One Big, Happy Evolutionary Biology Family

To generate robust evolutionary explanations, evolutionary biologists need to work together across the wide diversity of approaches that the field spans. Evolutionary biology generates inferences about events in the unobservable past. As a result, evolutionary explanations must piece together the evidence that is available today to infer processes that are lost in time. Evolutionary biology spans comparative approaches, which focus on variation between species; to population biology, which focuses on variation between individuals; to optimality models, which help discover why some variants prevail over others; and and evo-devo, which helps map the array of variation that can be presented to selection. Producing a maximally supported evolutionary explanation, say, for why mammals tend to have seven cervical vertebrae, why some clades are more morphologically diverse than others, or why spherical succulent plants tend to live in drylands, requires integrating information from all of these subdisciplinary approaches. This piecing together is necessary because each approach has its blind spots, and these blind spots are admirably filled by the strengths of the other approaches.

However, rather than focus on this complementarity and on the strengths of each approach, biologists often sieze on the weaknesses of a method and declare it scientifically deficient. The comparative method for the study of adaptation is the most frequent target of these criticisms. The comparative method for the study of adaptation finds evidence for adaptation in patterns of convergence (Martins, 2000). For example, the fact that all aerial animals have wings is good evidence of selection acting in similar ways in the context of aerial lifestyles. The comparative method has always been treated as the lesser little brother of “truly rigorous” population and experimental biology; see for example the attempt by Mayr (1982, p. 30) to mount a defense of comparative biology from decades of criticism on the part of population and experimental biologists. In his account, population and experimental biologists criticize comparative approaches for relying on observation rather than experimentation, and for leading to conclusions about past conditions and events that are not testable. Mayr notes that experiments rely on observations of the experimental subjects and conditions just as much as comparative studies do. He also notes that non-experimental inferences are subject to testing in numerous ways.

Ever since then, every few years, some biologist offers stern words of consternation over the supposedly egregious limitations of the comparative method. A much-cited example is Leroi et al. (1994), who note that “the adaptive (or nonadaptive) nature of traits cannot be determined from most comparative data” and deride “false hopes about the ability of comparative studies to resolve the processes of evolution” (p. 397), as though any single source of data can “resolve” questions about evolutionary processes. Grandcolas and D’haese (2003) trot out, to an audience well aware of it, the tired admonition that “Patterns cannot be taken as a direct indication of the occurrence of processes” (p. 484). Fuller (2005) note that the “comparative approach can identify the possible selective agents but will not often be conclusive” (p. 392), implying that there is a non-comparative source of information on evolutionary processes that is “conclusive.” Referring to comparative methods, Kluge (2005) “call[s] into question their validity as sources of increased knowledge of adaptation” (p. 653) and suggests that “laboratory and population studies” might somehow be ultimate arbiters of adaptation hypotheses. Maddison and Fitzjohn (2015) say that comparative biology “cannot give definitive tests of evolutionary mechanisms” (p. 127), another suggestion that some source really can provide “definitive” tests. Uyeda et al. (2021) complain that “observational datasets cannot definitively prove causal links between traits—correlation does not equal causation” (p. 1097). Auerbach et al. (2023) suggest that the comparative method “cannot address questions about evolutionary process” (p. 180). The list goes on. These objections all have the general message that the comparative method does “not provide conclusive proof” of adaptation. These objections are unreasonable. There is no single source of information that provides “conclusive proof” of any evolutionary process, and so no evolutionary approach–the comparative method, population biology, or any other– can hope to meet this criterion (De Santis, 2021).

Criticizing a source of evolutionary evidence for not providing “conclusive proof” is detrimental to evolutionary biology. Underestimating the contribution that some sources of direct evidence provide blocks the essential collaboration that must necessarily occur if biologists are to construct well-supported evolutionary explanations. Instead, it is necessary to forge mutual appreciation of the contributions of comparative biology, population and experimental biology, optimality modeling, evo-devo studies of the variants that can and can’t be presented to selection, together with explicit articulation of the untestable historical assumptions that are part and parcel of every evolutionary explanation (Griffiths, 1996). Only in doing so will scientists construct a maximally robust field of evolutionary biology.

To help in forging such a robust field, I turn the tradition of criticizing the comparative method on its head, by (very gently and I hope clearly ecumenically) pointing out the “fatal flaws” in population biology. As far as I can tell, there has never been an examination of population biology quite equivalent to the repeated highlighting of the “shortcomings” of comparative biology. The point that is so important to highlight is that the assumption that gives populational methods their relevance to evolutionary biology is that “the processes observed in a population-level study can plausibly be taken to be operative beyond the tiny slice of experimental time, to explain patterns across species.” This assumption cannot be tested by population methods, but can be tested by the comparative method. The comparative method can’t directly examine intrapopulational variation, heritability, or fitness, but populational methods can, hence the essential epistemic complementarity between the two: neither one has any supremacy over the other and both are crucial.

This exercise illustrates that, just as comparative methods have their weak points, so too population-level studies also have their glaring omissions. After, I show that the weak points in one are admirably addressed by the strengths of the other. This complementarity means that evolutionary biologists, by working together, can piece together robust accounts of evolutionary patterns. Taking a page from the critiques of the comaprative method for the study of adaptation cited above, I focus throughout on the study of natural selection, but my commentary applies to evolutionary explanation at large (indeed, explanation in any historical science (see for example Cleland, 2013)).

Is Population Genetics Evolutionarily Relevant?

Of course population genetics is relevant to evolutionary biology. Population biology broadly, including, among other studies, empirical analyses of genotype and phenotype frequencies within populations of a single species, artificial selection experiments, quantitative genetics, and mathematical modeling of intergenerational genetic change within populations, forms a cornerstone of evolutionary biology (Pigliucci, 2008; Templeton, 2021). It is the theoretical core around which the Modern Synthesis, and thus most of today’s evolutionary biology, is built (Dickins & Dickins, 2018). Its mathematical rigor is held up as a gold standard even by many comparative biologists (e.g. Butler & King, 2004; Steppan et al., 2002). This rigor means that the relevance of population biology to the effort to explain the immense diversity of all of life is often not even questioned (see Pigliucci, 2006). In this way, what exactly gives population genetics its relevance to evolutionary biology isn’t always recognized.

But this recognition is important. Here, I explain that recognizing where the major approaches within evolutionary biology– populational, comparative, and optimality– obtain their relevance to evolutionary biology at large leads to better-supported inferences of evolutionary process. This recognition helps biologists select their sources of direct evidence effectively and articulate the assumptions that they invoke. It shows why collaboration is essential between population, comparative, optimality, and evo-devo approaches, leading to a stronger evolutionary biology. To see where population biology obtains its relevance to evolutionary biology at large– and why we would even question it– we must examine its key assumption, which I explore below. But first, a quick look at why population studies are so appealing for evolutionary biology in the first place.

The Great Advantage of Population Studies

Evolutionary theory tells us that selection operates on heritable variation with fitness consequences among the members of populations. The monumental advantage of population biology is that it is the only means of examining these foundational phenomena directly. For example, body size affects everything from metabolic rate to reproductive output (West et al., 2003). So, it is reasonable to think that natural selection could act on body size, and the only way to examine this notion directly is with the tools of population biology. In Drosophila, researchers have found that body size does indeed vary across the members of a population (Flatt, 2020). What is more, they have found that this variation is heritable, with progeny tending to resemble their parents in body size (Lafuente et al., 2018). This heritable variation involves genetic factors (Oldham et al., 2000), turns out to be highly polygenic (Turner et al., 2011), and has different impacts on fitness given different selective contexts. All of this evidence together shows that in a given population of Drosophila, body size does indeed meet all of the criteria for evolution by natural selection and can evolve adaptively across generations. No other approach in evolutionary biology provides such direct empirical access to the very raw material that evolution by natural selection is made of– variation, heritability, and fitness. But population methods have a singular weakness: their tiny scale (Pigliucci, 2008). What props them up is the key assumption.

The key Assumption that Gives Population Biology its Evolutionary Relevance

Evolutionary biology aims to explain the vast sweep of diversity across all of life, but population biology examines trivially small slices of this immensity (Fig. 1). Even the most remorseless intelligent design miscreants happily admit that changes in gene frequency occur between populations, and that populations can respond to natural and artificial selection (Dembski & Witt, 2010). But making creationists happy is hardly the goal of evolutionary biology. Taken objectively and on their own, the exquisitely detailed descriptions provided by population biology experiments do not do much work at all in explaining the vast diversity of life. Their relevance to this great goal of evolutionary biology comes via the key assumption.

Fig. 1
figure 1

Population biology and the goal of evolutionary biology. Population biology studies examine very small numbers of individuals relative to the number of individuals of the focal species that exist in the wild, and certainly relative to all of the members of the species that have ever existed. In addition to small numbers of individuals, they examine tiny moments in time relative to the history of the entire species. The tiny scale of population biology is thrown into even sharper relief when viewed against the goal of evolutionary biology to explain the entire vast sweep of the living world. Despite its very small scale, population biology is indeed relevant to efforts to explain diversity at large. This relevance hinges on a crucial assumption, which is that the processes observed in micro-slices of populational time are operative along all the branches of the tree of life. The means of testing this assumption is the comparative method

Full size image

The key assumption is that the processes observed in population biology are operative in populations and times not directly observed (Leroi, 2000). In applying it, population biologists assume that relationships between genotype, phenotype, performance, and fitness at least qualitatively similar to those observed in their nanoscopic experiments apply to individuals and populations that they did not examine. Such extrapolations can extend to very great scales. Population biologists would tell us that it is plausible to think that Tyrannosaurus hatchlings in a clutch must have varied among one another just as less-charismatic but more tractable baby fruit flies do in their experiments. They would tell us that along uncounted, unobserved, and unobservable generations, giraffe mommies and daddies with longer necks probably tended to have giraffe babies with longer necks (Olson & Arroyo-Santos, 2015). The individual-level factors that bias or limit development are plausibly extrapolated beyond the study system and along the tree of life (Jones et al., 2003; Steppan et al., 2002). Adaptationists who have taken the admonitions of Gould and Lewontin’s “Spandrels” essay to heart even assume that genetic or developmental “constraints” can be inferred to have existed in the past (Arthur, 2004; Gould & Lewontin, 1979; Losos, 2011; Minelli, 2009; Olson, 2012, 2019, 2021; Olson et al., 2019; Olson & Arroyo-Santos, 2015; Rose & Lauder, 1996; Steppan et al., 2002, 2002; Sultan, 2015).

Assuming that the processes that they observe in their experiments can be extrapolated to evolutionary time at large might seem to require serious audacity on the part of population biologists. No matter how large a population biology experiment is, it only examines a paltry portion of the life of the focal species. Arabidopsis thaliana split from its common ancestor with A. lyrata perhaps 13 million years ago (Beilstein et al., 2010). An Arabidopsis experiment spanning 15 generations (Teng et al., 2009) by any account is an admirable achievement. But it covers just 1.2 × 10− 6 of the history of the species, hardly a proportion that can inspire much in the way of grand generalizations regarding the forces sha** life on earth. Even the longest selection experiments spanning tens of thousands of bacterial generations (Grant et al., 2020) necessarily examine minute slices of evolutionary time. It is true that the ages of bacterial species are notoriously hard to estimate (Kuo & Ochman, 2009), that their generation times can vary widely, and that even what a bacterial species might be are all vexing questions (Riley & Lizotte-Waniewski, 2009). But assuming that E. coli is, as has been estimated, 70 million years old (Kuo & Ochman, 2009; Ochman et al., 1999), and assuming a generation time of one day, then even examining an amazing 50,000 generations in an experiment (Grant et al., 2020) is just a small fraction of the zillions of generations stretching back in bacterial time (1.96 × 10− 6 or so). No matter how ambitious a population biology experiment using even the best-studied model species is, it leaves unexamined most of the individuals and most of the time that the species has been on earth.

The minuteness of population biology micro-slices is thrown into even sharper relief when the scope is widened beyond the model species and viewed in the context of the entire history of life. Explaining this entire history is the ultimate purview of evolutionary biology (Fig. 1), a goal that even the most committed population biologist would cleave to. The fulcrum that all of population biology balances upon, that gives population biologists license to claim that their mini-scale experiments have any relevance to the imposing expanse of evolutionary time, is nothing more substantial than an assumption. It is that processes identical or very similar to those observed directly can be assumed to be operative at every moment along every branch of the tree of life (Griffiths, 1996). As I will presently affirm whole-heartedly, this assumption is entirely justified, but in the interest of well-supported inferences and a better-integrated evolutionary biology, it bears prodding for a moment longer.

Population Biology: Without the Key Assumption, a Descriptive Idyll

The notion that population-level processes isomorphic to those observed in population biology experiments can be assumed to be operative over evolutionary time is so built-in to evolutionary biology that we practicing biologists often do not even notice it. But if you have not already, take a moment to picture a population biology experiment without this assumption. Population biologists take handfuls of individuals and describe their genotypes and phenotypes in meticulous detail. They subject the individuals to differing conditions of selection and document the microscopic changes that their interventions provoke. Remember, the assumption that similar phenomena of variation, heritability, and fitness are operative even in unseen populations is blocked. Without the key assumption that the phenomena observed are operative beyond the experiment, then all we are left with is a painstaking description of a handful of individuals from a micro-moment in the life of a single species. Nothing more. If the aim of science is to produce generalization, then population biology, without the key assumption, is barely science. If a main goal of evolutionary biology is to explain how the diversity of life on earth has arisen, then, in the absence of the key assumption, population biology becomes a collection of isolated anecdotes. Before turning to the ample justification underwriting the key assumption, there is one more point to press.

Population Biology Lacks the Tools to Justify its Evolutionary Relevance

The actual content of population biology experiments provide no means of testing the key assumption that gives the field broad relevance to evolutionary biology. A Drosophila experiment provides data only on the flies examined. It provides no direct evidence whatsoever regarding unexamined populations of flies contemporaneous with the ones studied, and certainly not the uncounted and unobservable ancestors that precede the studied flies. Left alone with only his or her model system, a population biologist has nowhere to turn to justify the key assumption. For the sake of good science, all evolutionary biologists need to be able to explain why the key assumption is justified.

The assumption that the processes observed in micro-slices of population biology time can be extrapolated to life at large is tested by patterns across species that are congruent with what would be expected if they were produced by processes identical or very similar to the processes observed in a population biology study. Take population biologists studying variation in natural populations in the proportions of bird wings (Carvalho Provinciato et al., 2018), or the results of centuries of artificial selection on pigeons (Domyan & Shapiro, 2017). These biologists will tell confident stories about similar heritable variation, with its indisputably dramatic performance consequences, as sha** the diversity of bird lifestyles. Witness albatrosses, which can cover hundreds of kilometers per day on their enormously long and slender wings. Witness forest-dwelling jays, with their short, broad wings ideal for flap** through dense shrubbery. Witness the broad wings and fingertip-like extremes of vulture wings, ideal for lolling all day on rising air. These biologists will invoke stories about heritable intrapopulational variation and its performance and fitness consequences leading to these striking form-function fits across species.

At least albatrosses, jays, and vultures are living today. In justifying the evolutionary relevance of their approach, population biologists will go so far as to tell stories about feathered reptiles scampering on branches more than 150 million years ago (Olson, 2021). They will say that some baby dinosaurs in a clutch had longer feathers, or more appropriate musculature, or better flap** skills than their less-fortunate brothers and sisters. They will tell you that these lucky individuals lived longer or had more offspring. They will fill in as many blanks as necessary between terrestrial feathered dinosaurs to end up with flying modern birds. The bold confidence of population biologists does not stop there. In their appeals to justify the relevance of their work, they will also point to other flying animals, such as insects, pterosaurs, and bats. They will tell stories almost identical to their bird ancestor scenario about the unseen and unseeable ancestral populations leading from terrestrial to flying insects, to pterosaurs from whatever pterosaurs evolved from, and from nonvolant mammals to stupendously able-flying bats (Olson, 2021). In other words, for the evidence necessary to test the assumption that underwrites the evolutionary relevance of their data, population biologists point to comparative evidence. And they are entirely right.

Comparative Biology to the Rescue

Comparative biology provides exactly the evidence needed to test the notion that the processes observable in the tiny windows that population biology provides can indeed be extrapolated beyond the population biology moment, even to interspecific time. Population biology is built on well-articulated mechanistic principles, making it easy to generate clear predictions about the patterns that should be observed on interspecific scales. If variation arises in similar ways generation after generation, if offspring tend to resemble their parents, and if this heritable variation is associated with predictable differences in tendencies to survive and reproduce, then there will necessarily be a fit between form and lifestyle. If there are relatively few ways of making a certain living, and if the possibilities of developmental variation are quite wide, then many organisms will exhibit similar forms even though they started from very different starting points. This is the phenomenon of convergent evolution (Losos, 2011). Convergence provides exactly the evidence necessary to test the notion that population-level processes shape life generally.

An example that is striking because it involves convergence from so many different starting points is the aquadynamic shape of fast-swimming animals. It is very conspicuous that any fast-moving aquatic animal is streamlined in such a way that requires a lower energy investment for a given displacement as compared to chunkier shapes. In this way, everything from fish, whales, seals, turtles, ichthyosaurs, squids, eurypterids, and insects can have (or had) remarkably similar streamlining energetics. This striking convergence is explained beautifully by appeal to the most standard of population-level processes. By virtue of their shape, some individuals born in each generation invest less energy to move a given distance than their conspecifics. These streamlined individuals have more energy to dedicate to esca** predators, to growth, and reproduction. Take a moment to acknowledge the level of detail that this scenario implies not just for one lineage but for all of these aquatic creatures.

Telling stories this detailed for so many lineages would seem outlandishly venturesome, but no one has been able to come up with a plausible alternative explanation. The possible alternative explanations that might be imagined founder before they even get going. Postulate for example that the strictures of development underwater somehow force living matter into an aquadynamic shape. If this were true, then no appeal to adaptation would be necessary. It would simply be the case that there is no other way that animal matter can build itself underwater. The counterexamples, in the form of frogfish, mata-mata turtles, caddisflies, oysters, and other lump-shaped relatives of more aquadynamic species, are so readily at hand that hand-waving about the possibilities development underwater is bound to be listless. No, apart from extrapolation of population-level processes of natural selection, there is simply no other account that so satisfactorily explains the aquadynamic shape of fast-moving aquatic animals. Biologists therefore accept the key assumption because it would explain the observed comparative patterns so well if it were true (Fig. 2) (Olson & Arroyo-Santos, 2015).

We can now restore to population biologists their ability to point to populations of the focal species outside of the studied ones, and even to other species: now evidence justifying the extrapolability of population biological processes abounds. What gives population biology its evolutionary relevance is the comparative method. And this population-comparative dependence goes both ways.

Fig. 2
figure 2

The structure of evolutionary explanations. Explanations in evolutionary biology, indeed all of science, have a “self-evidencing” structure. Self-evidencing explanations have as part of their structure the acceptance of key assumptions as a function of how well they would explain the data if they were true. In the case of evolutionary biology, biologists accept the assumptions that past populations existed and that processes very similar to those observed today were operative. The justification for this assumption is the pattern itself: if selection in past populations favored the presence of wings in the context of an aerial lifestyle in the same way in birds, bats, and insects, then this explains the presence of wings in these lineages today. No better assumption has been proposed, and so evolutionary biologists accept this assumption. The proportional importance of these assumptions can be lessened by adducing as many layers of direct evidence as possible, but the assumptions always remain. Diminishing the relative importance of the assumptions requires as many layers of direct evidence as possible, demanding collaboration between population, comparative, and other approaches, not separation

Full size image

Epistemic Interdependence

Just as population biology needs comparative biology, so comparative biology needs population biology. The comparative method gathers evidence regarding evolutionary process by examining variation across species. Larger species of terrestrial mammals have predictably thicker leg bones (Christiansen, 1999). If this pattern is due to adaptation, then it involves yet another extravagantly detailed story. It is that in each population of all of these species, there has been heritable variation such that other bone thickness-body mass combinations are possible. Presumably bones that are too thin are subject to breakage and their bearers have low fitness. Presumably bones that are too thick are admirably breakage-resistant but their bearers invest excessive energetic resources in just moving around. The solution favored by selection is the one that provides the just-right balance between resistance to breakage, locomotory efficiency, and energetic surpluses directed to reproduction. Whether they acknowledge it or not, in claiming convergent evolution by adaptation, comparative biologists invoke these detailed stories for each of the species they include in a study (Olson, 2021). They simply assume that variation is possible such that other slopes or intercepts of the bone thickness-body mass relationship could be observed were selection to favor a different one. They simply assume that intrapopulational differences in bone thickness are heritable. And they invoke Goldilocks stories of the just-right variant based on the existence of the pattern itself (Fig. 2). With its focus on variation across species, comparative biology is just as hamstrung in testing the assumptions that give it relevance to evolutionary biology as population biology is in testing its own justifying assumption. A key approach that can test the justifying assumption of the comparative method is population biology. The lesson is that neither population biology nor comparative biology can test the assumptions that give them relevance to evolutionary biology, but instead they each need one another (Cox & Logan, 2021; Griffiths, 1996; Mahler et al., 2017; Olson, 2012; Olson & Arroyo-Santos, 2015; Sinnott-Armstrong et al., 2022). Not understanding this epistemic interdependence leads to a lack of the necessary collaboration across comparative and populational perspectives, pointless criticisms of the supposed fatal flaws of the comparative method, and weaker evolutionary explanations. Instead, all well-supported evolutionary explanations are built from multiple layers of direct evidence (Fig. 2). Recognizing the structure of evolutionary explanation reveals that authors who think they have discovered fatal flaws in the comparative method are demanding a standard of informative conclusiveness that not just the comparative method but no single source of scientific evidence can provide (De Santis, 2021).

Conclusion: End any Vestige of a Population-Comparative Schism for a Stronger Evolutionary Biology

The implications of the epistemic dependence between comparative and population biology could not be more important for scientific practice. Both population biology and comparative biology have weaknesses that neither approach can remedy on its own. As a result, rather than criticizing an approach for its weakness, we should look across the evolutionary aisle for the complementary aspects of the other subdiscipline. A maximally supported inference of evolutionary process requires both comparative as well as populational evidence. Indeed, a maximally supported explanation requires information on developmental possibility and optimality modeling as well (Fig. 2). Development because the range of variants that can be presented to selection sometimes excludes, for reasons that have nothing to do with selection, combinations that would be favored if only they could be produced (Badyaev, 2011; Burt, 2001; Olson, 2012). Optimality modeling because which variants have the highest fitness should not be random with respect to performance but instead predictable given basic biophysics (Potochnik, 2009; Vincent & Brown, 2005). Crucial also to recognize is that constructing evolutionary explanations requires assumptions about unseen and unseeable ancestral populations (Griffiths, 1996; Olson & Arroyo-Santos, 2015; Pigliucci, 2006). Because a well-supported account of what happened in these unseeable populations can only be achieved by adducing as much evidence as possible from sources that are as complementary as possible, biologists must recognize the structure of evolutionary explanations and choose their sources of evidence carefully, so as to maximize thair complementarity. Recognizing the vital dependence between populational, comparative, optimality, and developmental evidence, and the assumptions that bind this evidence together in an explanatory framework, should overcome once and for all parochial notions of the superiority of one kind of evidence over another. In the process, biologists vastly improve their efforts to explain the diversity of life on earth.