UC Berkeley

Abstract Comparative Biogeography of Dune-Restricted Insects in the Desert Southwest by Matthew H Van Dam Doctor of Philosophy in Environmental Science, Policy and Management University of California, Berkeley Professor Kipling W. Will, Chair

The focus of my thesis is on the biogeography of dune-restricted insects, there are three main inquiries I addressed in my thesis: (1) Testing the effect distance may have on dispersal in the sand dunes fauna of the desert Southwest of North America. (2) A statistical evaluation of phylogenetic signal contained in different character systems in Trigonoscuta, using novel reproductive characters to delimit species. (3) Incorporating more precise natural history data into niche modeling and an evaluation of its effect on adaptation to climate change in the giant flower-loving flies Rhaphiomidas.

(1) I am specifically testing the hypothesis that Pleistocene river corridors mediated dispersal among sand dune-restricted taxa and aquatic taxa found in desert springs. From the seven different genera of dune restricted taxa and four aquatic desert taxa I examined, it appears that some taxa have used river corridors of sandy habitat to disperse. Although the patterns are largely concordant they differ markedly in dispersal times. Some having long distance dispersal occurring in the Pleistocene while others only disperse at the local level having disjunct distributions the result of much older vicariance events. Using both the phylogenetic pattern and date estimates allows for my conclusions not to be drawn from pattern alone. In addition I examined the effect distance has on jump dispersal and vicariance. I incorporated GIS estimates of dispersal distance in a connectivity matrix to model the connectivity of river corridors. Model selection was then used to see if the constraints and distance effect improved the fit of the model to the data, indicating if there was an actual effect of the Pleistocene river corridors used as dispersal pathways. This application of patterns, distance and dates allows for more statistical testing of biogeographic hypothesis and has implications for biogeography as a whole to explicitly test the effect of distance on biogeographic reconstructions.

(2) The genus Trigonoscuta consists of 65 described species with 98 subspecies. I investigated the validity of these species using a combined approach of morphology and molecular data to be able to properly delimit and reliably identify the species. For delimiting the species I examined the morphology of the male endophallus. The endophallus is the internal structure of the male genitalia, which is everted during mating. I derived a novel technique to evert the endophallus consistently. The characters of the endophallus alone repeated the relationships seen in the molecular data and providedcharacters to delimit and describe species using morphological criteria. Using the consistency index (CI), I evaluated the contribution of each morphological character across all the equally most parsimonious trees. I then used this data to see if there was statistical support for which character system provided more phylogenetic resolution. This supported the idea that using the endophallus outperformed more traditional characters and should be utilized in phylogenetic reconstruction more often.

(3) I explored how partitioning climate data by day influences niche model predictions, as estimated by the MaxEnt machine-learning algorithm, for taxa with constrained phenologies. I utilized the giant flower-loving flies Rhaphiomidas because of their discrete phenologies as my focal taxon. I compared the results of partitioned by day data with WorldClim data that is partitioned by moth to examine what effect using more precise data has on species distribution models. I also examined how phylogenetic signal in both life history traits and climate tolerances can be used to identify how they adapt to different climates.