Abstract

Many statistical models are available for detecting and quantifying ecological thresholds. The proper model can be selected through the comparison of the goodness-of-fit of multiple alternative models to the same data. In the context of selecting the appropriate threshold model, three sources of information should be used: patterns in model residuals, model predicted versus observed, and the posterior change point distributions. Applying alternative models to several ecological metrics measured along a total phosphorus gradient from a mesocosm experiment conducted in the Everglades, I conclude that changes in these metrics along the gradient are more likely to be gradual. The previously found strong threshold response is likely a result of the use of the step function model.

Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call