Abstract
BACKGROUND: Differences among species and among years in reproductive seasonality (the tendency for clusters of events to fall at approximately the same point in each year) and synchrony (amount of clustering of events within a year) have been intensively studied in bats, but are difficult to assess. Here, we use randomization methods with circular statistics to test for synchrony and seasonality of reproduction in three species of nectarivorous megachiropteran bats on Negros Island in the central Philippines. RESULTS: In Rousettus amplexicaudatus, estimated dates of birth were both highly synchronous and highly seasonal. In Macroglossus minimus, estimated births were seasonal and significantly clustered within years, but within each year births occurred over a broad period, indicating a low level of synchrony. In Eonycteris spelaea, estimated births were also seasonal and had statistically significant synchrony, with birth periods within years intermediate in synchrony between R. amplexicaudatus and M. minimus. All three species had a similar seasonal pattern, with two birth periods in each year, centered on March or April and August or September. In one species, R. amplexicaudatus, primigravid females (in their first pregnancy) produced their young in June and July, a birth period significantly different in timing from the two birth periods of older adult females. This more conservative pattern of young females may allow higher survival of parents and offspring at cost of a lost reproductive opportunity. There was weak evidence that in some years primigravid females of M. minimus might differ in timing from older adults. There were few significant differences in reproductive timing among different years, and those differences were generally less than two weeks, even during a severe drought in the severe el Niño of 1983. CONCLUSION: The results suggest that these species follow an obligately seasonal pattern of reproductive timing with very little phenotypic plasticity. The resampling methods were sensitive to differences in timing of under two weeks, in some cases, suggesting that these are useful methods for analyses of seasonality in wild populations of bats.