Introduction

Reproduction is a critical phase for most living organisms (^{Zortéa, 2003}). Bats have life history strategies that are distinctive from that of mammals of similar size. Because small mammals are characterized by multiple reproductive events and, in most species, by a small litter size (one, two, and up to three offspring), where gestation and lactation are relatively prolonged (^{Jerrett, 1979}; ^{Racey, 1982}; ^{Racey & Entwistle, 2000}). All species of bats present care to their offspring after birth, and after weaning, females can move away from their young and feed beyond the vicinity of the refuge (^{Clark, Leslie, & Carter, 1993}; ^{Burland, 1998}).

The reproductive strategies of bats exhibit considerable complexity and variability (^{Wilson, 1979}; ^{Racey & Entwistle, 2000}). Given that reproduction, particularly lactation, is energetically costly, the availability of food, which is directly correlated with precipitation, seems to be an important factor that determines the onset of this stage (^{Thompson, 1992}). Therefore, in places where the food is abundant throughout the year, animals can start reproduction at any time, but in an environment dominated by seasonal changes in the climate, where the availability of food varies, certain periods of the year are more suitable for reproduction than others. Depending on the latitude, reproduction may be limited by the seasonality of the habitats. For example, in temperate zones, where the climate varies dramatically between winter and summer, hibernation plays an important role in the reproductive cycles of species (^{Racey, 1978}; ^{Racey, 1982}). In tropical zones, although variations in temperature may not be extreme, there is a greater or lesser degree of seasonality in rainfall patterns (^{Racey, 1982}).

In most species of tropical bats, both insectivorous and frugivorous, reproductive activity is associated with precipitation. It can act directly on the onset of reproductive activity or indirectly by its effect on the control of flowering, fruiting, and increase of populations of insects that bats feed on (^{Bonaccorso, 1979}; ^{Humphrey & Bonaccorso, 1979}; ^{Racey, 1982}; ^{Cumming & Bernard, 1997}; ^{Estrada & Coates-Estrada, 2001}a, ^{Estrada & Coates-Estrada,2001}b). Births usually occur at the onset of rains and lactation occurs during the peak of this season (^{Racey, 1982}). Bats have developed seasonal breeding patterns to ensure that both offspring and lactating mothers find favorable conditions for their survival (^{Racey, 1982}; ^{Altringham, 1996}). Thus, they show a variety of annual reproduction patterns, which vary as much over the period of the year in which gestation occurs as in the space between them, in this way they relate to changes in climate (^{Jerrett, 1979}). In this regard, ^{Racey and Entwistle (2000}) present ten different reproductive patterns based on the studies of ^{Jerrett (1979}) and ^{Happold and Happold (1990}). Three are monoestrous type, when there is one litter per year (restricted seasonal monoestry, extended seasonal monoestry and aseasonal monoestry). Seven other patterns are of polyestrous type, when two and three litters per year happen (seasonal bimodal polyoestry with and without postpartum oestrus, seasonal multimodal polyoestry with postpartum oestrus, continuous bimodal polyoestry with postpartum oestrus, continuous multimodal polyoestry with and without postpartum oestrus, and aseasonal polyoestry).

An important aspect in the study of bats is population dynamics since it allows evaluating future trends in their populations and therefore, knowing the conservation status and defining protection strategies (^{Lemos-Espinal, Rojas-González, Zúñiga-Vega, & Jaime, 2005}). Knowledge of the population structure and dynamics (age categories, sex ratio and abundance) of bats over time allows estimating the number of individuals that support the population (^{Lemos-Espinal et al., 2005}). In cave bats that spend more than half of their lives in shelters and develop their main activities there, such as reproduction and rest, environmental conditions (temperature, humidity, air flow and light intensity) appear to be important factors in the selection of the refuge, where the population size of bat species may increase over time (^{Kunz, 1982}; ^{Hill & Smith, 1984}).

This study describes the reproductive patterns and population dynamics of seven bat species, two frugivorous species (Artibeus toltecus and Carollia perspicillata), four insectivorous (Mormoops megalophylla, Natalus mexicanus, Pteronotus fulvus and P. mesoamericanus) and one nectarivorous (Glossophaga soricina) in Mexico. Both Pteronotus species have also been treated as subspecies elsewhere: P. davyi subsp. fulvus and P. parnellii subsp. mesoamericanus, respectively. Although the reproductive patterns of these species have been studied previously in countries of America such as Costa Rica, Panamá and México (^{Fleming, Hooper, & Wilson, 1972}; ^{Bateman & Vaughan, 1974}; ^{Dinerstein, 1986}; ^{Bonaccorso, Arends, Genoud, Canton, & Morton, 1992}; ^{Iñiguez-Dávalos, 1993}; ^{Ramírez-Pulido, Armella, & Castro-Campillo, 1993}; ^{Boada, Burneo, De Vries, & Tirira, 2003}; ^{García-García, Santos-Moreno, & Rodríguez-Alamilla, 2010} ; ^{Torres-Flores, López-Wilchis, & Soto-Castruita, 2012}). Ecological studies of population dynamics and reproduction are relatively scarce. There is a large information gap that must be filled to understand the variability and specialization of these species throughout their distribution (^{Balmori, 1999}). In order to contribute to knowledge in this regard, this study describes the population dynamics and reproductive patterns of the seven aforementioned species. The hypothesis proposed is that bat reproductive patterns may be related to climatic factors such as temperature and precipitation.

Material and methods

Study site: The study was conducted in three roosts located in the Costa region of the State of Oaxaca, in southeastern México. The first site corresponds to a mine tunnel (15º54’52” N & 96º24’59” W and 1 110 m.a.s.l.) located in the municipality of Pluma Hidalgo. The other two sites are caves in the municipality of Santa Maria Huatulco: El Apanguito (15º51’58” N & 96º21’13.2” W and 695 m.a.s.l.) and Cerro Huatulco (15º50’59” N & 96º21’04.3” W and 475 m.a.s.l.). The climate in the area is semi-hot subhumid with rainfall in the summer (Aw₁) (Oficina Estatal de Información para el Desarrollo Rural Sustentable [^{OEIDRUS], 2005}), and temperature in the roosts varies throughout the year. At La Mina, the minimum temperature is 20.1 and the maximum is 23.5 °C. At El Apanguito, temperature ranges between 21 and 25.7 °C and in Cerro Huatulco, it ranges from 22.3 to 26.1 °C. The dominant vegetation in the area is medium subperennial rainforest with coffee plantations in the understory (^{Trejo, 2004}; ^{OEIDRUS, 2005}).

Field work: Monthly sampling was conducted between July 2016 and June 2017 including a dry (October-April) and a rainy (May-September) season, with a sampling effort of two nights per site. Bats were captured using a harp trap (Bat Conservation and Management, Inc., Carlisle, PA, U.S.A.), 1.5 m wide by 2 m high, which was placed at the entrance of the roosts between 18:00 and 00:00 hrs. The trap was checked every 20 min to collect any individuals found in the trap bag. The captured bats were then placed in 15 × 20 cm cloth bags for later processing. The species of each captured individual was determined with the aid of identification keys by ^{Medellín, Arita, and Sánchez (2007}) and ^{Álvarez-Castañeda, Álvarez, and González-Ruiz (2015}) and using the taxonomic classification system proposed by ^{Simmons (2005}). Sex and age class (young or adult) were also determined. In young males, testes are small, light colored, and covered with fur, whereas adults have larger, darker, and mostly hairless testes. Males were classified as sexually inactive or active depending on whether they had inguinal or scrotal testes, respectively. In young females, the nipples are small, light colored, and hairy, whereas in adult females, the nipples are larger, darker, and lack hair (^{Anthony, 1988}). Females were classified as reproductively inactive if their nipples were covered with hair, as gestational when the embryo could be felt in their abdomen, and as lactating when there was milk in the nipples and these were hairless (^{Kunz, 1996}). Finally, to avoid counting them again, each captured individual was marked on the forearm with an aluminum ring (National Band and Tag Company, Newport, Kentucky, USA) identified with a unique serial number. Rings corresponding to category A (2.9 mm) were used for Pteronotus fulvus and Natalus mexicanus, andcategory B (4 mm) rings were used for Artibeus toltecus, Carollia perspicillata, Glossophaga soricina, P. mesoamericanus, and Mormoops megalophylla. After recording the data of interest and marking individuals, they were released at their capture site. In order to carry out this study, a scientific collection license was granted for teaching purposes in the field of wildlife (20/ks-0112/10/16).

Data analyses: The sampling effort was calculated according to the formula proposed by ^{Medellín (1993}), where the dimensions of the harp trap (2 m length, 1.5 m width) were multiplied by the 6 h it was open during 12 months of sampling (69 nights in total; 22, 24 and 23 nights in Mina, El Apanguito and Cerro Huatulco, respectively). The result was expressed as m² net×h.

The classification proposed by ^{Racey and Entwistle (2000}) was followed to determine the reproductive pattern of each species. Since the largest distance between shelters is 11 km, it was unlikely to find two different reproductive patterns in the same species. This analysis was performed for each species in each roost and there were no differences in the pattern obtained. In addition, the low abundance and records of active individuals in some species (for example, the 29 ind. captured from Mormoops megalophylla in the Cerro Huatulco cave) did not allow to identify a pattern by roost, so it was decided to combine the data of each roost per species and show a single pattern. Temperature and relative humidity data were obtained with a WM-350 WindMate® Multi-function Weather Meter (Speedtech Instruments, USA) during the field work. Monthly precipitation data were obtained from a meteorological station near the study area (Santa María Huatulco, code 20 333) provided by the National Meteorological Service (Gobierno de México, n.d.).

The significance of differences in the number of individuals captured per species and month was evaluated statistically with a non-parametric Kruskal-Wallis test. Multiple generalized linear models (GLMs) of binomial distribution (0= reproductively inactive, 1= reproductive state) were used to investigate the relationships between the reproductive activity of the seven bats species and six explanatory variables: monthly precipitation (MPR), relative humidity (RHU), roost (ROO), season (SEA), sex (SEX) and temperature (TEM). The link function used was of the logit type. This resulted in a set of 63 possible candidate models. GLMs were run first for all species together and then for each species separately, therefore, in total 504 models were run. The best model was selected according to the Akaike Information Criterion (AIC). Analyzes were carried out in R programming language through the R Commander interface (^{R Core Team, 2020}).

Results

With a sampling effort of 1 242 m² net×h, a total of 5 836 bats from 14 species distributed in five families were captured and marked. The most abundant species were Pteronotus fulvus and P. mesoamericanus, representing 41 and 32.3 % of the captures, respectively (Table 1). The monthly abundance of each species fluctuated significantly (H= 13, df= 11, P= 0.044) throughout the entire study period (Fig. 1A, Fig. 1B, Fig. 1C). The mormopids Mormoops megalophylla, Pteronotus fulvus, P. mesoamericanus together with Natalus mexicanus showed a restricted seasonal monoestrous pattern while the phyllostomids Artibeus toltecus, Carollia perspicillata and Glossophaga soricina showed seasonal bimodal polyoestry pattern. Monthly abundance and reproductive patterns represent data of the three roosts evaluated.

Fig. 1 Variation of relative abundance of seven bat species in three roosts found on the Costa Region of Oaxaca, Mexico. X-axis numbers indicate the monthly captures. 

In general, of the 63 GLMs evaluated for the seven species, the best supported model (SEA + MPR + RHU + ROO + SEX + TEM, ω = 0.956) indicated that all the variables contribute to explain the reproductive activity of the seven bat species (Table 2). Estimated β coefficients indicated that the reproductive activity of the species was positively associated with females (β= 2.310, P< 0.001) captured in the El Apanguito cave (β= 1.501, P < 0.001) in the dry season (β= 1.713, P < 0.001). The monthly precipitation was less than 100 mm (β= 3.569, P< 0.001), relative humidity between 80 - 89 % (β= 2.994, P < 0.001) and temperature greater than 25 ºC (β = 3.941, P < 0.001) (Table 3). The reproductive pattern observed in Mormoops megalophylla was restricted seasonal monoestry, with the presence of males with scrotal testes between November and February and gestation occurred between February and April. This species did not use the El Apanguito cave for lactation and the colony abandoned the cave in May (Fig. 2A). No juveniles were recorded, 66.6 % of adults were males, and 33.4 % were females. The best model (MPR + RHU + ROO + TEM, ω = 0.993) indicated that four of the six variables contributed to explain reproductive activity (Table 2). Also, this was positively associated with individuals captured in El Apanguito (β= 1.571, P< 0.001) when the temperature was higher than 25 ºC (β = 1.752, P < 0.001), humidity had values of 70 - 79 % (β= 1.136, P= 0.003) and precipitation was less than 100 mm (β= 1.369, P < 0.001).

Pteronotus fulvus, the most abundant species at Cerro Huatulco and the second most abundant at El Apanguito (758 and 1 619 individuals captured, respectively), presented a restricted seasonal monoestrous pattern. Males with scrotal testes were observed from November to February. Gestation was observed between February and April and at the beginning of the rainy season. No lactating females were recorded in Cerro Huatulco and, due to the low abundance of lactating females in El Apanguito (two in May and two in June), it is most likely that lactation occurs in another unidentified roost (Fig. 2B). The best model (SEA + RHU + TEM, ω= 0.961) indicated that three of the six variables contributed to explain the reproductive activity (Table 2). Estimated β coefficients indicated that reproductive activity was positively associated with individuals caught in the dry season (β= 1.571, P < 0.001), when humidity was 70 - 79 % (β= 1.732, P= 0.005) and the temperature higher than 25 ºC (β= 1.935, P < 0.001) (Table 3).

Fig. 2 Reproductive patterns of seven bat species, Costa Region of Oaxaca, Mexico. Patterns represent the data of three roosts combined. X-axis numbers indicate the ratio of adult females to adult males in the monthly captures. 

The reproductive pattern observed in Pteronotus mesoamericanus was restricted seasonal monoestry, beginning with males with scrotal testes in November, followed by gestation starting in January and lasting until April, when it peaked (77 %). The first births occurred also in April and increased with the beginning of the rainy season in May and June (Fig. 2C). Captured individuals were mostly adults (98.6 %). The best model (MPR + RHU + ROO + SEX + TEM, ω= 0.997) indicated that five of the six variables contributed to explain reproductive activity (Table 2). Reproductive activity was positively associated with females (β= 1.870, P < 0.001) captured in the El Apanguito cave (β= 1.976, P < 0.001), when temperature was higher than 25 ºC (β= 3.917, P < 0.001), monthly precipitation less than 100 mm (β= 12.220, P < 0.001) and relative humidity greater than 90 % (β= 1.909, P < 0.001) (Table 3).

Natalus mexicanus presented a restricted seasonal monoestrous pattern beginning in March and April, with the occurrence of gestating females, followed by lactating females in May and June. There were no records of males with scrotal testes (Fig. 2D). All captured individuals were adults, except one juvenile individual, which was observed in June. The best model (MPR + RHU + ROO + SEX + TEM, ω= 0.905) indicated that five of the six variables contributed to explain reproductive activity (Table 2). It was positively associated with females (β= 1.999, P < 0.001) captured in El Apanguito (β= 3.163, P < 0.001), when humidity presented values between 80 - 89 % (β= 2.793, P= 0.004), the temperature was higher than 25 ºC (β= 1.526, P< 0.001) and precipitation less than 100 mm (β= 2.534, P < 0.001) (Table 3).

Artibeus toltecus, a phyllostomid bat, showed seasonal bimodal polyoestry pattern. The first reproductive period was observed from July to November and the second between January and April. Gestating females were observed of July to October, and November, and January-April. Males with scrotal testes were recorded from July to October and from January to April (Fig. 2E). The species did not use any roosts for lactation. Adult individuals comprised 88 % of captures and all juveniles were recorded in March. The best model (RHU + ROO + SEX + TEM, ω= 0.956) indicated that four of the six variables contributed to explain reproductive activity (Table 2). Reproductive activity was positively associated with females (β= 1.031, P= 0.004), when the relative humidity presented values between 80 - 89 % (β= 1.807, P= 0.003) and the temperature of 20 - 24 ºC (β= 2.401, P < 0.001) (Table 3).

Carollia perspicillata male data suggested a seasonal bimodal polyoestry pattern. In the first period, active males were only recorded from July to October and no pregnant or lactating females were recorded in this period. The second period was from January to April. Males with scrotal testes were recorded from January to April and gestating and lactating females were observed from January to April (Fig. 2F). Adult individuals represented 82 % of the captures and the rest were juveniles, 97 % of which were recorded in April. The best model (RHU + ROO + TEM, ω= 0.976) indicated that three of the six variables contributed to explain reproductive activity (Table 2). The estimated β coefficients indicated that this is positively associated when the relative humidity presents values between 80 - 89 % (β= 1.584, P= 0.005) and a temperature of 20 - 24 ºC (β= 1.760, P= 0.005) (Table 3).

Glossophaga soricina showed seasonal bimodal polyoestry pattern with two reproductive periods. The first occurred between August and November, and the second was observed from February to May. Males with scrotal testes were recorded from September to November, March and May. Gestating females were observed from August to November and from February to April (Fig. 2G). Adults comprised 75 % of the captures. The best model (MPR + ROO + SEX + TEM, ω= 0.738) indicated that four of the six variables contributed to explain reproductive activity (Table 2). Estimated β coefficients indicated that this was positively associated with females (β= 1.050, P= 0.004) captured in the Mine (β= 1.891, P < 0.001), when the precipitation was less than 100 mm (β= 1.073, P < 0.001) and the temperature of 20 - 24 ºC (β= 1.914, P < 0.001) (Table 3).

Discussion

According to the results obtained and the hypotheses proposed, the monthly abundances of bat species varied between shelters (except Glossophaga soricina) and only for Mormoops megalophylla and Pteronotus fulvus between seasons. In addition, for the seven species of bats, at least one reproductive pattern was identified. Bat species varied in the time of the beginning and the duration of their reproductive periods. The reproductive activity of the species is influenced by the six variables selected in the GLMs (season, sex, roost, temperature, precipitation and relative humidity), highlighting the temperature, which was present in the best model of the seven species studied, while humidity and precipitation were present in the models of six and four species, respectively. The roost, except for P. fulvus, also influences the reproductive activity of all the species of bats studied. Furthermore, in the best supported model of all the species and separately, the probability of reproductive activity was higher for females captured in El Apanguito or La Mina in the dry season (October-April), when the precipitation is less than 100 mm. Therefore, in bat species the beginning and duration of their reproductive periods will vary according to climatic factors and ecological characteristics of the species and its roosts.

In general, except Artibeus toltecus and Natalus mexicanus, the highest abundance of males with scrotal testes occurred in the mid-dry season, from November to January. Gestating females were observed at the end of the dry season, while births occurred at the beginning of the rainy season, in May, June and July. Reproductive activity may be synchronized with the beginning of the rainy season, mainly in monoestrous species, although in bimodal polyestrous species, at least one reproductive period coincides with the rainy season when there is a higher abundance of food resources available for lactating females that have increased energy requirements, as has been previously suggested by some authors (^{Fleming et al., 1972}; ^{Racey, 1982}; ^{Dinerstein, 1986}; ^{Estrada & Coates-Estrada, 2001}a). This reproductive strategy has been observed in species which base their diet on fruits and insects (^{Bradbury & Veherencamp, 1976}; ^{Racey, 1982}; ^{Cumming & Bernard, 1997}; ^{Racey & Entwistle, 2000}).

The influence of seasonality may not be so marked mainly in species that have more than one estrous cycle per year, that is, species that show reproductive activity during most of the year, where it has also been observed that the peaks of reproductive activity do not coincide significantly with the periods of greater food availability (for example C. perspicillata). This could be due to the type of food that the species consumes. For example, in C. perspicillata the main and most important element in its diet are fruits of Piper and Solanum, which bear fruit throughout the year (^{Fleming, 1991}; ^{Estrada, Coates-Estrada, & Meritt, 1993}). In addition, other authors have suggested that this species rather presents a generalist strategy, that is, it is able to consume other types of resources that are available during a particular season, being able to extend its diet to insects when the availability of Piper fruit is low (^{Mello, Schittini, Selig, & Bergallo, 2004}a, ^{Mello, Schittini, Selig, & Bergallo,2004}b).

Other important factors in bat reproductive activity, according to the GLMs, were relative humidity and temperature. In phyllostomids the probability of reproductive activity is higher when the relative humidity is 80 - 89 % and the temperature is 20 - 24 ºC, while in mormopids it occurs at temperatures higher than 25 ºC and lower relative humidity (70 - 79 %), in the case of Pteronotus fulvus and Mormoops megalophylla, and higher (90 %) for P. mesoamericanus. These conditions appear in the months of May and June and the most evident changes could be observed in El Apanguito, where the formation of maternity colonies of thousands of females of P. mesoamericanus, P. fulvus, and N. mexicanus was observed. In April, females of P. mesoamericanus dominated the roost, representing 90 % of the individuals present in the cave. In the same month, females coming from different sites congregated to complete gestation and the care and development of offspring (we corroborated through recaptures that some females came from both La Mina and Cerro Huatulco); whereas males of P. mesoamericanus leave the cave, showing segregation of sexes during this period. This behavior has been reported in Sinaloa (^{Bateman & Vaughan, 1974}) and Colima in Mexico (^{Torres-Flores et al., 2012}).

The restricted seasonal monoestry pattern observed in Pteronotus fulvus agrees with that reported by other authors (^{Wilson, 1973}; ^{Bateman & Vaughan, 1974}; Jiménez-Guzmán & Ceballos, 2005). Due to the low number of recorded lactating females (0.2 % in El Apanguito and absent in Cerro Huatulco), it can be inferred that lactation takes place in an alternative roost and is likely to occur between May and September, as has been reported by other authors in Mexico (^{Adams, 1989}; ^{Jiménez-Guzmán & Ceballos, 2005}; ^{Torres-Flores et al., 2012}). According to the study conducted by ^{Torres-Flores et al. (2012}) in Cueva El Salitre, Colima, Mexico, there are variations in time and duration of the reproductive period. In this study, reproduction occurred from December to January as shown by males with scrotal testes and gestational females from February to April, while in El Salitre active males were recorded in January-April, September, and November-December and gestational females in August-November and March-November. These variations could be explained by adjustments in the reproductive strategies of the species according to the habitat, since in El Salitre the habitat is a relict of lowland deciduous forest, while in this study it is subperennifolia medium forest.

The restricted seasonal monoestry pattern observed in Mormoops megalophylla and Natalus mexicanus also agrees with that reported in other studies (^{Sánchez-Hernández, Chavez-Tapia, Nunez-Garduño, Ceballos-Corona, & Gurrola-Hidalgo, 1985}; ^{Bonaccorso et al., 1992}; ^{Rezsutek & Cameron, 1993}; ^{Boada et al., 2003}; ^{Torres-Flores et al., 2012}). In the case of M. megalophylla, it was possible to corroborate through individuals marked in El Apanguito and recaptured in Cerro Huatulco, that the species presents sexual segregation during lactation. In May, females leave El Apanguito and move to another unidentified roost, while males go to Cerro Huatulco (colony formed exclusively by males). This behavior has been reportedin Venezuela (^{Bonaccorso et al., 1992}), Ecuador (^{Boada et al., 2003}), and México (^{Torres-Flores et al., 2012}). Although variations have been observed in dry forests where M. megalophylla does not completely abandon the roost, which, according to the hypothesis proposed by ^{Torres-Flores et al. (2012}), could indicate an adjustment in reproductive strategies according to habitat, food availability or competition.

^{Torres-Flores et al. (2012}) mentioned that when the reproductive period of the species begins in a refuge, the most abundant species will tend to use a larger perching area and will displace others whose abundance decreases drastically. This behavior could be observed in Pteronotus mesoamericanus in El Apanguito, where abundances increase considerably from March, with the arrival of pregnant females that cover a greater perching area. This may be why M. megalophylla, P. fulvus, and N. mexicanus are forced to leave the cave El Apanguito in order to carry out parturition and lactation in other alternative roost. However, the abandonment of the cave by these species could be related to their physiological requirements during the breeding season, because the best GLM model supported for this species (MPR + RHU + ROO + TEM) indicated that reproductive activity is mainly favored by environmental factors such as temperature (> 25 ºC), precipitation (< 100 mm) and relative humidity (70 - 79 %). In May, when El Apanguito cave is completely abandoned, the temperature was 21 ºC and 92 % relative humidity. Probably these unsuitable characteristics are what force the species to move to other sites. In addition, it is known that Mormopids have a preference for shelters with temperatures higher than 30 °C in other roosts (^{Rezsutek & Cameron, 1993}; ^{Ávila-Flores & Medellín, 2004}; Torres-Flores & Santos-Moreno, 2017; ^{Ayala-Téllez, Iñiguez-Dávalos, Olvera-Vargas, Vargas-Contreras, & Herrera-Lizaola, 2018}). Maximum temperatures are not higher than 27 °C in any of the shelters observed in this study. Therefore, it would be useful to investigate the variations in the environmental characteristics of the shelters, as well as the specific requirements of each species in subsequent studies to corroborate these hypotheses.

The seasonal bimodal polyoestry pattern observed in the phyllostomids Artibeus toltecus, Carollia perspicillata and Glossophaga soricina agrees with reports from México (Michoacán, ^{Sánchez-Hernández et al., 1985}; Jalisco, ^{Iñiguez-Dávalos, 1993}; Guerrero, ^{Almazán-Catalán, Sánchez-Hernández, Romero-Almaraz, Sánchez-Vasquez, & González-Pérez, 2015}; ^{Ramírez-Pulido et al., 1993}; Oaxaca, ^{García-García et al., 2010}; Colima, ^{Torres-Flores et al., 2012}), Costa Rica (^{Fleming et al., 1972}; ^{Dinerstein, 1986}), Panama (^{Fleming et al., 1972}), and Brazil (^{Mello & Fernández, 2000}). The best supported GLMs of the three species showed temperature and roost as important variables in common (A. toltecus: RHU + ROO + SEX + TEM, C. perspicillata: RHU + ROO + TEM, G. soricina: MPR + ROO + SEX + TEM), while the precipitation was only for G. soricina and the temporality for none of the three.

This reproductive pattern, seasonal bimodal polyoestry, is characterized by two estrous cycles per year, one at the end of the dry season (March-April) and another one at the end of the rainy season (July-August) (^{Hill & Smith, 1984}). In the populations studied, this pattern was observed in A. toltecus, while in G. soricina the second period begins later, in September-November in La Mina. The absence of lactating females of A. toltecus and G. soricina in La Mina, where the largest colonies are found (Table 1), indicates that births occur in one of the nearby tunnels between January and September as suggested (^{Webster & Jones, 1982}; ^{Álvarez & Álvarez-Castañeda 1991}; ^{Cloutier & Thomas, 1992}; ^{Almazán-Catalán et al., 2015}). Also, in these months the temperature and humidity in La Mina decrease below 19 ºC, the humidity between 85 - 90 % and the water level increases. Therefore, the availability of other nearby tunnels without water flow and with more suitable temperature and humidity (according to the best supported models, the probability of reproductive activity increases by three (C. perspicillata, OR= 3.201), six (G. soricina, OR = 6.147) and seven (A. toltecus, OR = 7.777) times more when the temperature goes from 20 - 24 ºC). It offers them the opportunity to give birth and lactation in a safer way, since, unlike the sampled tunnel, does not present a current of water throughout the year, posing a danger to the young if they fall and die from drowning.

In the case of G. soricina, it has been reported that it is a species that can perch in a variety of sites and forms maternity colonies (^{Álvarez, Willing, Jones & Webster, 1991}; ^{Uribe & Arita, 2005}). A colony may be in an abandoned house located 150 m from the sampled tunnel. The colony remained during the entire sampling period of this study; however, it was not possible to capture individuals and confirm that females go to that roost during birth and lactation. Unlike the tunnel with water flow, the abandoned house represents a better refuge for the growth of the offspring and against adversities.

The results of this study show that the reproductive cycles of the studied bat species are related to at least six variables (season, sex, roost, temperature, precipitation and humidity), highlighting the temperature, relative humidity and the roost site of the species. Future studies should increase the number of variables and include other variables, for example, weight, metabolic mass, or food availability, to help make more solid conclusions about the reproductive processes of Neotropical bat species. In addition, vaginal cytology studies can be complementary and generate more specific results of the time and duration of the reproductive patterns of bats.

Ethical statement: authors declare that they all agree with this publication and made significant contributions; that there is no conflict of interest of any kind; and that we followed all pertinent ethical and legal procedures and requirements. All financial sources are fully and clearly stated in the acknowledgements section. A signed document has been filed in the journal archives.