Finite study areas and vital rates: sampling effects on estimates of spotted owl survival and population trends

1. Evaluating the status of endangered wildlife depends upon well-designed field studies. Finances and logistics often constrain field studies to finite (limited-sized) areas where inductive inferences are needed to extrapolate results to populations. Although available quantitative techniques for analysing data are robust to many aspects of field investigations, few investigators assess the influence of their study area size on estimators of population parameters and subsequent inferences derived from those estimators. 2. We used mark-recapture to monitor an entire population of spotted owls Strix occidentalis in the San Bernardino Mountains of southern California (2140 km(2)) for which we knew the approximate true rate of survival. We defined hypothetical study areas of varying size by subsampling the population in increments of five territories; we then estimated apparent survival and emigration for non-juvenile and juvenile owls within each of these sample study areas to assess the influence of study area size on estimators of survival and population trends. 3. Estimated survival rates of juvenile spotted owls increased approximately fourfold from the smallest sample area to the largest ((phi) over cap (min) = 0.08, S (E) over cap = 0.03; (phi) over cap (max) = 0.33, S (E) over cap = 0.03). In contrast, estimates of apparent survival for non-juvenile owls did not vary with study area size (range (phi) over cap (non-juvenile) = 0.80-0.82, S (E) over cap = 0.01-0.03). 4. Juvenile emigration was extremely high in the smallest study area ((psi) over cap (juvenile) = 0.77, S (E) over cap = 0.09) and remained above 10% until > 62% of our study area (approximately 900 km(2)) was encompassed by a sample study area. Non-juvenile owls had low annual emigration probabilities from all sample study area sizes (range (psi) over cap (non-juvenile) = 0.00-0.02). 5. Although estimates of lambda (finite population growth rate) increased gradually from 0.828 to 0.903 as the subsample increased from 20 to 143 territories, these estimates were similar to the 'true' value. 6. Synthesis and applications. We provide direct estimates of the bias that sampling limited study areas has on emigration and mark-recapture estimators of survival. Our results demonstrate that permanent emigration from limited study areas can lead to underestimates of survival and population growth rates. In addition, our approach illustrates a technique for using multistate models to assess study design and estimator assumptions.