Estimating survival and movements using both live and dead recoveries: a case study of oystercatchers confronted with habitat change.
Animals facing partial habitat loss can try to survive in the remaining habitat or emigrate. Effects on survival and movements should be studied simultaneously since survival rates may be underestimated if emigrants are not considered, and since emigrants may experience reduced survival. We analysed movements and survival of adult wintering oystercatcher Haematopus ostralegus in response to the 1986-1987 partial closure of the Oosterschelde in the Dutch Delta. This reduced by one-third the tidal area of this major European wintering area for waders. We developed a novel variant of a multistate capture-recapture model allowing simultaneous estimation of survival and movement between sites using a mixture of data (live recaptures and dead recoveries). We used a two-step process, first estimating movements between sites followed by site-specific survival rates. Most birds were faithful to their ringing site. Winter survival was negatively affected by winter severity and was lowest among birds changing wintering site (i.e. moving outside of the Oosterschelde). During mild winters, survival rates were very high, and similar to before the closure in both changed and unchanged sectors of the Oosterschelde. However, the combined effect of habitat loss with severe winters decreased the survival of birds from changed sectors and induced emigration. The coastal engineering project coincided with three severe winters and high food stock, making assessment of its effects difficult. However, the habitat loss seems to have had less impact on adult survival and movements than did winter severity. Synthesis and applications. Human-induced habitat change may result in population decline through costly emigration or reduced survival or reproduction of individuals that stay. Long-term monitoring of marked individuals helps to understand how populations respond to environmental change, but site-specific survival and movement rates should be integrated in the same model in order to maximize the information yield. Our modelling approach facilitates this because it allows the inclusion of recoveries from outside the study area.