INTRODUCTION

Anthropogenic impacts are profoundly altering biological community composition (^{Newbold et al. 2015}). Generally, bat diversity declines with habitat degradation, resulting in loss of ecosystem function (^{Brosset et al. 1996}; ^{Brito De Oliveira et al. 2020}). However, the response of bats to anthropogenic disturbance varies according to context and to species (^{Herrera et al. 2018}). In the context of small-scale deforestation, in the Amazonian rainforests, some bat species had the highest abundance in mature forest while others were more abundant in agricultural areas or secondary forest (^{Willig et al. 2007}). Large-scale deforestation in French Guiana was associated with the loss of 48 mature-forest specialist species while the abundance of frugivorous and insectivorous bats increased (^{Brosset et al. 1996}). However, Phyllostomine bats in Mexico were more sensitive to habitat degradation than other bat clades because they were less abundant or absent in disturbed habitats (^{Fenton et al. 1992}). Vespertilionid bats in Peru were negatively affected by habitat disturbance (^{Wilson et al. 1996}).

In northern Central America, particularly in Honduras, the composition of bat assemblages in either undisturbed or disturbed ecosystems is still understudied (^{Turcios-Casco et al. 2020a}). The Honduran highlands are characterized mainly by cloud, pine, and pine-oak forests, grasslands, agricultural ecosystems, and urban areas. There are few large undisturbed forested areas remaining. While the number of bat studies has increased recently in Honduras, most recent publications have focused on first country records or range expansion (e. g., ^{Mora et al. 2020}). An exception is a study from subtropical moist forest in southern-central Honduras where 29 species of four families were found; from most to least abundant these families were: Phyllostomidae, Mormoopidae, Vespertilionidae, and Molossidae (^{Turcios-Casco et al. 2020b}). In this assemblage, the most abundant species were Artibeus inopinatus, A. jamaicencis, and Carollia perspicillata (^{Turcios-Casco et al. 2020b}). However, to our knowledge, no similar effort has been carried out in the highlands. This stresses the importance of more extensive studies of bat assemblages and breeding biology from northern Central America, especially from less studied areas, such as the Honduras highlands. Furthermore, the diversity and breeding phenology of bat assemblages at high elevations, defined by the sharp differences in rainfall between the rainy and dry season, may be affected by anthropogenic impacts due to urban expansion and deforestation for agricultural purposes (^{Fenton et al. 1992}; ^{Racey & Entwistle 2000}).

Reproductive patterns of bats have not been well-studied in most of northern Central America. Compared to bat assemblages of the Nearctic, where all species are insectivorous and seasonally monoestrous (^{Kunz 1982}), timing their reproduction to coincide with highest prey availability during the warmer months (^{Racey & Entwistle 2000}), bat assemblages of the Neotropics are ecologically diverse with insectivorous, frugivorous, nectarivorous, and sanguivorous species sharing biomes (^{McNab 1971}). This trophic diversity has led to a wider range of reproductive patterns in tropical species (^{Racey & Entwistle 2000}), from seasonal monoestry (i.e., a single litter during one climatic season) and seasonal polyestry (i.e., two litters per year tied to climatic seasons) to aseasonal polyestry (i.e., two or more litters per year irrespective of climatic season) (^{Jerrett 1979}), and variations thereof (^{Happold & Happold 1990}; ^{Racey & Entwistle 2000}). While monthly temperature is relatively more constant in the Neotropics compared to temperate zones, pronounced seasonal differences in rainfall may influence availability of food resources in areas characterized by distinct dry and wet seasons like the Pacific slope of Central America (^{Heithaus et al. 1975}; ^{Stoner 2001}).

Seasonal fluctuations in resource availability may drive food resource partitioning and explain reproductive timing among bat species with similar diets (^{Schoener 1974}). Most species time the most energetically costly periods of the reproductive cycle, such as gestation and lactation, to coincide with peak resource availability, which may occur at different periods in the year for each trophic group (^{Kurta et al. 1989}; ^{Racey & Entwistle 2000}). In those parts of Central America that are characterized by marked seasonal differences in rainfall, i.e., most of the Pacific Slope from Guatemala to northwest Costa Rica (^{Hastenrath 1966}), most flowering occurs during the dry season, when nectarivorous species may be reproductively active, while fruiting often occurs during the wet season, when frugivorous species may be reproductively active (^{Janzen 1967}; ^{Racey & Entwistle 2000}). However, some wide-ranging species exhibit plasticity in their reproductive patterns, showing bimodal polyestry in some parts of their range and aseasonal polyestry in others, based on local differences in resource availability and interspecific competition for these resources (^{Racey & Entwistle 2000}). In northern Central America, reproductive phenology of bat assemblages remains relatively understudied, compared to Mexico or Costa Rica (^{Ramírez-Pulido et al. 1993}; ^{Galindo-Galindo et al. 2000}; ^{Estrada & Coates-Estrada 2001}; ^{Tschapka 2005}; ^{Chaverri & Kunz 2006}; ^{Durant et al. 2013}).

Here we analyze the composition of a bat assemblage in a fragmented landscape in the Honduran highlands. Furthermore, we analyze sex ratios and document the reproductive phenology for the species of this bat assemblage. In addition, we were specifically interested in establishing the occurrence of Leptonycteris yerbabuenae, a species of conservation concern that was known only from three historical records in Honduras (^{Solari 2018}; ^{Turcios-Casco et al. 2020a}). While this species had never been reported from this part of Honduras, we observed that one of its food sources—Agave seemanniana—was abundant in our study area (^{Cole & Wilson 2006}; ^{Turcios-Casco et al. 2020a}, ²⁰²¹). The following hypotheses guide our study. We expected similar bat diversity to that documented in other studies with a comparable effort from Neotropical montane forests (e.g., ^{Briones-Salas et al. 2019}). Considering that mammals typically produce roughly equal numbers of sons and daughters (^{Rosenfeld & Roberts 2004}), we expected sexual equity in this bat assemblage. We expected that nectarivores would reproduce earlier than frugivores, specifically that nectarivores would be reproductively active during the dry season while frugivores would be reproductively active during the wet season (^{Racey & Entwistle 2000}). Since food resources for sanguivores are not governed by rainfall, we expected they would be lactating a-seasonally, i.e., throughout the year (^{Racey & Entwistle 2000}).

MATERIALS AND METHODS

Study area

We conducted this research in a primarily agricultural area on the Pacific Slope of south-central Honduras, along an elevational gradient between 1000–1700 m a.s.l., with small- scale cattle farming and subsistence crop farming, mixed with pine-oak forest, small patches of tropical deciduous dry forest on the south-facing slopes, and remnant patches of cloud forest at the highest altitudes. Local karst outcroppings provide natural refuges in the form of relatively small local caves, with a larger cave system located approximately 20 km to the west of the study area (^{Finch 1981}). We mist-netted in an area of approximately 52 km², in the municipalities of Santa Ana and San Buenaventura, about 10 to 15 km south of Tegucigalpa, in the department of Francisco Morazán, Honduras (Fig. 1). The study area included the slopes of Cerro de Hula (summit 1711 m a.s.l.) and Montaña de Izopo (summit 1860 m a.s.l.), and the intervening valley between these peaks. The local dry season ends in April, and the local wet season ends in November, with highest rainfall in the months of May, June, and September (ClimateSERV: https://climateserv.servirglobal.net).

Sampling protocol and data analyses

Between March 2014 and January 2015, we mist-netted on a weekly basis (weather permitting) at 34 sites throughout the study area, for 43 nights. Nine sites were visited twice during the study period because poor weather conditions during the first visit caused us to close the nets prematurely. Mist-netting sites were selected based on their potential as foraging sites or as corridors between foraging and roosting sites. During the period of local flowering of Agave seemanniana, an important food resource for Leptonycteris yerbabuenae (^{Cole & Wilson 2006}), i.e., roughly between May and September, we mist-netted near flowering agaves on seven nights, specifically targeting this species. Otherwise, site selection was standard for inventory purposes, i.e., potential foraging sites such as vegetation corridors, trails, forest edge, and other natural flyways. For each site and night, 8-10 nets were opened in an area of approximately 500 m². Mist-nets were placed near ground level and were opened at 6 pm and closed at midnight, weather permitting, and checked every 20 minutes (^{Hoffmann et al. 2010}). We used 36 mm mesh mist-nets of 12 x 2.6 m. The total sampling effort was obtained by multiplying the number of hours sampled by the number of nets used and the area in m² of each net. We sampled on average 4 nights per month; sometimes less due to inclement weather.

The bats captured were placed in individual cotton bags and taken to a field base located within each sample site, where they were identified. There, we measured forearm length and body mass of our captures and where necessary examined other morphological features to assist identification. Bats were identified using ^{Medellín et al. (1997)}, ^{Timm & LaVal (1998)}, and ^{Reid (2009)}. Additionally, we recorded sex, age class, and reproductive stage from our captures; in a handful of cases, individuals escaped before some or all of these data had been collected.

We sexed bats based on an examination of external sexual characters. We aged bats as juveniles, subadults, or adults by examining epiphyseal growth plate ossification in the metacarpus (^{Kunz & Anthony 1982}; ^{Brunet-Rossinni & Wilkinson 2009}). For adult females, we recorded reproductive status using the following categories: pregnant (i.e., presence of fetus felt by massaging lower abdomen); lactating (i.e., presence of milk in mammary glands); post-lactating (i.e., bare skin around mammary glands, flaccid and dark mammary glands); no evidence of sexual activity or “inactive” (i.e., abdomen normal, mammary glands invisible). We did not mark bats individually but for each individual noted the precise location of morphological features such as wing scratches, leg scratches, or broken dentures, to avoid collecting data from recaptures. Upon collecting all the information described above, bats were released at field bases, inside of the 500 m² area.

We evaluated sample sufficiency for Phyllostomidae by the observed and estimated species accumulation curve considering each night as a sampling unit. We estimated phyllostomid species richness by sample-based rarefaction using 200 randomizations performed with replacement. We followed the method of ^{Colwell et al. (2012)} and computed 95% confidence intervals based on the unconditional variance. We analyzed adult sex ratios with an exact binomial test of goodness-of-fit to test for sex skew for the 12 species where we captured more than 10 adult individuals. For this analysis, we considered two-tailed tests to be significant when P < 0.05. Relative abundance was calculated by dividing the number of captures for each species by the number of captures for all species. Our analysis was carried out in the R language and environment (version 4.0.2) using the package vegan (^{R Core Team. 2020}).

RESULTS

During 37890 m².h of mist-netting effort, we captured 594 bats representing 18 phyllostomid and four vespertilionid species (Table 1). Individuals we recognized as recaptures are not included in this result. For the Phyllostomidae, the species accumulation curve appears to approach an asymptote (Fig. 2). The estimated species richness using the sample-based rarefaction method indicates the occurrence of 19 phyllostomids for our study site, which suggests that 95% of the expected phyllostomids for this area were sampled. Three phyllostomids—Glossophaga soricina, Sturnira parvidens, and Chiroderma salvini—were found to be common in our study area, i.e., each represent more than 10% of all captures, and together make up half of our captures (Table 1). Four phyllostomids were uncommon (5-10% of all captures), while eleven phyllostomids and all four vespertilionids were rare at our study site, i.e., fewer than 5% of all captures (Table 1). Our efforts to target Leptonycteris yerbabuenae resulted in the capture of 26 individuals, which represent 4% of overall captures, divided over seven nights, with the highest number of captures—12 individuals—on 8 May 2014 (Table 1; Fig. 3A; Table S1). We also captured a male Choeronycteris mexicana on 19 June 2014, which represents 0.2% of overall captures (Table 1; Fig. 3B; Table S1).

We evaluated sex ratios for those species with more than 10 captures of sexed mature individuals, and found sexual equity for most species, except Desmodus rotundus (38% males; n = 40; 2-tailed binomial test: p = 0.04), Sturnira parvidens (32% males; n = 53; 2-tailed binomial test: p < 0.01), and Anoura geoffroyi (82% males; n = 11; 2-tailed binomial test: p = 0.03; Table 1).

We found physical evidence demonstrating reproductive activity in females for 16 out of 22 species captured in mist-nets (Table 2). Although we detected reproductive activity throughout the year, most species in this highland bat assemblage were reproductively active during or shortly after the wet season, i.e., May to November (Table 2; Fig. 4). We detected pregnancies in the sanguivorous Desmodus rotundus throughout the year (April, August, and November; Fig. 4). We detected pregnancies in nectarivorous phyllostomids at the end of the dry season (i.e., March), with a smaller peak in the wet season (i.e., July-September), and lactation early in the wet season (i.e., May-July), with a smaller peak early in the dry season (i.e., November) (Fig. 4). For Glossophaga commissarisi, we detected pregnancies only in August, i.e., in the middle of the wet season, and did not detect lactating females. For Glossophaga soricina, a common species among our captures, we detected pregnancies in March and September, i.e., at the end of the dry season and toward the end of the wet season; this bimodal pattern was also reflected in the lactation results, with a strong peak early in the wet season (May to July) and a weaker peak early in the dry season (November). For Leptonycteris yerbabuenae, the only reproductive signal we detected was a pregnancy in July, while for Choeroniscus godmani, two pregnancies were detected in August and one in November. We detected breeding activity in frugivorous phyllostomids throughout the year, with lactation peaking in June and December, i.e., during the wet season and into the early dry season (Fig. 4). Sturnira parvidens, a common species among our captures, showed a bimodal pattern with one pregnancy detected in March and lactation detected in May and June, i.e., early wet season, followed by a stronger peak in reproductive activity with pregnancies detected in August, September and November, and lactation between August and December, i.e., the second half of the wet season into the early dry season. A similar pattern was found for Chiroderma salvini, with pregnancies in April and August, and a lactation peak in May and June, followed by another lactation peak in November. For the three Artibeus encountered in our study site, we detected pregnancies only in Artibeus intermedius, one each in May and in November, with lactation detected in June. For Artibeus jamaicensis, we detected lactation in June, and then again between August and October, while for Artibeus lituratus, lactation was found only in August and September. The few insectivorous vespertilionids that we caught were reproductively active between April and July. For all species combined, we observed almost three times as many lactating bats than pregnant bats; this difference was especially marked in frugivores. These results suggest bimodal patterns for at least three species (Table 2).

Fig. 1 Location map of the study area. Indicated are land use and cover, and mist-netting sites. A: Cerro de Hula; B: Montaña de Izopo. Land cover data from Atlas Municipal Forestal y Cobertura de la Tierra (^{ICF 2015}). 

Juveniles (n = 68) made up 11% of our overall captures and were encountered among 14 species (Table 1). Almost half of all juveniles belonged to Glossophaga soricina, with young from May until October, with peaks in May (11) and September (7). We found three juveniles of Glossophaga commissarisi, all in August. We found juveniles of Desmodus rotundus in January, March, April, May, and September. For Chiroderma salvini, we found juveniles in March, October, and November. Juveniles of Sturnira parvidens were encountered from March to May and from August to November, while juveniles of Sturnira hondurensis were encountered in January, March, and May. Two juveniles of Choeroniscus godmani were encountered in August. Two juveniles of Artibeus toltecus were encountered in November. For the remaining 6 species, single juveniles were found, including a Leptonycteris yerbabuenae in September.

DISCUSSION

Despite the high level of habitat fragmentation and anthropogenic land use, we documented 22 species in our study area, i.e., a number considerably higher than that reported under similar effort from similar habitat in Oaxaca, Mexico, where nine species were encountered in montane forest fragments (^{Briones-Salas et al. 2019}). Another Mexican study, also from Oaxaca, found up to 17 species in primary pine-oak forest at a similar elevational range as our study site (^{Sánchez-Cordero 2001}). Our captures include nearly 20% of all Chiroptera reported from Honduras, and 64% of all Phyllostomidae reported from the department of Francisco Morazán (^{Turcios-Casco et al. 2020a}, ²⁰²¹). Two species had not yet been reported for Francisco Morazán: Myotis elegans and M. nigricans (^{Turcios-Casco et al. 2021}). We consider the relative abundance of Chiroderma salvini in this landscape, 10.6% of all captures, to be noteworthy, as this species is known to be scarce, with a relative abundance often less than 1%, in other parts of its range (^{Hice et al. 2004}; ^{Sánchez et al. 2007}; ^{Peña-Cuéllar et al. 2015}). In Honduras, from a subtropical moist forest located at ∼20 km from our study site, the relative abundance of Chiroderma salvini is 2% (^{Turcios-Casco et al. 2020b}). While we feel confident that we rejected virtually all recaptures from the same night based on recognition of morphological traits, it is possible that a small section of our captures were recaptures from sampling on previous nights in our study area. However, we do not believe this to have a major impact on our results because recapture rates in phyllostomid bats, under similar sample protocols, tend to be very low, i.e., 7-8% (^{Medellín et al. 2000}; ^{Bianconi et al. 2006}). Since on some nights we mist-netted close to flowering Agave seemanniana, our capture rates for Leptonycteris yerbabuenae are probably inflated and not representative of its relative abundance in this assemblage. This is further attested by the fact that we did not capture this species away from the immediate vicinity of agaves.

Table 1 Composition of understory bat assemblage from Cerro de Hula / Montaña de Izopo in the central Honduran highlands, department of Francisco Morazán, March 2014-January 2015. N = total individuals captured; RA = relative abundance. 

Table 2 Reproductive status of adult female bats captured in Cerro de Hula / Montaña de Izopo in the central Honduran highlands, department of Francisco Morazán, March 2014-January 2015. 

Nevertheless, our capture of 26 individuals, including 12 individuals on a single night, indicates that this species may be more common locally than previously thought (^{Turcios-Casco et al. 2020a}). Until recently, only three Honduran records were known, but a study documenting 18 individuals from two locations, including eight individuals caught during a single night, also suggest that this species is perhaps more common than was previously thought (^{Mora et al. 2020}). Since we only captured one individual that was reproductively active, more data is needed to confirm reproductive patterns of this species in Honduras. Our capture of a single male Choeronycteris mexicana (Fig. 2B; Table S1), also considered rare in Honduras, is noteworthy (^{Turcios-Casco et al. 2020a}). We know of only three other Honduran records (^{Davis et al. 1964}; ^{Ueda 2020}; ^{Turcios-Casco et al. 2021}). Leptonycteris yerbabuenae and Choeronycteris mexicana are species of conservation concern in Honduras (^{Hernández 2015}) and are listed as Near Threatened by the IUCN (^{Medellín 2016}; ^{Solari 2018}).

We provide information on the reproductive biology of 16 species, including bimodal polyestry for nectarivores such as Glossophaga soricina and frugivores such as Sturnira parvidens, Chiroderma salvini and Artibeus intermedius. Similar reproductive patterns have been found elsewhere for Glossophaga soricina in Costa Rica, Panama, and Mexico (^{Fleming et al. 1972}; ^{Ramírez-Pulido et al. 1993}; ^{Hernández-Aguilar & Santos-Moreno 2020}). ^{Calahorra-Oliart et al. (2021)} found evidence to split Glossophaga soricina into four species and propose to assign populations from Mexico and northern Central America to Glossophaga mutica, while noting that populations from southern Central America do not form part of this lineage and may comprise another cryptic species complex. Since this work is very recent, we here maintain the name Glossophaga soricina as used in ^{Turcios-Casco et al. (2020a)} and ^{Turcios-Casco et al. (2021)}. Some authors report year-round breeding for Sturnira parvidens (e.g., ^{Solari 2019}), but the pattern described from primary sources in Costa Rica, Panama, and Mexico, is clearly bimodal (^{Fleming et al. 1972}; ^{Sánchez-Hernández et al. 1986}). Bimodal polyestry has been suggested for Chiroderma salvini in Central American (^{Garbino et al. 2020}). In Mexico, the reproductive pattern of Artibeus intermedius is also bimodal polyestry (^{Montiel et al. 2011}). The sanguivorous Desmodus rotundus appeared to be reproductively active throughout much of the year. We detected pregnancies throughout the year, although lactation was only observed from April to June, i.e., at the beginning of the wet season. These patterns coincide with patterns found elsewhere in the ranges of this species (^{Wimsatt & Trapido 1952}; ^{Bonaccorso 1979}; ^{Delpietro et al. 2017}; ^{Aguirre 2019}).

Fig. 2 Accumulation curves for observed and estimated phyllostomid species richness in the Honduran highlands. Error bars are standard deviation around the estimated mean. 

Fig. 3 A = A male Leptonycteris yerbabuenae caught on 7 May 2014 in Santa Ana; B = A male Choeronycteris mexicana caught on 19 June 2014 on Montaña de Izopo, Santa Ana. 

Fig. 4 Female reproductive phenology of Phyllostomidae. Proportions are based on the total number of pregnant or lactating females per total number of adult females in each month between March 2014 and January 2015. The shaded area indicates the local wet season. 

For most species, our detected pregnancies signal was too weak to deduce patterns, attesting to the fact that, especially in the smaller species, early pregnancies can be difficult to detect in the field by palpation of the fetus in the ventral region (^{Kleiman & Davis 1979}; ^{Simmons & Voss 2009}; ^{Vela-Vargas et al. 2016}). For example, ^{Vela-Vargas et al. (2016)} found that 95% of females deemed reproductively inactive in the field based on external traits were found to be reproductively active based on vaginal smears. Nonetheless, our lactation results showed a general pattern: earlier breeding in nectarivorous phyllostomids at the end of the dry season and into the early wet season, while the frugivorous phyllostomids, a more diverse group consisting of more species, were reproductively active throughout the year. This result is expected and conforms to similar patterns found elsewhere in the Neotropics, especially in areas characterized by seasonal differences in rainfall (^{Graham 1987}; ^{Mena & Williams 2002}; ^{Wilson & Mittermeier 2019}). Bats from areas that are characterized by seasonal changes in rainfall are reported to experience interannual differences in reproductive peaks, following interannual climatic differences (^{Stoner 2001}). Thus, the specific patterns we reported here, based on a single annual cycle, may not be representative of every year.

Among nectarivorous Glossophaginae, we consider noteworthy an observation of a July pregnancy in Leptonycteris yerbabuenae, a species poorly known in Honduras (^{Medellín 2019}; ^{Turcios-Casco et al. 2020a}, ²⁰²¹). This species includes migratory populations that breed in July in northern Mexico and southwestern United States, as well as sedentary populations in the dry forests of southern Mexico that breed during January and February (^{Medellín 2019}). Nothing was known of its reproduction in Central America. The temporal pattern of our captures of Leptonycteris yerbabuenae, i.e., between May and September, coincides with what was reported for the same habitat in Honduras, i.e., March to August (^{Mora et al. 2020}), but is later than what was found in Pacific dry forest in Guatemala, i.e., March to June (^{López-Gutiérrez et al. 2015}). Since we captured this species only when mist-netting close to a preferred trophic resource, which bloomed in the study site between May and September, we do not know if the species remains present in the study area the rest of the year. Little is known about movements of Central American populations, which are thought to be sedentary and undertaking only altitudinal migrations (^{Mora et al. 2020}). Further studies of possible seasonal movements by Leptonycteris yerbabuenae in Honduras are justified.

We expected to find sexual equity among our captures, and indeed did so for all but three species: Desmodus rotundus, Anoura geoffroyi, and Sturnira parvidens. Among bats, male-biased sex ratios have been observed in Desmodus rotundus in Argentina (^{Delpietro et al. 2017}), while female-biased sex ratios have been reported for Anoura geoffroyi in Brazil (^{Baumgarten & Vieira 1994}) and for Sturnira parvidens in Mexico (^{Estrada & Coates-Estrada 2001}; ^{Iñiguez-Dávalos 2005}). Our study highlights the need to test sex-specific adaptive responses in these three species across multiple seasons, and along gradients of elevation, and agricultural ecosystems. Sexes may respond differently to habitat modifications, as ^{Rocha et al. (2017)} found for two frugivorous species in a human-altered landscape.

CONCLUSION

While our study area is not recognized as an important area for bat conservation in Honduras, its high phyllostomid diversity compared to similar habitats elsewhere, including two species of conservation concern, suggests that Honduran high- lands, even with fragmented landscapes, may represent significant biodiversity value. We were able to collect information on reproductive events of 16 bat species from a poorly studied area and habitat, the Honduran highlands, and for some regionally poorly known species like Chiroderma salvini and Leptonycteris yerbabuenae, despite sampling for only eleven months. We found that nectarivorous species breed mainly earlier in the year, at the end of the dry season, while frugivorous species tie the most demanding period of their breeding, i.e., lactation, to the wet season. Except for three species, Desmodus rotundus, Sturnira parvidens, and Anoura geoffroyi, most species of this bat assemblage show sexual equity. Finally, our Leptonycteris yerbabuenae results suggest that this species may be more common in Honduras, and present for a longer part of its life cycle, than was previously thought. Studies to establish the abundance, movement, reproduction, and feeding habits of Leptonycteris yerbabuenae in Honduras are needed to contribute to its conservation efforts.

Supplementary materials