Contemporary species are undergoing population declines and extinction at rates unprecedented in recorded history. These ongoing global biodiversity losses are largely caused by human overpopulation and other anthropogenic impacts on the environment such as natural habitat destruction driven by urbanization, deforestation, agriculture, pollution, overconsumption of natural resources, and climate change. Understanding how species are influenced by – and respond to – various changes in their environment is critical for predicting and mitigating future biodiversity loss. These predictions are challenging, however, because humans have been heavily modifying ecosystems for centuries – well before the advent of modern ecology as a field of study. Disentangling species responses to naturally occurring changes in their environment versus anthropogenic changes is thus extremely challenging.

Paleoecological studies of fossil organisms can help establish the baseline responses of biota to natural environmental changes at times before humans dominated terrestrial ecosystems. However, these studies have their own set of challenges. For example, it can be difficult to determine how representative a preserved fossil community is of the original living community because the fossil record is inherently incomplete and often biased. It is also difficult to quantify species-specific responses to environmental change if the identity of species is unknown or imprecise; and due to the fragmentary nature of the fossil record, it can be difficult to identify isolated elements to species. The incompleteness of the fossil record does not only apply to the organisms preserved, but also to the environmental data documenting the contexts in which they operated while alive and during preservation. Most paleontological assemblages are affected by time-averaging and incomplete depositional sequences to some degree. Depending on the severity of time averaging, and the resolution of data collected, these temporal gaps can erase fine-scale and geologically rapid events that are important for understanding ecological patterns and processes.

These unique opportunities and challenges of working with paleoecological data are what motivate my research. Within the scope of my dissertation, my goals are twofold. Foremost, I strive to quantify long-term biotic composition, diversity, and trait changes in response to pre-anthropogenic environmental change at population and community levels to establish baselines of organismal responses to natural ecosystem perturbations. However, to accomplish this, it is first necessary to quantify the strengths and limitations of paleontological data in these systems and maximize data resolution to mitigate erroneous interpretations. The main data types I focus on improving here are those of taxonomic fidelity and age control. The first three chapters of my dissertation focus on the former, using morphometric techniques to improve identification accuracy of closely related and morphologically similar species, thus extending paleoecological data resolution from genus to species for several taxa. The last two chapters of my dissertation focus on the latter, examining paleoecological data at various levels of temporal precision using a combination of radiocarbon-dated and time-averaged data to determine how analytical results and conclusions are affected by time-averaging. Once these limitations have been quantified and mitigated to the extent possible, I determine how the focal taxa of my study system were impacted by long-term environmental changes using multidisciplinary approaches. Chapter 3 focuses on intraspecific phenotypic responses to climate change using geometric morphometrics, Chapter 4 evaluates long-term changes in biotic community structure using diversity and trait metrics, and Chapter 5 quantifies the relative impacts of climate and biotic interactions on species niches over the last 50,000 years using stable isotope analysis. My study system for addressing all these topics is Rancho La Brea (RLB), a world renowned late Quaternary paleontological locality in Los Angeles, California, USA. I specifically examine the small mammals (e.g., rodents, lagomorphs, and soricomorphs) of this locality because they are ubiquitous across most Quaternary fossil assemblages, thus facilitating large sample sizes. In addition, small mammals are generally short lived and confined to small home ranges, so I am relatively certain that the paleoecological signals I track within samples are local and geologically instantaneous rather than substantially spatially or temporally averaged.

Results of the three taxonomic studies indicate that, although closely related and speciose small mammals are difficult to differentiate due to morphological variation and overlap, they can be identified to species with relatively good accuracy, usually > 80%, using quantitative techniques including morphometric and geometric morphometric measurements and statistical grouping analyses (Chapters 1-3). However, results can deviate considerably if data acquisition processes are not standardized. For example, geometric morphometric data collected by different personnel and, to a lesser extent, with different instruments can generate substantially different classification statistics (Chapter 2). It is therefore recommended that data acquisition procedures are standardized as much as possible to facilitate analytical replicability. Comparisons of time-averaged trait datasets (Chapters 4 and 5) to those with good age control (Chapter 5) further show that much information can be lost from geologically rapid events when data is time-averaged or time-binned versus continuous data. Such loss of information can then result in profoundly different interpretations regarding the probable drivers of observed paleoecological patterns (Chapter 5).

With these insights and limitations in mind, I show that local environments of RLB during the last glacial period (specifically Marine Isotope Stage (MIS) 3, ~60,000 to 29,000 years BP) were generally similar to that of the Los Angeles Basin today based on overall similarities between contemporary and fossil small mammal faunas (Chapter 4). Changes in taxonomic abundances and trait diversity among deposits of different mean ages suggest that the small mammal communities of RLB were responding to slight or moderate changes in temperature and precipitation during that time (Chapter 4). Unfortunately, precise information on the timing and pattern of environmental changes cannot be discerned at the community level due to the time-averaged nature of the deposits and faunas examined, combined with the variable climates during MIS 3. By subsequently examining the isotopic niches of individually-dated specimens, however, it becomes clear that geologically rapid environmental changes were occurring at RLB throughout the late Quaternary that largely reflect regional climate patterns (Chapter 5). Further, the isotopic niches of small mammals appear to be shaped more strongly by those climatic oscillations than by biotic interactions over the last 50,000 years. Insights on the paleoenvironments of RLB (Chapter 4) and climatic changes that likely occurred there during the late Quaternary (Chapter 5) have significant implications for studies of other RLB biota in that species responses to changing environments can be better contextualized now that those changes are better understood. In a broader context, my work quantifying geometric morphometric error (Chapter 2) and time-averaging error (Chapter 5) may facilitate best practices protocols for similar study systems. Finally, my taxonomic identification protocols for lagomorphs (Chapter 1) and woodrats (Chapter 3) should be useful for other small mammal studies because lagomorph remains are common at most late Quaternary sites and woodrat species are good indicators of paleoecological conditions and change.

contributors