Post-handling mortality during controlled field trials with marine grow-out Atlantic salmon, Salmo salar L

During the course of a multi-site clinical field trial comparing the field effectiveness of Atlantic salmon (Salmo salar L.) vaccines, an opportunity arose to examine the potential effects of repeated fish handling on the controlled trial outcome. Fish-level data obtained by seining, dip-netting and anesthetizing each of the fish among eight study populations comprising approximately 6000 individually tagged fish per group, provided detailed follow-up for the planned clinical trial. However, the stressful conditions generated during the sampling process resulted in variable post-handling mortality which also provided the unplanned opportunity to evaluate the effect of field procedures on survival. A multilevel mixed effects logistic regression model was used to analyze factors related to mortality occurring in the two-week period following scheduled fish handling events. Data on individual fish level characteristics such as weight, length, sex, physical deformities, lesions, vaccine group allocation, days since vaccination, days since transfer to seawater, and elapsed time spent in the seine were recorded for every fish at each sampling event. In total, the dataset included information on 58,923 fish from 13 sampling events, and contained a total of 648 mortalities. Vaccine group membership was not found to have a significant impact on mortality (p = 0.45). Fish with opercular and spinal deformities were found to have significantly higher odds of mortality compared to fish without deformities (odds ratio (OR) = 1.77, p = 0.001 and OR = 2.88, p < 0.001, respectively). Gender was found to be a significant predictor of mortality, with females having significantly greater odds of mortality compared to males in the two-week post-handling period (OR = 1.28, p = 0.002). The time of sampling relative to smolt transfer (i.e. first-year vs. second-year of seawater production) was found to significantly predict mortality (OR = 47.6 for second year vs. first year, p < 0.001). Significant interactions with the timing of the sampling event (first-year vs. second-year) were found for both weight and seine residence times (p < 0.001 for both). A 100 g increase in weight at the time of sampling, had significantly less impact on the mortality in sampling events during the second-year of saltwater production than those in the first-year (OR = 0.641 vs. 0.937). The impact of each additional hour in seine residence time was greater in first-year sampling events compared to second-year sampling events (OR = 2.38 vs. 1.28 respectively; both with p < 0.001). While the effect of vaccine group allocation on survival was not found to be influenced by the intensive sampling regime outlined in the controlled field trial protocol, fish characteristics such as spinal deformity, opercular deformity and sex were found to have a significant impact on mortality. The distribution of these population characteristics is difficult to control in field trials in a commercial setting, emphasizing the need for random allocation to evenly distribute these potential confounders to treatment groups. Furthermore, recording and incorporating such variables in the statistical analysis of field trials with repeated handling events provide an assessment of their outcomes