Annual survival estimation of migratory songbirds confounded by incomplete breeding site-fidelity: study designs that may help

Abstract

Many species of bird exhibit varying degrees of site–fidelity to the previous year’s territory or breeding area, a phenomenon we refer to as incomplete breeding site–fidelity. If the territory they occupy is located beyond the bounds of the study area or search area (i.e., they have emigrated from the study area), the bird will go undetected and is therefore indistinguishable from dead individuals in capture–mark–recapture studies. Differential emigration rates confound inferences regarding differences in survival between sexes and among species if apparent survival rates are used as estimates of true survival. Moreover, the bias introduced by using apparent survival rates for true survival rates can have profound effects on the predictions of population persistence through time, source/sink dynamics, and other aspects of life–history theory. We investigated four study design and analysis approaches that result in apparent survival estimates that are closer to true survival estimates. Our motivation for this research stemmed from a multi–year capture–recapture study of Prothonotary Warblers (Protonotaria citrea) on multiple study plots within a larger landscape of suitable breeding habitat where substantial inter–annual movements of marked individuals among neighboring study plots was documented. We wished to quantify the effects of this type of movement on annual survival estimation. The first two study designs we investigated involved marking birds in a core area and resighting them in the core as well as an area surrounding the core. For the first of these two designs, we demonstrated that as the resighting area surrounding the core gets progressively larger, and more “emigrants” are resighted, apparent survival estimates begin to approximate true survival rates (bias < 0.01). However, given observed inter–annual movements of birds, it is likely to be logistically impractical to resight birds on sufficiently large surrounding areas to minimize bias. Therefore, as an alternative protocol, we analyzed the data with subsets of three progressively larger areas surrounding the core. The data subsets provided four estimates of apparent survival that asymptotically approached true survival. This study design and analytical approach is likely to be logistically feasible in field settings and yields estimates of true survival unbiased (bias < 0.03) by incomplete breeding site–fidelity over a range of inter–annual territory movement patterns. The third approach we investigated used a robust design data collection and analysis approach. This approach resulted in estimates of survival that were unbiased (bias < 0.02), but were very imprecise and likely would not yield reliable estimates in field situations. The fourth approach utilized a fixed study area size, but modeled detection probability as a function of bird proximity to the study plot boundary (e.g., those birds closest to the edge are more likely to emigrate).
This approach also resulted in estimates of survival that were unbiased (bias < 0.02), but because the individual covariates were normalized, the average capture probability was 0.50, and thus did not provide an accurate estimate of the true capture probability. Our results show that the core–area with surrounding resight–only can provide estimates of survival that are not biased by the effects of incomplete breeding site–fidelity.