Effects of source–sink dynamics on harvest policy performance for yellow perch in southern Lake Michigan

We used a spatially explicit stochastic simulation model to evaluate whether source–sink population dynamics would affect performance of alternative harvest policies for yellow perch, Perca flavescens, in southern Lake Michigan. The model contained four management areas in southern Lake Michigan representing each U.S. state's waters. We parameterized the model such that all recruitment was produced by only one management area, considering each of the four areas to be the sole source in turn, and contrasted results with a base scenario where all areas produced recruits. We evaluated three types of harvest policies: constant-F, where fishing mortality was constant, and two state-dependent policies, where fishing mortality was constant above either 40% or 70% of unfished spawning stock biomass (B0) and decreased to 0 at 0% B0. We used four performance statistics to evaluate polices: (1) average percentage of B0 remaining, (2) percentage of years with low spawning stock biomass, (3) average recreational harvest, and (4) percentage of years with low recreational harvest. Performance of harvest policies differed predictably depending on which management area was the source because relative productivity of stock-recruitment relationships and growth patterns differed among source areas. Thus, if management areas on the western side of Lake Michigan were the source of most of the recruits, the fishery could support higher fishing mortality rates than if areas on the eastern side of the lake were sources. State-dependent harvest policies were less sensitive to assumptions about the source of recruits than were constant fishing mortality rate policies. The 0–70 policy was most robust to source–sink dynamics across most of the performance statistics and levels of fishing mortality.