Quantifying Molluscan Body Size in Evolutionary and Ecological Analyses: Maximizing the Return on Data-collection Efforts

How does the choice of size metric, specimen selection, and taxonomic level affect the results of macroevolutionary or ecological analyses? Four molluscan data sets are used to address this question as follows. First, the relationships among various size metrics are examined using a morphometric data set of Late Cretaceous–Oligocene veneroid bivalves. Second, the relationship between the size of bulk-sampled specimens and the size of species’ type specimens is examined using bulk-sampled bivalves and gastropods from the Coffee Sand (Upper Cretaceous, Mississippi). Third, the same relationship is examined using mollusk-dominated field censuses from the type Cincinnatian (Upper Ordovician, Ohio, Indiana, and Kentucky). Fourth, the relationship between the size of the type species of a genus and median species size is examined using literature-derived measurements of bivalve type specimens from the recent eastern Pacific continental shelf. Together these data sets provide estimates of the biases imposed by measuring different kinds of specimens and using different methods of estimating body size. The geometric mean of length and height is highly correlated with more complex morphometrically based metrics and is our preferred bivalve size metric. Bulk or randomly sampled specimens are significantly smaller than species’ type specimens for the Cretaceous dataset but significantly larger for the Ordovician dataset. Genus’ type-species size is an unbiased estimate of median species size. These results suggest that large-scale studies can use the size of the type species of a genus as an unbiased proxy for a typespecimen size of a genus’ median species, but that species’ type-specimen size is a biased proxy for bulk or randomly sampled specimens. In addition, this study emphasizes the importance of using a single type of measurement within studies and suggests that combining multiple types of specimens (e.g., type specimens and bulk-sampled specimens) could lead to substantive errors.