UC Berkeley

Most of our understanding about plant evolution comes from the study of angiosperms.Nevertheless, flower-free plants display an impressive diversity of morphologies, physiologies, and life-strategies that are not present in angiosperms. Bryophytes, in particular, differ in fundamental ways from angiosperms; notably they experience life as haploids, match their water content to that of the environment, and disperse through unprotected spores—like most ferns but in great contrast with seed plants. A deeper understanding of ‘the bryophyte way’ of being a plant and its evolutionary implications will boost our ability to understand, model, and interpret diversity patterns across all land plants. In this dissertation, I explore different aspects of liverwort evolution, arguably the least studied among the three lineages of bryophytes. In Chapter 1, I report the de novo assembly of a nuclear reference genome for Calasterella californica, a West Coast endemic liverwort. The assembly consists of 820 contigs with a coverage estimated at 41x. The BUSCO score was estimated at 95% which indicates a good completeness. This assembly constitutes only the fifth nuclear reference genome for liverworts, a lineage of ∼ 7200 species. The genome of C. californica is 520 Mbp long, almost double the size of M. polymorpha’s genome, but similar in size to Lunularia cruciata’s. A comparison of C. californica’s genome to M. polymorpha’s genome suggests that C. californica has eight autosomes and one sexual chromosome as previously suggested. The reference genome was obtained from a female liverwort, thus carrying a U chromosome. In Chapter 2, I report the assembly of the chloroplast genome of C. californica and use a dataset on the arrangement of coding regions in the chloroplast of liverworts to explore the potential of using genomic arrangement information for phylogenetic inference. The final chloroplast of C. californica is 122,592 bp long, and the assembly had a coverage of 1978 reads per bp. This plastome contains a total of 129 genes, including rRNAs and tRNAs. Overall, the chloroplast gene arrangement is very conserved across the 41 species of liverworts compared in this study. The most common differences observed, were the absence of genes or introns. There were no gene order rearrangements in this dataset. When using the structural data to infer phylogenetic relationships, I obtained topologies that, regardless of the inference method, on one hand recover some large groups like complex thalloid liverworts, but on the other hand disagree with our current understanding of liverwort phylogenetics for other clades. And finally, I found that structural changes (i.e., gene or intron loss) occur at a slower rate than nucleotide substitutions. But other structural changes, like changes on the type of tRNAs, have rates of change comparable to that of nucleotides. In Chapter 3, taking advantage of the newly generated reference genome, I use a genomic dataset to study the genetic diversity and structure of C. californica across its distribution range in California. C. californica occurs across the state and it is notably absent in the Central Valley; in fact the Coastal individuals are genetically differentiated from the individuals occurring in Sierra Nevada ranges. Additionally, the individuals occurring in the Sonora Desert appear to be the most genetically differentiated group. Overall, the Southern part of California is the region with the most diverse genetic pool. Finally, in Chapter 4, I develop a biogeographic model of evolution, PAw GeoSSE + J, that accounts for many of the challenges associated with studying an old, widespread, and vagile lineage like many liverworts. I apply this new model on a dataset of Aytoniaceae, the family that C. californica belongs to. Overall, the precision with which we can reconstruct the biogeographic history of Aytoniaceae is limited. But my results suggest that Aytoniaceae most likely originated in North America and became widespread during the Cretaceous, a time when continent configuration facilitated dispersal. The analysis suggests that during the Mesozoic, lineages of Aytoniaceae occupied Antarctica despite the fact that no extant Aytoniaceae occurs there. This inference is derived completely from the connectivity matrices that reflect the reachability of Antarctica during that period of time. Furthermore the parameter estimates obtained from the PAw GeoSSE + J analysis are biologically meaninful, suggesting for example that the rate of dispersal to adjacent regions was 10 times higher than the rate of dispersal to distant regions, and the rate of cladogenetic events during long distant dispersal was three times higher than during adjacent-dispersal.