The term "threshold" appears as a key word in more and more papers. However, there are misunderstandings of the few statistical methods for quantifying a threshold. Often, people seek for a one-size-for-all type of models with an option of producing the magic p-value. With several papers documenting statistical models for detecting and quantifying environmental and ecological thresholds, the questions is how to select the right model.
A statistical model for detecting a threshold comes with assumptions, just like a simple linear regression model. These assumptions may or may not fit the data in question and residual plots are always the best vehicle for evaluating these assumptions.
