Food habits of the threatened bat Leptonycteris nivalis (Chiroptera: Phyllostomidae) in a mating roost in Mexico

The objective of this study was to collect and present data on the food plants that a vulnerable migratory nectarivorous bat, Leptonycteris nivalis, used during its seasonal stay at a cave in Tepoztlan, Mexico. This cave is the only mating roost known for this species. Pollen grains from faeces and fur were examined. Little is known about the diet of this bat, especially in the southern half of its distribution. Pollen belonging to 10 plant species was found from five families: Cactaceae (Stenocereus beneckei), Bombacaceae (Ceiba aesculifolia, Pseudobombax ellipticum), Convolvulaceae (Ipomoea arborescens), Fabaceae (Calliandra houstoniana, Bauhinia ungulata), and Amaryllidaceae (Agave dasylirioides, A. horrida, A. inaequidens, A. salmiana). Bats showed the highest dietary diversity in November. There were no significant differences in diet between sexes. Of the four Agave species, two have paniculate inflorescences that are commonly attributed to bat pollination, while the other two had spicate inflorescences, which are considered to be primarily insect‐pollinated.


Introduction
Leptonycteris nivalis (Saussure, 1860) is one of the largest nectarivorous bats (Hensley and Wilkins 1988), attaining a body mass of up to 24 g. Its distribution ranges from south New Mexico and west Texas (USA) to south Mexico and Guatemala (Simmons 2005). Although their migration ecology is not well understood, preliminary evidence suggests that females remain in south-central Mexico during the mating season, between September and February (autumn to winter). In early spring (March), pregnant females apparently depart for northern areas to establish maternity roosts; at the end of summer, the young are weaned, maternity roosts disband, and adults and young bats migrate back toward southcentral Mexico (Téllez 2001).
Authors have studied certain aspects of the diet of L. nivalis mostly in the northern part of its distribution (Á lvarez and González 1970;Easterla 1972;Gardner 1977;Hevly 1979;Moreno-Valdez et al. 2004). Also, there is some indirect information on its diet, as a result of pollination system studies (i.e. Schaffer and Schaffer 1977;Eguiarte et al. 1987;Valiente-Banuet et al. 1997a, 1997bArizaga et al. 2000;Slauson 2000). However, in the southern and central part of its distribution there is little information about the diet of L. nivalis. There is no information on the diet of bats inhabiting Cueva del Diablo, Morelos, Mexico, which is the only mating roost known for this bat (Téllez 2001). Because the energy requirements of female bats increase during reproduction and lactation periods, it is feasible that their diets may differ from those of males during such periods (Kunz and Nagy 1988;Racey and Entwistle 2000). Indeed, several studies have suggested differences in food requirements between male and female nectarivorous bats (Sánchez-Casas and Á lvarez 2000; Riechers et al. 2003;Tschapka 2005). However, actual data about all these topics are scarce for the entire guild.
Leptonycteris nivalis is considered endangered by the IUCN (Chiroptera Specialist Group 2006) and by the US government (US Fish and Wildlife Service 1994). In Mexico L. nivalis is categorized as a threatened species (Diario Oficial de la Federació n 1991, 2002. The inclusion of this species in endangered species lists is a consequence of recent decreases in abundance that have been mainly attributable to various human-related factors (Wilson et al. 1985;US Fish and Wildlife Service 1994;Medellín 2003). The Mexican long-nosed bat (Leptonycteris nivalis) Recovery Plan (US Fish and Wildlife Service 1994) suggests that the two most important steps for the recovery of this bat are: gain more information about roost status (caves) and foraging habitats. The latter is particularly important because very little is known about the composition and distribution of their alimentary resources (US Fish and Wildlife Service 1994).
In order to provide crucial data for conservation and recovery purposes, the goals of our study were: (1) to document the diet of Leptonycteris nivalis and its temporal changes during their mating season in the Cueva del Diablo, Morelos, Mexico,and (2) to test whether there are dietary differences between males and females during the mating season.

Methods
Vegetation in the area includes temperate (Pinus spp., Quercus spp.) and tropical deciduous forests (Hoffmann et al. 1986). The climate is subtropical with most of the rainfall occurring in summer. The climatic pattern following García (1973) is (A)Ca(w 2 )(w)(i9)g: mean annual temperature 19.9uC, and rainfall is 1463.2 mm (INEGI 2003). The Cueva del Diablo is in the municipality of Tepoztlan, in the state of Morelos, Mexico (18u599N,99u039W), at an elevation of 1883 m a.s.l. The cave has 26 corridors and 11 galleries (Hoffmann et al. 1986). The bat colony in the main gallery was estimated to contain a maximum of about 3000 individuals between October and December of 2001 (Caballero 2004).
At each visit bats were captured inside the cave in a small corridor located 250 m from the entrance, where about 400 individuals typically roosted. A 6 m mist net (Avinet Inc., Dryden, NY, USA) was used. The net was open for 4 h during the period when bats returned to the cave after foraging (between 0:00 and 4:00 h). All captured animals were weighed (precision ¡0.5 g); forearm length (caliper, precision: ¡0.1 mm), sex, and the reproductive status of females (not reproductive, pregnant or post-lactating) and males (not reproductive or enlarged testicles) was scored.
Pollen was sampled from the fur of captured bats, using gel-safranine squares rolled over the bats' bodies (Medellín et al. 1983). Permanent microscope slides of each sample were prepared, sealing them with transparent nail polish. Faecal samples were obtained by keeping each captured animal in a clean cloth bag for ca 1-2 h. Each faecal sample was stored in a labelled paper bag. In order to prepare pollen grains for identification the acetolysis standard technique was used (Erdtman 1966), with the modification proposed by Horowitz (1992). This additional step eliminates organic matter present in the faeces (Á lvarez and González 1970;Horowitz 1992;Bruch and Pross 1999), with the exception of sporopollenin, chitin or pseudochitin which are the main components of the exine of pollen grains. Slides were prepared by mixing a subsample of the pollen purified in the acetolysis technique with melted glycerine gel (Á lvarez and González 1970).
Pollen grains were identified using a light microscope (Olympus, CH30; 406 and 1006) and by comparing the morphological properties and measurements of pollen grains from our samples to reference samples in the Palynological Collection of the Instituto de Geologia of the Universidad Nacional Autonoma de Mexico (UNAM) and the Palynological Collection of the MEXU Herbarium (Instituto de Biologia, UNAM) as well as with photographs and descriptions from other studies (Palacios 1968(Palacios , 1970González 1969;Quiroz et al. 1986). Slides of samples from faeces and fur were deposited in the Palynological Collection of the Instituto de Geologia (UNAM).
To determine an individual bat's diet, the presence/absence of pollen species was recorded per sample, following Thomas (1988). Chi-square tests were used with Yates' correction, to determine significant differences in total frequencies of pollen records between paniculate and spicate agaves and in plant use between sexes. Pollen grains from anemophilous plants (like pine and corn) and fungal spores were discarded, because their presence in the samples was considered accidental.

Results
Between September 2001 and February 2002 the Cueva del Diablo was visited approximately every 2 weeks for a total of 11 visits. During our visits a total of 136 individuals (116 males and 20 females) were captured, and 123 fur and 98 faecal samples were collected. Ten food plants were identified: Cactaceae (Stenocereus beneckei), Bombacaceae (Ceiba aesculifolia, Pseudobombax ellipticum), Convolvulaceae (Ipomoea arborescens), Fabaceae (Calliandra houstoniana, Bauhinia ungulata), and Amaryllidaceae (Agave dasylirioides, A. horrida, A. inaequidens, A. salmiana). All 10 species are new records for the diet of Leptonycteris nivalis, and two of the genera are also new (Stenocereus and Bauhinia) (Table I).
Pollen from all plant species was found in both fur and faecal samples, with the exception of Stenocereus beneckei, which was recorded only in the fur samples. Pollen of Ipomoea arborescens and Agave dominated the faecal samples and were the only two pollen groups present in more than 70% of the faecal samples; all remaining pollens were found in less than 30% of the samples. On fur Ipomoea arborescens was most frequently detected (93.49%) and the remaining plants were found at frequencies higher than 50%, with the exception of Bauhinia ungulata (4.87%) ( Figure 1).
The pollen of four species of Agave was identified in faecal samples, but in the fur samples identification of pollen was only possible to the level of genus, primarily due to the abundant debris in each gel-based slide that prevented the observation of species-specific diagnostic characteristics. Combining all four species of Agave together, 75% of the faecal samples and 57% of the fur samples showed Agave pollen (Figure 1). In faecal samples, Agave dasylirioides and A. horrida (spicate inflorescences with flowers arranged in rows along the elongated scape) dominate the samples with 54 and 45% respectively; while A. Insect exoskeleton parts were found in 28.57% of the faeces samples and 34.14% in fur samples. Insect parts were too finely chewed to allow identification. Butterfly scales were also found, but not included in the analysis. Flower-visiting insects leave scales and other remains on them, which easily stick to bat fur; but finding them in the faeces does not mean that the bat ate the insects.
The observed monthly frequencies differed between sample types ( Figure 2): on fur, the frequencies of many pollen types were high, but in faeces, only Ipomoea arborescens had particularly high frequencies in all months (.80% in monthly samples). In both sample types, Bauhinia ungulata showed low frequencies and was only found in October and November (Figure 2), which corresponded to months with the highest food species diversity. In December, many species showed a decrease in frequencies and the diet was dominated by Ipomoea arborescens.

Discussion
There is little information about the dietary habits of L. nivalis. Much more is known for the other Mexican species of the genus, L. curasoae (Miller, 1900). Apparently, many studies reporting information for L. nivalis were actually based on observations on L. curasoae (Arita and Humphrey 1988). Because of this confusion, information generated prior to Arita and Humphrey (1988) must be considered carefully. Taking into account only the published reports that are known to refer to L. nivalis with certainty, the number of plants reported in its diet consist of 21 species, 13 genera, and 10 families (Table I).
With the exception of two genera (Stenocereus and Bauhinia), all plant genera that we report in this paper had been previously reported to be part of the diet of Leptonycteris nivalis (Á lvarez and González 1970;Gardner 1977;Dobat and Peikert-Holle 1985). However, at a species level, our study adds 10 new food plants to the known diet of this rare, endangered bat.
The diet of L. nivalis probably tracks the spatio-temporal availability of local floral resources, while this bat inhabits the Tepoztlan cave. Opportunistic feeding behaviour has been well documented for both species of Leptonycteris in the northern parts of their distributions (Bogan et al. 2004;Fleming 2004;Moreno-Valdez et al. 2004). For example, Moreno-Valdez et al. (2004) found that the seasonal presence of a colony of an estimated 4000 L. nivalis in northern Mexico from April to October was highly correlated with availability of nearby flowering Agave. Although the importance of Bombacaceae and Ipomoea as food resources for L. curasoae in the southern areas of its distribution has been recognized (Fleming 2004), such a relationship was not previously known for L. nivalis.
The Morning Glory tree or cazahuate (Ipomoea arborescens) is a dominant plant in the Tepoztlan Mountain range and occurs throughout the tropical deciduous forest regions of Mexico (Ramírez 1944). It is therefore not surprising that this abundant plant dominates the diet of L. nivalis in all months. The proposed food shift of Leptonycteris between its northern (CAM metabolism plants, like Agave and Cactaceae) and south-centre roosts (C 3 metabolism plants, like Ipomoea), has been demonstrated by using carbon isotope analysis (Fleming et al. 1993;Téllez 2001). Our results show that in the southern part of the distribution, these bats are feeding on C 3 plants (mostly Ipomoea), and also on CAM plants (Agave). The difference is that in the north L. nivalis apparently only feeds on CAM plants (i.e. Schaffer and Schaffer 1977;Moreno-Valdez et al. 2004), principally Agave. In 24 faecal samples and 41 fur samples obtained during two visits (17 July and 2 August) to two maternity colonies in the state of Zacatecas (670.23 km N of El Diablo Cave), only pollen of Agave was detected.
Several authors have suggested that paniculate agaves (species with branched scapes and flowers in large umbellate clusters, like Agave salmiana and A. inaequidens) are pollinated chiefly by bats, while spicate agaves (species with unbranched scapes and with flowers forming directly on the main shoot, like A. dasylirioides and A. horrida) show pollination syndromes that suggest a predominance of entomophilic mechanisms (Schaffer and Schaffer 1977;Fleming et al. 1993). Thus, we did not expect to find pollen of spicate agaves in our samples from Leptonycteris nivalis, much less that the diet of L. nivalis in Central Mexico during the winter would be dominated by spicate agaves. However, other authors have also documented the presence of spicate agave pollen in the diet of bats such as L. curasoae and Choeronycteris mexicana (Tschudi, 1844) in southern USA (Cockrum and Hayward 1962;Schaffer and Schaffer 1977;Hevly 1979).
It has been suggested that spicate species of Agave that display certain chiropterophilous characteristics were originally bat pollinated and experienced a recent evolutionary shift towards insect pollination (Schaffer and Schaffer 1977). In temperate zones, this change toward entomophily could be explained by the unpredictable and asynchronous arrival of bats in relation to blooming of Agave (Arita and Humphrey 1988;Slauson 2000). Thus, plants may have developed flower characteristics that favour diurnal pollination by insects (Fleming et al. 1996). Although further research is needed, it is possible that this may also have happened with the spicate agave species that Leptonycteris nivalis feeds on near Cueva del Diablo. The high abundance of spicate agaves in the Tepoztlan Mountain range makes these plants a very attractive resource for the bats. Furthermore, in the case of A. dasylirioides, these plants frequently grow next to paniculate agaves like Agave inaequidens, forming dense clusters (A. García, personal communication). These clusters constitute high flower-density patches that might attract bats by maximizing feeding efficiency (Schaffer and Schaffer 1977;Eguiarte 1983) which in turn increases the pollination probability for both species.
The persistence of pollen grains in fur and faeces is influenced by many factors: gut transit time, the position of deposition on the animal's body, length of feeding time, and the amount of pollen grains produced by the flowers as well as timing of open flower and nectar secretion (Howell 1977;Heithaus 1982;Thomas 1988;Herrera and Martínez del Río 1998). These factors could contribute to the differences in plant frequencies between the two sample types. Therefore both methods should be used simultaneously but should be considered separately when collecting data on the diet of bats. For example, the flowers of Stenocereus beneckei open early in the evening, at approximately 18:30 h (Gentry 1982), so the bats visit them at the beginning of the night. When bats return to the cave groomed, there would be almost no S. beneckei pollen grains in the fur.
Differences in the dominance of certain pollen types in the faeces and fur might be related to the last meal the bats had before heading back to the roost. Both fur and faeces provide diet information, but probably over different time scales. Given the high food intake in nectar-feeding bats (Helversen and Winter 2003) it is possible that the faeces examined reflect only a rather short period of foraging immediately before the capture (Thomas 1988). Perhaps the most frequently detected type of pollen in the faeces might be the food plant that is most abundant in the area around the cave, and was eaten just before bats returned to the roost. If this is indeed the case, then capturing bats in their feeding grounds, as opposed to in their roost, might yield somewhat different results.
Our diet data did not reveal sexual differences in the diet of L. nivalis in Cueva del Diablo. During their stay in Tepoztlan, bats were mating but not forming a maternity colony. Thus, embryo development was only in its earliest stages and the physiological requirements at that time may not vary between the sexes (Kunz and Nagy 1988).

Conservation implications
Since the 1970s many studies have documented severe declines in the population of L. nivalis (Easterla 1972). The causes of this decrease have not been completely identified, but are probably related to human activities (Medellín 2003). The recovery plan for this species (US Fish and Wildlife Service 1994) recognizes destruction and modification of foraging and roosting habitat as the greatest threats to the species. For this reason, it was suggested that protection of all known roosts and protection of foraging habitat are the most important aspects to be considered for the recovery of L. nivalis (US Fish and Wildlife Service 1994). Our research addresses one of the major points of the Recovery Plan for Leptonycteris nivalis (US Fish and Wildlife Service 1994). Before our study, there was no detailed information on this species' diet in the southern parts of the range or during the mating season. According to our data, Ipomoea arborescens (Morning Glory tree) is the most important food plant of L. nivalis near this critical breeding site in Mexico. At other times of year and in the north of the distribution several species of Agave are the main source of food for this species (Fleming 2004;Moreno-Valdez et al. 2004). This information allows us to gain a wider perspective about the relevance of the vegetation types and foraging area that need to be preserved for the purpose of recovery of this species.
Many species of plants seem to rely on L. nivalis and other nectar-feeding bats for fruit production (Fleming 2004). Some plant species we found in the diet of this bat are of conservation concern in Mexico. Agave dasylirioides is listed as a threatened and endemic plant species (Diario Oficial de la Federació n 2002), while A. inaequidens and A. horrida are endemics to the floristic provinces of Sierra Madre Occidental and Sierra Madre del Sur, respectively (García-Mendoza and Galván 1995). Because of the potential symbiotic relationships between these plants and bats we need to protect both. By doing this, we ensure adequate supplies of food to bats as well as the successful pollination of plants.
Further studies should evaluate the role of bats in the pollination of plants such as the Morning Glory tree (Ipomoea arborescens), as well as the spicate and paniculate agaves in Central Mexico.