Evaluating Removal and Mark-Recapture Methods for Estimating Abundance of a Small, Non-Game Fish

Unbiased estimates of stream fish abundance are critical for sound fisheries management.  Most studies investigating the bias associated with estimates of stream fish abundance focus on salmonines, yet non-game fishes often comprise a major portion of many stream fish assemblages.  We evaluated mark-recapture (i.e., Lincoln-Peterson model with Chapman correction) and removal methods (i.e., models M_b and M_bh) for estimating the abundance of mottled sculpin (Cottus bairdii), a common non-game fish.  Specific objectives were to: (1) compare estimates of mottled sculpin abundance, (2) assess bias of removal methods by comparing estimated abundance to known abundance, and (3) evaluate closed-population model assumption.  We sampled eight streams via backpack electrofishing; each stream was sampled over 2-days.  On day one, fish were batch marked in three sections of a 90-m reach.  On day two, fish were captured and temporarily removed from the stream during four electrofishing passes; the number and marking status of fish was recorded during each pass.  Removal abundance estimates generated with program CAPTURE were significantly lower (range=22-58%) than mark-recapture estimates.  Moreover, the removal method underestimated (range=38-59%) known abundances of marked mottled sculpins.  Movement of marked fish was minimal among sections of the 90-m reach in all but one study stream.  Survival and mark retention of mottled sculpin after capture was 100% for fish retained overnight in stream enclosures (n=405 fish).  Our results suggest the closed-population assumption was valid in most streams and the removal method yielded negatively biased abundance estimates.  Consequently, we recommend using mark-recapture methods to estimate abundance of small, non-game fishes.