z-logo
Premium
Forecasting Walleye Abundance in Northern Wisconsin Lakes
Author(s) -
Madsen Eric R.
Publication year - 2008
Publication title -
north american journal of fisheries management
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.587
H-Index - 72
eISSN - 1548-8675
pISSN - 0275-5947
DOI - 10.1577/m06-201.1
Subject(s) - fishery , abundance (ecology) , environmental science , population , mixed model , geography , statistics , mathematics , biology , demography , sociology
Seventeen variations of a modified statistical catch‐at‐age model were applied to annual spawning adult mark–recapture population estimates (PEs) for walleye Sander vitreus from seven mixed‐fishery lakes in northern Wisconsin. In addition, 15 variations were applied to annual PEs for walleyes age 3 and older from Escanaba Lake. The model variations included different mortality, selectivity, and catchability assumptions, as well as different data weightings, averaging of forecasts, and naïve models using limited data. Population forecasts were examined retrospectively for each model variation to evaluate whether they were accurate enough to use as a basis for setting harvest quotas. Models were ranked according to how closely the forecasts matched the PEs and the abundance estimates fitted using all available data. A model that used estimated mortality and averaging of forecasts performed best for the large mixed‐fishery lakes, and a model that used fixed mortality and downweighted age composition data performed best for the small mixed‐fishery lakes. These model forecasts can be considered for setting harvest quotas on the mixed‐fishery lakes since they performed better than the naïve models, but the improvements were modest. For the large mixed‐fishery lakes, the mean absolute deviation and mean deviation of the forecasts from the PEs were 38% and 13%, respectively, for the naïve model using the previous year's PE as the forecast and 31% and 3%, respectively, for the best model; for the small mixed‐fishery lakes, these deviations were 34% and 16%, respectively, for this naïve model and 36% and 12%, respectively, for the best model. A naïve model using the previous year's fitted abundance as the forecast performed best for Escanaba Lake. Model forecasts performed poorly for Escanaba Lake, mostly because of the age composition of the population being estimated.

This content is not available in your region!

Continue researching here.

Having issues? You can contact us here