Nesting Survey

In 1993, we started collecting nesting data along 3.0 km of the Ibaruma beach on Ishigakijima Island, where sea turtle nests have been observed to be most abundant (Abe et al. 2004) (Fig. 1C). The nesting survey area gradually expanded to almost all beaches by 2002. Thus, nesting data collected from major beaches other than Ibaruma (Osaki [3.4 km], Hirakubo [1.3 km], Fusaki [2.1 km], Shiraho [7.0 km], and Yonehara [1.1 km] beaches) (Fig. 1C) were used for the analysis of the nesting trends on the entire Ishigakijima Island since 2002.

Nesting surveys were conducted during the day or night throughout the nesting season (that generally lasted from April to September) from 1993 to 2018. The surveys were conducted intermittently with various interval ranges that lasted anywhere between 1 d and about 1 mo. In this study, a nesting event was defined as the behavior that a female completed nesting activities (e.g., digging an egg chamber, laying eggs, and camouflaging the site), while a landing event included the behavior that a female came ashore but returned to the sea without any nesting success in addition to a nesting event. In the surveys, the numbers of landing and nesting events were counted; the date when those events occurred was estimated based on the conditions of crawl tracks. The nesting success was judged by whether a crawl track was accompanied by digging a body pit/egg chamber, abandoned nesting attempts, and emergence of hatchlings from the nests. The species of the turtles engaging in nesting activities was identified by the morphology of the nesting females, hatchlings, or adequately developed embryos in dead eggs and by the characteristics of the crawl tracks. Several typhoons passed by Ishigakijima Island every year. Thus, high tide caused by the typhoons washed out the traces of nesting sites and crawls. Therefore, in this study, we included the minimum count of sea turtle nesting or landing events.

If turtles were encountered during night patrols, the straight carapace length (SCL) of nesting individuals was measured. Then they were tagged internally using passive integrated transponder tags and externally using metal (inconel) and plastic tags on both hind flippers. These procedures took place after the turtles completed their nests or before they returned to sea after they had abandoned the nests. Moreover, a green turtle (SCL, 927 mm) nesting at Ibaruma beach was deployed with a Fastloc GPS-Argos tag (Mk10-AF, Wildlife Computers, Redmond, WA) in July 2014 and was tracked to examine the potential foraging habitat of green turtles nesting on Ishigakijima Island. The nest locations were often lost because high tide caused by typhoons washed out the nesting traces. The ratio of emergence success was investigated only for the nests the locations of which were still identified after hatchlings emerged.

Environmental Data

The mean sea surface temperature (SST) values during the annual nesting season for each turtle species (loggerhead: April–July; green and hawksbill: May–September) at the waters around Ishigakijima Island between 1982 and 2018 were obtained from the Japan Meteorological Agency database. Moreover, the mean air temperature was used to estimate the sand temperature in the clutch during the annual incubation period (loggerhead: April–August; green and hawksbill: May–October). The following conversion equation for beaches with light sand by Laloë et al. (2014) was used to estimate sand temperature from air temperature:

where T_sand represents the estimated sand temperature in the clutch incorporating metabolic heating (0.5°C, Laloë et al. 2014) and T_air represented the air temperature.

The annual data of Earth's magnetic field between 1980 and 2015 were used to compare the secular change in magnetic inclination at Ishigakijima Island and the nesting trend; these data were obtained by the Geospatial Information Authority of Japan.

Statistical Analysis

A generalized linear model (GLM) with a Poisson distribution and a log link function was used to determine the relationship between the number of nesting events in each species and SST. SST was treated as an explanatory variable (fixed effect). Moreover, GLMs with Gaussian and binomial distributions were used to examine the trends of the number of eggs in a single clutch and the emergence ratio of hatchlings over the research period, respectively. In this analysis, year was treated as an explanatory variable. Considering that data on nesting dates between 1993 and 2001 were available only for Ibaruma beach, the number of data was different between 1993–2001 and 2002–2018. Thus, the median date of all nesting events in the annual nesting season of each species was calculated to determine the long-term change of nesting dates on Ishigakijima Island. In this assessment, a GLM with Gaussian distribution was also used, and year was treated as an explanatory variable. We used the “lme4” package in the software R version 3.52 (R Development Core Team 2018) to run the GLM analyses. All data are reported as average ± standard deviation.

Nesting Trends

At Ibaruma beach, 1.8 ± 2.5 loggerhead (range, 0–9), 38.7 ± 22.1 green (range, 12–110), and 0.3 ± 0.6 hawksbill (range, 0–2) nests were laid annually from 1993 to 2018 (Fig. 2A, C, E). Meanwhile, the annual number of landing events for loggerhead, green, and hawksbill turtles was 3.4 ± 5.0 (range, 0–18), 94.5 ± 54.5 (range, 33–235), and 0.4 ± 0.6 (range, 0–2), respectively (Fig. 2A, C, E). The nesting trends showed that the number of loggerhead nests decreased drastically from 1993 to 1997. No loggerhead nest has been recorded at Ibraruma beach since 2013. Green turtle nests, on the other hand, increased gradually every year (Fig. 2A, C). Hawksbill turtles laid only a few nests every year (Fig. 2D).

View Figure

• Download as Powerpoint
• Download Figure

At the 6 major nesting beaches of Ishigakijima Island, 19.9 ± 14.3 loggerhead (range, 0–63), 76.1 ± 37.3 green (range, 37–170), and 3.0 ± 2.0 hawksbill (range, 0–6) nests were laid annually from 2002 to 2018 (Fig. 2B, D, F). Meanwhile, the number of landing events for loggerhead, green, and hawksbill turtles was 34.9 ± 23.7 (range, 1–97), 168.0 ± 71.8 (range, 68–306), and 5.1 ± 3.3 (range, 0–11), respectively (Fig. 2B, D, F). The nesting trends showed that the number of nesting loggerheads decreased after peaking at 2008 and then no nest was recorded at any of the major nesting beaches in 2018 (Fig. 2B); however, we counted 1 nest in a minor beach. The green turtles showed a similar nesting trend at Ibaruma beach, which gradually increased annually with considerable fluctuation (Fig. 2D); there was not a significant nesting trend for hawksbill turtles during the survey period (Fig. 2E).

The mean SSTs around the waters of Ishigakijima Island during the nesting seasons for loggerheads (April–July) and green and hawksbill turtles (May–September) between 1982 and 2018 were 26.9°C ± 0.3°C and 28.3°C ± 0.4°C, respectively. These temperatures gradually increased since 1982 (Fig. 3A). GLM analysis showed that the annual number of loggerhead nests decreased significantly between 2002 and 2018 with higher water temperatures (Wald test, n = 17, z = –2.56, p < 0.001; Fig. 3B); meanwhile, no temperature effect was observed for green (n = 17, z = 0.91, p = 0.36; Fig. 3C) and hawksbill (n = 17, z = –0.60, p = 0.55; Fig. 3D) turtles.

View Figure

• Download as Powerpoint
• Download Figure

Nesting Turtles

The mean SCLs were 872 ± 60 mm (n = 19), 979 ± 46 mm (n = 110), and 816 ± 29 mm (n = 11) for the loggerhead, green, and hawksbill turtles that we encountered during the surveys, respectively (Table 1). The tag identification of individuals revealed that green turtles nested 3.6 ± 1.1 times during a nesting season (Table 1). The mean remigration interval was 3.9 ± 1.2 yrs (Table 1). The turtle that we preliminary tagged in 1992 was encountered 7 times over 22 yrs (1992–2013). The remigration interval of this turtle was 3.5 ± 0.5 yrs. Meanwhile, the tagged loggerhead turtle was encountered only once at the next nesting event within a season. Furthermore, hawksbill turtles were not encountered again over the 26-yr research period (Table 1).

Table 1 Characteristics of 3 species of nesting sea turtles and their reproductive performance at Ishigakijima Island.^a

View Fullscreen

GPS-Argos tracking revealed that a nesting green turtle migrated to Iejima Island after her nesting season and stayed around there for 2 mo until the battery expired (Fig. 4). Two tagged nesting green turtles were found at the coastal area of Okinawajima Island (Fig. 4).

View Figure

• Download as Powerpoint
• Download Figure

Nesting Season

The nesting season of loggerhead turtles ranged from 31 March to 20 July, and the median was 23 May (Table 1; Fig. 5). In green turtles, the nesting season peak was about 1 mo after that of loggerheads (Fig. 5). Green turtle nests were observed mainly during the summer (May–September), while there were a few nests during the winter (November–January) (Table 1; Fig. 5). Hawksbills' peak nesting season was similar to that of green turtles as it ranged approximately from May to September (Table 1; Fig. 5).

View Figure

• Download as Powerpoint
• Download Figure

The median nesting date of loggerheads (Wald test, n = 25, t = 2.24, p < 0.05; Fig. 6A) and hawksbills (n = 19, t = 2.89, p < 0.05; Fig. 6C) was significantly delayed each year during the survey period (1993–2018); the nesting date of green turtles, on the other hand, did not shift significantly over the course of the research period (n = 26, t = –0.32, p = 0.76; Fig. 6B).

View Figure

• Download as Powerpoint
• Download Figure

Clutch Size, Emergence of Hatchlings, and Estimated Incubation Temperature

The number of eggs in a single clutch was similar between loggerhead (103.4 ± 23.3) and green (104.1 ± 22.2) turtles, while hawksbills laid more eggs (137.9 ± 26.3; Table 1). No significant trends were observed in the clutch sizes of loggerhead (Wald test, n = 261, t = –0.5, p = 0.61) and green (n = 844, t = –1.69, p = 0.09) turtles over the research period, but a significantly decreasing trend was observed in hawksbill turtles (n = 55, t = –2.6, p < 0.05).

The percentages of emergence success of hatchlings from the nest were 75.4% ± 23.5%, 80.0% ± 23.4%, and 84.2% ± 17.4% for loggerhead, green, and hawksbill turtles, respectively (Table 1). The presumed causes of emergence failure were the submergence of nests due to typhoons and the predation by ghost crabs (Ocypode spp.). No significant trends in emergence success were observed over the research period for all 3 species (loggerheads: Wald test, n = 251, z = –0.68, p = 0.50; greens: n = 804, z = 1.47, p = 0.14; hawksbills: n = 52, z = –0.40, p = 0.69).

The annual mean estimated sand temperatures in the clutch during the incubation period of loggerhead (April–August) and green and hawksbill (May–October) turtles were 30.1°C ± 0.5°C and 30.8°C ± 0.5°C between 1980 and 2018, respectively. These temperatures increased annually in a gradual manner, as was the case with SST (Fig. 7).

View Figure

• Download as Powerpoint
• Download Figure

Earth's Magnetic Field

The components of earth's magnetic field at Ishigakijima Island showed subtle annual changes between 1980 and 2015. The inclination angle increased gradually from 33.8° in 1980 to 35.5° in 2015. The total and vertical intensities of the magnetic field also increased from 43,889 and 24,396 nT in 1980 to 44,471 and 25,848 nT in 2015, respectively. Meanwhile, the horizontal intensity decreased gradually from 36,484 nT in 1980 to 36,184 nT in 2015.

Shift of Dominant Nesting Sea Turtle Species at Ishigakijima Island

Our nesting survey revealed that 3 sea turtle species showed different nesting trends at Ishigakijima Island over a quarter of a century; the dominant species among the nesting turtles was the green turtle. The dominant species on this island in the 1980s was the loggerhead turtle (Kamezaki 1991); therefore, the dominant species shift occurred from temperate loggerhead to tropical green turtles between the 1980s and 1990s. The same shift was reported at an adjacent island (Kuroshima Island) (Kondo et al. 2000).

Loggerhead Turtles

The carapace size of nesting loggerhead turtles was larger in the study area than it was in other major nesting beaches in Japan (Minabe and Yakushima Island) (Hatase et al. 2002, 2013). In Japanese nesting populations, larger turtles tend to be neritic foragers migrating to the continental shelf in the East China Sea to feed on benthic prey (Hatase et al. 2002, 2007). Therefore, loggerhead turtles nesting at Ishigakijima Island could have embarked on a foraging migration to the East China Sea after the nesting season.

Our results showed that the nesting population of loggerhead turtles on Ishigakijima Island declined from 1993 to 2005 but temporally increased between 2006 and 2008. This increasing trend corresponded to that of the entire nesting population of loggerheads in Japan (Kamezaki et al. 2003; Biodiversity Center of Japan 2016). However, although the entire Japanese loggerhead population increased substantially and continuously at least until 2012 (Biodiversity Center of Japan 2016), the Ishigakijima Island population decreased drastically after 2008 and became almost extinct in 2018. Thus, the decline of the Ishigakijima Island nesting population was, presumably, caused not by common factors affecting the entire Japanese population, such as the environmental condition of foraging grounds (Chaloupka et al. 2008b), but by a specific local factor. However, the recent population reduction (after 2013) could have been partially caused by a common factor because the entire Japanese population also decreased (Biodiversity Center of Japan 2016, 2019).

Ishigakijima Island is the southern (low-latitude) limit of the north Pacific population of loggerhead turtles' nesting distribution. Moreover, the SST around the Ishigakijima Island waters increased over the past decades, and this increment was significantly related to the reduction of nest numbers. Meanwhile, the emergence ratio of loggerhead hatchlings did not change over the research period. These facts may lead to a reliable hypothesis that the loggerheads that used to nest in Ishigakijima Island may not currently be able to access the waters around the island due to high temperatures; consequently, they shifted their nesting sites to regions that are more temperate. This hypothesis is supported by satellite tracking studies reporting that the range of SSTs that these turtles occupy in the central north Pacific is 15°C–25°C (Polovina et al. 2004); moreover, the maximum SST that juvenile and mature loggerhead turtles generally experience in the north Pacific is around 28°C (Kobayashi et al. 2008; Narazaki et al. 2015; J. Okuyama et al., unpubl. data, 2018). The SST during loggerheads' nesting season around Ishigakijima Island has been increasing annually and is currently close to 28°C (Fig. 3). This match between the SST around Ishigakijima Island and the upper SST range that loggerheads normally experience may have prevented them from nesting at Ishigakijima Island. Kondo et al. (2000) also suggested that there is a relationship between the decrease in nest numbers and SST at Kuroshima Island. A recent study on Mediterranean loggerhead populations reported the northward expansion of their nesting distribution and associated it with higher SSTs (Maffucci et al. 2016). Therefore, there is also the possibility that loggerheads previously nesting at Ishigakijima Island shifted their reproductive sites northward.

Brothers and Lohmann (2015) reported that the nest density of the north Atlantic loggerhead population varied with changes in the geomagnetic field because sea turtles locate their nesting beaches by using geomagnetic cues. In the case of Ishigakijima Island, all of the geomagnetic field components in the area changed subtly over the decades. However, all of the inclination and intensity isolines near Ishigakijima Island showed a southward shift. This geomagnetic shift direction was inverse to the shift of the nesting distribution of sea turtles, which followed a northward direction. Thus, the decline of the loggerhead nesting population at Ishigakijima Island was not likely due to the geomagnetic field shift.

The clutch size and emergence ratio did not change throughout the 26-yr period. Moreover, those values were within the general range of reproductive performance (Van Buskirk and Crowder 1994). The estimated incubation temperature in the clutches at Ishigakijima Island indicates that it would have probably produced a considerably female skewed sex ratio of hatchlings over the past 3 decades (according to the pivotal temperature of sex determination [28.6°C–29.7°C; Mrosovsky 1994]). Meanwhile, the incubation temperature has not yet reached lethal levels (31.6°C at the inflection point of survival rate; Matsuzawa et al. 2002); it has, however, been recently getting near the lethal line, although lethal incubation temperatures are quite different among nesting populations (Weber et al. 2011). Thus, further temperature increases may have a crucial impact on the nesting population of loggerhead turtles at Ishigakijima Island in the near future.

The increase in SSTs around the nesting beaches is reported to have caused the north Atlantic population of loggerheads to shift their nesting season a bit earlier to a more temperate period (Weishampel et al. 2004; Hawkes et al. 2007). The phenological shift of this species seems to be a reasonable response to climate change. However, our results showed a counterresponse in loggerhead turtles: the median nesting date moved roughly 20–30 d later and shifted to warmer dates over a quarter-century period. It is unknown why such a phenological shift occurred in loggerhead turtles. Further monitoring of nesting populations on Ishigakijima Island may help us understand the mechanism behind the delay in the nesting dates.

Green Turtles

In green turtles, there was no relationship between the annual number of nests and SST, unlike in loggerheads. The northern (high-latitude) limit of the nesting distribution of green turtles is a bit to the north of Ishigakijima Island (Biodiversity Center of Japan 2016). Thus, the water temperature around Ishigakijima Island was not likely cold enough to regulate the nesting distribution for green turtles.

The global population of nesting green turtles increased over the past 2–3 decades (Chaloupka et al. 2008a). Moreover, a recent increment in the nesting population of green turtles was also reported for the entire Japanese population (Biodiversity Center of Japan 2016; Kondo et al. 2017) and Lanyu Island, Taiwan, which is close to Ishigakijima Island (Cheng et al. 2018) (Fig. 1). At Ishigakijima Island, nesting green turtles showed natal philopatry (Nishizawa et al. 2011). Based on the assumption that age at maturity in green turtles is 30–40 yrs (Seminoff 2004), many green turtles born at Ishigakijima Island in the 1970s–1980s may have been recently recruited, thus contributing to the increase of the recent nesting population. Another possible reason for the recent increment is that some migrant turtles from the neighboring nesting populations might establish Ishigakijima Island as their nesting place. However, it should be noted that further warming beyond the lethal incubation temperature (31°C–33°C; Weber et al. 2011) may cause considerable reductions in hatchling emergence. We did not detect a nesting date shift in green turtles as in loggerheads. This may be because green turtle nesting sites may remain relatively fixed with regard to nesting beach temperatures even in a warming environment (Pike 2009).

GPS tracking and 2 reports of tag recapture indicate that the waters around Okinawajima Island may be one of the main foraging grounds for green turtles nesting at Ishigakijima Island. Indeed, coastal waters around Okinawajima provide several foraging grounds (Hayashi and Nishizawa 2015); the clutch frequency is slightly higher (2.93 ± 0.28 clutches/season; Van Buskirk and Crowder 1994), while the remigration interval is a bit lengthier than in other nesting populations (2.86 ± 0.23 yrs; Van Buskirk and Crowder 1994) but within the usual range (Mortimer and Carr 1987; Broderick et al. 2002).

Hawksbill Turtles

The time-series nesting data did not show a significant trend for hawksbill turtles. We detected their nests nearly every year in Ishigakijima Island, but the annual nest number was small. However, these nests tended to occur not at the same beach but sporadically at several beaches. This fact may indicate that hawksbill turtles on Ishigakijima Island did not have strong nest site fidelity based on the knowledge that they generally lay multiple egg clutches within a season (2.74 ± 0.22 clutches/season; Van Buskirk and Crowder 1994) and with strong site fidelity for nesting places (Kamel and Mrosovsky 2005). Ishigakijima Island is on the northern (high-latitude) peripheral nesting distribution of the northwestern Pacific population. The nests on this island could have been used not by hawksbill turtles returning to their natal home but by turtles migrating from neighboring sites (e.g., the Philippines) because of the lack of their orientation ability to find their natal beach. However, further studies may improve our understanding of the nest site selection of the peripheral population of hawksbill turtles.

To our knowledge, there have been no studies on the postnesting migration and potential foraging habitat of peripheral hawksbill turtle populations in the northern (high-latitude) nesting limit of the northwestern Pacific. Similarly, genetic characteristics of nesting hawksbills on Ishigakijima Island did not show a clear connectivity to potential foraging habitats (Nishizawa et al. 2012). Moreover, genetic diversity and a complex pattern of phylogeography of hawksbill turtles in the Indo-Pacific region made it difficult for researchers to understand how hawksbills expand their nesting distribution in the western Pacific and how they finally reach Ishigakijima Island (Okayama et al. 1999; Nishizawa et al. 2010, 2012; Vargas et al. 2015). Therefore, additional genetic and tracking studies may elucidate the linkage between Ishigakijima Island and neighbor nesting rookeries and the connection to potential foraging habitats.

Although Ishigakijima Island is on the northern peripheral nesting distribution, nesting performance characteristics, such as the number of eggs and hatchling emergence success, were similar to those reported in previous studies (Van Buskirk and Crowder 1994). Therefore, the beach conditions at Ishigakijima Island could be suitable for nesting hawksbill turtles. The hawksbill turtles showed the phenological shift of nesting date to a later period, as did loggerheads. It is unknown why such a phenological shift occurred in loggerhead and hawksbill but not green turtles. Further monitoring of nesting populations in Ishigakijima Island may help us understand the mechanism behind the delay in the nesting dates.