Stratification of Juvenile Migrant Data: Finding the Balance Between Bias and Precision

Over the last decade, monitoring of outmigrating juvenile salmonids has received increased attention due to the listing of many Pacific Northwest salmon and steelhead stocks under the Endangered Species Act. The objectives of many of these monitoring efforts are to collect data with levels of precision and accuracy (bias) sufficient to detect long-term trends and describe the outmigration. In most situations census counts are not feasible and for this reason mark-recapture experiments are often used to develop capture probabilities (trap efficiency) to derive abundance estimates. Capture probability heterogeneity during the outmigration can bias the abundance estimate. Yet temporally stratifying the mark-recapture data into short distinct trap efficiency trial strata will account for this bias. Conversely, estimate precision decreases as the number of efficiency trial strata increases. Pooling continuous homogenous efficiency trials together will increase estimate precision. Therefore, the process of stratifying or pooling mark-recapture data will inversely affect the estimate precision or bias. Because stocks listed as either threatened or endangered are inherently scarce or because capture probabilities are low, the ability to detect differences between efficiency trials is often low. In this presentation, we will evaluate the consequences of inappropriately pooling mark-recapture data on the bias of the abundance estimate. Given the level of stratification often required to minimize bias of the abundance estimate, we will also evaluate the likelihood of achieving precision levels, recommended by NOAA fisheries, to detect long-term trends and assess the recovery of threatened and endangered salmonids.