Assessing the Accuracy of Published Natural Mortality Estimators Using Rates Determined from Five Unexploited Freshwater Fish Populations

Abstract

Natural mortality is a vital demographic variable that is necessary for describing population dynamics and conducting fishery assessments. The instantaneous natural mortality rate (M) can be estimated through a variety of empirical methods, but they typically require extensive data collection (e.g., aging, tagging, and estimates of exploitation). Alternatively, M has been estimated using theoretically and empirically derived equations, some of which require minimal data, making their use attractive for application to information-limited fisheries. Most such equations were derived from marine stock assessments and have not been validated for freshwater fishes. We used age-frequency analyses to estimate M for unexploited populations of Largemouth Bass Micropterus salmoides, Smallmouth Bass M. dolomieu, Brown Bullheads Ameiurus nebulosus, Yellow Perch Perca flavescens, and Pumpkinseeds Lepomis gibbosus in the upper Hudson River, New York. We then compared our field-based estimates to M values predicted from nine published estimators. Using catch-curve regressions and the Robson and Chapman (1961) method, we estimated M for these populations with a sample of over 12,000 fish—nearly 2,000 of which were aged with otoliths. Variables used in the nine published estimators included maximum age, length, and weight; coefficients of the von Bertalanffy (VB) equation; the sample size used for aging; and average annual water temperature. The published estimators predicted M values that were fairly similar to our empirically derived values, although some estimators performed better than others. The mean of M values from the estimators was within ±0.12 units of the average derived from our age-frequency analyses. For the five estimators that utilized the VB growth relation, the M values were sensitive to changes in VB coefficients. For Largemouth Bass and Brown Bullheads, the unmodified VB growth functions did not conform to our observed length-at-age data, so we present modified VB equations for those species; coefficients derived from the modified equations appeared to provide more accurate predictions of M. Although the published estimators of M were primarily developed for marine fishes, they provided fairly accurate initial estimates of M (when bounded by uncertainty) for the five freshwater species we examined.

Received July 10, 2015; accepted November 30, 2015 Published online April 5, 2016

Acknowledgments

Funding for this project was provided by the General Electric Company. The views and opinions expressed in this paper are solely those of the authors.