9360 - Laserfiche WebLink

Home Browse Search

ISMP backwater sampling evaluation each species the same number of times. We also wanted to understand the relationship between the probability of detection of each species by the ISMP technique as a function of estimated fish abundance (from DMR sampling) in the backwater. To accomplish this, we used logistic regression (PROC GENMOD, SAS Institute 1993) with a logit link and a binomial error distribution to model the probability of detection of a species as a function of its estimated abundance (see below for abundance estimation techniques). A statistically significant effect would indicate that detection probability for a particular species was proportional in some form to its estimated abundance. The logistic equation was then re-arranged to predict detection probabilities at a range of fish abundance values. That information would be useful to evaluate how many fish would need to be present in a backwater to achieve a desired probability of detection. Abundance estimates were attempted using either capture-recapture or removal sampling. For capture-recapture DMR sampling, estimates were computed using a Lincoln-Petersen estimator for two-pass data or model Mt for three-pass data, which assumes a time-varying probability of capture (program CAPTURE,White et al. 1982). Abundance estimates for removal DMR sampling were calculated by maximum likelihood techniques. Because the ultimate goal of this research was to determine bias of estimates obtained by ISMP sampling relative to DMR sampling, abundance estimates had to be converted to fish densities for equitable comparisons. The ISMP data were converted by dividing the number of fish captured in the two seine hauls by the area seined. For DMR sampling, estimated abundance of each fish species in each backwater was divided by the total area of the backwater. Both ISMP and DMR density estimates were multiplied by 10, resulting in number of fish per 10 m2 of backwater area. -6-