EXPLORING MULTIDIMENSIONAL SPACE: TESTING THE INTERPRETATION OF ENVIRONMENT AND MULTIVARIATE ANALYSES USING DEVONIAN BRACHIOPOD COMMUNITIES IN NEVADA

Examining variations between paleocommunities is important in interpreting both biotic and abiotic processes, and multivariate statistics are often applied to this problem. Researchers use ordinations to delineate communities in multidimensional space, and the distribution of samples along the first axis is often interpreted to be controlled by an environmental factor such as depth. Environmental gradients can drive distributions and explain the patterns observed in ordination space, however direct tests of the validity of these interpretations are rare. Using high resolution brachiopod community data collected and published by J. G. Johnson in the 1990 Journal of Paleontology from the Lower and Middle Devonian of Nevada, the patterns observed in ordination space are compared with Johnson's interpretations of environment based on sedimentologic and geographic data. If the conventional assumption that environmental conditions control the distribution along the first axis is correct in this case, then Johnson's framework should be reflected in the distribution of communities in ordination space. Three of the faunal intervals from different epochs in Johnson's data set included abundant community samples along environmental gradients perpendicular to shore, from relatively deeper water to near shore. After using different distance measures and mapping on community and environmental information, the different community types described by Johnson are preserved, however the interpreted offshore to onshore gradients are not reflected by the order of the pattern. There are two explanations for this variance between observed patterns and inferred environments: 1) the environmental gradient is not one of the main determining factors in community distribution, and 2) the environmental interpretations do not reflect environmental conditions influencing the brachiopod communities. These results do not invalidate either environmental interpretations or multivariate analyses; rather caution should be used in the interpretation of both patterns.