Using Occupancy Models to Evaluate Gear Bias: Quantifying Detection Probabilities of Bridle Shiner When Using a Seine and Backpack Electrofishing Units

Bridle shiner (Notropis bifrenatus) is declining over most of its range and is listed as a species of concern in Connecticut.  Surveys conducted in the 1960s used seine nets and found bridle shiner at 56 locations while surveys in the 1990s using backpack electrofishing units found bridle shiner at 8 of the historic sites.  It is unclear what portion of observed declines might be a sampling artifact, confounding efforts to assess the conservation status.   An occupancy model approach was used to compare the probability of correctly detecting bridle shiner presence-absence in habitat patches when sampled with a seine or backpack electrofishing units.  Both gears were used three times in same habitat patches several days apart over three surveys in July/August 2012. A single season, multi-method model was built in program PRESENCE to estimate detection probabilities.  Electrofishing detection probability was low (0.36) and approximately half that of the seine (0.78) across 18 sampled patches.  Mean velocity was the most supported habitat covariate to detection, with higher relative currents improving detection performance of the backpack. It is possible the 1990s sampling underestimated the number of populations of bridle shiner, and a repeat survey of the historic locations using a seine is recommended.