Estimation and comparison of heritability and parent-offspring resemblance in dispersal probability from capture-recapture data using different methods: the Collared Flycatcher as a case study

Understanding the evolution of a trait requires analysing its genetic basis. Many studies have therefore estimated heritability values of different traits in wild populations using quantitative genetic approaches on capture-recapture data of individuals with known parentage. However, these models assume perfect individual detection probability, a hidden hypothesis that is rarely met in natural populations. To what extent ignoring imperfect detection may bias heritability estimates in wild populations needs specific investigation. We give a first insight into this question using dispersal probability in a patchy population of Collared Flycatchers Ficedula albicollis as an example. We estimate and compare heritability and parent-offspring resemblance in dispersal obtained from (1) quantitative genetic approaches ("classical'' parent-offspring regressions and more recent animal models) and (2) multi-state capture-recapture models accounting for individual detection probability. Unfortunately, current capture-recapture models do not provide heritability estimates, preventing a full comparison of results between models at this stage. However, in the study population, detection probability may be expected to be lower for dispersing compared to philopatric individuals because of lower mating/breeding success and/or higher temporary emigration, making the use of capture-recapture models particularly relevant. We show significant parent-offspring resemblance and heritable component of between-patch dispersal probability in this population. Accounting for imperfect detection does however not seem to influence the observed pattern of parent-offspring resemblance in dispersal probability, although detection probability is both sensibly lower than 1 and heterogeneous among individuals according to dispersal status. We discuss the problems encountered, the information that can be derived from, and the constraints linked to, each method. To obtain unbiased heritability estimates, combining quantitative genetic and capture-recapture models is needed, which should be one of the main developments of capture-recapture models in the near future.