The origin and evolution of the faunas inhabiting deep-sea hydrothermal vents and methane seeps have been debated for decades. These faunas rely on a local source of sulfide and other reduced chemicals for nutrition, which spawned the hypothesis that their evolutionary history is independent from that of photosynthesis-based food chains and instead driven by extinction events caused by deep-sea anoxia. Here I use the fossil record of seep molluscs to show that trends in body size, relative abundance and epifaunal/infaunal ratios track current estimates of seawater sulfate concentrations through the last 150 Myr. Furthermore, the two main faunal turnovers during this time interval coincide with major changes in seawater sulfate concentrations. Because sulfide at seeps originates mostly from seawater sulfate, variations in sulfate concentrations should directly affect the base of the food chain of this ecosystem and are thus the likely driver of the observed macroecologic and evolutionary patterns. The results imply that the methane-seep fauna evolved largely independently from developments and mass extinctions affecting the photosynthesis-based biosphere and add to the growing body of evidence that the chemical evolution of the oceans had a major impact on the evolution of marine life.
Hydrothermal vents and methane seeps in the deep sea are inhabited by highly specialized faunal communities that rely mostly on chemoautotrophic bacteria for nutrition . Typically, only a few species account for the vast majority of the biomass in these habitats, and these dominant species live in symbiosis with chemoautotrophic bacteria, a strategy called chemosymbiosis . Owing to their high degree of endemism, these vent and seep faunas were regarded geologically ancient ‘relic faunas' [3,4], but molecular age estimates and the fossil record indicate a Cretaceous to early Cenozoic origin of the major clades [5,6]. Oceanographic events such as widespread anoxia or biological triggers such as the appearance of whales, whose decaying carcasses are often colonized by a similar suite of animals and may therefore act as dispersal stepping stones, have been suggested as potential reasons for the geologically young origin of the vent and seep fauna [5,7,8]. Furthermore, because their diet is based on locally available reduced chemicals, mainly hydrogen sulfide and to a lesser extent methane and hydrogen , it was hypothesized that the evolutionary history of these faunas may be independent from that of photosynthesis-based ecosystems on the surface of our planet [6,10]. The ultimate source of sulfide at methane seeps is sulfate dissolved in seawater: it is produced by microbial, sulfate-dependent anaerobic oxidation of methane (AOM) . A recent non-steady-state box model of the global sulfur cycle indicated rapid fluctuations in seawater sulfate concentration through Earth's history that affected sulfate reduction rates at seeps  and consequently also the amount of released sulfide available to the biota. Therefore, the evolution and ecology of the seep fauna could have been affected by fluctuations in seawater sulfate concentrations. Here, patterns of size, relative abundance and epifaunal : infaunal ratios in the fossil record of seep molluscs are evaluated in the light of the potential effects of changing sulfate concentrations.
2. Material and methods
The analysis covers the last 150 Myr because the estimate of the sulfate concentrations covered the last 130 Myr  and because fossiliferous seep deposits older than Late Jurassic are scarce . Molluscs serve as model group because they have by far the richest fossil record among seep taxa and they have been extensively treated taxonomically in the past years by a small number of specialists, resulting in a relatively homogeneous dataset (electronic supplementary material, table S1). It consists of 678 records of species at 123 fossil seep deposits of Tithonian (Late Jurassic) to Miocene age, including relative abundance and size data. The geologic ages of the seep deposits are from the references cited in the electronic supplementary material, table S1, with updates provided by Kiel et al. . Relations between absolute ages and geological stages are based on the time scale of Gradstein et al. .
Most seep sites, recent and fossil, are dominated by a small number of very abundant species [15,16]; thus for the analysis of the dominant life habits at seeps only the most common taxa at each fossil site were analysed. A taxon was regarded as ‘most common’ at a given site when it was reported as ‘most common’ or ‘abundant’ or to ‘dominate’ the site, or when it was, according to the published specimen lists, at least three times more common than the next most common taxon. If two or three taxa occurred in large numbers, and those numbers were at least four times higher than that of the next most common taxon, all two or three taxa were regarded as ‘most common’.
Among the chemosymbiotic bivalves, Solemyidae, Nucinellidae, Thyasiridae and Lucinidae are regarded as infaunal, and the epifaunal and semi-epifaunal species of the Bathymodiolinae, Vesicomyidae and the fossil genus Caspiconcha are classified as epifaunal. Among the Vesicomyidae, very small sized species are probably infaunal , whereas species larger than 25 mm have been shown to be epifaunal [18,19]; thus vesicomyids smaller than 25 mm were assigned to the infauna. The Late Jurassic to Cretaceous bivalve Caspiconcha is regarded as chemosymbiotic because it is very large sized and very abundant at many seep deposits, and all large and abundant bivalves at modern seeps are chemosymbiotic [20,21]. While most bathymodiolin mussels have sulfur-oxidizing symbionts, some have acquired methanotrophic symbionts  and may thus rely only little on sulfide. However, only four to seven extant bathymodiolin species lost their sulfur-oxidizing symbionts entirely and this loss appears to be a geologically young event that took place only in early to middle Miocene time (approx. 16 Ma) . Thus, putative fossil, sulfide-independent bathymodiolins are unlikely to alter the conclusions of the present analysis.
Because the modelled seawater sulfate concentration curve shows two major shifts , the life habits of the most common taxa per site and the size ranges of the species were binned into sets of geological stages bordering these shifts (Tithonian to Barremian; Albian to Danian; middle Eocene to Miocene; figure 1). No data are available for the Aptian (Early Cretaceous) and Selandian—early Eocene (early Cenozoic). Among the Tithonian to Barremian records, 79% included size data and 70% included abundance data; among the Albian to Danian records, 89% included size data and 88% included abundance data; and for the middle Eocene to Miocene records, 67% included size data and 81% included abundance data (electronic supplementary material, table S1). To assess whether or not seep faunas of these three time intervals differed in average body size, the Kruskal–Wallis test and post hoc, pairwise Mann–Whitney tests were used; these statistical tests were performed using the software package past v. 2.17c .
3. Results and discussion
Two major shifts in marine sulfate concentrations have occurred since the Late Jurassic: a decrease in the mid-Cretaceous approximately 120 Ma due to the deposition of sulfur-bearing evaporites in the opening South Atlantic, and an increase in the early Eocene approximately 50 Ma presumably related to the erosion of sulfur-bearing evaporites during the early Himalayan orogeny . These two events coincide roughly with two major transitions in the evolution of the seep fauna (figure 1): a faunal turnover in the mid-Cretaceous followed by the demise of the dominant epifauna [21,25], and the appearance of many of the clades that inhabit modern vents and seeps in the middle Eocene, including the two clades of epifaunal chemosymbiotic bivalves that constitute a large portion of the biomass at modern vent and seeps (Bathymodiolinae and Vesicomyidae) [5,26]. Furthermore, the life habits of the most common taxa at seeps track the shifts in the modelled seawater sulfate concentrations (figure 1): during times of low sulfate concentrations (i) the proportion of epifaunal chemosymbiotic bivalves was low, (ii) the proportion of infaunal chemosymbiotic bivalves was high, and (iii) the proportion of epifaunal gastropods was high. A consequence of low sulfate concentrations at methane seeps should be that the rates of AOM were reduced and consequently less sulfide reached the sediment surface. Decreasing amounts of sulfide at the sediment surface should have been detrimental for epifaunal chemosymbiotic bivalves because fewer sulfides were available to sustain their sulfide-oxidizing symbionts, while epifaunal gastropods may have benefited from the reduced toxicity on the sediment surface.
Although infaunal chemosymbiotic bivalves were abundant during the Albian through early Cenozoic low sulfate interval, they were on average smaller than during times of higher sulfate concentrations: the median size is significantly smaller during the low sulfate interval than during the preceding and subsequent time intervals (figure 2a and table 1). Because infaunal chemosymbiotic bivalves (Solemyidae, Nucinellidae, Thyasiridae, Lucinidae) rely on sulfide for nutrition, this pattern suggests that these bivalves remained smaller as a reaction, or adaptation, to low sulfide availability. The change in size is seen not only among infaunal chemosymbiotic bivalves as a whole but also among genera that survived the shifts in sulfate concentrations (figure 2b). This indicates that the size changes were not driven by the appearance and disappearance of entire clades, but instead members of individual lineages were able to adapt to shifts in sulfide availability. Also, the epifauna as a whole was significantly smaller during the Albian through early Cenozoic low sulfate interval than during the preceding and the subsequent intervals (figure 2c and table 1). This analysis of the entire epifauna included also the gastropods; gastropods at seeps are not chemosymbiotic but feed on free-living bacteria and detritus, or are predators/scavengers . Thus the decrease in chemoautotrophic primary production induced by the low sulfate concentrations affected also secondary consumers. There is an interesting contrarian pattern among the diversity of a distinctive group of gastropods: the large, high-spired abyssochrysoids Abyssomelania, Ascheria and Humptulipsia are a conspicuous element among Early Cretaceous to Eocene seep faunas, with a peak diversity (though not size maximum) in the mid- to Late Cretaceous . This gastropod morphotype is absent from modern seeps where the only large gastropods are predatory neogastropods . Although the ecology of the large abyssochrysoids is unknown, morphologically similar caenogastropods are typically grazers on soft substrates . Thus, the peak diversity of the large abyssochrysoids in the mid- to Late Cretaceous supports the hypothesis that the less toxic soft sediments at seeps during times of low marine sulfate concentrations were beneficial for gastropods.
Did the shifting marine sulfate concentrations affect also life at hydrothermal vents? The fossil record of vent communities over the past 150 Myr is sparse  and does not allow an analysis as provided here for seep communities. Furthermore, sulfide at vents originates only partially from seawater sulfate that was reduced during interactions with the oceanic crust, while a variable amount comes from magmatic sources . However, the observation that vents and seeps are inhabited largely by members of the same taxonomic groups  and the assumption that this was also the case in the geological past, may allow the following speculation: a taxon that disappears only from seeps but not from vents should recolonize seeps once favourable conditions return, and this pattern should be particularly pronounced for epifaunal taxa owing to the preponderance of hard substrate at vents. Thus, the rise of new epifaunal clades (i.e. vesicomyids and bathymodiolins) after the strong rise of marine sulfate concentrations in the Eocene instead of the return of Cretaceous clades (i.e. Caspiconcha) may be interpreted as the result of the demise of epifaunal taxa also at hydrothermal vents during the Albian to early Cenozoic low sulfate interval.
Are there alternative explanations for the observed patterns? The distribution of seep carbonates through the investigated time internal is not uniform  but instead appears to be linked to seawater sulfate concentrations , raising the issue of sampling biases. Correlations between rock volume and faunal diversity or abundance can be contentious because they can either result from sampling bias or reflect a true biological signal, called species-area effect [34,35]. But in the present case, it is difficult to explain why more rocks should preferentially preserve more epifaunal taxa but fewer infaunal taxa, or produce opposing trends between epifaunal bivalves and epifaunal gastropods. The patterns should thus reflect the original ecology of the investigated seep communities to a relatively high degree.
Several hypotheses aim to explain the rise of the many new vent and seep taxa in the Eocene. The whale stepping stone hypothesis states that decaying whale carcasses act as dispersal stepping stones for vent and seep taxa and thus the rise of whales in the Eocene significantly expanded the dispersal capabilities of vent and seep taxa . Indeed, whale carcasses are sulfide-emitting hard substrates, which may have promoted epifaunal chemosymbiotic taxa in particular. One may ask, though, why during the Late Cretaceous large marine reptiles with their lipid-rich bones do not appear to have had the same effect [36,37]. The key may again be seawater sulfate concentrations: vertebrate carcasses emit sulfide because bone lipids are metabolized by sulfate-reducing microbes ; thus the low sulfate concentrations in the mid-Cretaceous to early Eocene may have reduced the extent of this process.
Molecular age estimates of various vent and seep clades including annelids, molluscs, shrimps, stalked barnacles and decapods suggest their origin or major radiation from the Eocene onward and it was argued that a putative extinction event at the Palaeocene–Eocene thermal maximum (PETM) extinguished incumbent clades and thus opened opportunities for new taxa [5,39]. However, the only study on vent and seep clades to date that attempted to estimate the impact of extinction events on molecular diversification rates found little evidence for extinction at the PETM . Furthermore, several apparently seep-restricted taxa, including the large abyssochrysoid Ascheria and Humptulipsia, survived both the end-Cretaceous mass extinction and the PETM [28,40,41], providing further evidence against a major role of extinction events in shaping the vent and seep fauna. Instead, it is suggested here that the high seawater sulfate concentrations from the middle Eocene onward resulted in increased sulfide availability at seeps, and possibly at vents and vertebrate carcasses too, providing an abundant food source to be exploited by a wide range of new taxa. A critical test for this hypothesis would be the discovery of seep faunas immediately preceding the early Eocene rise in sulfate concentrations, which are currently lacking.
My study indicates that the main macroecologic and evolutionary trends of the deep-sea methane-seep fauna through the last 150 Myr were driven by changes in seawater sulfate concentrations, which in turn were induced by plate tectonic processes . This implies that contrary to previous suggestions [5,39] the seep fauna indeed had an evolutionary history that is largely independent from the course of the photosynthesis-based biosphere and is buffered from extinction events affecting shallow-water ecosystems [6,10]. Furthermore, several seep-restricted taxa range across events of presumed global anoxia and extinction in the deep ocean, including the allegedly severe examples at the Cenomanian–Turonian boundary and the PETM, shedding doubt on the global and/or truly anoxic nature of these events. Metazoans have developed a diversity of symbiotic associations with different chemoautotrophic bacteria and take advantage of narrow redox boundaries . These range from external bacteria farming to highly integrated, intracellular symbioses, and the reduced chemicals for the symbionts are taken up from the water column, from the sediment, during continuous movements between oxic and anoxic environments, and bacterial growth can be supported by the active pumping of oxidized compounds into anoxic sediment, as in vestimentiferan tube worms . We are only beginning to understand the origin, timing and triggers of these adaptations . Shifting redox boundaries in the sediment through Earth's history, for example, in response to shifting sulfate concentrations, may well have played a significant role in this context. Many Palaeozoic and Mesozoic seep deposits were dominated by large sized and most probably filter-feeding brachiopods , but vent or seep sites dominated by filter-feeders are unknown in the present-day ocean. Recent work indicates, paradoxically, that these often superabundant brachiopods were better adapted to low rather than high sulfide flux, despite being restricted to the seep environment . Evaluating the role of low seawater sulfate concentrations during the Mesozoic  may provide new insights into such seep ecosystems with a radically different trophic structure than the modern ones. Thus, the results presented here add to the growing body of evidence [43,44] that the chemical evolution of the oceans had a major impact on the evolution of marine life.
Financial support was provided by the Deutsche Forschungsgemeinschaft through grants Ki802/6-1 and Ha1166/17-1.
I have no competing interests.
The manuscript benefited from insightful comments by Alex Nützel (Munich), Jörn Peckmann (Vienna) and three anonymous reviewers.
- Received November 25, 2014.
- Accepted January 26, 2015.
- © 2015 The Author(s) Published by the Royal Society. All rights reserved.