Incomplete bioinformatic filtering and inadequate age and growth analysis lead to an incorrect inference of harvested-induced changes.

Understanding the evolutionary impacts of harvest on fish populations is important for informing fisheries management and conservation and has become a growing research topic over the last decade. However, the dynamics of fish populations are highly complex, and phenotypes can be influenced by many biotic and abiotic factors. Therefore, it is vital to collect robust data and explore multiple alternative hypotheses before concluding that fish populations are influenced by harvest. In their recently published manuscript, Bowles et al, Evolutionary Applications, 13(6):1128 conducted age/growth and genomic analysis of walleye () populations sampled 13-15 years (1-2.5 generations) apart and hypothesized that observed phenotypic and genomic changes in this time period were likely due to harvest. Specifically, Bowles et al. (2020) documented differential declines in size-at-age in three exploited walleye populations compared to a separate, but presumably less-exploited, reference population. Additionally, they documented population genetic differentiation in one population pair, homogenization in another, and outlier loci putatively under selection across time points. Based on their phenotypic and genetic results, they hypothesized that selective harvest had led to fisheries-induced evolution (referred to as nascent changes) in the exploited populations in as little as 1-2.5 generations. We re-analyzed their data and found that (a) sizes declined across both exploited and reference populations during the time period studied and (b) observed genomic differentiation in their study was the result of inadequate data filtering, including retaining individuals with high amounts of missing data and retaining potentially undersplit and oversplit loci that created false signals of differentiation between time points. This re-analysis did not provide evidence for phenotypic or genetic changes attributable to harvest in any of the study populations, contrasting the hypotheses presented by Bowles et al. (2020). Our comment highlights the potential pitfalls associated with conducting age/growth analyses with low sample sizes and inadequately filtering genomic datasets.