Combining Molecular, Macroevolutionary, and Macroecological Perspectives on the Generation of Diversity

Abstract

Charles Darwin presented a unified process of diversification driven by the gradual accumulation of heritable variation. The growth in DNA databases and the increase in genomic sequencing, combined with advances in molecular phylogenetic analyses, gives us an opportunity to realize Darwin's vision, connecting the generation of variation to the diversification of lineages. The rate of molecular evolution is correlated with the rate of diversification across animals and plants, but the relationship between genome change and speciation is complex: Mutation rates evolve in response to life history and niche; substitution rates are influenced by mutation, selection, and population size; rates of acquisition of reproductive isolation vary between populations; and traits, niches, and distribution can influence diversification rates. The connection between mutation rate and diversification rate is one part of the complex and varied story of speciation, which has theoretical importance for understanding the generation of biodiversity and also practical impacts on the use of DNA to understand the dynamics of speciation over macroevolutionary timescales.

DARWIN'S VIEW OF SPECIATION

A general description of the evolutionary process is applicable to all levels of complexity, including the chemical level.

—Dayhoff and Eck (1969)

Charles Darwin tried out many alternative titles for the book describing his revolutionary theory of evolution. For example, a version sent to his friend and mentor Sir Charles Lyell in March 1859 was titled An Abstract of an Essay on the Origin of Species and Varieties Through Natural Selection (Tanghe 2019). At some point that year, he considered a simpler title, On the Mutability of Species (van Wyhe 2002). The book was eventually published in November 1859 under the lengthy descriptive title, On the Origin of Species by Means of Natural Selection, or the Preservation of Favoured Races in the Struggle for Life.

Curiously, some evolutionary biologists have claimed that Darwin's title was misleading (see Mallet 2008). Notably, Mayr (1970) stated that:

It is a familiar and often-told story how Darwin succeeded in convincing the world of the occurrence of evolution and how—in natural selection—he found the mechanism that is responsible for evolutionary change and adaptation. It is not nearly so widely recognized that Darwin failed to solve the problem indicated by the title of his work. Although he demonstrated the modification of species in the time dimension, he never seriously attempted a rigorous analysis of the problem of the multiplication of species, of the splitting of one species into two.

Yet the generation of diversity through the formation of new species was a central focus of the work that Darwin consistently referred to as “my Species Book.”¹ The draft titles for the book that came to be known as The Origin of Species hint at reasons why Darwin's hypothesis of species formation was not accepted by all researchers. These titles highlight Darwin's emphasis on the origin and accumulation of heritable variation as a force driving continuous divergence over time, providing an explanation of the origin of both varieties and species.

Darwin described a single process of divergence that begins with the generation of heritable variation. Differential survival of individuals by virtue of this variation leads to the accumulation of heritable changes, driving gradual divergence of character. Darwin's draft title—An Abstract of an Essay on the Origin of Species and Varieties Through Natural Selection—lumps species and varieties together because he considered both to be outcomes of the same underlying process of divergence. In Darwin's view, varieties are incipient species and species are well-marked varieties. Because of this, Darwin did not see the need to draw on any special processes associated with speciation: “… no clear distinction has been, or can be, drawn between species and well-marked varieties. It cannot be maintained that species when intercrossed are invariably sterile, and varieties invariably fertile” (Darwin 1869). It is the lack of a special process of speciation that Mayr objected to: “If one adopted Darwin's claim, then the problem of the origin of new species was virtually solved: every variety was at the threshold of becoming a new species, and it required only a little push of natural selection to complete the process” (Mayr 1992).

Darwin's unified theory of evolution is shown in the only figure in The Origin of the Species (Fig. 1). This figure shows how, at every time step, the descendants of a particular parent or ancestor are varied, so that they “fan out” from the ancestral type. Selection for extreme forms within the range of variability, driven by competition, will drive divergence, but divergence can also accumulate through chance fixation of neutral traits (Darwin 1859, p. 92). Darwin applies the same explanation to all levels of evolution. His description of the process is similar whether he is considering individuals—“the varying offspring of each species will try (only a few will succeed) to seize on as many and as diverse places in the economy of nature as possible”—(Darwin and Wallace 1858, p. 629) or species—“the more diversified the descendants from any one species become in structure, constitution, and habits, by so much will they be better enabled to seize on many and widely diversified places in the polity of nature, and so be enabled to increase in numbers” (Darwin 1859, p. 112). The same description also applies to higher-level lineages—“The more diversified in habits and structure the descendants of our carnivorous animal became, the more places they would be enabled to occupy” (Darwin 1859, p. 113). He even gives the same description for the macroecological level of the diversity of a region—“for more living beings can be supported on the same area the more they diverge in structure, habits, and constitution” (Darwin 1859, p. 128).

Figure 1.

Darwin's view of speciation is captured in the only illustration in The Origin of Species, which demonstrates Darwin's theory that selection favoring a reduction of competition would drive divergence and eventually generate diversity. (Image reproduced from Darwin 1859.) For additional information see darwin-online.org.uk.

There have always been objections to Darwin's insistence that species production is due solely to the gradual accumulation of small heritable variants. Firstly, since Darwinian speciation is a gradual process, we generally cannot witness the formation of a new species, making it hard to directly prove Darwin's gradual mechanism. Even Thomas Henry Huxley, that staunch defender of Darwin who was “prepared to go to the stake, if requisite” in support of chapters of The Origin on the fossil record and distribution of species (Huxley 1859), considered that Darwin lacked this crucial proof of his unified hypothesis (Huxley 1860). Secondly, many have argued that the differences between species are of a different kind from the variations that accumulate in populations or differ between varieties, so we cannot simply “scale up” population-level processes to explain the origin of discrete species. If this is the case, then we cannot explain macroevolution as the cumulative outcome of microevolution left to run over long timescales. Bateson (1894) summarized this view with the aphorism “variation is evolution.” In this mutationist view of evolution, it is the generation of special species-level variation that generates diversity, not the accumulation of the ordinary variation that accrues in populations every generation (Bateson 1894). Thirdly, critics insisted that Darwin had failed to provide a mechanism for the acquisition of reproductive isolation between populations (e.g., Gray 1860). This objection arose because Darwin considered that reproductive isolation did not require a special explanation, because it was “not a special endowment but incidental on slowly acquired modifications” (Darwin 1859). In Darwin's theory, it was the process of divergence itself that resulted, eventually, in the inability of species to interbreed.

The criticisms raised by Darwin's contemporaries are still debated today, and their resolution is now complicated by discipline specialization. The generation of variation (mutation), changing frequency of variations in populations (microevolution), diversification of lineages (macroevolution), and distribution of diversity in space (macroecology), considered by Darwin as aspects of a single evolutionary process, came to be studied by different kinds of biologists (geneticists, population biologists, paleontologists, systematists, and biogeographers), leading to a degree of disconnection (Fig. 2). But we now have both the data and the analytical tools to connect molecular change to macroevolutionary and macroecological patterns of biodiversity.

Figure 2.

Disciplinary spread: the stages of generation of biodiversity (in bold) are typically the focus of different fields of inquiry (in italics).

Such a connection was foreseen by one of the earliest pioneers of bioinformatics. Margaret Dayhoff's research spanned biochemistry, computing, evolution, and astronomy, and she promoted the idea of exploring the processes of macroevolution by focusing on evolution at the molecular level (see Bromham 2020). Dayhoff realized that comparison of DNA or protein sequences from the same gene in different species could reveal not just the patterns of change at the molecular level (e.g., Eck and Dayhoff 1966) but the history of diversification of lineages (e.g., Dayhoff and Eck 1969). Dayhoff's insight was not only theoretical but practical. By persuading researchers of the value of making their sequences publicly available in shared databases (Strasser 2010), she created the first comparative molecular databases, kick-starting the field of bioinformatics and allowing the comparison of genomic differences at all levels of biological organization. Dayhoff combined her skills in biochemistry and computing to produce the earliest model-based molecular phylogenies, a brilliant leap forward in combining molecular evolution and macroevolution. As molecular databases continue to grow exponentially, we can realize Dayhoff's vision of investigating evolutionary processes at all levels of change, from molecules to biodiversity.

There are many informative reviews of macroevolutionary and macroecological perspectives on speciation dynamics (e.g., Mittelbach et al. 2007; Vamosi et al. 2018; Tobias et al. 2020; Jablonski 2022). This review focuses on a part of the picture that is not always included in macroevolutionary and macroecological perspectives: the role of tempo and mode of molecular evolution in diversification dynamics. To show the connections between genome change, speciation, macroevolution, and macroecology, we can consider a question that links Darwin to Dayhoff: If speciation is driven by the accumulation of heritable variation, then does a greater rate of production of heritable variants increase the rate of generation of species? To explore this possible link, we need a way of comparing the rate of generation of new heritable variants to the rate of generation of new species for a sufficient number of lineages to reveal general patterns. Molecular phylogenies provide a universal framework that allows simultaneous exploration of the tempo and mode of molecular evolution and macroevolution. In this review, we consider how molecular phylogenies can be used to compare both the rate of molecular evolution between lineages and the rate of diversification. Then we will review the patterns that have emerged from molecular phylogenetic studies that compare the rate of molecular evolution to the rate of diversification. We will consider macroevolutionary and macroecological explanations for the link between rates of molecular evolution and rates of diversification. Finally, we will look at the possible practical implications of variation in both the rate of molecular evolution and the rate of diversification on our ability to infer macroevolutionary dynamics from phylogenies.

RATES OF MOLECULAR EVOLUTION VARY BETWEEN LINEAGES

Molecular phylogenies provide a way of investigating variation in the rate of molecular evolution by comparing the accumulation of differences in the same sequences in different lineages. Branch lengths of molecular phylogenies, typically expressed in terms of average substitutions per site for those sequences (or, for a timescaled tree, substitutions per site per million years), are not a direct estimate of mutation rate, because many of the mutations that occur each generation are lost from the population. Instead, sequence differences between species typically represent only those mutations that have been continuously copied from one generation to the next and have risen in frequency to become substitutions, present in all members of the population. Some substitutions will overwrite previous changes, but such “multiple hits” cannot be directly observed from a comparison between sequences. Because multiple hits cannot be counted, substitution rates are inferred using a model that uses the observed pattern of differences between sequences to ask how many changes are likely to have occurred in the past. Therefore, phylogenetic branch lengths do not represent direct estimates of substitution rate but model-based inference (i.e., principled guesses, Bromham 2019). Branch lengths of molecular phylogenies typically reflect only one class of genome change—single-base substitutions that change the sequence of DNA letters—not any other kinds of mutations, such as insertions, deletions, duplications, rearrangements, epigenetic marking, or transposon-driven mutagenesis. Despite these limitations, there are two particularly useful features of phylogenetic estimates of rates of molecular evolution: comparability across lineages and the use of different classes of substitutions to reflect different evolutionary mechanisms.

Molecular phylogenetic studies of variation in the rate of molecular evolution have the advantage that they can be calculated for any lineages for which comparable sequences are available. Increasing availability of genome-scale sequence data may increase the power of comparative studies to detect patterns of rate variation (e.g., Ivan et al. 2022). Genome sequencing is also allowing a new approach to investigating variation in mutation rate by using parent–offspring comparisons to estimate the per-generation mutation rate, reflecting mutations that occurred in gamete formation that did not render the gametes inviable (Campbell et al. 2021; Suárez-Menéndez et al. 2022; Bergeron et al. 2023). Pedigree-based estimates of mutation rates are not directly comparable to phylogenetically derived rates, because they are typically much higher than estimates of the synonymous substitution rate from sequence comparisons between species. This observation is referred to as the time dependency of rate estimates, which might be due to selection, multiple hits, demographic factors, or ascertainment bias (Ho et al. 2011).

Synonymous substitutions (changes to the DNA sequence of a gene that do not change the protein sequence) are expected to be neutral or only weakly affected by selection, so their rate of accumulation will be primarily determined by the mutation rate, although it is possible that some synonymous mutations may be subject to selection, for example, associated with splicing efficiency or mRNA stability (Chamary et al. 2006). Nonsynonymous substitutions (sequence change to a gene that changes the amino acid sequence) may also be neutral (with little or no effect on fitness), but they may also be under either positive or negative selection. For the genes typically included in molecular phylogenies, sites under strong positive selection are unlikely to represent more than a small proportion of the sequence differences. Nonsynonymous changes that are “nearly neutral” (typically with a slightly negative effect on fitness) will be more numerous and, while they will be removed by selection in very large populations, they can accumulate by drift in smaller populations. Nonsynonymous substitution rates can be impacted by mutation rate, selection, and population size (Eyre-Walker and Keightley 2007; Charlesworth 2009; Woolfit 2009). Because nonsynonymous substitutions change protein products, they may be more likely to generate incompatibilities between populations than synonymous changes (Orr and Turelli 2001).

Comparative phylogenetic studies of the rate of molecular evolution have revealed that variation in this rate between lineages is common and associated with a wide range of species traits. Most notably, smaller-bodied animals and plants tend to have faster rates of molecular evolution than their larger-bodied relatives. The cause of this body size effect is a matter of debate. It could be that small-bodied species copy their germline DNA more often per unit time, so accumulate more copy errors, which is consistent with the observation that generation time is one of the most consistent predictors of variation in the rate of molecular evolution (Thomas et al. 2010; Bergeron et al. 2023). But the relationship between generation time, genome copies, and mutation rate is not straightforward (Bromham 2011). An alternative explanation is that the mutation rate is adjusted to balance the relative costs of mutation and repair, which differ with longevity and fecundity (Hua et al. 2015). While species-specific differences in the rate of molecular evolution have often been attributed to differences in metabolic rate, there is little evidence that metabolic rate provides a good explanation for variation in the rate of molecular evolution, above and beyond its variation with other life-history traits, particularly size, generation time, and longevity (Lanfear et al. 2007). In addition to life-history correlates of variation in the rate of molecular evolution, many other species traits have been identified that are significantly associated with substitution rate variation, including location (e.g., depth in fish, Wright et al. 2011), lifestyle (e.g., parasitism in plants, Bromham et al. 2013), locomotion (e.g., flightlessness in insects, Mitterboeck and Adamowicz 2013), environment (e.g., temperature in lizards, Ivan et al. 2022), and behavior (e.g., extrapair mating, Bartosch-Härlid et al. 2003).

In summary, the rate of molecular evolution varies between species, and this variation is shaped by species traits, population dynamics, niche, and environment. Given that some of this variation appears to be best explained by differences in the genome-wide average mutation rate, we can conclude that the rate of supply of heritable variation differs between species. Could this have flowthrough effects to the rate of accumulation of variation in populations, and therefore the rate of lineage diversification?

EVIDENCE FOR LINKS BETWEEN RATES OF MOLECULAR EVOLUTION AND DIVERSIFICATION RATES

Biodiversity is notoriously uneven across time, space, and lineages, and at least some of the variation may be due to different rates of production of new species. Absolute rates of speciation have been estimated from paleontological data (e.g., Etienne and Apol 2009; Pires et al. 2017; Jablonski 2022), but comparable estimates are available for relatively few lineages. Molecular phylogenies provide a universal framework for estimating rates of species accumulation; by analyzing sequences from living (or recently extinct) species, the history of those lineages can be reconstructed, and the rate of addition of lineages over time inferred (Barraclough et al. 1998; Quental and Marshall 2010; Rabosky and McCune 2010). But, just as for rates of molecular evolution, we must be clear what the speciation rates derived from phylogenies represent. The distribution of nodes in a molecular phylogeny will reflect the speciation rate only if the tree contains all lineages descended from a common ancestor, which will only be the case if there has been no extinction and all living lineages are sampled. If not all lineages are included, due to incomplete sampling, extinction, or lack of taxonomic recognition of some species, then not all speciation events will be represented by nodes in the tree. Thus, phylogenies typically reflect the net diversification rate, the result of creation of new lineages by speciation, and loss of lineages by extinction (as long as sampling is complete, or at least not biased, Mynard et al. 2023).

Phylogenetic studies of correlates of diversification rate have been used to identify many correlates of variation in speciation rate, for example, diversification rate is lower in plants with specialized pollinators (Day et al. 2016) and birds with high dispersal ability (Claramunt et al. 2012), and higher in color-polymorphic birds (Hugall and Stuart-Fox 2012) and fish with low dispersal opportunity (Tedesco et al. 2017). But it is not straightforward to apply the same phylogenetic comparative approach to look for an association between the rate of molecular evolution and the rate of diversification, because both aspects of the evolutionary process are intertwined in molecular phylogenies. When diversification rate and molecular rate are both inferred from branch lengths of molecular phylogenies, they both are dependent on phylogenetic models that typically include assumptions about the distribution of node heights (aka the “tree prior” in Bayesian analyses) and how molecular rates vary between branches (the “rates prior,” see Bromham et al. 2018). These models are statistically convenient rather than biologically informed, yet changing the assumptions can alter branch lengths and the relative distribution of nodes (Bromham 2019), which can affect the accuracy of inference of diversification rates from molecular phylogenies (Duchêne et al. 2017). Estimates of the rate of molecular evolution and diversification from the same phylogenies are conflated because they are both inferred from the same data using the same (sometimes questionable) assumptions.

Sister pairs—phylogenetically nonoverlapping pairs of lineages that share a common ancestor to the exclusion of all other such pairs in the analysis—offer a simple way to avoid some of these difficulties of inferring node heights and branch lengths. For pairs of sister clades, we can ask whether the clade with more species tends to accumulate more substitutions in the same amount of time since they last shared a common ancestor (Barraclough et al. 1996; Lanfear et al. 2010b). For sister pairs analyses, species richness can be derived from any source, such as taxonomic lists of the number of species in a given lineage, avoiding the need for complete phylogenies. We do not need measures of absolute time, because we know both of the sister lineages of a pair started from the same common ancestor, so any difference in species number reflects a difference in speciation rate, extinction rate, or both (Cardillo 1999). Similarly, any difference in genetic distance between the pair of lineages implies that one lineage acquired substitutions at a faster rate than the other (as long as we have adequately inferred the relative number of overwritten substitutions). Each sister comparison contributes one phylogenetically independent data point to a test of the generality of the pattern that the sister lineage that has accumulated more species has also accumulated more substitutions (Barraclough et al. 1998).

Molecular phylogenetic analyses, including sister pairs studies, have revealed a common and consistent relationship between the rate of molecular evolution and the diversification rate, for example, in plants (Barraclough and Savolainen 2001; Davies et al. 2004; Lancaster 2010; Duchene and Bromham 2013; Bromham et al. 2015), foraminifera (Ezard et al. 2013), fish (Ritchie et al. 2022a), birds, and other reptiles (Eo and DeWoody 2010; Lanfear et al. 2010a; Iglesias-Carrasco et al. 2019). These observations are unlikely to be due to bias in estimating substitution rates because they have been reported from studies that use a wide range of sequences and different measures of the rate of molecular evolution across a range of taxonomic levels from genera within a single family of plants (Duchene and Bromham 2013) to comparisons between animal phyla (Fontanillas et al. 2007). The diversity of studies also suggests that the result is unlikely to be due to bias in the estimation of speciation rates, because a range of measures has been used, including taxonomic counts (Lanfear et al. 2013; Ritchie et al. 2022a), fossil data (Ezard et al. 2013), and estimates derived using a phylogenetic birth–death model (Eo and DeWoody 2010). Similar results have been found using a range of methods, including phylogenetic path lengths (Webster et al. 2003; Pagel et al. 2006; Ezard et al. 2013; Ritchie et al. 2022a), sister pairs (Barraclough et al. 1996; Barraclough and Savolainen 2001; Bromham et al. 2015), phylogenetically independent contrasts (PICs, Lancaster 2010), and phylogenetic generalized least squares (PGLS, Iglesias-Carrasco et al. 2019). Given the wide range of studies that have reported a link between diversification rates and molecular rates, for different taxa and for many genes, it seems reasonable to conclude that it is a common phenomenon, although possibly not universal (Webster et al. 2003; Pagel et al. 2006; Goldie et al. 2011).

POSSIBLE CAUSES OF THE LINK BETWEEN MOLECULAR EVOLUTION AND DIVERSIFICATION

How should we interpret the association between the rate of molecular evolution and the diversification rate? One explanation is that it reflects an acceleration of molecular evolution at the time of speciation, either through increased fixation of nearly neutral substitutions in small founder populations or through substitution of adaptive variants as populations adapt to new niches (Webster et al. 2003; Pagel et al. 2006). Speciation could potentially affect the mutation rate if population bottlenecks associated with speciation lead to fixation by drift of mutations that decrease copy fidelity or reduce DNA repair (Lynch et al. 2016), generating a burst of changes that adds to variation between lineages in the rate of molecular evolution (Janzen et al. 2022). However, to explain the observed patterns, a significant reduction in DNA repair efficiency would need to accompany most speciation events in a diverse array of plants and animals, and the effect would need to be substantial enough to influence estimates of substitution rate across a wide range of genes. In this view, speciation is special: bursts of substitutions around the time of speciation have a substantial effect on a lineage's rate of molecular evolution (Venditti and Pagel 2010). But there has not yet been a clear theoretical case made for this direction of causality, nor has empirical evidence been presented that this is the preferred explanatory path. Further work would be needed to make a case that speciation-accelerated mutation or substitution rates can account for the observed patterns of molecular evolution.

Alternatively, an increased rate of molecular evolution could drive higher rates of diversification. One consistent form of evidence for this direction of the causal arrow is that classes of substitutions that are generally considered to be neutral are associated with diversification rate, including synonymous substitutions (Barraclough and Savolainen 2001; Lanfear et al. 2010a; Duchene and Bromham 2013; Bromham et al. 2015; Ritchie et al. 2022a) and substitutions in nontranscribed sequences (Lancaster 2010). Diversification rates are also associated with substitutions at third codon sites, where two-thirds of base changes are synonymous (Venditti and Pagel 2010). Since the rate of change at neutral sites should be determined by the mutation rate (Kimura and Ohta 1971), these observations are interpreted as reflecting an association between diversification rates and the average mutation rate across the genome.

Since the average mutation rate varies between lineages, this could have flow-on effects to differences in diversification rate (Fig. 2). An increase in the average mutation rate should increase the rate of neutral substitutions, but it could also increase the rate of nearly neutral substitutions, which have a relatively minor effect on fitness and thus can be fixed by drift in smaller populations. Potentially, an increase in the rate of mutation could increase the rate of adaptive substitutions, if adaptation is mutation-limited or where standing variation is related to the mutation rate. For example, a higher mutation rate is associated with the emergence of new variants of concern in the SARS-CoV-2 virus (Tay et al. 2022). But adaptive substitutions are likely to make a very small contribution to branch lengths in most molecular phylogenies (Eyre-Walker and Keightley 2007) and thus seem unlikely to explain the observed link between the rate of molecular evolution and the diversification rate.

An increased rate of substitutions could result in the more rapid acquisition of reproductive incompatibility between populations. If substitutions that are neutral, nearly neutral, or advantageous in one population are deleterious in combination with substitutions from a different population, then the chance of forming a viable hybrid should decrease with an increasing number of substitutions, although the shape of this relationship might vary (Welch 2004). If increasing genetic divergence between parents results in reduced hybrid viability, it will reduce gene flow between populations, potentially leading to reproductive isolation between populations (Edmands 2002). However, some modes of speciation could weaken the association between substitution rate and speciation rate. If selection favors behavioral adaptations that reduce interpopulation mating, for example through mate recognition, ecological specialization, or temporal partitioning, then reproductive isolation could be triggered by substitutions in relatively few loci (e.g., Kronforst et al. 2013) and therefore might show little association with the overall substitution rate inferred from sequences used in molecular phylogenies. Species might also form rapidly with few substitutions if there are genomic changes that make large contributions to incompatibility, for example chromosome rearrangements or changes in ploidy that make hybrids between different karyotypes inviable (Presgraves 2010; Baack et al. 2015; Lucek et al. 2023).

An association between substitution rate and reproductive isolation has been observed for many taxa, sometimes in the form of a predictable relationship between genetic distance and hybrid incompatibility (speciation clock), for example in plants (e.g., Ferriol et al. 2023), fish (e.g., Bolnick and Near 2005), and amphibians (e.g., Wiens et al. 2006; Dufresnes et al. 2021). Such a relationship has been described by comparing genetic distance between pairs of taxa to gamete viability (e.g., Moyle and Nakazato 2010), embryo viability (e.g., Lijtmaer et al. 2003), or hybrid sterility (e.g., Liti et al. 2006). In such cases, the gradual acquisition of substitutions in a large number of genes seems to provide a realistic model for speciation (Seehausen et al. 2014; Dufresnes et al. 2021). But the form of the relationship between substitutions and incompatibility varies between case studies, from a linear relationship to a greater-than-linear “snowball,” or more complex dynamics, influenced by evolutionary mechanisms and genetic architecture (Gourbière and Mallet 2010; Kulmuni et al. 2020). For example, Mediterranean orchids with food-deceptive pollination strategies show a steady accumulation of postzygotic incompatibilities with time, consistent with a “speciation clock,” but sexually deceptive species in the same group, in which pollinator preference can be influenced by small numbers of changes to key genes, show no such relationship, and neither group shows a relationship between prezygotic isolation and genetic divergence (Scopece et al. 2007).

It seems reasonable to assume that, as the formation of independent species must ultimately rest on the acquisition of heritable traits that prevent interbreeding between populations, the rate of acquisition of genomic incompatibility should also increase the speciation rate, if the reduction in gene flow between populations allows continued divergence and reduces the chance that incipient species will collapse back into their parent lineage. Yet much remains to be done to provide empirical evidence to inform models of speciation completion (Kulmuni et al. 2020; Matute and Cooper 2021). In particular, there have been few studies that have directly compared rates of substitution to the formation of reproductive incompatibility to the average speciation rates across different lineages. A comparison of rates of acquisition of reproductive incompatibility against macroevolutionary estimates of speciation rates in birds and fruit flies found no clear relationship (Rabosky and Matute 2013). An alternative explanation for the link between substitution rate and diversification rate is through selection, if, as Darwin proposed, speciation is commonly driven by genetic changes associated with ecological divergence, reducing competition between nascent species and allowing for population growth and/or stability (Rieseberg and Willis 2007; Marques et al. 2019; Hernández-Hernández et al. 2021). However, it seems doubtful that this would result in accelerated substitution in the “housekeeping genes” typically included in phylogenetic studies.

Phylogenetic tests of the association between molecular evolution and macroevolution use species number, clade size, or number of nodes as an indicator of diversification rate. While it is generally assumed that the relationship is driven by differences in speciation rate, we should also consider whether the relationship could be driven, wholly or in part, by a link between mutation rate and extinction rate. To explain the observed positive association between net diversification and rates of molecular evolution, extinction rates would need to be lower in lineages with higher mutation rates. For example, it could be that lineages with a higher average mutation rate tend to have greater levels of standing genetic variation, which could provide a buffer against extinction in times of rapid environmental change (Lande 1995; Frankham 2005). Alternatively, if smaller populations have a higher risk of extinction, there could be an indirect association with mutation rates if larger-bodied species tend to have both lower mutation rates and smaller populations (Ohta 1972; Brüniche-Olsen et al. 2021). We currently lack evidence to evaluate these possible links between variations in mutation rates and extinction rates.

EVIDENCE LINKING MOLECULAR EVOLUTION, MACROEVOLUTION, AND MACROECOLOGY

The association between faster molecular evolution and greater species richness is correlational evidence that does not in itself reveal the cause of the association. However, there are good reasons to believe that increasing the genome-wide mutation rate could have flow-on effects on the speciation rate, and, thus far, there is little evidence or theory to support the opposite direction of causality, from speciation to genome-wide substitution rates. Alternatively, molecular rates and speciation rates could also be indirectly related through their association with a third variable, such as environmental energy, body size, or population size (Davies et al. 2004).

Several macroecological theories explain spatial variation in biodiversity through an environmental effect on mutation rates, either directly, via mutagenic effects of higher temperatures and ultraviolet (UV) light, or indirectly, via an influence of the environment on generation time or metabolic rate (Rohde 1992; Allen et al. 2006; Gillman and Wright 2014). These theories are compatible with observed patterns linking the rate of molecular evolution and the diversification rate. These macroecological hypotheses view the mutation rate as a predictable consequence of environmental energy, so that the speciation rate is predictable from a relatively small number of parameters. But the association between environment and molecular evolution is complicated because mutation rates are not simply passive reflections of rates of damage from extrinsic (e.g., UV light) or intrinsic (e.g., metabolic damage or copy errors) sources but evolve in response to the environment and life history (Bromham 2011).

For example, UV light causes a specific class of mutations (thymine dimers) as well as general DNA damage (through oxidative damage). Cells have a variety of response mechanisms (Britt 2004), including photoreactivation (specialized enzymes reverse dimers), mismatch repair (removes and replaces damaged bases), and SOS repair (damaged DNA patched via double-strand breaks and crossover). The variety of repair pathways within each cell differ in their efficiency and error-proneness, but the efficiency of UV repair also differs between species. At the extreme, bacteria living in high-altitude lakes, exposed to high UV as well as salt and heavy metals, have highly efficient DNA repair because otherwise, they could not persist in such a highly mutagenic environment (Albarracín et al. 2012). DNA repair efficiency varies with altitude in birds (e.g., Qu et al. 2021), with latitude in Drosophila (Svetec et al. 2016), and with water transparency in Daphnia (Miner et al. 2015). In plants, UV repair efficiency can vary both obligately and facultatively, and in many cases, repair levels can be increased in response to high UV exposure (Frohnmeyer and Staiger 2003). Organisms evolve in response to the level of threat of UV damage by investing more in damage mitigation to reduce mutation rate (e.g., Alton and Franklin 2017). As a result, rates of mutation estimated from phylogenies do not appear to increase with habitual environmental UV exposure (Davies et al. 2004; Bromham et al. 2015).

Similarly, temperature has been proposed as a driver of diversification rate via a direct influence on mutation rate (Rohde 1992) or indirectly by increasing metabolic rate, which generates more DNA-damaging reactive oxygen species (ROS) (Allen et al. 2006). Yet, there is a broad range of evolvable responses that can reduce the amount of ROS that “leaks” from the electron transport chain, mop up ROS, or repair DNA damage arising from metabolic by-products. This adjustment of repair responses may be why there is little evidence that species’ basal metabolic rate has explanatory power for variation in the rate of molecular evolution beyond its covariation with body size and generation time (Lanfear et al. 2007). Furthermore, the metabolic rate itself represents species’ adaptation to their thermal environment, rather than a merely passive response to external conditions (Clarke and Johnston 1999; Addo-Bediako et al. 2002). In this way, the selection on life history absorbs some of the variations in the abiotic environment.

Alternatively, variation in the rate of molecular evolution can be viewed as an evolvable part of a species’ life history, representing a balance between competing costs of DNA repair and mutation. “Slow” life history (increased generation length and longevity) is generally associated with lower rates of molecular evolution (Welch et al. 2008; Lanfear et al. 2013), possibly reflecting greater investment in DNA repair (Bromham 2009; Hua et al. 2015). If we view mutation rate as a life-history feature that is shaped by selection on DNA repair and copy fidelity in response to both external and internal mutagenic hazards, then it seems unlikely that we could accurately predict mutation rate (or diversification rate) from a small set of universal environmental parameters (Hua and Bromham 2017). In this case, mutation rate is expected to vary between species due to differences in life history, niche, or lifestyle, and this may influence the rate at which populations accumulate substitutions leading to genetic isolation and, ultimately, speciation.

We can examine the evidence for the different links in this explanatory chain for a number of well-studied taxonomic groups. Flowering plants illustrate how body size could drive a connection between mutation rates and diversification rates (Fig. 3). The rate of molecular evolution in plants is linked to life-history characteristics, including height and growth form (Smith and Donoghue 2008; Lanfear et al. 2013; Bromham et al. 2015). The synonymous substitution rate is correlated with the nonsynonymous substitution rate for all three genomes (chloroplast, mitochondrial, nuclear), consistent with the assumption that an increase in the mutation rate will flow through to an increase in the substitution rate (Bromham et al. 2015). Accumulation of substitutions is correlated with hybrid inviability in at least some plant lineages, although the association between genetic changes and reproductive incompatibility varies with speciation mode (Levin 2012; Baack et al. 2015). Selection for genetic changes that reinforce isolation, for example, traits influencing pollinator behavior, can reinforce the reduction in gene flow between sister species (Moyle and Nakazato 2010; Hopkins et al. 2012). Differences in the speciation rate of plant genera found in different geographic regions correlate with the substitution rate in a non-protein-coding sequence (Xiang et al. 2004), and plant families with more species tend to have faster rates of molecular evolution (Barraclough and Savolainen 2001; Davies et al. 2004; Duchene and Bromham 2013; Bromham et al. 2015). There is a latitudinal gradient in plant height (Moles et al. 2009), and latitude correlates with species richness in plants (Davies et al. 2004). In summary, in flowering plants, there is empirical evidence to support all links in the chain between molecular evolution and diversification, as well as evidence for an association between species richness and the rate of molecular evolution at a range of taxonomic levels (Fig. 3).

Figure 3.

Examples of connections between molecular evolution and diversification rates for flowering plants. For illustration, one example of a published case study is given for each connection (for more examples, see main text): (1) Moles et al. (2009); (2) Lanfear et al. (2013); (3) Bromham et al. (2015); (4) Moyle and Nakazato (2010); (5) Bromham et al. (2015); (6) Scopece et al. (2007); (7) Davies et al. (2004).

Fish show the complex connections between life history, environment, and diversification. In ray-finned fish, variation in life history and dispersal rates shape the relationship between molecular evolution and diversification (Fig. 4). Rate of molecular evolution is negatively correlated with body size in fish, but this pattern does not seem to be driven by metabolic rate or genome copy frequency but may be due to selection to reduce mutation rate in long-lived species (Qiu et al. 2014; Hua et al. 2015; Hirt et al. 2017; May et al. 2020). Fish in deeper waters or at higher latitudes tend to be larger, have longer generation times and higher longevity (Vila-Gispert et al. 2002; Drazen and Haedrich 2012), and this pattern is consistent with rates of molecular evolution being faster in fish in shallower waters and lower latitudes (Wright et al. 2011). Reproductive incompatibility increases with divergence time in many fish lineages (Russell 2003; Bolnick and Near 2005), although with a more complex relationship between genetic change and reproductive isolation in other groups (Lackey and Boughman 2017). Species richness is correlated with both synonymous and nonsynonymous substitution rates in ray-finned fish (Ritchie et al. 2022a), and in freshwater fish species richness correlates with phylogenetic depth, suggesting a relationship between speciation and substitution or time to accumulate substitutions (Miller and Román-Palacios 2021). Selection on sexually selected traits contributes to reproductive isolation (Boughman 2001). Body size is associated with diversification rate in ray-finned fish (Tedesco et al. 2017), suggesting the possibility that the relationship between molecular evolution and diversification rate could be an indirect association mediated by both varying with body size. However, body size does not wholly explain the association between synonymous substitution rates and species richness (Ritchie et al. 2022a). In summary, there is evidence for all links in the chain between the rate of molecular evolution and diversification in ray-finned fish (Fig. 4) (the association between mutation rate and substitution rate does not seem to have been directly tested but both synonymous and nonsynonymous rates are associated with diversification). However, the environmental patterns of diversification are more complicated than for many other taxa, shaped by particular habitats (e.g., coral reefs) and dispersal ability rather than a simpler environmental or latitudinal gradient (Tedesco et al. 2017).

Figure 4.

Examples of connections between molecular evolution and diversification rates for ray-finned fish. For illustration, one example of a published case study is given for each connection (for more examples, see main text): (1) Wright et al. (2011); (2) Drazen and Haedrich (2012); (3) Hua et al. (2015); (4) Bolnick and Near (2005); (5) Ritchie et al. (2022a,b); (6) Boughman (2001); (7) Tedesco et al. (2017).

Birds show how behavior could influence the relationship between molecular evolution and diversification through the impact of sexual selection on the tempo and mode of evolution (Fig. 5). In birds, the rate of molecular evolution is influenced by life history (Mooers and Harvey 1994) and sexual selection (Iglesias-Carrasco et al. 2019). Sexual selection could increase lineage average mutation rates through male-driven evolution, as greater sperm production results in greater DNA copy frequency per generation (Bartosch-Härlid et al. 2003). Hybrid inviability increases with the accumulation of genetic differences in some bird clades (e.g., Lijtmaer et al. 2003). The rate of population divergence is correlated to the lineage average speciation rate in birds (Harvey et al. 2017), although an analysis across many taxa failed to find a consistent relationship between the rate of acquisition of reproductive isolation and the speciation rate (Rabosky and Matute 2013). If sexual selection can influence rates of molecular evolution, and if the rate of accumulation of genetic differences is associated with hybrid incompatibility, then we should expect rates of lineage divergence to correlate with sexual selection. Consistent with this prediction, male-biased sexual selection is associated with higher rates of diversification in birds (Cally et al. 2021). This could contribute to broader geographic patterns in diversification. For example, sexual dimorphism has a latitudinal gradient (Cooney et al. 2022) matched by a greater rate of diversification in birds in the tropics (e.g., Cardillo 1999; Kennedy et al. 2014). Diversification rate is related to rates of molecular evolution in birds (Lanfear et al. 2010a). In summary, there is evidence to link rates of molecular evolution to rates of diversification in birds, particularly via sexual selection, which can increase both rates of molecular evolution and rates of diversification (Fig. 5).

Figure 5.

Examples of connections between molecular evolution and diversification rates for birds. For illustration, one example of a published case study is given for each connection (for more examples, see main text): (1) Cally et al. (2021); (2) Cooney et al. (2022); (3) Bartosch-Harlid et al. (2003); (4) Lijtmaer et al. (2003); (5) Lanfear et al. (2010a,b); (6) Cardillo (1999); (7) Harvey et al. (2017).

PRACTICAL IMPACTS ON ESTIMATES OF DIVERSIFICATION RATE

Estimates of speciation rate can be made from fossil data or taxonomies, but the majority of analyses of speciation rate are based on molecular phylogenies, using the distribution of branching events (nodes) through time. These analyses rely on either absolute or relative molecular dating to place nodes in time. While dates derived from molecular phylogenies or estimates of genetic distance are now ubiquitous in evolutionary biology, all such date estimates rely on assumptions about rates of molecular evolution (Welch and Bromham 2005). For example, Bayesian molecular phylogenetic methods are based on assumptions about the way that substitution rates change over the tree (typically a stochastic model that prefers small rate changes), the distribution of nodes in time (typically a birth–death model that prefers nodes compatible with a constant rate of speciation and extinction), and the relationship between calibrating information and nodes (specified through a prior distribution on calibrated nodes, Bromham et al. 2018). When these assumptions are violated, dates and branch lengths may be seriously in error (Bromham 2019).

Phylogenetic inference of diversification rates relies on assumptions about the way rates of molecular evolution change along the phylogeny. Most molecular phylogenetic methods use a stochastic model of rate change that does not take the influence of species traits into account (Bromham 2019). Substitution rate variation associated with species characteristics (e.g., body size or environment) can cause significant error in branch lengths (Ritchie et al. 2021, 2022a,b). This means that if there are correlates of the rate of molecular evolution that are not taken into account, branch lengths can be incorrectly inferred, influencing the relative placement of nodes in time, so that estimates of diversification rate from molecular phylogenies may be misleading.

Change in the rate of molecular evolution associated with species traits or diversification rates can also result in misleading inference of diversification dynamics (Duchêne et al. 2017; Sarver et al. 2019; Shafir et al. 2020). Errors in node heights (corresponding to the relative timing of speciation events) increase when rates of molecular evolution are linked to speciation rates (Duchêne et al. 2017; Ritchie et al. 2022b). The same problem of interaction between substitution rate, branch length, and estimates of time may influence attempts to describe a “speciation clock” by using genetic distance to estimate time (e.g., Bolnick and Near 2005; Stelkens et al. 2010) or to compare the age of species pairs to infer differences in the tempo and mode of speciation (e.g., Weir and Price 2011). Given the broad uptake of molecular phylogenetic analysis of variation in diversification rate, we will need to develop better ways of evaluating the possible influence of the tempo and mode of molecular evolution on the inference of speciation rates.

OPEN QUESTIONS FOR FUTURE RESEARCH

Enough evidence has now accumulated to put the link between molecular evolution and diversification on the intellectual map for evolutionary biologists. There are three key questions that should be prioritized for further study.

Firstly, could the relationship be an artifact of analysis? The growing number of studies that link substitution rate and diversification rate using a diverse range of measures and methods suggest not, but more thorough investigation would quell the nagging fear that, since both diversification rates and substitution rates tend to be measured from phylogenies, they might both be subject to shared measurement biases.

Secondly, if the relationship is not due to artifacts of analysis, is this a general property of evolving lineages? Nearly every taxonomic group investigated to date has supported an association between rates of molecular evolution and diversification rates; yet, a few cases do not, and the majority of clades have not been investigated. In particular, studies of fungal, microbial, and viral lineages are lacking. If there are exceptions, do they tell us anything about the nature of the relationship between molecular change and diversification in those clades in which it is detected?

Thirdly, if the first point (legitimacy) and second point (generality) are satisfied, what is the underlying cause of the association between the rate of molecular evolution and the diversification rate? Is it primarily due to drift-driven processes generating substitutions that are harmless in one population but deleterious in combination with substitutions from another population? Or does supply of variation enhance the rate of response to selection to adapt to new niches and avoid competition? Or is there an indirect link between diversification rate and molecular evolution, if both are influenced by the same life history or environmental factors?

CONCLUSION

Diversification is a complex process, influenced by a large number of factors including species biology, genomic architecture, environment, interactions with other species, chance, and more. Sufficient evidence has now accumulated that one of the factors contributing to this causal milieu is variation in the average mutation rate between lineages. This gives us an important insight into the process of divergence: supply of variation matters. Darwin (1859) begins The Origin of Species using domesticated pigeons to demonstrate how enhanced reproductive success of individuals with heritable variations can lead to dramatic divergence in form over many generations. Darwin considered that the process of generating diverse varieties of fancy pigeons from a common stock could be scaled up to any level of biodiversity (Fig. 1).

Consistent with Darwin's unified theory of speciation, the more genetic differences between species of pigeons, the less likely they are to be able to successfully interbreed (Lijtmaer et al. 2003), and, as is generally true for birds, the species-rich pigeon lineage has a faster average rate of molecular evolution than its more species-poor relatives (Lanfear et al. 2010a). We now have enough evidence to convince us that, if we wish to understand processes of speciation and determinants of the tempo and mode of evolution, we should consider the mutability of species.