Effects of spatial scale on assessment of dissolved oxygen dynamics in the Atchafalaya River Basin, Louisiana

Low levels of dissolved oxygen (DO) are potential stressors of fishes and invertebrates in freshwater and marine systems. Because of the incredibly nutrient-rich and warm nature of the Atchafalaya River Basin (ARB), microbial decomposition in backwater areas with limited flow often results in potentially stressful, if not lethal, DO levels during and after the annual flood pulse. As water temperatures increase and flood levels fall, DO levels often stratify, restricting fish habitat. In order to better guide development of water management projects, we investigated the relationship between DO and DO stratification (DOD; surface DO - bottom DO) and a suite of environmental variables potentially linked to DO and DOD. Based on continuous monitoring data collected from 2000 to 2007, we used a variable selection procedure to construct statistical generalized linear models to describe the relationship between DO levels and DOD and several potentially important predictor variables at three spatial scales: the entire sampled area of the ARB; individual water management units (WMUs) or subunits (WMSs); and individual sampling sites. We conducted these analyses using models that assume the normal distribution and compared them to models that incorporate alternate error terms based on other error term distributions. Goodness of fit statistics suggested that our analyses best described the relationship between DO and structuring variables at the WMU/WMS scale. At this scale, water temperature, river stage height, current velocity, and to a lesser extent, water depth appeared to be the most important measured variables that influence DO levels. We suggest that management decisions at the WMU and WMS scales will offer the greatest potential benefits for improvement of DO conditions in the ARB.