In the first three chapters, we considered some distributions commonly encountered in ecological studies, models that give rise to those distributions, and methods of determining the fit of a proposed distribution to the data. One of the models giving rise to the Poisson distribution assumes that each member of the species behaves independently of every other member of that species and that the probability that a species member is in a sampling unit is the same for all sampling units. This requires that each member of a species reacts with complete indifference to other members of that species or to the environment. This model is probably more appropriate for animal than for plant populations. For plant populations, a possible Poisson model assumes that every sampling unit contains a large number, n, of locations, each of which has the probability p of being occupied by a species member. Both models have very restrictive assumptions. Therefore, it is not surprising that the Poisson rarely provides a good fit to ecological data sets. At the same time, these models represent two cases that are not very biologically interesting. It is the interaction of plant and animals with members of their own species, with members of other species, and with their environment that makes ecology exciting. From this perspective, it makes sense first to determine whether the population departs from this most sterile case and then to describe the nature of those departures.
KeywordsSpatial Correlation Statistical Ecology Sampling Unit Wheat Yield Robust Estimator
Unable to display preview. Download preview PDF.