2.1 Geological setting

The Juan de Fuca Ridge (JdFR) (Fig. 1a) is an intermediate spreading-rate ridge between the Pacific and Juan de Fuca plates in the north-east Pacific Ocean. The Endeavour Segment (47^∘57^′ N, 129^∘06^′ W) (Fig. 1b) constitutes a ∼ 90 km long section of the JdFR, bounded to the north by the Middle Valley and to the south by the Cobb Segment. It is characterised by a 500–1000 m wide axial valley whose walls reach up to 200 m in height (Delaney et al., 1992). The five major vent fields – Sasquatch, Salty Dawg, High Rise, Main Endeavour Field (MEF) and Mothra – found along the Endeavour axial valley are spaced regularly, about 2–3 km apart.

The MEF (Fig. 1c) is the most active of the five hydrothermal fields, with the presence of high-temperature (370–390 ^∘C), actively venting sulfide edifices and diffuse low-temperature (10–25 ^∘C) venting areas (Delaney et al., 1992; Kelley et al., 2012). This vent field is the world's first hydrothermal marine protected area (Devey et al., 2007; Tunnicliffe and Thomson, 1999) and was selected as a target site for the deep-sea cabled observatory Ocean Networks Canada, providing unprecedented opportunities to better understand vent ecology. Within the MEF, Grotto (47^∘56.958^′ N, 129^∘5.899^′ W; Fig. 1d) is an active hydrothermal sulfide vent cluster (15 m long by 10 m wide by 10 m high) located at 2196 m depth that forms an open cove to the north. This edifice is characterised by high short-term variation in heat flux but is stable across years (Xu et al., 2014). Like many other MEF hydrothermal edifices, the site is largely colonised by dense assemblages of Ridgeia piscesae (Polychaeta, Siboglinidae) with their associated fauna (Bergquist et al., 2007; Govenar et al., 2002; Sarrazin et al., 1997; Urcuyo et al., 2007).

2.2 Faunal assemblage sampling

Sampling took place during the ONC (Ocean Networks Canada) oceanographic cruises Wiring the Abyss 2015 and 2016 from 25 August to 14 September 2015 on the R/V Thomas G. Thompson, and from 10 May to 29 May 2016 on the E/V Nautilus, respectively. Using the remotely operated vehicles (ROVs) Jason and Hercules, three samples of Ridgeia piscesae tubeworms and their associated fauna were sampled each year at different locations on the Grotto hydrothermal edifice (n=6; S1 to S6, Fig. 2). These samples corresponded to three distinct assemblages (assemblage III, assemblage IV and assemblage V low flow) according to Sarrazin et al. (1997). For each sample, a checkerboard of 7×7 mm squares was first placed on each tubeworm sample to estimate the surface area. Then, the first suction sample was taken to recover the mobile fauna, followed by collection of tubeworms and their associated fauna, which were placed in a BioBox using the ROV's mechanical arm. A final suction sample on the bare surface was performed to recover the remaining fauna. The final sampled surface area was filmed with the ROV camera to estimate its surface using imagery (see protocol in Sarrazin et al., 1997).

2.3 Sample processing

2.3.1 Sample processing and identification

On board, all faunal samples were washed over stacked sieves (250 and 63 µm mesh sizes). Macrofaunal specimens (> 250 µm) were preserved in 96 % ethanol and meiofauna (250 µm > × > 63 µm) in 10 % seawater formalin. In the laboratory, bushes of Ridgeia piscesae were thoroughly disassembled and each tube was washed and sieved a second time. All associated macrofaunal organisms were sorted, counted and identified to the lowest possible taxonomic level. Specimens whose identification was unclear were sent to experts for identification and/or description. Trophic guilds from the literature (symbiont host, bacterivore, scavenger/detritivore or predator) were compiled for each vent species (Table 1). For species with unknown diets, the compilation was based on trophic guilds identified from closely related species (within the same family). These were then compared to our isotopic results (see below).

Table 1Trophic guild and nutritional modes of macrofaunal species associated with the Ridgeia piscesae tubeworm assemblages of the Grotto hydrothermal edifice (Main Endeavour Field, Juan de Fuca Ridge). An asterisk (*) marks the original description of the species.

^a This species refers to Hesiospina legendrei that is currently under description (Bonifácio et al., 2018).
^b This species refers to Xylocythere sarrazinae that is currently under description (Tanaka et al., 2018).

Download Print Version | Download XLSX

2.4 Statistical analyses

In the present study, Ridgeia piscesae was regarded as a habitat builder and thus discarded from all statistical analyses. Species–effort curves were computed for each faunal sample collected to assess the robustness of the sampling effort. Local diversity (i.e. α diversity) was estimated for each tubeworm sample from several complementary indices (Gray, 2000) using the vegan package in R (Oksanen et al., 2017): species richness (S), exponential Shannon entropy (D), Simpson's (1-λ^′) indices of species diversity and Pielou's evenness index (J^′).

Table 2Univariate measures of macrofaunal community structure associated with Ridgeia piscesae tubeworm bushes on the Grotto edifice: sample surface area (SSA), tube height (TH), tube diameter (TD), tubeworm surface area (TSA), volume (V), species richness (S), exponential of Shannon entropy (D), Simpson's diversity index ($\mathrm{1}-{\mathit{\lambda }}^{\prime }$) and Pielou's evenness (J^′).

Download Print Version | Download XLSX

Figure 3Rarefaction curves for species richness in the six vent samples (S1 to S6) collected on the Grotto hydrothermal edifice.

Download

3.1 Species–effort curves, tubeworm structural complexity and diversity

The rarefaction curves (excluding Ridgeia piscesae; Fig. 3) showed that, overall, the collected samples (S1 to S6) gave a fairly good representation of the species diversity on the Grotto hydrothermal edifice. In 2015, sample S2 (24 species) and S3 (31 species) rarefaction curves seemed to reach a plateau. S1 consisted of a total of 28 macrofaunal species. The samples from the year 2016 exhibited lower species richness and did not reach an asymptote. Samples S4 and S5 both had a macrofaunal species richness of 19 species, while only 14 species were found in sample S6 (Fig. 3).

Tubeworm surface areas and volumes of the samples were used as proxies for habitat complexity provided by the engineer species R. piscesae (Table 2). Samples S1 and S3 were characterised by a similar degree of structural complexity, with a tubeworm surface of 4.27 and 4.26 m², respectively (Table 2). Sample S2 displayed a sampling area of less than half of that of S1 and S3 reaching a total tubeworm surface area of 1.63 m². Samples from 2016 were substantially smaller than those from 2015. Tubeworm surface areas varied between 0.1 and 0.57 m². To summarise, the tube surface areas of the different samples were as follow: S1 > S3 > S2 > S5 > S6 > S4 (Table 2). Sample volumes were also strongly correlated ($R_{\mathrm{adj}}^{\mathrm{2}}=\mathrm{0.82}$; p value = 0.008) to the tubeworm surface area for these Grotto samples and thus can be used as an easily measurable proxy to estimate habitat complexity.

Alpha diversity measures showed that S4 displayed the lowest diversity (D=2.605; 1-${\mathit{\lambda }}^{\prime }=\mathrm{0.550}$), followed by S6 (D=3.998; 1-${\mathit{\lambda }}^{\prime }=\mathrm{0.633}$) and S5 (D=4.348; 1-${\mathit{\lambda }}^{\prime }=\mathrm{0.697}$). The highest diversity values observed were S1 (D=5.377; 1-${\mathit{\lambda }}^{\prime }=\mathrm{0.728}$), slightly less than S2 (D=5.398; 1-${\mathit{\lambda }}^{\prime }=\mathrm{0.749}$) and S3 (Shannon D=6.053; 1-${\mathit{\lambda }}^{\prime }=\mathrm{0.778}$). The S2 and S3 samples showed a more even distribution (J^′) of individuals among species than the other samples. In contrast, S4 had the lowest evenness ($J^{\prime }=\mathrm{0.325}$) (Table 2). Species richness was significantly correlated with R. piscesae tube length ($R_{\mathrm{adj}}^{\mathrm{2}}=\mathrm{0.62}$; p value = 0.039).

Table 3Percentage abundance × 100 (% Ab.), faunal density (D. ind m⁻²), volume (V. ind m⁻³) and relative biomass × 100 (% Biom.) of the different macrofaunal taxa (> 250 µm) identified in the six sampling units (S1 to S6) on the Grotto edifice. The taxa were identified to the lowest possible taxonomical level.

Download Print Version | Download XLSX

3.2 Composition and structure of Grotto vent communities

The species list and abundances for each sample collected within the Grotto hydrothermal edifice are provided in Table 3. A total of 148 005 individuals representing 35 macrofaunal species (excluding Ridgeia piscesae) were identified in the six tubeworm bushes (S1 to S6) sampled on the Grotto edifice. Overall, gastropods (5 species) and polychaetes (18 species), respectively, accounted for 69.5 ± 18.8 % and 16.3 ± 11.8 % of the total macrofaunal abundance. The numerically most abundant species were the gastropods Lepetodrilus fucensis and Depressigyra globulus as well as the polychaete Amphisamytha carldarei. The highest macrofaunal densities were observed in samples S5 (59 450 ind m⁻²), S4 (23 784 ind m⁻²), S3 (15 452 ind m⁻²) and S2 (13 748 ind m⁻²), whereas S1 and S6 had the lowest densities, with 7212 and 6518 ind m⁻², respectively. A high percentage (30 %) of the species was only found in one or two samples.

More specifically, S1 was dominated by gastropod species such as L. fucensis, D. globulus and Provanna variabilis (Table 3). High densities contrasted with low biomass were observed for the ampharetid polychaete A. carldarei and the syllid polychaete Sphaerosyllis ridgensis. S2 was also dominated by L. fucensis, D. globulus and A. carldarei, with, however, a high proportion of ostracods Euphilomedes climax (Table 3). S3 was largely dominated by A. carldarei and, to a lesser extent, was almost equally dominated by L. fucensis and D. globulus. Polychaetes were also dominant, with the presence of S. ridgensis, the dorvilleid Ophryotrocha globopalpata and the maldanid Nicomache venticola. There were high densities of P. variabilis, the solenogaster Helicoradomenia juani, the halacarid Copidognathus papillatus, the ostracod Xylocythere sp. nov. and the pycnogonid Sericosura verenae (Table 3). S4 was dominated by L. fucensis and D. globulus and, to a lesser extent, by the alvinellid polychaete Paralvinella palmiformis (Table 3). S5 was also dominated by L. fucensis and D. globulus, followed by E. climax and P. variabilis (Table 3). Finally, S6 was also dominated L. fucensis and D. globulus and, to a lesser extent, by A. carldarei and the alvinellid polychaete Paralvinella pandorae (Table 3).

Figure 4Stable isotope bi-plots showing vent consumers' isotope signatures (mean δ¹³C versus δ¹⁵N values ± standard deviation) for the six vent assemblages sampled on the Grotto hydrothermal edifice. Each vent species is designated by a number: 1 – Ridgeia piscesae; 2 – Provanna variabilis; 3 – Depressigyra globulus; 4 – Lepetodrilus fucensis; 5 – Buccinum thermophilum; 6 – Clypeosectus curvus; 7 – Amphisamytha carldarei; 8 – Branchinotogluma tunnicliffeae; 9 – Lepidonotopodium piscesae; 10 – Levensteiniella kincaidi; 11 – Nicomache venticola; 12 – Paralvinella sulfincola; 13 – Paralvinella palmiformis; 14 – Paralvinella pandorae; 15 – Paralvinella dela; 16 – Hesiospina sp. nov.; 17 – Sphaerosyllis ridgensis; 18 – Ophryotrocha globopalpata; 19 – Berkeleyia sp. nov.; 20 – Protomystides verenae; 21 – Sericosura sp.; 22 – Euphilomedes climax; 23 – Xylocythere sp. nov.; 24 – Copepoda; 25 – Copidognathus papillatus; 26 – Paralicella cf. vaporalis; 27 – Helicoradomenia juani. Known trophic guilds are distinguished by a colour code: pink – symbiont; green – bacterivores; blue – scavengers/detritivores; red – predators. For more information on the interpretation of guilds, please consult the web version of this paper.

Download

3.3 δ¹³C and δ¹⁵N isotopic composition

δ¹³C values of the vent fauna ranged from −33.4 to −11.8 ‰ among the different samples (Fig. 4). In more details, δ¹³C values ranged from −33.4 to −13.5 ‰ for S1, from −33.4 to −15.4 ‰ for S2 and from −32.4 to −14.7 ‰ for S3. Samples from S4, S5 and S6 displayed slightly narrower δ¹³C ranges, varying from −30.3 to −12.5 ‰, from −31.3 to −14.8 ‰ and from −32.3 to −11.8 ‰, respectively; most species were enriched in ¹³C relative to the S1, S2 and S3 samples (Fig. 4). Overall, the gastropod Provanna variabilis (species no. 2) was the most depleted in ¹³C, with values around −32.2 ‰ (±1.2 ‰). In contrast, Ridgeia piscesae siboglinids (species no. 1) showed the highest δ¹³C values, with constant values around −14.7 ‰ (±1.0 ‰). The range of δ¹⁵N values in faunal assemblages varied between −8.5 and 9.4 ‰ (Fig. 4). Especially, S1 values ranged from 0.3 to 8.4 ‰, S2 from 0.4 to 9.2 ‰, S3 from −2.7 to 8.3 ‰, S4 from −1.3 to 8.7 ‰, S5 from −1.1 to 6.4 ‰ and S6 from −8.5 to 9.4 ‰ (Fig. 4). Generally, 15 species showed a δ¹⁵N > 5 ‰ in S1, S2 and S3 assemblages but only 4 species were over 5 ‰ in δ¹⁵N in S4, S5 and S6. In contrast to their δ¹³C values, P. variabilis and R. piscesae displayed similar and relatively stable δ¹⁵N values among samples with 0.3 ‰ (±0.8 ‰) and 1.5 ‰ (±1.1 ‰), respectively.

Figure 5Stable isotope bi-plots showing vent consumers' isotope signatures weighted by biomass per square metre of tubeworms (filled circles) for the six vent assemblages (S1 to S6) sampled on the Grotto hydrothermal edifice. Considered as a habitat, the biomass of Ridgeia piscesae (denoted by a triangle symbol) is not shown. Each vent species is designated by a number: 1 – Ridgeia piscesae; 2 – Provanna variabilis; 3 – Depressigyra globulus; 4 – Lepetodrilus fucensis; 5 – Buccinum thermophilum; 6 – Clypeosectus curvus; 7 – Amphisamytha carldarei; 8 – Branchinotogluma tunnicliffeae; 9 – Lepidonotopodium piscesae; 10 – Levensteiniella kincaidi; 11 – Nicomache venticola; 12 – Paralvinella sulfincola; 13 – Paralvinella palmiformis; 14 – Paralvinella pandorae; 15 – Paralvinella dela; 16 – Hesiospina sp. nov.; 17 – Sphaerosyllis ridgensis; 18 – Ophryotrocha globopalpata; 19 – Berkeleyia sp. nov.; 20 – Protomystides verenae; 21 – Sericosura sp.; 22 – Euphilomedes climax; 23 – Xylocythere sp. nov.; 24 – Copepoda; 25 – Copidognathus papillatus; 26 – Paralicella cf. vaporalis; 27 – Helicoradomenia juani. For legibility, the biomass of P. pandorae in collection S6 is not shown.

Download

3.4 Biomass distribution in Grotto vent food webs

The projection of the species isotopic ratios weighted by biomass is useful for estimating the relative contributions of the different trophic pathways within the vent assemblages (Fig. 5). In our study, there were similar patterns of biomass distribution in the six sampled assemblages. In all six samples, the ecosystem engineering polychaete Ridgeia piscesae (species no. 1) represented the highest proportion of biomass (69.3 ± 16 %). However, as it was considered to be a structuring species of the ecosystem, it was not included in the following biomass distribution analysis. With a total proportion of biomass ranging from 78.9 to 95.8 % (89.6 ± 6.8 %) across samples, gastropods seemed to play an important role in the trophic food web of communities associated with the siboglinid tubeworms. The gastropod biomass was dominated by Lepetodrilus fucensis (species no. 4), which accounted for 31.5 to 82.8 % (55.8 ± 18.3 %) of the total biomass. In addition to L. fucensis, the gastropods Depressigyra globulus (species no. 3), Provanna variabilis (species no. 2) and Buccinum thermophilum (species no. 5) showed relatively high proportions of biomass within the different samples, ranging from 5.6 to 36.6 % (16.5 ± 13.8 %), 0.6 to 26.3 % (10.9 ± 9.8 %) and 0 to 16.1 % (6.4 ± 6.8 %), respectively (Fig. 5). However, in some samples, other species also significantly contributed to the total biomass. For example, in S3, the polychaete Amphisamytha carldarei (species no. 7) contributed substantially (7.2 %) to the total biomass. Similarly, in S4, the polychaete Paralvinella palmiformis (species no. 13) contributed to 16.4 % of the total biomass.

4.1 Communities and diversity

Hydrothermal ecosystems of the north-east Pacific are dominated by dense populations of the tubeworm Ridgeia piscesae. In this study, a total of 36 macrofaunal species (including R. piscesae) were found in the six tubeworm bushes sampled on the Grotto edifice, which is consistent with previous knowledge of the community in the region (Bergquist et al., 2007). In this study, macrofaunal species richness was slightly lower than that observed at the Easter Island hydrothermal site on the Main Endeavour Field, where a total of 39 species has been identified in a single R. piscesae bush (Bergquist et al., 2007). Tsurumi and Tunnicliffe (2003) reported the presence of 39 macrofaunal species in 25 collections from the Axial Segment (JdFR), but lower values have been reported on other segments, with 24 species in seven collections from the Cleft Segment (JdFR) and 19 species in two collections from the CoAxial Segment (JdFR). These levels of diversity are lower than that found in Riftia pachyptila bushes on the East Pacific Rise, where species richness in eight collections reached 46 species (Govenar et al., 2005). Macrofaunal diversity was also lower than those obtained in ecosystem engineering mussel beds from Lucky Strike on the Mid-Atlantic Ridge, with 41 taxa identified (Sarrazin et al., 2015), or from the northern and southern East Pacific Rise, with richnesses of 61 and 57 species, respectively (Van Dover, 2003). Variation between sites and regions may be related to discrepancies in sampling effort and methodologies. Alternatively, faunal dissimilarities between biogeographic regions may be closely related to the geological context, species colonisation history, connectivity to neighbouring basins, presence of geographic barriers (transform faults, hydrodynamic processes, depths, etc.), stability of hydrothermal activity, age of the vent system and distances between sites (Van Dover et al., 2002).

Tubeworm bushes sampled on the Grotto edifice were characterised by the dominance of a few species such as Lepetodrilus fucensis, Depressigyra globulus and Amphisamytha carldarei. Numerical dominance by a few species is a pattern that has also been reported from other hydrothermal sites of the world oceans: the Mid-Atlantic Ridge (Cuvelier et al., 2011; Sarrazin et al., 2015), the East Pacific Rise (Govenar et al., 2005), JdFR (Sarrazin and Juniper, 1999; Tsurumi and Tunnicliffe, 2001) and the southern East Pacific Rise (Matabos et al., 2011). Polychaetes were the most diverse taxa, representing half of the macrofaunal species richness with 18 species. Similar results have been reported within R. piscesae bushes on Easter Island, with the identification of 23 polychaete species (Bergquist et al., 2007). Although the dominant species were similar among samples, variation between samples mainly involved the relative abundance of the few dominant species and the identity of the “rare” species. These variations may result from differences in sampling strategies between years. The areas sampled in 2016 were smaller than in 2015, and a problem with the sampling boxes may have led to the loss of some individuals, even though not visible from videos recorded by the submersible. Variation in species richness and diversity among samples may also depend on the presence of environmental gradients, created by the mixing between ambient seawater and hydrothermal effluents (Sarrazin et al., 1999). Unfortunately, no environmental data were recorded with our samples. However, physical and chemical conditions are known to change along the ecological succession gradient on the MEF from newly opened habitat characterised by high temperature and sulfide concentrations, colonised by the sulfide worm Paralvinella sulfincola, to mature communities in low diffuse venting areas characterised by low temperatures and sulfide concentrations and colonised by long skinny R. piscesae tubeworms (Sarrazin et al., 1997). Tubeworm samples S1 and S3 were visually recognised as type V low-flow assemblages (Sarrazin et al., 1997), an advanced stage in the ecological succession characterised by longer tubes (18.5 ± 3.3 cm) and thus a higher level of structural complexity. Both samples showed the highest species richness, diversity and most complex food webs, suggesting a strong influence of engineer species and the importance of biogenic structure in the diversification and persistence of the local resident fauna. By increasing the number of micro-niches available for vent species, the 3-D structure of R. piscesae bushes helps to increase the environmental heterogeneity and thereby promotes species richness and diversity at assemblage scales (Jones et al., 1997; Tsurumi and Tunnicliffe, 2003). As mentioned by several authors (Bergquist et al., 2003; Govenar et al., 2002; Tsurumi and Tunnicliffe, 2003), various ecological mechanisms may explain the influence of R.  piscesae tubeworms on local diversity: new habitats generated by tubeworm bushes provide (i) a substratum for attachment and colonisation, (ii) interstitial spaces among intertwined tubes, increasing habitat gradients and therefore the number of ecological niches, (iii) a refuge to avoid predators and to reduce the physiological stress related to abiotic conditions, and (iv) a control on the transport of hydrothermal vent flow and nutritional resource availability. Samples S2 and S5, also identified as type V low-flow assemblages (Sarrazin et al., 1997), presented shorter tube lengths than S1 and S3, which might explain the lower species richness in these two samples. High densities of A. carldarei in R. piscesae tubeworm bushes (up to 93.4 %) may be related to this ampharetid's tolerance to environmental conditions and, therefore, to their ability to take advantage of a wide range of ecological niches (McHugh and Tunnicliffe, 1994). Similar to L. fucensis, A. carldarei is characterised by early maturity and high fecundity, contributing to the success of this species in vent habitats (McHugh and Tunnicliffe, 1994). The dominance of gastropods L. fucensis and D. globulus as well as the relatively high abundance of the Paralvinella polychaete species in samples S4 and S6 suggest that they belong to lower succession levels, corresponding to transitory states between types III and IV assemblages described by Sarrazin et al. (1997). The latter two samples were characterised by low species richness and diversities. We hypothesise that the numerical dominance of gastropods negatively affected species diversity by monopolising space and nutritional resources and by potentially grazing new recruits, therefore reducing the settlement of other vent species. As suggested by Sarrazin et al. (2002), the development of tubeworm bushes along the successional dynamics leads to the diversification of ecological and trophic niches, which may increase the complexity of the food web.

4.2 Trophic structure of tubeworm assemblages

The Ridgeia piscesae tubeworm assemblages of the Grotto hydrothermal edifice harbour a relatively diverse heterotrophic fauna. The isotopic analyses conducted on the most dominant vent macrofaunal species within the bushes revealed a high degree of resemblance in trophic structure among the six faunal samples. In this study, the position of species in the food webs (trophic structure) was consistent with the ones reported in Bergquist et al. (2007), with, however, less variability in carbon and nitrogen stable isotopes.

Hydrothermal food webs are generally based on two main energetic pathways: the transfer of energy from symbionts to host invertebrates and the consumption of free-living microbial production (Bergquist et al., 2007). In the present study, the contrasting isotope compositions of the gastropods Provanna variabilis, Lepetodrilus fucensis and the polychaete R. piscesae suggest a wide range of isotopically distinct, symbiotic and/or free-living microbial production available to primary consumers. The high δ¹³C values of R. piscesae were associated with chemosynthetic endosymbiosis linked to thiotrophic symbionts (Hügler and Sievert, 2011). Despite R. piscesae contributing to 86 % of the total biomass, a low number of species displayed similar δ¹³C values, suggesting that specialist species deriving the majority of their food sources from siboglinid tubeworms are rare. Similar observations, where engineer species contribute to the community more as a habitat than as a food source, have been reported in R. piscesae tubeworm bushes from the Easter Island vent site (Bergquist et al., 2007) and in Bathymodiolus azoricus mussel bed assemblages on the Tour Eiffel hydrothermal edifice (Lucky Strike, Mid-Atlantic Ridge) (De Busserolles et al., 2009). The low degree of exploitation of this large biomass and potential food resource suggests that rather than playing a trophic role, R. piscesae would primarily play a structuring role in vent ecosystems. Nevertheless, the δ¹³C and δ¹⁵N values of polynoids Branchinotogluma tunnicliffeae and Lepidonotopodium piscesae were consistent with a predatory diet including R. piscesae tubeworms. Moreover, the predation of tubeworms by polynoids is often observed, as shown on a video sequence from the ecological observatory module TEMPO-mini, deployed on the Grotto hydrothermal edifice (ONC observatory; Video S1 in the Supplement). The ¹³C-depleted stable isotope compositions of P. variabilis suggest a possible symbiosis with chemoautotrophic bacteria or reliance on feeding on a very specific free-living microbial community that depends on a ¹³C-depleted carbon source (Bergquist et al., 2007). To date, no study has reported the presence of chemoautotrophic symbionts in P. variabilis, but symbioses have been described for other species from the Provannidae family (Windoffer and Giere, 1997). With an intermediate δ¹³C composition between R. piscesae and P. variabilis, L. fucensis gastropods seem to represent a major energetic pathway in these vent communities. Food webs obtained in this study revealed that most vent species display a δ¹³C similar to L. fucensis, but with slightly higher δ¹⁵N values. Through its high densities, large biomasses and its position at the base of the food webs, we suggest that L. fucensis could play an important role in structuring vent communities. The wide-ranging feeding strategies of this limpet may exert a high pressure on the availability of nutritional resources for others vent species. Likewise, the range of feeding strategies open up more opportunities for L. fucensis, allowing this limpet to take advantage of resources that other species cannot. The four Paralvinella species observed in our samples, which are described as suspension and/or deposit feeders (Desbruyères and Laubier, 1986; Tunnicliffe et al., 1993), displayed low or negative δ¹⁵N values. These lowest δ¹⁵N values may be related to the nutrition of a microbial pool based on local nitrogen sources. In fact, the ammonium produced during the microbial degradation of organic matter appears to be usually ¹⁵N-depleted (Lee and Childress, 1996). Amongst these Paralvinella species, Paralvinella sulfincola, Paralvinella palmiformis and Paralvinella dela shared the same isotopic niche while P. pandorae displays a distinct isotopic composition.

Like in many vent food webs (Van Dover and Fry, 1994; Levesque et al., 2005; Limén et al., 2007), Grotto primary consumers were dominated by grazers and deposit feeders (Table 1). The high diversity, densities and biomass of bacterivores emphasise the importance of free-living bacteria in the establishment and maintenance of the structure of the vent food web (Bergquist et al., 2007). The bacterivore guild was mainly represented by the gastropods P. variabilis, Depressigyra globulus and L. fucensis and the polychaetes P. sulfincola, P. palmiformis, P. pandorae and P. dela. Like Paralvinella grasslei and Paralvinella bactericola at vent sites of the Guaymas Basin (Portail et al., 2016), the alvinellid species found at Grotto had comparable δ¹³C values but different δ¹⁵N signatures. Paralvinella pandorae showed a depleted δ¹⁵N signature relative to other alvinellid species. A previous study of isotope variability among three sympatric alvinellid species, P. palmiformis, P. sulfincola and P. pandorae on the JdFR reported that their differences in δ¹⁵N isotope composition could be closely related to the presence of food-source partitioning and/or to spatial segregation (Levesque et al., 2003). The comparatively small size of P. pandorae compared with other alvinellid species (Desbruyères and Laubier, 1986; Tunnicliffe et al., 1993) may be linked to the presence of interspecific competition for food resources and/or a diet based on an isotopically distinct microbial source. The wide range of δ¹³C signatures in bacterivores, coupled with the high interspecific variability in the isotopic space, suggest a large, diversified microbial pool in the hydrothermal ecosystem and high variability in isotope ratios in dominant microbial taxa. Detritivore/scavenger species were observed at an intermediate trophic level, between the bacterivore and predator feeding guilds. This guild was represented by a low number of species including the gastropod Buccinum thermophilum, the ampharetid Amphisamytha carldarei and the orbiniid Berkeleyia sp. nov. The predator feeding guild was represented by the highest δ¹⁵N values. High predator diversity was found in our vent assemblages and was associated with a wide range of δ¹³C values, covering the isotopic spectrum of lower trophic-level consumers (i.e. bacterivores as well as scavengers/detritivores). This guild of predators appears to be dominated by polychaetes, which tend to show the highest δ¹⁵N values. The syllid Sphaerosyllis ridgensis, the polynoid Levensteiniella kincaidi and the hesionid Hesiospina sp. nov. displayed the highest δ¹⁵N values, suggesting that they play the role of predators in the benthic food web. Similarly, the solenogaster Helicoradomenia juani consistently displayed higher δ¹⁵N values than other molluscs, indicating a predator trophic position. The presence of Zoarcidae eelpouts Pachycara gymninium and Oregoniidae spider crabs Macroregonia macrochira on the Grotto edifice, not sampled but observed in the video recorded by the TEMPO-mini ecological module, could also play the role of predator within the ecosystem. Except for the polynoid L. kincaidi, whose isotopic variability seemed to reveal a nutrition based on highly diversified food resources, stable isotope analyses conducted on predators revealed narrow ranges of δ¹³C and δ¹⁵N values at the species scale, suggesting the dominance of specialist feeding strategies, as was the case for bacterivores. An accurate assessment of the isotopic composition of food sources and a description of the meiofaunal communities would be necessary to further increase our understanding of the functioning of these chemosynthetic communities and their trophic structures.

4.3 Ecological niche partitioning

Vent species on the Grotto hydrothermal edifice exhibit high isotopic heterogeneity that reflects the complexity of vent ecological food webs. The distribution of species in the bi-dimensional isotopic space depends on their diets, environmental conditions and biotic interactions, which together define the concept of species ecological niche (Newsome et al., 2007) or the realised species trophic niche (Bearhop et al., 2004). Here, the fact that most of the isotopic space was occupied by isotopically distinct species shows that the available food resources are partitioned within the community. Although the δ¹⁵N variability among primary consumers did hinder our inference of trophic levels based on nitrogen isotopes, these communities are unlikely to host more than three trophic levels, given the overall δ¹⁵N ranges. Moreover, although predators were quite diverse, they only represented a minor part of the biomass, suggesting that Grotto vent communities are mostly driven by bottom–up processes. Food webs of chemosynthetic ecosystems – such as hydrothermal vents and cold seeps – do not appear to be structured along predator–prey relationships but rather through weak trophic relationships among co-occurring species (Levesque et al., 2006; Portail et al., 2016). Habitat and/or trophic partitioning are important structuring processes at the community scale (Levesque et al., 2003; Levin et al., 2013; Portail et al., 2016). Our results corroborate those from Axial Volcano in the JdFR (Levesque et al., 2006) and the Guaymas basin (Portail et al., 2016), where habitat heterogeneity induces spatial partitioning of trophic niches, leading to a spatial segregation of species and species coexistence (Levesque et al., 2006). Although the observed isotope variability (standard deviations) in Grotto vent species suggests the occurrence of both trophic specialists and generalists within the assemblages, the majority of vent species exhibited low standard deviations, suggesting a predominantly specialist feeding behaviour. As already shown in previous studies of vent sites with alvinellids (Levesque et al., 2003) and sulfidic sediments at methane seeps with dorvilleid polychaetes (Levin et al., 2013) in the north-east Pacific, food partitioning may occur between different species of the same or related taxonomic family, allowing species coexistence through the occupation of distinct trophic niches. Our study confirmed this pattern for alvinellid polychaetes of the genus Paralvinella (Levesque et al., 2006). Hydrothermal vent gastropods were numerically dominant in all Ridgeia piscesae bushes collected on the Grotto edifice, and their isotope compositions were fairly diverse. Gastropods exhibit great diversity in feeding strategies, and as a result they are found in a wide variety of niches where they exploit many food sources (Bates et al., 2005; Bates, 2007). The isotope composition of Provanna variabilis indicated low δ¹³C and δ¹⁵N values. Lepetodrilus fucensis gastropods had higher δ¹³C and δ¹⁵N values than P. variabilis but a similar range of δ¹³C as Clypeosectus curvus and Depressigyra globulus. However, these latter two species occupy an upper position in the trophic structure of their communities. The great ecological success of L. fucensis in vent habitats may be attributed to a combination of several characteristics. First, this species is characterised by a broad trophic plasticity that includes (i) grazing on siboglinid tubeworms and hard substrata (Fretter, 1988), (ii) active suspension feeding (Bates, 2007), and (iii) harbouring filamentous bacterial epibionts in its gills, which – via endocytosis – may contribute to the animal's nutritional requirements (Bates, 2007; Fox et al., 2002). In addition, the early maturity, high fecundity and continuous gamete production of L. fucensis may help to maintain the large populations on the edifice (Kelly and Metaxas, 2007). Stacking behaviour near fluid emissions also suggests that L. fucensis is an important competitor for space and food in the community (Tsurumi and Tunnicliffe, 2003). L. elevatus, the ecological equivalent of L. fucensis on the East Pacific Rise, is prey for the vent zoarcid fish Thermarces cerberus; the reduced limpet population promotes the successful settlement and growth of sessile benthic invertebrates such as tubeworms (Micheli et al., 2002; Sancho et al., 2005). The potential absence of an equivalent predator for L. fucensis and the biological characteristics detailed above may explain its ecological success on the north-east Pacific vent sites. In contrast, the nutrition of D. globulus is based on the grazing of organic matter only (Warén and Bouchet, 1989). However, its small size allows it to exploit interstitial spaces that are not available to larger fauna (Bates et al., 2005). Finally, P. variabilis was relatively less abundant than the other two species but appeared to exploit a different thermal niche than L. fucensis and D. globulus (Bates et al., 2005). On the other hand, the isotope composition of Buccinum thermophilum clearly differentiates that species from the other gastropods with higher δ¹³C signatures. Differences in the diets of co-occurring species may contribute to the high abundance and diversity of vent gastropods through niche partitioning (Govenar et al., 2015).

Habitat specialisation among co-occurring vent species may drive differences in their diets (Govenar et al., 2015), facilitating species coexistence in heterogeneous habitats such as hydrothermal ecosystems. We hypothesised that vent food webs display a small-scale spatial structure that is linked to the 3-D architecture of the biogenic structures generated by engineer species. This would promote high interspecific trophic segregation. The spatial segregation of trophic niches by environmental gradients limits the occurrence of biotic interactions such as predation and competition for resources between species sharing a common spatial niche (Levesque et al., 2006). Vent food webs may therefore be structured through the interplay between the availability and diversity of food sources and the abiotic and biotic conditions structuring species distribution.