Speciation is primarily regarded as an ancestral split that results in two distinct taxonomic units, and proceeds in stages along a continuum from initiation (i.e., population divergence) to completion (i.e., reproductively isolated species). Establishing how and why populations diverge, including the primary mechanisms influencing these events is a major objective for evolutionary scientists. Focusing on incipient forms, researchers attempt to disentangle the antagonistic nature of selection, genetic drift, and gene flow in the speciation process.
In chapter 1, I investigate the phylogenetic relationships of 14 closely related taxa within the mallard complex (Anas spp.) that underwent a radiation within the past one million years. Using mitochondrial DNA (mtDNA) and 20 nuclear loci for one to five individuals per taxon, I further examine how recombination and hybridization affect species tree reconstructions. In general, relationships within major clades were robust to treatment of recombination (i.e., ignoring or filtering) and inclusion or exclusion of hybridizing taxa, but branch lengths and posterior support were sensitive to both treatments. Of the 14 taxa, the most confounded relationships were those within the New World (NW) group comprising the sexually dichromatic mallard (Anas platyrhynchos) and the monochromatic American black duck (A. rubripes; "black duck"), mottled duck (A. fulvigula), and Mexican duck (A. [p.] diazi). Finally, I address discordance between nuclear, morphometric, and mitochondrial trees, particularly with regard to the placement of the Hawaiian duck (A. wyvilliana), Philippine duck (A. luzonica), and two spot-billed ducks (A. zonorhyncha and A. poecilorhyncha) and discuss how alternative modes of speciation (i.e., hybrid speciation) may lead to variance in these relationships.
In Chapter 2, I attempt to disentangle the evolutionary relationships of the New World (NW) group using mtDNA and 17 nuclear loci for a larger per taxon sample size (24-25 individuals per taxon). In general, whereas both Florida and Gulf Coast mottled ducks were differentiated from one another and from the other taxa (mean ΦST = 0.024 - 0.064), mallards, American black ducks, and Mexican duck were not significantly differentiated among nuclear markers (mean ΦST < 0.020). Using coalescent methods to estimate rates of gene flow between mallards and each of the monochromatic taxa generally supported hybridization, but I could not reject complete isolation for any pairwise comparison. Furthermore, species tree reconstructions revealed that phylogenetic relationships were sensitive to stochastic sampling of individuals likely due to incomplete lineage sorting or hybridization. I conclude that members of the NW Mallard group appear to be adaptive incipient morphs, and that future work should focus on genomic regions under selection to better understand the stage and process of speciation in this group.
In Chapter 3, I use restriction site associated DNA (RAD) sequencing methods to generate a pseudorandom sampling of 3,563 autosomal and 172 sex-linked (Z chromosome) markers scattered across the genome to more rigorously test the mechanism of speciation between Mexican ducks (N = 105 individuals from six Mexican states and two US states) and mallards (N = 17). Specifically, I aim to determine the stage of speciation and whether speciation has been driven by few loci with large effects versus many loci with small effects, plumage associated differentiation, or genetic drift. Marker comparisons between mallards and Mexican ducks revealed strong discordance among autosomal ΦST = 0.014), sex-linked (mean ΦST = 0.091), and mtDNA (ΦST = 0.12) markers. In general, divergence at autosomal loci followed a stepping stone model, with a gradual transition in genotypic frequencies from North to South. In contrast, Z-linked markers followed an island model of divergence, with a sharp transition in genotypic frequencies at the geographic boundary between mallards and Mexican ducks. In contrast, both autosomal (meanΦST = 0.012) and Z-linked markers (mean ΦST = 0.018) were tightly correlated among Mexican duck sampling groups. These results suggest that, whereas genetic drift is likely influencing structure among Mexican duck populations and between Mexican ducks and mallards at autosomal loci, selection is likely influencing Z-chromosome structure between Mexican ducks and mallards. The latter finding is consistent with the evolution of post-mating isolation between Mexican ducks and mallards. Finally, I report that contemporary hybridization with mallards is likely limited to the northern edge of the Mexican duck's range, and that those from inland Mexico appear to be "pure" and follow an isolation-by-distance model of divergence. In conclusion, these results suggest that mallards and Mexican ducks are at the earliest stages of parapatric divergence with the Z chromosome at a later stage - relative to autosomal chromosomes - of divergence, which is being driven by selection on few loci with large effects.
In Chapter 4, I test another mechanism of speciation - whether the Hawaiian duck evolved via hybrid speciation. Following from the results of Chapter 1, where I presented compelling evidence of mitochondrial-nuclear-morphological discord in the phylogenetic placement of this species, I sequenced a larger sample size of Hawaiian ducks (N = 15 individuals) and its putative parental species, the Laysan duck (A. laysanensis; N = 21 individuals) and mallard (N = 25 individuals). I demonstrated that the Hawaiian duck's genome was a mosaic of mallard (59%) and Laysan duck (41%) polymorphisms. Moreover, gene flow estimates revealed significant non-zero gene flow from the Laysan duck into the Hawaiian duck under a mtDNA-like topology (Hawaiian sister to mallard) or from the mallard into the Hawaiian-Laysan duck ancestor under a nuDNA-like topology (Hawaiian sister to Laysan). Thus, regardless of the tree topology used, gene flow from the non-sister species is necessary to explain extant genetic diversity in Hawaiian ducks, further supporting a genomic mosaic. This work is one of few well-supported cases for hybrid speciation in homoploid systems, and highlights the potential for such events on island systems where the hybrid descendants can become geographically isolated from the parental species.
In Chapters 1 and 4, I found no nuclear variation in Laysan ducks, which is a critically endangered species. Consequently, in Chapter 5, I developed a PCR-based protocol to examine diversity within the Major Histocompatibility Complex (MHC) I gene in Laysan ducks. Particular attention has been given to MHC genes due to their direct correlation to an individual's immunity. The haplotype-specific primers allowed for direct genotyping after gel electrophoresis based on the presence/absence of their respective amplicons. Using the developed techniques, a total of eight unique haplotypes were isolated and assayed across 21 Laysan duck individuals from Laysan Island (N = 10) and Midway Atoll (N = 11). The protocol provides a simple, cost-effective method for isolating haplotypes and monitoring existing MHC variation in Laysan ducks that can be implemented in admixture schemes within captive breeding programs to maximize heterogeneity prior to reintroduction.
In conclusion, divergence and speciation within the mallard complex has been driven by a number of mechanisms, including allopatric divergence, parapatric divergence, and hybrid speciation. These results demonstrate the value of multi-taxa, multi-marker comparisons in resolving complex evolutionary relationships. Furthermore, each chapter builds on previous chapters, illustrating the utility of addressing speciation from macroevolutionary scales (e.g., phylogenetics), which generate testable hypotheses, to progressively more microevolutionary scales for testing those hypotheses. Given their incipient stage and evolutionary heterogeneity, the mallard complex is an excellent system for studying the effects of various evolutionary mechanisms and demographies in the speciation process.