Population viability analysis (PVA) is a central tool for conservation decision-making. To ensure the reliability of PVA projections, it is important to identify factors that can introduce biases. This study addresses two general but overlooked related issues in PVAs that can significantly affect the reliability of their projections. First, we examined the effects of using apparent versus true survival estimates on PVA outcomes. Second, we incorporated emigration and immigration into the models to assess their influence on the accuracy of projections based on each type of survival estimate. To evaluate these concerns, we implemented PVAs structured by age, sex, and breeding status using as a study system a threatened raptor population (Bonelli's eagle, Aquila fasciata) from which apparent and true survival were available (2008–2020). The performances of PVA projections based on each survival type and dispersal process were assessed by evaluating their fit to census data. Our findings revealed that using apparent survival underestimated census data, while true survival showed a considerably better fit. However, models including dispersal processes showed that each survival type may only deliver precise projections at very specific levels of emigration and immigration. Given the potentially large differences found when using true and apparent survival, we suggest that the relevance of this issue should be elevated to that of other widely reported PVA limitations. Recently developed, accessible analytical methods may permit an easier estimation of true survival and dispersal processes. Otherwise, calibrating projections against observed data may be fundamental to test the adequacy of survival estimates.
Read full abstract