Understanding the dynamics of trends within evolving lineages

The study of evolutionary trends is one of the oldest and most intriguing topics in evolutionary biology and paleobiology (McNamara 1990). Workers since Cuvier, Lyell, and Owen have wanted to know if the fossil record demonstrates “progression” within temporal sequences of related organisms. Regardless of whether changes in the average values of morphological characters are progressive in any meaningful sense, these changes are still of great interest. In practice, questions about trends are most commonly framed by paleontologists in terms of “complexity” (however defined) or body size (McShea 1998a). Research on evolutionary trends has intensified over the last few years, bringing several fundamental conceptual issues to a head. Here I analyze these conceptual problems, concluding that paleontologists should largely abandon a key method used in most studies: comparing among-species morphological distributions in successive time slices, irrespective of phylogenetic patterns. These data supposedly may distinguish random evolution, constant directional trends, and diffusion away from morphological boundaries. However, many other simple evolutionary dynamics may result in the same nonrandom trends, making it difficult to distinguish qualitatively distinct mechanisms using time-slice data. Thus, I will try to show that comparisons between ancestral and descendant morphologies need to be made instead. The fundamental problem here is not so much mathematical or statistical as conceptual. Null hypotheses have not always been framed rigorously, and the logical connection between underlying hypotheses and methods for testing these hypotheses has sometimes been weak. Although paleontologists have long since embraced the use of null models, they still do not employ a single, straightforward definition of “randomness” (Eble 1999). As a result, different authors use different null hypotheses, and depending on their conceptual outlooks they may interpret the same kinds of patterns as either confirming or refuting the existence of trends. Considering the volume of literature and the century-long …