Niches for neglected ideas?
David Houle
Abstract
Over the last 25 years, a small but persistent group of biologists have argued that evolutionary biology is overly focused on genetic inheritance and processes that change variant frequencies, and that this focus blinds mainstream evolutionary biologists to other important but neglected aspects of evolution. The list of neglected phenomena championed in Evolution Evolving is identical to those in previous articles advocating an “extended evolutionary synthesis” (Laland et al., 2014, 2015; Pigliucci & Müller, 2010). It must be noted that Evolution Evolving is subtitled “The developmental origins of adaptation and diversity,” and the authors try to make the case that the same neglected aspects of evolution that are prominent in the extended synthesis should be studied from a “developmentalist” perspective. Development is generally defined as the processes that shape the morphological phenotype of multicellular organisms (e.g., Gilbert, 2006). It does not include many of the phenomena that figure prominently in Evolution Evolving, such as learning, culture, non-genetic inheritance, or the causes of fitness. More generally, the aspects of evolution featured by advocates of the Extended Synthesis apply equally well to unicellular organisms that do not have development and the behavioral, biochemical, and physiological processes that evolve alongside development. While one can learn a good deal about development in this book, Evolution Evolving is first and foremost a presentation of, and justification for, the Extended Synthesis. I will use the name of the first author, Lala,1 instead of the phrase Lala et al. to refer to the authorship of Evolution Evolving with apologies to Tobias Uller, Nathalie Fiener, Marcus Feldman, and Scott Gilbert. Lala emphasize five aspects of evolution are relatively neglected: First, there is the role of development in shaping phenotypes. Second is niche construction, the processes by which individuals may shape their selective environments. Third is non-genetic inheritance, where factors other than genes are transmitted from parents to offspring. Fourth is phenotypic plasticity, where the phenotype is shaped by interactions between the organism and its environment. Fifth, evolvability, the responsiveness of a population to evolutionary forces, itself evolves. I refer to these aspects of evolution as the Neglected Five. I agree with Lala that the Neglected Five are interesting, actually or potentially important, and in some senses understudied. My own work has been greatly concerned with two of these aspects, the role that processes that produce phenotypes, including development, play in shaping phenotypic variation and the study of evolvability. Great questions about evolution and adaptation hinge on the relationship between genes, environments and phenotypes (Houle et al., 2010). The role of evolvability in shaping evolution has genuinely surprised me (Houle et al., 2017), and the processes that shape evolvability together form an important unsolved problem (Tsuboi et al. in press). We have a great deal more to learn about each of the Neglected Five, and their further study will continue to enrich evolutionary analysis. Evolution Evolving presents a wide variety of evidence to support the importance of the Neglected Five. The chief pleasure in reading this book comes from the thought-provoking and clearly explained empirical case studies. To take just a few examples, I learned a lot about how the microbiome of woodrats facilitates the evolution of the ability to subsist on a poisonous plant, how turtles evolved shells, and how cave fish evolved to inhabit their dark and cold world. Any reader of this book will learn a lot of biology they have not otherwise encountered. While I do not agree with the interpretation of all of the authors’ examples, the end notes clarify how they came to their conclusions and direct the reader to the primary literature. The second section of the book is a concise summary of animal development, with emphasis on its interaction with environmental factors. This section and the biological examples are the nutritious middle of this wide-ranging book. This nutritious middle is unfortunately the filling of a sandwich made with unappetizing bread. The bottom layer of the sandwich is a moldy interpretation of the history of evolutionary biology. In a nutshell, the historical claim of this book is that the Neglected Five are understudied because the architects of the modern synthesis in the first sixty years of the 20th century insisted that the Neglected Five were unimportant, and the dead hand of their prejudices prevents their proper appreciation today. The top layer of the sandwich consists of forward-looking claims that once the “developmentalist” perspective is properly appreciated, the Neglected Five will be elevated as fundamental causes of evolution equal in stature to mutation, drift and natural selection. In a few cases, these reinterpretations seem justified. For example, the authors make a strong case that cultural inheritance of learned behaviors is a different mode of evolution where the random mutational process and genetic inheritance that underlies neo-Darwinian evolution is replaced by behavioral innovation of variation, guided by experience, and inherited through a fully non-genetic process. The key to the importance of niche construction is ecological inheritance, which is at least an implicit assumption of nearly all models of adaptation, but the inheritance of modifications of the environment by the evolving population is both real and rarely modeled. In Chapter 9 (p. 124 ff), Lala et al. make the excellent point that there is less thought given to the ways that phenotype-environment interactions actually cause fitness differences. Population genetics is a discipline devoted to making predictions about how evolution proceeds once we know genotypic fitnesses, but predicting how a phenotypic change will affect fitness is only haphazardly studied. I feel the lack of such “source laws” (Sober, 1984, p. 50) in my work when I try to deal with the complexities of multivariate evolution, where a single variant may affect many traits, and there is no systematic way to predict the fitness effects on the majority of those traits. However, the bulk of the arguments that the Neglected Five will eventually revolutionize evolutionary biology are weak. For example, the authors believe that developmental bias, plasticity, and niche construction are directive forces in evolution, and that epigenetic inheritance is as important as mendelian inheritance. As the reader is likely aware, both the historical claims and the Extended Synthesis that the authors promote have been criticized, and this book prompted me to carefully reread some of these critiques (Charlesworth et al., 2017; Dickins & Rahman, 2012; Futuyma, 2017; Scott-Phillips et al., 2014; Wray et al., 2014). Sadly, these criticisms of previous iterations of the Extended Synthesis apply equally well to much of this book. For a brief, yet comprehensive, summary of the reasons that the evidence does not support reordering evolutionary theory, I suggest Futuyma (2017). I offer just two examples of the “muddle and misrepresentation” (Laland et al., 2014) that the authors’ arguments in favor of the Extended Synthesis generate. In Chapter 7, Lala lays out the case for the widely known phenomenon “that developmental systems generate some trait combinations more readily than others,” (p. 88) a phenomenon the authors term “developmental bias.” They further argue that developmental bias is a “cause of evolution” (p. 99). Lala notes that many biologists conceive of developmental bias as a constraint on evolution, and claim that this is misleading because it implies that “selection is fully responsible for both the generation of variation and the sorting of this variation,” which denies development “any explanatory responsibility for interesting aspects of evolution” (p. 88). As prime evidence that the traditional view of developmental bias is flawed, Lala offers Charlesworth et al. (1982). To the attentive reader, however, it is apparent that Charlesworth et al. have a virtually identical view of the role of developmental bias as Lala. Charlesworth et al. wrote their paper to counter Gould’s (1980) argument that neo-Darwinism “is effectively dead.” Part of Gould’s argument was that neo-Darwinism ignored the role of development in structuring variation, the same claim that Lala makes about contemporary evolutionary biology. Charlesworth et al.’s section on developmental constraints was written to counter Gould’s claim that neo-Darwinists believe that development and genetic correlations don’t matter for evolution. Charlesworth et al. devote a full page to the importance of correlations between traits in structuring the response to selection. They summarize that section by saying “the rate and direction of phenotypic change depend crucially on the pattern of genetic variation available for natural selection or random genetic drift to act upon” (p. 478), a clear endorsement of the idea that developmental bias matters for evolution. Lala’s historical analysis fails to capture the intellectual currents in evolutionary biology over the last 60 years (Futuyma, 2017; Wray et al., 2014). Their reconstruction of history instead erects a series of straw men that they can struggle to overcome. In presenting the idea that development is a cause of evolution, Lala features the example of a “domestication syndrome,” where independently domesticated animals share many of a suite of traits rarely seen in wild populations. In mammals those traits include piebald coloration, smaller jaws, teeth and brains, and floppy ears. To explain the syndrome, Lala favors the hypothesis offered by Wilkins and colleagues (Wilkins et al., 2014) that selection for tameness promotes reduced reactivity of the sympathetic nervous system by reducing the population of neural crest cells from which the sympathetic nervous system is derived. The domestication syndrome follows as an indirect response because neural crest cells also underpin the development of the traits involved in the syndrome. I find this hypothesis plausible, although others do not (Johnsson et al., 2021), and even its proponents note that the hypothesis is as yet unproven (Wilkins et al., 2021). While this example may be controversial, the underlying premise that some selection leads to indirect responses because of the pleiotropic consequences of development is surely correct. Let’s play out the logical consequences of accepting Wilkins et al.’s hypothesis. Lala argues that the Wilkins hypothesis is evidence that developmental bias plays a “creative” role in evolution by “steering” adaptation (p. 99). For me, creativity and steering imply not just change, but change in a particular direction. Where does direction come from in this example? Not from developmental bias. Lala’s definition of developmental bias is a suitably general one, incorporating all factors above the genomic level that shape the population of phenotypic variants found in a population. Hence it includes both the mean phenotype and the pattern of variation and covariation around that mean phenotype. The term “bias,” however, is in one respect an unfortunate one: in statistics, “bias” refers to the directional departure of estimates from their true population values. Lala would presumably agree that this statistical concept of directional bias has no application to development, as there is no null pattern of development for a population to deviate from. Development requires the entire organismal system to proceed, and that system does not itself drive animals towards domestication syndrome, as the continued existence of boars, rabbits, and foxes with large teeth and erect ears shows. In Wilkins et al.’s (2014) hypothesis, variation in neural crest cell abundance is the handle for selection to affect tameness, but it is selection for tameness that imparts directionality to the syndrome. While we can agree that floppy ears in domesticated animals may be caused by a decrease in neural crest cell abundance, it is essential to Wilkins et al.’s hypothesis that the decrease in those cells was caused by selection, not bias. Contrast Lala’s developmental bias with the well-known concept of mutation bias, which occurs when mutation is biased in direction. A simple example is transition-transversion bias: The null hypothesis is that each mutation at a particular nucleotide is equally likely, predicting that transversions will be twice as common as transitions, but in most organisms, mutations are biased towards transitions over transversions. These and other mutation biases can “steer” CG content, codon usage and even adaptation towards particular states (e.g., Cano et al., 2023; Hershberg & Petrov, 2010). As a graduate student searching for a suitable dissertation project, I was inspired by the work of Richard Lewontin, one of the biologists that Lala dedicate Evolution Evolving to. When I read Lewontin’s (1974) book The Genetic Basis of Evolutionary Change, I enthusiastically embraced the first five chapters of the book, but I can still vividly remember how disheartening I found the final chapter, entitled “The genome as the unit of selection.” In this chapter, Lewontin makes the case that an adequate understanding of evolution requires not just the study of genotypic fitnesses, but the study of how these fitnesses are caused by interactions with the genetic background they are embedded in, culminating in his recommendation to study fitness and evolution at the genomic level, 25 years before the publication of the first metazoan genome! This felt like a slap in the face to my younger self: How was I supposed to study evolution at the genomic level? I reread the rest of the book many times, but could not bear to return to the depressing Chapter 6 for many years. My difficulty with Lewontin’s Chapter 6 was not that I wanted to dismiss Lewontin’s famous aphorism that “Context and interaction are of the essence” (Lewontin, 1974, p. 318), but that I simply could not imagine how to make context and interaction of the essence in my own work, especially not at a genomic level. Stated more academically, I did not react as an ontological or epistemic reductionist (Brigandt & Love, 2017) asserting that evolution is best be studied by ignoring context. I was acting as a graduate student with limited resources, trying to find a feasible dissertation project. Circumstances usually force us to be methodological reductionists, as time and money are always limited, despite our desires to incorporate the complexities of life into our science. While I agree with Lala that the Neglected Five are relatively understudied, my experience with putting Lewontin’s advice into practice suggests a more obvious explanation for this than cultural bias: they are difficult to study. Lewontin (1970) identified phenotypic variation, differential fitness and inheritance of those phenotypic differences as the essential prerequisites for evolution by natural selection. Lala uses this list to make the point that most studies of natural selection treat these features as independent (p. 39), while the Neglected Five all involve interactions between Lewontin’s essential components. The interaction of mutation and the existing process of development generates developmental bias. The interactions of genotypes, development and environmental factors, which we conceptualize as plasticity, alter phenotypic variation. In niche construction, phenotypes interact with the environment, potentially altering relative fitnesses, and generating genotype-environment covariance. Interactions between environmental conditions and processes of non-genetic inheritance can predictably alter the phenotypes in future generations. Regardless of the potential importance of interactions, it is an unfortunate fact that convincing studies of interactions require larger sample sizes, more complex designs and more sophisticated analyses than studies that hold some or all of the interacting factors constant. For example, studying phenotypic variation and fitness in a plant population in a single novel environment can tell you whether there is natural selection on the population and enable predictions about what traits will respond to selection. To connect such predictions to plasticity, one would have to replicate genotypes, then plant them in a biologically relevant range of environments. Each of these additional steps requires careful planning, time, and money. An additional factor that makes studies of the genotype-phenotype map and developmental bias difficult is that developmental and systems biology are still focused on deciphering the constituents of the map. Biologists still struggle to discover the function of many of the genes that produce phenotypic variation. Developmental processes themselves are complex and highly interactive, as described by Lala in the second section of the book. Developmental biases will remain hard to characterize until the constituents of the map that produces them are clear. Evolution Evolving and the literature on the Extended Synthesis imagine one way in which neglected aspects of evolution can be incorporated into an existing field: they favor elevating these aspects to principles equal in importance to the prevailing fundamental principles of evolution, including undirected mutation, natural selection, and inheritance. Okasha (2021) instead suggests that important phenomena have historically been incorporated into evolutionary biology through a process he terms endogenization. Endogenization is using the fundamental principles of evolutionary biology to explain phenomena previously used as background conditions to the study of evolution. For example, the existence of inherited phenotypic variation was a critically important assumption for Darwin. Given Darwin’s ignorance of genetics, he could not explain the existence of inherited variation within the confines of his evolutionary theories, although he could readily observe it. Since Darwin’s day, the discovery of Mendelian inheritance and the development of population genetics have us many fully evolutionary for phenotypic variation. These include of selection, and genetic drift of Okasha (2021) also notes that some biological phenomena require no evolutionary as they are consequences of or list of principles consists of such system of inheritance is a for biological evolution to is in system of inheritance. selection is a logical once mutation inherited variation in fitness. This suggests a clear for when a phenomenon or process should be elevated to a it is a key of evolutionary but does not have a likely evolutionary we should treat it as a It clear to me that of the Neglected Five can clear this does this mean that we are ignoring processes that can be from it. one could argue that the study of inherited variation is not to the study of evolution. Okasha many phenomena to his list of including the existence of two and the of biological from genes to and on to multicellular organisms, and populations. These are clearly of not importance to evolutionary biology. Their existence is the point for further evolutionary analyses of and relevant to Evolution Evolving, Okasha also suggests that of the Neglected Five have been into evolutionary biology. Okasha argues that development bias is a of the evolution of the genotype-phenotype which can by selection on the of the including the it or as an indirect response to selection for particular phenotypes. we that the genotype-phenotype map it is clear evolvability of the Neglected Five. Okasha argues that niche construction has been through models of the interaction between mendelian processes and environmental he the niche construction models of and two of the authors of Evolution Evolving as the hard work of that et al., As noted niche construction models involve ecological inheritance, a of the that to some is given by the and of the environment and not fully to endogenization. In I to the of the Neglected Five and note that this book does one to about despite its I in with most of Lala’s to the of study of the Neglected Five. For example, on Lala et al. We are the causes of fitness traits will be by and traits will be through to be questions for evolutionary biology. other evolutionary we to the of are the of and makes organisms good at A developmental perspective offers and to such statistical this were the of the book, I would be to it. of The no of I and for me into this by and Futuyma prompted me to it.