Using Bayesian growth models to reconstruct small-mammal populations during low-trapping periods

Small-mammal populations that fluctuate in size often undergo periods of low trappability, which could be an important factor contributing to low-density estimates based on trapping efforts. Age cohort analysis is commonly used to estimate population parameters of animals that are harvested. The method is based on known age at death that can be used for Bayesian hierarchical growth models. It is interesting to see if similar methods, hitherto conducted on long-living species, can be used on live-trapping data on short-lived and fast-growing small mammals. Using data from live-trapping surveys of Apodemus sylvaticus in Iceland, we adjusted growth curves to individual body weights to predict their birth dates. The estimated birth dates were used to estimate population density and recruitment. These were then compared with other data sources. We found out that density estimates, based on numerical methods [modified nodal analysis (MNA)], underestimated population density during the period of low trappability and that recruitment occurred up to 100 days earlier than was observed by capture-mark-recapture (CMR) analysis and MNA. This study suggests that cohort analysis can be conducted on short-lived small mammals during periods when estimates based on CMR or numerical analysis fail because of low sample sizes. Furthermore, it is possible to use body weight of live-trapped individuals to estimate age. This is important in terms of ethics and conservation as such methods can be conducted without harming or killing the animals. We believe that live-trapping data obtained during a peak period in population density can be a useful aid when describing population parameters of previous months when low trappability prevents direct measurements