NEW USES FOR OLD DATA: IMPACT OF ANALYTICAL METHODS AND PALEOCOLOGICAL DATA ON INTERPRETATIONS OF EVOLUTIONARY MODE OF MIOCENE CHESAPEAKE GROUP MOLLUSC

Kelley’s 1979 PhD thesis was one of the first studies designed to test punctuated equilibria (PE) by determining the relative frequency of stasis and gradualism for all sufficiently abundant mollusc lineages in a stratigraphic sequence (Maryland Miocene Chesapeake Group). Traits were chosen to quantify a range of aspects of morphology (size, shape, hinge, ornament, internal anatomy) and bivariate methods were used to determine temporal changes in growth patterns by regressing each trait against shell length at each stratigraphic level. Patterns were interpreted as consistent with PE in 82% of cases; canonical variate analysis of the data yielded mixed results, but stasis still dominated patterns. Kelley (1989) and Kelley & Hansen (2001) applied the same methods to traits useful in bivalve defense against shell-drilling predation and found that 42% of predation-related characters showed gradual change. In contrast, traits useful in species discrimination overwhelmingly (91%) were characterized by PE.

We reanalyzed the results of Kelley & Hansen (2001) using the paleoTS framework developed by Hunt (2006) to fit statistical models of evolutionary dynamics (directional evolution, unbiased random walk = URW, and stasis) to Kelley’s dataset. We compared these models to one that incorporated naticid gastropod drilling frequency normalized to shell length as a covariate. Model ranking techniques based on the Akaike Information Criterion (AIC) were used to assess the total empirical support each model received.

In contrast to the 2001 results, no support for directional evolution was found, even among traits linked to predation defense. Instead, 54% and 36% of cases are best fit by the URW and stasis models, respectively, with the covariate model best supported in 10% of cases. Of traits useful in species discrimination, 15% were best fit by the covariate model, suggesting a possible role for ecology in evolution. However, none of the predation-related characters was best supported by the covariate model; 65% corresponded to URW and 35% to the stasis model. Results may differ from the 2001 study due to accessing additional models, effects of small sample sizes, and/or the fact that paleoTS ranks models using AIC instead of a hypothesis test against a null and thus provides a superior approach to comparing competing models.