The thermal niche of Neotropical nectar‐feeding bats: Its evolution and application to predict responses to global warming

Abstract The thermal niche of a species is one of the main determinants of its ecology and biogeography. In this study, we determined the thermal niche of 23 species of Neotropical nectar‐feeding bats of the subfamily Glossophaginae (Chiroptera, Phyllostomidae). We calculated their thermal niches using temperature data obtained from collection records, by generating a distribution curve of the maximum and minimum temperatures per locality, and using the inflection points of the temperature distributions to estimate the species optimal (STZ) and suboptimal (SRZ) zones of the thermal niche. Additionally, by mapping the values of the STZ and SRZ on a phylogeny of the group, we generated a hypothesis of the evolution of the thermal niches of this clade of nectar‐feeding bats. Finally, we used the characteristics of their thermal niches to predict the responses of these organisms to climate change. We found a large variation in the width and limits of the thermal niches of nectar‐feeding bats. Additionally, while the upper limits of the thermal niches varied little among species, their lower limits differ wildly. The ancestral reconstruction of the thermal niche indicated that this group of Neotropical bats evolved under cooler temperatures. The two clades inside the Glossophaginae differ in the evolution of their thermal niches, with most members of the clade Choeronycterines evolving “colder” thermal niches, while the majority of the species in the clade Glossophagines evolving “warmer” thermal niches. By comparing thermal niches with climate change models, we found that all species could be affected by an increase of 1°C in temperature at the end of this century. This suggests that even nocturnal species could suffer important physiological costs from global warming. Our study highlights the value of scientific collections to obtain ecologically significant physiological data for a large number of species.


| INTRODUCTION
The thermal niche of a species, defined as the range of temperatures where it is able to live, is one of the main determinants of its ecology and biogeography (Bozinovic, Ferri-Yáñez, Naya, Araújo, & Naya, 2014). It is determined by the species size and shape, and its ability to survive in, or adapt to places with different temperature regimes (Angilletta, Niewiarowski, & Navas, 2002;Porter & Kearney, 2009). It is strongly associated with the species metabolic costs and its physiological capacities to withstand thermal variation (Spicer & Gaston, 1999). From an energetic perspective, it is closely limited by the species capacity to provide the energy needed to maintain its metabolic costs under different ambient temperatures (Bell, Bartholomew, & Nagy, 1986).
Endothermy is a physiological strategy mainly used by mammals and birds, which allow them to maintain an almost constant body temperature, independently from ambient temperature (Schmidt-Nielsen, 1997). It is achieved by using the heat generated by body functions to control body temperature within a range of environmental conditions that favors its metabolic functions, enabling it to survive in places with highly variable conditions (Angilletta et al., 2002). Overall, this internal heat is the incidental result of the routine metabolism of animals, and its costs are energetically low (Bozinovic et al., 2014). However, when endothermic animals are confronted with extreme temperatures, they use specialized mechanisms to maintain stable body temperature (i.e., the use of large amounts of energy to increase heat production, or water loss by evapotranspiration to cool down; Scholander, Hock, Walters, Johnson, & Irving, 1950). As a result of this, by facing different ambient temperatures along its geographic distribution, a species confronts areas within its thermal niche that have different metabolic costs, some of them low, but some exceptionally high (Angilletta et al., 2002).
The division of the thermal niche of a species in areas of optimal conditions, with low physiological costs, and areas of suboptimal conditions, with higher physiological costs, is not a new idea in the field of ecological physiology. Brett (1952) defined the area where environmental conditions are optimal for the survival of the members of a species as its tolerance zone (TZ), and the zone where environmental conditions reduce individual survival by increasing physiological costs, as the species resistance zone (RZ). Regarding the thermal niche and physiological capabilities of endotherm animals, the TZ corresponds to temperature ranges where species have lower metabolic rates or present metabolic costs, that while higher, can be easily covered by their energy intake, reducing their effects on the animal capacity to survive and reproduce. Some authors have suggested that this temperature range is associated with the species thermoneutral zone, because the thermoneutral zone limits provide an index of an endotherm's temperature comfort range (Bozinovic et al., 2014;McNab, 2012). The RZ comprises temperatures above and below the tolerance zone, where metabolic costs increase as the species move away from the thermoneutral zone to a point where individuals cannot survive for long periods of time (Cossins & Bowler, 1987).
Despite the fact that understanding the TZ and the RZ of different species, and its evolution, allow us to understand the thermal ecology and capacity of animals to adapt and survive climate change, we have scarce knowledge of the thermal tolerances of most organisms (Araújo et al., 2013). This is mainly the result of the demanding experimental methods needed to determine the thermal limits of animals.
Additionally, laboratory measurements of metabolic responses of animals to different temperatures (e.g., thermoneutral zones) cannot be used to infer the capacity of animals to find their thermal niches in complex natural environments (Porter & Kearney, 2009). However, with the inclusion of geographic information system (GIS) in scientific disciplines like biology, we can infer a species' TZ and RZ using temperature information related to geographic information data that can be found in museum specimens. This geographically linked information offers us the advantage of integrating the interaction between physiological capacities and environmental factors over large geographic areas, allowing us to obtain physiological information at the level of species, and not only at level of individuals. Furthermore, this information combined with phylogenetic approaches has the potential to provide insights on the evolution of the thermal niches of animals.
The family Phyllostomidae comprises a clade of bats endemic to the Americas (Fleming, Geiselman, & Kress, 2009). This group presents the biggest diversity of diets for a family of vertebrates (Gardner, 1977), including specialized nectarivory, a diet found in the members of the subfamily Glossophaginae. The species of this diverse clade display several features associated with their sweet diet, such as long and narrow snouts, a reduction in the number of functional teeth, an elongated and projectable tongue, and several digestive and renal traits that allow them to cope with their sugary water diet (Carstens, Lundrigan, & Myers, 2002;Schondube, Herrera-M, & Martínez del Rio, 2001). As a result of their dependence on floral nectar as a source of energy, the evolution of this group of nectar-feeding bats occurred in the tropics, where the diversity of plants is high, forming close-ties with the plants species from which they obtain their food (von Helversen & Winter, 2003;Valiente-Banuet, Arizmendi, Rojas-Martínez, & Domínguez-Canseco, 1996). Because these mammals have high metabolic rates (Voigt & Speakman, 2007), and nectar is a resource that varies widely in time and space (Chalcoff, Aizen, & Galetto, 2006), they tend to live on the verge of a negative energy balance (Ayala-Berdon, Schondube, Stoner, Rodriguez-Peña, & Martínez Del Rio, 2008;Ayala-Berdon et al., 2011;Cruz-Neto & Abe, 1997;von Helversen & Winter, 2003). Consequently, we can expect nectar-feeding bats to be very sensitive to extreme temperatures, and the metabolic costs associated with them.
In this study, we determined the thermal niche of 23 species of nectar-feeding Neotropical bats based on collection records from public databases. Additionally, by using a phylogenetic approach, we propose a hypothesis of the evolution of the thermal niche in Glossophaginae. Finally, we related the characteristics of the thermal niches with some of the physiological capacities of these bats, and the ecological conditions in which these species have evolved. We highlight the usefulness of public databases, along with spatial tools, to reveal critical insights to understand the evolution of the thermal niches of bats and their potential adaptation capabilities at a time in which planetary climate is changing fast due to anthropogenic factors (Parmesan & Yohe, 2003).
Locality records for each species were obtained from the Global Biodiversity Information Facility (GBIF; www.gbif.org). We treated databases conservatively by comparing each locality collection with the areas of species distribution based on Hall (1981), Gardner (2007), and Reid (2009). All collection records that were clearly outside the mentioned ranges were excluded from further analyses (following Elith et al., 2006 andJaramillo &Martínez, 2014). All records with the same geographic coordinates were considered as a single locality. As small differences in geographic coordinates might represent important differences in environmental conditions, especially when a geographic region presents a complex topography (i.e., mountain regions of Mexico, Central America, and the Andes), nearby collection points (≥1 km) were considered as independent localities. The number of collection localities for each species is shown in Table 1. Additional data on all the unique localities used for our analysis can be consulted in Appendix S1 (supporting information).

| Calculation of the thermal niches
Our concept of thermal niche includes an important difference in the definition of TZ and RZ from the concept generated by the work of Brett (1952). While TZ and RZ were defined in the past at an individual level, measuring the survival time of individual animals at different temperatures, our concept reflects the response of a species to environmental temperatures. We determined the TZ and RZ of Glossophaginae bats using our unique localities database. For each of our studied bat species, we plotted the distribution of the number  To describe the thermal niche, we generated a database of minimum and maximum temperatures of all unique localities for each species. We used monthly temperature data obtained at www.worldclim.
org (Hijmans, Cameron, Parra, Jones, & Jarvis, 2005). This procedure was performed using ESRI ArcGIS © version 10 (Redlands, CA 1999CA -2010. For minimum temperatures, we used the value of the coldest temperature recorded for each locality. Because bats are nocturnal animals that use shelters during the day, they are not exposed to daily maximum temperatures. Unfortunately, there is no available information on maximum night temperatures at global level. Therefore, to determine the maximum temperature limits of the thermal niche of bats, we used the minimum value of the maximum temperatures recorded at each locality as a conservative proxy for maximum temperatures that bats could find at night. We then calculated the thermal niche obtaining the STZ and SRZ separately for each species. As previously mentioned, to calculate thermal STZ and SRZ, we constructed two curves for each species, one for the minimum temperatures, and the second for the maximum temperatures. We calculated the left inflection point for the minimum temperature curve (STZ lower ), and the right inflection point for the maximum temperature curve (STZ upper ; Figure 1). We considered the two inflection points as the limits of the STZ, while the limits of SRZ were calculated using the minimum (SRZ lower ) and maximum values (SRZ upper ) of each one of the temperature distribution curves (Figure 1).

| Reconstruction of ancestral states and evolution of thermal niche in nectar-feeding bats
Understanding the evolution of thermal niches requires a robustly supported phylogenetic hypothesis as framework. Thus, our taxon sampling is in line with the current knowledge of the evolutionary history Phylogenetic hypothesis within the subfamily Glossophaginae are well-supported and, in general, they agree with the current understanding of the systematics of the group (e.g., Amador et al., 2016;Dávalos et al., 2014). We sampled 100% of the genera and 88% of the species of the subfamily Glossophaginae analyzed by Rojas et al. (2016). In addition to the species considered by Rojas et al. (2016), we included Leptonycteris nivalis in our analyses.
In order to investigate how STZ and SRZ have changed throughout the evolutionary history of nectar-feeding bats, we inferred the ancestral states of all four elements of their thermal niche (STZ lower , STZ upper , SRZ lower , and SRZ upper ). As mentioned previously, in order to conduct the reconstruction of ancestral states and the evolution of the thermal F I G U R E 1 Method used to determine the thermal niche of Neotropical nectar-feeding bats. We delimitated the Species Tolerance Zone (STZ) and Species Resistance Zone (SRZ) using a database of unique localities. We constructed two curves for each species, one for the minimum temperatures (solid line), and the second for the maximum temperatures (dash line). We calculated the left inflection point for the minimum temperature curve (STZ lower ), and the right inflection point for the maximum temperature curve (STZ upper ), while the limits of SRZ were calculated using the minimum (SRZ lower ) and maximum values (RSZ upper ) of each one of the temperature distribution curves. See methods section for more details niche of nectar-feeding bats, we followed the phylogenetic hypothesis from Rojas et al. (2016). To place L. nivalis in that phylogenetic hypothesis, we ran a Bayesian inference using the 119 morphological characters described by Carstens et al. (2002), and 658 base pairs of the mitochondrial gen Cytochrome Oxidase subunit 1. Even though our phylogenetic analyses included only a subset of the full sampling, and different characters from those used in Rojas et al. (2016), it was adequate to determine the position of L. nivalis relative to other congeneric species. These analyses kept morphology and genetic data separately and were based on eight independent runs and 10,000,000  (Maddison & Maddison, 2015). In a descriptive manner, we used "+" and "−" symbols to indicate species that presented a higher or lower temperature value in relation to their ancestral state in figure 3. In similar fashion, we used "0" if we did not found changes between the ancestral value of the thermal variable and the values calculated for the current thermal niches.
Finally, we conducted phylogenetic signal analyses on the TZ lower , STZ upper , SRZ lower and SRZ upper data of the bat species included in our phylogeny. Phylogenetic signal is defined "as a tendency for related species to resemble each other more than they resemble species drawn at random from the tree" (Blomberg & Garland, 2002). To determine the existence of a phylogenetic signal, we calculated Blomberg's K (Blomberg, Garland, & Ives, 2003) using R package Picante (v. 3.0.2) (Kembel et al., 2010;R Core Team, 2016). showed stabilized variances for maximum temperatures. We decided to include species that did not showed stabilized variances in temperature data in our study because there is very little information on those species (see Table 1), and our analyses provide a starting point to understand their thermal biology, and to promote research on this topic. For these species, our results need to be considered as preliminary.

| The thermal niche of Neotropical nectarfeeding bats
Nectar-feeding bats thermal niches showed a high variability in the SRZ lower limit (5.5 ± 5.3°C, mean ± SD), while the SRZ upper limit tended to be similar among species (32.5 ± 1.6°C), and did not exceed 35°C for any species (

| Ancestral states of the thermal niche
The

| DISCUSSION
Thermal niches of the nectar-feeding bats of the subfamily Resistance Zone. Third, we discuss the evolution of the thermal niche in this clade of nectar-feeding bats. And finally, we use the characteristics of the thermal niche to understand the capabilities of these organisms to withstand changes in ambient temperature generated by anthropogenic climate change.

| Laboratory studies of thermal biology of nectarfeeding bats
There is limited information on the metabolic responses to temperature of Glossophaginae bats from laboratory studies. The existing research has determined lethal temperatures, and/or metabolic curves (Scholander curves) for only seven species, and we do not have values of critical and lethal temperatures for all of them (Arends, Bonaccorso, & Genoud, 1995;Carpenter & Graham, 1967;Cruz-Neto & Abe, 1997;McManus, 1975;McNab, 1969;Soriano, Ruiz, & Arends, 2002). The trait best studied in the laboratory is  McNab, 1989;Arends et al., 1995), ignore the capacity of animals to obtain energy, or use energy reserves, and do not allow us to determine the net energetic cost of a shift in ambient temperature for the bats. If an animal has a high capacity to acquire energy, like nectar-feeding bats do, a small, or even a large, increase in metabolic costs due to thermoregulation could be irrelevant for the species under natural conditions but not in the laboratory when experimental individuals were fasting (Ayala-Berdon, Schondube, & Stoner, 2009). This could explain why the lower critical temperature data was not correlated with minimum field temperatures.
This problem may seriously cripple our capability to use metabolic curves to understand the thermal niches of nectar-feeding bats in a real ecological context.
Geographic presence data from natural history museums records integrate the physiological characteristics of species (that define their fundamental niche) with environmental factors. The intersection of intrinsic (physiology) and extrinsic factors (bionomic and scenopoetic niche axes sensu Hutchinson′s 1978), determine the capacity of a species to be present and survive at a given geographic site (Peterson et al., 2011). While from a geographic locality we can obtain environmental (i.e., temperature, precipitation among others), and topographic data (i.e., elevation, slope), this type of information also conceals data on microhabitat, species interactions and diet quality, by proving a biogeographic context for the species (Peterson et al., 2011). This perspective of the niche, from the point of view of some of its axes, while myopic (sensu Newsome, Martinez del Rio, Bearhop, & Phillips, 2007), offers us a snapshot of the costs and benefits that animals face in the field, and provides critical information to understand when laboratory physiological data are relevant to understand the ecology of a species.

| Upper and lower limits of the thermal niche in nectar-feeding bats
Our results indicate that nectar-feeding bats have an average SRZ upper value of 32.5°C ± 1.6°C, while the mean value of the SRZ lower was 5.5 ± 5.3°C. We also found that the SRZ upper values showed a phylogenetic signal. The SRZ upper values we found in our research are similar to those reported by a study that synthesized the thermal tolerances of a large number of terrestrial ectotherm and endotherm organisms from a wide arrange of geographic areas (Araújo et al., 2013), and those of a comparative study of 85 species of rodents (Bozinovic et al., 2014). Both Araújo et al. (2013) and Bozinovic et al. (2014) found that the upper limit of the thermal niche was shared by most species of mammals, and was located close to 34°C, while the lower limit of the thermal niche was labile. In our study group, this low variation in the thermal upper limits of the species can be explained by two complementary hypotheses: (1) high environmental temperatures are less variable than cold temperatures (Addo-Bediako, Chown, & Gaston, 2000;Boher, Godoy-Herrera, & Bozinovic, 2010), generating an upper limit to the thermal niche that varies less than its lower limit (Araújo et al., 2013;Bozinovic et al., 2014) and (2)  conditions. This would allow the higher limits of the thermal niche to be controlled by biochemical thermal limits, while the lower limits of the thermal niche could vary more in response to differences in energetic acquisition/thermodynamic effects of species present in colder localities (Araújo et al., 2013).

| The evolution of thermal niches in nectarfeeding phyllostomid bats
The  et al., 2016), with the first nectar-feeding species in the subfam- having colder thermal niches. Koopman (1981) proposed that the genus Anoura was "fairly primitive" (meaning basal) in this clade, suggesting that the Choeronycterines evolved associated with the cooler conditions found in the mountain areas of South and Central America, with a subsequent adaptation of some group to warmer conditions (i.e., genera Choeroniscus, Lychonycteris, and Musonycteris; Koopman, 1981;Gardner, 2007). This is supported by the altitudes at which these different genera of bats are generally found (Ceballos & Oliva, 2005;Eisenberg, 1989;Eisenberg & Redford, 1999;Gardner, 2007;Redford & Eisenberg, 1992).
Our results indicate that the clade of the Glossophagines had an ancestral state of their thermal lower limits related to warmer temperatures. This clade includes six genera associated with the humid and arid tropical low lands of South America, the Caribe, Central America and Mexico (Eisenberg, 1989;Gardner, 2007;Redford & Eisenberg, 1992;Rojas et al., 2016;Silva, 1979;Villa-R., 1967). Our study suggests that this clade may have evolved their thermal niches as an adaptation to warmer climate conditions. Rojas et al. (2016) suggested that the basal clade of the Glossophagines (composed by

| Relationship between thermal niches and the ability of Neotropical nectar-feeding bats to withstand global warming
By relating the thermal niches of our study species with models of climate change, we formulated a conservative hypothesis of the responses of nectar-feeding bats to changes in ambient temperature.
Several simulations of anthropogenic caused climate change (i.e., CMIP5, RCP4,5, RCP6,0, RCP8,5) project increases in temperature at the end of this century that vary from 1.5 to 2.8°C for different tropical and subtropical areas in America (Diffenbaugh & Giorgi, 2012;IPCC 2014 However, because we are assuming a constant temperature increase in all localities, the values of our projections need to be considered with caution. Furthermore, we found a negative relationship between the percentage of localities that moved its temperature values outside the SRZ and the width of the thermal niche of the different species (R 2 = .19, p = .037). Additionally, the thermal niche width was positively related to the size of the geographic distribution area of our study species.
The relationship between these two results suggests that species with restricted distributions could be more affected by global change, as have been previously stated by Walther et al. (2002).
Finally, our study shows that locality data obtained from natural history museums could provide crucial information to determine the physiological parameters of species. The method we proposed here to describe the thermal niche of Neotropical nectar-feeding bats, by using temperature data linked to locality records, offers the possibility to work with large number of species, and generates physiological data that are ecologically relevant in a critical moment of history calling for urgent action to address anthropogenic climate change.

ACKNOWLEDGMENTS
We like to thank Danny Rojas, who kindly shared with us the phylogenetic tree we used to conduct our phylogenetic analyses. We also want to thank Bryan Carstens for sharing with us the morphology data we used to conduct the phylogenetic analysis in an earlier version of this manuscript. Additionally, we thank Demetris Christopoulos who solved all our doubts related to the method used to calculate the inflection points of the temperature distribution curves and Jorge Cortés who help us to conduct the phylogenetic signal analyses.
SOG acknowledges the scholarship and financial support provided by the National Council of Science and Technology and PAEP-UNAM.