Species collisions, admixture, and the genesis of biodiversity in poison frogs

This is the authors' abstract. We don't add key points for this paper.

In a manner analogous to the use of atom collisions in the study of particle physics, admixture between divergent populations or species (i.e., genome collisions) can shed light on the fundamental nature of biological groups (e.g., species) and the genetic basis of barriers to gene flow between them (Buerkle & Lexer, 2008). And like atom collisions, which can result in both fusion and emission of novel particles, collisions between divergent genomes can both erode and create biological diversity. For example, whereas demographic or genetic swamping can result in loss of biodiversity, hybrid speciation can lead to gains. From a conservation standpoint, a better understanding of the causes and prevalence of these outcomes is necessary to identify and protect processes that maintain biodiversity and promote adaptive evolution. But the role of admixture in generating novelty becomes especially contentious when one of the hybridizing populations is a narrow endemic or critically endangered. In a From the Cover article in this issue of Molecular Ecology, Ebersbach et al. (2020) analzyed the evolutionary outcomes and conservation implications of hybridization between a critically endangered South American poison frog, Oophaga lehmanni (Myers & Daly, 1976), and its congener, O. anchicayensis (Posso-Terranova & Andres, 2018). Using trait data, reduced-representation genome sequences, and statistical population genomics, Ebersbach et al. (2020) identified three poison frog populations of hybrid origin. Frogs from these populations exhibit novel color pattern phenotypes and genomes that are mosaics of ancestry segments inherited from O. lehmanni and O. anchicayensis. Historical demographic inferences suggest that these admixed populations did not originate through recent, human-mediated translocations associated with the pet trade, but rather are of older origin and have persisted through time as isolated, admixed entities (i.e., taxa). This study highlights the role of admixture and subsequent evolution of hybrid genomes in the genesis of biodiversity, and adds support to the idea that divergence via admixture may be a somewhat common facet of the evolutionary process. Many early studies in ecological genetics focused on color pattern variation, and specifically on the evolutionary processes that maintain color polymorphisms in natural populations (Ford, 1977). This emphasis has continued to the present day, and we now know more about the genetic basis and evolution of color and color patterns than most other traits (e.g., Martin & Orgogozo, 2013). Color pattern polymorphisms often exhibit relatively simple, modular genetic architectures (e.g., Lindtke et al., 2017; Van Belleghem et al., 2017), and adaptive introgression of color pattern loci has been documented in multiple species (e.g., Dasmahapatra, Walters, & Briscoe, 2012). Harlequin poison frogs (genus Oophaga) exhibit particularly striking levels of variation in color and color patterns (i.e., spots and stripes), with different morphs being found at large and small (e.g., a few kilometers) spatial scales (Ebersbach et al., 2020) (Figure 1). Past work suggests that the frog's patchy distribution, drift, and natural and sexual selection likely contribute to the geographic mosaic of color pattern morphs, but the possible contribution of hybridization to this phenotypic diversity had not been thoroughly investigated until now. Ebersbach et al. (2020) take a population genomics approach to show that hybridization between O. lehmanni and O. anchicayensis gave rise to a series of admixed populations that exhibit novel color patterns (i.e., a novel combination of orange and red spots and stripes) (Figure 1). Specifically, principal components analysis (PCA) of 49 morphological variables (including color patterns), shows that the populations are phenotypically distinct from both parental species. Population genomic analyses with PCA-based methods and Bayesian admixture models (i.e., ADMIXTURE, fast STRUCTURE and introgress) confirm the hybrid origin of this series of populations. The genomic analyses suggest the admixed populations comprise late-generation hybrids (i.e., Fn), with a near or complete absence of early-generation hybrids (e.g., F1, BC1, or F2), and thus, that most matings are between hybrids with low rates of contemporary gene flow into the admixed populations. Bayesian genomic cline analysis further shows that the genomes of the hybrids are made up of a mosaic of ancestry segments originating from each species (i.e., segments of chromosomes from each species) (Gompert & Buerkle, 2011). For some regions of the genome, most of the hybrid frogs have ancestry segments from O. lehmanni (positive α outliers for genomic clines), whereas for other regions of the genome, most of the frogs have ancestry from O. anchicayensis (negative α outliers). Other genomic regions show evidence of restricted introgression, with ancestry segments confined mostly to alternative genomic backgrounds, and perhaps to different admixed populations (β outliers). Although these analyses do not directly demonstrate that the genomic mosaic of ancestry in the hybrids is responsible for the novel color patterns observed in the hybrid frog populations, this is certainly a plausible hypothesis and one put forward by the authors (Ebersbach et al., 2020). More generally, shuffling of parental allele combinations in admixed populations can readily create genetically distinct populations and phenotypic novelties (e.g., Elgvin et al., 2017; Rieseberg et al., 2003). This occurs because recombination and independent assortment in hybrids break up parental ancestry segments (i.e., chromosomes with parental gene combinations), and generate new gene combinations with alleles from each of the hybridizing lineages. These new gene combinations can result in transgressive phenotypes, that is, phenotypes beyond the range observed in either species. Whereas individuals with transgressive phenotypes can be observed in early-generation hybrids, entire populations exhibiting novel gene combinations and phenotypes can become established if gene flow between the admixed population and parental species declines or ceases. In such cases, recombination erodes the size of ancestry segments, and ancestry segments begin to fix by genetic drift or selection in a process known as genome stabilization (Buerkle & Rieseberg, 2008) (Figure 2). This can include selection against developmental incompatibilities or for adaptation to a novel biotic or abiotic environment (Schumer et al., 2018). The outcome of this process is a population or species of hybrid origin that is genetically divergent, and often phenotypically distinct, from each of the parental species. Past debates have considered whether or not populations arising through the process of hybridization and genome stabilization represent hybrid species, with the prevailing notion being that hybrid species must exhibit near complete reproductive isolation from the parent taxa and that reproductive isolation must have evolved as a direct consequence of hybridization (e.g., Schumer, Rosenthal, & Andolfatto, 2014). However, even in cases where other processes facilitate reproductive isolation after admixture, or when reproductive isolation and genome stabilization are incomplete, the outcome is nevertheless a genetically distinct population or group. This is especially true when the genomes of the hybrids include non-trivial contributions from each species (e.g., more than 10%), as the admixed population would necessarily differ, at least genetically and almost certainly in ecologically relevant phenotypes, from any non-admixed population. There is increasing evidence that such distinct, stabilized (or semi-stabilized hybrid populations) might not be uncommon and could represent a non-negligible component of biological diversity (regardless of taxonomic status or decisive evidence of hybrid speciation per se) (e.g., Ebersbach et al., 2020; Elgvin et al., 2017; Gompert et al., 2014; Gross & Rieseberg, 2005; Schumer et al., 2018). Such results are consistent with verbal models that suggest repeated bouts of geographic isolation and admixture can fuel adaptive radiations (Seehausen, 2004). This has implications for conservation efforts and policy, especially in cases like Oophaga where one or both hybridizing taxa are endangered (Ebersbach et al., 2020). Hybridization can unlock novel phenotypes not found in either parental population and increase the adaptive capacity of populations in changing environments (Hamilton & Miller, 2016). Despite this, a heavy focus on the “purity” of species prevails in government policy. A 2015 review of conservation policies in the U.S. and Canada found that only 16% provided guidelines for management of hybrids, and of these, 46% denied protection for individuals of hybrid origin (Jackiw, Mandil, & Hager, 2015). More importantly within the context of hybrid origins, the large majority of hybrid policies (77%) made no attempt to distinguish between levels of hybridization (i.e. F1 hybrids vs. backcrosses or early vs. late-generation hybrids) (Jackiw et al., 2015). This lack of nuance leaves policy administrators with few means for differentiating between the evolutionary potential of stable, late-generation hybrid populations (as in Oophaga; Ebersbach et al., 2020) vs. cases of ongoing, maladaptive genetic swamping. Furthermore, in cases where admixture is a frequent driver of diversification, yet species-level divergence remains low (e.g., Gompert et al., 2014), conservation policy may under-protect the biological diversity of entire taxa (vonHoldt, Brzeski, Wilcove, & Rutledge, 2018). Hybridization is a nuanced process, and requires equally nuanced treatment in conservation policy. Our ability to detect complex patterns of admixture and gene flow between populations is better than ever before (e.g., Ebersbach et al., 2020). Whereas early genetic analyses based on a limited number of markers (i.e. microsatellites) were lacking in power, whole and partial genome methods have revealed that admixture is both a frequent and widespread evolutionary process (vonHoldt et al., 2018). This opens the door for using admixture as a tool to deepen our understanding of the genomic processes underlying speciation and diversification. As the study of atom collisions enabled the discovery of fundamental properties of particle physics, so the study of admixture (genome collisions) may elucidate how fundamental evolutionary processes mold and maintain biological diversity.