Parsimony in evolutionary theory: Law or methodological prescription?

Parsimony (Ockham's razor) is in widespread use in phylogenetic reconstruction (evolution takes the shortest route), however it is not quite obvious which is the rank that this principle should have in evolutionary theory. Parsimony is not of a single kind but, on the contrary, is at least of two kinds: ontological and methodological. Ontological parsimony involves an assumption about the “simplicity of nature”. Methodological parsimony is a purely logical precept, a case of the broad practical principle not to believe anything for which there is no evidence. The two kinds of parsimony are not compatible with one another. The ontological hypotheses that reality is simple has been refuted many times in the history of science, and evolution is not an exception to this. In spite of the fact, that direct evolutionary changes have higher probability than the ones that take “unnecessary” steps, evolutionary parsimony is merely a methodological precept, not a law of evolution. Probability is not enough to give evolutionary parsimony a rank of ontological axiom. Therefore, the reasons to use the principle of evolutionary parsimony are only methodological. A definition of evolutionary parsimony is: as long as no evidence is available to suggest an alternative pathway evolution may be considered to occur in the most parsimonious way.     

Phylogenetic approaches for studying diversification

Estimating rates of speciation and extinction, and understanding how and why they vary over evolutionary time, geographical space and species groups, is a key to understanding how ecological and evolutionary processes generate biological diversity. Such inferences will increasingly benefit from phylogenetic approaches given the ever‐accelerating rates of genetic sequencing. In the last few years, models designed to understand diversification from phylogenetic data have advanced significantly. Here, I review these approaches and what they have revealed about diversification in the natural world. I focus on key distinctions between different models, and I clarify the conclusions that can be drawn from each model. I identify promising areas for future research. A major challenge ahead is to develop models that more explicitly take into account ecology, in particular the interaction of species with each other and with their environment. This will not only improve our understanding of diversification; it will also present a new perspective to the use of phylogenies in community ecology, the science of interaction networks and conservation biology, and might shift the current focus in ecology on equilibrium biodiversity theories to non‐equilibrium theories recognising the crucial role of history.     

Simulation as experiment: a philosophical reassessment for biological modeling

Some scientific modelers suggest that complex simulation models that mimic biological processes should have a limited place in ecological and evolutionary studies. However, complex simulation models can have a role that is different from that of simpler models that are designed to be fit to data. Simulation can be viewed as another kind of experimental system and should be analyzed as such. Here, I argue that current discussions in the philosophy of science and in the physical sciences fields about the use of simulation as an experimental system have important implications for biology, especially complex sciences such as evolution and ecology. Simulation models can be used to mimic complex systems, but unlike nature, can be manipulated in ways that would be impossible, too costly or unethical to do in natural systems. Simulation can add to theory development and testing, can offer hypotheses about the way the world works and can give guidance as to which data are most important to gather experimentally.     

CHAOS AND UNPREDICTABILITY IN EVOLUTION

The possibility of complicated dynamic behavior driven by nonlinear feedbacks in dynamical systems has revolutionized science in the latter part of the last century. Yet despite examples of complicated frequency dynamics, the possibility of long‐term evolutionary chaos is rarely considered. The concept of “survival of the fittest” is central to much evolutionary thinking and embodies a perspective of evolution as a directional optimization process exhibiting simple, predictable dynamics. This perspective is adequate for simple scenarios, when frequency‐independent selection acts on scalar phenotypes. However, in most organisms many phenotypic properties combine in complicated ways to determine ecological interactions, and hence frequency‐dependent selection. Therefore, it is natural to consider models for evolutionary dynamics generated by frequency‐dependent selection acting simultaneously on many different phenotypes. Here we show that complicated, chaotic dynamics of long‐term evolutionary trajectories in phenotype space is very common in a large class of such models when the dimension of phenotype space is large, and when there are selective interactions between the phenotypic components. Our results suggest that the perspective of evolution as a process with simple, predictable dynamics covers only a small fragment of long‐term evolution.     

Weighted parsimony outperforms other methods of phylogenetic inference under models appropriate for morphology

One of the lasting controversies in phylogenetic inference is the degree to which specific evolutionary models should influence the choice of methods. Model‐based approaches to phylogenetic inference (likelihood, Bayesian) are defended on the premise that without explicit statistical models there is no science, and parsimony is defended on the grounds that it provides the best rationalization of the data, while refraining from assigning specific probabilities to trees or character‐state reconstructions. Authors who favour model‐based approaches often focus on the statistical properties of the methods and models themselves, but this is of only limited use in deciding the best method for phylogenetic inference—such decision also requires considering the conditions of evolution that prevail in nature. Another approach is to compare the performance of parsimony and model‐based methods in simulations, which traditionally have been used to defend the use of models of evolution for DNA sequences. Some recent papers, however, have promoted the use of model‐based approaches to phylogenetic inference for discrete morphological data as well. These papers simulated data under models already known to be unfavourable to parsimony, and modelled morphological evolution as if it evolved just like DNA, with probabilities of change for all characters changing in concert along tree branches. The present paper discusses these issues, showing that under reasonable and less restrictive models of evolution for discrete characters, equally weighted parsimony performs as well or better than model‐based methods, and that parsimony under implied weights clearly outperforms all other methods.     

Application of network methods for understanding evolutionary dynamics in discrete habitats

In populations occupying discrete habitat patches, gene flow between habitat patches may form an intricate population structure. In such structures, the evolutionary dynamics resulting from interaction of gene‐flow patterns with other evolutionary forces may be exceedingly complex. Several models describing gene flow between discrete habitat patches have been presented in the population‐genetics literature; however, these models have usually addressed relatively simple settings of habitable patches and have stopped short of providing general methodologies for addressing nontrivial gene‐flow patterns. In the last decades, network theory – a branch of discrete mathematics concerned with complex interactions between discrete elements – has been applied to address several problems in population genetics by modelling gene flow between habitat patches using networks. Here, we present the idea and concepts of modelling complex gene flows in discrete habitats using networks. Our goal is to raise awareness to existing network theory applications in molecular ecology studies, as well as to outline the current and potential contribution of network methods to the understanding of evolutionary dynamics in discrete habitats. We review the main branches of network theory that have been, or that we believe potentially could be, applied to population genetics and molecular ecology research. We address applications to theoretical modelling and to empirical population‐genetic studies, and we highlight future directions for extending the integration of network science with molecular ecology.     

Rhythmic and non-rhythmic attractors in asynchronous random Boolean networks

In multi-component, discrete systems, such as Boolean networks and cellular automata, the scheme of updating of the individual elements plays a crucial role in determining their dynamic properties and their suitability as models of complex phenomena. Many interesting properties of these systems rely heavily on the use of synchronous updating of the individual elements. Considerations of parsimony have motivated the claim that, if the natural systems being modelled lack any clear evidence of synchronously driven elements, then random asynchronous updating should be used by default. The introduction of a random element precludes the possibility of strictly cyclic behaviour. In principle, this poses the question of whether asynchronously driven Boolean networks, cellular automata, etc., are inherently bad choices at the time of modelling rhythmic phenomena. This paper focuses on this subsidiary issue for the case of Asynchronous Random Boolean Networks (ARBNs). It defines measures of pseudo-periodicity between states and sufficiently relaxed statistical constraints. These measures are used to guide a genetic algorithm to find appropriate examples. Success in this search for a number of cases, and the subsequent statistical analysis lead to the conclusion that ARBNs can indeed be used as models of co-ordinated rhythmic phenomena, which may be stronger precisely because of their in-built asynchrony. The same technique is used to find non-stationary attractors that show no rhythm. Evidence suggests that the latter are more abundant than rhythmic attractor. The methodology is flexible, and allows for more demanding statistical conditions for defining pseudo-periodicity, and constraining the evolutionary search.     

Assessing variation in life‐history tactics within a population using mixture regression models: a practical guide for evolutionary ecologists

Mixed models are now well‐established methods in ecology and evolution because they allow accounting for and quantifying within‐ and between‐individual variation. However, the required normal distribution of the random effects can often be violated by the presence of clusters among subjects, which leads to multi‐modal distributions. In such cases, using what is known as mixture regression models might offer a more appropriate approach. These models are widely used in psychology, sociology, and medicine to describe the diversity of trajectories occurring within a population over time (e.g. psychological development, growth). In ecology and evolution, however, these models are seldom used even though understanding changes in individual trajectories is an active area of research in life‐history studies. Our aim is to demonstrate the value of using mixture models to describe variation in individual life‐history tactics within a population, and hence to promote the use of these models by ecologists and evolutionary ecologists. We first ran a set of simulations to determine whether and when a mixture model allows teasing apart latent clustering, and to contrast the precision and accuracy of estimates obtained from mixture models versus mixed models under a wide range of ecological contexts. We then used empirical data from long‐term studies of large mammals to illustrate the potential of using mixture models for assessing within‐population variation in life‐history tactics. Mixture models performed well in most cases, except for variables following a Bernoulli distribution and when sample size was small. The four selection criteria we evaluated [Akaike information criterion (AIC), Bayesian information criterion (BIC), and two bootstrap methods] performed similarly well, selecting the right number of clusters in most ecological situations. We then showed that the normality of random effects implicitly assumed by evolutionary ecologists when using mixed models was often violated in life‐history data. Mixed models were quite robust to this violation in the sense that fixed effects were unbiased at the population level. However, fixed effects at the cluster level and random effects were better estimated using mixture models. Our empirical analyses demonstrated that using mixture models facilitates the identification of the diversity of growth and reproductive tactics occurring within a population. Therefore, using this modelling framework allows testing for the presence of clusters and, when clusters occur, provides reliable estimates of fixed and random effects for each cluster of the population. In the presence or expectation of clusters, using mixture models offers a suitable extension of mixed models, particularly when evolutionary ecologists aim at identifying how ecological and evolutionary processes change within a population. Mixture regression models therefore provide a valuable addition to the statistical toolbox of evolutionary ecologists. As these models are complex and have their own limitations, we provide recommendations to guide future users.     

Colour spaces in ecology and evolutionary biology

The recognition that animals sense the world in a different way than we do has unlocked important lines of research in ecology and evolutionary biology. In practice, the subjective study of natural stimuli has been permitted by perceptual spaces, which are graphical models of how stimuli are perceived by a given animal. Because colour vision is arguably the best‐known sensory modality in most animals, a diversity of colour spaces are now available to visual ecologists, ranging from generalist and basic models allowing rough but robust predictions on colour perception, to species‐specific, more complex models giving accurate but context‐dependent predictions. Selecting among these models is most often influenced by historical contingencies that have associated models to specific questions and organisms; however, these associations are not always optimal. The aim of this review is to provide visual ecologists with a critical perspective on how models of colour space are built, how well they perform and where their main limitations are with regard to their most frequent uses in ecology and evolutionary biology. We propose a classification of models based on their complexity, defined as whether and how they model the mechanisms of chromatic adaptation and receptor opponency, the nonlinear association between the stimulus and its perception, and whether or not models have been fitted to experimental data. Then, we review the effect of modelling these mechanisms on predictions of colour detection and discrimination, colour conspicuousness, colour diversity and diversification, and for comparing the perception of colour traits between distinct perceivers. While a few rules emerge (e.g. opponent log–linear models should be preferred when analysing very distinct colours), in general model parameters still have poorly known effects. Colour spaces have nonetheless permitted significant advances in ecology and evolutionary biology, and more progress is expected if ecologists compare results between models and perform behavioural experiments more routinely. Such an approach would further contribute to a better understanding of colour vision and its links to the behavioural ecology of animals. While visual ecology is essentially a transfer of knowledge from visual sciences to evolutionary ecology, we hope that the discipline will benefit both fields more evenly in the future.     

Engineering microbial systems to explore ecological and evolutionary dynamics

A major goal of biological research is to provide a mechanistic understanding of diverse biological processes. To this end, synthetic biology offers a powerful approach, whereby biological questions can be addressed in a well-defined framework. By constructing simple gene circuits, such studies have generated new insights into the design principles of gene regulatory networks. Recently, this strategy has been applied to analyze ecological and evolutionary questions, where population-level interactions are critical. Here, we highlight recent development of such systems and discuss how they were used to address problems in ecology and evolutionary biology. As illustrated by these examples, synthetic ecosystems provide a unique platform to study ecological and evolutionary phenomena that are challenging to study in their natural contexts.     

Organisms as natural purposes: the contemporary evolutionary perspective

Kant's conception of organisms as natural purposes raises a challenge to the adequacy of mechanistic explanation in biology. Certain features of organisms appear to be inexplicable by appeal to mechanical law alone. Some biological phenomena, it seems, can only be accounted for teleologically. Contemporary evolutionary biology has by and large ignored this challenge. It is widely held that Darwin's theory of natural selection gives us an adequate, wholly mechanical account of the nature of organisms. In contemporary biology, the category of the organism plays virtually no explanatory role. Contemporary evolutionary biology is a science of sub-organismal entities-replicators. I argue that recent advances in developmental biology demonstrate the inadequacy of sub-organismal mechanism. The category of the organism, construed as a 'natural purpose' should play an ineliminable role in explaining ontogenetic development and adaptive evolution. According to Kant the natural purposiveness of organisms cannot be demonstrated to be an objective principle in nature, nor can purposiveness figure in genuine explain. I attempt to argue, by appeal to recent work on self-organization, that the purposiveness of organisms is a natural phenomenon, and, by appeal to the apparatus of invariance explanation, that biological purposiveness provides genuine, ineliminable biological explanations.     

Experimental strategies to assess the biological ramifications of multiple drivers of global ocean change—A review

Marine life is controlled by multiple physical and chemical drivers and by diverse ecological processes. Many of these oceanic properties are being altered by climate change and other anthropogenic pressures. Hence, identifying the influences of multifaceted ocean change, from local to global scales, is a complex task. To guide policy‐making and make projections of the future of the marine biosphere, it is essential to understand biological responses at physiological, evolutionary and ecological levels. Here, we contrast and compare different approaches to multiple driver experiments that aim to elucidate biological responses to a complex matrix of ocean global change. We present the benefits and the challenges of each approach with a focus on marine research, and guidelines to navigate through these different categories to help identify strategies that might best address research questions in fundamental physiology, experimental evolutionary biology and community ecology. Our review reveals that the field of multiple driver research is being pulled in complementary directions: the need for reductionist approaches to obtain process‐oriented, mechanistic understanding and a requirement to quantify responses to projected future scenarios of ocean change. We conclude the review with recommendations on how best to align different experimental approaches to contribute fundamental information needed for science‐based policy formulation.     

Assessing the performance of DNA barcoding using posterior predictive simulations

Accurate estimates of biodiversity are required for research in a broad array of biological subdisciplines including ecology, evolution, systematics, conservation and biodiversity science. The use of statistical models and genetic data, particularly DNA barcoding, has been suggested as an important tool for remedying the large gaps in our current understanding of biodiversity. However, the reliability of biodiversity estimates obtained using these approaches depends on how well the statistical models that are used describe the evolutionary process underlying the genetic data. In this study, we utilize data from the Barcode of Life Database and posterior predictive simulations to assess the performance of DNA barcoding under commonly used substitution models. We demonstrate that the success of DNA barcoding varies widely across DNA substitution models and that model choice has a substantial impact on the number of operational taxonomic units identified (changing results by ~4–31%). Additionally, we demonstrate that the widely followed practice of a priori assuming the Kimura 2‐parameter model for DNA barcoding is statistically unjustified and should be avoided. Using both data‐based and inference‐based test statistics, we detect variation in model performance across taxonomic groups, clustering algorithms, genetic divergence thresholds and substitution models. Taken together, these results illustrate the importance of considering both model selection and model adequacy in studies quantifying biodiversity.     

Human behavioral ecology and niche construction

We examine the relationship between niche construction theory (NCT) and human behavioral ecology (HBE), two branches of evolutionary science that are important sources of theory in archeology. We distinguish between formal models of niche construction as an evolutionary process, and uses of niche construction to refer to a kind of human behavior. Formal models from NCT examine how environmental modification can change the selection pressures that organisms face. In contrast, formal models from HBE predict behavior assuming people behave adaptively in their local setting, and can be used to predict when and why people engage in niche construction. We emphasize that HBE as a field is much broader than foraging theory and can incorporate social and cultural influences on decision‐making. We demonstrate how these approaches can be formally incorporated in a multi‐inheritance framework for evolutionary research, and argue that archeologists can best contribute to evolutionary theory by building and testing models that flexibly incorporate HBE and NCT elements.     

Failed refutations: further comments on parsimony and likelihood methods and their relationship to Popper's degree of corroboration

Kluge's (2001, Syst. Biol. 50:322-330) continued arguments that phylogenetic methods based on the statistical principle of likelihood are incompatible with the philosophy of science described by Karl Popper are based on false premises related to Kluge's misrepresentations of Popper's philosophy. Contrary to Kluge's conjectures, likelihood methods are not inherently verificationist; they do not treat every instance of a hypothesis as confirmation of that hypothesis. The historical nature of phylogeny does not preclude phylogenetic hypotheses from being evaluated using the probability of evidence. The low absolute probabilities of hypotheses are irrelevant to the correct interpretation of Popper's concept termed degree of corroboration, which is defined entirely in terms of relative probabilities. Popper did not advocate minimizing background knowledge; in any case, the background knowledge of both parsimony and likelihood methods consists of the general assumption of descent with modification and additional assumptions that are deterministic, concerning which tree is considered most highly corroborated. Although parsimony methods do not assume (in the sense of entailing) that homoplasy is rare, they do assume (in the sense of requiring to obtain a correct phylogenetic inference) certain things about patterns of homoplasy. Both parsimony and likelihood methods assume (in the sense of implying by the manner in which they operate) various things about evolutionary processes, although violation of those assumptions does not always cause the methods to yield incorrect phylogenetic inferences. Test severity is increased by sampling additional relevant characters rather than by character reanalysis, although either interpretation is compatible with the use of phylogenetic likelihood methods. Neither parsimony nor likelihood methods assess test severity (critical evidence) when used to identify a most highly corroborated tree(s) based on a single method or model and a single body of data; however, both classes of methods can be used to perform severe tests. The assumption of descent with modification is insufficient background knowledge to justify cladistic parsimony as a method for assessing degree of corroboration. Invoking equivalency between parsimony methods and likelihood models that assume no common mechanism emphasizes the necessity of additional assumptions, at least some of which are probabilistic in nature. Incongruent characters do not qualify as falsifiers of phylogenetic hypotheses except under extremely unrealistic evolutionary models; therefore, justifications of parsimony methods as falsificationist based on the idea that they minimize the ad hoc dismissal of falsifiers are questionable. Probabilistic concepts such as degree of corroboration and likelihood provide a more appropriate framework for understanding how phylogenetics conforms with Popper's philosophy of science. Likelihood ratio tests do not assume what is at issue but instead are methods for testing hypotheses according to an accepted standard of statistical significance and for incorporating considerations about test severity. These tests are fundamentally similar to Popper's degree of corroboration in being based on the relationship between the probability of the evidence e in the presence versus absence of the hypothesis h, i.e., between p(e|hb) and p(e|b), where b is the background knowledge. Both parsimony and likelihood methods are inductive in that their inferences (particular trees) contain more information than (and therefore do not follow necessarily from) the observations upon which they are based; however, both are deductive in that their conclusions (tree lengths and likelihoods) follow necessarily from their premises (particular trees, observed character state distributions, and evolutionary models). For these and other reasons, phylogenetic likelihood methods are highly compatible with Karl Popper's philosophy of science and offer several advantages over parsimony methods in this context.     

The augmentation algorithm and molecular phylogenetic trees

Summary The augmentation procedure of G.W. Moore leads to correct estimates of the total number of nucleotide substitutions separating two genes descendent from a common ancestor provided the data base is sufficiently dense. These estimates are in agreement with the true distance values from simulations of known evolutionary pathways. The estimates, on the average, are unbiased: they neither overaugment nor underaugment seriously. The variance of the population of augmented distance values reflects accurately the variance of the population of true distance values and is thus not abnormally large due to procedural defects in the algorithm.The augmented distances are in agreement with stochastic models tested on real data when the latter take proper account of the restricted mutability of codons resulting from natural selection.When the experimental data base is not dense, the augmented distance values and population variance may underestimate both the true distance values and their variance. This has a logical consequence that there exist significant and numerous errors in the ancestral sequences reconstructed by the parsimony principle from such data bases.The restrictions, resulting from natural selection, on the mutability of different nucleotide sites is shown to bear critically on the accuracy of estimates of the total number of nucleotide replacements made by stochastic models.     

A macroecological approach to evolutionary rescue and adaptation to climate change

Despite the widespread use of ecological niche models (ENMs) for predicting the responses of species to climate change, these models do not explicitly incorporate any population‐level mechanism. On the other hand, mechanistic models adding population processes (e.g. biotic interactions, dispersal and adaptive potential to abiotic conditions) are much more complex and difficult to parameterize, especially if the goal is to predict range shifts for many species simultaneously. In particular, the adaptive potential (based on genetic adaptations, phenotypic plasticity and behavioral adjustments for physiological responses) of local populations has been a less studied mechanism affecting species’ responses to climatic change so far. Here, we discuss and apply an alternative macroecological framework to evaluate the potential role of evolutionary rescue under climate change based on ENMs. We begin by reviewing eco‐evolutionary models that evaluate the maximum sustainable evolutionary rate under a scenario of environmental change, showing how they can be used to understand the impact of temperature change on a Neotropical anuran species, the Schneider's toad Rhinella diptycha. Then we show how to evaluate spatial patterns of species’ geographic range shift using such models, by estimating evolutionary rates at the trailing edge of species distribution estimated by ENMs and by recalculating the relative amount of total range loss under climate change. We show how different models can reduce the expected range loss predicted for the studied species by potential ecophysiological adaptations in some regions of the trailing edge predicted by ENMs. For general applications, we believe that parameters for large numbers of species and populations can be obtained from macroecological generalizations (e.g. allometric equations and ecogeographical rules), so our framework coupling ENMs with eco‐evolutionary models can be applied to achieve a more accurate picture of potential impacts from climate change and other threats to biodiversity.     

The method of parsimony: an experimental test and theoretical analysis of the adequacy of molecular restoration studies

The ability of the principle of parsimony to accurately reconstruct molecular evolutionary pathways from an analysis of amino acid or nucleic acid sequences from extant organisms is tested by direct comparison with a known pathway. Topological errors occur under specified conditions. Importantly, given no errors in the topology, and error-free experimental sequences, the ancestral sequences inferred by the parsimony principle err significantly, the magnitude of the error increasing with the distance of the nodal sequence from the present. These errors are irreducible as an inherent consequence of any evolutionary process in which chance processes operate within the constraints imposed by Darwinian selection. Formulae are derived which predict the errors in the ancestral sequences from a knowledge of only the internodal distances. The parsimony solution is not a reliably good solution. It is necessary to develop a detailed understanding of the interaction between chance processes and natural selection to further advance our understanding of molecular change in proteins and nucleic acids.     

Biogeography and ecology: towards the integration of two disciplines

Although ecology and biogeography had common origins in the natural history of the nineteenth century, they diverged substantially during the early twentieth century as ecology became increasingly hypothesis-driven and experimental. This mechanistic focus narrowed ecology''s purview to local scales of time and space, and mostly excluded large-scale phenomena and historical explanations. In parallel, biogeography became more analytical with the acceptance of plate tectonics and the development of phylogenetic systematics, and began to pay more attention to ecological factors that influence large-scale distributions. This trend towards unification exposed problems with terms such as ‘community’ and ‘niche,’ in part because ecologists began to view ecological communities as open systems within the contexts of history and geography. The papers in this issue represent biogeographic and ecological perspectives and address the general themes of (i) the niche, (ii) comparative ecology and macroecology, (iii) community assembly, and (iv) diversity. The integration of ecology and biogeography clearly is a natural undertaking that is based on evolutionary biology, has developed its own momentum, and which promises novel, synthetic approaches to investigating ecological systems and their variation over the surface of the Earth. We offer suggestions on future research directions at the intersection of biogeography and ecology.     

Mechanistic models of population-level phenomena

This paper is about mechanisms and models, and how they interact. In part, it is a response to recent discussion in philosophy of biology regarding whether natural selection is a mechanism. We suggest that this debate is indicative of a more general problem that occurs when scientists produce mechanistic models of populations and their behaviour. We can make sense of claims that there are mechanisms that drive population-level phenomena such as macroeconomics, natural selection, ecology, and epidemiology. But talk of mechanisms and mechanistic explanation evokes objects with well-defined and localisable parts which interact in discrete ways, while models of populations include parts and interactions that are neither local nor discrete in any actual populations. This apparent tension can be resolved by carefully distinguishing between the properties of a model and those of the system it represents. To this end, we provide an analysis that recognises the flexible relationship between a mechanistic model and its target system. In turn, this reveals a surprising feature of mechanistic representation and explanation: it can occur even when there is a mismatch between the mechanism of the model and that of its target. Our analysis reframes the debate, providing an alternative way to interpret scientists’ “mechanism-talk”, which initially motivated the issue. We suggest that the relevant question is not whether any population-level phenomenon such as natural selection is a mechanism, but whether it can be usefully modelled as though it were a particular type of mechanism.