The problem of variation

Chapter 7 maps out a broad framework for considering the problem of variation in evolution. Under the neo-Darwinian view that variation merely plays the role of supplying random infinitesimal raw materials, with no dispositional influence on the course of evolution, a substantive theory of form and its variation is not required to specify a complete theory of evolution. This view has been breaking down from the moment it was proposed, and is now seriously challenged by results from evo-devo, comparative genomics, molecular evolution, and quantitative genetics. For instance, the multivariate generalization of quantitative genetics indicates that selection cannot possibly act as an independent governing force. Replacing a theory of variation as fuel with a theory of variation as a dispositional factor will require, at minimum, an understanding of tendencies of variation (source laws), and an understanding of how those tendencies affect evolution (consequence laws).

Climbing Mount Probable

Chapter 8 provides the formal basis to recognize biases in the introduction of variation as a cause of evolutionary biases. The shifting-gene-frequencies theory of the Modern Synthesis posits a “buffet” view in which evolution is merely a process of shifting the frequencies of pre-existing alleles, without new mutations. Within this theory, mutation is represented like selection or drift, as a “force” that shifts frequencies. Yet, within a broader conception of evolution, a second kind of causal process is required: an introduction process that can shift a frequency upwards from 0, which selection and drift cannot do. Abstract models demonstrate the influence of biases in the introduction process in one-step and multi-step adaptive walks. Such biases do not require mutation biases per se, but may arise from effects of development, and from the differential accessibility of alternative forms in abstract possibility-spaces.

The revolt of the clay

Chapter 9 presents an empirical case for the importance of mutational biases, based on studies of adaptation traced to the molecular level. Where Chapter 8 identified a variational cause of bias that does not depend on neutral evolution, absolute constraints, or high mutation rates, this chapter focuses on how quantitative biases in ordinary nucleotide mutations influence adaptive evolution. It uses published studies of parallel adaptation in nature and in the laboratory. The natural studies include both (1) cases of recent local adaptation, e.g., evolution of resistance to insecticides and herbicides, and (2) cases of fixed changes, e.g., altitude adaptation via changes in hemoglobins, spectral tuning of photoreceptors used in color vision, and so on. The results indicate that the kinds of changes that happen most often in adaptation are the kinds favored by simple biases in mutation, e.g., transition-transversion bias.

Ordinary randomness

Chapter 2 addresses how well the biological process of mutation is described by some of the ordinary meanings of “chance“ or “randomness“ in science: lack of purpose or foresight, uniformity (homogeneity), stochasticity, indeterminacy, unpredictability, spontaneity, and independence (chance). Ordinary mutations exhibit various kinds of heterogeneity (nonuniformity), e.g., by genomic position, or by cell-cycle state. The occurrence of mutations is affected by various conditions inside the cell, e.g., the spectrum of replication errors is shaped by the composition of DNA precursor pools. Many of the processes that lead to mutation are spontaneous in the sense of emerging internally, but some processes reflect external effects such as radiation or uptake of foreign DNA. Though most of the processes that lead to mutations are “macroscopic,” some processes (e.g., damage caused by radioactive decay or electromagnetic radiation) implicate quantum indeterminacy.

Mutational mechanisms and evolvability

Well-studied cases of programmed DNA rearrangements, e.g., somatic recombination in the emergence of specific antibodies, suggest a rubric for specially evolved mutation systems: they amplify the rates of specific types of mutations (by orders of magnitude), subject to specific modulation, using dedicated parts, with the favored types of mutations being used repeatedly. Chapter 5 focuses on six types of systems that generate mutational diversity in a focused manner, often in an ecological context that makes sense of such a specialized feature, e.g., immune evasion or phage-host coevolution: cassette shuffling, phase variation (switching), CRISPR-Cas defenses, inversion shufflons, diversity-generating retro-elements, and mating-type switching. The emergence and influence of these systems relates to the concept of evolvability, here expressed in terms of three types of claims: evolvability as fact (E1), evolvability as explanans (E2), and evolvability as explanandum (E3).

Evolutionary Randomness

Contemporary defenses of the randomness doctrine refer not to ordinary meanings of randomness, but to a special evolutionary meaning by which mutation is said to be independent of, or uncorrelated with, something like environment, selection, evolution, or adaptation. Chapter 4 addresses whether this type of claim is justified, either empirically or on the basis of principles of biology. In fact, claims of evolutionary randomness typically fail for obvious reasons. Mutation in a particular locale is not walled off from the rest of biology, but is sensitive to conditions such as gene expression and the concentration of DNA precursors, and this leads to nonindependence with conditions and even with fitness. The incipient fitness effects of mutational intermediates typically are not involved in feedback loops that enhance (diminish) the chances of completing a beneficial (harmful) mutation, but this narrow sense of independence has little practical value.

Randomness as Irrelevance

Under the neo-Darwinian theory, selection is the potter and variation is the clay: peculiarities or regularities of variation may emerge from internal causes, but these are ultimately irrelevant, because selection governs the outcome of evolution. Chapter 6 addresses this sense of “randomness” as irrelevance or unimportance, featuring (1) an analogical-metaphysical argument in which mutation is equated with raw materials or fuel, or is said to act at the “wrong level” to be an evolutionary cause; (2) direct empirical arguments; (3) mechanistic claims, e.g., claims about the ability of the “gene pool” to maintain variation, or of selection to be creative; (4) methodological claims to the effect that selection is amenable to study, but not mutation; and (5) an explanatory claim to the effect that mutation, though perhaps influential, only affects the boring parts of evolution. Appendix D provides quotations on the theme of unimportance.

Introduction: a curious disconnect

Chapter 1 begins with a synopsis of the central argument concerning models of evolution (and theories of causation) that incorporate a mutational introduction process, using a study of laboratory adaptation that shows proportional effects of a 50-fold range of rates for different mutations. The exploration of the role of variation in this book covers mutation and randomness, the neo-Darwinian dichotomy of selection and variation, the shifting-gene-frequencies theory of the Modern Synthesis (and its relation to population-genetic “forces“), developmental bias, self-organization, the emergence of evolvability as a major topic, and the causes of parallel adaptation. This chapter provides a guide to the remainder of the book, and explains how the main arguments relate to more familiar topics such as evo-devo, the distinctiveness of molecular evolution, the “directed mutations“ controversy, and debates about the adequacy of a “Modern Synthesis.”

Moving on

Chapter 10 includes a synopsis of key points from previous chapters as well as reflections on changing explananda, notions of causation, and the importance of identifying testable theories. The ongoing delay in recognizing the introduction process as a dispositional evolutionary cause reflects the lasting influence of the shifting-gene-frequencies theory, and a lack of influence of molecular studies of evolution. Evolutionary discourse proceeds as if the major issues are defined relative to the morphology and behavior of large charismatic animals, yet evolutionary biologists themselves focus increasingly on molecules and microbes. Verbal theories of causation play an important role in determining what causes are possible and what they may explain. In contemporary debates on the status of “evolutionary theory,” the pressure to defend or reject a flexible “Synthesis” distorts history and spawns confusion over what makes a theory. Testable theories, not loosely defined traditions, are what make science distinctive.

Practical Randomness

Chapter 3 addresses the idea of randomness as a simplifying assumption, beginning with a discussion (using examples from phylogenetics) of the reasons that scientists employ simplifying assumptions that are known to be incorrect. That is, some ways of thinking about mutation may be useful, even if they are only approximately correct. Approximations come at a cost, and thus the practical use of an approximation, e.g., the assumption that mutation is uniform when it really is not, is a matter of weighing costs and benefits. The application of probabilistic reasoning to problems of mutation may be understood as an extension of logic that does not rely on any concept of “randomness.” In this context, references to “chance” or “randomness” as something that exists in the physical world, rather than in our minds, represent what E.T. Jaynes calls a “mind projection fallacy.”