Disentangling historical relationships within Poeciliidae (Teleostei: Cyprinodontiformes) using ultraconserved elements.

Poeciliids (Cyprinodontiformes: Poeciliidae), commonly known as livebearers, are popular fishes in the aquarium trade (e.g., guppies, mollies, swordtails) that are widely distributed in the Americas, with 274 valid species in 27 genera. This group has undergone various taxonomic changes recently, spurred by investigations using traditional genetic markers. Here we used over 1,000 ultraconserved loci to infer the relationships within Poeciliidae in the first attempt at understanding their diversification based on genome-scale data. We explore gene tree discordance and investigate potential incongruence between concatenation and coalescent inference methods. Our aim is to examine the influence of incomplete lineage sorting and reticulate evolution on the poeciliids' evolutionary history and how these factors contribute to the observed gene tree discordace. Our concatenated and coalescent phylogenomic inferences recovered four major clades within Poeciliidae. Most supra-generic level relationships we inferred were congruent with previous molecular studies, but we found some disagreements; the Middle American taxa Phallichthys and Poecilia (Mollienesia) were recovered as non-monophyletic, and unlike other recent molecular studies, we recovered Brachyrhaphis as monophyletic. Our study is the first to provide signatures of reticulate evolution in Poeciliidae at the family level; however, continued finer-scale investigations are needed to understand the complex evolutionary history of the family along with a much-needed taxonomic re-evaluation.

Multi-model approaches to phylogenetics: Implications for idealization.

Phylogenetic models traditionally represent the history of life as having a strictly-branching tree structure. However, it is becoming increasingly clear that the history of life is often not strictly-branching; lateral gene transfer, endosymbiosis, and hybridization, for example, can all produce lateral branching events. There is thus motivation to allow phylogenetic models to have a reticulate structure. One proposal involves the reconciliation of genealogical discordance. Briefly, this method uses patterns of disagreement - discordance - between trees of different genes to add lateral branching events to phylogenetic trees of taxa, and to estimate the most likely cause of these events. I use this practice to argue for: (1) a need for expanded accounts of multiple-models idealization, (2) a distinction between automatic and manual de-idealization, and (3) recognition that idealization may serve the meso-level aims of science in a different way than hitherto acknowledged.

Phylogenomics of the world's otters.

Comparative whole-genome analyses hold great power to illuminate commonalities and differences in the evolution of related species that share similar ecologies. The mustelid subfamily Lutrinae includes 13 currently recognized extant species of otters,1-5 a semiaquatic group whose evolutionary history is incompletely understood. We assembled a dataset comprising 24 genomes from all living otter species, 14 of which were newly sequenced. We used this dataset to infer phylogenetic relationships and divergence times, to characterize patterns of genome-wide genealogical discordance, and to investigate demographic history and current genomic diversity. We found that genera Lutra, Aonyx, Amblonyx, and Lutrogale form a coherent clade that should be synonymized under Lutra, simplifying the taxonomic structure of the subfamily. The poorly known tropical African Aonyx congicus and the more widespread Aonyx capensis were found to be reciprocally monophyletic (having diverged 440,000 years ago), supporting the validity of the former as a distinct species. We observed variable changes in effective population sizes over time among otters within and among continents, although several species showed similar trends of expansions and declines during the last 100,000 years. This has led to different levels of genomic diversity assessed by overall heterozygosity, genome-wide SNV density, and run of homozygosity burden. Interestingly, there were cases in which diversity metrics were consistent with the current threat status (mostly based on census size), highlighting the potential of genomic data for conservation assessment. Overall, our results shed light on otter evolutionary history and provide a framework for further in-depth comparative genomic studies targeting this group.

A multispecies coalescent model for quantitative traits.

We present a multispecies coalescent model for quantitative traits that allows for evolutionary inferences at micro- and macroevolutionary scales. A major advantage of this model is its ability to incorporate genealogical discordance underlying a quantitative trait. We show that discordance causes a decrease in the expected trait covariance between more closely related species relative to more distantly related species. If unaccounted for, this outcome can lead to an overestimation of a trait's evolutionary rate, to a decrease in its phylogenetic signal, and to errors when examining shifts in mean trait values. The number of loci controlling a quantitative trait appears to be irrelevant to all trends reported, and discordance also affected discrete, threshold traits. Our model and analyses point to the conditions under which different methods should fare better or worse, in addition to indicating current and future approaches that can mitigate the effects of discordance.

Population delimitation across contrasting evolutionary clines in deer mice (Peromyscus maniculatus).

Despite current interest in population genetics, a concrete definition of a "population" remains elusive. Multiple ecologically and evolutionarily based definitions of population are in current use, which focus, respectively, on demographic and genetic interactions. Accurate population delimitation is crucial for not only evolutionary and ecological population biology, but also for conservation of threatened populations. Along the Pacific Coast of North America, two contrasting patterns of geographic variation in deer mice (Peromyscus maniculatus) converge within the state of Oregon. Populations of these mice diverge morphologically across an east-west axis, and they diverge in mitochondrial DNA haplotypes across a north-south axis. In this study, we investigate these geographically contrasting patterns of differentiation in the context of ecological and evolutionary definitions (paradigms) of populations. We investigate these patterns using a new and geographically expansive sample that integrates data on morphology, mitochondrial DNA, and nuclear DNA. We found no evidence of nuclear genetic differentiation between the morphologically and mitochondrially distinct populations, thus indicating the occurrence of gene flow across Oregon. Under the evolutionary paradigm, Oregon populations can be considered a single population, whereas morphological and mitochondrial differentiations do not indicate distinct populations. In contrast, under the ecological paradigm morphological differentiation indicates distinct populations based on the low likelihood of demographic interactions between geographically distant individuals. The two sympatric but mitochondrially distinct haplogroups form a single population under the ecological paradigm. Hence, we find that the difference between evolutionary and ecological paradigms is the time-scale of interest, and we believe that the more chronologically inclusive evolutionary paradigm may be preferable except in cases where only a single generation is of interest.