A Statistically Valid Model of the Morphoedaphic Index
The morphoedaphic index (MEI) has been criticized because of the use of ratio variables in linear regression. Computationally simple, the continued use of the index is questionable given the widespread access fisheries biologists now have to computerized statistical packages. We present a statistically valid analogue to the MEI, the morphoedaphic model (MEM), that utilizes multiple regression to characterize the morphometric and fertility properties of lakes to predict annual fish yield. Surface area, lake volume, and total dissolved solids (TDS) are used to predict annual fish yield for the lake and to derive associated confidence limits. Predicted yield of the newly derived model was compared with predictions from the original MEI Comparisons were also made based on models derived from Ontario sport and commercial fisheries data sets. The MEM derived from these partitioned data sets more accurately modelled the observed long-term yields for these lakes. Analysis of the remaining outliers suggests that several additional variables and stratification may be required to further develop the precision of the statistical model.