Many environmental variables have been suggested to explain species distributions. However, species distributions generally cannot be evaluated statistically because of inadequate sampling procedures. The purpose of the present study is to outline a method of analysis using statistical models.

A pilot study in the Choptank River, Maryland, indicated that Elphidium clavatum density decreases progressively upstream, whereas Ammobaculites exiguus and Ammonia beccarii densities change very little. From this pilot study, three stations were selected for detailed analysis.

Four foraminiferal samples were taken monthly at each station for a year. Temperature, salinity, oxygen, and chlorophyll a, b, and c were measured each month at every station. A general multiple regression-analysis of variance model was constructed containing 21 parameters for environmental variables, station differences, overall periodic differences, and interaction of station and periodic differences. This model was compared with several containing fewer parameters. Restricted models containing 15 parameters sufficiently accounted for observed species densities in each example.

The set of environmental variables is significant at the 95% level for all species, but none is significant individually. However, relatively large values of regression coefficients for chlorophylls, especially b, suggest that food (amount and kind) is important in determining species densities. Each species exhibits periodicity, and for each, periodicity differs at the three stations.

Results indicate that the use of statistical models permits greater understanding of relations between foraminiferal species and environments. Such understanding will be of great value in paleoecology.

End_of_Article - Last_Page 710------------