Potential limitations of behavioral plasticity and the role of egg relocation in climate change mitigation for a thermally sensitive endangered species

Abstract Anthropogenic climate change is widely considered a major threat to global biodiversity, such that the ability of a species to adapt will determine its likelihood of survival. Egg‐burying reptiles that exhibit temperature‐dependent sex determination, such as critically endangered hawksbill turtles (Eretmochelys imbricata), are particularly vulnerable to changes in thermal regimes because nest temperatures affect offspring sex, fitness, and survival. It is unclear whether hawksbills possess sufficient behavioral plasticity of nesting traits (i.e., redistribution of nesting range, shift in nesting phenology, changes in nest‐site selection, and adjustment of nest depth) to persist within their climatic niche or whether accelerated changes in thermal conditions of nesting beaches will outpace phenotypic adaption and require human intervention. For these reasons, we estimated sex ratios and physical condition of hatchling hawksbills under natural and manipulated conditions and generated and analyzed thermal profiles of hawksbill nest environments within highly threatened mangrove ecosystems at Bahía de Jiquilisco, El Salvador, and Estero Padre Ramos, Nicaragua. Hawksbill clutches protected in situ at both sites incubated at higher temperatures, yielded lower hatching success, produced a higher percentage of female hatchlings, and produced less fit offspring than clutches relocated to hatcheries. We detected cooler sand temperatures in woody vegetation (i.e., coastal forest and small‐scale plantations of fruit trees) and hatcheries than in other monitored nest environments, with higher temperatures at the deeper depth. Our findings indicate that mangrove ecosystems present a number of biophysical (e.g., insular nesting beaches and shallow water table) and human‐induced (e.g., physical barriers and deforestation) constraints that, when coupled with the unique life history of hawksbills in this region, may limit behavioral compensatory responses by the species to projected temperature increases at nesting beaches. We contend that egg relocation can contribute significantly to recovery efforts in a changing climate under appropriate circumstances.


| INTRODUC TI ON
Anthropogenic climate change is widely considered a major threat to global biodiversity (Foden et al., 2013;Parmesan & Yohe, 2003;Poloczanska et al., 2013), with 15%-37% of Earth's species potentially "committed to extinction" by 2050 (Thomas et al., 2004). The ability of a species to exhibit compensatory responses to climatedriven environmental changes will determine its likelihood of survival; species more able to adjust to new environments or adapt to local climatic conditions will have a greater likelihood of persisting than those that cannot (Sinervo et al., 2010). Because the influence of climate change can vary among taxa and geographic regions (Parmesan, 2007), species may adapt in a variety of ways to mitigate unfavorable conditions (Bellard, Bertelsmeier, Leadley, Thuiller, & Courchamp, 2012), including evolutionary changes (Shefferson, Mizuta, & Hutchings, 2017) and spatiotemporal shifts in behavior (Chen, Hill, Ohlemüller, Roy, & Thomas, 2011;Yang & Rudolf, 2010).
For instance, maternal nest-site choice can compensate for climatic variation among populations of the Australian water dragon (Physignathus lesueurii; Doody et al., 2006). Similarly, behavioral plasticity in painted turtles (Chrysemys picta bellii) can allow females to match shade cover over nests with prevailing environmental conditions to influence the sex ratio of offspring (Refsnider & Janzen, 2012).
Given their complex life histories and reliance on marine and terrestrial habitats during their lifecycle, it is unclear how sea turtles will respond to climate-driven change in these environments.
Given potential limitations of plastic compensatory responses of sea turtles to accelerated changes in thermal conditions of nesting beaches, it is possible that sea turtles will be unable to adapt quickly enough to offset negative consequences to population demographics. In such cases, human intervention may be required to ensure population persistence. Relocation of sea turtle eggs as a management strategy used to increase hatchling production and enhance population recovery is ubiquitous worldwide (Chacón-Chaverri & Eckert, 2007;Formia, Tiwari, Fretey, & Billes, 2003;García, Ceballos, & Adaya, 2003;Naro-Maciel, Mrosovsky, & Marcovaldi, 1999;Patino-Martinez, Marco, Quinones, & Hawkes, 2012b). By utilizing internationally recognized best practices throughout the egg relocation process (Eckert et al., 1999), many of the concerns about possible undesired biological outcomes (Mrosovsky, 2006;Pilcher & Enderby, 2001;Prichard, 1980) can be avoided or mitigated (Kornaraki, Matossian, Mazaris, K E Y W O R D S egg relocation, environmental policy, mangrove estuaries, nest-site selection, reproductive behavior, sand temperature, sea turtle, sea-level rise, species redistribution, temperaturedependent sex determination Matsinos, & Margaritoulis, 2006;Patino-Martinez, Marco, Quinones, Abella, et al., 2012a). Because temperatures are predicted to increase substantively in Central America over a relatively short period, the influence of sea turtle egg relocation on the thermal regimes of nest environments, primary sex ratios, and hatchling fitness compared with in situ clutches is a top research priority, particularly for severely depleted populations of highly endangered species.

Critically endangered hawksbill turtles (Eretmochelys imbricata)
in the eastern Pacific Ocean belong to one of the least resilient (Fuentes, Pike, Dimatteo, & Wallace, 2013) and most threatened marine turtle regional management units (RMU) in the world (Wallace et al., 2011), with fewer than 700 adult females nesting along 15,000 km of Latin American coastline (Gaos et al., 2017). Further, >70% of this nesting activity is concentrated on low-relief beaches in mangrove estuaries at Bahía de Jiquilisco in El Salvador and Estero Padre Ramos in Nicaragua (Gaos et al., 2017;Liles, Peterson, Seminoff, et al., 2015b)-ecosystems that are particularly vulnerable to increasing global temperatures and sea-level rise (Gilman, Ellison, Duke, & Field, 2008).
In this study, we investigated whether behavioral plasticity in this species is likely to be able to compensate for projected climate change and what the role of egg relocation may be as a mitigation strategy. The objectives of our study were to (a) estimate sex ratios and physical condition of hatchling hawksbills under natural and manipulated conditions ( Figure 1) and (b) generate and analyze thermal profiles of nest environments. Our results provide the first empirical assessment of the efficacy of nest protection strategies for this severely depleted RMU. Based on our findings, we offer recommendations for mitigation strategies that complement potential plastic adaptive responses to climate change demonstrated by nesting hawksbills in mangrove ecosystems.

| MATERIAL S AND ME THODS
Our study was conducted at Bahía de Jiquilisco (13°13′N, 88°32′W) in El Salvador and Estero Padre Ramos (12°48′N, 87°28′W) in Nicaragua, which are located on the western and eastern borders of Gulf of Fonseca on the Pacific coast of Central America, respectively ( Figure 2). Hawksbill nesting occurs primarily during the rainy season between May and September, with a peak in June and July. Contrary to typical contiguous open-coast beaches used by nesting hawksbills in other oceanic regions (Loop, Miller, & Limpus, 1995;Mrosovsky, 2006), hawksbills at these two sites nest on low-relief beaches scattered within mangrove estuaries (Gaos et al., 2017;Liles, Peterson, Seminoff, et al., 2015b).
Bahía de Jiquilisco is located on the south-central coast of El Salvador and has hawksbill nesting habitat (42.1 km) comprised of eight distinct fine-grained sand beaches with three hatcheries and one in situ nest protection area ( Figure 2). A fragmented mosaic of second-growth coastal forest and small-scale fruit tree plantations 10-15 m wide from the high water line is present at most nesting beaches (Liles, Peterson, Seminoff, et al., 2015b). Moderate development exists in some nesting areas, particularly along eastern and western Punta San Juan, eastern and western Isla Madresal, and northern Isla San Sebastian.
Estero Padre Ramos is situated on the northwestern Pacific coast of Nicaragua and consists of eight distinct fine-grained sand beaches (12.8 km), with one hatchery and one in situ nest protection area ( Figure 2). Intact secondary coastal forest extends >100 m landward from the high water line at most beaches (Liles, Peterson, Seminoff, et al., 2015b (Liles et al., 2016;Liles, Peterson, Lincoln, et al., 2015a). Consequently, conservation organizations purchase eggs encountered and/or collected by local residents for protection to prevent their sale   We relocated all clutches deposited at other beaches to a hatchery (Table 1), except during 2010 and 2011, when some clutches were relocated to an area of beach near the hatchery because the hatchery had reached capacity or was not yet operational.
For clutches relocated on the beach or to a hatchery, we measured the dimensions of original nest cavities and attempted to emulate these dimensions in artificial nests. We relocated most clutches <12 hr after deposition to minimize movement-induced mortality during transfer and reburial (Limpus, Baker, & Miller, 1979).

| Hatchling sex ratios and physical condition
Although direct methods for estimating hatchling sex ratios, such as histological evaluation of gonads, are highly accurate for sexing individual hatchlings, they are logistically infeasible to perform on endangered species. Indirect methods-including nest temperature and incubation duration-are reliable proxies when direct methods are infeasible (Wibbels, 2003). Because financial and logistical constraints prohibited us from recording nest temperatures at Estero Padre Ramos in 2010-2011, we used incubation duration values obtained for offspring-producing nests to estimate primary sex ratios at both sites to provide results that are commensurable across sites and among years.
We used published data for hawksbills that related incubation duration to sex ratio based on constant temperature incubator experiments to convert the incubation duration of each clutch into hatchling sex ratio (Godfrey et al., 1999). For incubation duration calculations, the incubation period was calculated as the number of days between the date and hour of clutch deposition and the date and hour of first hatchling emergence. For nests where the date of emergence was unavailable (n = 50 nests, 2.5% of total) or where no hatchlings emerged but were found alive during exhumation (n = 30 nests, 1.5% of total), we used the average incubation duration of the nest protected using the same strategy immediately before and after the nest without date of emergence or with live hatchlings that did not emerge. We used a one-to four-day correction factor for the hatching-to-emergence interval in overall hatchling sex ratio calculations to establish a range of mean values that accounts for potential differences in the amount of time it takes a hatchling to emerge from the nest after hatching, which would affect incubation duration estimates (Godfrey et al., 1999;. We calculated the overall sex ratio for each protection strategy within and across sites, and among years, and for specific comparisons among nest protection strategies and between sites, we used a three-day correction factor based on nests that showed a marked temperature signal at hatching (mean = 2.9 ± 0.2 days, n = 3; King, Cheng, Tseng, Chen, & Cheng, 2013).

| Thermal profiles of sand and nests
To measure intrabeach variation in temperature during the hawks- Loggers had an accuracy of ±0.2°C (per manufacturer specifications) and recorded the temperature every 30 min. We averaged recorded values to give a mean daily temperature for each logger, which facilitated comparisons with previous studies (e.g., Glen & Mrosovsky, 2004;Kamel & Mrosovsky, 2006b;Hawkes et al., 2007). Loggers that were stolen (n = 4 at Estero Padre Ramos), lost due to beach erosion (n = 4 at Bahía de Jiquilisco), or did not function properly during data collection (n = 4 at Bahía de Jiquilisco) were excluded from analyses. Ramos resulted in loss of temperature data for the open sand zone and deforested area.
To protect hawksbill clutches deposited on beaches where in situ protection and relocation on the beach were infeasible, shaded hatcheries were constructed at nesting beaches at both sites that typically operated from 1 May to 31 October annually and whose dimensions varied according to the capacity required for relocated clutches (Table 1). We buried loggers in the center of each hatchery at the two depths at Bahía de Jiquilisco in 2012-2015 (n = 2 or 3 hatcheries) and at Estero Padre Ramos in 2015 (n = 1 hatchery; Table 1). Temperature was recorded every 30 min and then averaged to obtain a mean daily temperature for each logger. Loggers that malfunctioned during data collection (n = 2 at Bahía de Jiquilisco) were not included in analyses.
To measure temperature in hawksbill nests during the incubation

| Statistical analyses
We used version 4.0.3 of Girondot's (1999) method to convert incubation duration of hawksbill clutches protected at our sites into hatchling sex ratios. Two-way analysis of variance (ANOVA) was used to test for differences among the three nest protection strategies in each of 10 parameters of incubation regime (i.e., nest temperature-minimum, maximum, mean of entire period, mean of thermosensitive period-during incubation, incubation duration, and nest depth) and hatchling condition (i.e., hatching success, offspring sex ratios, hatchling mass, and hatchling length) at Bahía de Jiquilisco and Estero Padre Ramos, and among years.
We also used a two-way ANOVA to test for differences in sand temperature within and among the six nest environments be-

| Nest distribution and protection strategies
We
TA B L E 4 Two-way ANOVA results for differences in each of four hatchling condition variables among three nest protection strategies (in situ, relocated on beach, and hatchery) among years at Bahía de Jiquilisco, El Salvador ( Overall, mean nest temperature during the entire incubation period was 30.4 ± 1.1°C (n = 274 clutches), with slightly higher temperatures in clutches protected in situ (30.7 ± 1.0°C, n = 44) than clutches relocated to hatcheries (30.3 ± 1.1°C, n = 218). Mean nest temperature during the middle third of the incubation period was likewise higher in clutches protected in situ (30.6 ± 1.3°C, n = 44) than clutches relocated to hatcheries (29.9 ± 1.1°C, n = 218; overall, 30.1 ± 1.2°C, n = 276). There was little difference in nest temperature between sites (Table 2), but significant differences among nest protection strategies at both sites (Table 3).

Incubation duration was similar at Bahía de Jiquilisco and Estero Padre
Ramos (Table 2), but significant differences existed among protection strategies and years at both sites (

| Sand temperature in beach, deforested, and hatchery environments
Sand temperatures at all logger locations exhibited temporal and spatial variation at Bahía de Jiquilisco ( There were significant differences in temperature between sand depths in woody vegetation, deforested areas, and hatcheries and among years at Bahía de Jiquilisco (  Figure 4f). In all nest environments, fluctuations in daily temperature were greater at the 30-cm than at the 60-cm depth, regardless of mean daily temperature (Figure 4a-d).
We suspect differences in overall hatching success reflect distinct biophysical conditions of beaches in mangrove estuaries, such as presence of extremely fine-grained sand. Because sand grain size affects water and gas flux (Ackerman, 1980), sand consisting of small particle sizes could have interstitial spacing and high water content that inhibits respiratory gas exchange of developing embryos (Ackerman, 1997), which could lower hatching success. For example, nesting beaches at Bahía de Jiquilisco consist of a high proportion (90.1%) of sand particle sizes measuring ≤0.125 mm (Y. Flores, unpublished data), which is substantially smaller than sand grain sizes reported for hawksbill nesting beaches in other geographic regions (Ditmer & Stapleton, 2012;Dobbs et al., 1999;Zare, Vaghefi, & Kamel, 2012).
We found significantly lower hatching success in clutches protected in situ (43.6%) than clutches relocated on the beach (46.2%) or in hatcheries (58.6%) at both sites (Table 4). This difference probably arises primarily from differences in microenvironmental conditions during incubation (Eckert & Eckert, 1990;Kornaraki et al., 2006;Revuelta et al., 2015), such as the amount of organic content (e.g., roots and leaves) in the sand, which is likely lower in hatcheries due to removal of organic material during hatchery preparation. This is consistent with hawksbill clutches in Antigua (Caribbean), where hatching success increased as a function of decreasing organic content in the sand (Ditmer & Stapleton, 2012).
Hatchery preparation processes could further favorably alter conditions of nest environments by lowering sand compaction within the hatchery enclosure, which could facilitate respiratory gas exchange of developing embryos (Garrett, Wallace, Garner, & Paladino, 2010).
We estimate that hawksbill nesting beaches produced 66.0%-81.0% female hatchlings across nest protection strategies at our sites, with a slightly higher percentage of females produced at Bahía de Jiquilisco than Estero Padre Ramos (Table 2). Our results represent lower female-biased sex ratios than reported at many sea turtle nesting beaches in other ocean basins (Hawkes et al., 2009;Poloczanska, Limpus, & Hays, 2009;Wibbels, 2003), but female production was more pronounced in clutches protected in situ, with 88.9%-96.2% and 68.1%-88.3% females at Bahía de Jiquilisco and Estero Padre Ramos, respectively. Clutches relocated to hatcheries at Estero Padre Ramos experienced a significant shift in sex ratios from highly male-biased in 2010-2011 to highly female-biased in 2012-2015 (Figure 3d). This shift is likely due to a change in hatchery location from a site with 100% overstory vegetation cover to an area with less cover (77.7%; Table 1), combined with climatic factors-such as cooler ambient temperature and increased precipitation associated with La Niña-reflected by longer incubation durations across protection strategies at Estero Padre Ramos. We attribute the higher percentage of female hatchlings produced at Bahía de Jiquilisco primarily to the degraded condition of coastal forest at many beaches relative to the higher-quality habitat that is available to nesting turtles at Estero Padre Ramos (Liles, Peterson, Seminoff, et al., 2015b), including areas where clutches are protected in situ. Indeed, vegetation cover can predict nest temperatures (Kamel, 2013) and hatchling sex (Janzen, 1994), which highlights the importance of preserving and restoring natural vegetation cover at hawksbill nesting beaches.
Hatchling length and mass differed among nest protection strategies and among years (Table 4), with hatchlings that were smaller and weighed less from clutches protected in situ than clutches relocated on the beach or in hatcheries (Table 2). Previous studies indicate that nest temperature is inversely correlated with hatchling body size, where warmer nests produce hatchlings with smaller carapaces and flippers, but that nest temperature did not influence hatchling mass (Booth, Feeney, & Shibata, 2013;Maulany, Booth, & Baxter, 2012;Wood, Booth, & Limpus, 2014). Hatchlings with larger carapaces and flippers are likely to crawl faster and employ more thrust while swimming than smaller hatchlings (Ischer, Ireland, & Booth, 2009;Janzen, Tucker, & Paukstis, 2000), which may allow them to more quickly navigate away from near-shore predators to offshore waters and thus increase their chance of survival (Booth, 2017;Wood et al., 2014).

| Warmer sand temperatures at the deeper depth
Our data on seasonal sand temperature in nest environments delineate temporal and spatial differences in hawksbill nesting environments at Bahía de Jiquilisco and Estero Padre Ramos. We found sand temperatures generally decreased from ocean to forest, with woody vegetation and hatcheries cooler than other nest environments (Figure 4e,f), which is consistent with thermal patterns reported for some hawksbill nesting beaches (Kamel, 2013;Kamel & Mrosovsky, 2006a), but contrasts with studies at other hawksbill nesting beaches that detected no difference between unshaded and shaded areas (Glen & Mrosovsky, 2004;Mrosovsky, Bass, Corliss, Richardson, & Richardson, 1992).
For most beach and hatchery environments at Bahía de Jiquilisco and Estero Padre Ramos, mean sand temperature was higher at the deeper depth (Figure 4e,f), which contrasts with the prevailing paradigm that temperatures are lower at deeper depths (Glen & Mrosovsky, 2004;Hill, Paladino, Spotila, & Santidrián Tomillo, 2015;Naro-Maciel et al., 1999). For example, Laloë, Esteban, Berkel, and Hays (2016) (Cartwright, 1974), the temperature of shallow groundwater (<10 m) can be 1-2°C greater than the mean annual surface temperature (Anderson, 2005) which can be further amplified in heavily shaded areas (Lewis & Wang, 1998), such as in woody vegetation and hatchery environments at our sites (Figure 4e,f).
Given that 90% (n = 564 clutches annually) of hawksbill reproductive output in the eastern Pacific is concentrated at five nesting sites within only one degree latitude (12°35′-13°35′N; Gaos et al., 2017), highly specific biophysical (e.g., sand morphology and ocean currents) and human-induced (e.g., depredation and beach development) conditions govern viability of these areas as suitable nesting habitat, suggesting that latitudinal redistribution to exploit other Central American beaches where similar climatic patterns are projected to occur seems unlikely Santidrián Tomillo et al., 2012).
Shifts in nesting phenology have been observed for some sea turtle populations (Azanza-Ricardo et al., 2017;Patel et al., 2016;Weishampel et al., 2004). Because sand temperatures at Bahía de Jiquilisco and Estero Padre Ramos generally decreased over the nesting season in all nest environments at both depths (Figure 4a-d), the decrease in temperature between the beginning (April-May) and end (September-October) of the nesting season-which is reflected in shorter incubation durations and higher percentage of female hatchlings produced during the first half than the second half of the nesting season (Figure 3a,b)suggests that hawksbills could respond to projected temperature increases by nesting later in the season to exploit cooler temperatures.
Additionally, turtles that currently nest in September-October at both sites may have an adaptive advantage (Valladares et al., 2014), highlighting the importance of protecting the nests of these individuals, even if their numbers are relatively fewer during those later months.
Some turtle populations appear to be capable of spatially adapting nest placement to align the current thermal niche of the nest environment with changing climatic conditions, whereas others seem relatively inflexible. For example, female painted turtles from five distinct populations across their geographic range that were translocated to a common environment differed in choice of nesting date and nest depth, but did not differ in shade cover, resulting in similar incubation regimes across populations despite differences in local climate at their locations of origin (Refsnider & Janzen, 2012). In contrast, individual hawksbills in the Caribbean are highly consistent in their nest microhabitat preferences, including vegetative cover above nests within and between years (Kamel & Mrosovsky, 2005, 2006b, suggesting that female hawksbills are relatively constrained in their ability to alter nesting behavior (Kamel, 2013). Although hawksbills at Bahía de Jiquilisco and Estero Padre Ramos are highly consistent in their selection of vegetative cover, they exhibit locally specific adaptations shaped by microhabitat differences at each site (Liles, Peterson, Seminoff, et al., 2015b). For example, nest placement by hawksbills at Bahía de Jiquilisco is restricted to the narrow tract of secondary forest measuring 10-15 m wide adjacent to the high water line at most beaches, whereas nest placement at Estero Padre Ramos extends nearly twice the distance inland within intact second-growth forest that is present >100 m landward from the high water line at most beaches (Liles, Peterson, Seminoff, et al., 2015b).
Such adaptations may indicate the potential for development of compensatory responses to climate variability through nest-site choice.
However, mangrove ecosystems are among the most threatened tropical environments in the world, with deforestation rates as high as 3.6% per year in the Americas (Valiela, Bowen, & York, 2001), suggesting that future degradation of forest habitat may impair its ability to buffer against increasing temperatures (Patrício et al., 2017). Coastal forests at our sites are confronted with the persistent threat of conversion by competing land uses, and forests along nesting beaches at Bahía de Jiquilisco have already experienced substantial alteration that restricts nest-site selection by hawksbills (Liles, Peterson, Seminoff, et al., 2015b). Our findings suggest that inability to halt the continued fragmentation of intact woody vegetation will progressively replace cooler male hatchling producing refugia (28.5°C) for naturally incubating clutches with markedly warmer woody vegetation border (30.2°C) and deforested (31.9°C) areas, increasing the probability of highly female-biased sex ratios (Poloczanska et al., 2009) and ultimately, climate-driven egg and hatchling mortality (Santidrián Tomillo et al., 2012).
The ability of egg-burying species to alter nest depth to compensate for increasing temperatures has been advanced as a possible adaptive strategy, presumably under the basic assumption that nest environments become cooler with increasing depth (Davenport, 1989;Fuentes & Porter, 2013). Our results, however, indicate that adjustment of nest depth by hawksbills is unlikely to compensate for climate change in mangrove estuaries. First, we detected higher temperatures at the deeper depth in most nest environments at both sites (Figure 4e,f). Second, the water table is at a depth of 50-85 cm during the nesting season at many beaches, which can be expected to become shallower as sea levels rise and further constrict suitable nest environments (Pike, 2014 The accelerated rate at which climate change is projected to occur, together with other interacting anthropogenic threats, may outpace the biological capacity of sea turtles to adapt (Fuentes, Hamann, & Limpus, 2010). The inability of sea turtles to adaptively respond through behavioral or evolutionary mechanisms (e.g., adjust pivotal temperature; Davenport, 1989) would require that humans intervene to prevent local extinctions, such as watering, shading, and clutch relocation to modify sand temperatures and reduce egg and hatchling mortality (Hill et al., 2015;Jourdan & Fuentes, 2015;Wood et al., 2014). Indeed, we found that hawksbill clutches relocated on the beach and protected in shaded hatcheries had higher hatching success, produced higher proportions of male offspring, and produced fitter hatchlings than clutches protected in situ at Bahía de Jiquilisco and Estero Padre Ramos. However, we are not suggesting egg relocation as a panacea that should be employed without careful consideration of local conditions, species biology, and conservation objectives. Previous studies have highlighted negative consequences of hatcheries using poor management practices, such as low hatching success (Boulon, Dutton, & Mcdonald, 1996), biased sex ratios of hatchlings (Morreale, Ruiz, Spotila, & Standora, 1982), and increased hatchling mortality (Pilcher & Enderby, 2001). We contend, however, that egg relocation can contribute substantively to recovery efforts under appropriate circumstances. Our results underscore the importance of empirical assessments to evaluate potential mitigation strategies for severely depleted populations of highly endangered species that may be unable to respond sufficiently to climate change.

ACK N OWLED G M ENTS
We thank local egg collectors and residents of Bahía de Jiquilisco and Estero Padre Ramos for their trust and collaboration. We acknowledge numerous people and organizations for their assistance with this study, in- Development for consistently funding hawksbill conservation initiatives that facilitated data collection for this study. We appreciate constructive comments on earlier drafts from three anonymous reviewers.

CO N FLI C T O F I NTE R E S T
None declared.

AUTH O R CO NTR I B UTI O N S
MJL, TRP, JAS, ARG, BPW, and MJP conceived and designed the study. MJL, EA, AVH, VG, SC, and JU collected data. MJL and MJP carried out data analyses. MJL led writing of the manuscript with input and critical review from all authors.

DATA ACCE SS I B I LIT Y
Hawksbill hatchling and temperature data can be accessed in the Dryad Digital Repository https://doi.org/10.5061/dryad.33rq371.