Data consisting of samples of species' presences in association with measurements on a set of environmental variables can be used to determine environmental factors separating the species. It the multiple discriminant model is modified by a covariance extraction of time effects applied to the within—species, and total, deviation squares and cross products matrices prior to the discriminant analysis, then temporally varying environmental parameters can be included. If the distribution of sampling in space is consistent over time, then factors separating the species in time, as well as in space, can be determined; if it is not, then separation in space can still be determined even if the samples were collected at different times. The multiple discriminant model is analogous to the Hutchinsonian niche model; its use is illustrated with artificial data, and it is then applied to data from a benthic stream community to demonstrate heterogeneity of niche sizes, and separation of species' niches in space and in time by different environmental factors related to substrate type and water depth. Trophically similar species are more environmentally separated than are trophically different species; the separation is spatial for herbivore—detritivores, and temporal for carnivores.