Improving Weight-Length Relationships in Fish to Provide More Accurate Bioindicators of Ecosystem Condition

Abstract

Bioindicators are effective tools for evaluating ecosystem condition. Weight-length models are essential to using fish as bioindicators, providing expected weights for healthy fish of given lengths. The traditional model, W(L) = aLb, is widely used and fits many fish taxa but is error-prone and has undesirably large uncertainties.This study evaluated a proposed improvement, replacing ? with scaling parameter L1: W(L) = 1000(L/L1)b. The primary hypothesis was that the proposed model would have lower mean parameter uncertainties than the traditional model and smaller uncertainties in most data sets, yielding more accurate bioindicators. The models were compared for 160 data sets including 94 taxa containing 14,102data points. Each set was fit to the traditional model and the proposed improvement with appropriate regression techniques. The improved model yielded lower uncertainties for L1 but similar uncertainties to the traditional model for b. Lower L1 uncertainties provide more sensitive bioindicators. The secondary hypothesis was supported: L1 shows promise as a new bioindicator because its value increases when fish are stressed by suboptimal conditions including the Deepwater Horizon oil spill, oyster reef destruction, and overpopulation of invasive species. L1 is sensitive, accurate, and valuable in conjunction with condition factor to assess environmental well-being.