https://orcid.org/0000-0002-7347-5758
Abstract
In this Perspective we show the value of studying living organisms in the field to understand their history. Darwin’s finches are an iconic example of the early stages of speciation in a young adaptive radiation that produced 18 species in little more than a million years. The question they pose is how and why so many species originated and diversified rapidly. A long-term study of four species of finches on the small island of Daphne Major, combined with genomic investigations, provide some answers in terms of extrinsic and intrinsic factors. Beak size and shape, as well as body size, are key heritable features involved in both ecological and reproductive isolation, and their evolution by natural selection was caused by competitor species during prolonged droughts. Introgressive hybridization of related species is rare but recurring, apparently widespread, increases genetic variation, and does not incur a fitness cost. Hybridization can produce a new species. We use a phylogeny based on whole genome sequences of the four finches to infer morphological transitions in their radiation. Several lines of evidence indicate that some species are missing from the early phase of the radiation due to extinction. Combining these results, we re-cast the classical allopatry-then-sympatry theory of adaptive radiation as a competition-selection-hybridization process that generates a diversity of species.
Introduction
Darwin’s finches have been heralded as a classical example of adaptive radiation (Grant 1986, 1999, Schluter 2000, Stroud and Losos 2016, Gillespie et al. 2020) ever since David Lack published his pioneering study (Lack 1945, 1947) (Box 1). Like many other and larger radiations (Rabosky and Glor 2011, Stroud and Losos 2016, Fleischer et al. 2022, Miles et al. 2023, Muñoz et al. 2023, Ngoepe et al. 2023, Title et al. 2024), they pose a problem for evolutionary biologists—the problem of explaining how and why many species diversified relatively rapidly from a shared ancestor (Grant and Grant 2008a). In this Perspective we draw upon discoveries and insights from a long-term field study and more recent genomic analyses to track changes during the radiation and explain how and why it developed.
The finches are a prime example of the early stages of speciation in a young adaptive radiation when ecological divergence has proceeded dramatically without being accompanied by the kind of genetic divergence that impairs fertility and viability between divergent lineages that are seen in some other radiations (Coyne and Orr 2004, Price 2008). At least 18 species evolved from a common ancestor (Fig. 1) in a relatively short time of 1 to 2 Myr (Barker et al. 2015, Lamichhaney et al. 2015): 17 in the Galápagos archipelago (Fig. 2) and one on Cocos Island. They vary in size more than four-fold, from an 8 g warbler finch (Certhidea olivacea and Certhidea fusca) to a 35 g large ground finch (Geospiza magnirostris) and vegetarian finch (Platyspiza crassirostris). They vary conspicuously in beak size and shape, and to a much lesser extent in plumage. Their morphological disparity is unusual and evolved unusually rapidly; it is greater than theoretically predicted and observed in a large and taxonomically diverse sample of species of comparable age (Uyeda et al. 2011, Grant and Grant 2014, Arnold 2023). Also unusual is their behavioural disparity, which includes the use of a tool to extract cryptic arthropods from woody tissues and exploitation of seabirds for blood and the contents of their eggs (Grant 1986, 1999, Grant and Grant 2008a).
Phylogeny of Darwin’s finches. The topology is inferred from a maximum likelihood tree based on all autosomal sites in the genome. Dating of nodes is indicated in thousands of years with confidence intervals. Symbols refer to islands: C (Cocos), D (Darwin), E (Española), F (Fernandina for G. difficilis), F (Floreana for C. pauper), G (Genovesa), I (Isabela), L (San Cristóbal), M (Marchena), P (Pinta), S (Santiago), Wolf (W), and Z (Santa Cruz). Tree finches are in green and ground finches are in blue. Species previously classified as Geospiza difficilis are shown in red. Note C. olivacea and C. fusca are in the genus Certhidea, and C. heliobates, C. pauper, C. pallidus, C. parvulus, and C. psittacula are in the genus Camarhynchus. From Lamichhaney et al. (2015).
Map of the Galápagos islands. From Grant and Grant (2008a).
We begin with the classical theory of adaptive radiation and the essential ingredients of speciation and then extend it to accommodate new discoveries of introgressive hybridization and hybrid speciation from field and genetic studies on one small island, Daphne Major (area 0.34 km2). We conclude by using the results to interpret evolution in the past, thereby bridging the gap between the Daphne Major (contemporary) microcosm and the Galápagos (phylogenetic) macrocosm.
Ecology
The classical theory of adaptive radiation
Adaptive radiations develop by repeated speciation, each one adding to the accumulating morphological and ecological diversity. There is widespread agreement that speciation generally results from genetic differentiation of separate populations in allopatry (or parapatry), followed in many cases by coexistence and further differentiation in sympatry (Mayr 1963, Grant 1986, 1999, Coyne and Orr 2004, Price 2008, Schluter 2009, Nosil 2012): this is the allopatry-then-sympatry or ATS model. Natural selection in the sympatric phase minimizes competition with other species for resources, the likelihood of interbreeding, or both.
Natural selection on Daphne Major Island
All components of the classical theory of adaptive radiation are amenable for study with living organisms in a well-chosen system. Darwin’s finches have several advantages. They are easy to observe, capture, and mark individually; growth is almost complete within 60 days of hatching; lifetime fitness is measurable on small islands; environments are in the natural state on many islands; species occur together and separately on different islands; and no species is known to have become extinct through human activity—at species level the radiation is intact.
A continuous 40-year study of four species of ground finches on Daphne Major (Fig. 3) demonstrated that natural selection and evolution are recurring phenomena under extreme weather conditions. The first selection episode occurred in 1977 during a prolonged drought. The population of Geospiza fortis (medium ground finch) declined by approximately 80% as a result of starvation and size-selective mortality. Finches with large beaks survived better than those with small beaks because they were able to crack open and extract seeds from the large and hard fruits of Tribulus cistoides after the supply of small and easy-to-handle seeds had been severely depleted (Boag and Grant 1981). Measurable evolution occurred in the next generation (Grant and Grant 1993, 2014) because beak size is a highly heritable trait in this species (Boag and Grant 1978, Boag 1983, Keller et al. 2001).
The Daphne finch quartet. From Grant and Grant (2014). Upper left: small ground finch, G. fuliginosa. Upper right: medium ground finch, G. fortis. Lower left: large ground finch, G. magnirostris. Lower right: cactus finch, G. scandens. From Grant and Grant (2014).
This discovery supports the first part of the ATS theory, adaptive change in allopatry. Support for the second part, adaptive divergence in sympatry, came much later, after a large competitor species (Geospiza magnirostris) established a breeding population on Daphne Major in 1982, gradually increased in numbers, and then outcompeted the large members of the resident G. fortis for a diminishing supply of large seeds during a drought lasting from 2003 to 2005. As a result, the morphological difference between the two species increased, and remained enhanced for the next 7 years, hence it is an example of competitively induced character displacement (Grant and Grant 2006). Fitness (survival) of G. fortis individuals varied not only with phenotype (body and beak size) but with genotype. Individuals with the S variant associated with small beaks at HMGA2, a transcription factor locus, survived twice as well as those with the L variant associated with large beaks (Lamichhaney et al. 2016; Fig. 4).
Shift in allele frequencies at the HMGA2 locus in G. fortis caused by selection during a drought in 2004. The alleles were associated with small (blue) and large (red) beaks, respectively. Frequencies before 1983 were estimated from correlations with beak size after 1983. From Grant and Grant (2024).
Ecology of speciation
Why did the Darwin’s finch radiation produce at least 18 species and not, say, just two? The necessary conditions for multiple speciation in adaptive radiations are environments with diverse ecological opportunities, genetic potential for evolutionary change, and sufficient time (Burress and Tan 2017, Meier et al. 2019, Gillespie et al. 2020, McGee et al. 2020). Extrinsic potential for evolutionary change in allopatry is substantial because there are many islands in the Galápagos archipelago and they differ greatly in size, elevation, and ecological conditions (Abbott et al. 1977, Smith et al. 1978, Schluter and Grant 1984, Grant 1986, 1999). And since the islands are within reach of dispersing individuals there is a high potential for coexistence (Grant 1986, 1999). It is the combination of these two geographical factors that makes Galápagos so favourable for the multiplication of species. Where these are lacking there is no speciation. The single and well-isolated Cocos Island has just a single Darwin’s finch species (Lack 1947, Coyne and Price 2000).
Coexistence and limiting similarity
Given rich ecological potential, what sets a limit to the number of species? A plausible answer is a combination of resources, principally food, and competition for them, as well as spatial isolation (Lack 1947). The evidence for competitive constraints on coexistence is indirect, morphological, and consistent with the observed character displacement on Daphne Major. Beak size is an index of diet, and interspecific differences in beak size are indices of dietary differences. Most pairs of coexisting granivore species (fuliginosa, fortis, and magnirostris) are separated by a gap of more than two standard deviations from the mean in both beak length and width (or depth) of each species (Fig. 5). The Daphne Major resident fortis and immigrant fuliginosa (from Santa Cruz or Santiago) pair is instructive because they fail to coexist. They are exceptional in differing by no more than two standard deviations in both length and width. The difference is a threshold that must be exceeded for coexistence of these granivorous species. It is an empirical measure of limiting similarity (Abrams 1983, see also Pimm et al. 2023, for an example with doves).
Limiting similarity of coexisting ground finches. Differences in beak dimensions between coexisting pairs of closely related species—G. fortis and G. fuliginosa (blue) and G. fortis and G. magnirostris (red). Differences are measured as the gap between two standard deviations above the mean of the smaller species and two standard deviations below the mean of the larger species. Negative values (< 0) indicate overlap of the two distributions. The star indicates two populations that fail to coexist: G. fortis and G. fuliginosa on Daphne Major Island. From Grant and Grant (2024).
Genomics of morphological diversity
Species diversified in beak size and proportions, adapted to exploiting different food resources. Genomic studies provide deeper insights into the origins of this diversity by showing how quantitative, polygenic traits evolve independently, despite constraints from positive genetic correlations between traits (Grant and Grant 1994). The genetic potential for adaptive evolution (Naciri and Linder 2020, Yusuf et al. 2020, Bomblies and Peichel 2022) complements ecological potential, and is known to be strong from three observations: the high heritabilities of beak traits and body size in all species that have been studied so far (Grant and Grant 2008a); the presence of just a few regulatory transcription factors with large individual and collective effects on those traits (Enbody et al. 2023; Fig. 6); and selection and/or drift at these loci when population sizes are small (Lamichhaney et al. 2018).
Genome wide association study of beak size (PC1) and shape (PC2), including sex and body weight as covariates (upper), and body size including sex as a covariate (lower). The cut-off for genome-wide significance at -log10(P value) is 7.7 and shown as a horizontal line. IGF2 in the lower figure just falls below the level of significance. Adapted from Enbody et al. (2023).
Several transcription factor loci are involved in gene regulation during beak development (Lamichhaney et al. 2015, Chaves et al. 2016, Lawson and Petren 2017, Rubin et al. 2022), but just six (Fig. 6) explain almost half (45%) of beak size variation among Geospiza species, or 29% after the correlated effects of body size have been controlled (Enbody et al. 2023). Prominent among them are four tightly linked genes including HMGA2 at a locus on chromosome 1A, and another gene, ALX1, 7 Mb away. Tight linkage signifies potentially coordinated function. The first locus, a supergene, is associated primarily with beak size variation (Lamichhaney et al. 2016), and partly independently, with body size variation (Enbody et al. 2023). The second locus, also proposed as a supergene (Almén et al. 2016), is associated with beak shape variation (Lamichhaney et al. 2015, Rubin et al. 2022). Despite the physical proximity of the two loci, beak size and shape vary independently to some degree because a recombination hotspot lies between them (Rubin et al. 2022, Endoby et al. 2023). Independent evolution of beak shape is further modulated by the association of each of the three beak dimensions—length, depth, and width—with different signalling molecules expressed independently in different tissues at different stages of embryonic development (Abzhanov et al. 2004, 2006, Mallarino et al. 2011).
Reproduction
Reproductive isolation
For sustained coexistence, species must be reproductively and ecologically independent, or ‘isolated’ from each other to a large degree (Lack 1947). What are the barriers to interbreeding that isolate the species? For many passerine bird species, the answer would be differences in plumage and courtship behaviour (Price 2008, Price-Waldman et al. 2020, Delhey et al. 2023). The answer for Darwin’s finches is morphology (Lack 1947) and song (Bowman 1983). Experiments with stuffed museum specimens of ground finches have demonstrated species discrimination by appearance alone (Lack 1947, Ratcliffe and Grant 1983a, b), and experiments with playback of tape-recorded song demonstrated species discrimination by acoustic cues alone (Ratcliffe and Grant 1985, Grant and Grant 1989). The two sets of cues function in the context of courtship and mate-choice. Song is learned from the father in association with parental morphology in the first 5 or 6 weeks of life (Bowman 1983) when offspring are being fed as nestlings and fledglings. The barrier for species in the genus Geospiza is thus a combination of genetically inherited (morphology) and culturally inherited (song) traits.
Hybridization
The barrier to interbreeding occasionally leaks. Beginning with observations on Daphne Major in 1976 (Boag and Grant 1984), a major discovery in recent years has been the prevalence of introgressive hybridization in both ground finches (Grant and Grant 1989, 1992, 1996a, Grant et al. 2005, Farrington et al. 2014, Lawson et al. 2016) and tree finches (Kleindorfer et al. 2014, Peters et al. 2017, Kleindorfer and Dudaniec 2020). It was a major discovery because the incidence of hybridization was believed to be negligible, not only in Darwin’s finches (Lack 1947, Bowman 1961, 1983) but among birds in general (Mayr 1963).
An important question is the fitness of the hybrids and their offspring. Interbreeding on Daphne Major and backcrossing to one or other of the parental species occur without a reduction in survival or reproductive fitness (Grant and Grant 1992, 2020). Geospiza fortis occasionally breeds with immigrant fuliginosa and resident scandens, and even though fuliginosa and scandens do not interbreed on this island scandens receives alleles from fuliginosa via the conduit of fortis (Grant and Grant 2020). Alleles thus spread through interspecific networks.
The role of beak morphology in courtship coupled with the fitness of hybrid offspring and backcrosses implies that speciation is ecologically driven by natural selection, and reproductive isolation is an incidental effect (Fisher 1930, Dobzhansky 1940) or a by-product of the morphological divergence that began in allopatry (Grant 1986, 1999, Schluter 1996, 2000, Grant et al. 2000, Grant and Grant 2008a).
Homoploid hybrid speciation
Perhaps the most outstanding discovery of recent years has been speciation by hybridization: outstanding because it was completely unexpected from theoretical and empirical research. A male ground finch immigrated to Daphne Major in 1981, and 2 years later it bred with a fortis female. Genomic analysis identified the male as a Geospiza conirostris (large cactus finch) that had flown from Española Island more than 100 km away in the south-east of the archipelago (Lamichhaney et al. 2018). Backcrossing to fortis ensued, but by generation 3 the members of this new lineage were breeding with each other. Homogamy continued for three generations until the end of the study and without any outbreeding. The lineage was thus functioning as an incipient species, reproductively isolated from both fortis and scandens and showing no sign of diminished fitness (Grant and Grant 2014).
The lineage revealed how a change in proportions can evolve by selection or drift when the population is small. The population exhibited a shift in the allometric relationship between beak and body size. Its members were intermediate between the two parental species in average body size, as expected, but were identical in beak size to one of them, conirostris. Furthermore, they resembled conirostris more than fortis in frequencies of variants at genetic loci associated with beak size (Rubin et al. 2022, Grant and Grant 2024). Oligogenic regulation of beak traits and small population size are likely to have been key factors in this abrupt shift.
The origin of species by hybridization
Theoretically, a new species may arise from hybridization in two ways. The first is by the production of a hybrid species, as just described: hybrid speciation. The second is by unidirectional or reciprocal introgression and the gradual fusion of populations of two species, generating a new one: introgressive speciation. Convergence of fortis and scandens on Daphne Major is a speciation trajectory of this sort. Both genetically and phenotypically the two species differed in 2012 by approximately half the difference that separated them 40 years earlier. Most members of each population breed with each other and not with the other species, but if this process of convergence continues two species will have been reduced to one. Hypothetically this will be a new species if it is reproductively isolated to a large degree when encountering fortis and scandens populations elsewhere in the archipelago.
An amplified theory of adaptive radiation
In the light of these findings on hybridization it is apparent that the classical theory of adaptive radiation is insufficient. With its focus on ecology and geography, the theory does not address the source of genetic variation and so overlooks the important point that speciation occurs by fusion as well as by fission (Grant and Grant 2008b, Meier et al. 2023). The ecogeographical theory of the ATS model should be modified and re-cast as a theory of competition, selection, and hybridization (CSH) that generates a diversity of species. The context remains geographical, but the essential processes are ecological, behavioural, and genetical. Hybridization should be explicitly incorporated, not just because it can generate new species but because introgression increases genetic variance, contributes to the high heritability of beak traits, and enhances the potential for evolutionary change (Grant and Grant 1994). This is part of the reason why the adaptive radiation was relatively rapid. Hybridization can be adaptive (Hedrick 2013).
Introgressive hybridization is important for three additional reasons. First, introgression without loss of fitness shows that species are genetically compatible in the early stages of speciation. Before genetic incompatibilities arise, populations are mainly if not entirely isolated from each other by a pre-mating behavioural barrier to interbreeding that is set up by learning through sexual imprinting (Grant and Grant 1996b, 1997, 1998). Introgression may have been influential throughout Darwin’s finch history, as there is evidence from whole genome studies of gene exchange early in the radiation, and reticulation in the most recent branches (Lamichhaney et al. 2015, Rubin et al. 2022, Enbody et al. 2023).
Second, introgression may influence mutation rates. Carleton et al. (2020) have suggested that hybridization may elevate genetic variation via mutation rates by disrupting mechanisms of transposable element suppression (see also Craddock 2016). Endogenous retroviruses in Darwin’s finches may do the same, although at present their functions are not known (Hill et al. 2022).
Third, introgression explains why some populations are genetically variable in the ecologically significant traits of beak size and shape and body size. In combination with environmental heterogeneity (Grant et al. 1976), introgression explains why phenotypic variation is maintained within populations at different levels in time and place (Grant and Grant 2019, 2024), and how exchange of haplotype blocks of the genome contributes to the formation of new species (Rubin et al. 2022).
Thus, a combined ecological and hybridization theory explains more features of the Darwin’s finch radiation than either does alone. Studies of cichlids (Irisarri et al. 2018, Marques et al. 2019, Meier et al. 2019, 2023) and some other fish species (e.g. Richards et al. 2021) have had similar success in accounting for extensive radiations by adding introgression to ecological opportunity.
Introgression and coexistence
Introgression throws a different light on the ecological basis of coexistence. The failure of fuliginosa to coexist with fortis on Daphne Major is now seen to be due to either competition for small seeds in dry seasons, to hybridization, or more likely to both, since both competition for food and mate discrimination and choice are functions of beak size similarity (Ratcliffe and Grant 1983b, Grant and Grant 1997). The two mechanisms may jointly account for the statistically non-random assemblages of species on islands that are structured by morphology, correlate with island size, and are interpreted ecologically (Abbott et al. 1977, Schluter and Grant 1984).
History
The environment and the increase in finch diversity
It seems highly likely that environmental change contributed to the radiation of Darwin’s finches. The Galápagos archipelago has been anything but stable. New islands were formed by volcanic activity and by repeated depression of sea level at times of glacial maxima, whereas many islands were lost by submergence when sea level was at its highest level during warm inter-glacial periods. The net result was an increase in number of islands over the past 1 Myr, and an increase in number of finches approximately in parallel (Grant and Grant 1996a). Several islands the size of Daphne Major—possible speciation hotspots—may have contributed to the increase in finch diversity (Grant and Grant 1996a, Ali and Aitchison 2014, Geist et al. 2014).
Ecological opportunities for diversification would have increased with the invasion of new plants and arthropods from the South American continent (Grant and Grant 2024). We suggest such invasions occurred about 250 000 years ago when a large number of ground and tree finch species (seven) originated in a concentrated period of about 50 000 years (Fig. 1). Included amongst them is the cactus finch Geospiza scandens. Its evolution may have been triggered by the arrival of the Opuntia cactus at this pivotal time of rapid transition from a glacial maximum to a rapid warming and increase in the number of islands as the sea level rose, followed by a rapid reversal (Kawamura et al. 2007, Ali and Aitchison 2014). A dated phylogeny of Opuntia in Galápagos is needed to test this suggestion.
Morphological diversity of the adaptive radiation
Morphological divergence of species took place primarily along two major axes of structural variation: body (weight) and beak size, and beak shape (Fig. 7). The A group of 11 species in Figure 7 fit this relationship closely, whereas the B group clearly deviate from it. Average beak shape of the group A species varies with average body size owing to different scaling relationships of beak length and depth (or width). Beak depth increases isometrically with body size (weight), whereas beak length increases allometrically at half that rate (Grant and Grant 2024). The net result is an increase in bluntness with increasing body size.
Mean morphology of all Darwin’s finch species. Bill shape of most species (group A) varies systematically in relation to body size. Shape (length/depth) becomes blunter with increasing body size because beak depth increases isometrically, whereas beak length increases allometrically at half the isometric rate (Grant and Grant 2024). Some species of tree and ground finches (group B) deviate markedly from this relationship: Camarhynchus heliobates, Camarhynchus parvulus, Camarhynchus pauper, Camarhynchus psittacula, Geospiza acutirostris, Geospiza fuliginosa, and Geospiza fortis. The remaining 11 species constitute group A: Certhidea olivacea, Certhidea fusca, Pinaroloxias inornata, Platyspiza crassirostris, Camarhynchus pallidus, Geospiza difficilis, Geospiza scandens, Geospiza propinqua, Geospiza septentrionalis, Geospiza conirostris, and Geospiza magnirostris. Two populations of G. difficilis are identified by triangles, one (solid) for the Pinta population in group B and one (empty) for the Santiago population (group A). The link between them is a possible indication of how group B species evolved from group A species. For clarity, only one population of each of the remaining species has been included. The star indicates the approximate position of Asemospiza obscura, the most likely closest modern relative of Darwin’s finches, and the square is at the position of Pinaroloxias inornata (Grant and Grant 2024).
The strong negative relationship points to a constrained upper limit to average beak shape for a given body size. The limit at the constraint line appears to be set by properties of the developmental system that are shared by all members of the Darwin’s finch group (Fritz et al. 2014, Al-Mosleh et al. 2021, 2023). Older radiations of birds such as the honeycreepers of Hawaii (Pratt 2005, Tokita et al. 2017, Navalón et al. 2019, Fleischer et al. 2022) and the vangids of Madagascar (Jønsson et al. 2012, Reddy et al. 2012) have been less constrained in beak shapes and are much more diverse (Grant and Grant 2008a).
Tracking morphological changes
To track the changes during the radiation, we begin with a molecularly-based phylogeny that spans a period of approximately 1 Myr (Fig. 1). As we use it here, the phylogeny is not an end in itself but a means to an end, that is the interpretation of important adaptive transitions as species multiplied.
We assume the ancestral morphological state is approximated by the morphology of Asemospiza (Tiaris) obscura, a strong candidate for the closest living relative of the finches (Sato et al. 2001, Burns et al. 2002, Funk and Burns 2018). It is smaller than most Darwin’s finches and has a blunter beak than all the small species (Fig. 8), which suggests that the ancestor of Darwin’s finches ate small buds, seeds, and fruits. The first derived species in the genus Certhidea is three-quarters the size of this presumptive ancestor-relative, whereas the beak is more than twice as pointed in being both longer and shallower. It eats insects, spiders, pollen, and nectar. Transformation from Asemospiza to Certhidea morphology would have involved changes in these two beak dimensions in opposite directions. This cannot be accomplished simply by antagonistic selection if their beak traits were genetically correlated strongly and positively, as is the case with contemporary ground finches (Grant and Grant 1994). Therefore, the evolutionary route from granivore to insectivore, from Asemospiza (seedeater) to Certhidea (warbler finch), was probably non-linear and took a long time (Figs 1, 7). For simplicity it is represented by straight lines in Figure 7 and by a split in the ancestral lineage at about the mid-point into two, one of which gave rise to the Certhidea branch. The second phase from a hypothetical midpoint, which is close to contemporary Cocos finch morphology, was faster than the first phase because it was less constrained.
Reconstruction of the early stages of the finch radiation. The ancestral state is represented by Asemospiza obscura, and the Certhidea and Platyspiza populations are the derived states. In this hypothetical reconstruction the first Certhidea populations (stage 2) evolved from a Asemospiza-like ancestor via an assumed intermediate represented by a circle (stage 1), which is similar to the modern Cocos finch (Pinaroloxias inornata) (see also Fig. 7). Platyspiza populations evolved much later (stage 3), and at about the same time (see Fig. 1) Cocos Island was colonized by Pinaroloxias. From Grant and Grant (2024).
Missing species and the problem of extinction
A parsimonious assumption for interpreting a radiation is that all species that existed are present now. Definitive evidence in the form of fossils can be used to test the assumption and to estimate ancestral phenotypes. This has been done with Hawaiian honeycreepers (Fleischer et al. 2022), cichlid fish in Lake Victoria (Ngoepe et al. 2023), and other vertebrates (e.g. Bell 1988, Steadman et al. 2015), but sufficiently old fossils are lacking in Darwin’s finches, and so is finch environmental DNA (e-DNA) that might be useful. Instead, indirect evidence must be used, and it indicates extinction has left an imprint on finch radiation.
The large transformation from the ancestral state to Certhidea morphology may have taken place by slow anagenesis over a long time under warmer conditions than today (Grant and Grant 2014), that is from the origin of the Darwin’s finch lineage at about 2.6 Mya (Funk and Burns 2018) to the first split in the phylogeny at 1.4 Mya, according to mtDNA dating (Lamichhaney et al. 2015). First, for this to have happened without the extinction of intermediate species seems unlikely given that several islands were present during that period (Grant and Grant 1996a).
Second, extinction probably occurred in the next period in the history of the finches. After the ancestral Certhidea split into two species (Fig. 1), the next development in the radiation followed between a quarter and half a million years later with the formation of a lineage that split into two. One branch led to modern Platyspiza crassirostris, the vegetarian finch (Lamichhaney et al. 2015), and the other to all other species in the radiation including Pinaroloxias inornata (Cocos finch) and Geospiza difficilis (sharp-beaked ground finch). Remarkably, Certhidea and Platyspiza span the full range of size and beak shape variation of the whole radiation. Since several species occupy this range now, others with intermediate morphologies probably existed in the past, became extinct and were replaced by new species.
Third, later stages in the radiation provide additional indications of extinction. The circled group of tree finch and ground finch species that deviate from the line in Figure 7 evolved relatively recently, in the last 200 000 to 400 000 years. They are not a connecting link between the original Asemospiza-like ancestor and the Pinaroloxias-like species at the midpoint, rather they are a cluster of relatively recently diverged species. Hence, the original evolutionary trajectory was likely to have been reversed. Since several species occupy this morpho-space now, one or more did so earlier and have become extinct.
The early extinctions may have been caused solely by geophysical factors: volcanic activity, climate change, or island submergence. Alternatively, extinction may have been partly caused by competitive interactions with later species. If this was the case it implies that the adaptive radiation took place not only by divergent natural selection on variation in populations but also by a sorting process of selective replacement of some species and retention of others.
Ontogeny of the radiation
The slow beginning of the radiation accompanied by a substantial morphological reorganization is counter to a theoretical expectation of rapid generation of diversity—numbers of species and their disparity—soon after a founding species invades a new region or adaptive zone (Simpson 1953) and encounters abundant ecological opportunities for diversification (Stroud and Losos 2016, Burress and Tan 2017, Lim and Marshall 2017, Meier et al. 2019, Gillespie et al. 2020). A subsequent decrease in diversification rates is the rule (Harmon et al. 2023) to which Darwin’s finches are an exception (Lamichhaney et al. 2015, Reaney et al. 2020, Vinciguerra and Burns 2021). Speciation rate was not uniform, it accelerated in the past 300 000 years (Lamichhaney et al. 2015). Nevertheless, an early burst of differentiation may have occurred, with extinction later removing the evidence (e.g. Rabosky and Lovette 2008, Henao Diaz et al. 2019). Future genomic analyses may allow a test of the extinction hypothesis if extant species retain the signatures of ‘ghost’ lineages now extinct in the form of different rates of evolution in different parts of the genome (Gopalkrishnan et al. 2018, Zhang et al. 2019, Hibbins and Hahn 2022, Pawar et al. 2023).
From microcosm to macrocosm
The present illuminates the past despite the difference in temporal scale. Periods of directional selection on Daphne Major associated with extreme events and interspersed with long periods of no selection, or at most weak fluctuating or stabilizing selection, constitute an example on the microscale of patterns that have been identified on the macroscale: approximate stasis for millennia or longer followed by relatively rapid evolutionary change and speciation when organisms were confronted with new environmental challenges (Saarinen and Lister 2023, Vermeij 2023). As a second example, evolutionary divergence of fortis on Daphne Major across 40 years is equivalent to the divergence of eight of eleven populations of fortis on different islands (Fig. 9). The divergence is a substantial fraction of the differences between this species and its sister species, fuliginosa and magnirostris, that originated about 250 000 years ago (Grant and Grant 2024). These comparisons indicate speciation is potentially rapid and limited more by ecological opportunity than by genetic potential. Together with hybridization as a factor in speciation, the two examples show how Daphne Major is a microcosm of the Galápagos macrocosm of an adaptive radiation.
Time and space equivalence in evolution. Divergence of G. fortis across 40 years on Daphne is compared with mean morphology of conspecific populations on other islands. Daphne Major points (purple) are annual joint means, and means of the other island populations are in blue. The Daphne Major points span a range of values equivalent to the means of eight of the 11 island populations. The symbol X (green) indicates the average position of specimens in museum collections (corrected to live measurements) from the first half of the twentieth century. From Grant and Grant (2024).
Summary
Darwin’s finches are a prime example of the early stages of speciation in a young adaptive radiation that produced 18 species in little more than a million years.
The question they pose is how and why so many species originated and diversified rapidly. Studies of contemporary populations can provide answers.
Evolution of beak size and shape by natural selection is shown to occur under drought conditions.
Interspecific competition determines the direction of evolutionary change.
Beak size and shape are keys features involved in both ecological and reproductive isolation.
Closely related species hybridize with no detectable loss of fitness. Therefore, speciation is ecologically driven, and reproductive isolation is a consequence.
Hybridization can yield a new species in two ways by fusion of genomes.
We use a phylogeny based on whole genome sequences to infer morphological transitions in the radiation. From this we find three lines of evidence suggesting that species are missing from the early phase of the radiation due to extinction.
Extinctions raise the possibility that the adaptive radiation took place not only by divergent natural selection on variation in populations but also by a competitive sorting process of selective replacement of some species and retention of others.
These findings provide guidelines for future research by identifying what needs to be known for a better understanding of why some groups of organisms diversify: the need for fossil or molecular evidence of extinct species and the need for data on past environments and how they change to complement the insights gained by long-term study of populations in their natural environment.
Attempts to understand the evolution of Darwin’s finches began with observations, measurements, and classification that allowed inferences to be made of history (Swarth 1931, 1934, Lack 1945, Bowman 1961). However, reconstructions of phylogenetic relationships were limited by being restricted to phenotypes of evolutionary interest (Givnish 1997) and by a lack of a fossil record. The restriction was only lifted with the advent of statistical tools and laboratory methods for genetic analysis (Petren et al. 1999, Sato et al. 2001, Burns et al. 2002, Lamichhaney et al. 2015). There now exists a genomic basis for determining three fundamental features of the radiation—the number of species, their patterns of relatedness, and origin—(Lamichhaney et al. 2015). The most likely ancestor has been identified as a relative of the seedeater tanagers (Asemospiza) (Sato et al. 2001, Burns and Naoki 2004, Funk and Burns 2018), and hence Darwin’s finches are tanagers (Thraupidae) and not fringillid finches as formerly believed (Lack 1947).
