Discussion
We characterise the reproductive phenology of two free-tailed bat species in south-eastern Kenya – Mops condylurus and Mops pumilus – and present quantitative information on the synchronicity, magnitude, and timing of their birth pulses. Our data are the highest resolution available for these species across their distributions (Mutere 1973; Happold and Happold 1989; Vivier and Merwe 1997). This has allowed us to empirically estimate birth-pulse synchrony for the first time in M. condylurus and M. pumilus . This provides ecological information relevant for understanding mechanisms of transmission and maintenance for ebolaviruses and other zoonotic pathogens in bat populations (Peel et al. 2014; Hayman 2015).
For M. condylurus , observed patterns in reproduction had important distinctions to those in existing literature in Kenya and across their broader range. Notably, our estimate of birth-pulse synchronicity, based on more detailed data, was considerably longer than previous “best-guess” estimates (~8.5 weeks, compared to 3-4 weeks) (Mutere 1973; O’Shea and Vaughan 1980; Happold and Happold 1989; Vivier and Merwe 1997) (Table S1). We also observed proportionally more females than expected based on previous estimates of male-to-female ratios (1:1.06 vs 2.6:1), but proportionally fewer reproductively active females than expected (16.6% non-reproductively active females, vs 0%) (Happold and Happold 1989). These findings will have countering effects on the potential magnitude of the birth pulse – a population with proportionally more females will have greater potential to produce more susceptible juveniles, but only if reproductively active.
For M. pumilus , the timing of each reproductive stage was wide and largely non-defined, and birthing spanned nearly the entire study period (11 of 12 weeks). The absence of a discernible birth pulse in our data may be consistent with a pattern of continuous breeding, as described elsewhere for this species (e.g., Uganda) (Marshall and Corbet 1959; Mutere 1973; Happold and Happold 1989) or reflect a mix of two, tight birth pulses in November and March-April (O’Shea and Vaughan 1980) (Table S2). The male-to-female ratio of the M. pumilus population was lower than anticipated. Female-defence polygyny is the mating system proposed for M. pumilus , with observations of harems including one male and up to 21 females within a roost (McWilliam 1988; Bouchard 2001). The sex ratios observed in this study would suggest a more neutral male-to-female roost composition – contradicting what would be observed for a female-defence polygyny strategy. The proportion of reproductively active females were in line with previous literature (~90%), though previous estimates vary widely dependant on how many birth pulses are observed for that location (e.g., 93% of females have been observed to birth to 3 litters in Ghana, and 80% in Malawi. The proportion of females achieving the maximum of 5 births/year has not been determined) (Marshall and Corbet 1959; Mutere 1973; McWilliam 1976; Happold and Happold 1989).
Theory suggests that the large within-pulse asynchrony observed for M. condylurus and M. pumilus should increase the probability of long-term viral maintenance within populations and is consistent with an ecological trait of a reservoir host. Birthing events of African molossid bats have been linked to human EVD spillover previously (Hranac et al. 2019), supporting the plausibility of molossid bats as reservoir hosts, and that seasonal birthing drives epidemic cycles in these species. The effects of birth pulse magnitude and timing on infection dynamics are more difficult to predict from existing theory. The intervals between birthing pulses are uneven for M. condylurus – assuming peak parturition in March and November, intervals are roughly April-October (seven months) and December-February (three months). This contrasts to even intervals used in prior models of bi-annual birthing (Hayman 2015). The potential magnitude of the birth pulse will be driven (in part) by the male-to-female ratio observed in the population, the proportion of females reproducing per birth pulse, population size, and the maturation period, which are ecological features of M. condylurus that differ from generic bat parameters used in Hayman (2015). Species-specific models are needed to interpret how, specifically, identified attributes of the birth pulse may interact with other features of M. condylurus -ebolavirus ecology to influence infection dynamics.
It is worthwhile noting that we cannot determine whether lactating non-pregnant females caught early in the M. condylurus birth pulse (~January/February) were females that had recently given birth in the recorded birth pulse, as interpreted, or had given birth in the prior birth pulse and were not yet (or not going to be) pregnant for the recorded birth pulse. If the latter is true, our estimate of synchronicity for M. condylurus could be much lower than 8.5 weeks. However, we observed a wide span of gestational stages throughout the study period. Moreover, an even wider span of gestational stages has been observed at this location during this timeframe, from previous capture events by the authors (Appendix S2). 10, 21 and 56 M. condylurus at early-, mid-, and late-stage gestation stages, respectively (here, directly determined by dissection and examination of foetus size) were captured over a 7-day capture period in early March 2019 (Appendix S2). As the gestation period for M. condylurus is 85-90 days (~3 months), this range of gestation stages (particularly those observed within a one-week period in 2019) implies an equally wide period of birthing. These data support a long birth pulse, as interpreted.
In addition, our estimate of non-reproductive females during the pulse (16.6%) could be higher than the true estimate if females at early-gestation were miscategorised as non-reproductive. The observation of early gestation by dissection of adults in late March 2019 underscores this possibility (Appendix S2). However, we would note that non-reproductive adult females were observed throughout the study period, including April, when gestation and/or lactation should be obvious. As such, it is reasonable to conclude that less than 100% of females were pregnant immediately prior to the birth pulse, which is less than previous estimates (Mutere 1973; Happold and Happold 1989; Vivier and Merwe 1997).