Low conservatism of the climatic niche of sea turtles and implications for predicting future distributions

Low conservatism of the climatic niche of sea turtles and implications for predicting future distributions A. D. MAZARIS, 1, D. VOKOU, 1 V. ALMPANIDOU, 1 O. TÜRKOZAN, 2 AND S. P. SGARDELIS 1 1 Department of Ecology, School of Biology, Aristotle University of Thessaloniki, 54124 Thessaloniki, Greece 2 Department of Biology Adnan Menderes University, 09010 Aydin, Turkey Citation: Mazaris, A. D., D. Vokou, V. Almpanidou, O. Türkozan, and S. P. Sgardelis. 2015. Low conservatism of the climatic niche of sea turtles and implications for predicting future distributions. Ecosphere 6(9):169. http://dx.doi.org/10. 1890/ES15-00053.1 Abstract. Understanding how the ecological niches of species are formulated across space is critical for modeling the current and predicting future distributions under climate change. Here, we examine how stable are the realized niches of distinct populations of the endangered green sea turtle (Chelonia mydas) which are defined on the basis of genetic and demographic data and are distributed around the globe. We used 906 georeferenced records of green turtle nesting sites that are embedded within eight Regional Management Units to compare climatic niche for each one of the distinct units. We used an asymmetric overall index and a decision-tree model to assess niche conservatism through a comparison of temperatureand precipitation-based bioclimatic variables that characterize the climatic niche breadth of the spatially distinct population segments. We found a high degree of variance in climatic space, which also lacks any latitudinal pattern. Environmental determinants vary significantly across the studied population units: variables that have been long-recognized as critical for nesting (e.g., maximum temperature of the warmest month) showed only low overlap between some of different regional entities. Our results contribute substantially to the current debate on the predictive power of species distribution models that use only climate variables as predictors when aggregating data from various populations. We argue that studies on climatic-niche evolution and divergent ecological selection mechanisms are required before attempting to identify suitable habitats for a species, describe climatic widths and search for plausible expansions of geographical shifts attributed to climate change. Key words: climatic niche overlap; distribution forecast; latitude; marine turtles; niche divergence; uncertainty. Received 30 January 2015; revised 29 April 2015; accepted 1 May 2015; published 29 September 2015. Corresponding Editor: D. P. C. Peters. Copyright: Ó 2015 Mazaris et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any médium, provided the original autor and source are credited. http://creativecommons.org/licenses/by/3.0/ E-mail: amazaris@bio.auth.gr INTRODUCTION Understanding the factors that determine species range limits and, hence, affect species distribution patterns is a critical issue in ecology (Guisan and Thuiller 2005). Of the full range of environmental conditions that allow a species to prevail (the species fundamental niche sensu Hutchinson 1957), only a part is realized, due to various constraints (Soberón 2007). Assessing the stability of realized niches may be critical for projecting accurately current distributions to the future (Wiens et al. 2010). The increasing scientific evidence on the impact of climate change on species distribution adds more weight to the requirement for the assessment of niche conservatism, which is defined as the degree to which species retain their ancestral ecological characteristic demonstrated in their distributions (Wiens and Graham 2005). Thus, to safely predict v www.esajournals.org 1 September 2015 v Volume 6(9) v Article 169

the species responses to changing climatic conditions, knowledge of the patterns of ecological differentiation is critical (Pearman et al. 2008, Pellissier et al. 2013). An important feature of many widely distributed species is that they may consist of populations subjected to different factors that in turn shape differently their realized niches (Davis and Shaw 2001). If geographically disconnected populations of the same species share the same realized niche, extrapolating the distribution at a global scale and predicting potential distribution changes in the context of global warming would be feasible (Randin et al. 2006, Peterson et al. 2007). However, locally prevailing environmental conditions may have driven genetically isolated populations into phenotypic variations and niche evolution resulting into diverging realized niches (Lawson and Weir 2014). Therefore, the lack of knowledge of the degree of niche conservatism among populations, raises concerns on the accuracy of quantifying niche breadth and projecting future potentially suitable habitats for the entire species range globally (Lavergne et al. 2010). Sea turtles are wide-ranging species that spend most of their life in the water (Poloczanska et al. 2009). Mature females visit coastal sites for reproduction traveling up to thousands of kilometers from the foraging to the nesting sites (Lutz et al. 2002, James et al. 2005). Sea turtles are ectothermic animals exhibiting temperature-dependent sex determination (Hays et al. 2014), with all phases of their reproduction indissolubly connected to climate. In this context, the impact of temperature, and precipitation, upon the biology and behavior of sea turtles is well documented (Hays et al. 2003, Houghton et al. 2007, Hawkes et al. 2009, Saba et al. 2012, Mazaris et al. 2013). Given this, it is expected that climate change will affect sea turtle population dynamics and nesting behavior in various ways (Poloczanska et al. 2009, Fuentes and Abbs 2010, Pike 2013a). However, we have only limited knowledge regarding climatic niche conservatism of the sea turtle nesting populations. Such information is critical not only to understand the climatic constraints in species distribution, but also to predict population responses to climate change and propose and apply accordingly specific mitigation measures at various scales (Wiens et al. 2010). Towards this direction, Pike (2013b) attempted to predict suitable nesting habitats of sea turtles at a global scale. In his study, the predicted potentially suitable habitats cover various areas around the world for species, which are known to inhabit only one small region. Based on this output, it becomes apparent that given the complexities of the species life history (Lutz et al. 2002) and the uncertainties which accompany the species distribution models (Elith and Leathwick 2009), there are many requirements that should be satisfied for a thorough evaluation of the influence of climate upon sea turtle distributions (Witt et al. 2010). For sea turtles, geographically defined population segments have been proposed by combining various types of information (e.g., genetic, demographic, satellite telemetry) in order to address different management and research challenges (Wallace et al. 2010). The distinction of these regional management units (RMUs) offers an excellent case study for evaluating niche conservatism providing insights on whether global projections of the species distribution are feasible. To explore the niche conservatism of sea turtles for climatic conditions, we focus our analysis on the green turtle (Chelonia mydas), which is one of the most widely distributed species in the planet having nesting sites across a broad latitudinal range (Seminoff et al. 2007). We use temperatureand precipitation-based bioclimatic variables that correspond to the nesting sites of the green turtles RMUs in order to characterize the climatic niche breadth of these spatially distinct units. The basic goals of our study are to investigate (1) whether there is a climatic niche overlap of the regional entities and thus test if the niche conservatism hypothesis holds true for the species, and (2) whether the realized niches of the distinct sea turtle units are driven or not by the same climatic factors. METHODS Nesting sites and climate data Our initial database consisted of a total of 1170 georeferenced records of green turtle nesting sites distributed across the world. These were obtained from the State of the World s Sea Turtles v www.esajournals.org 2 September 2015 v Volume 6(9) v Article 169

database (Halpin et al. 2009, Kot et al. 2013; SWOT Reports volumes I VII; http://seamap.env. duke.edu/swot). Nesting sites are found within 17 RMUs, as proposed by Wallace et al. (2010). Sites that were located within two RMUs are omitted from our analysis since they could not be used to define the climatic niche of distinct RMUs (see below for details), limiting our dataset to 1016 records. At the original paper by Wallace et al. (2010), RMUs have been defined on the basis of genetic and demographic information for a total of 935 sites which were available at the time of their research. Given that the available records on nesting sites have increased (n ¼ 1016), we assigned additional sites to the RMUs that were located within the shortest distance (Wallace, personal communication). We used an initial set of 19 bioclimatic variables, available via the WorldClim database, representing annual and seasonal trends and extreme or limiting environmental factors regarding temperature and precipitation (Hijmans et al. 2005; www.worldclim.org). Current climatic conditions were representative of the period 1950 2000 (Hijmans et al. 2005) and consisted of grid cells at a spatial resolution 2.5 arcmin, about 4 3 4 km, at the equator. Given that temperature and precipitation conditions strongly affect sea turtle reproductive performance and output (Hays et al. 2003, Poloczanska et al. 2009, Saba et al. 2012, Pike 2013a), selection of one variable between highly correlated pairs was suggested, on the basis of being biologically meaningful for nesting activities, i.e., successful reproduction (see also Pike 2013a, b). A total of 9 independent climatic variables were maintained for the analysis (all pairwise correlations,0.85): annual mean temperature, mean temperature diurnal range, maximum temperature of the warmest month, temperature annual range, mean temperature of the driest quarter, mean temperature of the warmest quarter, precipitation seasonality, precipitation of the driest quarter and precipitation of the warmest quarter. Given that RUMs are derived by merging spatial information such as nesting site location and satellite telemetry (Wallace et al. 2010), they are actually expanded across large areas. Still, some of these regional entities include only a limited number of nesting sites (e.g., the North- West Pacific RMU includes 7 nesting sites). In order to offer a greater margin for estimated climatic niche, we limited our analyses to 8 of the 17 RMUs (number of sites ¼ 906) each of which include more than 50 nesting sites and are located across a broad latitudinal range (21848 0 Sto248013 0 N; Fig. 1a, b). Under this conservative approach, we ensure that results of low climatic niche conservatism (based on temperature- and precipitation-based bioclimatic variables) are not driven by the abundance of occurrences and would actually offer a representative picture for the species. Analysis For each RMU, we extracted the maximum and minimum values of each variable across all of the nesting locations. Next, the difference between the maximum and the minimum values of each climatic variable was defined as the climatic range for this variable. For a climatic variable k, the degree of overlap between the RMU i over RMU j was then calculated using an asymmetric index climate overlap k;i;j ¼ minðr max;i; R max;j Þ maxðr min;i ; R min;j Þ O k;i where R max,i, R max,j, and R min,i,, R min,j are the maximum and minimum values of the climatic variable k, for RMU i and j and O k,i is the climatic range of variable k in the RMU i. The index takes values between zero and one that indicate the portion of the climatic range of RMU i that is overlapped by the range of RMU j. One means that the whole range of RMU i is within the range of the j RMU (i is completely overlapped by j ). Zero means that the two RMUs are not overlapping. The overall climatic niche overlap of a RMU over another one was then calculated as the product of the overlaps estimated for the nine climatic variables that were examined; the lack of overlap in at least one of the studied variables would result to the absence of niche overlap of the one RMU over the other. If the assumption that niche conservatism hypothesis holds for the green turtle RMUs, then the estimated similarity between the climatic niche overlap would be rather constant and high. This should hold for all pairwise estimates of the overall climate overlap. For each pair of RMUs compared, we obtained v www.esajournals.org 3 September 2015 v Volume 6(9) v Article 169

Fig. 1. (a) Distribution of green sea turtle nesting sites (n ¼ 906; red circles), (b) along the 8 Multi-scale Regional Management Units for green sea turtles analyzed in the present study (after Wallace et al. 2010). two generally different estimates of overlap. To check for the niche conservatism hypothesis we maintained only the largest of the two overlap values produced and examined the properties of their distribution. A highly right skewed distribution and a high frequency of observation at low values of overlap, even using the largest values, would be indicative of the lack of strong signal of niche conservatism. To test whether the climatic breadth of the sea turtle s niche in the different RMUs is driven by geographical constraints, we employed a Spearman correlation analysis between the degree of overlap for each climatic variable and the latitudinal range of the RMUs. To examine if the observed correlations were confounded by sample size, we checked if the residuals were related to the number of nesting sites in each RMU. To investigate whether the RMUs of green turtles are separated after climatic variables, we employed a non-parametric supervised learning method by the means of a Decision Tree model. The rpart package in R (R Development Core Team 2013) was used for growing and pruning the decision trees. A multivariate analysis of variance was performed applying the Lawley- Hotelling trace statistic to test for statistically significant differences of bioclimatic parameters per nesting site by RMU considered a factor Hooper s squared trace correlation coefficient (a generalization of R 2 for multivariate data) was used to assess the goodness of fit of the multivariate model. An analysis of deviance for univariate model fits (i.e., distinction of RMUs by each bioclimatic parameter) was conducted, with p-values generated by a residual resampling permutation with 1000 iterations. Univariate statistics were employed to identify the climatic variables that best explain the variation in a RMU. The manylm function of the mvabund package in R was used to perform the former analysis. v www.esajournals.org 4 September 2015 v Volume 6(9) v Article 169

Fig. 2. Estimates of the climate overlap asymmetric index for all pairs of RMUs. The boundaries of the box indicate the 25th and 75th percentiles, the line within the box marks the median, and cups, below each box indicate the 90th and 10th percentiles. RESULTS Climatic variation across green turtle RMUs All studied climatic variables showed a large variation across RMUs (Appendix). Annual mean temperature ranged from 24.08 6 2.58C at the nesting sites of the Pacific Southwest RMU to 27.08 6 0.48C at the nesting beaches located at the Pacific West Central RMU. Of all studied RMUs, the nesting sites of the Pacific South Central showed the lower variation in mean temperature diurnal range (6.2 6 0.48C), with sites located at the Pacific East RMU been subjected to higher variation in the same bioclimatic variable (10.38 6 2.38C). A great variation was further obtained at the maximum temperature of the warmest month, with Pacific South Central, Pacific West Central and Atlantic South Caribbean RMUs showing the more constant patterns among their nesting sites. The same RMUs were also subjected to the lower variation of the annual range of temperature, across their nesting beaches. Mean temperature of the driest quarter was highly variable across the nesting sites of the Pacific Southwest RMU while the mean temperature of the warmest quarter demonstrated a low variation among most of the sites for the majority of the RMUs, with the exception of Pacific Southwest and Pacific South Central RMUs. The nesting sites of the Indian Northwest RMU showed a remarkable variation in precipitation seasonality (93.23 6 42 mm) with the more stable patterns obtained in Pacific South Central and the Atlantic South Caribbean RMUs. Precipitation of the driest and warmest quarters followed a rather similar pattern across the RMUs, with lower variation at the sites of the Atlantic South Caribbean and the Atlantic Northwest RMUs. Climatic niche overlap For each climatic variable, estimates of asymmetric climatic overlap index are presented in Fig. 2. The highest climatic overlap was obtained for the mean diurnal temperature range (median ¼ 0.91) and the mean temperature of the driest quarter (median ¼ 0.82), whereas the lowest index values corresponded to the maximum temperature of the warmest month (median ¼ 0.63). Although, in several RMUs, we obtained a large overlap for a number of climatic variables, still the overall climatic niche overlap was maintained at very low levels. Even when the highest observed values of overlap index were used for each pair of RMUs, the outputs were described by a highly right skewed distribution with a medium value of v www.esajournals.org 5 September 2015 v Volume 6(9) v Article 169

Fig. 3. Distribution of climatic overlap values for all pairs of the eight (8) multi-scale Regional Management Units (RMUs) of green sea turtles studied. Climatic niche overlap of a RMU over another one was calculated as the product of the overlaps estimated for the nine climatic variables with value of one (1) indicating a complete overlapped of a RUM over another and zero values being indicative of the lack of climate overlapping. 0.091 (Fig. 3) and only four RMUs pairs showing an overlap higher than 0.2. The latitude of the each RMU varies between 12 and 27 degrees. We found no significant relationship between the range of the climatic variables and the latitudinal range of the RMUs (in all cases p. 0.05) with the exception of two marginally positive correlations: of the temperature annual range (R 2 ¼ 0.534, p ¼ 0.40) and the mean temperature of the warmest quarter of the year (R 2 ¼ 0.525, p ¼ 0.42). Analysis of the asymmetric overlap index for all pairs of RMUs revealed no correlation between the latitudinal and the climatic overlaps ( p. 0.05). Bioclimatic variable driving realized niches of the geographical distinct green turtle units A significant separation of the RMUs on the basis of the climatic variables was observed (Lawley-Hotelling trace statistic ¼ 473.2, p, 0.01) with multivariate analysis of variance demonstrating an adequate fit (Hooper s R 2 ¼ 0.4283, p, 0.01). The decision tree analysis revealed that closely related clusters in terms of bioclimatic conditions did not consist of RMUs located at the same ocean basis but were rather mixed along the geographical space; the pruned tree is presented in Fig. 4. RMUs were separated initially after the maximum temperature of the warmest month, while both precipitation and temperature related variables were responsible for further divisions of the RMUs. DISCUSSION The results of this study provide evidence of limited climatic niche conservatism among the distinct geographical units of green sea turtles. We found that ambient climatic determinants vary considerably along the distribution range of the species. The width of the climatic variables is not driven by the width of the latitudinal distribution of distinct segments, suggesting that factors other than geographical range constrain ecological niche within each unit. These results do not support the use of aggregate information on the localities of different populations for predicting the spatial and temporal stability of the realized niches of the given species. Our findings raise even more concerns on the use of climate envelope models for predicting current and future distribution of the species (see Pearson and Dawson 2003, Hijmans and Graham 2006,) without considering other climatic (e.g., v www.esajournals.org 6 September 2015 v Volume 6(9) v Article 169

Fig. 4. Structure of the decision tree grouping the eight Multi-scale Regional Management Units (RMUs) of green sea turtles according to the level of similarity of different bioclimatic variables based on temperature (8C) and precipitation (mm), along with the two (a, b) main clusters of RMUs resulted by the decision tree on the basis of maximum temperature of the warmest month depicted in a geographical scale. sea temperature) or environmental variables (e.g., beach topography) which are important drivers of sea turtles nesting. The observed divergence of the climatic niche that green sea turtle populations experience could reflect an adaptation from an evolutionary perspective. Evidence suggests that thermal tolerance affects profoundly the intraspecific phylogeography of green sea turtles (Bowen and Karl 2007), providing support to this hypothesis. Following the ancient phylogenetic separation between Indo-Pacific and Atlantic v www.esajournals.org 7 September 2015 v Volume 6(9) v Article 169

Ocean populations (Bowen and Karl 2007), the Atlantic basin is further subdivided into contemporary breeding units that were possibly created as a response to alterations of the climate and the biotic environment (Bowen et al. 1989). Other evidence suggests that over the last interglacial period, green turtles in northern Australia have been adapted to changing temperature conditions by shifting their timing of nesting (from austral summer to the austral winter; Dethmers et al. 2006). Differences in local environmental conditions that sea turtle populations experience due to the geographic separation may have further accelerated the adaptation and/or phenotypic plasticity, with evidence on adaptation regimes reported for various behavioral and biological processes at different spatial scales. At a coarse scale, increasing evidence supports the existence of diverging nesting ecology in green sea turtles. At such scales studies demonstrate that breeding populations nest all over the year where climatic conditions are appropriate (e.g., Malaysia) but limit their nesting season mainly within 4 months (May August), in the northern hemisphere, as an adaptation to the strong temperature seasonality (Poloczanska et al. 2009). Even at a finer scale, Weber et al. (2012) showed an adaptive differentiation at a nesting site, with diverging thermal tolerances of green turtle embryos, which are adapted to the contrasting incubation temperatures at different locations. Our analyses were restricted to some climatic variables affecting the selection and usage of nesting habitats (Pike et al. 2013a, b). However, a series of other environmental factors may also play a critical role in these processes. For example, the geographic distribution of nesting can be associated with sea surface current conditions that facilitate successful migration of offspring (Putman et al. 2010). Wind exposure could also act as a determinant of the spatial distribution of nesting sites (Garcon et al. 2010), while extreme weather events like cyclone activity may also influence the distribution of nesting sites at local scales (Fuentes et al. 2010). The existence of all these physiological and energetic constraints for the selection of nesting site further highlight the need to incorporate local threats that are connected with climate change and specific species traits that mediate or reflect responses to local conditions (Garcia et al. 2014). Our analysis showed that the climatic niche and the observed overlap of climatic variables were not related to the latitudinal width of each population. Previous studies, examining climatic niche overlap between sister species at temperate and tropic regions, have reached contradictory conclusions, showing highest thermal overlap in the tropics (bats, birds, frogs, lizards and snakes; Cadena et al. 2012), a larger niche overlap at high latitudes (salamanders; Kozak and Wiens 2007), no latitudinal pattern (frogs; Hua and Wiens 2010) or mixed latitudinal trends depending on the climatic variables (birds; Lawson and Wier 2014). In regions with limited observed overlap, species have probably experienced stronger divergent selection pressure driven by climate and may have been more prone to climate induced geographical separation and climatic niche specialization (Kozak and Wiens 2007, Lawson and Wier 2014). The RMUs studied here have broad latitudinal extents, with some of them located or even expanding through large areas over the northern and southern hemispheres. Given that geography could exhibit an important role in climatic niche divergence, and broad latitudinal ranges include geographical heterogeneity in climate (Lawson and Wier 2014), the lack of a relationship between latitudinal and climatic niche overlaps is likely to suggest that differences in selection of suitable habitats are driven by local adaption. Biotic factors contribute into shaping species niche, modifying the limits determined by climatic conditions (Soberón 2007, Wiens 2011). In the case of sea turtles, human induced pressure could be considered as an additional such biotic factor. Intensive harvesting of sea turtles, which in some cases dates back to three millenia (Allen 2007), has resulted to widespread losses of nesting habitats (McClenachan et al. 2006, Bell et al. 2007). In this case, the climatic width of a RMU inferred by the current distribution of the nesting sites would be actually limited to a subset of location of varying conditions where species could prevail. The full understanding of the potential impact of such range contractions would require detailed information on past nesting areas but also on climatic niche evolution rates. v www.esajournals.org 8 September 2015 v Volume 6(9) v Article 169

Various methods have been proposed for measuring ecological niche overlap including species distribution modelling and ordination techniques (Broennimann et al. 2012). Given that the development of species distribution models could themselves been subjected to various sources of uncertainly (e.g., Elith and Leathwick 2009, Beale and Lennon 2012), the conservative index applied here allowed us to study niche differentiation as defined by a corresponded unique set of climatic variables of each population (Kozak and Wiens 2007). In addition, the application of a set of multivariate analyses offered us a statistical rigor background for ranking climatic variables according to their importance in delimiting niche; a step which is considered a critical in studying ecological niche properties which, however, lacking from assessments made by the use of ordination methods (Broennimann et al. 2012). Species distribution models offer an empirical background for linking current environmental data with species occurrences, with climatic envelope modeling techniques, as part of this family of models, utilizing climatic data for deriving predictions of the potential occurrences and assessing vulnerability to future changes. However, it is important to mention that the success of models used to predict current and future distribution of species relies on our ability to accurately describe their ecological niche (Pearson and Dawson 2003). In the case of sea turtles, the actual selection of the nesting site is driven by a series of factors that operate at a local scale (e.g., grain size, beach furniture, slope, vegetation) and thus could not easily be captured within global-scale datasets. The importance of such factors becomes more pronounced if we consider that even neighboring sites might be subjected to different number of sea turtle nesters. On the basis of our findings, it becomes apparent that broad scale climate data are questioned regarding their ability to provide a large scale assessment of potential nesting suitability, and thus we further caution that accurate predictions should be built upon a comprehensive understanding of the ecological niche of the species, by considering environmental drivers that operate at finer spatial scale. Contributing to these issues, we encourage future studies to address potential changes in estimated climatic niche by addressing loss of historic or colonization of new nesting sites, as these inputs could serve as a tool for evaluating the efficiency of the niche models. In this context, predictions based on climate alone may not be sufficient for safeguarding sea turtle populations but could represent a valuable tool towards building the resilience of existing and potential nesting grounds. Such outputs could serve as a first step towards assessing the vulnerability of the nesting populations with a potential assessment of the relationship between the environmental niche and population size allowing to detect different responses at the margins of species range (e.g., Russell et al. 2015). Our study expands the results of Pike (2013b), who modeled the plausible current climatic distribution of sea turtles, assuming that nesting locations of each species currently occupy the entire climate width of the species. The lack of climatic similarity of RMUs along with the undoubtedly linkage between climatic variables and nesting activity, suggests that although main climatic factors act as drivers for the selection of nesting environments over time, the local conditions, at the end, structure thermal tolerances for each population. To further contribute to the open debate on the reliability of climate impact models predicting species distribution (see Heikkinen et al. 2007), we strongly caution that spatial climatic trends at a global scale cannot accurately predict the distribution of sea turtle nesting sites. ACKNOWLEDGMENTS This work was co-financed by the European Union and the Greek State, Ministry of Education and Religious Affairs/General Secretariat for Research and Technology (O. P. Competitiveness and Entrepreneurship (EPAN II), ROP Macedonia Thrace, ROP Crete and Aegean Islands, ROP Thessaly Mainland Greece Epirus, ROP Attica). The work of OT was supported by The Scientific and Technological Research Council of Turkey (Project code: 113Z437). LITERATURE CITED Allen, M. S. 2007. Three millennia of human and sea turtle interactions in Remote Oceania. Coral Reefs 26:959 970. Beale, C. M., and J. J. Lennon. 2012. Incorporating uncertainty in predictive species distribution modelling. Philosophical Transactions of the Royal v www.esajournals.org 9 September 2015 v Volume 6(9) v Article 169

Society B 367:247 258. Bell, C. D., J. L. Solomon, J. M. Blumenthal, T. J. Austin, G. Ebanks-Petrie, A. C. Broderick, and B. J. Godley. 2007. Monitoring and conservation of critically reduced marine turtle nesting populations: lessons from the Cayman Islands. Animal Conservation 10:39 47. Bowen, B. W., and S. A. Karl. 2007. Population genetics and phylogeography of sea turtles. Molecular Ecology 16:4886 4907. Bowen, B. W., A. B. Meylan, and J. C. Avise. 1989. An odyssey of the green sea turtle, Chelonia mydas: Ascension Island revisited. Proceedings of the National Academy of Sciences USA 86:573 576. Broennimann, O., M. C. Fitzpatrick, P. B. Pearman, B. Petitpierre, L. Pellissier, N. G. Yoccoz, W. Thuiller, F. Marie-Josée, C. Randin, N. E. Zimmermann, C. H. Graham, and A. Guisan. 2012. Measuring ecological niche overlap from occurrence and spatial environmental data. Global Ecology and Biogeography 21:481 497. Cadena, C. D., et al. 2012. Latitude, elevational climatic zonation and speciation in New World vertebrates. Proceedings of the Royal Society B 279:194 201. Davis, M. B., and R. G. Shaw. 2001. Range shifts and adaptive responses to Quaternary climate change. Science 292:673 679. Dethmers, K. E. M., D. Broderick, C. Moritz, N. N. Fitzsimmons, C. J. Limpus, S. Lavery, S. Whiting, M. Guinea, R. I. T. Prince, and R. Kennett. 2006. The genetic structure of Australasian green turtles (Chelonia mydas): exploring the geographical scale of genetic exchange. Molecular Ecology 15:3931 3946. Elith, J., and J. R. Leathwick. 2009. Species distribution models: ecological explanation and prediction across space and time. Annual Review of Ecology Evolution and Systematics 40:677 697. Fuentes, M. M. P. B., and D. Abbs. 2010. Effects of projected changes in tropical cyclone frequency on sea turtles. Marine Ecology Progress Series 412:283 292. Fuentes, M. M. P. B., M. Hamann, and C. J. Limpus. 2010. Past, current and future thermal profiles of green turtle nesting grounds: implications from climate change. Journal of Experimental Marine Biology and Ecology 383:56 64. Garcia, R. A., M. B. Araujo, N. D. Burgess, W. B. Foden, A. Gutsche, C. Rahbek, and M. Cabeza. 2014. Matching species traits to projected threats and opportunities from climate change. Journal of Biogeography 41:724 735. Garcon, J. S., A. Grech, J. Moloney, and M. Hamann. 2010. Relative exposure index: an important factor in sea turtle nesting distribution. Aquatic Conservation: Marine and Freshwater Ecosystems 20:140 149. Guisan, A., and W. Thuiller. 2005. Predicting species distribution: offering more than simple habitat models. Ecology Letters 8:993 1009. Halpin, P. N., et al. 2009. OBIS-SEAMAP: the world data center for marine mammal, sea bird, and sea turtle distributions. Oceanography 22:104 115. Hawkes, L. A., A. C. Broderick, M. H. Godfrey, and B. J. Godley. 2009. Climate change and marine turtles. Endangered Species Research 7:137 154. Hays, G. C., A. C. Broderick, F. Glen, and B. J. Godley. 2003. Climate change and sea turtles: a 150-year reconstruction of incubation temperatures at a major marine turtle rookery. Global Change Biology 9:642 646. Hays, G. C., A. D. Mazaris, and G. Schofield. 2014. Different male vs. female breeding periodicity helps mitigate offspring sex ratio skews in sea turtles. Frontiers in Marine Science 1:43. Heikkinen, R. K., M. Luoto, R. Virkkala, R. G. Pearson, and J. H. Korber. 2007. Biotic interactions improve prediction of boreal bird distributions at macroscales. Global Ecology and Biogeography 16:754 763. Hijmans, R. J., S. E. Cameron, J. L. Parra, P. G. Jones, and A. Jarvis. 2005. Very high resolution interpolated climate surfaces for global land areas. International Journal of Climatology 25:1965 1978. Hijmans, R. J., and C. H. Graham. 2006. The ability of climate envelope models to predict the effect of climate change on species distributions. Global Change Biology 12:2272 2281. Houghton, J. D. R., A. E. Myers, C. Lloyd, R. S. King, C. Isaacs, and G. C. Hays. 2007. Protracted rainfall decreases temperature within leatherback turtle (Dermochelys coriacea) clutches in Grenada, West Indies: ecological implications for a species displaying temperature dependent sex determination. Journal of Experimental Marine Biology and Ecology 345:71 77. Hua, X., and J. J. Wiens. 2010. Latitudinal variation in speciation mechanisms in frogs. Evolution 64:429 443. Hutchinson, G. E. 1957. Population studies: animal ecology and demography, concluding remarks. Cold Spring Harbor Symposia on Quantitative Biology 22:415 427. James, M. C., R. A. Myers, and C. A. Ottensmeyer. 2005. Behaviour of leatherback sea turtles, Dermochelys coriacea, during the migratory cycle. Proceedings of the Royal Society B 272:1547 1555. Kot, C. Y., A. DiMatteo, E. Fujioka, B. Wallace, B. Hutchinson, J. Cleary, P. Halpin, and R. Mast. 2013. The State of the World s Sea Turtles online database: data provided by the SWOT team and hosted on OBIS-SEAMAP. Oceanic Society, Conservation International, IUCN Marine Turtle Specialist Group (MTSG), and Marine Geospatial v www.esajournals.org 10 September 2015 v Volume 6(9) v Article 169

Ecology Lab, Duke University. http://seamap.env. duke.edu/swot Kozak, K. H., and J. J. Wiens. 2007. Climatic zonation drives latitudinal variation in speciation mechanisms. Proceedings of the Royal Society B 274:2995 3003. Lavergne, S., N. Mouquet, W. Thuiller, and O. Ronce. 2010. Biodiversity and climate change: integrating evolutionary and ecological responses of species and communities. Annual Review of Ecology, Evolution, and Systematics 41:321 350. Lawson, A. M., and J. T. Weir. 2014. Latitudinal gradients in climatic-niche evolution accelerate trait evolution at high latitudes. Ecology Letters 17:1427 1436. Lutz, J. A., J. A. Musick, and J. Wyneken. 2002. The biology of sea turtles. Volume II. CRC Press, Boca Raton, Florida, USA. Mazaris, A. D., A. S. Kallimanis, J. D. Pantis, and G. C. Hays. 2013. Phenological response of sea turtles to environmental variation across a species northern range. Proceedings of the Royal Society B 280:20122397. McClenachan, L., J. B. C. Jackson, and M. J. H. Newman. 2006. Conservation implications of historic sea turtle nesting beach loss. Frontiers in Ecology and the Environment 4:290 296. Pearman, P. B., A. Guisan, O. Broennimann, and C. F. Randin. 2008. Niche dynamics in space and time. Trends in Ecology and Evolution 23:149 158. Pearson, R. G., and T. P. Dawson. 2003. Predicting the impacts of climate change on the distribution of species: Are bioclimate envelope models useful? Global Ecology and Biogeography 12:361 371. Pellissier, L., et al. 2013. Thermal niches are more conserved at cold than warm limits in arctic-alpine plant species. Global Ecology and Biogeography 22:933 941. Peterson, A. T., M. Papes, and M. Eaton. 2007. Transferability and model evaluation in ecological niche modeling: a comparison of GARP and Maxent. Ecography 30:550 560. Pike, D. A. 2013a. Forecasting range expansion into ecological traps: climate-mediated shifts in sea turtle nesting beaches and human development. Global Change Biology 19:3082 3092. Pike, D. A. 2013b. Climate influences the global distribution of sea turtle nesting. Global Ecology and Biogeography 22:555 566. Poloczanska, E. S., C. J. Limpus, and G. C. Hays. 2009. Vulnerability of marine turtles to climate change. Advances in Marine Biology 56:151 211. Putman, N. F., J. M. Bane, and K. J. Lohmann. 2010. Sea turtle nesting distributions and oceanographic constraints on hatchling migration. Proceedings of the Royal Society B 277:3631 3637. Randin, C. F., T. Dirnbock, S. Dullinger, N. E. Zimmermann, M. Zappa, and A. Guisan. 2006. Are niche-based species distribution models transferable in space? Journal of Biogeography 33:1689 1703. R Development Core Team. 2013. R: a language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. Russell, D. J., S. Wanless, Y. C. Collingham, B. J. Anderson, C. Beale, J. B. Reid, B. Huntely, and K. C. Hamer. 2015. Beyond climate envelopes: bioclimate modelling accords with observed 25year changes in seabird populations of the British Isles. Diversity and Distributions 21:211 222. Saba, V. S., C. A. Stock, J. R. Spotila, F. V. Paladino, and P. S. Tomillo. 2012. Projected response of an endangered marine turtle population to climate change. Nature Climate Change 2:814 820. Seminoff, J. A., B. Schroeder, S. McPherson, E. Possardt, and K. Bibb. 2007. Green sea turtle (Chelonia mydas) 5-year review: summary and evaluation. National Marine Fisheries Service and US Fish and Wildlife Service Report 130. National Oceanic and Atmospheric Administration, Silver Spring, Maryland, USA. Soberón, J. 2007. Grinnellian and Eltonian niches and geographic distributions of species. Ecology Letters 10:1115 1123. Wallace, B. P., et al. 2010. Regional management units for marine turtles: a novel framework for prioritizing conservation and research across multiple scales. PLoS ONE 5(12):e15465. Weber, S. B., A. C. Broderick, T. G. G. Groothuis, J. Ellick, B. J. Godley, and J. D. Blount. 2012. Finescale thermal adaptation in a green turtle nesting population. Proceedings of the Royal Society B 279:1077 1084. Wiens, J. J. 2011. The niche, biogeography and species interactions. Philosophical Transactions of the Royal Society B 366:2336 2350. Wiens, J. J., et al. 2010. Niche conservatism as an emerging principle in ecology and conservation biology. Ecology Letters 13:1310 1324. Wiens, J. J., and C. H. Graham. 2005. Niche conservatism: integrating evolution, ecology, and conservation biology. Annual Review of Ecology Evolution and Systematics 36:519 539. Witt, M. J., L. A. Hawkes, M. H. Godfrey, B. J. Godley, and A. C. Broderick. 2010. Predicting the impacts of climate change on a globally distributed species: the case of the loggerhead turtle. Journal of Experimental Biology 213:901 911. v www.esajournals.org 11 September 2015 v Volume 6(9) v Article 169

SUPPLEMENTAL MATERIAL ECOLOGICAL ARCHIVES The Appendix is available online: http://dx.doi.org/10.1890/es15-00053.1.sm v www.esajournals.org 12 September 2015 v Volume 6(9) v Article 169