Estimating recruitment and survival in partially monitored populations

In evolutionary and ecological studies, demographic parameters are commonly derived from detailed information collected on a limited number of individuals or in a confined sector of the breeding area. This partial monitoring is expected to underestimate survival and recruitment processes because individuals marked in a monitored location may move to or recruit in an unobservable site. We formulate a multi-event capture-recapture model using E-SURGE software which incorporates additional information on breeding dispersal and the proportion of monitored sites to obtain unbiased estimates of survival and recruitment rates. Using simulated data, we assessed the biases in recruitment, survival and population growth rate when monitoring 10-90% of the whole population in a short- and a long-lived species with low breeding dispersal. Finally, we illustrate the approach using real data from a long-term monitoring program of a colony of Scopoli's shearwaters Calonectris diomedea. We found that demographic parameters estimated without considering the proportion of the area monitored were generally underestimated. These biases caused a substantial error in the estimated population growth rate, especially when a low proportion of breeding individuals were monitored. The proposed capture-recapture model successfully corrected for partial monitoring and provided robust demographic estimates. Synthesis and applications. In many cases, animal breeding populations can only be monitored partially. Consequently, recruitment and immature survival are underestimated, but the extent of these biases depends on the proportion of the area that remains undetected and the degree of breeding dispersal. We present a new method to obtain robust and unbiased measures of survival and recruitment processes from capture-recapture data. The method can be applied to any monitored population regardless of the type of nests (e.g. artificial or natural) or breeding system (e.g. colonial or territorial animals), and it only relies on an estimate of the proportion of the monitored area. The unbiased estimates obtained by this method can be used to improve the reliability of predictions of demographic population models for species' conservation and management.