The dissertation research presented herein addresses two questions on possible ecological drivers of mammalian diversity dynamics at macroevolutionary time scales. The first question is whether key intrinsic biological traits that are tightly correlated with body size (e.g., reproductive rates) have strong influence on the extinction probability of mammalian taxa at the generic level. The second question is whether, within a regional mammalian fauna, the ecological composition of carnivores (as inferred from their dental morphology) responds in predictable manners to shifts in the ecological composition of non-carnivores that represent their potential prey.
In preparation for the ecological analyses of carnivore and non-carnivore compositional changes through time, concentrated effort was made to advance the taxonomy of carnivorous mammals from the middle Eocene of southern California. As part of this effort, the first chapter describes a carnivoramorphan that sheds a new light on the origin and early evolution of crown-group carnivorans. The new taxon, Lycophocyon hutchisoni, exhibits stages of dental and basicranial evolution that are intermediate between earlier carnivoramorphans and the earliest representatives of canoid carnivorans. The evolutionary affinity of the new taxon was determined by a cladistic analysis of previously-published and newly-acquired morphological data for 30 Paleogene carnivoramorphans. The most-parsimonious trees identified L. hutchisoni as a basal caniform carnivoran, and placed (1) Tapocyon robustus, Quercygale angustidens, "Miacis" sylvestris, "M." uintensis, and "M." gracilis inside or outside the Carnivora, (2) nimravids within the Feliformia, and (3) the amphicyonid Daphoenus outside the crown-group Canoidea. Parsimony reconstructions of ancestral character states suggest that loss of the upper third molars and development of well-ossified entotympanics that are firmly fused to the basicranium (neither condition is observed in L. hutchisoni) are not associated with the origin of the Carnivora as traditionally thought, but instead occurred independently in the Caniformia and the Feliformia. A discriminant analysis of the estimated body weight and dental ecomorphology predicted a mesocarnivorous diet for L. hutchisoni, and the postcranial morphology suggests a scansorial habit. Thus, Lycophocyon hutchisoni illuminates the morphological evolution of early caniforms leading to the origin of crown-group canoids. Nevertheless, considerable uncertainty remains with respect to the phylogenetic origin of the Carnivora. The minimum date of caniform-feliform divergence is provisionally suggested to be either 47 million years ago or 38 million years ago, depending on the position of "Miacis" sylvestris within or outside the Carnivora, respectively.
The second chapter investigates the relationship between body size as a proxy for various intrinsic biological traits of key importance and extinction probability as measured by durations of genera in the fossil record. Preservation of mammalian diversity requires a concentrated effort to identify biological correlates of vulnerability to environmental perturbations. Studies of living mammals and late-Quaternary extinctions frequently point to large body size as a correlate--if not necessarily a determinant--of elevated extinction risk in mammalian species, and this correlation is often attributed to slow reproductive rates and lower population densities of large taxa. At the same time, biological patterns of extinction risk above the level of species have received much less attention, despite their relevance to conservation of evolutionary history embedded in ecological types that are more inclusive than individual species. I examined the North American fossil record of modern and some extinct families of terrestrial mammals to test whether extinction probability (or, more precisely, inter-regional extirpation probability) of genera, as measured by their durations in geologic time, scales with body size across 7 orders of magnitude in body weight. After adjusting observed generic durations for significant paleontological sampling bias against small taxa, generalized least-squares regression analyses showed no correlation between estimated body weights and durations in 221 Oligo-Pleistocene genera ranging from shrews to rhinoceroses. The same lack of correlation was observed for subsets of the data that (a) approximated basic trophic divisions (small/large herbivores, "insectivores," and carnivores) or (b) were grouped by the timing of extinctions, suggesting that the overall pattern is not clouded by trophic and temporal variations in the relationship between size and vulnerability. The only notable deviation from this pattern was the significantly shorter durations of carnivorans compared to other taxonomic/ecological groups. Qualitatively identical results were obtained by analyses of durations and inter-birth interval lengths expected from body weights. Thus, in general, the population-biological expectation of higher extinction risk for large and slow-reproducing mammals was not supported for the genera that lived prior to significant anthropogenic influence. Two non-exclusive hypotheses are offered to explain this apparent mismatch: (1) the size-biased extinctions since the late Quaternary and elevated extinction risk for living large mammals signify an abnormal state of diversity dynamics brought about by human-induced reduction of large-mammal populations to critical levels, below which demographic or environmental stochasticity alone can threaten slow-reproducing taxa of low population density; (2) large mammalian species indeed have higher probabilities of extinction, but replacement of lost species within genera compensates for this pattern, resulting in comparable durations of large and small taxa at the genus level. The corollary of the first hypothesis is that, in normal times, thriving large mammals are no more likely--and perhaps less likely--to reach the critical population size than small mammals. The second hypothesis, if true, would indicate that extinction processes are distinct across levels of phylogenetic hierarchy and that prediction of future extinctions at supraspecific levels should not simply rely on extrapolation of extinction risk for individual species, especially if some of the species constituting a genus of interest are poorly known.
Building on the taxonomic work on the middle-Eocene carnivores of southern California, I investigate in the final chapter the matches and mismatches between shifts in the ecological compositions of mammalian carnivores and other mammals that constitute their potential prey at the macroevolutionary time scale of approximately 6-9 million years. The middle-Eocene fossil record of southern California, which includes a diverse array of carnivores and particularly rich record of small mammals, was analyzed. Appearance event ordination was used to estimate the relative ages of fossil-bearing localities and their associated assemblages. Using the predicted temporal ranges of carnivore taxa and locality-level occurrence data for non-carnivores (ultimately grouped into time bins), it was found that changes in the distribution of taxa, and possibly taxonomic abundance as well, across morphological categories (defined by estimated body weight, arboreal versus non-arboreal habit, and ecologically-informative dental morphology) are largely discordant between carnivores and non-carnivores in the study system, except for the overall increase in the number of taxa in both groups. Analysis of morphological-compositional variation and factors that correlate with taphonomic disparity lend support to the interpretation of observed diversity fluctuations through time in non-carnivores. The findings raise additional questions about the controls of carnivore diversity--for example, what promotes the appearance of new morphotypes--and predictability of their extinctions.
