Imagine a set of territories each subdivided into 5m grid cells. Each grid cell represents a spatial unit, and that unit can have any number of variables associated with it - e.g. habitat type/subtype, slope/aspect, management status, etc. We are interested in analyzing time since fire (TSF) effects for a species that is fire dependent. We have long-term datasets, but the most complete is simply the TSF at each point at a given point in time. So for example we have a TSF for each cell in each territory each nesting season. Territories are mapped and we can thus describe the TSF for a territory. Now, imagine one territory has a mean (across all cells) TSF of 10, and a St.D. of 0: The whole territory burned 10 years ago. Another has a mean of 10, and a St.D. of 20, thus has a much more varied fire history influencing demography of the species in question.
My question is how would you quantify the differences in TSF across these and other territories for each year? Because the means and St.D. will vary greatly, would it be potentially appropriate to use (in a linear model) an interaction between mean and variance as a metric: e.g. Y = mean + StD + mean*StD? Other ideas?
Although there may be other data available, assume for the moment that we only have the TSF data. Thanks for any suggestions or thoughts.