Throwing down a genomic gauntlet on fisheries-induced evolution

Beginning with studies on crypsis and camouflage, the hypothesis that predators can generate evolutionary change in their prey has a long and rich history (1). Few predators, however, rival humans in their potential to generate selection responses and concomitant phenotypic change on contemporary timescales. In the 1930s, J. B. S. Haldane (2) mused that fishing would be an ideal candidate for such “observable evolution” within a human lifetime, proceeding “with extreme and abnormal speed.” However, it was not until the late 1970s that research on fisheries-induced evolution (FIE) gained a substantive scientific foothold, beginning with thought-provoking work on Canadian whitefish (Coregonus clupeaformis) (3) and Pacific salmon (Oncorhynchus spp.) (4). Backed by the logical premise that high fishing mortality (selective or not) can generate evolution in heritable traits, a great deal of research on FIE has been empirically fueled by the reductions in age (and often size) at maturity routinely observed in heavily fished natural populations (5). Laboratory studies have repeatedly found that highly intense selection (e.g., 75 to 90% mortality per generation), imposed by ageand size-based culling of obligatorily semelparous experimental populations, can alter life histories and gene frequencies (6–8). These experiments stimulated a slew of mathematical models that have predicted undesirable consequences of FIE—altered life histories, reduced recovery potential, lowered per capita population growth (r), decreased sustainable yields—against a backdrop of FIE-induced genetic change that might prove difficult or impossible to reverse (9, 10). Calls for evolutionarily enlightened management have ensued (11, 12). Despite a theoretically strong conceptual basis, evidence of genetic change unequivocally attributable to wild-capture fisheries has been elusive (10, 13). Among the top five threats to biodiversity, evidence for genetic trait change is strongest for studies of pollution and weakest for studies of overexploitation (and habitat change) (14). Determining whether phenotypic change in declining populations is the result of evolution, as opposed to other influences on growth, survival, and fitness, or gene flow from adjacent populations, has proven challenging (5, 9, 10, 13, 15). A recent paper in PNAS (16) threatens to pierce a 21st-century narrative that FIE has been broadly and negatively implicated in effecting phenotype change in exploited fish populations (9–11). Pinsky et al. (16) examine two geographically distinct populations (“stocks” in management parlance) of Atlantic cod (Gadus morhua) to test for evidence of genomic change before and after fishery-induced population collapses: northern cod (southeastern Labrador to Newfoundland’s Grand Banks) and northeast Arctic cod (northern coastal Norway and the Barents Sea). Historical and contemporary tissues (northern cod: 1940, 2013; northeast Arctic cod: 1907, 2011, 2014), collected from the same locations over time (Twillingate, Canada; Lofoten, Norway), provided sufficient DNA to detect 346,290 singlenucleotide polymorphisms (SNPs) in 113 individuals. These cod stocks are strong candidates for exploring temporal shifts in gene frequencies and genetic variation resulting from FIE. Both experienced decades of overfishing, population biomass declining by 90% or more over roughly 30-y periods, and both exhibited considerable phenotypic change in life history, reflected by 30 to 35% (2 to 3 y) reductions in age at maturity (10, 16). However, Pinsky et al.’s analyses of whole-genome sequence data yielded no evidence of unusually strong shifts in allele frequencies over time (16). Large genomic changes were absent. Distinct signals of strong selective sweeps over time, anywhere in the genome, were lacking. There was no evidence of a substantive loss of genetic diversity. Effective population sizes remain high (∼10,000 to 36,000). In short, there was little evidence of outlier loci suggestive of rapid evolution. On the other side of the ledger, the authors acknowledge that their methods and sample sizes could