Biogeographic patterns of modern benthic shallow-water molluscs and the roles of temperature and palaeogeographic legacy

Our results offer new insights into the provincialism of the world’s shallow-water benthic molluscs across various scales. Using a comprehensive dataset and a new analytical approach without a priori defined units, we find evidence for the overarching importance of temperature for biogeographic structuring. Below, we discuss and show the persistent relevance of temperature, or more specifically, the linkage of molluscs to cold or warm water masses and their variation through geological time, on marine biogeography. Moreover, we find deviations of our new biogeographic scheme from previous ones and elaborate on the potential causes.

The impact of temperature and ocean currents

All our analyses indicate latitudinal differences, with a strong distinction of polar versus circum-temperate–tropical faunas. This is especially evident from the nMDS (Fig. 2) and higher relationships among clusters of the hierarchical cluster analyses (Fig. 1). The latitudinal distinction is clearest for families and genera, but less so for species, which present a more refined regionalisation. These finer-scale differences in species composition are mirrored by the better fit in the analyses with global sea-surface temperature compared to genera and families (Fig. 3).

Our results for species are similar to what was found in a previous study on global shallow-marine benthic faunas¹⁸. These authors tested for the prediction power of temperature on fixed, existing biogeographic schemes, while our analyses are not based on any given boundaries or a priori assumptions.

The link with temperature reminds conspicuously of the latitudinal diversity gradient^{8,9,19,20,34,35,36,37}. Various hypotheses have been brought forward to explain latitudinal differences in species richness and specifically how it relates to variation in temperature. Metabolic and speciation rates are higher, and generation cycles are shorter in warm climates, while extinction rates tend to be lower^8,11,35,36. A higher demand in nutrients with increasing temperatures boosts biotic interactions such as competition and predation^9,10.

Consequently, a smaller set of species is adapted to the cold waters of the world’s polar regions, which accounts not only for differences in species richness but also for the biogeographic signatures we observe here. Many of those species belong to genera and families that are also present in temperate and tropical areas. Conversely, many genera and families are consistently restricted to tropical regions. The comparatively weak relationship between taxonomic composition and temperature for genera and families shows that related species can be adapted to very different climatic regimes, causing many genera and families to occur in a broad range of temperatures. This is particularly the case for species-rich genera³⁸.

Aside from the general latitudinal temperature pattern affecting biogeographic relationships on a larger scale, individual warm- and cold-water currents influence the distribution of benthic shallow-water molluscs and often contribute to sharpen the transitions. We find that the boundaries of many biogeographic regions identified by our cluster analyses coincide with the convergence of ocean currents (Fig. 4). Examples are numerous: the Warm-temperate Southeast Pacific coincides with the limits of the Humboldt (Peru) Current both in the north and the south. The Magellanic region links to the Cape Horn and Falkland currents. The Arctic-type faunas extend in the northwestern Atlantic down to mid latitudes due to the cold-water Labrador Current, and in the northeastern Pacific even further as a result of the California Current. Wedged in between the cold-water California Current in the north and the Humboldt Current in the south is the Tropical Eastern Pacific region supplied by warm Pacific equatorial currents. The North Atlantic Current brings warm water towards northern Europe, causing the distinction of European and Arctic faunas. The transition between Temperate Southern Africa and Tropical Eastern Atlantic regions coincides with the transition of the Benguela Current and the Atlantic South Equatorial Current. The Tropical Indo-Pacific region in the Indian Ocean is delimited in the south by the South Indian and West Australian currents. In the northwestern Pacific, the Liman, Oyashio, and Anadyr Currents bring cold water to the Japan Sea as well as northeastern Honshu, marking the transition between Cold- and Warm-temperate Northwest Pacific faunas (Fig. 4).

Biogeographic regions as defined by the hierarchical cluster analysis for species (compare Fig. 1A) and current ocean circulation patterns (modified from https://www.britannica.com/science/ocean-current). Light blue colours indicate warm currents, dark blue show cold currents. Note that directions for the Monsoon Drift here refer to Northern Hemisphere winter conditions; it alternates during summer. Incoherent clusters were omitted in this figure. C.H.C – Cape Horn Current; F.C. – Falkland Current; L.C. – Liman Current; T.S. – Tsushima Current. Map was created with R v. 4.3.2 and edited with CorelDRAW X8, https://www.coreldraw.com/.

Partly, these patterns are traceable also for genus-level compositions, and in rarer cases also for family-level data. Particularly the Warm-temperate Southeast Pacific region (although it is not found coherent for family-level data because of outliers), the boundaries between Arctic and warm-temperate/tropical faunas, as well as between Antarctic and Southern Ocean/Magellanic faunas are visible across systematic ranks (Fig. 1, 4). This may in part be related to abrupt shifts in temperature regimes due to strong currents, such as the Humboldt Current pulling the equatorward limit of the temperate zone to the north. For genera, the delimitation of Northern Atlantic and European faunas differs from that for species, with northern European faunas showing a higher affinity towards Arctic than western European to Mediterranean faunas.

For the most part, these trends reflect large-scale climatic trends (polar vs. temperate/tropical), with the exception of the Warm-temperate Southeast Pacific region. Its persistence across all systematic ranks indicates that the Humboldt Current is an important driver of biogeographic distinction. This may root in its specific hydrological properties: because of intensive coastal upwelling, it is the most productive of all eastern boundary currents and has yielded more fish catch than any other ocean region in the world^39,40. The Humboldt Current System is affected by a variety of seasonal to centennial fluctuations, such as the El Niño phenomenon⁴⁰. As a result, the associated biogeographic unit, variably termed Warm Temperate Southeastern Pacific², Humboldt Current Large Marine Ecosystem⁴¹ or Peruvian Province⁴², harbours a unique mollusc fauna.

Ocean circulation, palaeogeography, and biogeography

An aim of this paper is to identify the influence of temperature and oceanographic history on the distribution of benthic molluscs. Yet, these factors cannot be considered separately, but they are in fact tightly interrelated. The reconfiguration of land bridges and seaways over the past million years has had severe effects on global ocean currents and, therefore, temperature^43,44,45,46. Particularly the legacy of Cenozoic changes is still traceable nowadays. Many gateways, be it the opening of a seaway or the formation of a land bridge, are reflected in elevated biogeographic distances between clusters. In the following, we discuss the most important reconfigurations of gateways and ocean currents in the geological past, especially highlighting those that link to the boundaries of the here observed biogeographic units. The most important information is summarized in Table 1.

A major distinction concerns Western and Eastern Atlantic tropical faunas, which show strong differences on the species level but comparatively little variation on the genus level (Fig. 1). In addition to geologically more recent events and reconfigurations of ocean circulations, this pattern in part reflects the old age underlying the opening of the Atlantic Ocean in the Mesozoic: The Northern Atlantic opened in the Jurassic⁴⁷, the Southern Atlantic separated Africa and South America during the mid-Cretaceous⁴⁸.

In the north, the Tropical Western Atlantic is delimited by the southward flowing cold-water Labrador Current. The onset of that circulation pattern (or proto-versions of it) is a matter of current debate; dates range from as early as the latest Cretaceous to the Miocene⁴⁹. The old age is mirrored in high species-level differences between the Tropical Western Atlantic and Arctic faunas, being almost as high as between Western and Eastern Atlantic tropical faunas.

Similarly, New Zealand faunas were found clearly distinct in most analyses. The continent of Zealandia diverged from Gondwana back to the Late Cretaceous⁵⁰, leading to long-term isolated faunal development of shallow marine biota⁵¹, which explains at least in part the observed faunistic separation.

The full opening of the Drake and Tasmanian passages to deep oceanic circulation during the latest Eocene to early Oligocene led to the formation of the Antarctic Circumpolar Current^52,53. In addition to polar temperatures, this old oceanographic system explains the high faunal distances of the Antarctic to any other faunal region. On average, distances are higher than between the Arctic and adjacent regions.

The formation of the Gomphotherium land bridge between Africa and Europe closed the Tethys Seaway in the Early Miocene²⁵. Faunistic connectivity for benthic molluscs ceased even several million years before, causing an isolated development of Indian Ocean and Mediterranean faunas since ca. 21 Myr ago⁵⁴. Combined with the different thermal regimes (tropical versus warm-temperate), this contributes to the high faunistic distance between the respective clusters in our analyses. The current anthropogenically induced biotic immigration from the Indo-Pacific into the Mediterranean since the opening of the Suez canal in 1869, as well as continuing global warming^55,56,57,58, tend to attenuate this differentiation.

The Indonesian Passage between the Indian and Pacific oceans has been restricted from approximately 11 Myr ago (early Late Miocene) onward but has remained open until today⁵⁹. This rather young geological tendency may explain why the Tropical Indo-Pacific is still a rather homogenous cluster for species and genera, despite major diversification of this tropical diversity hotspot already since the Miocene^27,30. Faunistic differences within the Tropical Indo-Pacific relate to restricted seas (e.g. Red Sea, Persian Gulf) or subregions, such as the northern Indian Ocean coastlines, rather than a distinction between Pacific and Indian Ocean faunas (Fig. 1).

The Humboldt Current System, leading to the particular biogeographic signature discussed above, is a rather young geological feature in its present form. While the characteristic upwelling signatures have prevailed at least since the Late Oligocene, an increased contribution of cold Antarctic waters starting ca. 6 Myr ago coincided with major turnovers in the Miocene and Pliocene^42,60. This is also reflected in our analyses: the region’s fauna is closely related to Antarctic-type species assemblages.

The faunistic distance between the Tropical Western Atlantic and Eastern Pacific has its origins in the younger geological history of the region: the closing of the Central American Seaway and formation of the Isthmus of Panama accordingly. The gateway was ultimately closed by ca. 2.8 Ma⁶¹, but deep ocean circulation had been constrained already since the Miocene^62,63,64. This strongly increased both the intensity and seasonality of the North Atlantic Current and Gulf Stream considerably⁶². This geologically young development had an impact on species distributions but apparently not (yet) on genus-level similarities, which show no distinction among Pacific and Atlantic faunas (Fig. 1).

Apart from the geological time these ocean circulation patterns have existed, variation of their intensity and seasonality throughout Earth history, and particularly during the Quaternary icehouse climate (e.g. related to Milankovitch-scale climatic oscillations), have had additional impact on the composition and distribution of benthic marine life – including but not limited to molluscs. Going into detail is beyond the scope of this globally-focused study, but examples exist^65,66,67,68. Future studies may test for a correlation between faunistic distances and the evolutionary age of modern faunas, particularly under the consideration of fossil assemblages.

Previous biogeographic studies – where do molluscs diverge?

The biogeographic zonation constructed by the cluster analyses largely mirrors existing studies based on a larger set of marine organisms^2,6,7. Our results are particularly similar to those of Kocsis et al.⁷, based on a network analysis of a dataset containing bryozoans, brachiopods, bivalves, gastropods, zooxanthellate stony corals, echinoderms, and decapods. For example, their regions termed Arctic, European, Tropical East Atlantic, South African, Tropical East Pacific, Tropical Indo-Pacific, Temperate Australian, and New Zealandian match the regions in our species-level cluster analyses (Fig. 1A) almost perfectly. Differences are noted for the Warm-temperate Southeast Pacific, which is delimited in our analyses as cluster separate from the Antarctic but not distinguished by Kocsis et al.⁷, and the Cold-temperate Northwest Pacific, which we find to be part of the cluster also including the Yellow Sea.

Costello et al.⁶ used a different approach using pre-defined sea areas as well as a larger dataset, including additionally selected groups of algae, annelids, nematodes, sponges, and fish. While a few selected regions, e.g. the Tropical East Pacific, Southern Africa, Southern Australia, New Zealand, and Antarctica, largely match ours, the majority does not. The empirical zonation presented by Spalding et al.² matches largely our regions, but the borders differ when observed in detail. Also when considering studies on specific mollusc clades or biogeographic regions^18,69, the results slightly deviate from ours, but yet again this may relate to different data sources or analytical approaches.

Especially since the results of Kocsis et al.⁷ largely match ours, we expect that the considerable discrepancies compared to other studies either root in the different methodology or the type of faunas included or both, rather than assuming a major difference of the biogeographic signals of molluscs compared to other groups. Particularly, the inclusion of nektonic and planktonic taxa by Spalding et al.² and Costello et al.⁶ certainly affects the broader picture, since taxa with these types of lifestyle differ markedly in their dispersal potential.

Limitations

Studies like this one relying on large databases come with certain limitations. While the amount of data stored in OBIS and GBIF is impressive, the data represent unbalanced input combining many different types of source and concepts. We have identified a number of points that potentially causes variation in the dataset, but we do not expect a major bias of our results and conclusions.

Geographic coverage: varying research focus and intensity over the past centuries and different efforts to make data available via online portals among researchers have led to considerable geographic variation in the amount and spread of occurrence data available through OBIS and GBIF^70,71. In our particular case, the shallow-water regions off North America, northern Europe, Australia, and New Zealand have provided large sets of occurrences to these databases and are overrepresented in the dataset (Figure S1). Additionally well covered are the shores of countries like Japan, China, South Korea, South Africa, as well as certain hotspots such as the Philippines, New Caledonia, Hawai’i, and many other tropical island archipelagos. Comparatively poor is the record for parts of the southern Atlantic, especially along western Africa, the western and northern Indian Ocean, as well as coast segments in the Arctic and Antarctic (Figure S1). While we do not expect polar faunas to yield significantly higher diversity or different taxonomic compositions with more intensive sampling, the underrepresentation of tropical to subtropical regions in the Atlantic and Indian oceans most likely has an effect on the observed biogeographic patterns (compare Ondo et al.⁷² for geographic biases in plants and Borgelt et al.⁷³ for a conservation perspective). Probably, this bias contributed to the limited capability of the cluster analyses to find clear biogeographic boundaries in the Southeastern Atlantic⁶⁹ or allowing for a more refined hierarchy of clusters in the Tropical Indo-Pacific. Future research and data mining effort should be directed toward filling these gaps.

Taxonomic resolution and accuracy: a large portion of the occurrence records only provide genus or family names (9.6% and 2.6%, respectively). In other cases, the reliability of identifications is questionable. The latter is especially true for unrevised museum collections bearing partly outdated identifications or widely known misidentifications that have not been accounted for. All these aspects concur with a seriously declining taxonomic expertise over the past decades^74,75,76. We fear that this detrimental trend will continue to threaten data quality also in coming years. Without the input of experts, high-resolution studies on the species level will become increasingly difficult and unreliable.

Taxonomic currentness: many clades have not been revised in the past decades or generally remain poorly known because of small body size or identification problems (e.g. species with few diagnostic characters). This problem affects entire groups such as mega-diverse families of microgastropods (e.g. Triphoridae, Eulimidae, Pyramidellidae, or the conoidean families Mangeliidae and Raphitomidae), or nondescript bivalves such as protobranchs or Thyasiridae^77,78,79.

Taxonomic practice: different schools around the world have used different taxonomic and systematic approaches and conventions and continue to do so, causing larger datasets to host a convolute of inconsistent and partly contrasting concepts⁸⁰.

Methodology: most species are still identified purely based on external characters rather than anatomical, genetic or genomic data or a combination thereof, leading to species being defined using different species concepts. This variation has potentially huge impact on studies of biodiversity, biogeography, and conservation biology^80,81.

Unknown diversity: several studies over the past years have stressed the importance of cryptic diversity and the potentially huge amount of undetected cryptic species^82,83,84.

Technical issues: inconsistencies and variation in data quality in huge databases such as OBIS and GBIF have been shown to be potentially problematic for biodiversity analyses^85,86,87. These link particularly to varying precision or erroneous records of geographic coordinates, partly as a function of georeferencing or projection issues, records of marine taxa on land or vice versa, and records of extinct taxa or still living taxa in the fossil record not marked as such. Although we have followed existing pipelines to account for such problems^85,86 (see Methods chapter), a number of erroneous records may have been overlooked.

Environmental factors: here, we focus on temperature as a prime factor explaining differences in taxonomic composition. In the future, also other factors, such as seasonal temperature variation, nutrients, spatial complexity, and the nature of predominant substrate, may be taken into consideration to further specify potential drivers of biogeographic structure.

Lacking quantitative framework to explain the relationship between compositional differences among larger regions and the age of a connecting seaway or disrupting land bridge: Unfortunately, the paucity of such gateways limits the possibility of a reliable statistical inference. In addition, the ages available for many of them are poorly constrained or involve long temporal ranges of uncertainty. A similar problem concerns the ages of ocean currents, some of which have existed for millions of years but may have significantly shifted in their intensity through geological time (see e.g. discussion on the Humboldt Current). As such, not their onset but their intensification may have sparked biogeographic provincialisation. More data and better age constraints are needed to allow for a reliable quantitative assessment.

Despite all these technical and conceptual biases, we are confident that our interpretations hold true on the global scale. Nonetheless, we encourage further research to fill the abovementioned gaps to limit potential biases for future studies and allow a more refined picture of biogeographic relationships.

Conclusion and outlook

Our findings show the importance of selected factors for biogeographic structure. Temperature and palaeogeographic legacy, the latter adding east–west barriers to the persistent north–south temperature thresholds, both have had a significant impact on benthic shallow-water marine life. More precisely, temperature explains a significant proportion of benthic mollusc biogeographic relationships. This association is mirrored by the match between the biogeographic units defined by our cluster analyses and present-day ocean circulation patterns. The similarities among biogeographic units at least partly relate to palaeogeographic history and associated changes in ocean currents. While we lack sufficient data for a quantitative approach, older palaeogeographic events tend to correspond to higher distances. However, in part this is also the result of different temperate regimes. Future studies may focus on disentangling the temperature versus legacy effect and their individual contributions to the observed biogeographic patterns, as well as include additional factors.

Our results have various implications. First, our analyses on three systematic levels indicate major differences in the biogeographic relationships and the influence of temperature on them. Considering the high-resolution picture presented by the species-level data compared to the genus or family level, we emphasize the need of utilizing species data, both for extant and fossil studies, for meaningful and detailed results. This also entails the pressing need for more taxonomic experts being able to identify taxa to the species level – 12.2% of the taxon records in the final dataset are identified on genus and family level only.

Second, the rapid climate change nowadays entails massive alterations of sea-surface temperature that strongly affect ocean currents. A prime example is the predicted break-down of the Atlantic Meridional Overturning Circulation within the century^88,89. Direct or indirect anthropogenic activity also impacts the observed biogeographic structure in other ways, e.g. through habitat destruction, pollution, overfishing, or the introduction of invasive species^55,90. Not only do these stressors affect the boundaries of biogeographic units, they ultimately threaten the existence of unique ecosystems, faunas, and species, many of which have a long and rich legacy. The strong relationship with temperature we show stresses the urgent need for countermeasures against climate change to avoid shifts in biogeographic structure of marine benthic life on earth. The picture we draw here based on molluscs is readily transferrable to other benthic groups.