We introduce a generalized form of the common first-order (exponential) dec
ay model, that has potential utility for describing contaminant declines in
environmental applications, particularly when declines are a mixture of ma
ny underlying processes. The exponent on contaminant concentration is left
as a free parameter allowing the order of the reaction to be determined by
the data. The mixed-order model is more flexible than models with the expon
ent determined a priori, facilitating an improved fit to observed behavior.
We demonstrate the utility of this model, and compare it to two other mode
ls, by estimating PCB concentration declines in Lake Michigan fishes.