Mitochondrial DNA divergence among popylations of American shad (Alosa sapidissima): how much variation is enough for mixed-stock analysis?

We investigated the geographic distribution of the mitochondrial DNA (mtDNA) diversity of American shad from 15 North American rivers in 1992 with the intent of assessing sampling efficiency for future mixed-stock analysis. We observed 116 haplotypes among the 988 individuals assayed. Because no single or group of haplotypes completely discriminated river stocks or regional complexes, we investigated haplotype frequencies as stock descriptors. Analysis of four unique indices of haplotype divergence indicated that including rather than suppressing restriction site heteroplasmy increased resolution; however, the final results were not overwhelmingly dependent on this choice. A redundancy of variation among restriction enzymes diminished information returns rapidly after considering the six best enzymes, caused by physical linkage of restriction sites on the mtDNA molecule. Stock discriminatory power was tested by computing allocation efficiencies of mtDNA characters. When each individual was temporarily removed from the data set and reallocated to the various candidate populations on the basis of haplotypic similarity, 28% of the reallocations were correct, a fourfold increase over random success. We demonstrate that although the specific stock identity of individuals cannot be confidently established, the haplotypic arrays from baseline stocks can support stock identification and mixed-stock analysis for shad because rivers support stock-specific haplotype frequencies.