts. (B) numbers of mohua per October five-minute count in the Maclennan Range 1998-2002 estimated from a Poisson GLM. Error bars show SEM while lowercase letters indicate significant differences from a post-hoc analysis of means by Fishers LSD test. Dashed lines denote the timing of aerial 1080 drop, dotted lines indicate when predator irruption was first observed. the study area during autumn 1999 (G. lead to unusually high densities of stoats Loh, pers comm., DOC, 2006). Similar (O’Donnell & Hoare, 2012), which then beech mast events were seen in Eglinton also prey on native birds. This can have Valley ( National Park) during devastating effects, for example killing up 1999 and 2000, overlapping with several to 50 % of nesting females in a mohua anecdotal reports of major increases in population and destroying nearly 70 % rodent abundance in many areas of the of the nests (Elliott, 1996, King, 1983). South Island during 2000 and 2001 While the predator irruption was first (Dilks et al., 2003). High densities of observed in our study area in late 1999, rodents after such beech mast events can a significant reduction in both mohua 6 New Zealand Natural Sciences 42 (2017) occupancy and abundance was not de- to the development of the “Battle for our tected until 2001. Again, it is likely that Birds” campaign in 2014 in reponse to a the lower number of monitoring surveys signifcant beech mast event (Department conducted post 1080 control potentially of Conservation, 2016b, Department of masked the detection of an earlier effect Conservation, 2016c). Research con- by decreased statistical power. ducted in the Dart Valley following 2014 In any case, two years of decline follow- aerial 1080 control indicated mohua ing a major irruption event are expected nesting success of 89 % following 1080 for mohua populations in situations with- control, compared to only 58 % recorded out predator management (O’Donnell et in 2011, another season with very high al., 1996), and in 2002, both occupancy rat and stoat abundance (Department and abundance of mohua recovered. of Conservation, 2015). Based on these Based on the available literature it seems results, further 1080 control is planned in highly likely that the mohua population 2016 in response to another high beech decline observed in our study was due seeding event. Finally, more research is to predation by abundant mammalian warranted to better understand the indi- pests. Our results also showed that mo- rect effects of pest control on non-target hua numbers in occupied areas remained species over multiple years and where pos- stable throughout the study period. In sible an integrated approach taking into combination, this indicates that mohua account all pest species for management responded to predation by disappearing in an area should be adopted. completely from some areas, while num- bers remained constant in others. This Acknowledgements suggests that some parts of the study area are marginal habitat for mohua during We gratefully acknowledge the assistance predator irruptions, and similarly, mohua of Dr Jennifer Brown (University of territories decreased with increased preda- Canterbury) and Ian Westbrooke tor abundance in the Eglinton Valley (Department of Conservation) that (Dilks et al., 2003). Mohua population both provided invaluable advice on how declines can otherwise be associated with best to analyse this data. We acknowledge extreme cold spells (Dilks, 1999), but Graeme Loh and other DOC field no such extreme weather events were staff who diligently conducted all the recorded throughout the study period. bird surveys. We also acknowledge Additionally, because the experimental the TB Free Inc. (now OSPRI) who design of the study did not encompass contracted this research project. Two a control area, natural variation cannot anonymous reviewers are thanked for be ruled out completely as a driver of their constructive comments that helped changes in mohua abundance as well. to improve a previous draft of the Accordingly, future studies should adopt manuscript. a BACI (Before-After Control-Impact) design (Underwood, 1994) with con- References sistent sampling effect to assess impacts on threatened populations over Baber, M., Brejaart, R., Babbitt, K., multiple years. Lovegrove, T. & Ussher, G. 2009. In conclusion, this and complemen- Response of non-target native birds tary research on other bird species lead to mammalian pest control for kokako Katzenberger & Ross: Mohua abundance following aerial 1080 control 7

(Callaeas cinerea) in the Hunua Ranges, Dilks, P.J. 1999. Recovery of mohua New Zealand. Notornis: 56, 176–182. (Mohoua ochrocephala) populations Bates, D., Maechler, M., Bolker, B. & following predator control in the Walker, S. 2015. Fitting Linear Mixed- Eglinton Valley, Fiordland, New Effects Models using lme4.Journal of Zealand. Notornis: 46, 323-332. Statistical Software: 67, 1-48. Eason, C., Miller, A., Ogilvie, S. & Choquenot, D. 2006. Bioeconomic Fairweather, A. 2011. An updated modeling in conservation pest review of the toxicology and management: Effect of stoat control ecotoxicology of sodium fluoroacetate on extinction risk of an indigenous (1080) in relation to its use as a pest New Zealand passerine, Mohua control tool in New Zealand. New ochrocephala. Conservation Biology: Zealand Journal of Ecology: 35, 1-20. 20, 480-489. Elliott, G.P. 1996. Productivity and Dawson, D.G. & Bull, P.C. 1975. mortality of mohua (Mohoua Counting birds in New Zealand ochrocephala). New Zealand Journal of forests. Notornis: 22, 101-109. Zoology: 23, 229-237. Department of Conservation. 2015. Fox, J. & Weisberg, S. 2011. An R Battle for our birds – a success story. Companion to Applied Regression. Wellington, PO Box 10420, New Second ed. Sage, Thousand Oaks, CA. Zealand, from http://www.doc.govt. Hothorn, T., Bretz, F. & Westfall, P. 2008. nz/our-work/battle-for-our-birds/ Simultaneous Inference in General updates/june-2015/. Parametric Models. Biometrical Department of Conservation. 2016a. Journal: 50, 346-363. Operational Planning for Animal Innes, J.G., Kelly, D., McC. Overton, Pest Operations. Wellington, PO J. & Gillies, C. 2010. Predation and Box 10420, New Zealand, from other factors currently limiting New http://www.doc.govt.nz/nature/pests- Zealand forest birds. New Zealand and-threats/animal-pests/managing- Journal of Ecology: 34, 86-114. animal-pests/standard-operating- King, C.M. 1983. The relationships procedures/. between beech (Nothofagus sp.) Department of Conservation. 2016b. seedfall and populations of mice (Mus Battle for our Birds. Wellington, PO musculus), and the demographic and Box 10420, New Zealand, from www. dietary responses of stoats (Mustela doc.govt.nz/battleforourbirds. erminea), in three New Zealand Department of Conservation. 2016c. forests. Journal of Animal Ecology: 52, The science behind Battle for our Birds. 141-166. Wellington, PO Box 10420, New Murphy, E., Robbins, L., Young, J. Zealand, from www.doc.govt.nz/ & Dowding, J. 1999. Secondary battleforourbirds. poisoning of stoats after an aerial 1080 Dilks, P., Willans, M., Pryde, M. & Fraser, poison operation in Pureora Forest, I. 2003. Large scale stoat control to New Zealand. New Zealand Journal of protect mohua (Mohoua ochrocephala) Ecology: 23, 175-182. and kaka (Nestor meridionalis) in Nugent, G., Whitford, J., Sweetapple, P., the Eglinton Valley, Fiordland, New Duncan, R. & Holland, P. 2010. Effect Zealand. New Zealand Journal of of one-hit control on the density of Ecology: 27, 1-9. possums (Trichosurus vulpecula) and 8 New Zealand Natural Sciences 42 (2017)

their impacts on native forest. Science Vienna, Austria: R Foundation for for Conservation, 304: Wellington, Statistical Computing. Department of Conservation. 64 p. Robertson, H.A., Dowding, J.E., Elliott, O’Donnell, C.F.J. 1993. Mohua G.P., Hitchmough, R.A., Miskelly, (yellowhead) Recovery Plan. Wellington, C.M., O’Donnell, C.J.F., Powlesland, N.Z.: Threatened Species Recovery R.G., Sagar, P.M., Scofield, R.P. & Plan. Series 6. Department of Taylor, G.A. 2012. Conservation Conservation. status of New Zealand birds, 2012. O’Donnell, C.F.J. 1996. Monitoring New Zealand Threat Classification mohua (yellowhead) populations in Series 4. Department of Conservation, the South Island, New Zealand, 1983- Wellington: 26 p. 93. New Zealand Journal of Zoology: Sadleir, R. 2000. Evidence of possums 23, 221-228. as predators of native . In T. O’Donnell, C.F.J. & Dilks, P.J. 1986. Montague, ed. The brushtail possum: Forest Birds in South Westland. biology, impact and management of an Status, Distribution and Habitat Use. introduced marsupial. Lincoln, N.Z.: Wellington, N.Z.: New Zealand Manaaki Whenua Press: 126-131. Wildlife Service occasional publication. Underwood, A.J. 1994. On beyond BACI: Department of Internal Affairs. sampling designs that might reliably O’Donnell, C.F.J., Dilks, P.J. & Elliott, detect environmental disturbances. G.P. 1996. Control of a stoat (Mustela Ecological Applications: 4, 3-15. erminea) population irruption to Veltman, C.J. & Westbrooke, I.M. 2011. enhance mohua (yellowhead) (Mohoua Forest bird mortality and baiting ochrocephala) breeding success in New practices in New Zealand aerial 1080 Zealand. New Zealand Journal of operations from 1986 to 2009. New Zoology: 23, 279-286. Zealand Journal of Ecology: 35, 21-29. O’Donnell, C.F.J. & Hoare, J.M. 2012. Veltman, C.J., Westbrooke, I.M., Quantifying the benefits of long-term Powlesland, R.G. & Greene, T.C. integrated pest control for forest 2014. A principles-based decision bird populations in a New Zealand tree for future investigations of native . New Zealand New Zealand birds during aerial 1080 Journal of Ecology: 36, 131-140. operations. New Zealand Journal of R Core Team. 2016. R: A language and Ecology: 38, 103-109. environment for statistical computing.