Introduction
Over the past century, global seabird populations have undergone catastrophic population collapses and it is estimated that the monitored seabird populations have declined by 69.7% since 1950 (Paleczny et al. Reference Paleczny, Hammill, Karpouzi and Pauly2015). Consequently, nearly half of all seabird species around the globe are currently in extinction crisis (Croxall et al. Reference Croxall, Butchart, Lascelles, Statersfield, Sullivan, Aymes and Taylor2012). Empirical evidence suggests that documented seabird declines are likely associated with various natural and anthropogenic factors such as predation by introduced predators, decline of prey stocks driven by climate change, harvesting, fisheries bycatch, competition with fisheries for shared prey, marine pollution, and loss of nesting habitats (Frederiksen et al. Reference Frederiksen, Wanless, Harris, Rothery and Wilson2004, Lewison et al. Reference Lewison, Oro, Godley, Underhill, Bearhop, Wilson, Ainley, Arcos, Boersma, Borboroglu, Boulinier, Frederiksen, Genovart, Gonzalez-Solis, Green, Gremillet, Hamer, Hilton, Hyrenbach, Martinez-Abrain, Montevecchi, Phillips, Ryan, Sagar, Sydeman, Wanless, Watanuki, Weimerskirch and Yorio2012, Spatz et al. Reference Spatz, Newton, Heinz, Tershy, Holmes, Butchart and Croll2014). Quantitative estimates of their population status and long-term trends represent an essential first step to inform conservation actions (Paleczny et al. Reference Paleczny, Hammill, Karpouzi and Pauly2015). To this end, many seabird population monitoring efforts have been made globally (Wooller et al. Reference Wooller, Bradley and Croxall1992, Anker-Nilssen and Røstad Reference Anker-Nilssen and Røstad1993, Anker-Nilssen Reference Anker-Nilssen2010).
The Japanese archipelago harbours 40 breeding seabird species including several endemic and rare species such as the recently rediscovered Bryan’s Shearwater Puffinus bryani (Kawakami et al. Reference Kawakami, Eda, Horikoshi, Suzuki, Chiba and Hiraoka2012) and re-described Bannerman’s Shearwater Puffinus bannermani (Kawakami et al. Reference Kawakami, Eda, Izumi, Horikoshi and Suzuki2018). Hence, Japan is a globally important seabird-rich region (Higuchi et al. Reference Higuchi, Minton and Katsura1995). Previous studies have roughly summarized the status of Japanese seabirds and suggested potential decreases/increases in species throughout Japan before 1984 (Hasegawa Reference Hasegawa1984) or locally before 2000 (Osa and Watanuki Reference Osa and Watanuki2002). However, due to the poor availability and accessibility of synthesized seabird breeding records, no nationwide assessment of population trends in Japanese seabirds has been conducted.
Recently, an extensive database covering a range of historical seabird breeding records throughout Japan was made available. This database synthesizes published records in the academic literature and unpublished records compiled by citizen observers and public monitoring projects. The “Monitoring Site 1000” project by the Japanese Ministry of the Environment contributed greatly to this database, with count data for 25 seabird species on a total of 77 islands since 2004 (Ministry of the Environment of Japan 2015). In the current study, we used this database to quantify nationwide population trends (i.e. annual population growth rates) in 10 major seabird species breeding in Japan over the past three decades. We employed a Bayesian state-space modelling approach because it can provide less biased estimates of demographic parameters by accounting for missing values and various types of observational errors (Kéry and Schaub Reference Kéry and Schaub2012), both of which are typical problems among long-term monitoring data. This study is the first to determine long-term population trends in seabirds breeding in Japan.
Materials and Methods
Japan seabird colony database
We created a database including information on 40 seabird species within Phaethontidae, Diomedeidae, Procellariidae, Hydrobatidae, Sulidae, Phalacrocoracidae, Laridae, and Alcidae, with 3,300 breeding records from a total of 389 colonies with exact locations throughout the Japanese archipelago (Figure 1). The dataset used in this study is based on the latest version of this database (January 2018). The data were collected by various contributors including both citizens and researchers using the following methods: for species that nest on open, flat ground (e.g. albatrosses and gulls), numbers of nests and/or adults during the incubation period were estimated. For most colonies, observers counted these metrics directly. For colonies having areas difficult to access, observers counted numbers of nests and/or adults in multiple plots within a colony, and then the averaged densities were multiplied by the total area of the colony. Both metrics were countable in most cases, but adult counts were often prioritised when nest counts were difficult to obtain. For burrow nesters (e.g. shearwaters, storm petrels, and several alcid species), a stratified sampling method was applied; nests were counted in plots of several square meters (plot sizes varied by species and colony) on multiple habitats with different ground structures (e.g. proportion of vegetation cover) within a colony. The estimated mean density for each habitat was then multiplied by the total area of respective habitat in the colony; these multiplied estimates were then summed. For species that nest on cliffs or rock-ledges (e.g. cormorants and murres), incubating or brooding adults (i.e. nests) were counted using panoramic photographs or sketches from vantage points covering the total area of a colony. For species that nest in crevices on cliffs or rock-ledges (e.g. puffins and guillemots), numbers of active nests were estimated by observing prey-provisioning adults from vantage points within a colony. We also counted adults on the sea near colonies, because these numbers can be positively related to the number of active nests. Hasebe et al. (Reference Hasebe, Fukuda, Senzaki and Watanuki2015) showed that the maximum number of Spectacled Guillemots observed on the sea during the early incubation period was moderately related to the number of active nests. For colonies where multiple counts were performed within a given year, we used the maximum numbers of adults and/or nests as representative.
Figure 1. Locations of seabird colonies in Japan. Smaller circles and larger squares indicate all colonies of the 40 species included in the Japan colony database and the focal colonies used in the population trend analysis, respectively.
Population trend analysis
Focal species and study timeframe
Of the 40 species in our database, we focused on 10 species for which we had a reasonable amount of data (Short-tailed Albatross Phoebastria albatrus, Leach’s Storm-petrel Oceanodroma leucorhoa, Japanese Cormorant Phalacrocorax capillatus, Pelagic Cormorant Phalacrocorax pelagicus, Black-tailed Gull Larus crassirostris, Slaty-backed Gull Larus schistisagus, Common Murre Uria aalge, Spectacled Guillemot Cepphus carbo, Tufted Puffin Fratercula cirrhata and Rhinoceros Auklet Cerorhinca monocerata. These species were monitored for > 4 years on at least one colony since 1980 using the methods described above (Figure 1, Table 1). Thus, the timeframe of this study was 36 years (1980–2015). Although several focal species (i.e. Short-tailed Albatross, Common Murre, Spectacled Guillemot, Tufted Puffin) are known to have declined by that time (Hasegawa and Degange Reference Hasegawa and Degange1982, Osa and Watanuki Reference Osa and Watanuki2002), the analysis from the year 1980 can provide a better understanding of how Japanese seabirds responded to recent environmental changes. We analysed a total of 18 colonies where single or multiple species breed (Figure 1). Several focal colonies are exposed to increasing predation threat recently from introduced predators such as rats Rattus spp. and/or feral cats Felis catus (e.g. Teuri Island, Rishiri Island, and Kabushima Island) (Tomita et al. Reference Tomita, Mizutani, Fujii, Sugiura, Yanai, Asano and Niizuma2010, Reference Tomita, Sato and Iwami2016), as well as native predators such as White-tailed Eagle Haliaeetus albicilla (e.g. Teuri Island and Daikoku-jima) (Ministry of the Environment of Japan 2015). Although some other colonies were not included in the analysis, our focal colonies accounted for a substantial portion of the current national population size of each species (Table 1). We therefore presume that our analysis represents the nationwide population trends of each species. Trends of the other 30 species and those of our focal species before 1980 were not estimated due to substantial data deficiencies within the database.
Table 1. Sample sizes for population trend analysis and estimated population sizes for each focal species. In the “sample size for analysis”, numbers of adults and numbers of nests represent sums of the latest counts for all focal colonies. Proportion represents the percentage of the respective national population size of each species. National population size represents the sum of the numbers of nests or adults among all colonies where surveys were conducted after 2000. CR, Critically Endangered; VU, Vulnerable; EN, Endangered; LC, Least Concern.
Bayesian state-space modeling
To estimate temporal trends in seabird populations, we employed a Bayesian state-space model (Kéry and Schaub Reference Kéry and Schaub2012). The state-space modelling approach is robust against missing values (Clark and Bjørnstad Reference Clark and Bjørnstad2004), which are typical in sparse long-term datasets like that used in this study (Terui et al. Reference Terui, Ishiyama, Urabe, Ono, Finlay and Nakamura2018). Furthermore, this modelling approach allowed us to isolate observation errors from state processes (i.e. true population dynamics), leading to less biased estimates of essential demographic parameters (e.g. population growth rates) (Clark and Bjørnstad Reference Clark and Bjørnstad2004, Kakinuma et al. Reference Kakinuma, Terui, Sasaki, Koyama, Jamsran, Okuro and Takeuchi2017). These separate models were created as follows.
Observation model
We assumed that nest (NEST tij) and adult counts (ADULT tij) in year t and colony i for species j followed a Poisson distribution:
$$NES{T_{tij}}\simPoisson\left( {exp\left( {log\,{n_{tij}} + \varepsilon {1_{tij}}} \right)} \right)$$
$$ADUL{T_{tij}}\simPoisson\left( {exp\left( {log\,{b_j} + log\,{n_{tij}} + \varepsilon {2_{tij}}} \right)} \right),$$
where n tij is the “true” nest abundance index, which is estimated using both nest and adult count data. The species-specific conversion parameter, b j, links the true nest abundance index to adult counts. The log-transformed conversion parameter, log b j, was drawn from a normal distribution with a hyper-mean of log B and variance of σb2. The parameters ε1tij and ε2tij are observation errors (year- and site-specific) drawn from normal distributions with a mean of 0 and variance of σnest2 or σadult2, respectively. The standard deviations, σnest or σadult, represent the degree of observation errors caused by either ecological or artificial factors (Kéry and Schaub Reference Kéry and Schaub2012). Thus, the model enables us to apply both nest and adult count information simultaneously, while accounting properly for the influence of unavoidable observation uncertainties. We incorporated the observation error terms (σnest or σadult) because the compiled dataset included observations conducted by various investigators and thus may involve artificial errors that could not be accommodated by a Poisson distribution. For species whose data are confined to either adult or nest abundance only (C. monocerata, U. aalge, P. pelagicus, P. albatrus, and O. leucorhoa), the species-specific conversion parameters b j were drawn randomly from a normal distribution with hyperparameters (log B and σb2). This imputation does not compromise species-specific information, because the model estimates demographic parameters mainly using available data for these focal species (either nest or adult counts).
Process model
We decomposed true nest abundance n tij into its temporal and spatial components. The temporal component captures population trends over observed periods, whereas the spatial component accounts for spatial (among-colony) variation in nest abundance (Amano et al. Reference Amano, Okamura, Carrizo and Sutherland2012). Specifically, nest abundance in year t at colony i was described as follows:
$$\log \,{n_{tij}} = {\alpha _{tj}} + {\beta _{ij}}.$$
In equation 4, αtj and βij are the temporal and spatial components of nest abundance, respectively, for species j. The spatial variation term βij accounts for among-colony variation in nest abundance and is assumed to follow a normal distribution with a mean of 0 and variance of σβ2.
We then employed a simple exponential growth model to describe seabird population dynamics:
$${\alpha _{\left( {t + 1} \right)j}} = \log \,{r_{tj}} + {\alpha _{tj}},$$
where the parameter log r tj is a species- and year-specific population growth rate (log-scale) drawn from a normal distribution as log r tj ∼ Normal (log r mean,j, σtemp,j2), where log r mean,j is the mean population growth rate for species j. The parameter σtemp,j2 governs the degree of temporal variation in log r tj. The parameter log r mean,j was drawn from a normal distribution with hyperparameters log r global (mean) and σr2 (variance). This hierarchical model structure increases parameter estimate accuracy, as data-poor species can borrow information from data-rich species (Gelman and Hill Reference Gelman and Hill2007, Hanioka et al. Reference Hanioka, Yamaura, Yamanaka, Senzaki, Kawamura, Terui and Nakamura2018).
Vague priors were assigned for the parameters: i.e. truncated normal distributions for σnest, σadult, σb, σβ, σtemp,j, and σr (mean = 0, variance = 1,000, range = 0–100) and normal distributions for log B, α1j, and log r global (mean = 0, variance = 1000). We fitted the model to the data using the JAGS ver. 4.1.0 software and the “runjags” package in R ver. 3.3.1 software, which allowed parallel computation of multiple Markov-chain Monte Carlo (MCMC) chains. Three MCMC chains were run for 22,500 steps with an initial 7,500 burn-in period, and 500 MCMC samples were stored for each chain. Convergence was assessed by determining whether the R-hat indicator of each parameter approached 1 (Gelman and Hill Reference Gelman and Hill2007).
Based on estimated nest abundances, we calculated population index as:
$$Population\,inde{x_{tj}} = \exp \left( {{\alpha _{tj}}} \right)/\exp \left( {{\alpha _{1j}}} \right),$$
where the denominator exp(α1j) is the estimated nest abundance in year 1980 for species j. Population indices provide relative measures of population trends in comparison to the base year 1980 (Amano et al. Reference Amano, Székely, Koyama, Amano and Sutherland2010). We also estimated averages of log-transformed population growth rates log r tj (i.e., geometric mean) over 10- (2005–2015), 20- (1995–2015), and 30-year (1985–2015) periods to reveal period-specific population trends.
Results
Four of 10 species (Black-tailed Gull, Slaty-backed Gull, Common Murre, and Tufted Puffin) declined significantly in at least one of the past 10-, 20-, or 30-year periods (Figure 2, Table 2; Table S1 in the online supplementary material). Two alcid species declined steeply during the past 30 years, decreasing to 3% (Common Murre) and 13% (Tufted Puffin) compared to the base year 1980 (Figure 2). Two gull species showed consistent decline throughout all three study periods (Tables 2, Table S1), decreasing to 28% (Black-tailed Gull) and 35% (Slaty-backed Gull) (Figure 2). In contrast, four species (Short-tailed Albatross, Pelagic Cormorant, Rhinoceros Auklet, and Spectacled Guillemot) increased significantly in at least one of the past 10-, 20-, or 30-year periods (Figure 2, Table 2, Table S1). No apparent global trends were observed in the other two species (Leach’s Storm Petrel, Japanese Cormorant) throughout the three study periods, but their wide 95% credible intervals suggested that their trends might not be detectable due to data paucity of these species (Figure 2, Table 2, Table S1).
Figure 2. Population indices estimated by the Bayesian state-space model. These indices are relative measures of population trends in comparison to the base year, 1980. Solid lines indicate median estimates; broken lines indicate associated 95% credible intervals.
Table 2. Population growth rates (log r tj) averaged for 10- (2005–2015), 20- (1995–2015), and 30-year (1985–2015) periods estimated by the Bayesian state-space model. Median values and 95% credible intervals of posterior distributions are shown. Significance indicated in bold.
Discussion
Current global efforts to conserve seabirds, including those in Japan, have almost exclusively focused on narrow-ranging rare species with the belief that those species are facing near-future extinction risk. However, the analysis performed using our extensive dataset suggests that, in addition to nationally threatened species (Common Murre and Tufted Puffin), wide-ranging common species (Black-tailed and Slaty-backed Gull) are also rapidly declining. The declines of common species may deserve attention because they are often dominant in an ecosystem and have effects on ecosystem function and associated ecosystem services (Elliott et al. Reference Elliott, Wilson, Taylor and Beggs2010, Inger et al. Reference Inger, Gregory, Duffy, Stott, Voříšek and Gaston2015). Although some colonies that were not included in the analysis might show different trends, our results should reflect their nationwide population trends because our focal colonies accounted for the bulk of the current national population size of each species (Table 1).
Our modelling approach integrated nest and adult count data to mitigate the influence of data scarcity on parameter estimates (see Methods). Nevertheless, our results must be interpreted with some caution. As with ordinary statistical approaches, significant population trends, if present, are less likely to be found for species with fewer data. In our dataset, this issue may be relevant for Leach’s Storm-petrel and Japanese Cormorant, for which we could not detect any population trends: this result might arise just because of the scarcity of the data (Table 2). The uncertainty of estimated trends in data-poor species can be improved by, at least in part, continued or increased monitoring efforts in the future. Therefore, incorporating further data into the current model may represent an important next step to better assess population trends of Japanese seabirds.
Drivers associated with the detected trends in focal species
Drivers associated with the detected trends in our focal species have not been fully addressed. However, the results of the current study allow us to suggest some contributing factors to these trends, which are not mutually exclusive.
Species with decreasing population size
The declines in two gull species over the past three decades might be associated with elevated predation by introduced and/or native predators. For example, many colonies of both gull species are exposed to increasing predation pressure from the native White-tailed Eagle due to its population expansion in northern Japan since the 1990s (Ministry of the Environment of Japan 2015). Predation by foxes and feral cats on adult Black-tailed Gulls has also been observed recently in several colonies (Tomita et al. Reference Tomita, Mizutani, Fujii, Sugiura, Yanai, Asano and Niizuma2010, Reference Tomita, Sato and Iwami2016).
The declines in gulls might also be linked to a reduction in prey availability associated with decadal-scale changes in atmospheric and ocean temperatures in the Pacific Ocean (Chavez et al. Reference Chavez, Ryan, Lluch-Cota and Niquen2003). For example, Japanese populations of the Japanese sardine Sardinops melanostictus, a main prey of Slaty-backed Gulls (Watanuki Reference Watanuki1988), collapsed in the late 1980s when local waters in the north-west Pacific warmed and became unproductive (Wada and Jacobson Reference Wada and Jacobson1998). Additionally, local stock of the sandlance Ammodytes personatus, which is the major diet of Black-tailed Gulls around Teuri Island, collapsed in the mid-1990s and never recovered. In parallel with these collapses, the population declines in Slaty-backed Gull and Black-tailed Gull have been observed on its historically largest colony (Daikokujima Island) and one of the largest colonies (Teuri Island), respectively (Figure S1).
The nationally threatened Common Murre and Tufted Puffin have experienced declines in the last 30-year period, but not in the past 10- and 20- year periods. Previous studies have suggested that drastic declines in both species occurred before our study timeframe, likely due to driftnet entanglement (Osa and Watanuki Reference Osa and Watanuki2002). The influence of driftnet entanglement on these species around Japan is underappreciated but is expected to become weaker in the future. For example, salmon driftnet fishery in the Sea of Japan, which is reported to severely impact the population of Common Murre on its largest national colony (Teuri Island) (Osa and Watanuki Reference Osa and Watanuki2002), had declined since the mid-1970s and the operation ended officially in 2011 (Mizusawa Reference Mizusawa2011). Moreover, driftnet fishery in Japan’s Exclusive Economic Zone in the north-west Pacific Ocean, which might have negative impacts on the populations of Tufted Puffin in eastern Hokkaido, has been declining gradually since the mid-1990s (Kataoka and Mizuno Reference Kataoka and Mizuno1999). However, despite the current implementation of conservation efforts for both species (Kataoka and Mizuno Reference Kataoka and Mizuno1999, Hasebe et al. Reference Hasebe, Aotsuka, Terasawa, Fukuda, Niimura, Watanabe, Watanuki and Ogi2012), our results suggest that their population sizes in Japan are considerably reduced and approaching extirpation, although the Common Murre population on Teuri Island has been reported to be gradually recovering (Hasebe et al. Reference Hasebe, Aotsuka, Terasawa, Fukuda, Niimura, Watanabe, Watanuki and Ogi2012).
Species with increasing population size
The positive population growth of the Short-tailed Albatross is consistent with that reported in previous studies (Finkelstein et al. Reference Finkelstein, Wolf, Goldman, Doak, Sievert, Balogh and Hasegawa2010), and may be attributable to intensive conservation efforts on the Torishima colony (e.g. nest site enhancements, decoy deployment) (Hasegawa and Degange Reference Hasegawa and Degange1982, Sato et al. Reference Sato, Momose, Tsurumi, Hiraoka, Mitamura and Baba1998, Melvin et al. Reference Melvin, Parrish, Dietrich and Hamel2003, Sato Reference Sato2009, Finkelstein et al. Reference Finkelstein, Wolf, Goldman, Doak, Sievert, Balogh and Hasegawa2010). Additionally, reduced mortality by fisheries bycatch in their wintering areas may be contributing to population improvement (Zador et al. Reference Zador, Punt and Parrish2008). Increasing trends among the other three species support the findings of previous local studies of the Pelagic Cormorant and Rhinoceros Auklet (Watanuki et al. Reference Watanuki, Kondo and Nakagawa1988, Osa and Watanuki Reference Osa and Watanuki2002) and of the largest Teuri Island Spectacled Guillemot colony (Hasebe et al. Reference Hasebe, Fukuda, Senzaki and Watanuki2015). It may also be possible that the positive trends in Rhinoceros Auklet and Spectacled Guillemot reflected to the recent declines in sympatric Black-tailed and Slaty-backed Gulls for two reasons. First, both Rhinoceros Auklet and Spectacled Guillemot might experience “predator release” on their colonies because gulls impede feeding rate of these Alcidae species as well as predate their eggs, chicks and adults (Finney et al. Reference Finney, Wanless, Harris and Monaghan2001, Senzaki et al. Reference Senzaki, Suzuki and Watanuki2014). Second, the declines in gulls might increase at-sea food availability of shared prey such as the sandlance and the Japanese anchovy Engraulis japonicus (Takahashi et al. Reference Takahashi, Kuroki, Niizuma, Kato, Saitoh and Watanuki2001, Yoda et al. Reference Yoda, Tomita, Mizutani, Narita and Niizuma2012). These factors might have positive effects on survival, reproduction, and ultimately population growth in Rhinoceros Auklet and Spectacled Guillemot. However, we acknowledge that the drastic Spectacled Guillemot decline occurred before our study timeframe (before 1980), and the current population size of this species is not still attaining historical levels (Senzaki et al. Reference Senzaki, Hasebe, Kataoka, Fukuda, Nishizawa and Watanuki2015).
Future research directions and implications for conserving Japanese seabirds
Despite increasing monitoring efforts for seabirds in Japan during the past several decades (Kazama et al. Reference Kazama, Ito, Niizuma, Sakurai, Takada, Sydeman, Croxall and Watanuki2010), we were unable to estimate population trends in most species breeding in Japan due to data deficiency and unreliability. Thus, the following issues must be addressed in future studies to better guide conservation efforts for Japanese seabirds.
First, the ongoing Monitoring Site 1000 project plays a pivotal role in seabird monitoring in Japan. Focal colonies are mainly selected as subjects for narrow-ranging rare species as well as many common and wide-ranging species. However, for some colonies of wide-ranging species, surveys or estimates of total numbers of nests have not been conducted due to lack of manpower or data deficiencies for the total breeding area (e.g. Brown Booby Sula leucogaster, Streaked Shearwater Calonectris leucomelas, and Wedge-tailed Shearwater Puffinus pacificus). Thus, in addition to the Monitoring Site 1000 project, it may be important to establish monitoring systems in cooperation with local governments, local conservation organisations, and other groups that conduct research at these breeding sites. Moreover, it is important to collect population demographic data such as adult survival and recruitment. These data allow us to estimate more reliable long-term persistence of each species (Genovart et al. Reference Genovart, Bécares, Igual, Martínez-Abraín, Escandell, Sánchez, Rodríguez, Arcos and Oro2018, Sanz-Aguilar et al. Reference Sanz-Aguilar, Massa, Lo Valvo, Oro, Minguez and Tavecchia2009, Reference Sanz-Aguilar, Igual, Tavecchia, Genovart and Oro2016).
Second, natural and/or anthropogenic drivers of common seabird declines must be identified. Although an array of factors have been suggested as potential threats, such environmental data are not available for most seabird colonies (Jones et al. Reference Jones2016; but see Emura et al. Reference Emura, Furuya, Ando and Deguchi2015). For example, seabird bycatch from commercial fisheries is believed to have substantial negative impacts on populations in many seabird species such as albatrosses and petrels (Croxall et al. Reference Croxall, Butchart, Lascelles, Statersfield, Sullivan, Aymes and Taylor2012). However, there are very limited data on seabird bycatch around Japan, especially in the coastal zone that most breeding seabirds in Japan depend heavily on for foraging, despite the fact that coastal fisheries such as fixed-net fishery operate commonly in the coastal zone across Japan (Ministry of Agriculture, Forestry and Fisheries 2019). This data deficiency limited our ability to quantify the impacts of environmental factors. Therefore, another important task is to monitor potential threats to seabird populations while compiling relevant existing data. Filling this gap will also allow us to examine why some species have declined while others do not.
Finally, applying modern technical or statistical methods to current monitoring programmes may facilitate the accurate estimation of population trends in species that are difficult to monitor using traditional methods. For example, we were unable to estimate trends in most nocturnal burrow-nesting species (e.g. Ancient Murrelet Synthliboramphus antiquus, Japanese Murrelet Synthliboramphus wumizusume, Band-rumped Storm-petrel Oceanodroma castro). However, Oppel et al. (Reference Oppel, Hervias, Oliveira, Pipa, Silve, Geraldes, Goh, Immler and McKown2014) estimated the population size of a nocturnal species (Cory’s Shearwater Calonectris borealis) at one of its colonies by combining acoustic monitoring and habitat mapping. Other examples include a mark–recapture method to correct nest occupancy rates (Sutherland and Dann Reference Sutherland and Dann2012) and aerial/satellite-based digital image surveys (Buckland et al. Reference Buckland, Burt, Rexstad, Mellor, Williams and Woodward2012, Fretwell et al. Reference Fretwell, LaRue, Morin, Kooyman, Wienecke, Ratcliffe, Fox, Fleming, Porter and Trathan2012). Nevertheless, given the limited budgets and time available for monitoring and practical conservation efforts, further discussion and empirical evidence are clearly needed to determine which research avenues and conservation targets, including those discussed above, should be prioritised (Brooke et al. Reference Brooke, Bonnaud, Dilley, Flint, Holmes, Jones, Provost, Rocamra, Ryan, Surman and Buxton2017).
Supplementary Material
To view supplementary material for this article, please visit https://doi.org/10.1017/S0959270919000352
Acknowledgements
We are grateful to all the researchers and citizen scientists who collected field data used in this study. We also thank H. Nakai, Y. Hamada, and the city of Izumo for providing data.
Data used in this study was compiled in accordance with the current laws of Japan and with relevant guidelines and regulations. This study was funded by a Japanese Society for the Promotion of Science (JSPS) KAKENHI grant (No. 17J00646). All data supporting this study are available online at http://www.sizenken.biodic.go.jp/seabirds/index.php (Environmental Agency of Japan).
