INTRODUCTION
In the Neotropics, fruit-eating bats are important for forest regeneration in areas that have been cleared of trees (Cubiña & Aide Reference CUBIÑA and AIDE2001, Gorchov et al. Reference GORCHOV, CORNEJO, ASCORRA and Jaramillo1993, Medellín & Gaona Reference MEDELLÍN and GAONA1999, Muscarella & Fleming Reference MUSCARELLA and FLEMING2007), because they feed on a variety of fruits from pioneer plants (Gorchov et al. Reference GORCHOV, CORNEJO, ASCORRA and Jaramillo1993, Lopez & Vaughan Reference LOPEZ and VAUGHAN2004) and cover long distances when foraging (Heithaus & Fleming Reference HEITHAUS and FLEMING1978). Yet, natural forest recovery on abandoned pastures is often slow (Aide et al. Reference AIDE, ZIMMERMANN, HERRERA, ROSARIO and SERRANO1995, Uhl et al. Reference UHL, BUSCHBACHER and SERRAO1988) and limited to short distances of 20 m or less from the forest edge (Cubiña & Aide Reference CUBIÑA and AIDE2001, Wildeven & Kuzee Reference WILDEVEN and KUZEE2000), indicating that seed availability or inaccurate directionality of seed transfer may limit the rate of succession at deforested sites. Seed rain dynamics may become particularly complex once succession has advanced, as fruit production of pioneer plants can be substantial and bats feeding on fruits may return frequently to older forests in search of roosts in large hollow trees (Rothenwoerer et al. Reference ROTHENWOERER, BECKER and TSCHAPKA2011). Therefore, we here studied the directionality of seed transfer mediated by fruit-eating bats between a primary forest and an adjacent disturbed area that has been undergoing succession for more than 15 y (Hartshorn & Hammel Reference HARTSHORN, HAMMEL, McDade, Bawa, Hespenheide and Hartshorn1994). In particular, we asked whether fruit-eating bats are more likely to transfer seeds from the mid-successional deforested area into the adjacent forest than in the opposite direction. We investigated the spatial origin of fruits consumed by short-tailed fruit bats by using an isotopic approach. This approach is based on the observation that stable carbon isotope ratios – given as δ13C in the δ-notation and expressed as the per mille (‰) difference between the sample and an international standard (Vienna PDB for stable carbon isotopes) – differ in leaves of C3 plants between deforested and forested areas (Medina & Minchin Reference MEDINA and MINCHIN1980). For example, plant δ13C differed by more than 5‰ between forest sites (−34.5‰) and open areas at 50 m from the forest edge (−28.4‰) in an Amazonian edge habitat (van der Merwe & Medina Reference VAN DER MERWE and MEDINA1989). Thus, seeds of C3 plants may also differ in isotopic composition according to whether they originate from inside or outside dense forests. To elucidate whether the micro-climatic situation of forest gaps (more solar radiation and higher ambient temperatures) could cause elevated δ13C values in individuals of the genus Piper (similar to those growing on successional sites), we tested for differences in plant δ13C collected from three sites: forest understorey, forest gaps and successional areas.
We first validated whether plant organs (leaves, fruit pulp and seeds) are similar in isotopic composition in plants of the genus Piper. Then, we tested whether individuals (Piper spp.) growing in the forest interior are isotopically similar, even when growing under different microclimatic conditions, such as in the understorey and in forest gaps. We then compared δ13C of Piper individuals between forest (pooled data of understorey and gap sites) and successional sites. Furthermore, we tested if δ13C of defecated seeds (δ13Cseeds) is altered by the digestive process in a feeding experiment with C. sowelli. Then, we used data on δ13Cseeds collected from free-ranging bats to estimate the spatial origin of fruits consumed by bats. We tested the hypothesis that short-tailed fruit bats (Carollia spp.) are more likely to carry Piper seeds from deforested areas of mid-successional stage into forests than the opposite direction, when bats forage outside but roost inside the forests (Kelm et al. Reference KELM, WIESNER and VON HELVERSEN2008). We predicted seeds defecated by free-ranging short-tailed fruit bats (Carollia spp.) to be more enriched in 13C than plant matter of forest Piper, if bats consume infructescences of Piper individuals predominantly from outside the forest.
MATERIALS AND METHODS
Study site
This study was carried in the forest and adjacent deforested areas of mid-successional stage around La Selva Biological Station (Costa Rica, 10°25′, 84°00′W) in March and November 2010. Habitats and vegetation are described in detail in Hartshorn & Hammel (Reference HARTSHORN, HAMMEL, McDade, Bawa, Hespenheide and Hartshorn1994). Briefly, disturbed sites of our study area are recovering from deforestation for more than 15 y, and thus carry large numbers of pioneer plants (Hartshorn & Hammel Reference HARTSHORN, HAMMEL, McDade, Bawa, Hespenheide and Hartshorn1994). We expected plants from the forest interior to differ isotopically from those of successional sites, because in dense forests, leaves of understorey plants are more enriched in 12C than in 13C due to the so-called ‘canopy effect’ (Ehleringer et al. Reference EHLERINGER, KIN, FIELD, SUN and KUO1987, Medina & Minchin Reference MEDINA and MINCHIN1980, Rex et al. Reference REX, MICHENER, KUNZ and VOIGT2011, Voigt Reference VOIGT2010). The ‘canopy effect’ is caused by photosynthetic recycling of 13C-depleted CO2 produced by soil respiration (van der Merwe & Medina Reference VAN DER MERWE and MEDINA1989, Reference VAN DER MERWE and MEDINA1991; Vogel Reference VOGEL1978) and by low light intensity at ground level affecting photosynthesis (Farquhar et al. Reference FARQUHAR, O'LEARY and BERRY1982). Since the ‘canopy effect’ is absent outside the forest, leaves of C3 plants growing at successional areas are more enriched in 13C than leaves of C3 plant growing under a dense canopy (Medina & Minchin Reference MEDINA and MINCHIN1980, van der Merwe & Medina Reference VAN DER MERWE and MEDINA1989, Reference VAN DER MERWE and MEDINA1991).
Study species
Among the several hundred species of fruit-eating bats worldwide, bats of the genus Carollia play presumably a paramount role for forest regeneration and succession in the Neotropics (Muscarella & Fleming Reference MUSCARELLA and FLEMING2007). Species of this genus are highly mobile (Heithaus & Fleming Reference HEITHAUS and FLEMING1978) and occur at high densities in disturbed habitats and primary forests where they forage in the understorey (Estrada et al. Reference ESTRADA, COATES-ESTRADA and MERITT1993, Faria Reference FARIA2006, Rex et al. Reference REX, CZACZKES, MICHENER, KUNZ and VOIGT2010, Reference REX, MICHENER, KUNZ and VOIGT2011). Short-tailed fruit bats (Carollia spp.) feed mostly on infructescences of pioneer plants of the Piperaceae (Fleming Reference FLEMING1988, Lopez & Vaughan Reference LOPEZ and VAUGHAN2004). Thus, short-tailed fruit bats may be particularly efficient in transferring seeds from forests into recently cleared sites (Kelm et al. Reference KELM, WIESNER and VON HELVERSEN2008). In our study, we focused on three co-existing short-tailed fruit bats: C. castanea (Allen 1890), C. sowelli (Baker, Solari & Hoffmann, 2002) and C. perspicillata (Linné 1758).
Intra-individual variation of stable carbon isotope ratios in individuals of the genus Piper
To facilitate sample collection, we first tested whether δ13C of leaves, fruit pulp and seeds differed within the same Piper individuals. If δ13C of leaves is similar to that of fruit pulp and seeds, we decided to collect leaf material instead of infructescences for reason of simplicity. Overall, we collected ripe infructescences and leaves from 10 shrubs representing three species: P. generalense (four individuals), P. sancti-felicis (four), P. urostachyum (two). Plant species were identified with help of a key (http://sura.ots.ac.cr/local/florula3/fr_species.php). All plant samples were stored in plastic containers and dried until constant mass at 40°C.
Samples were analysed at the Leibniz Institute for Zoo and Wildlife Research as described in Voigt et al. (Reference VOIGT, REX, MICHENER and SPEAKMAN2008, Reference VOIGT, SÖRGEL and DECHMANN2010). All 13C/12C were expressed relative to the international standard (V-PDB) using the δ notation (‰) according to the following equation δ13C = [(Rsample/Rstandard) – 1] × 1000 where R is the ratio 13C/12C. The analytical precision was better than 0.08‰ (1 standard deviation; SD) based on repeated measures of internal laboratory standards. We tested for isotopic differences among leaves, fruit pulp and seeds by using a repeated-measures analysis of variance (ANOVA). All statistical tests were performed two-tailed assuming an alpha value of 5%. We presented values as mean ± 1 SD if not mentioned otherwise.
Inter-individual variation of stable carbon isotope ratios in individuals of the genus Piper
We collected leaves from Piper individuals opportunistically from mature forest (understorey and gaps) and adjacent deforested areas at an average distance of 205 ± 94 m (median: 184 m; range: 57–340 m; n = 13) and 334 ± 225 m (184 m; 57–792 m; n = 22) from the forest edge respectively. Distances between collecting sites and forest edge did not differ between forest and succession sites (Mann–Whitney U-test: U = 97, U' = 189, P = 0.12). We categorized plants as understorey Piper, gap Piper and Piper of successional sites, because we expected the largest differences in ambient CO2 concentration and ambient temperature among these three habitats. Leaf material was prepared, stored and analysed as outlined above. We tested whether individuals of the genus Piper differed in δ13C among the habitats using a Kruskal–Wallis test. We then subsumed data of understorey and gap Piper as those of forest Piper. We tested whether leaves of forest Piper were more depleted in 13C than those of Piper individuals from successional sites using a Mann–Whitney U-test. We repeated the analysis with a Student's t-test after removing two outliers (highly enriched in 13C) from the dataset of the deforested site to assess the sensitivity of the analysis.
Feeding experiment
We captured 13 Carollia sowelli during 20 nights of mist-netting using two to three mist-nets (6–9 m length, 2.5 m height, Ecotone, Gdynia, Poland) set up at ground level between 19 and 22 h. Bats were identified according to Timm & LaVal (Reference TIMM and LAVAL1998) and transferred to flight cages (1 m3), where they were kept in groups of two to three for a maximum of 8 d at semi-ambient conditions. Bats were fed with a mixture of mashed bananas and honey water with ad libitum access to water. At 17 h, we arbitrarily selected two to four individuals from the captive population for experiments. Every experimental bat was kept singly in a small, clean linen bag until feeding started.
Prior to experimental nights, we collected several ripe infructescences from nearby Piper hispidum shrubs at 16 to18 h. From all infructescences, we transferred a small subsample in a plastic container for isotope analysis. Before analysis, samples were dried until constant mass at 40°C in a drying oven (Heraeus Function line 7000, Thermo Electron Corporation, Langenselbold, Germany). All animals used for feeding experiments were fasting for at least 12 h prior to the experiment. We fed each animal at least half and up to one complete infructescense. At ~90 min post-feeding, we collected faeces from the bags in which the bats stayed after feeding. We removed seeds from the faecal samples and dried them until constant mass at 40°C in a drying oven. All bats were released at the site of capture after the experiments. We tested whether δ13Cseeds differs before and after passage of the intestinal tract using a Wilcoxon matched-pairs test.
Stable isotope data of free-ranging short-tailed fruit bats (Carollia spp.)
Bats of the genus Carollia were captured as described above at four understorey sites: in the mature forest at a distance of at least 100 m from the forest edge, and at three sites in adjacent deforested area at a minimum distance of 100 m and a maximum distance of 300 m from the forest edge. We collected faecal samples only from bats of the genus Carollia and released all other bats immediately. For seed collection, bats were put singly into clean linen bags for at least 60 min. From these faecal samples we separated seeds into plastic vials filled with ethanol. After marking bats with a wing puncture (3 mm diameter) using a sterile biopsy punch, we released bats at the site of capture. The hole caused by wing puncturing ensured that we did not collect samples repeatedly from the same individuals. Seeds were dried, prepared and analysed as described above. We then used a two-way ANOVA to test whether δ13Cseeds differed among species and capture sites. Finally, we tested if defecated seeds are more enriched in 13C than plant matter of forest Piper by using a Mann–Whitney U-test.
RESULTS
Stable carbon isotope ratios of individuals of the genus Piper
In our intra-individual analysis, we found that mean δ13C of leaves (−33.5‰ ± 1.9‰), seeds (−33.1‰ ± 2.6‰) and fruit pulp (−33.5‰ ± 2.3‰) did not differ within the same plant (n = 10; F2 ,29 = 2.16, P = 0.144). Thus, we used δ13C of leaves (δ13Cleaves) as a predictor for the δ13C value of fruit pulp and seeds bats are consuming.
For the inter-individual comparison, we collected leaf material from 35 plants representing 12 Piper species. In the forest interior, we collected plant material from nine P. generalense, five P. sancti-felicis, two P. urostachym, two P. peracuminatum, one P. glabrescens, one P. paulownifolium, and one P. sp. 3. At the successional site, we collected plant material from eight P. sancti-felicis, two P. auritum, one P. colonense, one P. biolleyi, one P. sp. 1 and one P. sp. 2. δ13C of leaves differed among Piper growing in forest understorey, in forest gaps and at successional sites (Kruskal–Wallis test: KW = 13.4; P = 0.0012; Figure 1). Leaves from successional areas were more enriched in 13C (median δ13Cleaves = −31.8‰) than those from the forest understorey (−35.8‰ Dunn's multiple comparison test: mean rank difference = −15.4; P < 0.01) and those from forest gaps (−33.6‰; mean rank difference = −9.6; P < 0.05). Piper individuals from forest understorey and gaps did not differ in δ13Cleaves (mean rank difference = −5.8; P > 0.05).
Figure 1. Stable carbon isotope ratio (δ13CV-PDB; ‰) of leaves collected from Piper individuals in forest interior, forest gaps and in successional areas (a); significant pairwise differences are indicated by a horizontal line above the boxes). Stable carbon isotope ratio (δ13C; ‰) of defecated seeds collected in the forest interior and at the successional areas (b). Box margins indicate the 25 and 75 percentiles, whiskers the 5 and 95 percentiles, the solid lines within the boxes the median and the dashed lines within the boxes the mean.
Overall, Piper individuals growing in the forest interior were more depleted in 13C by 2.5‰ than those growing in deforested areas (Mann–Whitney U test: n1 = 14; n2 = 21; U = 45; U' = 276, P = 0.0006). The median of δ13Cleaves equalled −33.9‰ for forest sites (−34.3‰ ± 1.8‰) and −31.9‰ for successional sites (−31.8‰ ± 1.9‰). In an intraspecific comparison, we found that leaves of P. sancti-felicis growing in the forest was more depleted in 13C (34.0‰ ± 1.7‰) than conspecifics growing at the successional sites (32.2‰ ± 0.8; Mann-Whitney U-test: n1 = 8, n2 = 5, U = 5, U' = 35, P = 0.0295). Leaves from two individuals of the deforested site (both Piper auritum) were substantially more enriched in 13C than the other leaf material (δ13Cleaves = −27.7‰ and −27.9‰). Excluding these two outliers from analysis still yielded a significant difference in δ13Cleaves between sampling sites, i.e. δ13Cleaves from plants collected from forest sites were more depleted in 13C (−34.3‰ ± 1.8‰) than that of plants from the deforested area (−32.4‰ ± 0.9‰; Mann–Whitney U-test: n1 = 12; n2 = 21; U = 45; U' = 207, P = 0.0026).
Feeding experiment
In an experiment with 13 captive C. sowelli, we collected faeces 90 min after bats consumed infructescences of Piper hispidum (δ13C = −29.8‰ ± 0.8‰). δ13Cseeds of defecated seeds averaged −30.0‰ ± 1.1‰ which did not differ from the δ13C of seeds prior to consumption (t 12 = 0.68, P = 0.51).
Stable isotope data of free-ranging short-tailed fruit bats (Carollia spp.)
In total, we captured 21 Carollia castanae, 21 C. perspicillata, 16 C. sowelli in similar numbers per species in forest and deforested areas (G-test: Gadj = 1.37, P > 0.05; Table 1). We obtained about the same number of faecal material from each species in both habitats (G-test: Gadj = 3.7, P > 0.05; Table 1). δ13Cseeds did not vary among species and sites (species: F 2,34 = 13.2, P = 0.09; sites: F 1,34 = 0.0, P = 0.988; species × sites: F 2,34 = 0.91, P = 0.18). Thus, we pooled data for all Carollia species and capture sites (δ13Cseeds = −28.8‰ ± 2.2‰). Seeds defecated by short-tailed fruit bats were significantly more enriched in 13C than seeds from Piper individuals growing in the forest interior (Mann–Whitney U-test; n1 = 20; n2 = 40; U = 36, U' = 844, P < 0.0001).
Table 1. List of short-tailed fruit bat species captured at the forest and the successional site. Number of faecal samples collected from each species at each site.
DISCUSSION
We used stable carbon isotope ratios of defecated seeds (δ13Cseeds) collected from three species of Neotropical short-tailed fruit bats (Carollia spp.) to infer the origin of consumed Piper infructescenses in a Costa Rican edge habitat. We show that almost all infructescenses consumed by short-tailed fruit bats shortly before capture were not harvested from Piper individuals of the forest interior but of the successional site, even when bats were captured in the forest interior.
Our approach was based on several assumptions that required prior validation. First, we showed that δ13C did not deviate among leaves, fruit pulp and seeds within the Piper individuals. This facilitated sample collection, because we were able to use δ13Cleaves as a proxy for δ13Cseeds. Second, we demonstrated that Piper individuals differed in δ13C depending on the site where they grew. Individuals from forest sites were more depleted in 13C by more than 2‰ than those from deforested areas, even when individuals of the same species were compared between habitats. This pattern is best explained by the ‘canopy effect’ that leads to lower δ13C values in forest plants in relation to plants outside dense forests (van der Merwe & Medina Reference VAN DER MERWE and MEDINA1989). However, the detected isotopic contrast between the forest and the successional area was not as strong as, for example, in an Amazonian edge habitat (5‰ in van der Merwe & Medina Reference VAN DER MERWE and MEDINA1989). Possibly, this is due to the advanced successional stage of our deforested site, which already included medium-sized trees and layers of leaf litter. This could have increased the effect of soil respiration on ambient CO2 and thus lowered δ13C of plant matter (Farquhar et al. Reference FARQUHAR, O'LEARY and BERRY1982, van der Merwe & Medina Reference VAN DER MERWE and MEDINA1989, Reference VAN DER MERWE and MEDINA1991). However, we suspect that δ13C of plant matter is even higher in individuals from early succession sites. This could also explain why some defecated seeds were more enriched in 13C than plants of our successional sites. Third, our feeding experiment revealed that the isotopic composition was not altered by digestion, probably because the passage time through the bat's intestine is too rapid (~20–40 min in C. perspicillata: Laska Reference LASKA1990). Consequently, we could use δ13Cseeds to infer the spatial origin of infructescenses consumed by bats.
We found no statistical differences in δ13C of defecated seeds among the three Carollia species, indicating that they consumed infructescences from the same area. However, we cannot rule out the possibility that the isotopic differences among species were too small to be detectable with the given sample size. Indeed, previous studies have found differences in stable isotope ratios in tissues of Carollia species in our and other study areas (Rex et al. Reference REX, CZACZKES, MICHENER, KUNZ and VOIGT2010, Reference REX, MICHENER, KUNZ and VOIGT2011; York & Billings Reference YORK and BILLINGS2009), suggesting that short-tailed fruit bats have well-defined dietary niches. Despite these differences, species-specific data obtained from our study were similar, i.e. we found that seeds defecated by free-ranging short-tailed fruit bats were more enriched in 13C than plant matter from the forest site. This implies that all three Carollia species were more likely to carry seeds from successional areas into the forest than in the opposite direction.
Several alternative methods can be used to assess the origin of defecated seeds and the directionality of seed transfer (reviewed in Voigt et al. Reference VOIGT, KELM, BRADLEY, ORTMANN, KUNZ and PARSONS2009). Almost all of these methods are based on plant species identification; either by seed morphology, by identifying seedlings grown from defecated seeds, or by molecular genetic tools. The major disadvantage of these methods is that the spatial origin of plants cannot be revealed unambiguously when plant species grow in both forest and successional habitats (e.g. P. sancti-felicis in this study). Growing seedlings from defecated seeds may also bias results when some species fail to grow. The disadvantage of the isotopic approach is that it builds on natural contrasts or gradients in δ13C that may depend largely on local topography, forest type and climatic conditions. We therefore suggest that isotopic approaches in the study of seed transfer should always involve testing underlying assumptions, e.g. a description of isotopic gradients and contrasts. Finally, a combination of isotopic and conventional methods seems to be particularly suitable for elucidating seed transfer in tropical forests.
In summary, bats of the genus Carollia transferred seeds from disturbed areas of mid-successional stage into the forest. The major reason underlying this pattern might be that (1) short-tailed fruit bats are specialists on Piper which are more dominant in disturbed habitats than in dense forests, and that (2) Carollia bats lack daytime roosts or even feeding roosts in disturbed areas which have been cleared of large trees (Kelm et al. Reference KELM, WIESNER and VON HELVERSEN2008). The lack of suitable roosts may cause bats to commute between roosts in the forest and individuals of the genus Piper outside of the forest. In our study area, short-tailed fruit bats are probably responsible for a constant influx of Piper seeds and faecal nutrients from the successional area into the forest. Transferred seeds and nutrients could potentially alter the plant composition in the edge habitat either directly by seed rain or indirectly by nutrient enrichment of forest soil. It remains to be tested whether the direction of seed transfer mediated by short-tailed fruit bats could be altered or even reversed when roosts outside of the forest are available to bats as well. This could be achieved either by preserving fragments with old-growth forest or by installing artificial roosts (Kelm et al. Reference KELM, WIESNER and VON HELVERSEN2008).
ACKNOWLEDGEMENTS
We thank Karin Sörgel and Anja Luckner for stable isotope analyses at the IZW and Ivailo Borrisov, José Gonzales, Sabine Greiner, Daniel Lewanzik, Sarah Schmidt and Karin Schneeberger for help in the field. The experiments complied with the current laws of Costa Rica.
