Although the Galápagos Islands are renowned for the role their biota has played in the development and refinement of Natural Selection (e.g., Darwin 1859; Grant and Grant 2008; Hedrick 2019), they are important to the geoscience community with regards to mantle-plume hotspots, which are a fundamental feature of the solid-Earth system (e.g., Morgan 1971; Harpp et al. 2005; Villagómez et al. 2007; Whattam and Stern 2015; Ali and Meiri 2023; Soderman et al. 2023). A basic requirement of practically all geological investigations is a robust chronological framework. Early data for the Galápagos were provided by Cox and Dalrymple (1966), whose two-pronged radiometric- and magnetostratigraphic-dating programme indicated that the various islands were all, apparently, less than 2 Mya (n.b., material from what is now known to be the oldest island, San Cristóbal, was not examined). Later studies (Bailey 1976; Hall 1983; Geist et al. 1985; Hickman and Lipps 1985) pushed back the maximum age to ca. 4 Mya (Fig. 1). The geological work was then extended by Christie et al. (1992) and Sinton et al. (1996) to the submerged volcanic massifs on the eastern Galápagos platform and to the sea-bed to the east on the Carnegie Ridge’s western end (Figs 2 and 3). The dredge-haul samples they recovered included once-subaerial rocks (rounded clasts, weathered materials, etc.) with some yielding ages ranging from 2.8 to 9.1 Mya. Subsequently, Werner et al. (1999, 2003), Mix et al. (2003) and Sinton et al. (2017, 2018) expanded the coverage and showed that the Galápagos plume had been spawning islands in the vicinity of the Cocos-Nazca plate-spreading ridge since at least the early Middle Miocene (Fig. 2).

Figure 1. 

Map showing the ages of the various islands of the Galápagos Archipelago (figures in parentheses; Geist et al. 2014), plus the estimated time intervals when some nearby seamounts were subaerial (number ranges in brackets; Sinton et al. 2017, 2018). Also depicted are dredge-haul locations were once-subaerial rocks have yielded radiometric age dates (Christie et al. 1992; Sinton et al. 1996; yellow circles with red rims). The base image was created using GeoMapApp (Ryan et al. 2009). The ocean bed down to 1000 m water-depth is shaded white.

Figure 2. 

Map showing the key physiographical features of the eastern equatorial Pacific and the adjacent land areas. The base chart was generated using GeoMapApp (Ryan et al. 2009). The yellow-black dashed lines show the edges of the various tectonic plates based on Bird (2003), although the boundary types (spreading ridge, transform or subduction) are not depicted. Note that the origin of the Malpelo Ridge is contested; Werner et al. (2003) were uncertain whereas Marcaillou et al. (2006) thought it to be a former part of the Cocos Ridge that has now been displaced to the south. Plotted isobaths are for 1000 m and 2000 m. Yellow circles indicate locations where dredge-hauls recovered volcanic rocks that sometime after their formation experienced subaerial weathering/erosion such that today some form rounded pebbles or cobbles (Christie et al. 1992; Sinton et al. 1996). Many such symbols have associated numbers, which denotes the radiometric ages of the recovered clasts (Mya). Red triangle denotes Ocean Drilling Program (ODP) Site 1239, where a 6.8–million-year hiatus was identified between Middle Miocene (c. 14.6 Mya) and the Upper Miocene (c. 7.8 Mya) sedimentary packages (Mix et al. 2003). Notably, the lower unit includes sand grains (0.2–2.0 mm diameter), which is suggestive of nearby land, their mafic compositions indicating a basaltic massif.

Figure 3. 

Bathymetric-topographic profiles across key parts of the eastern equatorial Pacific. The data were downloaded from GeoMapApp (Ryan et al. 2009). (A) and (B) are along the main axes of the Galápagos platform-Carnegie Ridge and the western Galápagos platform-Cocos Ridge respectively, with their traces shown in Figure 2. A third section (C) provides insight into the general morphology of the ocean floor of the eastern Pacific; note its slightly lower base-level. This reflects the increased age of the crust, which formed at the Galápagos Spreading Centre (Fig. 2; also see O’Connor et al. 2007) and thermal-cooling subsidence (Parsons and Sclater 1977; Hillier and Watts 2005; Kearey et al. 2009). Depicted on (A) and (B) are the various age-dated dredge-haul samples that experienced subaerial erosion (see Fig. 3) are in their approximate transferred positions (yellow circles). Concerning the ‘plate up-flexing’, see Turcotte et al. (1978) and Kearey et al. (2009).

Appreciation of the morphology of the Galápagos platform-Carnegie Ridge and the Cocos Ridge can be gleaned from a series of bathymetric cross-sections across the various physiographic highs that are shown in Figures S1, S2, and S3. Significantly, there are sizable tracts of sea-floor that are < 2000 m deep, with a number of patches < 1000 m (Fig. 2), that when the effects of lithospheric-cooling subsidence together with geological time are considered (e.g., Parsons and Sclater 1997; Hillier and Watts 2005; Ali and Aitchison 2014), it is certain that some of these could once have been emergent (see Orellana-Rovirosa and Richards 2018; Ali and Meiri 2023).

One of the problems with biogeographical studies of insular systems is the lack of an analytical framework that can be used to quantitively evaluate the colonization history of a biotic assemblage. This shortfall has, however, been recently addressed through the introduction of ‘Colonization Profile’ analysis (Ali and Hedges 2023). The procedure comprises three stages: (i) establishing colonization intervals for each of the clades, (ii) constructing the colonization profile(s), and (iii) generating equation values for use as both a quality-control metric, as well as for simulating and testing hypothetical arrival histories (e.g., temporally stochastic overwater dispersal, focused influx along a temporary land-bridge).

A clade’s ‘colonization interval’ represents the earliest and latest possible instants the lineages’ ancestors could have arrived on the island/archipelago. The initial estimation uses molecular-clock data, the former value being provided by the stem node’s older uncertainty age, while the latter one is given by the crown node’s younger uncertainty age. If the original publications do not include confidence intervals for the relevant nodes, then the preceding and succeeding divergences (i.e., of the parent and child taxa) are used. Subsequently, it may be possible to refine each clade’s interval using palaeontological data, island-emergence information (for volcanic landmasses constructed upon deep ocean floor) and regional or global extinction events. The first might pull back the latest time of arrival when direct proof of presence predates the crown-young age. The second and third may push forward the earliest time of arrival; a clade could not have been present before the landmass’ birth, nor could its descendants have survived the environmental devastation. Where on-island or on-archipelago divergence has not taken place, or relevant data are unavailable, the younger end of the colonization interval is set at 0 Mya.

A ‘colonization profile’ is then constructed in the form of a histogram by ‘stacking’ all of the colonization intervals. Its shape reflects the overall history of arrivals with spikes and troughs marking influx concentrations and reduced arrivals. The plot is referred to as the ‘actual’ data profile.

Stage three involves calculating equation values for use in quality control and model testing. Here all of the colonization intervals are plotted against their colonization interval mid-points (CIMP). This is followed by the stem-old and the crown-young ages each having best-fit lines applied with equation values deduced. Thereafter, a ‘back-modelled’ profile can be generated that uses the equation values linked to the ‘actual’ CIMP age values. Ideally, the ‘actual’ and ‘back-modelled’ profiles show close correspondence. This is assessed through two quantitative metrics. The first is a simple measure of ‘fit’ based on (C_tot−C_nm)/C_tot, where C_tot is the total number of cells associated with the ‘actual’ data and C_nm is the number of non-matching cells. The second makes use of the chi-squared test; for each 1-Myr age-bin there is an observed value related to the simulation, as well as an expected value based on the ‘actual’ and/or the ‘back-modelled’ data. Histograms are used to display the data, and in most cases the enveloping lines for the ‘actual’ and ‘back-modelled’ profiles are overlain for comparison.

Information sources of the node-age determinations for the various clades are presented below (see also Table 1), with ‘stem’ and ‘crown’ abbreviated to ‘st.’ and ‘cr.’ For processing purposes, the various clades are allocated to one of two types: land-bound vertebrates (eight reptiles, three rodents; labelled 1–11); and non-land-bound vertebrates (ten winged birds, including the now-flightless Galápagos cormorant, plus the Galápagos fur seal, Galápagos sea lion, and the Galápagos penguin; labelled A–M). Two other vertebrate groups were omitted due to their limited connections with the Galápagos’ terrestrial environments: marine fishes (the archipelago hosts no native primary freshwater fishes) and the green sea turtle, Chelonia mydas (see Jensen et al. 2019).

Regarding the target study groups, several clades had to be excluded due to them lacking phylogenetic data, for instance the dark-billed cuckoo (Coccyzus melacoryphus), Galápagos martin (Progne modesta), South American hoary bat (Aeorestes villosissimus), and western red bat (Lasiurus blossevillii brachyotis).

The colonization interval literature sources are: 1, Chelonoidis niger tortoises, st. and cr. – Kehlmaier et al. (2023); 2, main clade of the Phyllodactylus geckos, st. and cr. – Koch et al. (2016) and Torres-Carvajal et al. (2016); 3, Amblyrhynchus-Conolophus iguanas, st. and cr. – Malone et al. (2017); 4, Pseudalsophis racer snakes, st. and cr. – Zaher et al. (2018); 5, †Megaoryzomys curioi rodent, st. – estimated by Ali and Fritz (2021) based on Leite et al. (2014) and Ronez et al. (2021); 6, Phyllodactylus darwini gecko, st. – Koch et al. (2016) and Torres-Carvajal et al. (2016); 7, Nesoryzomys rodents, st. and cr. – Castañeda-Rico et al. (2019); 8, main clade of the Microlophus lava lizards, st. and cr. – Benavides et al. (2009); 9, eastern clade of the Microlophus lava lizards, st. and cr. – Benavides et al. (2009); 10, Aegialomys galapagoensis rodent, st. – Castañeda-Rico et al. (2019); 11, Phyllodactylus gilberti gecko, st. – Koch et al. (2016) and Torres-Carvajal et al. (2016); A, Phalacrocorax harrisi cormorant, st. – Burga et al. (2017); B, Mimus mockingbirds, st. – Arbogast et al. (2006), cr. – Nietlisbach et al. (2013); C, Zenaida galapagoensis dove, st. – Valente et al. (2015); D, Darwin’s finches, st. and cr. – Funk and Burns (2018); E, Laterallus spilonota rail, st. – Chaves et al. (2020); F, Spheniscus mendiculus penguin, st. – Gavryushkina et al. (2017); G, Arctocephalus galapagoensis fur seal, st. – Yonezawa et al. (2009); H, Pyrocephalus vermillion flycatchers, st. and cr. – Carmi et al. (2016); I, Myiarchus magnirostris flycatcher, st. – Sari and Parker (2012); J, Zalophus wollebaeki sea lion, st. – Asadobay et al. (2023); K, Buteo galapagoensis hawk, st. – do Amaral et al. (2009); L, Phoenicopterus ruber Caribbean flamingo, st. – Frias-Soler et al. (2022); M, Dendroica petechia yellow warbler, st. – Browne et al. (2008), see also Chaves et al. (2012). As Clades 5, 6, 10, 11, A, C, E, F, G, I, J, K, L, and M do not have crown nodes, the younger bound of each’s colonization interval is set at 0 Mya. Regarding the giant tortoises (Clade #1), we note the recent study of Torres et al. (2024), which used a total evidence matrix (morphology combined with mitochondrial DNA information). However, this resulted in a highly unlikely phylogenetic topology with the African crown group testudinine Stigmochelys pardalis placed basal to all other extant testudinids, i.e., the testudinine and xerobatine tortoises, thereby compromising any age estimates.

The initial presentation of the colonization intervals is shown in Figure 4. Notably, those for the land-bound vertebrates have wide ranges (Fig. 4A), with four of the eleven groups having CIMP ages greater than that of the oldest island (4 Mya). This contrasts with the flying and swimming vertebrates, which collectively have narrower ranges (Fig. 4B), with all of the CIMP values <4 Mya.

Figure 4. 

Colonization interval data for Galápagos’ vertebrate clades. (A) is for the land-bound lineages, (B) is for the swimming and flying groups that are made up of endemic species. Clade key (A): 1, Chelonoidis niger tortoises; 2, main clade of the Phyllodactylus geckos; 3, Pseudalsophis racer snakes; 4, Amblyrhynchus-Conolophus iguanas; 5, †Megaoryzomys curioi rodent; 6, Phyllodactylus darwini gecko; 7, Nesoryzomys rodents; 8, main clade of Microlophus lava lizards; 9, eastern clade of the Microlophus lava lizards; 10, Aegialomys galapagoensis rodent; 11, Phyllodactylus gilberti gecko. Clade key (B): A, Phalacrocorax harrisi cormorant; B, Mimus mockingbirds; C, Zenaida galapagoensis dove; D, Darwin’s finches; E, Laterallus spilonota rail; F, Spheniscus mendiculus penguin; G, Arctocephalus galapagoensis fur seal; H, Pyrocephalus vermillion flycatchers; I, Myiarchus magnirostris flycatcher; J, Zalophus wollebaeki sea lion; K, Buteo galapagoensis hawk; L, Phoenicopterus ruber Caribbean flamingo; M, Dendroica petechia yellow warbler. In (B), the asterisks highlight the three swimming clades (F, G, and J).

Figure 5A shows the various colonization intervals in Figure 4A with the colonization interval mid-point (CIMP) plotted against the interval range. The accompanying graph (Fig. 5B) presents all of the stem-old and the crown-young age dates, with best-fit lines calculated, respectively (1.721 * CIMP) + 0.114, where r² = 0.983 and (0.278 * CIMP) – 0.114, where r² = 0.607 (n = 11).

Figure 5. 

(A) Individual colonization intervals plotted against their mid-point values. (B) Best-fit lines through the stem-old ages and the crown-young ages are used to generate simulated colonization profiles (Fig. 4). CIMP, colonization interval mid-point; cr., crown; st. stem.

The colonization profile for the ‘actual’ data is shown in Figure 6A. In comparison, the ‘back-modelled’ data profile, Figure 6B, has a ‘fit’ of 0.889 with P = 1. The plot reiterates the strong ‘older than the islands’ issue. Indeed, if all of the ancestral colonizers had arrived at 4 Mya, the colonization profile would show an obvious mismatch (Fig. 6C), with ‘fit’ values around 0.38 and p ≤ 0.001.

Figure 6. 

Galápagos vertebrate colonization profiles. (A) is the ‘actual’ data (grey shaded region and red line) for the land vertebrates, while (B) is the ‘back-modelled’ data (grey shaded region and blue line, with the ‘actual’ data line [red] overlain). (C) is a simulation based upon all of the colonizations taking since the oldest extant island (San Cristóbal) formed, with an arrival every c. 381 kya starting at 4 Mya and ending at c. 190 kya, a scenario that is termed ‘constant rate’. (D) is the profile for the non-land vertebrate clades, with the ‘actual’ and ‘back-modelled’ land-vertebrate lines added. Where appropriate, ‘fit’ and p values are shown on the respective plots.

The colonization profile for the non-land-bound vertebrates, both endemic and non-endemic, is shown in Figure 6D (respectively 10 flying and 3 swimming clades). Its shape is conspicuously different to that of land-bound groups and indicates that it comprises more recent arrivals.

The assembled dataset makes use of divergence dates from more than twenty different studies that were published over a 15-year interval (Browne et al 2008; Kehlmaier et al. 2023). Hence, they differ in terms of the quality of the data as well as the level of the analyses. However, a general bias is unlikely to have impacted them all, and we are confident that the features we outline below, as well as the conclusions that follow, are likely to be valid. At the very least, perceived weakness may spur discussion and/or further work.

Early studies reporting ‘older than the islands colonizations’ included Wright (1983) for the geckos and lava lizards, Lopez et al. (1992) for the lava lizards, and Rassmann (1997) for the marine and land iguanas (Amblyrhynchus, Conolophus). The data compiled for this study indicate that the condition applies to a least one lineage and up to seven others (Fig. 7A,B). There are a number of reasons why this might be the case. Firstly, the ancestral colonizers arrived on a now-drowned landmass; as explained above, the Galápagos hotspot has been generating islands in the eastern equatorial Pacific since at least the Middle Miocene, and likely without a break. This has to be the case with the oldest Phyllodactylus gecko clade (#2), plus it probably also applies to the Amblyrhynchus-Conolophus iguanas (#3) and the Pseudalsophis racer snakes (#4), and possibly the Chelonoidis niger tortoises (#1), the †Megaoryzomys curioi rodent (#5), the Phyllodactylus darwini gecko (#6), Nesoryzomys rodents (#7), and the main clade of the Microlophus lava lizards (#8). However, for some of these groups, the supposed sister species on the mainland could have been incorrectly identified and closer relatives actually exist, or off-archipelago extinction has removed intermediate lineages, thus increasing the ages of the stem nodes as an artefact. Notably, the ‘lost taxa’ issue was pondered in early molecular studies of the Galápagos tortoises (e.g., Caccone et al. 1999, 2002). Recent dating of the split between Chelonoidis niger and its genetically closest extant relative C. chilensis, which is found in parts of Argentina, Bolivia, and Paraguay, yields a stem-node uncertainty range of 23.5–9.5 Mya (Kehlmaier et al. 2023; these numbers are consistent with the independent studies of Sánchez et al. 2017 and Poulakakis et al. 2020). Significantly, though, Chelonoidis-genus subfossils have been recovered from numerous Pliocene and younger sedimentary formations in western South America, Central America, and the Caribbean, with uppermost Pleistocene ‘giants’ found in coastal areas of mainland Ecuador (Kehlmaier et al. 2017 and references therein; Ali and Fritz 2021).

Figure 7. 

Stem (red) and crown (blue) intervals for the various vertebrate clades and how they relate to the extant islands. The land-bound groups are shown in (A) and (B), while the non-land-bound lineages are in (C) and (D). The data are plotted using two different y-axis measures, the colonization interval mid-points (A) and (C) and the crown-old ages (B) and (D). For those clades where on-archipelago diversification has not taken place, or no relevant age data are available, the crown-young ages are set at 0 Mya. In such cases, the associated horizontal bars have been given a width of –0.10 to 0.10 Mya to make them visible.

Comparing the land-vertebrate colonization record with that of the flying and swimming vertebrates

The colonization profile for the flying and swimming vertebrate clades is unlike that for the land-bound groups (Fig. 7C,D). Notably, ten of the clades postdate the oldest island. The conspicuous lack of old components suggests that those taxa that can cross easily to the archipelago have either experienced higher rates of turnover, or that following the initial colonization a steady trickle of mainland migrants has slowed genetic differentiation, thus lowering the ages of the stems and, where relevant, the crowns (e.g., Leaché et al. 2014; Tiley et al. 2023), or that both processes have operated.

Concerning the turnover explanation, the Galápagos Islands are located in the eastern equatorial Pacific and the biota is subject to significant environmental stresses imposed by the El Niño–La Niña climate cycles (Wang and Fiedler 2006; Chen et al. 2019). Notably, the phenomenon has deleteriously impacted a wide variety of taxa (Table 1 of Dueñas et al. 2021). The hot, wet El Niño state negatively affects the marine feeders, decimating their food chains, while the cool, dry La Niña phase poses a greater challenge to the land-based foragers (Hedrick 2019). Moreover, the system, which has a somewhat erratic periodicity of 3–7 years (Fig. 8), appears to have operated since at least the Late Miocene, although with intervals of reduced intensity as compared to the modern configuration (e.g., Galeotti et al. 2010; Weiss et al. 2017; Drury et al. 2018). Thus, a potential extirpation mechanism is in place that might regularly pressure any recent arrivals, the older clades presumably having evolved such that sufficient numbers of their members can ride out the extremes of the climatic phases. A suggested example of a taxon being affected in such a manner is the Galápagos penguin Spheniscus mendiculus, which has a colonization interval of 1.88 to 0 Mya. Notably, its small, geographically-confined population shows an almost complete lack of genetic diversity, which is thought to result from it having passed through multiple bottleneck events (Arauco-Shapiro et al. 2020). This contrasts with the marine iguana Amblyrhynchus cristatus whose colonization interval is 12.5 to 2.5 Mya (Table 1); Steinfartz et al. (2007) showed it experienced little genetic loss following the severe El Niño event of 1997–1998 (Fig. 8), despite the depletions of populations on various islands reaching up to 90%.

Figure 8. 

Temperature anomalies associated with the El Niño-La Niña climate system. The data are from NOAA’s Climate Prediction Center (https://origin.cpc.ncep.noaa.gov/products/analysis_monitoring/ensostuff/ONI_v5.php, accessed April 2024) and are presented as 3-month means centred on January, April, July, and October. El Niño phases are associated with increased precipitation, whereas La Niña intervals are comparatively dry.