PRERELEASE STRATIFICATION IN TAG-RECOVERY MODELS WITH TIME-DEPENDENCE

A statistical model is developed for estimating time-dependent surviva l and recovery rates of tagged populations that are stratified into gr oups at release. For populations of fish, the groups might include dif ferent sexes or stocks, or they might be distinguished by the location or method of capture. In the model developed here, survival and recov ery rates are formulated as linear combinations of parameters that inc lude the effects of time, group, and time-group interactions. Methods of testing the statistical significance of these potential sources of variation are described and illustrated with the recoveries of differe nt stocks of anadromous striped bass (Morone saxatilis). For the strip ed bass data, elimination of time- and stock-specific sources of varia tion substantially improved the precision of survival estimates withou t introducing important bias. During 1988-89 the average survival of b oth stocks was 0.78 (SE = 0.04). Calculations of statistical power ind icate that stock-specific differences in survival of +/- 0.2 could hav e been detected about 50% of the time, given the low rates of recovery and numbers of fish released.