Predicting postrelease survival in large pelagic fish

Sharks, turtles, billfish, and marine mammals are frequently caught accidentally in commercial fisheries. Although conservationists and fisheries managers encourage the release of these nontarget species, the long-term outcome of released animals is uncertain. Using blue sharks (Prionace glauca), we developed a model to predict the long-term survival of released animals based on analysis of small blood samples. About 5% of the sharks were landed in obviously poor condition (lethargic and unresponsive to handling); these moribund sharks were sampled and euthanized. A subset of the remaining sharks was sampled and tagged with pop-up satellite archival tags (PSATs). Each of the PSATs that reported data (11 tags) showed that the sharks roamed at sea for at least 3 weeks post-release. Five variables differentiated moribund sharks from survivors: plasma Mg2þ (moribund, 1.57 6 0.08 mM; survivor, 0.98 6 0.05 mM; P , 0.00001), plasma lactate (moribund, 27.7 6 4.1 mM; survivor, 5.80 6 2.96 mM; P , 0.001), erythrocyte heat shock protein 70 (Hsp70) mRNA (relative levels: moribund, 3.96 6 0.53; survivor, 1.00 6 0.29; P , 0.005), plasma Ca2þ (moribund, 3.70 6 0.14 mM; survivor, 3.13 6 0.11; P , 0.005), and plasma Kþ (moribund, 7.01 6 0.66 mM; survivor, 5.12 6 0.44 mM; P , 0.05). These analyses were used to develop logistic regression models that could ‘‘predict’’ the long-term survival of captured sharks, including a larger group of sharks that we sampled but did not tag. The best logistic model, which incorporated Mg2þ and lactate, successfully categorized 95% of fish of known outcome (19 of 20). These analyses suggest that sharks landed in an apparently healthy condition are likely to survive long term if released (95% survival based on biochemical analyses; 100% based on PSATs).