Abstract
Fish that are produced in hatcheries often produce fewer surviving adult offspring than do wild fish when both spawn in the wild. This difference in fitness is likely due to inadvertent selection for adaptation to the hatchery environment. Size at release is positively correlated with survival at sea. Therefore, selection should favor traits that promote fast growth in the hatchery even if those traits are maladaptive in the wild. In that case, changing hatchery conditions to reduce the variance in size at release would reduce the opportunity for domestication selection. Here we test whether grading by size and raising each size group separately can substantially reduce the variance in size at release. We graded a mix of 15 full-sibling families of winter run steelhead (Oncorhynchus mykiss) juveniles (initial n = 375 fish/tank) into small, medium, and large body size groups. We then promoted growth in the initially-small fish with excess food and low densities, and slowed growth in the initially-large fish with restricted feedings and high densities. We successfully held back the growth of the initially-large fish, but the initially-small fish never caught up with the controls, despite being raised under ideal conditions. This result suggests that inherent physiological or behavioral factors, rather than interactions with larger fish, restrict the growth of small fish. Both the total variance among individual fish and the variance among family means was reduced by the grading treatment, although only the reduction in total variance among individuals was statistically significant when comparing intraclass correlation values (ICC). The reduced variance due to slowing the growth of the initially-large fish indicates that grading could possibly be a tool used by hatcheries to lessen selection on salmonids. However, it would first be necessary to better understand the factors limiting growth in the smaller fish.