UNIVARIATE ANALYSIS OF ECOPHENOTYPIC VARIATION: BACK FROM THE CUTTING EDGE

Techniques of multivariate morphometrics are used widely to assess morphological variation among fossil organisms both stratigraphically and geographically. Unfortunately, the computation of multivariate statistics is decades ahead of the interpretation. Because of the complexity of the assumptions made regarding multivariate data, rigorous hypothesis testing is rarely possible. Instead, the morphometric methods are used typically to provide numbers to enable the investigators to prepare a graphic representation of their results. On the other hand, hypothesis testing with univariate statistics is straightforward. Moreover, with careful experimental design one is not limited to deciding if morphology is different but can determine if it differs among environments so as to be biologically interesting. Geographic variation in morphology that is the result of response to environment can be evaluated with a model II or mixed-model analysis of variance. The ostracode species Sansabella bolliaformis (Ulrich & Bassler, 1906) is marked by a terminal adult stage that is morphologically quite distinct from its A-1 instar. An analysis of variance model assesses morphological differences within localities, among localities within environmental tracts, and among environmental tracts. It suggests biologically significant morphological variation within this species among such ancient, nearshore environmental tracts as tidal flat, lagoon, carbonate shoal, and open sea.