UC Berkeley

Dramatic climate changes are already occurring, yet their ecological impacts remain difficult to predict as we lack both a detailed understanding of how key indicator species will react as well as large-scale models that are ground-truthed using paleontological data. In my dissertation, I focused on these challenges. I examine how climate change not only changes species' size and range, but can elicit evolutionary responses in phenotype and alter population variation. Detailed reconstructions of species' paleontological ranges test predictions made by species distribution models (SDMs) as well as their underlying assumptions. Population-level analyses of shifts in morphological variation through time allow the prediction of future evolutionary responses and the assessment of susceptibility to environment stress due to reduced natural variation. I focus on the how climate changes might affect California's ecosystems, developing quantitative methods to look at how morphological shape changes across space and through time in ecologically important small mammals, Microtus (voles). Looking back at their response to paleoecological transitions, I make predictions about how they might react in the future.

In Chapter 1, I use geometric morphometrics of vole dentition to establish a potential new paleoclimate proxy. First, I test if Microtus morphology correlates with paleoclimate signals, using an abundant California species, Microtus californicus. Geometric morphometrics (quantitative shape analysis) and partial least squares (PLS) analyses reveal geographic signals in the shape of the first lower molar (m1) of this species. M. californicus m1s are relatively straight in the northwest, cooler, moister portion of California and more curved in the southeast, hotter, drier portion of the state. These tooth shape changes may be a result of selection related to different vegetation ultimately controlled by climate, and therefore diet, within the species' range. The pattern in m1 shape persists when phylogeographic hypotheses are taken into account, indicating that the climate signal is significant independent of intraspecific groupings. This method reveals a geographic/climatic signal linked with morphological variation across the range of M. californicus and adds an important proxy to reconstruct past climates at fine spatiotemporal scales.

Another challenge addressed in my dissertation is that of species-level identification of fossils, which is necessary to maximally interpret the biostratigraphy, evolution, and paleoecology of Quaternary vertebrate localities. Microtus fossils typically are preserved only as isolated teeth, making identification difficult. In Chapter 2, I distinguish between the five species of Microtus living in California today (M. californicus, M. longicaudus, M. montanus, M. oregoni and M. townsendii) using geometric morphometrics on only their m1. Discriminant analysis on the resulting projected shapes correctly classifies extant specimens of known species 95 percent of the time, enabling vole identification from several important Quaternary fossil localities. In addition, I demonstrate the importance of using jackknife metrics for evaluating discriminant analyses and make some initial identifications that recognize the first extralimital fossil Microtus specimens in California.

Using the identification methods established in Chapter 2, I then trace range shifts in Pacific-coast Microtus species throughout the Quaternary. I reinterpret fossil localities that contain Microtus specimens and compare Quaternary range change to patterns of species' range shifts over the last 100 years. I find range contractions in all five species examined, but individual congeners react differently and to different climate variables. Using these detailed range reconstructions based on newly identified fossils, I examine whether SDMs projected into the past can predict that past habitats would have been suitable for the fossils found therein. I find that nearly half of the extralimital fossil specimens, those fossils found outside the modern range of the species, are living in habitats in the past that were predicted by the SDMs. The species that are not predicted by the SDMs all have niche reconstructions that are strongly influenced by precipitation, leading to the hypothesis that paleoclimate reconstructions may be incorrectly modeling precipitation variables. In addition, I find that two species of Microtus reacted similarly to climate change earlier in the Quaternary as they have over the last 100 years. Although one species, M. longicaudus, has experienced much more rapid range contraction 100 years than previously in the Quaternary.

In my final chapter, I trace morphological variation in M. californicus m1s through time and across California over the last 45 ky. I compare the effects of climate change on this species at the population and species levels. I find that m1 shape is significantly correlated with mean annual precipitation in this species. As California has become increasingly arid, M. californicus appears to have lost some of the m1 variation that is correlated with high-precipitation climate regimes. Although, at the species level I have yet to see any evidence that climate change has resulted in a change in Quaternary range limits. However, future climate change is likely to stress the species to the extent that populations in increasingly wet or dry habitats may be extirpated.

Throughout my dissertation, I demonstrate the utility of using detailed paleontological data in examining the past and future effects of climate change. In being able to identify fossil specimens to the species level, we can begin to directly compare modern and paleontological ecology studies. Using those identifications, we can create accurate paleorange reconstructions that can be used to test predictive distribution models and explore the underlying factors maintaining range limits. Once a species is identified, high-resolution quantification of morphological features, such as teeth, using geometric morphometrics can allow us to trace shifts in variation through time. This allows us to see the history of a species variation so that we can determine whether it retains enough variation to react to future stress.