a. Mixed Linear Effects modeling results (GLMM, using a logit link function) for 28 d survival.

Random-effects covariates included in all models were replicate tagging days and surgeons. Fixed-effect covariates tested against survival were: Treatment, fork length, weight, water temperature at tagging, presence of foreign material (0,1), sutures (0,1,2), suture tearing (0,1), inflammation or ulceration at suture site (0,1,2), and healing at suture site (1,2). See Table 1 for the full list of post-incision metrics. For 28-d survival we evaluated models at three stages: Variable at time of tagging, incision metrics after 7d for survivors, and 2 –way interactions among covariates. Only models with delta AIC values within approximately 2 units from the best model were included in the table, although the process included all subsets as possible models. b. Final GLMM linear mixed model.