Multinomial regression for analyzing macroinvertebrate assemblage composition data
Macroinvertebrate species composition data are often expressed as proportional abundances when assessing water-quality conditions or responses to disturbance. Proportional abundances represent the probability of belonging to one of many mutually exclusive and exhaustive groups (taxa). Proportional abundances have some unique properties that must be considered when analyzing these data: 1) the probabilities of group membership must sum to 1 and 2) a change in any 1 group affects all other groups.