Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

Short-term movement patterns, population estimates and breeding biology of an island endemic bird, the Tristan Thrush

Short-term movement patterns, population estimates and breeding biology of an island endemic... Revista Brasileira de Ornitologia 27(4): 245–252. ARTICLE December 2019 Short-term movement patterns, population estimates and breeding biology of an island endemic bird, the Tristan Thrush 1,4 1 1 2 3 Peter G. Ryan , Ben J. Dilley , Delia Davies , Trevor Glass & Fitsum Abadi FitzPatrick Institute of African Ornithology, DST–NRF Centre of Excellence, University of Cape Town, South Africa. Tristan Conservation Department, Edinburgh of the Seven Seas, Tristan da Cunha, United Kingdom. Department of Fish, Wildlife and Conservation Ecology, College of Agricultural, Consumer and Environmental Sciences, New Mexico State University, Las Cruces, United States of America. Corresponding author: pryan31@gmail.com Received on 18 May 2019. Accepted on 22 October 2019. ABSTRACT: The Tristan Thrush Turdus eremita is the only land bird that survived human colonisation of the main island of Tristan da Cunha and is listed as “Near Threatened ”. Population estimates are confounded by the thrushes' inquisitive and gregarious nature -1 as well as limited information on their movements. We report the first measures of nest densities on Nightingale Island: 6 nests·ha -1 in Phylica arborea woodland and 4–5 nests·ha in tussock habitat, which suggests that the population is approximately double the previous estimate. At Inaccessible Island, we individually color ringed 110 thrushes over two months to track their short-term movements and estimate the local population size. Individuals moved up to 950 m along the coast, but 96% of resightings were < 100 m. A Bayesian data augmentation approach estimated that some 260 thrushes visited the core study area, a 200-m stretch of cobble and boulder beach where birds come to drink, bathe and forage. This result suggests that the population on Inaccessible Island also is substantially larger than reported previously. We estimate the total population to be 8000–15,000 Tristan Thrushes. The main need is a population estimate for the nominate subspecies on the main island of Tristan. KEY-WORDS: Turdus eremita, Inaccessible Island, Nightingale Island, Bayesian population estimate. INTRODUCTION and its offshore islets, Middle and Sto ltenhoff (Elliott 1957, Richardson 1984, Ryan et al. 2011). The species The Tristan Thrush or Star chy Turdus eremita is endemic is listed as “Near Threatened ” globally, and although the to the Tristan da Cunha Archipelago in the central South population is thought to be stable, there is a need for o o Atlantic Ocean (37 S; 12 W). Until recently it was placed up-to-date population estimates (BirdLife International in its own genus, Nesocichla, reflecting its morphological 2017). Current estimates are extrapolated from crude adaptations to life on oceanic islands: reduced wings, density estimates made by field workers. In the early robust legs and a brush-tipped tongue adapted for lapping 1970s, Richardson (1984) suggested that there were 40– up egg contents (Lowe 1923, Rand 1955, Fraser et al. 60 pairs at Tristan, 300–500 pairs at Nightingale Island, 1994). However, genetic sequence data suggest that it is 20–40 pairs at Middle, 10–20 pairs at Stoltenhoff and nested within the South American radiation of Turdus 100–500 pairs at Inaccessible Island. His estimate for thrushes (Klicka et al. 2005, Voelker et al. 2007), and like Inaccessible Island was particularly crude because he was the other land birds found at Tristan (Ryan et al. 2013, unable to spend much time on that island. Subsequent Stervander et al. 2018), it probably reached the islands in field work on Inaccessible Island in the 1980s improved the last few million years. the estimate there to 850 pairs (Fraser et al. 1994), giving The thrus h is the only land bird that survived the current total population estimate of around 6000 human colonisation of the main island of Tristan, where birds (BirdLife International 2017). Estimates to date a small population of the nominate subspecies persists have focused on breeding pairs because the thrushes' mainly above 300 m elevation on the steep coastal inquisitive and gregarious nature (e.g., Hagen 1952, scarps and upper plateau or “base” (Ryan 2007, BirdLife Fraser et al. 1994) confounds estimates of the density of International 2017). Thrushes are common at the two non-breeding individuals using transect or point counts. uninhabited islands in the archipelago: T. e. gordoni on We studied thrushes at two of the three islands Inaccessible Island and T. e. procax on Nightingale Island in the Tristan Archipelago: on Inaccessible Island, we Tristan Thrush movements and population size Ryan et al. thus the sex of at least some birds could be inferred. aimed to individually mark a sample of thrushes to Fledged chicks were observed from early November, but better understand their movements and estimate their no juveniles were ringed, and thus all data refer to birds population size; and on Nightingale Island we aimed to accurately estimate nest densities and gather data on at least 1 year old. Thrushes were mainly caught with breeding phenology over three seasons of fieldwork. hand nets, but a few birds were caught in mist nets set for finches. Thrushes initially received two color-rings Lastly, we aimed to use these data to refine the population as well as a metal ring, allowing them to be identified estimate of Tristan Thrushes in the ar chipelago. individually without recapture. However, we ran out of distinct color combinations after ringing 110 birds, and METHODS from 09 November, birds in the study area received a metal ring only (no further thrushes were ringed outside the study area). Ringing ceased on 23 November when Breeding biology and nest densities we ran out of metal rings, but resighting observations We recorded the contents of all nests found. Tristan continued until 25 November, the day before we left the Thrush nests are har d to locate during incubation and island. early chick stage (Fraser et al. 1994), but are conspicuous Ringing and resighting effort was concentrated along a 200-m long stretch of cobble and boulder once the chicks start to beg noisily when being fed. We beach backed by low tussock-covered cliffs centered on estimated the approximate age of chicks in nests from the nestling descriptions in Fraser et al. (1994). Authors Runaround Beach (Fig. 1). Several fresh-water springs B.J.D. and D.D. systematically recorded nests while and seeps emerge along the base of the cliffs in this conducting research on Nightingale Island in 2015 and area, attracting thrushes to drink and bathe. Thrushes also forage along the shoreline (Fraser et al. 1994, Ryan 2017. Additional ad hoc observations were made by P .G.R. & Dilley 2019), and are much easier to resight in this on Nightingale Island in October–November 2007 and Inaccessible Island in November 1999–February 2000, habitat than in the dense vegetation that covers most of September–December 2004 and October–December the island. We attempted to resight color-ringed Thrushes 2009. and catch unringed individuals at Runaround Beach on 20 days from 29 September to 25 November. Most visits In 2015 and 2017, thrush nests were recorded involved two observers, and lasted at least 2 h. On 19 systematically on Nightingale Island along the path from the huts to the Ponds (1.1 km) and in First Wood, a 4.3 and 25 November, visits lasted 4 h, and were divided into ha patch of Phylica arborea woodland, which was searched two sessions, recording individuals present in each 2-h intensively for nests. The pathway, which runs through period to give a total sample of 22 observation events. We calculated individual detection probabilities as the tall Spartina arundinacea Tussock grassland, was used proportion of observation periods when color-ringed daily to reach First Wood, so most nests were found that reached the large chick stage. To estimate nest densities birds were resighted after initial ringing (n = 12–21 in Tussock grassland, we assumed the path represented observation periods). a strip transect between 20–40 m wide (10–20 m either Most sighting effort occurred a long the coast. In addition to the main study area, the shoreline between side of the path, 2.2–4.4 ha). All nests were within 10 m Runaround Beach and Blenden Hall Bay, where there of the path, but the path created a favourable foraging area and might have biased the distribution of thrush nests is a small research hut (Fig. 1), was checked for ringed (e.g., several pairs of Brown Skua Catharacta antarctica thrushes on each visit to the study area. The coast between breed along the path, and thrushes often scavenge from Tern Rock and West Point (Fig. 1) also was searched on 55 days from 22 September to 26 November, and any carcasses of petrels killed by the skuas). We divided the ringed thrushes noted. Occasional visits also were made path into three sections: the lower 300 m runs through mostly level ground close to Northern Rockhopper to the accessible section of coast to Warren's Cliff, east of Penguin Eudyptes moseleyi colonies; the next 330 m is also Runaround Beach, and between West Point and Dirleton mostly level ground but lacks penguin colonies, and the Point (Fig. 1). On 01 December we also checked 1.7 km of the northeast coast between the Waterfall and north of final 470 m is on sloping terrain. Salt Beach. Resighting data at Inaccessible Island Systematic searches for ringed thrushes away from the coast were complicated by the dense vegetation, We ringed Tristan Thrushes on Inaccessible Island especially on the coastal slopes. The main paths followe d o o on the western side of the island, although the East Road (37 18'S; 12 41'W) from 22 September to 23 November was used much less than the West Road to access the 2018, which coincides with the species' breeding season. Only females develop well-defined brood patches, and plateau (Fig. 1). In addition, a few thrushes were ringed Revista Brasileira de Ornitologia 27(4): 2019 Tristan Thrush movements and population size Ryan et al. Figure 1. The Tristan Ar chipelago (A), Inaccessible Island (B), and main study area at Runaround Beach (black rectangle). Red lines show Tristan Thrush movements > 100 m (closed cir cle = ringing location; open circle = resighting location). th around our campsite on the eastern plateau at Denstone where Z = 1, if the i individual was a member of the o o Junction (37 17.61'S; 12 40.43'W). All distances moved population and 0 otherwise, and  is the inclusion and habitat areas were estimated from Google Earth. probability. The closed population (N) was then estimated as the sum of Z (i.e., N = Z ). Second, we modeled the i i Estimating thrush abundance encounter histories of each individual conditional on the latent variables (Z ) as Our initial plan was to mark 50 birds on the coast and 50 on the plateau in areas that would be visited frequently Y ~Bernoulli(Z p ) it i it over the two months we were on the island, but it soon th became apparent that the number of birds was greater where Y if the i individual was alive and detected at it than anticipated, and we focused most attention on a occasion t and 0 otherwise. p is the detection probability it th single coastal site where we attempted to ring as many of the i individual at occasion t. We imposed two possible thrushes as possible. Repeated visits to this site, coupled constraints on detection probability: time and individual with regular checks along the adjacent coastline, allowed heterogeneity (i.e., individual variation in detection us to gauge movement distances and use a Bayesian probability) (Table 1). Under M , M and M models, we t h ht approach to estimate the numbers of thrushes visiting used a random-effect approach to model t he time effect the main study site. We used closed population models and individual heterogeneity. Hence,  and  are time t i (Otis et al. 1978) to estimate abundance of the Tristan and individual random effects, respectively. Bir ds marked Thrush population at the core study area at Runaround with metal-rings only were given a detection probability Beach. All models were implemented using a Bayesian of zero after first capture, because there were no resighting data augmentation approach (Royle & Dorazio 2012), data for these individuals. boosting the observed 147 encounter histories with a large To implement the models in a Bayesian framework, number of individuals (M) that had all-zero encounter we specified vague prior distributions for all model histories. We used a hierarchical approach to describe parameters: U (0,1) for , N(0, 0.01) for  , and U(0, 10) the models. First, we used a Bernoulli distribution to for  and  . For each model, we ran three independent determine whether an individual was a member of the Markov Chain Monte Carlo (MCMC) iterations of true population exposed to sampling. That is, length 100,000 with a burn-in of 50,000, and a thinning rate of 50. We used the Brooks-Gelman-Rubin R statistic Z ~Bernoulli() (Brooks & Gelman 1998) to assess the convergence of the Revista Brasileira de Ornitologia 27(4): 2019 Tristan Thrush movements and population size Ryan et al. Table 1. Models fitted to the Tristan Thrush resighting data in the core study area at Inaccessible Island, and associated DIC and DIC scores. Variation in detection Model logit model for p DIC score DIC probability (p) logit(p ) =  +  + it 0 t i M Individual + time 2475.6 0.0 2 2 ht ~ N(0,  );  ~ N(0,  ) t  i logit(p ) =  + it 0 t M Time 2584.0 108.4 t 2 ~ N(0,  ) logit(p ) =  + it 0 i M Individual heterogeneity 2674.2 198.6 ~ N(0,  ) M Null logit(p ) =  2789.2 313.6 0 it 0 MCMC chains to the targeted posterior distributions. likely occur. Based on the inferred ages of chicks when Both the R statistic and visual inspections of the MCMC nests were found, most eggs hatched from early October plots indicated model convergence. The model with to late November. Our results suggest a crude density of -1 -1 the smallest Deviance Information Criterion (DIC; 3–10 nests·ha in tussock habitat and ~6 nests·ha in Spiegelhalter et al. 2002) was considered the best model. Phylica woodland. All analyses were performed using the JAGS software (Plummer 2003), called from R (R Core Team 2018) Movements at Inaccessible Island using the package jagsUI (Kellner 2018). Of the color-ringed thrushes, 71 were caught at the core study site, 30 were caught farther west along the coast RESULTS between Runaround Beach and Blenden Hall Bay or immediately inland around the hut (Fig. 1), and 9 were Nest densities at Nightingale Island caught on the plateau (seven at Denstone Junction) or high on the West Road (two). The resighting probability We found 60 nests over the study period. All thrush nests of color-ringed thrushes in the core study area ranged from along the path were in Tussock stands (n = 32). Nests 0–0.83, with 32% of birds observed on less than 10% of in First Wood (n = 28) were found in Tussock stands visits, and 85% seen on less than 50% of visits (Fig. 2). (9), Asplenium ferns (11), Carex sedge (5), in small rock Overall, the most common number of resightings was 0 crevices sheltered by hanging Scirpus grass (2), and in (Fig. 1 in Appendix I), even though most color ringing New Zealand Flax Phormium tenax (1). Flax is an invasive away from the study area ceased in mid-October, allowing species at Tristan, so this nest was moved ~ 1 m to an more than a month for ringed birds to be resighted. A adjacent Carex stand to allow the flax plant to be removed; few birds were seen regularly in the same area. Some of the female continued to incubate, and all three eggs these birds held breeding territories and subsequently hatched. Tussock nests were 0.2–1.4 m off the ground, were seen with fledglings, whereas others seemingly were but other nests were within 5–10 cm of the ground. Eight non-breeders that foraged along the shoreline (see Ryan eggs at 4 nests measured 30.7 ± 0.8 mm (29.3–31.6 mm) & Dilley 2019). Most non-breeders remained on more in length and 21.7 ± 0.6 mm (20.9–22.4 mm) in width. or less the same stretch of coast throughout, but one bird Clutch size at 9 nests with eggs averaged 2.1 ± 0.6 (1–3 moved ~120 m west midway through the study period, eggs), but some clutches may have been incomplete. remaining in this new site until we left the island. Average brood size at 55 nests was 2.4 ± 0.6 chicks, with The pattern of a few resident individuals and larger 51% of nests containing three chicks. The number of numbers of transient birds was obvious at the hut, where thrush nests found along the main path to the Ponds was catching and resighting effort was arguably greatest. 18 in 2015 and 14 in 2017, with densities tending to Fourteen thrushes were caught at the hut. Of these, one decrease with distance from the coast and elevation (Table pair of birds was resident, being seen virtually every day 2). A total of 28 nests were found in Phylica woodland at ranging 50 m between the hut and the adjacent coast; First Wood in 2017. We were unable to tell whether any this pair nested in dense tussock next to the beach in nests were repeat lays after breeding failure, but this is front of the hut. Of the remaining 12 thrushes, one was unlikely to greatly influence the estimate of the number seen regularly in Blenden Hall Bay, 90 m from the hut, of breeding pairs because few nests were found during two were resighted once in the core study area (~950 m incubation or the small chick stage, when most failures from the hut; resighted 28 and 32 days after ringing), one Revista Brasileira de Ornitologia 27(4): 2019 Tristan Thrush movements and population size Ryan et al. Table 2. Numbers of Tristan Thrush nests in Tussock grassland a long the path from the huts to First Wood on Nightingale Island in 2015 and 2017. Number of nests Density* Habitat Distance -1 (nests·ha ) 2015 2017 Level ground with 300 m 8 5 5.4–10.8 penguins Level ground, no 330 m 4 6 3.8–7.6 penguins Sloping ground 470 m 6 3 2.4–4.8 *assumes average number of nests and strip transect 20–40 m wide; see methods for further details. was seen once in Phylica trees 20 m behind the hut (44 data (Table 1). The posterior mean thrus h population days after ringing), and the other 8 were not seen again. visiting Runaround Beach under this model was ~258 A similar pattern also occurred at the plateau camp, birds (95% credible interval, CRI: 212–314), and this although only 7 thrushes were ringed here: 2 were seen estimate was largely independent of M, providing M on most visits to the camp, 1 was seen once (recaptured ≥ 200 (Table 1 in Appendix I, Fig. 2 in Appendix I). in a mist net the day after initial capture), and 4 were not The rate at which t hrushes were ringed at the study site seen again. showed no evidence of decreasing over time (Fig. 3), Most resightings were within 100 m of the original supporting the large population estimate for the site. ringing site (96%, n = 427 of 446). However, we frequently The posterior mean detection probability was 0.031 observed thrushes making flights > 100 m, mainly along (95% CRI: 0.016–0.053), and posterior estimates for the shore. One male that defended a territory at the  and  were 0.91 (0.63–1.29) and 1.92 (1.57–2.39), western edge of the study area frequently chased birds out respectively, confirming the considerable temporal and of sight along the coast to the west (at least 200 m), but individual variation in detection probability. was not observed outside the study area. The maximum movement distance recorded was 950 m, from the hut to the core study area (n = 2). Two other birds ringed in DISCUSSION Blenden Hall Bay moved as far as the study area (700– 900 m), and 1 ringed near Tern Rock also was seen in The main challenge to estimating the population size of Tristan Thrushes is their inquisitive nature. As Hagen the core study area (500 m). The farthest movement of birds ringed in the study area was to the Tern Rock area (1952) noted “they immediately fly to meet every new (400–500 m, n = 4 of 71 birds). No ringed birds were thing”, including people. As a result, any attempt to seen inland from the study area, but access to the dense estimate the population density by random transects Spartina tussock in this area was restricted to the East or point counts is confounded by the unknown radius over which thrushes are attracted. Individually marking Road, which was only visited a few times. In the more accessible Blenden Hall area, one thrush caught in a mist thrushes does not directly solve this problem, because the net midway between the hut and Wilkins' Copse, 130 m act of catching and ringing the birds makes them more inland (Fig. 1), was subsequently seen regularly on the cautious around people (Fraser et al. 1994, pers. obs.). adjacent coast. Current population estimates are crudely estimated from the approximate density of breeding pairs (Richardson Thrush numbers in the study area 1984, Fraser et al. 1994). Richardson (1984) estimated at Inaccessible Island 300–500 pairs on Nightingale Island, but our observations show that this is conservative. The population estimate for the core study area was based We present the first estimates of nest densities for Tristan Thrushes. On Nightingale Island, the estimate on encounter histories for 76 color-marked thrushes (71 -1 ringed at the site and 5 birds ringed along the adjacent for Phylica woodland at First Wood (6 nests·ha ) is coast as far west as the hut) and 71 metal-only ringed likely to be fairly accurate, as the entire area was searched birds caught at the study sight after we ran out of color intensively. It is likely that some thrushes re-lay if their combinations. Average resighting rate of the 76 ringed initial breeding attempt fails but this is unlikely to have inflated t he nest count because most nests were only thrushes in the 12 2-h observation periods in the core study area after color-ringing ceased was 29 ± 6% (21– recorded at the chick stage and there is no indication 38%). The model including individual heterogeneity and of double brooding (the breeding season is short, with a time effect (M ) provided the best fit to the resighting adults starting to moult from late November while still ht Revista Brasileira de Ornitologia 27(4): 2019 Tristan Thrush movements and population size Ryan et al. feeding fledged chicks; Fraser et al. 1994). Estimating this species. Despite ample opportunities for birds to be nest density for tussock habitat is complicated by the resighted, and high resighting rates for some individuals, fact that the path is not a random transect through this the most common result was for color-ringed birds to be habitat. However, incidental observations made while not seen again (Fig. 2 & Fig. 1 in Appendix I). More walking through tussock away from the path suggest observations are needed to assess whether such birds that the density of thrush nests along the path is not remain in areas of dense tussock, not readily accessible -1 atypical, and a density of 4–5 nests·ha is thought to to observers, or are more vagile. Until this conundrum is be a reasonable average across the island. Extrapolating resolved, we will struggle to derive accurate population these densities based on the total area of each habitat at estimates. Nightingale Island (160 ha tussock and 11 ha woodland) Despite the issue of possible transient birds, the suggests a population of some 700–850 breeding pairs, number of thrushes caught at the core study site at roughly double that estimated by Richardson (1984). Inaccessible Island was impressive, with no decrease in the This number of bree ding pairs suggests a total population rate at which unringed birds were caught over 2 months of 2000–3500 thrushes on Nightingale Island, including (Fig. 3). In fact, the proportion of unringed birds caught pre-breeders (few if any thrushes breed until they are at least two years old, Hagen 1952). Our breeding observations confirm that clutch sizes on Nightingale Island are consistently larger than those on Inaccessible Island. Roughly half of all broods on Nightingale were of three chicks, whereas three-egg clutches have not been recorded on Tristan or Inaccessible Island (Elliott 1957, Richardson 1984, Fraser et al. 1994, P.G.R. unpub. data). Quite what causes this difference is not known. Elliott (1957) reported that eggs of T. e. procax on Nightingale Island were larger than those of T. e. gordoni on Inaccessible Island, but our egg measurements at Nightingale averaged smaller (30.7 × 21.7 mm) than those reported by Elliott (1957) of 33.5 × 22.7 mm. Thrush densities are thought to be lower at Inaccessible than at Nightingale Island. Fraser et al. -1 (1994) estimated thrushes occurred at 1.0 pair·ha in -1 Figure 2. The probability that a Tristan Thrush color-ringed in tussock grassland, and 0.2–0.8 pairs·ha in three habitats the core study area was resighted in subsequent visits (white = on the island plateau. However, densities are hard to 0 probability; n = 71). assess, particularly along the coastal cliffs where nests are concealed in the densest, most inaccessible Tussock grass (Fraser et al. 1994), and as on Nightingale Island, probably are conservative estimates. We attempted to gain an independent estimate of thrush density by color- ringing a large sample of birds at Inaccessible Island. Our results support previous records on the movement of ringed Tristan Thrushes. The 36 thrus hes color-ringed at Inaccessible Island in 1982/83 were largely sedentary, with regular movements of around 250 m between the hut and Wilkins' Copse, and a maximum movement of 800 m (Fraser et al. 1994). Similarly, of the “nearly 100” thrushes Elliott (1957) ringed at the landing area on Nightingale Island, none was resighted more than 300 m away, suggesting they seldom undertake large movements. However, 2 of the 5 color-ringed birds resighted at Inaccessible Island from 1987–1989/90 #! $% &"! were observed only once over the three seasons (Fraser Figure 3. The cumulative number of Tristan Thrushes ringed at et al. 1994), suggesting that they are not regular visitors the core study area from 22 September to 23 November 2018. to areas accessible to observers. Our results confirm that Color-ringing ceased from 09 November, making it faster to individual variation in resighting rates is characteristic of process birds and thus increasing the catch rate. Revista Brasileira de Ornitologia 27(4): 2019 ! " ! Tristan Thrush movements and population size Ryan et al. decreased over the course of the study, as individuals REFERENCES became more wary of being approached with a hand BirdLife International. 2017. The IUCN red list of threatened species net – had this not been the case, even more birds would Turdus eremita. e.T22708541A117286443 (Access on 25 April have been caught later in the study. On the final day of 2019). observations (25 November) we estimated that there Brooks S.P. & Gelman A. 1998. General methods for monitoring were almost as many unringed birds as there were color- convergence of iterative simulation. Journal of Computational and Graphical Statistics 7: 434–455. ringed and metal-only birds, which lends credence to the Elliott H.F.I. 1957. A contribution to the ornithology of Tristan da Bayesian estimate of around 260 birds visiting the study Cunha group. Ibis 99: 545–586. site. If we assume that thrushes from the lower half of the Fraser M.W., Ryan P.G., Dean W.R.J. Briggs, D.J. & Moloney C.L. coastal cliffs (i.e., up to c. 200 m elevation) visit the core 1994. Biology of the Tristan Thrush Nesocichla eremita. Ostrich study site, and travel up to 500 m along the coast (likely a 65: 14–25. Hagen Y. 1952. The bir ds of Tristan da Cunha. Results of the Norwegian minimum estimate, given the limited movement of most Expedition to Tristan da Cunha 1937–1938. 20: 1–248. birds), then the population at the site represents birds from Kellner K. 2018. jagsUI: a wrapper around ‘rjags’ to streamline ‘JAGS’ a catchment area of around 20 ha. Increasing the radius analyses. R package version 1.5.0. along shore to the maximum movement recorded would Klicka J., Voelker G. & Spellman G.M. 2005. A molecular phylogenetic analysis of the “true thrushes” (Aves: Turdinae). increase this to around 40 ha. These catchment areas give -1 Molecular Phylogenetics and Evolution 34: 486–500. a density of 6.5–13 thrushes·ha , which if extrapolated Lowe P.R. 1923. Notes on some land birds of the Tristan da Cunha across the coastal scarp (500 ha), suggests a population group collected by the ‘Quest’ Expedition. Ibis 65: 511–528. of roughly 3250–6500 fully grown thrushes. The density Otis D.L., Burnham K.P., White G.C. & Anderson D.R. 1978. on the plateau (1000 ha) is perhaps 1/3 to 1/2 that along Statistical inference from capture data on closed animal populations. Wildlife Monographs 62: 1–135. the coast (Fraser et al. 1994), giving a crude population of Plummer M. 2003. JAGS: a program for analysis of Bayesian graphical 5500–11,000 thrushes on Inaccessible Island. rd models using Gibbs sampling. Vienna: Proceedings of the 3 Elliott (1957) does not specify the period over International Workshop on Distributed Statistical Computing which he ringed “nearly 100” thrushes in the immediate (DSC 2003). R Core Team 2018. R: a language and environment for statistical vicinity of the landing-place on Nightingale Island, but computing. Vienna: R Foundation for Statistical Computing. his experience mirrors ours on Inaccessible Island, and Rand A.L. 1955. The origin of the land birds of Tristan da Cunha. supports our inference of a large thrush population Fieldiana Zoology 37: 139–166. from nest densities on Nightingale Island. Combined Richardson M.E. 1984. Aspects of the ornithology of the Tristan da with recent estimates of thrush populations on Middle Cunha group and Gough Island, 1972–1974. Cormorant 12: 123–201. (5–10 pairs) and Stoltenhoff Islands (10–20 pairs; Royle J.A. & Dorazio R.M. 2012. Parameter-expanded data Ryan et al. 2011) and “hundreds” on Tristan da Cunha augmentation for Bayesian analysis of capture - recapture models. Island (BirdLife International 2017), we estimate the Journal of Ornithology 152: S521–S537. total population to be 8000–15,000 Tristan Thrushes. Ryan P.G. 2007. Field guide to the animals and plants of Tristan da Cunha and Gough Island. Newbury: Pisces Publications. Although this increases the global population, we Ryan P.G. & Dilley B.J. 2019. Intertidal foraging by Tristan Thrushes. recommend the species status should remain as “Near Ostrich 90: 179–181. Threatened ” globally. A more accurate estimate of the Ryan P.G., Glass N. & Ronconi R.A. 2011. The plants and birds of tiny population on Tristan is a conservation priority. Stoltenhoff and Mi ddle Islands, Tristan da Cunha. Polar Record 47: 86–89. Ryan P.G., Klicka L.B., Barker F.K. & Burns K.J. 2013. The origin of finches on Tristan da Cunha and Gough Is land, central South ACKNOWLEDGEMENTS Atlantic Ocean. Molecular Phylogenetics and Evolution 69: 299– We thank the Island Council and Administrator for Spiegelhalter D.J., Best N.G., Carlin B.R. & van der Linde A. 2002. approving our visit to Inaccessible Island. Maëlle Connan Bayesian measures of model complexity and fit. Journal of the Royal Statistical Society, Series B: Statistical Methodology 64: 583–616. and Erica Sommer gave assistance in the field. Funding Stervander M., Ryan P.G., Melo M. & Hansson B. 2018. The origin for land bird surveys at Tristan was received from Darwin of the world's smallest flightless bird, the Inaccessible Island Rail Plus. The South African National Antar ctic Programme Atlantisia rogersi (Aves: Rallidae). Molecular Phylogenetics and provided logistical support, and we are especially grateful Evolution 130: 92–98. Voelker G., Rohwer S., Bowie R.C.K. & Outlaw D.C. 2007. to the captain and crew of the “Geo Searcher” for getting Molecular systematics of a speciose, cosmopolitan songbird us safely off the island. We thank William R. Gould for genus: defining the limits of, and relationships among, the Turdus helpful discussions on the statistical analysis. Fitsum thrushes. Molecular Phylogenetics and Evolution 42: 422–434. Abadi thanks The New Mexico Agricultural Experiment Associate Editor: Lilian T. Manica. Station (NM Fitsum-2017H). Revista Brasileira de Ornitologia 27(4): 2019 Tristan Thrush movements and population size Ryan et al. APPENDIX I Table 1. The estimated population size (n) of Tristan Thrushes in the core study area as a function of M, the number of augmented individuals that had all-zero encounter histories. M n (95% credible interval) 50 193 (183–197) 100 233 (206–247) 150 256 (212–293) 200 258 (212–309) 300 258 (212–314) 500 259 (212–312) 1000 259 (214–313) & !   " Figure 1. The frequency distribution of the number of times color-ringed Tristan Thrushes were resighted at Inaccessible Island in 2018 (white = birds ringed in the core study area; grey = other areas). Figure 2. Posterior distributions of Tristan thrush abundance (N) from the best model (M ), under different values of M. The ht posterior distribution of N was not right truncated for M ≥ 200. Revista Brasileira de Ornitologia 27(4): 2019 & ! http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Ornithology Research Springer Journals

Short-term movement patterns, population estimates and breeding biology of an island endemic bird, the Tristan Thrush

Loading next page...
 
/lp/springer-journals/short-term-movement-patterns-population-estimates-and-breeding-biology-qLb3UVyT49
Publisher
Springer Journals
Copyright
Copyright © Sociedade Brasileira de Ornitologia 2019
eISSN
2178-7875
DOI
10.1007/bf03546070
Publisher site
See Article on Publisher Site

Abstract

Revista Brasileira de Ornitologia 27(4): 245–252. ARTICLE December 2019 Short-term movement patterns, population estimates and breeding biology of an island endemic bird, the Tristan Thrush 1,4 1 1 2 3 Peter G. Ryan , Ben J. Dilley , Delia Davies , Trevor Glass & Fitsum Abadi FitzPatrick Institute of African Ornithology, DST–NRF Centre of Excellence, University of Cape Town, South Africa. Tristan Conservation Department, Edinburgh of the Seven Seas, Tristan da Cunha, United Kingdom. Department of Fish, Wildlife and Conservation Ecology, College of Agricultural, Consumer and Environmental Sciences, New Mexico State University, Las Cruces, United States of America. Corresponding author: pryan31@gmail.com Received on 18 May 2019. Accepted on 22 October 2019. ABSTRACT: The Tristan Thrush Turdus eremita is the only land bird that survived human colonisation of the main island of Tristan da Cunha and is listed as “Near Threatened ”. Population estimates are confounded by the thrushes' inquisitive and gregarious nature -1 as well as limited information on their movements. We report the first measures of nest densities on Nightingale Island: 6 nests·ha -1 in Phylica arborea woodland and 4–5 nests·ha in tussock habitat, which suggests that the population is approximately double the previous estimate. At Inaccessible Island, we individually color ringed 110 thrushes over two months to track their short-term movements and estimate the local population size. Individuals moved up to 950 m along the coast, but 96% of resightings were < 100 m. A Bayesian data augmentation approach estimated that some 260 thrushes visited the core study area, a 200-m stretch of cobble and boulder beach where birds come to drink, bathe and forage. This result suggests that the population on Inaccessible Island also is substantially larger than reported previously. We estimate the total population to be 8000–15,000 Tristan Thrushes. The main need is a population estimate for the nominate subspecies on the main island of Tristan. KEY-WORDS: Turdus eremita, Inaccessible Island, Nightingale Island, Bayesian population estimate. INTRODUCTION and its offshore islets, Middle and Sto ltenhoff (Elliott 1957, Richardson 1984, Ryan et al. 2011). The species The Tristan Thrush or Star chy Turdus eremita is endemic is listed as “Near Threatened ” globally, and although the to the Tristan da Cunha Archipelago in the central South population is thought to be stable, there is a need for o o Atlantic Ocean (37 S; 12 W). Until recently it was placed up-to-date population estimates (BirdLife International in its own genus, Nesocichla, reflecting its morphological 2017). Current estimates are extrapolated from crude adaptations to life on oceanic islands: reduced wings, density estimates made by field workers. In the early robust legs and a brush-tipped tongue adapted for lapping 1970s, Richardson (1984) suggested that there were 40– up egg contents (Lowe 1923, Rand 1955, Fraser et al. 60 pairs at Tristan, 300–500 pairs at Nightingale Island, 1994). However, genetic sequence data suggest that it is 20–40 pairs at Middle, 10–20 pairs at Stoltenhoff and nested within the South American radiation of Turdus 100–500 pairs at Inaccessible Island. His estimate for thrushes (Klicka et al. 2005, Voelker et al. 2007), and like Inaccessible Island was particularly crude because he was the other land birds found at Tristan (Ryan et al. 2013, unable to spend much time on that island. Subsequent Stervander et al. 2018), it probably reached the islands in field work on Inaccessible Island in the 1980s improved the last few million years. the estimate there to 850 pairs (Fraser et al. 1994), giving The thrus h is the only land bird that survived the current total population estimate of around 6000 human colonisation of the main island of Tristan, where birds (BirdLife International 2017). Estimates to date a small population of the nominate subspecies persists have focused on breeding pairs because the thrushes' mainly above 300 m elevation on the steep coastal inquisitive and gregarious nature (e.g., Hagen 1952, scarps and upper plateau or “base” (Ryan 2007, BirdLife Fraser et al. 1994) confounds estimates of the density of International 2017). Thrushes are common at the two non-breeding individuals using transect or point counts. uninhabited islands in the archipelago: T. e. gordoni on We studied thrushes at two of the three islands Inaccessible Island and T. e. procax on Nightingale Island in the Tristan Archipelago: on Inaccessible Island, we Tristan Thrush movements and population size Ryan et al. thus the sex of at least some birds could be inferred. aimed to individually mark a sample of thrushes to Fledged chicks were observed from early November, but better understand their movements and estimate their no juveniles were ringed, and thus all data refer to birds population size; and on Nightingale Island we aimed to accurately estimate nest densities and gather data on at least 1 year old. Thrushes were mainly caught with breeding phenology over three seasons of fieldwork. hand nets, but a few birds were caught in mist nets set for finches. Thrushes initially received two color-rings Lastly, we aimed to use these data to refine the population as well as a metal ring, allowing them to be identified estimate of Tristan Thrushes in the ar chipelago. individually without recapture. However, we ran out of distinct color combinations after ringing 110 birds, and METHODS from 09 November, birds in the study area received a metal ring only (no further thrushes were ringed outside the study area). Ringing ceased on 23 November when Breeding biology and nest densities we ran out of metal rings, but resighting observations We recorded the contents of all nests found. Tristan continued until 25 November, the day before we left the Thrush nests are har d to locate during incubation and island. early chick stage (Fraser et al. 1994), but are conspicuous Ringing and resighting effort was concentrated along a 200-m long stretch of cobble and boulder once the chicks start to beg noisily when being fed. We beach backed by low tussock-covered cliffs centered on estimated the approximate age of chicks in nests from the nestling descriptions in Fraser et al. (1994). Authors Runaround Beach (Fig. 1). Several fresh-water springs B.J.D. and D.D. systematically recorded nests while and seeps emerge along the base of the cliffs in this conducting research on Nightingale Island in 2015 and area, attracting thrushes to drink and bathe. Thrushes also forage along the shoreline (Fraser et al. 1994, Ryan 2017. Additional ad hoc observations were made by P .G.R. & Dilley 2019), and are much easier to resight in this on Nightingale Island in October–November 2007 and Inaccessible Island in November 1999–February 2000, habitat than in the dense vegetation that covers most of September–December 2004 and October–December the island. We attempted to resight color-ringed Thrushes 2009. and catch unringed individuals at Runaround Beach on 20 days from 29 September to 25 November. Most visits In 2015 and 2017, thrush nests were recorded involved two observers, and lasted at least 2 h. On 19 systematically on Nightingale Island along the path from the huts to the Ponds (1.1 km) and in First Wood, a 4.3 and 25 November, visits lasted 4 h, and were divided into ha patch of Phylica arborea woodland, which was searched two sessions, recording individuals present in each 2-h intensively for nests. The pathway, which runs through period to give a total sample of 22 observation events. We calculated individual detection probabilities as the tall Spartina arundinacea Tussock grassland, was used proportion of observation periods when color-ringed daily to reach First Wood, so most nests were found that reached the large chick stage. To estimate nest densities birds were resighted after initial ringing (n = 12–21 in Tussock grassland, we assumed the path represented observation periods). a strip transect between 20–40 m wide (10–20 m either Most sighting effort occurred a long the coast. In addition to the main study area, the shoreline between side of the path, 2.2–4.4 ha). All nests were within 10 m Runaround Beach and Blenden Hall Bay, where there of the path, but the path created a favourable foraging area and might have biased the distribution of thrush nests is a small research hut (Fig. 1), was checked for ringed (e.g., several pairs of Brown Skua Catharacta antarctica thrushes on each visit to the study area. The coast between breed along the path, and thrushes often scavenge from Tern Rock and West Point (Fig. 1) also was searched on 55 days from 22 September to 26 November, and any carcasses of petrels killed by the skuas). We divided the ringed thrushes noted. Occasional visits also were made path into three sections: the lower 300 m runs through mostly level ground close to Northern Rockhopper to the accessible section of coast to Warren's Cliff, east of Penguin Eudyptes moseleyi colonies; the next 330 m is also Runaround Beach, and between West Point and Dirleton mostly level ground but lacks penguin colonies, and the Point (Fig. 1). On 01 December we also checked 1.7 km of the northeast coast between the Waterfall and north of final 470 m is on sloping terrain. Salt Beach. Resighting data at Inaccessible Island Systematic searches for ringed thrushes away from the coast were complicated by the dense vegetation, We ringed Tristan Thrushes on Inaccessible Island especially on the coastal slopes. The main paths followe d o o on the western side of the island, although the East Road (37 18'S; 12 41'W) from 22 September to 23 November was used much less than the West Road to access the 2018, which coincides with the species' breeding season. Only females develop well-defined brood patches, and plateau (Fig. 1). In addition, a few thrushes were ringed Revista Brasileira de Ornitologia 27(4): 2019 Tristan Thrush movements and population size Ryan et al. Figure 1. The Tristan Ar chipelago (A), Inaccessible Island (B), and main study area at Runaround Beach (black rectangle). Red lines show Tristan Thrush movements > 100 m (closed cir cle = ringing location; open circle = resighting location). th around our campsite on the eastern plateau at Denstone where Z = 1, if the i individual was a member of the o o Junction (37 17.61'S; 12 40.43'W). All distances moved population and 0 otherwise, and  is the inclusion and habitat areas were estimated from Google Earth. probability. The closed population (N) was then estimated as the sum of Z (i.e., N = Z ). Second, we modeled the i i Estimating thrush abundance encounter histories of each individual conditional on the latent variables (Z ) as Our initial plan was to mark 50 birds on the coast and 50 on the plateau in areas that would be visited frequently Y ~Bernoulli(Z p ) it i it over the two months we were on the island, but it soon th became apparent that the number of birds was greater where Y if the i individual was alive and detected at it than anticipated, and we focused most attention on a occasion t and 0 otherwise. p is the detection probability it th single coastal site where we attempted to ring as many of the i individual at occasion t. We imposed two possible thrushes as possible. Repeated visits to this site, coupled constraints on detection probability: time and individual with regular checks along the adjacent coastline, allowed heterogeneity (i.e., individual variation in detection us to gauge movement distances and use a Bayesian probability) (Table 1). Under M , M and M models, we t h ht approach to estimate the numbers of thrushes visiting used a random-effect approach to model t he time effect the main study site. We used closed population models and individual heterogeneity. Hence,  and  are time t i (Otis et al. 1978) to estimate abundance of the Tristan and individual random effects, respectively. Bir ds marked Thrush population at the core study area at Runaround with metal-rings only were given a detection probability Beach. All models were implemented using a Bayesian of zero after first capture, because there were no resighting data augmentation approach (Royle & Dorazio 2012), data for these individuals. boosting the observed 147 encounter histories with a large To implement the models in a Bayesian framework, number of individuals (M) that had all-zero encounter we specified vague prior distributions for all model histories. We used a hierarchical approach to describe parameters: U (0,1) for , N(0, 0.01) for  , and U(0, 10) the models. First, we used a Bernoulli distribution to for  and  . For each model, we ran three independent determine whether an individual was a member of the Markov Chain Monte Carlo (MCMC) iterations of true population exposed to sampling. That is, length 100,000 with a burn-in of 50,000, and a thinning rate of 50. We used the Brooks-Gelman-Rubin R statistic Z ~Bernoulli() (Brooks & Gelman 1998) to assess the convergence of the Revista Brasileira de Ornitologia 27(4): 2019 Tristan Thrush movements and population size Ryan et al. Table 1. Models fitted to the Tristan Thrush resighting data in the core study area at Inaccessible Island, and associated DIC and DIC scores. Variation in detection Model logit model for p DIC score DIC probability (p) logit(p ) =  +  + it 0 t i M Individual + time 2475.6 0.0 2 2 ht ~ N(0,  );  ~ N(0,  ) t  i logit(p ) =  + it 0 t M Time 2584.0 108.4 t 2 ~ N(0,  ) logit(p ) =  + it 0 i M Individual heterogeneity 2674.2 198.6 ~ N(0,  ) M Null logit(p ) =  2789.2 313.6 0 it 0 MCMC chains to the targeted posterior distributions. likely occur. Based on the inferred ages of chicks when Both the R statistic and visual inspections of the MCMC nests were found, most eggs hatched from early October plots indicated model convergence. The model with to late November. Our results suggest a crude density of -1 -1 the smallest Deviance Information Criterion (DIC; 3–10 nests·ha in tussock habitat and ~6 nests·ha in Spiegelhalter et al. 2002) was considered the best model. Phylica woodland. All analyses were performed using the JAGS software (Plummer 2003), called from R (R Core Team 2018) Movements at Inaccessible Island using the package jagsUI (Kellner 2018). Of the color-ringed thrushes, 71 were caught at the core study site, 30 were caught farther west along the coast RESULTS between Runaround Beach and Blenden Hall Bay or immediately inland around the hut (Fig. 1), and 9 were Nest densities at Nightingale Island caught on the plateau (seven at Denstone Junction) or high on the West Road (two). The resighting probability We found 60 nests over the study period. All thrush nests of color-ringed thrushes in the core study area ranged from along the path were in Tussock stands (n = 32). Nests 0–0.83, with 32% of birds observed on less than 10% of in First Wood (n = 28) were found in Tussock stands visits, and 85% seen on less than 50% of visits (Fig. 2). (9), Asplenium ferns (11), Carex sedge (5), in small rock Overall, the most common number of resightings was 0 crevices sheltered by hanging Scirpus grass (2), and in (Fig. 1 in Appendix I), even though most color ringing New Zealand Flax Phormium tenax (1). Flax is an invasive away from the study area ceased in mid-October, allowing species at Tristan, so this nest was moved ~ 1 m to an more than a month for ringed birds to be resighted. A adjacent Carex stand to allow the flax plant to be removed; few birds were seen regularly in the same area. Some of the female continued to incubate, and all three eggs these birds held breeding territories and subsequently hatched. Tussock nests were 0.2–1.4 m off the ground, were seen with fledglings, whereas others seemingly were but other nests were within 5–10 cm of the ground. Eight non-breeders that foraged along the shoreline (see Ryan eggs at 4 nests measured 30.7 ± 0.8 mm (29.3–31.6 mm) & Dilley 2019). Most non-breeders remained on more in length and 21.7 ± 0.6 mm (20.9–22.4 mm) in width. or less the same stretch of coast throughout, but one bird Clutch size at 9 nests with eggs averaged 2.1 ± 0.6 (1–3 moved ~120 m west midway through the study period, eggs), but some clutches may have been incomplete. remaining in this new site until we left the island. Average brood size at 55 nests was 2.4 ± 0.6 chicks, with The pattern of a few resident individuals and larger 51% of nests containing three chicks. The number of numbers of transient birds was obvious at the hut, where thrush nests found along the main path to the Ponds was catching and resighting effort was arguably greatest. 18 in 2015 and 14 in 2017, with densities tending to Fourteen thrushes were caught at the hut. Of these, one decrease with distance from the coast and elevation (Table pair of birds was resident, being seen virtually every day 2). A total of 28 nests were found in Phylica woodland at ranging 50 m between the hut and the adjacent coast; First Wood in 2017. We were unable to tell whether any this pair nested in dense tussock next to the beach in nests were repeat lays after breeding failure, but this is front of the hut. Of the remaining 12 thrushes, one was unlikely to greatly influence the estimate of the number seen regularly in Blenden Hall Bay, 90 m from the hut, of breeding pairs because few nests were found during two were resighted once in the core study area (~950 m incubation or the small chick stage, when most failures from the hut; resighted 28 and 32 days after ringing), one Revista Brasileira de Ornitologia 27(4): 2019 Tristan Thrush movements and population size Ryan et al. Table 2. Numbers of Tristan Thrush nests in Tussock grassland a long the path from the huts to First Wood on Nightingale Island in 2015 and 2017. Number of nests Density* Habitat Distance -1 (nests·ha ) 2015 2017 Level ground with 300 m 8 5 5.4–10.8 penguins Level ground, no 330 m 4 6 3.8–7.6 penguins Sloping ground 470 m 6 3 2.4–4.8 *assumes average number of nests and strip transect 20–40 m wide; see methods for further details. was seen once in Phylica trees 20 m behind the hut (44 data (Table 1). The posterior mean thrus h population days after ringing), and the other 8 were not seen again. visiting Runaround Beach under this model was ~258 A similar pattern also occurred at the plateau camp, birds (95% credible interval, CRI: 212–314), and this although only 7 thrushes were ringed here: 2 were seen estimate was largely independent of M, providing M on most visits to the camp, 1 was seen once (recaptured ≥ 200 (Table 1 in Appendix I, Fig. 2 in Appendix I). in a mist net the day after initial capture), and 4 were not The rate at which t hrushes were ringed at the study site seen again. showed no evidence of decreasing over time (Fig. 3), Most resightings were within 100 m of the original supporting the large population estimate for the site. ringing site (96%, n = 427 of 446). However, we frequently The posterior mean detection probability was 0.031 observed thrushes making flights > 100 m, mainly along (95% CRI: 0.016–0.053), and posterior estimates for the shore. One male that defended a territory at the  and  were 0.91 (0.63–1.29) and 1.92 (1.57–2.39), western edge of the study area frequently chased birds out respectively, confirming the considerable temporal and of sight along the coast to the west (at least 200 m), but individual variation in detection probability. was not observed outside the study area. The maximum movement distance recorded was 950 m, from the hut to the core study area (n = 2). Two other birds ringed in DISCUSSION Blenden Hall Bay moved as far as the study area (700– 900 m), and 1 ringed near Tern Rock also was seen in The main challenge to estimating the population size of Tristan Thrushes is their inquisitive nature. As Hagen the core study area (500 m). The farthest movement of birds ringed in the study area was to the Tern Rock area (1952) noted “they immediately fly to meet every new (400–500 m, n = 4 of 71 birds). No ringed birds were thing”, including people. As a result, any attempt to seen inland from the study area, but access to the dense estimate the population density by random transects Spartina tussock in this area was restricted to the East or point counts is confounded by the unknown radius over which thrushes are attracted. Individually marking Road, which was only visited a few times. In the more accessible Blenden Hall area, one thrush caught in a mist thrushes does not directly solve this problem, because the net midway between the hut and Wilkins' Copse, 130 m act of catching and ringing the birds makes them more inland (Fig. 1), was subsequently seen regularly on the cautious around people (Fraser et al. 1994, pers. obs.). adjacent coast. Current population estimates are crudely estimated from the approximate density of breeding pairs (Richardson Thrush numbers in the study area 1984, Fraser et al. 1994). Richardson (1984) estimated at Inaccessible Island 300–500 pairs on Nightingale Island, but our observations show that this is conservative. The population estimate for the core study area was based We present the first estimates of nest densities for Tristan Thrushes. On Nightingale Island, the estimate on encounter histories for 76 color-marked thrushes (71 -1 ringed at the site and 5 birds ringed along the adjacent for Phylica woodland at First Wood (6 nests·ha ) is coast as far west as the hut) and 71 metal-only ringed likely to be fairly accurate, as the entire area was searched birds caught at the study sight after we ran out of color intensively. It is likely that some thrushes re-lay if their combinations. Average resighting rate of the 76 ringed initial breeding attempt fails but this is unlikely to have inflated t he nest count because most nests were only thrushes in the 12 2-h observation periods in the core study area after color-ringing ceased was 29 ± 6% (21– recorded at the chick stage and there is no indication 38%). The model including individual heterogeneity and of double brooding (the breeding season is short, with a time effect (M ) provided the best fit to the resighting adults starting to moult from late November while still ht Revista Brasileira de Ornitologia 27(4): 2019 Tristan Thrush movements and population size Ryan et al. feeding fledged chicks; Fraser et al. 1994). Estimating this species. Despite ample opportunities for birds to be nest density for tussock habitat is complicated by the resighted, and high resighting rates for some individuals, fact that the path is not a random transect through this the most common result was for color-ringed birds to be habitat. However, incidental observations made while not seen again (Fig. 2 & Fig. 1 in Appendix I). More walking through tussock away from the path suggest observations are needed to assess whether such birds that the density of thrush nests along the path is not remain in areas of dense tussock, not readily accessible -1 atypical, and a density of 4–5 nests·ha is thought to to observers, or are more vagile. Until this conundrum is be a reasonable average across the island. Extrapolating resolved, we will struggle to derive accurate population these densities based on the total area of each habitat at estimates. Nightingale Island (160 ha tussock and 11 ha woodland) Despite the issue of possible transient birds, the suggests a population of some 700–850 breeding pairs, number of thrushes caught at the core study site at roughly double that estimated by Richardson (1984). Inaccessible Island was impressive, with no decrease in the This number of bree ding pairs suggests a total population rate at which unringed birds were caught over 2 months of 2000–3500 thrushes on Nightingale Island, including (Fig. 3). In fact, the proportion of unringed birds caught pre-breeders (few if any thrushes breed until they are at least two years old, Hagen 1952). Our breeding observations confirm that clutch sizes on Nightingale Island are consistently larger than those on Inaccessible Island. Roughly half of all broods on Nightingale were of three chicks, whereas three-egg clutches have not been recorded on Tristan or Inaccessible Island (Elliott 1957, Richardson 1984, Fraser et al. 1994, P.G.R. unpub. data). Quite what causes this difference is not known. Elliott (1957) reported that eggs of T. e. procax on Nightingale Island were larger than those of T. e. gordoni on Inaccessible Island, but our egg measurements at Nightingale averaged smaller (30.7 × 21.7 mm) than those reported by Elliott (1957) of 33.5 × 22.7 mm. Thrush densities are thought to be lower at Inaccessible than at Nightingale Island. Fraser et al. -1 (1994) estimated thrushes occurred at 1.0 pair·ha in -1 Figure 2. The probability that a Tristan Thrush color-ringed in tussock grassland, and 0.2–0.8 pairs·ha in three habitats the core study area was resighted in subsequent visits (white = on the island plateau. However, densities are hard to 0 probability; n = 71). assess, particularly along the coastal cliffs where nests are concealed in the densest, most inaccessible Tussock grass (Fraser et al. 1994), and as on Nightingale Island, probably are conservative estimates. We attempted to gain an independent estimate of thrush density by color- ringing a large sample of birds at Inaccessible Island. Our results support previous records on the movement of ringed Tristan Thrushes. The 36 thrus hes color-ringed at Inaccessible Island in 1982/83 were largely sedentary, with regular movements of around 250 m between the hut and Wilkins' Copse, and a maximum movement of 800 m (Fraser et al. 1994). Similarly, of the “nearly 100” thrushes Elliott (1957) ringed at the landing area on Nightingale Island, none was resighted more than 300 m away, suggesting they seldom undertake large movements. However, 2 of the 5 color-ringed birds resighted at Inaccessible Island from 1987–1989/90 #! $% &"! were observed only once over the three seasons (Fraser Figure 3. The cumulative number of Tristan Thrushes ringed at et al. 1994), suggesting that they are not regular visitors the core study area from 22 September to 23 November 2018. to areas accessible to observers. Our results confirm that Color-ringing ceased from 09 November, making it faster to individual variation in resighting rates is characteristic of process birds and thus increasing the catch rate. Revista Brasileira de Ornitologia 27(4): 2019 ! " ! Tristan Thrush movements and population size Ryan et al. decreased over the course of the study, as individuals REFERENCES became more wary of being approached with a hand BirdLife International. 2017. The IUCN red list of threatened species net – had this not been the case, even more birds would Turdus eremita. e.T22708541A117286443 (Access on 25 April have been caught later in the study. On the final day of 2019). observations (25 November) we estimated that there Brooks S.P. & Gelman A. 1998. General methods for monitoring were almost as many unringed birds as there were color- convergence of iterative simulation. Journal of Computational and Graphical Statistics 7: 434–455. ringed and metal-only birds, which lends credence to the Elliott H.F.I. 1957. A contribution to the ornithology of Tristan da Bayesian estimate of around 260 birds visiting the study Cunha group. Ibis 99: 545–586. site. If we assume that thrushes from the lower half of the Fraser M.W., Ryan P.G., Dean W.R.J. Briggs, D.J. & Moloney C.L. coastal cliffs (i.e., up to c. 200 m elevation) visit the core 1994. Biology of the Tristan Thrush Nesocichla eremita. Ostrich study site, and travel up to 500 m along the coast (likely a 65: 14–25. Hagen Y. 1952. The bir ds of Tristan da Cunha. Results of the Norwegian minimum estimate, given the limited movement of most Expedition to Tristan da Cunha 1937–1938. 20: 1–248. birds), then the population at the site represents birds from Kellner K. 2018. jagsUI: a wrapper around ‘rjags’ to streamline ‘JAGS’ a catchment area of around 20 ha. Increasing the radius analyses. R package version 1.5.0. along shore to the maximum movement recorded would Klicka J., Voelker G. & Spellman G.M. 2005. A molecular phylogenetic analysis of the “true thrushes” (Aves: Turdinae). increase this to around 40 ha. These catchment areas give -1 Molecular Phylogenetics and Evolution 34: 486–500. a density of 6.5–13 thrushes·ha , which if extrapolated Lowe P.R. 1923. Notes on some land birds of the Tristan da Cunha across the coastal scarp (500 ha), suggests a population group collected by the ‘Quest’ Expedition. Ibis 65: 511–528. of roughly 3250–6500 fully grown thrushes. The density Otis D.L., Burnham K.P., White G.C. & Anderson D.R. 1978. on the plateau (1000 ha) is perhaps 1/3 to 1/2 that along Statistical inference from capture data on closed animal populations. Wildlife Monographs 62: 1–135. the coast (Fraser et al. 1994), giving a crude population of Plummer M. 2003. JAGS: a program for analysis of Bayesian graphical 5500–11,000 thrushes on Inaccessible Island. rd models using Gibbs sampling. Vienna: Proceedings of the 3 Elliott (1957) does not specify the period over International Workshop on Distributed Statistical Computing which he ringed “nearly 100” thrushes in the immediate (DSC 2003). R Core Team 2018. R: a language and environment for statistical vicinity of the landing-place on Nightingale Island, but computing. Vienna: R Foundation for Statistical Computing. his experience mirrors ours on Inaccessible Island, and Rand A.L. 1955. The origin of the land birds of Tristan da Cunha. supports our inference of a large thrush population Fieldiana Zoology 37: 139–166. from nest densities on Nightingale Island. Combined Richardson M.E. 1984. Aspects of the ornithology of the Tristan da with recent estimates of thrush populations on Middle Cunha group and Gough Island, 1972–1974. Cormorant 12: 123–201. (5–10 pairs) and Stoltenhoff Islands (10–20 pairs; Royle J.A. & Dorazio R.M. 2012. Parameter-expanded data Ryan et al. 2011) and “hundreds” on Tristan da Cunha augmentation for Bayesian analysis of capture - recapture models. Island (BirdLife International 2017), we estimate the Journal of Ornithology 152: S521–S537. total population to be 8000–15,000 Tristan Thrushes. Ryan P.G. 2007. Field guide to the animals and plants of Tristan da Cunha and Gough Island. Newbury: Pisces Publications. Although this increases the global population, we Ryan P.G. & Dilley B.J. 2019. Intertidal foraging by Tristan Thrushes. recommend the species status should remain as “Near Ostrich 90: 179–181. Threatened ” globally. A more accurate estimate of the Ryan P.G., Glass N. & Ronconi R.A. 2011. The plants and birds of tiny population on Tristan is a conservation priority. Stoltenhoff and Mi ddle Islands, Tristan da Cunha. Polar Record 47: 86–89. Ryan P.G., Klicka L.B., Barker F.K. & Burns K.J. 2013. The origin of finches on Tristan da Cunha and Gough Is land, central South ACKNOWLEDGEMENTS Atlantic Ocean. Molecular Phylogenetics and Evolution 69: 299– We thank the Island Council and Administrator for Spiegelhalter D.J., Best N.G., Carlin B.R. & van der Linde A. 2002. approving our visit to Inaccessible Island. Maëlle Connan Bayesian measures of model complexity and fit. Journal of the Royal Statistical Society, Series B: Statistical Methodology 64: 583–616. and Erica Sommer gave assistance in the field. Funding Stervander M., Ryan P.G., Melo M. & Hansson B. 2018. The origin for land bird surveys at Tristan was received from Darwin of the world's smallest flightless bird, the Inaccessible Island Rail Plus. The South African National Antar ctic Programme Atlantisia rogersi (Aves: Rallidae). Molecular Phylogenetics and provided logistical support, and we are especially grateful Evolution 130: 92–98. Voelker G., Rohwer S., Bowie R.C.K. & Outlaw D.C. 2007. to the captain and crew of the “Geo Searcher” for getting Molecular systematics of a speciose, cosmopolitan songbird us safely off the island. We thank William R. Gould for genus: defining the limits of, and relationships among, the Turdus helpful discussions on the statistical analysis. Fitsum thrushes. Molecular Phylogenetics and Evolution 42: 422–434. Abadi thanks The New Mexico Agricultural Experiment Associate Editor: Lilian T. Manica. Station (NM Fitsum-2017H). Revista Brasileira de Ornitologia 27(4): 2019 Tristan Thrush movements and population size Ryan et al. APPENDIX I Table 1. The estimated population size (n) of Tristan Thrushes in the core study area as a function of M, the number of augmented individuals that had all-zero encounter histories. M n (95% credible interval) 50 193 (183–197) 100 233 (206–247) 150 256 (212–293) 200 258 (212–309) 300 258 (212–314) 500 259 (212–312) 1000 259 (214–313) & !   " Figure 1. The frequency distribution of the number of times color-ringed Tristan Thrushes were resighted at Inaccessible Island in 2018 (white = birds ringed in the core study area; grey = other areas). Figure 2. Posterior distributions of Tristan thrush abundance (N) from the best model (M ), under different values of M. The ht posterior distribution of N was not right truncated for M ≥ 200. Revista Brasileira de Ornitologia 27(4): 2019 & !

Journal

Ornithology ResearchSpringer Journals

Published: Dec 1, 2019

Keywords: Turdus eremita; Inaccessible Island; Nightingale Island; Bayesian population estimate

References