Over geological time, glacial-interglacial cycles affected the geographical range of marine species. Typically, this has been documented by tracking the long-distance dispersal of tropical, shallow-water mollusc species in archipelagos during the last phase of glacial terminations or the early phase of an interglacial episode. Many studies conducted in the Macaronesian archipelagos (i.e., the Azores, Madeira, Selvagens, Canaries and Cabo Verde) support this view. To date, however, such studies exclude data from full glacial periods, owing to difficulties in accessing the geological record of lower sea-level glacial episodes. Here we demonstrate, for the first time, the range expansion into the tropics of cold-water/temperate species during two glacial episodes (MIS 4 and MIS 6), using the Macaronesian region as a case study. For that, we innovate by using megatsunami deposits to unveil biogeographic processes and patterns noticed in the conglomerates of Tarrafal (Santiago Island, Cabo Verde) and of Teno Bajo (Tenerife, Canary Islands), which are interpreted to have been emplaced by megatsunamis triggered by volcanic flank collapses occurred at ~68 ka (MIS 4) and ~170 ka (MIS 6), respectively. Our results detect that not only latitudinal, long-distance dispersal of marine molluscs occurred toward the tropics (mainly between archipelagos, and between European and African shores towards the Macaronesian archipelagos), but also longitudinal range expansion. Moreover, both MIS 4 and MIS 6 megatsunami deposits yielded a high biodiversity (expressed both by species richness and diversity indices) when compared with raised beach sediments. This new finding must be added to the distinctive sedimentological and textural characteristics of tsunami deposits. Finally, we demonstrate that four mollusc species reported from the Teno glacial MIS 6 tsunami deposits, and several temperate and sub-tropical bivalve and gastropod species reported from the Tarrafal glacial MIS 4 tsunami deposits spread to the Canaries and Cabo Verde, respectively, establishing viable populations in those archipelagos. Thus, these species provide evidence of geographical range expansion of marine species from mid-latitudes to low latitudes by means of long-distance, equatorward dispersal of benthic, shallow-, cold-water/temperate marine molluscs between archipelagos and from continental shores to oceanic islands during glacial periods.

Cabo Verde, Canaries, glacial periods, marine molluscs, Megatsunami deposits, MIS 4, MIS 6, oceanic islands, palaeobiogeography

The island biogeography equilibrium theory (MacArthur and Wilson 1963, 1967) and its subsequent advances (Whittaker et al. 2008, 2010; Triantis et al. 2012; Fernández-Palacios et al. 2016) generally has lacked a marine point of view, a situation that changed with topical efforts by Ávila et al. (2009a, 2016a, 2018, 2019), Hachich et al. (2015, 2016, 2020) and Pinheiro et al. (2017). Thus, the understanding of biogeographic patterns and processes of long-distance dispersal in marine species among oceanic archipelagos, and between continental shores and archipelagos, is crucial for extending island-biodiversity evolution to marine ecosystems (Trakhtenbrot et al. 2005; d’Aloia et al. 2015). Ávila et al. (2019) were the first to explore in detail the impact of glacial periods on shallow-water marine species. Their Sea Level Sensitive dynamic model of marine island biogeography put forward several predictions that can be tested on the effects of such glacial episodes. Due to the reduction of the insular shelf area available around islands during glacial periods, and for geotectonically stable or subsiding islands, it was proposed that: (1) extinction rates and extinction debts are higher during glacials in comparison with interglacials; (2) the extirpation rates of thermophilic species and shallow-water species associated with soft sediments will be higher during glacials in comparison with interglacials; (3) the extinction of endemic species will be higher during glacials; and (4) the rates of speciation will be lower during glacial episodes and higher during interglacials. However, only recently, fossiliferous megatsunami deposits emplaced during glacial periods became available to study.

Global climate changes influence the expansion and contraction of the geographical range distribution of species through different dispersal processes, constituting a common and well-studied biogeographic process, especially in the terrestrial realm (Channell and Lomolino 2000; Gaston 2003; Huang et al. 2023). This scenario is known as the “interglacial expansion hypothesis” (Qian and Ricklefs 2000) or the “expansion-contraction model” (Hofreiter and Stewart 2009), and many examples are known, such as arctic-alpine species like Dryas integrifolia Vahl (Tremblay and Schoen 1999) or plant species distributed in low latitudes of eastern Asia, such as Pinus armandii Franch. (Liu et al. 2022). For terrestrial taxa, records for both fossil and extant species document poleward shifts stemming from the range expansion of tropical species during interglacial times in addition to the simultaneous range contraction of temperate species (Clark et al. 1999; Jackson and Blois 2015). Similarly, for extant marine species, the geographical range contraction of benthic, shallow-, temperate/cold-water species to higher latitudes (and/or to deeper waters) are also documented as a result of the current global warming (Hiddink et al. 2015), as well as the range expansion of benthic, warm-water species to higher latitudes (Perry et al. 2005; Lima et al. 2007; Ling et al. 2009; Baptista et al. 2021). The fossil record from volcanic oceanic islands also attests to the range expansion of benthic, tropical, shallow-water marine molluscs towards higher latitudes during interglacials through long-distance dispersal (García-Talavera et al. 1978; Meco et al. 2002; Cabero et al. 2013; Ávila et al. 2015; Martín-González et al. 2016; Melo et al. 2022a, b), as well as the range contraction of thermophilic species (with associated local disappearances from some archipelagos) during the Pliocene and Pleistocene climatic cooling (Landau et al. 2007; Ávila et al. 2016b). Planktonic marine species (e.g., foraminifera) are commonly used to reconstruct past sea-surface temperatures (Kucera et al. 2005). These authors have also demonstrated an equatorward (i.e., a latitudinal geographical range expansion of cold-water/temperate planktonic species/morphs towards the tropics) during glacial episodes. These phenomena also relate to range expansions of tropical planktonic species/morphs towards higher latitudes during interglacials (Pailler and Bard 2002; Alessandro et al. 2018, and references therein).

In contrast, few studies have considered the theoretical expected range expansion of benthic-, shallow-, cold-water/temperate marine species towards the tropics during glacial episodes, the most relevant being those by Paulus (1950), Mars (1958), Froget et al. (1972), Malatesta and Zarlenga (1986), Raffi (1986), Di Geronimo et al. (2000), and more recently, Urra et al. (2023). These authors focused on the Mediterranean area and reported on Atlantic “Northern Guests”, also known as “Boreal Guests” (BG), i.e., cold-water marine molluscs whose geographic ranges expanded towards lower latitudes, with Atlantic individuals entering and establishing viable populations in the Mediterranean, during glacial episodes; these species have thus important palaeoclimatic significance. Taviani et al. (1991) is among the only studies that deal with the geographic range expansion of BG molluscs along the Atlantic coasts of Europe. However, BG are hard to access either because they are usually deposited under several tens or even hundreds of meters of seawater (e.g., 150–350 m depth; Mars 1958), and/or such deposits are seldom preserved on account of erosion by rising sea levels during the subsequent interglacial episode (Ávila et al. 2022). Exceptions may occur when littoral deposits formed during glacial episodes have been raised by later tectonic or isostatic coastal uplift. However, few places experience uplift rates high enough to exceed the rate of local sea-level rise to preserve such a deposit (Peltier 2004; Seppä et al. 2012). We herein innovate by using glacial-age megatsunami deposits from the Macaronesian islands to infer range shift patterns of littoral marine molluscs during glacial periods, by means of long-distance dispersal.

Massive gravitational flank failures of volcanic island edifices, although infrequent at short geological timescales, may involve considerable volumes of collapsed material, on the order of tens to hundreds of km³ (e.g. Holcomb and Searle 1991; Carracedo 1999; Day et al. 1999; Masson et al. 2002, 2008). As they typically occur catastrophically, they often trigger megatsunamis (sensu Goff et al. 2014) with initial wave magnitudes of hundreds of meters (e.g. Moore and Moore 1984; McMurtry et al. 2004a; Giachetti et al. 2011; Abadie et al. 2012; Ramalho et al. 2015; Paris et al. 2017). These giant waves are capable of inundating near-source coastlines up to several kilometres inland and hundreds of meters in elevation. Megatsunami deposits attesting to such extreme inundations have been documented on several volcanic archipelagos, namely the Hawaiian Islands (Moore and Moore 1984; Moore 2000; McMurtry et al. 2004a, b), the Canary Islands (Hürlimann et al. 2004; Pérez-Torrado et al. 2006; Ferrer et al. 2013; Paris et al. 2017), and the Cabo Verde Archipelago (Paris et al. 2011, 2018; Ramalho et al. 2015; Madeira et al. 2020; Costa et al. 2021). Hence, large tsunami events occurring during glacial periods and able to transport large amounts of marine sediment onshore, away from the erosive action of interglacial high sea levels, represent a geological process providing access to unique glacial fossil assemblages of benthic, shallow-water marine species, otherwise inaccessible.

This study focuses on the very rich fossiliferous sediments deposited during glacial periods by the impact of two megatsunamis: the conglomerates of Tarrafal (Santiago Island, Cabo Verde), and Teno Bajo (Tenerife, Canary Islands). Critically, the Tarrafal conglomerates are here documented in detail for the first time in regard to species composition and palaeobiogeographic relationships. These sediments were deposited by a megatsunami triggered by the flank collapse of the neighbouring Fogo Island (Day et al. 1999), at approximately 68 ka (Cornu et al. 2021), and with an estimated run-up height of 270 m (Ramalho et al. 2015). A preliminary analysis by Paris et al. (2011) revealed the presence of three species of corals, calcareous algae, molluscs (three species of gastropods and at least four species of bivalves) and skeletons of Bryozoan zooids. None of the specimens were found in original growth positions. According to Paris et al. (2011), the conglomerates represent at least, two main tsunami waves that transported sediments both during inflow and outflow. A minor role is attributed to the impact of the following waves (whose effects were probably limited to a surface reworking of the deposits), as a result of their smaller amplitude and of the decreasing amount of sediment still available offshore. Paris et al. (2011) also noticed that the marine fauna of the Tarrafal conglomerate was different from the present-day and Last Interglacial deposits from Sal and Boavista Islands (Zazo et al. 2007). However, this observation was not further substantiated.

In the Canaries Archipelago, fossiliferous conglomerates in the Teno massif, northwestern Tenerife, were also interpreted as the result of megatsunamis (Ferrer et al. 2013; Paris et al. 2017). Paris et al. (2017) reported 85 gastropod taxa, 31 bivalves, six corals, and one scaphopod species, representing a high biodiversity and a mixing of faunas from different environments (infra-circalittoral fauna being dominant).

The fossiliferous sediments of these two megatsunamis events were deposited during glacial periods [the Tarrafal deposits been dated 65–84 ka (Ramalho et al. 2015) and attributed to the flank collapse of nearby Fogo volcano, which probably took place at ca. 68 ka (Cornu et al. 2021), whereas the Teno deposits result from a flank collapse dated of ca. 170 ka (Coello-Bravo et al. 2014)], with similar coeval sea level at about 80 m below present sea level in both cases (Miller et al. 2011). Critically, both events resulted in deposits emplaced several tens of meters above Holocene sea level, ensuring their preservation. Thus, both deposits provide an exceptional opportunity to test the hypothesis of equatorward geographical range expansions of benthic, temperate/subtropical, shallow-water marine species during glacial episodes in an insular environment.

Accordingly, this study aims to: 1) screen the tsunami sediments to identify any cold-water/temperate marine molluscan species/assemblages preserved therein; 2) investigate their palaeoclimatic significance, based on their present geographical range and biogeographical patterns; 3) compare the glacial Marine Isotope Stage 4 (MIS 4, i.e., the period between 57 and 71 ka) molluscan fauna of the Tarrafal megatsunami deposit with the recent and the Last Interglacial fauna of Cabo Verde; 4) compare the glacial Marine Isotope Stage 6 (MIS 6, i.e., the period between 130 and 191 ka ago) molluscan fauna of the Teno megatsunami deposit with the recent and the Last Interglacial fauna of Canary Islands; and 5) test and draw conclusions on the theoretically expected equatorward geographical range expansion of benthic, shallow-, cold-water/temperate marine species during glacial episodes.

In common with other fossiliferous deposits left by megatsunamis (e.g., McMurtry et al. 2004b; Pérez Torrado et al. 2006; Paris et al. 2017), a mixture of deep- and shallow-water species (including pelagic species), as well as an assortment of species with very different preferential habitats and ecological conditions is detected in the Tarrafal sediments. This allochthonous, highly biodiverse assemblage, with well-preserved micromollusc species, is attributed to the peculiar genesis of the fossiliferous deposit resulting from a megatsunami, with a rapid deposition and minor reworking of the bioclasts. Such an outcome is reported for similar deposits existing elsewhere, such as the Teno deposits in Tenerife, emplaced during the glacial MIS 6 (Coello-Bravo et al. 2014; Martín-González et al. 2016; Paris et al. 2017). The extremely high biodiversity of these megatsunami deposits is emphasized when compared with raised beach sediments sampled similarly (Fig. 3). In a series of studies, Ávila and co-workers sorted, identified and counted molluscs from 24 samples of 1 kg each, from fossiliferous sands of the MIS 5e (Last Interglacial) deposits of Santa Maria Island (Ávila 2005; Ávila et al. 2009b, 2015). In total, they found 136 marine molluscan-specific taxa and counted 25,090 specimens (Ávila et al. 2009a, b, 2015), which contrasts with the present study, where the 7 samples of 1 kg each, of fossiliferous sediments from the glacial MIS 4 deposit of Tarrafal, yielded 209 marine molluscan taxa (Table 1), but only 2,257 specimens (Table 2). A similar pattern was found by Coello-Bravo et al. (2014), who identified 117 marine molluscan taxa and counted 1,941 specimens collected from the glacial MIS 6 Teno tsunami deposits. The distinctive sedimentological and textural characteristics that confirm marine fossiliferous conglomerates as tsunami deposits include distribution of the conglomerates at variable elevations, many of them out of the range of marine highstand deposits; chaotic textures and the extremely poorly-sorted distribution of clast size; mixing of sediment and bioclast assemblages (planktonic, benthonic, littoral and terrestrial) from different provenances; and fossils never found in life position – for recent reviews on the subject see Goff et al. (2014) and Paris et al. (2018). In addition to these characteristics, this work highlights another trait of megatsunami deposits: the extremely high palaeobiodiversity and evenness of their fossil content.

Moreover, the few studies that analyse palaeobiodiversity in tsunami deposits using diversity indices applied to quantitative fossil data (e.g., Coello-Bravo et al. 2014; Martín-González et al. 2016; this study) point to high evenness index values in the fossil record in these sediments, in stark contrast to highstand fossil deposits. This apparent non-occurrence of abundant/dominant species on tsunami deposits was an unexpected result, yet predicted by Behrensmeyer et al. (2000). Fossiliferous marine sediments deposited during highstands (e.g., MIS 5e) are typically the result of several hundred or even a few thousand years of gradual accumulation of biological remains, mainly on supratidal locations, and are usually associated with several storm events (Ávila et al. 2010). Because of this long-duration process, the resulting death assemblage (which mirrors the populations of the local biocoenose and involves the steady accumulation and deposition of their faunal remains; cf. Lyman 2008) is a time-averaged collection of species/specimens. Moreover, and unless taphonomical biases interfere, it is expected that the most abundant species present in the biocoenose to also be the most abundant species present in the death assemblage. In the long term, it is also expected that these species will be the most abundant in the fossil assemblage. This is so, because specimens of dominant species will increase in number faster than rare ones, in the context of long-term sedimentation. Hence, time-averaged samples are biased towards higher taphonomical durability species and to the most abundant species (Behrensmeyer et al. 2000; Kidwell 2013). In contrast, megatsunami deposits are instantaneous events at geological time scales, being responsible for a single-event fossil assemblage. Additionally, tsunami sedimentary deposits mostly result from en-mass transport and deposition by long-wavelength inundation (Paris et al. 2018; Madeira et al. 2020), rather than repetitive energetic mechanisms (such as swell- and wind-driven waves), with much less sorting and destruction, resulting in very-well preserved fossil assemblages more representative of the full range of faunas from the ecological zones they sample. As such, during such catastrophic events, a few mega-waves produce an instantaneous sampling of a varied biological community that includes living and dead remains, the latter also time-averaged, collected from different depths and habitats – thus the usual presence on tsunami deposits of permanent and transient species (e.g., pelagic species) –, and by its transport and deposition high on the slopes of the islands. As a result, the mixed-zonation fossil content found in tsunami deposits is highly biodiverse and might be characterized by the absence of dominant species (as only a few waves are involved). Correspondingly high values of species richness estimates (Hill’s q = 0 and ES50), Shannon’s entropy index (Hill’s q = 1), and the inverse of Simpson’s concentration index (Hill’s q = 2) are explained by this relationship (cf. Table 2).

Three species of bivalves presently absent from the Cabo Verdean islands (Tetrarca tetragona, Ostrea stentina, and Venus verrucosa) were found in the MIS 5e (Melo et al. 2022a) and in the MIS 4 fossil record of Cabo Verde (this work); the populations of these species therefore must have been extirpated from the archipelago during the last glacial episode, after MIS 4. Our data also show that, during the glacial MIS 4, several temperate and sub-tropical bivalve and gastropod species were able to reach the Cabo Verdean islands through long-distance dispersal (cf. Table 4); these species were subsequently extirpated from the malacofauna of the archipelago, most probably during the present interglacial. Most of the 24 species listed in Table 4 have in the present-day their southernmost range distribution at the Canaries (10 bivalves and 12 gastropods; Fig. 4) or at the Northwest-African coasts (Atlantic Morocco, from the Straits of Gibraltar south to Cape Vert in Senegal, including Western Sahara and Mauritanian shores; 7 bivalves and 7 gastropods). These 24 species were not present in Cabo Verde during the Last Interglacial (Cabero 2009; Cabero et al. 2013; Melo et al. 2022a, b) and so, their presence during the glacial MIS 4 conclusively supports the equatorward long-distance dispersal and successful establishment of viable populations, with the consequential geographical range expansion of benthic, shallow-, cold-water/temperate marine species to volcanic oceanic islands during glacial episodes.

Interestingly, four mollusc species listed in Table 5 and reported from the glacial MIS 6 Teno’s tsunami deposits (e.g., Coello-Bravo et al. 2014; Martín-González et al. 2016; Paris et al. 2017), also support an equatorward geographical range expansion of temperate/cold-water species towards lower latitudes. None of these species presently occur in the Canary waters and their southernmost present geographical distribution is in the Mediterranean Sea, Madeira and Selvagens archipelagos.

Fig. 4 shows that, during glacial times, the most probable direction of the detected range expansion of shallow-water marine species was from medium latitudes towards the Cabo Verde Archipelago (MIS 4, Fig. 4A, B), and to the Canary Islands (MIS 6, Fig. 4C). As today, also during the MIS 4 and MIS 6 glacial episodes, the dispersal of these mollusc species most probably occurred during larval stages (in the case of planktotrophic species), aided by sea-surface currents and trade winds, or by rafting in natural floating devices (in the case of non-planktotrophic species), e.g., wooden logs, algae, as attached eggs/juveniles/adults (Scheltema 1971, 1989, 1995). Our results, which are based in the glacial fossil record (MIS 4 of Cabo Verde and MIS 6 of Canaries), allow us to speculate that the inferred equatorward dispersal of marine species (Fig. 4A–C) will only be possible if and when the prevailing oceanographic barriers that exist today, such as the Canary Current (nowadays located between the Canaries and Cabo Verde archipelagos) shifts its position to a more southern location, thus allowing the arrival of mid-latitude species to Cabo Verde. Alternatively, if the Canaries Current weakens during glacial episodes, the equatorward range expansion of marine species may also occur. In both cases (shifting of the position of the Canary Current or weakened Canary Current), some cooling of the sea-surface temperatures at Cabo Verde latitudes is expected to occur in glacial times, enabling settlement of temperate/cold-water species. In contrast, the detected longitudinal dispersal of species (in this case, from the continental African shores of Senegal and Mauritania towards Cabo Verde – Fig. 4A, B) is expected to occur if the West African upwelling, that today dominates along the Senegal shores, increases in intensity. Indeed, mid-latitude North Atlantic records clearly show a decrease in sea surface temperature (SST) during both MIS 4 and MIS 6 (e.g. Iberian margin: Martrat et al. 2007; Azores: Naafs et al. 2013) caused by the weakening of the Atlantic Meridional Overturning Circulation (AMOC) (McManus et al. 2004; Lippold et al. 2009; Böhm et al. 2014). The AMOC reduction has favoured the southward displacement of the Intertropical Convergence Zone (ITCZ) triggering very dry conditions over northwestern Africa and intensified NE trade winds (Castañeda et al. 2009) which contribute for huge amounts of dust supply to the adjacent ocean (Skonieczny et al. 2019). Increased NE trade winds contributed for the intensification of the upwelling in this region and therefore for high marine productivity (Abrantes 1991; Adkins et al. 2006).

Finally, we wish to emphasize that Patiño et al. (2017) conducted a collaborative horizon-scanning approach to identify the 50 most fundamental questions for the continued development of the field of island biogeography. Although the preliminary list of 187 questions contained 27 related to “Palaeoecology and Palaeobiogeography”, only one of these remained in the final list of 50 questions. This reflects not only the low percentage of current biologists that identify “Palaeoecology and Palaeobiogeography” as their field of expertise (Patiño et al. 2017), but also the usual lack of knowledge of biologists in the fields of Geology and Palaeontology. However, island biogeography theory requires the integration of state-of-the-art studies on these three areas of research (Biology, Palaeontology and Geology), and therefore, the formation of multidisciplinary teams is a superior approach for this task. Our study demonstrates, for the first time in the Atlantic Ocean Macaronesian archipelagos, the theoretically expected equatorward geographical range expansion of benthic, shallow-, cold-water/temperate marine species during glacial episodes, a conclusion that was only possible because of the joint efforts of an inter-disciplinary team of biologists, palaeontologists and geologists.

Figure 4. 

Most probable direction of the detected range expansion processes of mollusc taxa towards the Cabo Verde Archipelago during the glacial MIS 4 (A. Bivalvia; B. Gastropoda), and to the Canary Islands during the glacial MIS 6 (C – Gastropoda). Numbers indicate the number of taxa that underwent such range expansion processes. CAB – Cabo Verde Archipelago; CAN – Canaries Archipelago; SEL – Selvagens Archipelago; MAD – Madeira Archipelago; AZO – Azores Archipelago; SEN – Senegal; MAU – Mauritania; WES – Western Sahara; MOR – Atlantic Morocco; MED – Mediterranean. Delimitation of the landmasses from the Portuguese Hydrographic Institute available data (https://www.hidrografico.pt/op/33).

Glacial-age marine fossiliferous deposits are seldom preserved worldwide due to erosion by rising sea levels during the subsequent interglacial episode. Megatsunami events occurring during glacial times are thus one of the best ways to preserve such glacial fossil assemblages, given that they transport and deposit large amounts of sediment onshore, away from the erosive action of subsequent interglacial high sea levels. Moreover, as demonstrated by this study, megatsunami deposits have extremely high palaeobiodiversity and evenness, and these features may constitute additional criteria to distinguish megatsunami deposits from normal sea level highstand deposits.

To conclude, this is the first study in the Atlantic domain that provides conclusive evidence of long-distance dispersal of shallow-water marine mollusc species among different archipelagos and from continental shores to oceanic islands during glacial periods, with both latitudinal and longitudinal inferred range shifts that resulted in successful geographical range expansion processes of marine species from medium to low latitudes. In times of accelerated global changes, knowing the distributions of species during Pleistocene glacial and interglacial periods is crucial to our understanding of biogeography, ecology, and evolution over the past one million years in fragile marine ecosystems such as those of archipelagic settings.

This work was done in the remit of project PTDC/CTA-GEO/28588/2017 – LISBOA-01-0145-FEDER-028588 UNTIeD, co-funded by the European Regional Development Fund – ERDF, through Programa Operacional Regional de Lisboa (POR Lisboa 2020), and by FCT – Fundação para a Ciência e a Tecnologia I.P. SPA acknowledges the project M1.1.A/INFRAEST CIENT/A/001/2021 – Base de Dados da PaleoBiodiversidade da Macaronésia, funded by Direção Regional da Ciência e Tecnologia, Governo Regional dos Açores, and his research contract with BIOPOLIS: https://doi.org/10.54499/2023.07418.CEECIND/CP2845/CT0001. JM was funded by FCT project CVPLUME (PTDC/CTE-GIN/64330/2006), and (PIDDAC) – UIDB/50019/2020 to Instituto Dom Luiz. RC was supported by grant SFRH/BD/60366/2009 from FCT, Portugal. CSM benefited from a PhD grant M3.1.a/F/100/2015 funded by the Regional Government of the Azores. This work also benefited from FEDER funds, through the Operational Program for Competitiveness Factors – COMPETE, and from National Funds, through FCT: UIDB/50027/2020 (CIBIO/INBIO), UID/GEO/50019/2013 (Instituto Dom Luiz), and POCI-01-0145-FEDER-006821 and LA/P/0048/2020 (CIBIO/INBIO), as well as through the Regional Government of the Azores (M1.1.a/005/Funcionamento-C-/2016, CIBIO-A; M3.3.B/ORG.R.C./005/2021).

SPA, RP and RR conceived the ideas; funding acquisition: RR, SPA, RP, and EMG; SPA, RP, RR, CSM and JM collected the samples during fieldwork at Santiago Island; SPA identified and counted the fossil molluscs, with contributions from CSM, GCA, EMG and ER; GMM and SPA performed the statistical analysis; and SPA, RR, MEJ and JM led the writing. All authors have read and agreed with the final version of the manuscript.

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.