2016 journal article

Comparison of Visual Survey and Mark-Recapture Population Estimates of a Benthic Fish in Hawaii

TRANSACTIONS OF THE AMERICAN FISHERIES SOCIETY, 145(4), 878–887.

co-author countries: United States of America 🇺🇸
Source: Web Of Science
Added: August 6, 2018

Abstract Visual surveys are conducted to rapidly estimate population densities of stream fishes, often without calibration against more established or more widely used methods to determine precision and accuracy or to correct for potential biases. We compared population density estimates from a visual survey (VS) point quadrat method widely used in Hawaii with estimates from “in hand” individual and batch mark–recapture (BMR) methods. Visual survey sampling and individual mark–recapture (IMR) sampling were conducted in three watersheds that represent gradients of land use and prevalence of nonnative poeciliid fishes on the Island of Hawaii. Focusing on adult O‘opu Nākea Awaous stamineus , VSs were conducted prior to IMR events to allow direct comparisons of results independent of location and time. Density estimates of O‘opu Nākea from VS and IMR samplings were strongly correlated, although VS estimates were generally higher and underrepresented exceptionally large fish. Batch mark–recapture estimates of O‘opu Nākea densities were conducted for comparison with VSs at 13 sites across the archipelago. Estimates of VSs were not significantly different from BMR estimates. Estimates of VSs also exhibited less variance than did BMR estimates across sites. General linear models showed that the relationship between VS and IMR estimates varied significantly among watersheds but not seasons and that land use was associated with a greater mismatch between VS and BMR estimates of population density. These findings indicate that visual surveys using a point quadrat method are an efficient and accurate approach for estimating the abundance of small benthic fishes, such as O‘opu Nākea, in wadeable streams and that obtaining absolute densities or size distributions from VS methods would benefit from a calibration with IMR not BMR estimates. Received July 15, 2015; accepted February 23, 2016 Published online June 29, 2016