Population structure of sigmodontine rodents through age estimation by individual growth models

Abstract

Ecological studies aimed at identifying and characterizing seasonal demographic patterns in rodent populations often face methodological difficulties. First, imperfect detectability could lead to biased abundance estimates and, in particular, yield biased proportions of individuals of different age groups. Second, age determination methods that require killing animals are undesired, particularly in longitudinal studies. In this work, we develop a strategy to overcome those problems by using growth models of two readily in-the-field measurable traits (body length and mass) to obtain age estimates based on recapture data of two species of sigmodontine rodents. We used extrapolated birth dates as a complement to data obtained from visual examination of adults’ genitals to assess reproductive activity quantitatively. Our method revealed that births showed a yearly cycle of high and low frequency, but occurred throughout the year without fully stopping. Moreover, we found that young individuals (age < 2 months) are not detectable and also a large fraction of the adult population was not detected until several months after they began to be detectable. However, adding the youngest group of the population to typical minimum-number-alive abundance estimates, based on age estimates, allowed us to obtain a complete picture of populations’ age structures and abundance cycles, partly solving the detectability problem. Overall, our results suggest that the proposed age determination method has a great potential for exploiting longitudinal data and may be particularly useful for conservation programs and epidemiological studies, where demographic patterns play an important role.