Effects of Antecedent Streamflow and Sample Timing on Trend Assessments of Fish, Invertebrate, and Diatom Communities

Abstract

Detecting trends in biological attributes is central to many stream monitoring programs; however, understanding how natural variability in environmental factors affects trend results is not well understood. We evaluated the influence of antecedent streamflow and sample timing (covariates) on trend estimates for fish, invertebrate, and diatom taxa richness and biological condition from 2002 to 2012 at 51 sites distributed across the conterminous United States. A combination of linear regression and Kendall-tau test for trends were used to evaluate covariate influence on trend estimates. Adjusting for covariates changed the magnitude of trend estimates in two-thirds of cases on average by 21%, most often reducing the estimated magnitude of the trend. Additionally, covariates influenced the interpretation of over one-third of trend estimates by either strengthening or weakening trends after adjustment. Our findings clearly indicate that antecedent streamflow and sample timing influences trend estimates and subsequent interpretation. Accounting for covariates during trend analysis will enhance stream monitoring programs by providing a better understanding and interpretation of estimated changes in biological endpoints at monitored sites. Failure to account for antecedent streamflow and sample timing may lead to mischaracterization of a trend and/or misunderstanding of potential causes.