MODELING THE IMPACT OF TIME-AVERAGING ON THE ABILITY TO RESOLVE THE STRUCTURE OF PHENOTYPIC INTEGRATION IN THE FOSSIL RECORD

Patterns of phenotypic trait variation and covariation within a population indicate a structure to the raw material from which natural selection might draw. Developmental systems can be foundational to these (co)variance structures by biasing variation through phenomena such as canalization (homeostasis) and integration. The structure of developmental bias makes explicit predictions regarding directions of phenotypic disparification. Testing such predictions—and thus determining the extent to which developmental bias served as an evolutionary constraint—requires the comparison of empirical covariance matrices through time and/or across a phylogeny. In this respect, the fossil record is uniquely suited to understanding the role developmental bias may have on phenotypic evolution. Nevertheless, the fossil record presents challenges to such study. Time-averaging of evolving populations, for example, might blur or distort observed (co)variance structures relative to any “true” (instantaneous) structure. The extent to which such taphonomic overprint affects the ability to resolve evolutionary pattern is poorly known.

Here we present the results of a series of simulations designed to model the effects of time-averaging across a variety of scales and under different evolutionary regimes. Unsurprisingly, higher rates of evolution and degrees of time-averaging are shown to effect estimation of a (co)variance structure more than slow rates and minimal averaging. However, other more complex factors include the mode of evolution of the (co)variance structure, for example along a line of least resistance, by a random walk, or with varying degrees of integration and trait variation. The implications of our findings have meaningful impact on comparisons not only between fossil samples, but also against modern sampling. Time-averaging is not a desirable feature of any sample in an evolutionary study but understanding its impact on a variety of evolutionary rates and modes provides indications of how much of a factor it may play in the important work of detecting developmental bias in the fossil record.