RICHARD A. KRAUSE, JENNIFER A. STEMPIEN, MICHAŁ KOWALEWSKI, ARNOLD I. MILLER
PALAIOS 22 (1), 60-73, (1 January 2007) https://doi.org/10.2110/palo.2005.p05-122r
Images in the monographic literature represent an important but relatively untapped resource for paleontologists. In particular, they could provide vast amounts of body size data. It is possible, however, that images of specimens represent a biased sample of the fossil record. Thus, the quality of these data must be assessed before body size estimates from the literature can be used in analyses.
Two complementary datasets were constructed for a group of bivalve and brachiopod species from the Paleozoic and the Cenozoic. The monograph dataset consisted of length measurements taken from all unique images of a species in a monograph. The counterpart bulk dataset consisted of comparable measurements taken from a set (n > 10) of bulk-collected specimens of the same species acquired from the same locality as those figured in the monograph. These paired datasets were used to assess the quality of monographic data.
Bias direction and magnitude were assessed by using the bulk sample of a species as an estimate of its underlying size-frequency distribution. Bias was estimated for each monographed specimen by calculating its percentile-value in relation to the size-frequency distribution for that species. All species groups had mean values within the 70th to 85th percentile range, indicating a significant bias toward monograph specimens that are larger than the mean of the bulk sample. The consistency of bias was evaluated by comparing the monograph sample mean to the bulk sample mean for each species. When compared in bivariate scatter plots, all species groups yielded significant regression lines with slopes near unity, indicating highly consistent, yet predictable, bias in each case. This trend persisted when the data were grouped taxonomically, geographically, or by year of monograph publication.
These results indicate that size measurements of monographed specimens of bivalves and brachiopods consistently record similar size classes for most species. This bias is easy to remove and doing so renders size data from images in monographs useful for macroevolutionary studies of body size.