Effect of taphonomy and methodological decisions on inferences of spatio temporal distribution of molluscan assemblages and its paleobiological implications

Abstract

Marine biodiversity changes through time and space. Identifying the drivers of such change is becoming especially important in the context of recent anthropogenic biodiversity loss. Shallow marine molluscan assemblages have long been recognized as good indicators of overall marine biodiversity and the health of the ecosystem at a regional scale. Their long geologic span and abundance in the fossil record also make useful diversity indicators of past ecosystems. Their complex ecosystem and durable shells enable their fossil record to be a reliable indicator of ecological interactions including predation and competition. A comparison of live assemblage (LA) and time-averaged death assemblage (DA) also provides important ecological insight into the changes in the molluscan community through time. Before one can use these signals for inferring spatio-temporal patterns from molluscan fossil assemblage, however, it is important to recognize that various taphonomic and methodological artifacts can potentially affect the accurate ecological signal. In this thesis, I tried to assess the influence of taphonomy and methodological decisions (such as sampling protocol, analytical method, and data categorization) on ecological inferences using time-averaged molluscan death assemblages. Using statistical modeling, I also proposed ways to recognize such influences and account for them. The first research problem explored how the degree of spatial live-dead similarity of an assemblage (spatial fidelity) is affected by the degree of post-mortem transportation in a tropical marine setting with a high sedimentation rate and high frequency of storms. Shells can be transported both within and out-of-habitat depending on the energy conditions of the surrounding habitats. Largescale mixing is more common in siliciclastic settings with a narrow shelf, high sedimentation rate, and those that are frequented by episodically high-energy events. By studying the live-dead (L-D) fidelity and modeling size-frequency distribution