Spatial structure and the estimation of zooplankton biomass in Lake Erie

Data collected by an optical plankton counter were analyzed to quantify the horizontal spatial patterns of zooplankton biomass in Lake Erie during 1994. During spring and late summer, an optical plankton counter was towed along a total of 11 transects, ranging from 26 to 67 km in length, distributed among the three major basins of the lake. A general model of spatial variation was used to identify large-scale trends, cyclic components at intermediate scales, and small-scale autocorrelation. Significant trends in biomass were found along all transects, while significant spatial autocorrelation underlying these trends was present in all but one case. The observed spatial autocorrelation was then removed, and standard statistical methods were used to estimate mean biomass and confidence limits. In most cases, the larger-scale spatial trends remained after removing autocorrelation. Regression methods were therefore used to estimate mean biomass in the presence of these trends. Relatively little overlap in confidence bands of the regression estimates were found from different regions of the lake. Mean biomass estimates for the entire take from two seasons, when spatial trends and autocorrelation were ignored, fell outside the 95% confidence limits of estimates that did account for these spatial structures more than 50% of time. The relatively large change in biomass within and among many transects show that spatial variability of zooplankton biomass should be incorporated in future zooplankton studies in Lake Erie and the other Great Lakes.