Introduction
Currently, threats to seabirds are increasing due to climate change (e.g. global warming) and direct anthropogenic activities, including the introduction of non-native species at breeding sites, bycatch, overfishing, microplastic pollution, and construction of offshore wind farms (Dias et al. Reference Dias, Martin, Pearmain, Burfield, Small and Phillips2019). In this context, it is essential to elucidate the spatiotemporal dynamics of seabirds for identifying the marine areas that are a significant part of their life-cycles, which serves as the foundational knowledge underpinning conservation efforts (Gonzáles-Solís and Shaffer Reference Gonzáles-Solís and Shaffer2009). However, the movements of seabirds are highly dynamic and variable. Although the global flyways of land- and coastal-migrating birds are well known, the global flyways of pelagic seabirds that travel long distances over the sea, especially small species, remain to be investigated. Recently, a project led by BirdLife International identified six major “marine flyways” for seabirds based on remote-tracking data for 48 species registered in the Seabird Tracking Database (https://www.seabirdtracking.org/) (BirdLife International 2023; Morten et al. Reference Morten, Beal, Bonnet-Lebrun, Carneiro, Dias and Rouyer2023, Reference Morten, Carneiro, Beal, Bonnet-Lebrun, Dias and Rouyer2025), covering less than 14% of the world’s approximately 360 seabird species, and including only three small-sized species (approximately ≤100 g). Moreover, although the most significant places for landbirds have been successfully identified under the Important Bird and Biodiversity Areas (IBAs) programme, also led by BirdLife International, the identification of marine IBAs remains on-going (BirdLife International 2009; Carr et al. Reference Carr, Trevail, Koldewey, Sherley, Wilkinson and Wood2022; Lascelles et al. Reference Lascelles, Taylor, Miller, Dias, Oppel and Torres2016). Therefore, it is crucial to continue gathering spatiotemporal data on a wide range of seabird species.
Remote tracking is a very useful and common methodology for investigating areas where seabirds stay throughout their annual life-cycle, including the migratory period. Technologies such as satellite transmitters, Global Positioning System (GPS) loggers, and light-level geolocators have facilitated the long-term tracking of large- and medium-sized seabirds. These technologies have not only helped elucidate the migration routes and important foraging areas of seabirds during their breeding and non-breeding seasons but also assisted in the discovery of their unique behaviours during their movements (Phalan et al. Reference Phalan, Phillips, Silk, Afanasyev, Fukuda and Fox2007; Lempidakis et al. Reference Lempidakis, Shepard, Ross, Matsumoto, Koyama and Takeuchi2022; Shaffer et al. Reference Shaffer, Tremblay, Weimerskirch, Scott, Thompson and Sagar2006; Trevail et al. Reference Trevail, Nicoll, Freeman, Le Corre, Schwarz and Jaeger2023). However, the long-term tracking of small seabirds remains challenging, as their body size limits the weight of tracking devices. Nonetheless, the recent downsizing of light-level geolocators (<1 g) has made it possible to track the smallest-sized seabird group, namely the Storm-petrel Hydrobatidae (Halpin et al. Reference Halpin, Pollet, Lee, Morgan and Carter2018; Medrano et al. Reference Medrano, Hernández-Montoya, Saldanha, Bedolla-Guzmán and González-Solís2024; Militão et al. Reference Militão, Sanz-Aguilar, Rotger and Ramos2022; Pollet et al. Reference Pollet, Hedd, Taylor, Montevecchi and Shutler2014, Reference Pollet, Ronconi, Leonard and Shutler2019).
In this study, we focused on a migratory small seabird, Swinhoe’s Storm-petrel Hydrobates monorhis, which breeds mainly on a limited number of islands in the north-western Pacific Ocean and its marginal seas, including those around Japan. As the population size of this species is declining (BirdLife International 2018), it is significant to know its fundamental movement ecology in detail. However, the migration route of the species and the connectivity between its breeding and wintering areas remain unknown. Information regarding its wintering range is also limited, although it is likely to winter primarily in the north-western Indian Ocean (Carboneras et al. Reference Carboneras, Jutglar, de Juana, Kirwan, del Hoyo, Sargatal, Christie and de Juana2021). In this study, we captured Swinhoe’s Storm-petrels at Koyajima (also known as Koyashima), Fukuoka, Japan, and tracked their movements using tiny light-level geolocators to elucidate their migratory routes, wintering areas, and activity patterns in migration and wintering periods.
Methods
Study species and study area
Swinhoe’s Storm-petrel is classified as “Near Threatened” on the International Union for Conservation of Nature (IUCN) Red List (BirdLife International 2018) and “Vulnerable” on the Red List of the Ministry of the Environment, Japan (Ministry of the Environment, Japan 2020). The main breeding islands of the species are remote islands in Japan, Korea, the Russian Far East, Taiwan, and China. Recent studies have also implied breeding in the northern Atlantic on an extremely small scale (Carboneras et al. Reference Carboneras, Jutglar, de Juana, Kirwan, del Hoyo, Sargatal, Christie and de Juana2021). It is often observed in the straits of Southeast Asia (Brickle Reference Brickle2017; Poole et al. Reference Poole, Davison and Rajathurai2014) and in the north-western Indian Ocean and its marginal seas during its non-breeding season (Baidya et al. Reference Baidya, Bhagat, Dharwadkar and Gauns2017; Bailey et al. Reference Bailey, Pocklington and Wrllrs1968; Campbell et al. Reference Campbell, Smiles, Roberts, Judas and Pedersen2017; Praveen Reference Praveen2013).
In Japan, recent evidence for active breeding of this species has been reported at only six sites (Sato Reference Sato2022; Sato et al. Reference Sato, Karino, Oshiro, Sugawa and Hirai2010), whereas the other three sites are known to be historical or potential breeding sites (Nakamura and Furunaka Reference Nakamura and Furunaka2022; Sato Reference Sato2022; Sato et al. Reference Sato, Karino, Oshiro, Sugawa and Hirai2010). Our study site was Koyajima (34°13′54N, 130°06′42E), Fukuoka, Japan, a small islet located approximately 54 km north-west of Kyushu Island and 1 km south of Okinoshima, with an area of 1.89 ha and the highest point at 29 m a.s.l. This islet has been a part of the World Cultural Heritage area “The Sacred Island of Okinoshima and Associated Sites” in the Munakata Region since 2017 (UNESCO World Heritage Centre 2017). The breeding population size of Swinhoe’s Storm-petrel was estimated to be 391 individuals in 1974 (Environmental Agency of Japan 1975), but was severely reduced by the invasion of non-native brown rats Rattus norvegicus in 1987 and 2009 (Sato Reference Sato2022; Takeishi Reference Takeishi1987). The rats were exterminated within a year of the first invasion and within two years of the second (Sato Reference Sato2022), and the bird population is on the path to recovery from depredation damage. A total of 93 individuals were banded there in two nights in the breeding season in 2023 (Biodiversity Centre of Japan 2024).
Capture and geolocator attachment
We captured 10 Swinhoe’s Storm-petrels at Koyajima using mist nets during the nights of 9–10 and 12–13 August in 2022 and attached rings and light-level geolocators (0.45 g, Intigeo-W30A9-SEA; Migrate Technology Ltd, Cambridge, UK) to them. The devices were set to record the maximum light level every 5 minutes, and immersion in saltwater (0 or 1) every 30 seconds, which were summarised into 10-minute intervals (range: 0–20). The geolocators were attached to a metal ring using duct tape (Tesa 4613; Tesa, Hamburg, Germany) and superglue. The total weight of the attachments was less than 2% of the body weight of each bird at capture. We attempted to retrieve geolocators using mist nets over two nights (21–23 July 2023). Birds were captured with permission from the Kyushu Regional Environment Office (permission numbers: 2201214 and 2305181) and the Munakata Taisha Shrine, with the latter being the islet owner.
Data analyses
The light data were retrieved from geolocators and analysed following Lisovski et al. (Reference Lisovski, Bauer, Briedis, Davidson, Dhanjal-Adams and Hallworth2020) using the free software R ver. 4.3.1 (R Core Team 2023). We used a light threshold of 1.35 lux to define sunrise and sunset transitions using the R package TwGeos (Lisovski et al. Reference Lisovski, Sumner and Wotherspoon2016) for twilight annotation. When the automatically determined sunrise and sunset times appeared to be inaccurate, likely because the light sensor was shaded by feathers or other obstacles, we manually adjusted the inferred times by checking the daily light curves. Subsequently, Markov Chain Monte Carlo (MCMC) simulations were used to estimate the geolocation of each tracked individual using the R package SGAT (Wotherspoon et al. Reference Wotherspoon, Sumner and Lisovski2013). We first calculated the zenith angle for zero and the median deviation under the gamma distribution based on the post-deployment “open sky” calibration period. We also assumed a mean movement speed of 12 km/hour following a gamma distribution (shape = 1.2, rate = 0.1) based on GPS tracking during the breeding season (Nakahara et al. unpublished). At the start of the MCMC simulations, we obtained the initial path of each individual using the function “thresholdPath”, setting the “tol” value to 0.05. In addition, we used a simple spatial mask to assign a lower prior probability to land than to sea, as Storm-petrels are normally pelagic (Carboneras et al. Reference Carboneras, Jutglar, de Juana, Kirwan, del Hoyo, Sargatal, Christie and de Juana2021). Geolocation estimation was performed using an Estelle model, which was run for 3,000 initial iterations, followed by five tuning steps of 300 iterations each, and finally run for 5,000 iterations to ensure convergence, from which the median geolocations were obtained. Estimating locations around the equinoxes is difficult owing to similar day lengths during these periods across the globe. Therefore, we excluded output locations within 14 days before and 25 days after the autumn equinox (9 September–18 October) and 25 days before and 14 days after the spring equinox (24 February–4 April) (defined as the autumn and spring equinox periods, i.e. AEP and SEP, respectively). The estimated geolocation data are accessible via the Seabird Tracking Database (ID: 2441).
The start and end dates of migration were defined when either of the following situations was satisfied: (1) just after the last no light record for over a day after breeding season or just before the first no light record for over a day before breeding season, or (2) beginning or end of the movement in one direction from the breeding or wintering sea areas. The absence of light recorded throughout the day likely indicated that the bird was in the nest on the breeding islet. If the start date of the autumn migration was not determined because of the autumn equinox, the date was defined as AEP. When the start day of spring migration was not determined because of the spring equinox, it was defined as SEP.
To identify the wintering sea area, we performed a fixed kernel density estimation (KDE) using the median geolocations over three months (from 15 November 2022 to 15 February 2023, 93 days) with the R package adehabitatHR (Calenge Reference Calenge2006). We calculated 10–95% KDE for each tracked bird and defined 95% KDE as its wintering area and 50% KDE as its core area. As the calculated KDEs partially included land areas owing to the generally large errors of geolocation estimated from light-level data (Phillips et al. Reference Phillips, Silk, Croxall, Afanasyev and Briggs2004), the portions of the KDEs overlapping with land were excluded from the wintering areas. Calculated areas were represented in km2 and rounded to the nearest 10,000.
To examine activity patterns across day/night and seasons, we calculated the diurnal and nocturnal immersion rates during the migrations and wintering periods for each individual. In this analysis, the inferred and manually adjusted times of sunrise and sunset were used to determine daytime and night-time. Wet and dry data summarised into 10-minute intervals were aggregated separately for day and night periods, although intervals that included sunrise or sunset times were removed from this calculation. The immersion rate was calculated from wet and dry counts for each day/night period across three seasons (autumn and spring migration and wintering season). When the beginning/end of the migrations was uncertain due to the equinox periods, we only used the days showing clear longitudinal movement. The wintering season was set as the same period of the KDE analysis. We fitted a generalised linear mixed model (GLMM) with a binomial error distribution and logit link to examine the effects on the immersion rate. The response variable was the wet and dry counts for each day/night period. Day/night and seasons and their interaction were included as fixed effects. Individual ID was included as a random effect. We performed the analyses using the R package lme4, and post-hoc multiple comparisons using the R package emmeans.
Results
Autumn migration
We recaptured 4 of the 10 individuals (IDs: CH693, CH695, CH699, and CH701) and successfully retrieved light-level data from their geolocators. Although the geolocations of the CH693 during the autumn migration period were suspected to be erroneous owing to sensor errors and were not included in the results, the geolocations of the other three birds during the autumn migration period and of all four individuals during the wintering and spring migration periods were available for analysis.
During the autumn migration, all tracked birds departed from the breeding areas between late August and the AEP (Table 1). However, the latitudinal data for either the first or second half of the migration routes of three birds could not be estimated because of a large overlap with the AEP. CH695 travelled south from Koyajima to the South China Sea and reached the Indian Ocean via the Sunda Strait before the AEP (Figure 1A). It most likely arrived in its wintering area during the latter part of the AEP. CH699 most likely departed from Koyajima during the AEP and moved west into the Arabian Sea after the AEP (Figure 1B). CH701 appears to have departed from Koyajima after being released, briefly stayed around the Ryukyu Islands, and then moved west into the Arabian Sea after the AEP (Figure 1C). For CH699 and CH701, both the start and end dates of migration were determined, with durations of 56 and 78 days, respectively (Table 1).
Table 1. Data of tracked Swinhoe’s Storm-petrels. AEP and SEP represent the autumn and spring equinox periods, respectively
* Calculated as the shortest distance between the breeding island and the centre of the wintering area.
Figure 1. Autumn migration routes of tracked Swinhoe’s Storm-petrels in 2022. A red star on each map indicates the capture site of the tracked individuals. Cross marks represent the estimated locations during the autumn equinox period (AEP), where latitudinal estimates are highly uncertain. When it is difficult to determine the exact departure or arrival date due to the equinox period, all estimated locations during that time are shown on the maps.
Spring migration
During the spring migration, all four birds departed from the Arabian Sea between the SEP and late April, although the actual start of migration was difficult to determine because of gradual southward movements during the late wintering period (Table 1 and Figure 2). They travelled eastward over the Indian Ocean and turned northward around the Sunda Islands (Figure 2A–D). The straits and sea areas crossed before reaching the Ryukyu Islands varied among the birds. CH693 passed through the Strait of Malacca and moved northward over the South China Sea (Figure 2A). CH695 crossed one of the straits of the Lesser Sunda Islands, the Makassar Strait, and the waters off the western coast of the Philippines (Figure 2B). CH699 travelled eastward through the Sunda Strait and Java Sea and then moved north between Sulawesi and New Guinea to the Philippine Sea (Figure 2C). CH701 likely passed through the Sunda Strait and moved north via the Makassar Strait to the waters off the western coast of the Philippines (Figure 2D). After reaching the Ryukyu Islands, all birds travelled along the archipelago and returned to their breeding sea areas (Figure 2A–D). The travel distance during the spring migration exceeded 13,000 km (Table 1). The durations of spring migration were identified for CH693 and CH701 and were 47 and 62 days, respectively, wherein gradual movement to the south at a small scale during the late wintering season was not included (Table 1 and Figure 2). No clear stop-over area for a long-term stay could be identified, although only CH699 showed a short-term stay at the Philippine Sea (Figure 2C).
Figure 2. Spring migration routes of tracked Swinhoe’s Storm-petrels in 2023. A red star on each map indicates the capture site of the tracked individuals. Cross marks represent the estimated locations during the spring equinox period (SEP), where latitudinal estimates are highly uncertain. All estimated locations during SEP are shown on the maps.
Wintering area
The wintering sea areas were in the Arabian Sea, located 6,656 ± 623 km (mean ± standard deviation) from the breeding colony (Table 1), and varied between individuals. CH693 and CH695 used areas off the western coast of India (Figure 3A and B), whereas CH699 primarily used areas around the Gulf of Oman (Figure 3C). CH701 used areas around the Gulf of Aden, ranging from the coast of the Somali Peninsula to the Arabian Peninsula (Figure 3D). The sizes of the wintering areas and their core areas varied among individuals, ranging from about 270,000 km² to 1,260,000 km² and 60,000 km² to 420,000 km², respectively (Table 1).
Figure 3. Wintering range of tracked Swinhoe’s Storm-petrels (15 November 2022 to 15 February 2023).
Immersion patterns
The interaction between day/night and season significantly affected immersion rates (GLMM, P <0.0001, see Supplementary material Table S1). Post-hoc tests revealed that immersion rates were significantly higher during daytime than at night across all seasons (Figure 4 and Table S2). However, the difference between daytime and night-time was greatest during the wintering season, followed by spring migration, and smallest during autumn migration (Figure 4 and Table S2). Diurnal immersion rates were highest during the wintering season, followed by spring migration, and then autumn migration (Figure 4 and Table S3). On the other hand, nocturnal immersion rates were also highest during wintering season, followed by autumn migration, and lowest during spring migration, although overall values were very low (Figure 4 and Table S3).
Figure 4. Means of the diurnal and nocturnal immersion rates with confidence intervals. The number shown below each bar is the number of day/night periods used in the analysis, which do not necessarily correspond to the total number of calendar days in each seasonal period.
Discussion
Our tracking data for Swinhoe’s Storm-petrel revealed large-scale movements across the north-western Pacific and Indian oceans. Although the Marine Flyways Project by BirdLife International identified six large Marine Flyways, movement patterns across multiple oceans have not yet been detected (BirdLife International 2023; Morten et al. Reference Morten, Carneiro, Beal, Bonnet-Lebrun, Dias and Rouyer2025). We also confirmed that Swinhoe’s Storm-petrels breeding in the north-western Pacific used a wide area of the Arabian Sea during winter. Although migratory connectivity was inferred from direct observational records, it was not previously confirmed. In this study, we highlighted the characteristics of the migratory route and wintering area and identified important marine areas for the conservation of Swinhoe’s Storm-petrel.
The migration pattern of Swinhoe’s Storm-petrel is characterised by large-scale movements not only in a north–south direction but also in an east–west direction. Although north–south movements are prevalent among migrants breeding in East Asia (Yong et al. Reference Yong, Heim, Chowdhury, Choi, Ktitorov and Kulikova2021), large-scale migrations that include east–west movements are uncommon, with a few exceptions. The east–west movement from East Asia to South Asia or Africa has been documented in several migratory landbirds, such as the Grey-headed Lapwing Vanellus cinereus, Common Cuckoo Cuculus canorus, Amur Falcon Falco amurensis, and an Asian subspecies of the Common Swift Apus apus pekinensis (Dixon et al. Reference Dixon, Batbayar and Purev-Ochir2011; Lee et al. Reference Lee, Kang, Lee, Kim, Jin and Bae2023; Lei et al. Reference Lei, Li, Kuang and Liu2021; Mellone Reference Mellone, Panuccio, Mellone and Agostini2021; Zhao et al. Reference Zhao, Zhao, Wu, Mu, Yu and Kearsley2022). Their migration pattern is linked to the detours of a geographical barrier stretching north-west to south-east, namely the Himalayas (Dixon et al. Reference Dixon, Batbayar and Purev-Ochir2011; Mellone Reference Mellone, Panuccio, Mellone and Agostini2021; Zhao et al. Reference Zhao, Zhao, Wu, Mu, Yu and Kearsley2022), and wind conditions during migration periods (Lee et al. Reference Lee, Kang, Lee, Kim, Jin and Bae2023; Mellone Reference Mellone, Panuccio, Mellone and Agostini2021). However, the migration pattern of Swinhoe’s Storm-petrel is different from that of landbird species, as the movement of Swinhoe’s Storm-petrel is entirely over the sea. It is reasonable to suggest that Swinhoe’s Storm-petrel does not choose an inland east–west migration route on the Eurasian Continent because of the difficulty to forage on land. Although detouring the Eurasian Continent increases the migration distance, it may not impose a severe energetic cost on the species, as pelagic seabirds can forage and rest anytime and anywhere at sea. This hypothesis is supported by the finding that the tracked individuals do not exhibit distinct stop-over areas.
It is difficult to explain the ultimate factor of the large-scale east–west movement, but the identification of the wintering area of the tracked Swinhoe’s Storm-petrel offers deep insight into why it uses the Arabian Sea far to the west of its breeding site. The Arabian Sea is likely a food-rich area for petrels because blooms of phytoplankton and green dinoflagellate Noctiluca scintillans are observed in winter. One of the major causes of this bloom is the convective north-east monsoon mixing of seawater and the supply of nitrate to the upper ocean (Wiggert et al. Reference Wiggert, Jones, Dickey, Brink, Weller and Marra2000). It occurs from November to March (Anjaneyan et al. Reference Anjaneyan, Kuttippurath, Kumar, Ali and Raman2023), which roughly coincides with the period during which the Swinhoe’s Storm-petrel remains in the Arabian Sea. As blooms lead to an increase in zooplankton and fish (Asha Devi et al. Reference Asha Devi, Vimalkumar, Padmakumar, Lathika, Maneesh and Sudhakar2021; Dwivedi et al. Reference Dwivedi, Chauhan, Solanki, Raman, Matondkar and Madhu2012), the Arabian Sea may be a suitable area for Swinhoe’s Storm-petrel to remain in the non-breeding season. The use of such productive sea areas has also been observed in Streaked Shearwaters Calonectris leucomelas breeding on a remote island in the East China Sea. This species is concentrated in a sea area approximately 400 km west of its colony for foraging, which is a nutrient-rich area derived from the Changjiang River plume (Matsumoto et al. Reference Matsumoto, Yamamoto, Kawabe, Ohshimo and Yoda2016).
Our data clarified the important sea areas used during migration and wintering as the first step toward the conservation of Swinhoe’s Storm-petrel. During its migration, the Swinhoe’s Storm-petrel changes direction from east–west to north–south in spring, and likely from north–south to east–west in autumn in Southeast Asia. In this region, it approaches the Malay Peninsula and the Sunda Islands, passing through narrow straits. These coastal sea areas are likely bottlenecks, where many petrels are concentrated during migration. In the future, it will be necessary to investigate how human activities are conducted and how they affect the movement of Swinhoe’s Storm-petrel through a relatively more accurate tracking of many individuals and more observations in the area. Moreover, it is essential to investigate the wintering ecology and resource availability of Swinhoe’s Storm-petrel in the changing Arabian Sea environment. It has been suggested that the weakening of the winter monsoon because of the warming of the Arabian Peninsula relative to the southern Indian Ocean may affect the dynamics of plankton blooms in the Arabian Sea, and the intensification of Noctiluca in plankton blooms could potentially disrupt the food web (Goes et al. Reference Goes, H.d.R, Al-Hashimi, Buranapratheprat, Glibert, Berdalet, Burford, Pitcher and Zhou2018; Parvathi et al. Reference Parvathi, Suresh, Lengaigne, Izumo and Vialard2017; Sarma et al. Reference Sarma, Baliarsingh, Pandi, Lotliker and Samanta2022). Monitoring how such changes affect the life of Swinhoe’s Storm-petrel is among the next key issues.
Immersion patterns showed that Swinhoe’s Storm-petrel tends to be nocturnal during non-breeding seasons. This tendency is consistent with patterns observed in other Hydrobatidae species, such as the European Storm-petrel H. pelagicus, Townsend’s Storm-petrel H. socorroensis, and Ainley’s Storm-petrel H. cheimomnestes (Medrano et al. Reference Medrano, Hernández-Montoya, Saldanha, Bedolla-Guzmán and González-Solís2024; Militão et al. Reference Militão, Sanz-Aguilar, Rotger and Ramos2022). However, flying activity increased during migration periods, particularly during the daytime, suggesting that migratory movements occur not only at night but also during the day. Moreover, diurnal migratory movements were longer during autumn migration than in spring migration. There may be variation in energy costs between migration periods, potentially influenced by factors such as wind conditions (Thorne et al. Reference Thorne, Clay, Phillips, Silvers and Wakefield2023). It may be important to consider the various environmental factors during migration periods for understanding the navigational abilities of Swinhoe’s Storm-petrel.
Tracked Swinhoe’s Storm-petrel (mean ± SD = 44.9 ± 2.6 g) showed large-scale seasonal movements over several thousand kilometres similar to the movements previously observed in other Hydrobatidae species. The distance from the breeding colony to the wintering area was longer than that observed for smaller sized species, the European Storm-petrel (30.3 ± 1.9 g) moving from Benidorm Island in the Mediterranean Sea to the northern Atlantic (Militão et al. Reference Militão, Sanz-Aguilar, Rotger and Ramos2022), and Townsend’s Storm-petrel (33.7 ± 4.1 g), and Ainley’s Storm-petrel (38.2 ± 13.3 g) moving from Morro Prieto Islet off the Baja California Peninsula to South America and North Hawaii, respectively (Medrano et al. Reference Medrano, Hernández-Montoya, Saldanha, Bedolla-Guzmán and González-Solís2024). In fact, it was even longer than that observed for a larger species, the Fork-tailed Storm-petrel H. furcatus (55.1 ± 3.8 g), moving from Gillam Island in Canada to the north-eastern Pacific (Halpin et al. Reference Halpin, Pollet, Lee, Morgan and Carter2018). However, this distance was similar to, or shorter than, the distances observed in a similarly sized species, Leach’s Storm-petrel H. leucorhous (approximately 41.9–48.1 g) moving from Bon Portage Island in Canada to the South Atlantic or from Gillam Island to the south-eastern Pacific (Halpin et al. Reference Halpin, Pollet, Lee, Morgan and Carter2018; Pollet et al. Reference Pollet, Ronconi, Leonard and Shutler2019). This small-sized seabird taxon is likely capable of flying relatively long distances over the sea, similarly to that observed for other taxa in the Procellariiformes. Moreover, the body size of Hydrobatidae may be associated with migration distances. However, knowledge of the movement patterns of Hydrobatidae species is still lacking, especially among species breeding in the north-western Pacific region, where six species, including Swinhoe’s Storm-petrel, breed. Further studies and the acquisition of more tracking data for this taxon are required to avoid underestimating the risk of decline.
Supplementary material
The supplementary material for this article can be found at http://doi.org/10.1017/S095927092510021X.
Acknowledgements
We are grateful to the staff of the Munakata Taisha Shrine, offices for the World Heritage areas of Munakata City and Fukuoka Prefecture, and the Kyushu Regional Environment Office for their generous support. We also thank Masayoshi Takeishi, who provided us with much information on the Swinhoe’s Storm-petrel breeding in Koyajima; Seiya Nishi, who supported our field surveys; Yoshinobu Miyasaka, the captain of the fishing boat Ebisu-maru, who helped us visit Koyajima; Akiko Shoji and the laboratory students, who told us how to attach a geolocator. During the preparation of this paper, we used ChatGPT to enhance language and improve readability. After using this service, we have carefully reviewed and edited the content as needed, and we take full responsibility for the content of the publication. This study was supported in part by a grant-in-aid from the Inui Memorial Trust for Research on Animal Science, Japan, and a grant from the Nihon Seimei Foundation for Environmental Issues.
