OUP user menu

Keystone resource (Ficus) chemistry explains lick visitation by frugivorous bats

Adriana Bravo, Kyle E. Harms, Louise H. Emmons
DOI: http://dx.doi.org/10.1644/11-MAMM-A-333.1 1099-1109 First published online: 14 September 2012


Geophagy is a widespread behavior among plant-eating animals. In the Neotropics, mineral licks are activity hot spots for frugivorous bats (Stenodermatinae). Bats drink mineral-rich water accumulated in soil depressions made by geophagous mammals. Two mechanistic hypotheses have been proposed to explain this behavior: licks are reliable sources of limiting nutrients, especially sodium; and licks provide substances that render dietary toxins less harmful. We assessed the former by examining bats’ diets in conjunction with lick chemistry in the Peruvian Amazon. We found that most bats that visit licks belong to the subfamily Stenodermatinae and are specialists on Ficus fruits—a keystone resource. In addition, although Ficus fruits are good sources of some minerals, their sodium content is limited in relation to the physiological requirement of a small mammal. In contrast, bats of the subfamily Carolliinae supplement their fruit diets with insects, potential sources of sodium. Complementary results among diets, Ficus chemistry, and lick-water chemistry strongly support the sodium-limitation hypothesis for bat lick use and suggest a mechanistic link between bats and ecosystem engineers that make soil-borne resources available. Because sodium is an essential nutrient for vertebrates and Ficus is a keystone resource for many animal species, our results may have implications for the community of frugivorous vertebrates in areas where sodium is limited. Licks may play a critical role as sodium sources and thus they should be considered as important conservation targets.

Key words
  • Amazonia
  • bats
  • Carolliinae
  • figs
  • limiting resources
  • mineral licks
  • sodium
  • Stenodermatinae

Geophagy, the deliberate consumption of soil, is a widespread behavior among plant-eating animals (Gilardi et al. 1999; Klaus and Schmid 1998; Krishnamani and Mahaney 2000; Lee et al. 2010). This unique phenomenon has captured the attention of researchers, who have attempted to determine its underlying mechanisms. As a result, a set of hypotheses ranging from physiological to social explanations has been proposed (Burger and Gochfeld 2003; Davies and Baillie 1988; Gilardi et al. 1999; Klaus and Schmid 1998; Mahaney et al. 1995).

In the Neotropics, natural soil licks are activity hot spots for several frugivorous bats that drink water that accumulates in puddles (Bravo et al. 2008, 2010b) created by geophagous mammals (Izawa 1993; Klaus and Schmid 1998; Tobler et al. 2009). Even though eating soil and drinking water differ in whether the bulk of the ingested material is solid versus liquid, 2 non–mutually exclusive, mechanistic hypotheses have been proposed to explain these behaviors: licks are reliable sources of limiting nutrients, especially sodium (Bravo et al. 2010b; Brightsmith et al. 2008; Emmons and Stark 1979; Powell et al. 2009); and licks provide substances that render dietary toxins less harmful (Brightsmith et al. 2008; Gilardi et al. 1999; Voigt et al. 2008). These hypotheses have mainly been addressed by examining characteristics of soil or water preferred by animals that visit licks (Bravo et al. 2010a, 2010b; Brightsmith and Aramburu 2004; Emmons and Stark 1979; Izawa 1993; Powell et al. 2009; Wilson 2003). However, few studies have examined the organisms' diets in conjunction with substances consumed at licks (e.g., Brightsmith et al. 2008; Gilardi 1996). Here, we studied a group of Neotropical frugivorous phyllostomatid bats (Stenodermatinae) whose members in the southeastern Peruvian Amazon regularly visit natural soil licks to drink water that has accumulated in puddles (Bravo et al. 2008, 2010b) and compared them to related species (Carolliinae) that rarely visit licks, to determine whether the chemistry of preferred fruits is consistent with the nutrient-limitation hypothesis.

Although frugivorous bats of the subfamilies Carolliinae and Stenodermatinae (family Phyllostomidae) are common in Neotropical assemblages (Gardner 2008), puddles at licks are visited nearly exclusively by stenodermatine bats (Bravo et al. 2010b; Voigt et al. 2007). Reproductive females visit licks out of proportion to their relative abundances in their respective populations (Bravo et al. 2008, 2010b; Voigt et al. 2007).

The striking difference in visitation patterns to licks by carolliine and stenodermatine bats coincides with a striking difference between their general diets, at least as reported in the literature. Most stenodermatines are consistently reported to specialize on Ficus fruits (Ascorra et al. 1996; Giannini and Kalko 2004; Kalko et al. 1996)—keystone resources in Neotropical forests (sensu Terborgh 1986). Sturnira species are exceptions to this general pattern, because they feed mainly on Solarium fruits (Fleming 1986). In contrast, carolliines are consistently reported to be Piper specialists (Ascorra et al. 1996; Fleming 1988; Giannini and Kalko 2004). Both the nutrient-limitation and the dietary-toxin hypotheses could be consistent with these dietary patterns. If nutritional requirements are not met by—or dietary toxins are present in—the diets of lick-visiting stenodermatines, members of that group may seek substances (e.g., mineral nutrients or clay) at licks, especially during reproduction, that is, periods of high nutritional demand (Barclay and Harder 2003; Nelson et al. 2005).

Licks are sources of mineral-rich water (Bravo et al. 2010b; Izawa 1993). Lick water contains a consistently high concentration of selected minerals, especially sodium, in comparison to other water sources also available to bats in regions where licks are found (Bravo et al. 2010b). Bravo et al. (2010a) also showed experimentally that stenodermatine bats prefer lick water over other water sources. Thus, it is very likely that lick water provides 1 or more important resources to bats.

Determining the nutritional quality of plant species consumed by carolliine and stenodermatine bats should provide additional insights into which substances bats may be seeking at licks. In particular, low concentrations of key mineral nutrients in the diets of lick-visiting stenodermatines compared to nonvisiting carolliines would support the nutrient-limitation hypothesis. Accordingly, here we address the hypothesis that key minerals are limited in diets of stenodermatine bats in the Peruvian Amazon. We determined the diet compositions of 22 bat species and assessed their correlation with lick visitation. We also determined the nutritional composition of Ficus and Piper fruits collected in the area of study and analyzed the concentrations of 4 key minerals, as well as nitrogen (as a measure of protein).

Materials and Methods

Study site.—We conducted this study at Los Amigos Conservation Concession in Madre de Dios, southeastern Peruvian Amazon (12°30′–12°36′S, 70°02′–70°09′W). This private concession protects > 140,000 ha of lowland tropical forest. Average annual temperature for 2005–2007 ranged from 23.9°C to 24.TC, and annual rainfall ranged from 2,152 to 2,682 mm.

Bats' associations with licks.—To determine whether there was a bias by stenodermatine bat species for visiting licks, we compared bat assemblages at licks, forest, and gap sites using species' abundances at each site, as in Bravo et al. (2010b), but supplemented with an analysis of similarity (Magurran 2004). A visual representation of similarities among bat assemblages sampled at lick and nonlick sites was generated with a nonmetric multidimensional scaling analysis, using Bray–Curtis dissimilarities (Gotelli and Ellison 2004). We also compared the total abundances of each bat species captured among site types (lick, forest, or gap; data from Bravo et al. [2010b]) using a goodness-of-fit G-test, for species with expected values larger than 5 individuals (Sokal and Rohlf 1995). To handle and process bats in this study, we followed guidelines of the American Society of Mammalogists (Sikes et al. 2011) and protocol 08-017 approved by the Institutional Animal Care and Use Committee from Louisiana State University.

Fecal samples and diet analyses.—To characterize bats' diets, we collected fecal samples from bats captured at 3 places along the Los Amigos River. At each place from September to November 2005, we captured bats at a lick and a forest site. From July to September 2007 and February to April 2008 we added a gap site to each lick-forest pair. At licks, we used a single 6-m mist net, which generally captured bats at a rate that allowed 2 or 3 people to process them. At gap and forest sites we deployed five to ten 6-m mist nets along previously opened trails. We opened the nets from dusk (∼1730 h) to midnight (2400 h). Because of the large numbers of bats at licks, we closed and reopened those nets as necessary (see Bravo et al. [2008, 2010b] for more-detailed information about sites and methods used to capture bats). After capture, each bat was aged, identified, measured, sexed, and weighed. In addition, we collected fecal samples from the cotton bag where the bat was kept temporarily (no more than 30 min).

Items identified in fecal samples using a dissecting microscope were classified as insects, pulp, seeds, or soil. With the assistance of an experienced field botanist, seeds were classified as Cecropia (Moraceae), Ficus (Moraceae), Philodendron (Araceae), Piper (Piperaceae), Solanum (Solanaceae), Vismia (Clusiaceae), family Cucurbitaceae, or undetermined species. Because of small sample sizes (8% of total fecal samples collected), we grouped samples with seeds of Philodendron, Solanum, Vismia, and Cucurbitaceae into 1 category for analysis. To determine whether stenodermatine bats were associated with a particular diet, we examined the relationship between diet composition (Cecropia, Ficus, Piper, insects, etc.) and bat species using a correspondence analysis for all fecal samples collected across all sites (Gotelli and Ellison 2004). Then, because of our particular interest in the diets of frugivorous bats of the subfamilies Carolliinae and Stenodermatinae, we grouped bats as carolliines, stenodermatines, and “others.” Next, we compared the proportion of each item in the diet across bat subfamilies using generalized linear models with Poisson distributions (Crawley 2007). We tested the effect of diet–subfamily interaction by comparing a saturated model with a model without the interaction using an analysis of deviance that used a chi-square test (Crawley 2007). We then tested the equality of proportions of the most common food items (Cecropia, Ficus, and Piper) for each bat group using a goodness-of-fit G-test (Sokal and Rohlf 1995).

Fruit sampling and analyses.—From February to April 2008 and July to August 2008, we collected ripe fruits from Ficus and Piper species. Twice a week 1 of us (AB) systematically walked along the approximately 50-km trail system of the Los Amigos Biological Station, which covers both floodplain and terra firme forest. We collected intact ripe Ficus fruits from beneath the crowns of fig trees. We collected ripe infructescences (maturity of fruits was gauged by their softness) directly from adult individuals of Piper species. On a given walk, when no ripe infructescences were found, we enclosed unripe ones with a soft mesh cloth to prevent bat consumption until they became soft and were collected a few days later. We collected botanical samples to identify each plant to species. We oven-dried fruits at ∼60°C for ∼15 h. Dried fruits Were analyzed for 12 elements (boron, calcium, copper, iron, magnesium, manganese, nitrogen, phosphorus, potassium, sodium, sulfur, and zinc) by the Soil Testing and Plant Analysis Laboratory at the Louisiana State University Agricultural Center (http://www.lsuagcenter.com) using the following procedure. First, 5 ml of concentrated HNO3 was added to a minimum of 0.5 of g ground, dry plant matter. Second, after 50 min, 3 ml of H2O2 was added and the sample was digested on a heat block for 2.75 h. Finally, samples were cooled and diluted to measure the concentration of minerals using inductively coupled plasma spectrometry. Although we were interested in the mineral content of fruits, we also determined the concentration of nitrogen because of increased protein demand during reproduction (Speakman 2008; Studier and Wilson 1991). Nitrogen concentration was analyzed via dry combustion of a 0.1-g sample using a Leco carbon-nitrogen analyzer (Leco Corp., St. Joseph, Michigan). Concentrations were provided in parts per hundred (%) for most minerals. Sodium and nitrogen concentrations were provided in parts per million (ppm). For comparative purposes we converted parts per hundred to parts per million when necessary.

We explored patterns of both mineral and nitrogen concentrations among fruits of Ficus and Piper with a principal component analysis (Gotelli and Ellison 2004). In addition, using an a priori contrasts analysis of variance (Gotelli and Ellison 2004), we compared the concentrations of nitrogen and 4 key minerals (calcium, magnesium, potassium, and sodium) between Ficus and Piper species. We adjusted the alpha level for all contrasts using the Bonferroni correction method (Gotelli and Ellison 2004). All statistical analyses were performed in R (R Development Core Team 2007).


Bat species' associations with licks.—Stenodermatine bats showed a strong preference for natural soil licks. There was a significant difference between the bat assemblage at licks compared to the qnes at forest (R = 0.94, P = 0.001) and gap (R = 0.96, P = 0.001) sites. On the other hand, there was no significant difference between the bat assemblages at forest and gap sites (R = 0.05, P = 0.28; Fig. 1). These specific results are supported by tables presented in Appendices I–VI. Sixteen of 17 stenodermatine species analyzed were overrepresented at licks compared to nonlick site types (Fig. 2; Appendix V contrast, Carollia brevicauda and C. perspicillata were more common in forest sites and gaps compared to licks (Fig. 2; Appendix V).

Fig. 1

Ordination plot for assemblages of bats captured at lick, forest, and gap sites using a nonmetric multidimensional scaling analysis (stress = 9.83).

Fig. 2

Ordination plot for the correspondence analysis of 22 bat species and 8 dietary items. Bat species are abbreviated as Artibeus lituratus (Al), A. obscurus (Ao), A. planirostris (Ap), Carollia brevicauda (Cb), C. castanea (Cc), C. perspicillata (Cp), Chiroderma salvini (Cs), C. trinitatum (Ct), C. villosum (Cv), Mesophylla macconnelli (Mm), Phyllostomus elongatus (Pe), P. hastatus (Ph), Platyrrhinus brachycephalus (Pb), P. helleri (Phe), P. infuscus (Pi), Phylloderma stenops (Ps), Rhinophylla pumilio (Rp), Sturnira lilium (SI), Thyroptera tricolor (Tt), Uroderma bilobatum (Ub), Vampyriscus bidens (Vb), and Vampyriscus pusilla (Vp). The “Other” category of diet includes seeds of Araceae, Clusiaceae, Cucurbitaceae, and Solanaceae, and “Und” accounts for undetermined species. Circles and diamonds indicate stenodermatine and carolline bats, respectively. Gray circles indicate bat species overrepresented at licks (P < 0.001), whereas gray diamonds show bat species underrepre-sented at licks compared to nonlick sites (P < 0.001). G-values and number of bats at each site are presented in Appendix II.

Composition of bats' diets.—We collected a total of 245 bat fecal samples: 103 samples from 16 bat species captured at natural soil licks, 60 from 12 species captured at forest sites, and 82 from 10 species captured at gaps. At licks, samples were obtained from 15 frugivores of the family Phyllostomidae: 2 carolliines and 13 stenodermatines (Appendix IV). At forest sites and gaps, all but 1 fecal sample belonged to bats of the family Phyllostomidae. Fecal samples from Carollia species were common in forest and gap site types (63% and 88% of total samples, respectively) and rare at licks (∼3% of total samples; see Appendixes IVVI).

There was a clear distinction between the diets of carolliine and stenodermatine species. Most stenodermatine species clustered as Ficus specialists, whereas all carolliine species clustered toward a more diverse diet, mostly composed of Piper but complemented with insects and other fruits (Fig. 2). Eight fecal samples from 5 stenodermatine species captured at licks also contained small amounts of soil. None of the carolliine fecal samples contained obvious soil. In addition, there was a significant interaction between diet composition and the bats' groupings (carolliine, stenodermatine, and “other,” Du = 183.14, P < 0.01). Stenodermatine bats preferred Cecropia (G2 = 23.6, P < 0.001) and Ficus (G2 = 109.9, P < 0.001), whereas Carollia species preferred mostly Piper (G2 = 84.8, P < 0.001) fruits.

Mineral and nitrogen concentration in Ficus versus Piper species.Ficus and Piper fruits differed in their mineral and nitrogen contents (n = 10 Ficus species and 6 Piper species; Fig. 3; see Appendix I for concentration values). From the principal component analysis ∼50% of the total variation was explained by the first 2 components. Principal component 1 explained 31%, whereas principal component 2 explained 21% (Fig. 3). Nitrogen and sulfur contributed the most to principal component 1 (loadings −0.476 and −0.415, respectively), whereas boron and calcium were most influential for principal component 2 (−0.54 and −0.504, respectively; Appendix III).

Fig. 3

Biplot for the two 1st principal components from the principal component analysis of nutrient content of Ficus and Piper fruits.

Ficus fruits had higher concentrations of calcium and potassium compared to Piper fruits (Ca: t1 = 22.92, P < 0.001; K: t1 = 5.50, P < 0.001). In contrast, Piper fruits had higher concentrations of nitrogen compared to Ficus fruits (t1 = −14.90, P < 0.001). No significant differences were found in the concentrations of magnesium (t1 = −1.09, P = 0.3) and sodium (t1 =−2.45, P = 0.03; Table 1).

View this table:
Table 1

Maximum, minimum, and average mineral (calcium [Ca], magnesium [Mg], potassium [K], and sodium [Na]) and nitrogen (N) concentrations in Ficus and Piper fruits. Results of the contrasts analysis of variance between genera are shown by the P-values. An asterisk (*) indicates significant differences for alpha values corrected by the Bonferroni method. % = parts per hundred, ppm = parts per million.

Ca (%)1.808Ficus insipida0.240Ficus sp. 50.7960.599Piper sp. 20.115Piper augustum0.269<0.01*
Mg (%)0.403Ficus americana0.121Ficus jurunesis0.2580.430Piper sp. 10.203Piper sp. 50.2560.30
K(%)2.671Ficus sp. 11.073Ficus americana1.8761.955Piper augustum1.270Piper sp. 21.637<0.01*
Na (ppm)39.391Ficus maxima5.077Ficus sp. 417.40346.000Piper sp. 25.182Piper sp. 520.6280.03
N(%)1.512Ficus maxima0.791Ficus juruensis1.2022.759Piper sp. 31.435Piper sp. 51.757<0.01*


Stenodermatine bats as Ficus specialists.—Most stenodermatine bats in southeastern Peru are Ficus specialists. In spite of the fecal sample size collected (245 samples) relative to the total number of bats captured (2,409 individuals—Bravo et al. 2010b), the consistency in dietary composition suggests that most stenodermatine species in southeastern Peru are Ficus specialists, consistent with other dietary studies in Panama (Giannini and Kalko 2004). Stenodermatines also had a strong preference for licks compared to other common phyllostomid bats (i.e., carolliine bats—Bravo et al. 2008, 2010b).

Contrary to stenodermatine bats, carolliine species had a more diverse diet. Carollia spp. were associated with a diet composed mainly by Piper species (as suggested by Fleming [1988] and Giannini and Kalko [2004] for Central America), but complemented with other fruit species and insects (as found by York and Billings [2009]). Carollia spp. are usually common in open areas, such as gaps, where Piper plants are common (Dumont 2003; Thies and Kalko 2004). However, neither Carollia spp. nor Piper plants were common at licks (open areas). Piper plants were not common, possibly because. of the frequent trampling of small plants by larger geophagous mammals (A. Bravo, pers. obs.). The low number of Carollia spp. found at licks suggests that they do not need lick water as much as do stenodermatine bats.

Carolliine and stenodermatine bats also feed on Cecropia, a relatively abundant tropical genus of trees that produces fruits continuously throughout the year (Dumont 2003). It is also known that frugivorous bats consume some large-seeded plant species (Lobova et al. 2009), seeds of which cannot pass through the bats' guts. Although we recorded the presence of pulp in fecal samples, the contribution of large-seeded fruits may be underestimated. However, many studies have demonstrated that Ficus and Piper species constitute the main component of stenodermatine and carolliine diets, respectively (Ascorra et al. 1996; Giannini and Kalko 2004; Gorchov et al. 1995), likely because contrary to large-seeded species, Ficus and Piper fruits are available year-round (Janzen 1979; O'Brien et al. 1998; Terborgh 1986).

Ficus and Piper nutritional patterns.Ficus and Piper species clearly differed in their nitrogen and mineral concentrations. Nitrogen, the main constituent of proteins (Morris 1991), was present in higher concentrations in Piper fruits than in Ficus fruits. Herbst (1986) and Fleming (1988) likewise presented evidence for a similar pattern of nitrogen concentrations in Piper compared to other fruit species in Central America. However, although some studies have suggested that bats cannot obtain sufficient proteins from Ficus fruits compared to Piper (Herbst 1986; Morrison 1980; Studier and Wilson 1991), Wendeln et al. (2000) found higher concentrations of protein in Ficus insipida (7.9% in dry pulp and 8.5% in seeds) than previously reported, concluding that Ficus was a good source of protein (nitrogen) for bats. In southeastern Peru, concentrations of nitrogen in Ficus and Piper species are higher than in Central America. Thus, frugivorous bats in that region seem to acquire adequate amounts of nitrogen and protein from their fruit sources.

Ficus fruits are rich in calcium. They have higher calcium concentrations than Piper fruits. This calcium-rich pattern for Ficus has been reported for species from around the tropics (Gilardi 1996; Nagy and Milton 1979; O'Brien et al. 1998; Wendeln et al. 2000). Therefore, it is unlikely that Ficus-specialist bats (stenodermatines) face calcium constraints from having a fruit diet. Although Piper fruits contain lower calcium than Ficus fruits, they have enough for Carollia spp. to meet calcium demands of small mammals (5,000 ppm for mice— National Research Council 1995). Furthermore, stenodermatine as well as carolliine species consume Cecropia fruits, which often contain high concentrations of calcium (13,300 ppm—Nagy and Milton 1979). Accordingly, frugivorous bats in southeastern Peru seem able to meet their needs of calcium from their fruit diets.

Unlike calcium, fruits in the southeastern Peruvian Amazon have significantly lower sodium concentrations than fruits in other tropical regions (Appendix III). Similar to Gilardi (1996), who reported 28.86 ppm ± 21.02 SD of sodium for 8 Ficus species collected in southeastern Peru, we found an average of 17.4 ±11.5 ppm and 20.63 ± 15.96 ppm for Ficus and Piper fruits, respectively. In general, it is expected for most plants to contain low concentrations of sodium because contrary to vertebrates, physiologically plants require low concentrations of sodium (Morris 1991). However, compared to other sites in the tropics (Nagy and Milton 1979; O'Brien et al. 1998; Wendeln et al. 2000), sodium seems to be more limited in fruits of southeastern Peru. In Central America, Wendeln et al. (2000) reported a sodium concentration ∼100 times higher in Ficus fruits (1,690 ppm average for 14 Ficus spp.) than what we found in this study. These differences in sodium concentrations among sites may be explained by the reduction in sodium availability in areas located further inland compared to areas close to the ocean (Kaspari et al. 2008). For Piper species data are limited. A single study by Studier et al. (1995) reports an average sodium concentration of 730 ± 60 ppm for species of Piper from northeastern Peru, which is substantially higher than our findings. This difference between northeastern and southeastern Peru may be due to historical processes such as the mid-Miocene marine incursion through the Maracaibo Basin in northern South America (Hoorn 1993; Vonhof et al. 1998). This incursion could have increased the sodium availability in the soils where Studier et al. (1995) collected the samples. Thus, based on the results of our study and others conducted in the same region (i.e., Brightsmith et al. 2008; Gilardi 1996), we conclude that sodium in southeastern Peru is more limited for vertebrate folivores and frugivores than in other regions.

Ficus and Piper species contain sufficient concentrations of magnesium and potassium for bats. Average concentrations of magnesium for both genera (2,580 ppm for Ficus and 2,560 ppm for Piper) surpassed the demands for maintenance and reproduction estimated for small mammals (500 and 600–700 ppm, respectively—National Research Council 1995). Frugivorous bats can thus meet their magnesium demands from their diets. Although there were differences in the concentrations of potassium between Ficus and Piper fruits, both genera contained enough to meet the maintenance and reproductive requirements estimated for small mammals (2,000–3,600 ppm—National Research Council 1995). The concentrations found in this study are similar to those of other localities in the tropics (Gilardi 1996; Nagy and Milton 1979; O'Brien et al. 1998; Wendeln et al. 2000).

Sodium-limitation hypothesis and bat's lick visitation.Ficus chemistry supports the sodium-limitation hypothesis for lick visitation by stenodermatine bats, one of the most speciose Neotropical bat assemblages (Gardner 2008). Sodium, an essential element for osmoregulation, nerve impulses, and muscular function in vertebrates (Michell 1995), is found in significantly lower concentrations in Ficus fruits in the southeastern Peruvian Amazon compared to other geographical regions (Nagy and Milton 1979; O'Brien et al. 1998; Wendeln et al. 2000). Consequently, bats or other animals feeding primarily on fig fruits, or other plants and plant parts with low sodium content, may potentially face sodium constraints unless they supplement their diets with high-sodium sources such as licks. The daily minimal requirements of sodium estimated for small mammals (500 ppm [National Research Council 1995]; which increases during reproduction [Michell 1995]) exceed by 30-fold the concentrations in the Ficus fruits analyzed in the present study. For stenodermatine bats, the daily requirement for an adult Arabeus jamaicensis is 14 mg sodium animal−1 day−1 (Studier and Wilson 1991). If A. jamaicensis feeds exclusively on Ficus with 1,690 ppm of sodium (as in Central America [Wendeln et al. 2000]), bats would need to ingest approximately 10 fruits per day to meet the minimal sodium requirements. In southeastern Peru, a frugivorous bat would need to ingest more than 100 Ficus fruits per day. Because flying to search for fruits demands high levels of energy (Korine et al. 2004; Speakman 2008), it is possible that bats choose less costly mechanisms to supplement their low-sodium fruit diets, such as the use of natural licks, especially during reproduction. Furthermore, high concentrations of potassium in plant tissue can decrease the assimilation of sodium (Weeks and Kirkpatrick 1976), thus sodium deficiency in bats would not be ameliorated only by increasing the consumption of potassium-rich plants.

The complementary results observed in this study among patterns of lick visitation by stenodermatine bats, their specialized Ficus diet, and the low sodium content in Ficus fruits, with the consistently high concentration of sodium in lick water reported by Bravo et al. (2010a, 2010b) strongly support the sodium-limitation hypothesis as an explanation for lick visitation by stenodermatine frugivorous bats in the southeastern Peruvian Amazon. An alternative explanation for lick visitation by bats is that clay renders dietary toxins less harmful for stenodermatine bats (Voigt et al. 2008). However, because ripe Ficus fruits contain low concentrations of secondary compounds (Janzen 1979; Wendeln et al. 2000), this hypothesis does not seem to be the main explanation for lick visitation by Ficus-specialist bats.

Because Piper species also had low concentrations of sodium, carolliine bats feeding exclusively on Piper could potentially face sodium limitation. However, as in other studies, we found that carolliines supplement their diets with insects. Although we were not able to identify insects found in fecal samples, York and Billings (2009) report a variety of insects in the diets of 6 Carollia species. If these insects had significantly higher concentrations of sodium compared to Piper fruits (as found by Studier et al. [1994]—540 ppm for 181 lepidopteran species and 1,660 ppm for 43 coleopteran species from a temperate forest), we could suggest that insects may function as supplementary sources of sodium for carolliine bats. However, data are limited on insect sodium content for the Neotropics. So, further investigation into the mineral content of insects consumed by carolliines is required to completely understand sodium intake in carolliines.

The results of our study are consistent _ with studies conducted on parrot geophagy in the same region of Peru. Parrots consume sodium-rich soils from licks (Brightsmith and Aramburu 2004; Brightsmith et al. 2008; Emmons and Stark 1979; Powell et al. 2009) and plants consumed by parrots have low concentrations of sodium compared to plants from other regions (Brightsmith et al. 2008; Gilardi 1996). In addition, whereas the presence and use of licks by parrots in South America is concentrated in regions where sodium is relatively scarce, licks are absent in nutrient-poor regions, such as the Guianan and Brazilian shields, where it is predicted that plants would have high concentrations of toxins for defense (Lee et al. 2010). Thus, although clay consumption at licks also may provide protection from plant dietary toxins (Gilardi et al. 1999), Lee et al. (2010) suggested sodium limitation is the most-parsimonious explanation for parrot geophagy, similar to our study.

Although there is still no clear evidence to suggest that licks may have an effect on animal biodiversity at large scales, the geographic limitation of sodium in fruits, especially Ficus, could have certain implications on local community structure. Ficus is considered a keystone species in the tropics and is consumed by a great variety of organisms (Janzen 1979; O'Brien et al. 1998; Terborgh 1986). In areas such as southeastern Peru, where Ficus fruits have low sodium concentrations, Ficus-specialist bats would not be able to survive on a diet with such low sodium concentrations. Thus, bat communities would potentially be impoverished if mineral licks were not present. Comparative studies of frugivorous bat assemblages, as well as detailed patterns of sodium content in Ficus and access to natural licks across sites at a continental scales, could provide insights into the mechanisms maintaining tropical bat diversity. In addition, our results suggest an important mechanistic link between frugivorous bats and terrestrial ecosystem engineers (e.g., geophagous tapirs and peccaries [Beck et al. 2010]) that make soil-borne resources available. Bats drink sodium-rich water that accumulates in soil depressions made by larger geophagous mammals that visit licks, such as tapirs and white-lipped peccaries (Tobler et al. 2009). Accordingly, bats benefit from the mechanical action of these mammals at licks. In addition, the role of bats as important dispersers of Ficus (Ascorra et al. 1996; Giannini and Kalko 2004; Kalko et al. 1996) benefits the whole community of Ficus consumers, including tapirs and peccaries. However, large ungulates are among preferred prey of hunters (Bodmer 1995), who use licks as hunting sites (A. Bravo, pers. obs.). If ungulates were extirpated from local communities where bats use licks, the availability of sodium sources for bats would be imperiled, which could potentially affect bat communities and processes of seed dispersal. Consequently, the present study not only provides a key piece of evidence to explain lick visitation by bats, but also reveals an intricate net of interrelationships among tropical plants and mammals that may have numerous implications for understanding and preserving tropical rain-forest ecosystems.


We thank the Peruvian Institute of Natural Resources (INRENA) for providing research permits 070-2005-INRENA-IFFS-DCB, 080-2007- INRENA-IFFS-DCB, and 007-2008-INRENA-IFFS-DCB to conduct this study. For help in the field we thank the Asociación para la Conservacion de la Cuenca Amazónica rangers, Y. Arteaga, M. Bravo, F. Carrasco, S. Claramunt, F. Cornejo, M. Cruz, Z. Ordoñez, A. L. Rodales, M. Rodriguez, and W. Torres. We also thank N. Pitman, J. Ramos, and K. Salas for help with logistics. We are grateful to S. Claramunt, J. Eberhard, S. Galeano, F. Galvez, M. Gavilanez, M. Hafner, L. Hooper-Bui, J. Myers, V. Remsen, R. Stevens, and P. Stouffer for their insightful comments on this study and the manuscript. Financial support was provided by the Amazon Conservation Association (graduate student grant and seed grant award), American Society of Mammalogists, Bat Conservation International, Graduate Student Association—Biograds of Louisiana State University, Idea Wild, Louisiana Office of Environmental Education, Louisiana State University Graduate School, Rufford Small Grants, and the United States National Science Foundation (to KEH).

Appendix I

View this table:

Concentrations in parts per million (ppm) and parts per hundred (%) of 12 elements for fruits of 10 species of Ficus and 6 species of Piper collected in Los Amigos Conservation Concession in Madre de Dios, southeastern Peru. Replicates for a species represent fruit samples collected from different individuals (trees for Ficus and shrubs for Piper).

Family and speciesBoron (ppm)Calcium (%)Copper (ppm)Iron (ppm)Magnesium (%)Manganese (ppm)Nitrogen (%)Phosphorus (%)Potassium (%)Sodium (ppm)Sulfur (%)Zinc (ppm)
Ficus americana21.7170.85010.225118.1840.395337.7111.2320.1281.00625.5610.11323.266
Ficus americana23.1730.67910.08678.0470.412392.6361.2100.1271.14038.5230.11026.446
Ficus insipida16.1641.7259.050153.9070.38542.2451.2430.1732.14819.6830.14517.646
Ficus insipida16.7801.8918.69093.3830.41161.2921.2670.1822.18425.9120.15418.507
Ficus juruensis14.6620.8609.78466.6800.11513.6730.7640.1371.33517.3650.06419.161
Ficus juruensis15.9490.9639.75276.7620.12715.3920.8190.1461.39423.5580.07220.673
Ficus maxima25.4110.8896.49985.1910.23325.8661.5170.1752.44542.0980.13815.884
Ficus maxima25.4220.9006.67966.4870.23936.2341.5070.1802.43736.6830.13716.707
Ficus sp. 118.3830.7937.84047.1420.19126.4391.5100.1952.61915.2840.13616.604
Ficus sp. 119.0560.7497.21146.4440.18525.1001.4820.1882.72228.6660.13516.790
Ficus sp. 214.2850.40913.734287.2870.19112.9031.3080.2222.1028.7090.10923.664
Ficus sp. 214.5080.43812.634149.3040.18812.2171.2810.2242.1375.5670.11024.652
Ficus sp. 314.9010.7096.27137.4070.288177.5361.0280.1282.2267.3010.0858.742
Ficus sp. 413.7470.7187.58748.5860.298210.8721.0120.1241.9788.4630.08811.519
Ficus sp. 413.8960.7796.15831.8340.277203.0440.9790.1111.9641.6910.0808.764
Ficus sp. 414.1370.7005.45526.0410.281186.5261.0670.1262.10212.8700.0859.055
Ficus sp. 413.5910.7296.24925.9530.281174.9850.9950.1231.98612.8770.0848.545
Ficus sp. 516.4130.23610.74433.8290.203144.3451.0910.1111.65618.3530.06914.891
Ficus sp. 515.2320.24410.35535.9500.204145.6041.1040.1101.59813.5190.06815.402
Ficus sp. 610.8970.81917.22250.5580.25654.7471.4110.1491.2859.0750.10224.332
Ficus sp. 611.4100.74515.53149.2130.26450.0101.4840.1731.4204.6820.09622.255 ’
Ficus sp. 611.7650.68815.23744.5910.25758.8881.1280.1231.3916.2990.09021.656
Piper augustum8.9010.12020.14721.0910.262127.5581.7190.1932.19025.5310.13814.683
Piper augustum8.7890.10822.26314.9920.20690.4911.5280.1662.11314.2830.10915.112
Piper augustum8.9450.09722.09313.4620.170127.6511.4250.1552.09514.7290.10915.122
Piper augustum7.3530.13717.46323.4330.177106.5671.8040.1961.4248.9150.11211.095
Piper sp. 19.0360.15717.15342.8310.440644.9681.6670.1981.69712.7800.14211.532
Piper sp. 19.0990.16016.04638.7220.420616.1661.5730.1791.7206.3710.13311.388
Piper sp. 226.2080.630. 18.96238.2140.20264.5712.1870.2121.44055.2890.14127.495
Piper sp. 213.3290.56814.07154.2250.230208.5891.8900.1991.10036.7110.14232.397
Piper sp. 321.5240.39319.48738.1900.307918.7522.7590.2371.54439.4320.18951.857
Piper sp. 414.5630.37228.57083.6150.248639.5231.6640.1941.38723.138-0.13734.101
Piper sp. 59.3700.25110.23320.3070.21029.4351.3680.2781.4896.7690.15512.871
Piper sp. 58.7160.24310.38820.6750.19721.8541.5020.2441.4503.5960.15212.525

Appendix II

View this table:

Number of captures of bat species of the subfamilies Carolliinae and Stenodermatinae (Phyllostomidae) at lick, forest, and gap sites in the Peruvian Amazon. Number of captures in bold indicates bat species overrepresented at the site. An asterisk (*) indicates P < 0.001 from comparisons among licks versus forest versus gap sites. Bat nomenclature follows Gardner (2008).

Subfamily and speciesLickForestGapG-value
Carollia brevicauda3182924.6*
Carollia perspicillata7402625.76*
Rhinophylla pumilio241033.51*
Artibeus lituratus2082621251.95*
Artibeus obscuras2104018237.04*
Artibeus planirostris3183611470.49*
Chiroderma salvini54118.65*
Chiroderma trinitatum1462304*
Chiroderma villosum641132.49*
Platyrrhinus brachycephalus723139.6*
Platyrrhinus helleri23841480.19*
Platyrrhinus infuscus584388.84*
Sphaeronycteris toxophyllum1839.55*
Sturnira lilium295118.65*
Uroderma bilobatum26523536.72*
Uroderma magnirostrum8982142.71*
Vampyriscus bidens891186.8*
Vampyriscus pusilla2724.6*
Vampyrodes caraccioli2146.14*

Appendix III

View this table:

Loading values for the first 2 principal components (PCI and PC2) from the principal component analysis of the mineral content of Piper and Ficus fruits.


Appendix IV

View this table:

Diet composition of bats (Phyllostomidae) captured at licks in the southeastern Peruvian Amazon. Numbers of fecal samples containing each constituent are presented. Abbreviations are as follows: Araceae (Ara.), Clusiaceae (Clu.), and undetermined (Und.). Bat nomenclature follows Gardner (2008).

Subfamily and speciesCecropiaFicusPiperAra./Clu.PulpSoilInsectsUnd.
Carollia brevicauda2
Carollia perspicillata31
Artibeus lituratus92
Artibeus obscurus4121
Artibeus planirostris11212
Chiroderma salvini31
Chiroderma trinitatum22
Chiroderma villosum311
Platyrrhinus brachycephalus41
Platyrrhinus helleri111
Platyrrhinus infuscus842
Sturnira lilium1211
Uroderma bilobatum821
Vampyriscus bidens1
Vampyriscus pusilla1

Appendix V

View this table:

Diet composition of bats (Phyllostomidae) captured in forest sites in the southeastern Peruvian Amazon. Numbers of fecal samples containing each constituent are presented. Abbreviations are as follows: Araceae (Ara.), Clusiaceae (Clu.), Cucurbitaceae (Cue), Solanaceae (Sol.), and undetermined (Und.). Bat nomenclature follows Gardner (2008).

Subfamily and speciesCecropiaFicusPiperAraVCluVCuc/Sol.PulpSoilInsectsUnd.
Phylloderma stenops1
Phyllostomus elongatus23
Phyllostomus hastatus4
Carollia brevicauda3611
Carollia perspicillata181331
Rhinophylla pumilio1621
Artibeus obscuras121
Artibeus planirostris21
Chiroderma trinitafum1
Mesophylla macconnelli1
Platyrrhinus infuscus1
Sturnira lilium2

Appendix VI

View this table:

Diet composition of bats captured in gaps in the southeastern Peruvian Amazon. Numbers of fecal samples containing each constituent are presented. Abbreviations are as follows: Araceae (Ara.), Clusiaceae (Clu.), Solanaceae (Sol.), and undetermined (Und.). Bat nomenclature follows Gardner (2008).

Family, subfamily, and speciesCecropiaFicusPiperAra./Clu./Sol.PulpSoilInsectsUnd.
Phyllostomus elongatus1111
Phyllostomus hastatus2
Carollia benkeithi61112
Carollia brevicauda2121312
Carollia perspicillata319344
Rhinophylla pumilio2212
Artibeus obscurus1
Mesophylla macconnelli1
Sturnira lilium1
Thyroptera tricolor1


  • Associate Editor was Ricardo A. Ojeda.

Literature Cited

View Abstract