Oh, what a tangled web we weave

Abstract

Divergence with Genetic Exchange. By Michael L. Arnold. Oxford University Press, Oxford, UK, 2016. xv + 251 pp. US $125 (hardback), ISBN 978-0-19-872602-9; US $64.95 (paperback) ISBN 978-0-19-87551-1. Although I have a great respect and fondness for Darwin’s illustration of the evolutionary process as a simple, bifurcating treelike structure, such a model has been falsified by a wealth of data. Instead, the genotypic, phenotypic, ecological and evolutionary relationships among organisms are much more complex. (p. 175) An idea that is receiving a lot of press these days is the notion that the “Tree of Life” as a description of the organizational framework of biological diversity should be reconceived as something other than an irregularly bifurcating hierarchy. Not long ago (Brower, 2019), I reviewed a popular account about interspecific gene flow, David Quammen’s Tangled Trees, in which I pointed out some core logical flaws in the book’s argument. Here I address another relatively recent manifesto of similar, if more empirically bolstered, rhetoric advancing the thesis that the Tree of Life should be replaced by a webbier conceptual framework. Although the threat to the tree metaphor, firmly established since Darwin’s time, may seem inconsequential, systematists need to be cognizant of the rising tide of heterodoxy lapping at the conceptual foundations of their field. After a stint with Australian grasshoppers, Michael Arnold has spent his career studying a charismatic three-way hybrid zone among the “Louisiana irises”, Iris fulva, Iris brevicaulis and Iris hexagona, and writing books emphasizing the creative role of gene flow in speciation among other taxa (Arnold, 1997; 2006; 2009). Each of these books has been a plea to fellow evolutionists to unfetter themselves from the blinkers of Ernst Mayr’s allopatric speciation model and to entertain the role of interspecific gene flow as a progressive force in evolutionary diversification. Arnold’s fourth book is buoyed by the torrent of genomic data from plants and animals that has inundated the systematic literature in the past decade, allowing him to invoke instance after instance of ostensible introgression and hybrid speciation across the spectrum of biological diversity. He clearly feels that he is on the winning side of history in his advocacy of the prevalence of these processes: “I want to express my profound relief concerning the degree to which evolutionary biologists in general have embraced a model of diversification that includes some measure of genetic exchange” (p. vii). I am not going to spend a lot of time describing the particulars of the book’s content, mainly composed of confirmatory vignettes drawn from the classic and more recent empirical literature. The reader is presented with accounts of gene flow in many groups, for example, Darwin’s finches, Italian sparrows, goatweeds, sunflowers, plague bacteria, bats, mice, bears, horses, monkeys, humans and butterflies. Instead, I will focus on the book’s peripatetic structure, which struck me as deliberately obfuscatory. Arnold’s accounts jump back and forth between viruses, bacteria, plants, animals and the occasional fungus. The purpose of this hopscotch seems to be to overwhelm the reader with a collage of examples of horizontal transfer, genome duplication, hybridization, introgression and putative hybrid speciation, and to homogenize the variety of these different reproductive and nonreproductive processes under the big tent of “genetic exchange”. Just because a phenomenon (or group of phenomena) is widespread or even prevalent among a variety of groups, that does not necessarily mean it represents a driving force of evolution. One could as easily argue that speciation occurs in spite of pervasive gene flow. No systematist will argue that speciation is marked by the instantaneous origin of synapomorphies, that monophyly is unencumbered by homoplastic character distributions, or that interspecific gene flow never occurs (as amply refuted by Arnold’s many examples). Indeed, the key to the success of the Darwinian and Hennigian Revolutions is that systematics provides a unified and robust hierarchical structure for describing the pattern of biological diversity, in spite of these inconveniences. Without this epistemological framework, it is not possible to fruitfully consider adaptation, convergence and other processes of interest to evolutionary biologists. For example, one cannot measure or even think about the admixture of Neanderthal alleles into “humans” unless you have a concept of what distinguishes “Neanderthal” and “human” as separate entities. Even if systematists countenance some “genetic exchange” with the divergence that is their main focus, it is critical to acknowledge that are different kinds of gene flow with different impacts among different taxa. For example, viruses can move DNA around and integrate themselves into the genomes of their hosts. Viruses are not alive, do not reproduce themselves and maintain their existence solely by virtue of horizontal transfer. Thus, there is no “phylogeny” of viruses. Likewise, it is common knowledge that conjugation, transformation and other forms of DNA exchange occur routinely enough among bacteria to have allowed for observed instances of evolution of pathogenicity and drug resistance. On account of these processes, the taxonomy of “prokaryotes” remains somewhat phenetic and arbitrary. Among eukaryotes, however, the tree-like pattern of relationships is less ambiguous, and interspecific gene flow may not be so prevalent or straightforwardly adaptive in many cases. Yes, eukaryotes contain mitochondria and in many cases chloroplasts of bacterial origin, but those ancient symbiotic assimilations do not obscure patterns of relationship. Indeed, DNA from these organelles have been some of the most popular markers for phylogenetic inference of relationships among the organisms that contain them. Despite a surge in reports of interspecific hybridization (cf. Mallet, 2007)––perhaps in part due to the redefinition of species as diagnosable entities rather than “biological” entities––most eukaryotic species are not known to engage in interspecific reproduction. Among those that do, many hybrid zones are tension zones, in which hybrid offspring, particularly F2 and parental backcross individuals, are less fit than parental forms due to Dobzhansky–Muller incompatibilities and other selective disadvantages (Mallet and Barton, 1989; Coyne and Orr, 2004; Harrison and Larson, 2016). Arnold paints a rosy picture of such cases, arguing that even though the average hybrid genotype might be less fit, some individuals inherit combinations of alleles that are above average, offering opportunities for adaptive novelty. But superior genotypes do not replicate themselves in sexually reproducing taxa, so the chances of such a hopeful monster passing along its heterotic advantages to offspring are modest, at best. Many observed instances of interspecific hybridization are due to the direct or indirect action of humans, either moving taxa around, or changing the environment in ways that bring formerly isolated taxa in contact with one another. The Florida panther and the red wolf are respective examples. Once, a Florida panther was part of a genetically distinct lineage; now, it just a panther that lives in Florida. In his chapter on conservation, Arnold argues that we should not be so quick to judge the impact of humans on the integrity of species as “unnatural”, because humanity also is a part of nature, and, as he has already established in previous chapters, species really did not have that much integrity anyway, even before the Anthropocene. Of course, one could make the same argument about extinction: all species are mortal, so why preoccupy ourselves with the loss of polar bears? (After all, they are really the same thing as grizzly bears.) From a taxonomic perspective, dosage compensation and other selective constraints against polyploidy, as well as the inability to self, make stasipatric speciation (White, 1978) much more onerous in animals than it is in plants. But homoploid hybrid speciation, in which hybrid offspring are reproductively compatible with their parental species and must avoid backcrossing with them in order to become established as an independent and stable “third way”, is the biggest challenge of all (Buerkle et al., 2000; Barton, 2001; Barton, 2013; Schumer et al., 2014). Arnold quotes Mallet (2007): “ … there is no reason why homoploid hybrid species would be rarer in animals than in plants”. Here is a reason: plants can self-pollinate, which although not as certain a route to genetic isomorphism as parthenogenesis, allows one individual to reproduce variations on its genotypic theme by itself. Interspecific F1 animals need to find a mate and the demographic odds favour that mate being a parental form rather than a fellow interspecific hybrid. In sum, Arnold’s indiscriminate mustering of this heterogeneous array of mechanisms to argue for a “web of life” conception across all of biological diversity seems to me to jump to conclusions in a manner that demonstrates hostility to systematics and advocates epistemological chaos. Arnold says, “the concept of genetic exchange explored in this book is intentionally independent of species concepts” (p. 24). This is a disingenuous or delusional claim: one cannot talk or even think about “species x” hybridizing with “species y” without a preconception that “species x” and “species y” are epistemologically reified as separate entities. As I pointed out in my review of Quammen’s book, neither can one conceive of, or empirically discover, “foreign” genes being introduced into an organism’s genome without an a priori theory of what constitutes that organism’s “native” DNA. The notion that the study of interspecific gene exchange is not dependent on an underlying species concept, even if that concept is unstated, is preposterous, and Arnold’s demurral to explicitly commit to one or another of these is a casuistic feint that allows him wiggle room for semantic obfuscation. Nevertheless, his references to “clades” and “lineages” throughout the book, not to mention the regular employment of Linnean binomina, reveals the underlying epistemological necessity of specific distinctness and a hierarchical structure for effective communication about taxa. Naturally, Arnold denies that species need to be “reproductively isolated”––the requirement of Mayr’s Biological Species Concept (BSC). The BSC is obviously the antithesis of Arnold’s agenda, because applying its criterion would lump together any taxa that are capable of interbreeding, relegating gene flow solely to an intraspecific level. Rejecting the interbreeding criterion is hardly novel: for 40 years, cladists have argued that recognizing species based on that symplesiomorphy is undesirable (Rosen, 1979). Literal application of the BSC means that lions and tigers, wolves, dogs and coyotes, modern humans and Neanderthals would each become single, spatiotemporally variable species, and “speciation with genetic exchange” would be semantically eliminated, rather than the subject of empirical investigation. Arnold asserts that under the Phylogenetic Species Concept (PSC) (he cites Cracraft, 1989), “the requirement of monophyly in species definition eliminates the possibility of genetic exchange in the origin of new species” (p. 29). The premise of this statement is, of course, false: there is nothing about a diagnosably distinct, irreducible cluster that requires or even implies monophyly, and neither Hennig (1966), nor Cracraft (1989), nor Nixon and Wheeler (1990) claimed that species must be “monophyletic”; in fact, the last asserted that the concept of monophyly is not applicable to individual species-level taxa. If one claims to dismiss a species concept, one ought at least to represent its criteria correctly. The PSC is at least as compatible with modest levels of gene flow as other species concepts which Arnold discusses. For example, the Genealogical Species Concept (at least as described here) seems internally contradictory: “the most inclusive group of organisms having potential for genetic and/or demographic exchangeability” (Templeton, 1989) is plainly at odds with the statement Arnold attributes to no less than four different Templeton papers: “cohesion species may participate in some degree of genetic exchange (e. g., syngameon) with other lineages and yet maintain their species status” (p. 31). The element that each of the concepts (as presented by Arnold) has in common is metaphysicality––species are things that exist in the world because of some defining property they possess or lack, such as monophyly or the ability to interbreed. By contrast, my understanding of the PSC is that it recognizes and circumscribes a species as an empirical hypothesis based on the observation of one or more fixed character state differences between it and its sister species (another hypothesis). By contrast to species as entities that meet some ontological criterion at present, or even in the future, species-as-hypotheses are less constrained and more testable. Ironically, it may well be the finer divisions of biological diversity under the PSC that has led to the recent recognition of so many new cases of “interspecific hybridization”. If, for example, the number of species of birds doubled as a result of abandoning Mayrian BSC definitions for Cracraftian PSC definitions, then presumably about half of the latter sort of species are capable of hybridizing with at least one other species. A nonspecialist reader cannot, in general, directly evaluate from Arnold’s text the empirical validity of the nuggets of evidence he draws to advance his argument. To bolster his claims, Arnold meticulously cites the literature––there are 61 pages of references (25% of the book). Of course, it is better to cite your sources than not to cite them, but Arnold’s strategy is not only exhaustive, but exhausting, and one cannot help but wonder if this is a deliberate stratagem to foist the burden of proof from the author onto the reader: Don’t believe me? Then go look at these ten references…. Of course, for every reference that Arnold cites, there could be another one he omitted that might offer an alternative interpretation of the “facts”1. The book would have made a stronger case if Arnold had tried to measure the prevalence or importance of gene flow to speciation objectively, rather than simply reporting favorable evidence. Even if the reporting is unbiased, the empirical underpinnings of these examples are in many cases black boxes. The nature of the genomic evidence that has lately advanced Arnold’s agenda is largely inscrutable to critical evaluation. Very few reviewers have the wherewithal to wade through reams of supplemental material to assess whether the data actually support the conclusions asserted in a report (I recall a recent four-page Science paper on chimpanzees with 129 supplemental pages appended to it!). A critical reader of such genomic data dumps might wonder: What is the evidentiary value of SNPs––anonymous genetic polymorphisms extracted from random segments of genomic sequence––as markers of specific distinctness or introgression? What is a “signature of gene flow” in a sliding window examination of FST? What is the null hypothesis? Were the measures of significance corrected for multiple comparisons? How many comparisons were there? Many of the analytical techniques that purport to reveal genetic admixture, such as Patterson’s D (aka “ABBA-BABA”; Durand et al., 2011) or Structure (Pritchard et al., 2000) assume that their markers are selectively neutral, but the appeal of introgression stories is mainly due to the claims of adaptive novelty that this mode of inheritance offers. You can’t have it both ways. Once again: the epistemological framework that makes the notion of “incongruence” meaningful is disagreement of some subset of data with an a priori baseline phylogenetic pattern. The falsifying evidence referred to in the epigram at the top of this review is what systematists call “homoplasy”. Systematics is not population genetics writ large. If there has been a paradigm shift in the evolutionary biological community’s interpretation of incongruent sharing of characters across specific boundaries, it seems to relate to a desire to explain homoplasy as homology due to nonphylogenetic patterns of inheritance, or simply to jettison the concepts of homology and homoplasy altogether, and revert instead to phenetic descriptions of levels of admixture. Contrary to Arnold’s claim, the tree is a pattern, not a process, and it is the conflation of pattern and process that leads to the sort of nihilistic rhetoric this book espouses. To cladists, such efforts represent the epitome of unfalsifiable ad hoc hypotheses. As Darwin (1859:358) said, “ … classification is not arbitrary like the grouping of stars in constellations.”