Stable isotope signatures and nutritional sources of some dominant species from the PACManus hydrothermal area and the Desmos caldera

Deep-sea hydrothermal vents in the western Pacific are increasingly explored for potential mineral extraction. The study of the composition of the food web plays an important guiding role in the ecological protection and restoration of potential mining areas. The general picture of the nutritional sources of species should be established to assess the potential impacts of future mining activities on the biological composition and food sources. To provide basic information, we analyzed the carbon and nitrogen stable isotope ratios of the dominant macrofauna (mussels, commensal scale worms, crustaceans, gastropods, and vestimentiferans) at three different sites in the PACManus hydrothermal area and the Desmos caldera. The δ13C ratio was significantly different between species: mussels and commensal scale worms showed lighter δ13C ratios, whereas crustaceans showed heavier ratios. In terms of δ15N, mussels had the lowest values and the crustaceans had the highest values. By taking into account these stable isotope signatures, we were able to develop inferences of the food sources for vent community organisms. We found that the food web was based on various species of chemoautotrophic bacteria. Mussels appeared to rely primarily on sulfur-based endosymbionts, which use the Calvin–Benson–Bassham (CBB) cycle and RuBisCO form I as the CO2-fixing enzyme. Commensal polychaetes mostly obtained their nutrition from their hosts. Crustacean species were omnivorous, feeding on chemosynthetic bacteria, sedimentary debris, or even animals according to the local environment. In contrast, gastropods relied mainly on symbiotic bacteria with some supplementary consumption of detritus. Vestimentiferans obtained food from symbiotic bacteria using the RuBisCO form II enzyme in the CBB cycle and may have several symbionts using different fixation pathways. Although most macrofauna relied on symbiotic chemoautotrophic bacteria, our study suggested a closer trophic relationship between animals. Therefore, to evaluate the potential impacts of deep sea mining, it is necessary to study the cascade effects on the food web of the whole ecosystem. Before exploiting deep-sea resources, further systematic investigations concerning the protection of deep-sea ecosystems are necessary.


Introduction
Hydrothermal vents occur in geologically active seafloor areas, such as mid-ocean ridges and volcanic hotspots [1,2]. Geothermally heated fluids emitted from the seafloor are rich in reduced gases and metals [3], and dense invertebrate communities thrive in the areas where these fluids mix with colder and oxygenated seawater [4]. Most hydrothermal deposits containing significant amounts of metals are of economic interest to mining companies [5]. Almost 20% of all known global vent fields currently fall within mining exploration leases [6], so basic information on the diversity and connectivity of vent systems is necessary for establishing effective environmental management framework [7].
The stability of the food web is one of the most important factors affecting ecological stability. There are complicated food relationships between different species to maintain stability. Understanding the food sources of the dominant species is a crucial step determining the food web and nutritional relationships. Oceanwide, several inorganic carbon sources underpin various ecosystems. Because deep-sea vent communities exist so far outside the photic zone and the ordinary photosynthetic basis for many food webs is unavailable, questions regarding the nutritional resources of those species have sustained scientific interest since their discovery [8]. The main carbon and nitrogen sources for hydrothermal vent fauna have been identified as being primarily of local origin rather than being imported from photic zone communities [9, 10]. As with surface systems, carbon dioxide and methane are the primary carbon sources utilized by autotrophic organisms [11,12], but these are fixed by chemoautotrophic bacteria rather than organisms utilizing photic energy. Such bacteria form the base of the food chain, oxidizing reduced materials in vent fluids and utilizing the energy released to fix carbon [13][14][15]. Vent community invertebrates always inhabit with these bacteria that exist freely or symbiotically as endobionts which can be extracellular or intracellular, or as epibionts [16]. Animals derive the vast majority of their nutrition from these relationships [17][18][19]. Epipelagic photosynthetic primary production may also support part of the trophic system, depending on the environmental characteristics of the vent system. Some upper sublittoral vents and upper bathyal vents have a mixed photosynthetic-chemosynthetic system, with photosyntheticallyderived organic matter as a nutritional source [20][21][22][23].
Stable isotope analysis is a useful approach for investigating such trophic interactions, evaluating community structure, and examining the trophodynamics of ecological communities [18,24,25]. The δ 13 C isotopic signature is a good indicator of consumed food sources since carbon trophic discrimination is small, ranging from 0 to 1.5 ‰ between successive trophic levels [26]. The ratio of the stable isotopes of nitrogen (δ 15 N) can be used to estimate trophic position, with typical enrichment in δ 15 N increasing by 3 to 4 ‰ per trophic level [27,28].
Carbon origins are distinct stable isotope ratios that are useful for determining the origin of energy. In surface layers, stable isotopes of dissolved carbon dioxide are relatively constant at approximately 0 ‰ [28]. The withdrawal of carbon to form carbonates involves little isotopic fractionation, whereas the uptake of dissolved inorganic carbon during planktonic photosynthesis involves a larger kinetic fractionation that results in algal values of approximately -19 ‰ to -24 ‰ [28]. Particulate organic matter (POM) in the oceans predominantly reflects a marine planktonic origin. However, POM could also reflect microbial signal (bacteria-archaea) at vent systems, especially in the vent plume or close to the vent emissions. Biological fractionation, the input of organically produced material, its oxidation, and changes in temperature and pH all influence the composition of isotopes [29,30]. δ 13 C values in dissolved inorganic carbon (DIC) are generally lower at depth compared to the surface water layers [29,31], in some places by -9 ‰ [32]. Methane is generally the main inorganic carbon source at cold seep fields [33], with values ranging from -60 ‰ to -110 ‰ [32,[34][35][36]. The type of inorganic carbon source used by producers in hydrothermal vent ecosystems can be estimated according to stable isotope signatures. The carbon isotopic ratios of animal tissues are lower or equal to -40 ‰when they used methane-based energy sources [13]. These signatures have been exploited in a number of previous trophic studies of hydrothermal vent communities around the world [18,22,24,25,37,38]. Most recently, stable isotope ratios have been used to examine the relationships within the trophic network in the Guaymas Basin [39], revealing variable nutritional pathways during the life time of the hydrothermal vent snail Ifremeria nautilei and barnacle Eochionelasmus ohtai manusensis from the Manus Basin, Western Pacific [40].
The Eastern Manus Basin (EMB), from the Bismarck Sea, north-east of Papua New Guinea, Western Pacific, is the location of a number of hydrothermal areas, including the Southeast Ridges region (SER) (Fig 1). The SER is a rift zone of pre-existing island arc crust that contains several sigmoidally-shaped volcanic ridges [41]. Both the Papua-Australia-Canada-Manus (PACManus) and the Desmos caldera hydrothermal areas are located in this region. The PAC-Manus site is a polymetallic type of mineral deposit consisting of sphalerite, chalcopyrite, bornite, wurtzite, pyrite, marcasite, enargite, tennantite, galena, Pb-As-sulfosalt, gold, covellite, digenite and chalcocite [5]. The Onsen site at the Desmos caldera has a typical acid-sulfate type of mineralization consisting of enargite, covellite, chalcopyrite, pyrite and marcasite [5]. Both vent fields are considered important sources of potential mineral wealth, and there is basically no information on the trophic relationships in the community. In these areas, a steady discharge of vent fluid supports well-developed biological communities, including bacterial mats, molluscs, tube worms, crabs, anemones, holothurians, crustaceans and fishes [42][43][44][45]. The species composition is similar across the Manus Basin fields, but the relative abundance varies from field to field, possibly as a result of the amount of time the communities have existed [43]. Overall, the vent fauna displays closer affinity with the communities of the North Fiji and Lau Basins than the Mariana Trough vent region [43].
This work presents the trophic structures of some dominant assemblages from the PACManus hydrothermal area and the Desmos caldera using stable isotope analysis. Specifically, our aims were: to provide descriptive information about the stable isotope signatures of some dominant species and to investigate potential food sources and nutritional relationships within the communities in order to lay the groundwork for subsequent studies and mineral extraction.

Sample collection and processing
A research cruise to the PACManus hydrothermal area and the Desmos caldera was attended in June 2015 by the R.V. Kexue, and sampling at depth was conducted using the remotely operated vehicle (ROV) Faxian belonging to the Chinese Academy of Sciences. Samples were collected from three sites several meters away from vents, and general information about these sites is shown in Fig 1 and (Fig 1D) was situated southeast of Desmos caldera at 1912 m depth. Several low-  temperature seeps (~1˚C) and inactive chimneys were observed around the sampling field. Abundant crustaceans were present on and between small pieces of rock. We collected a large number of specimens at each site. The squat lobster Munidopsis lauensis and the shrimp Rimicaris vandoverae were collected using a suction pump coupled to a rotating carousel of acrylic collection bottles. The gastropods Ifremeria nautilei (Site Dive #32) and Provanna nassariaeformis (Site Dive #36), the mussel Bathymodiolus manusensis, and the vestimentiferans species Arcovestia ivanovi were collected using a scoop controlled by an ROV arm and brought to the surface in a closed and thermally insulated biobox. The commensal scale worms Branchipolynoe spp. were found in the mantle cavity of B. manusensis. Samples were preserved on ice immediately upon arrival on board. Three or five voucher specimens of each species were preserved in a seawater-formalin solution for further taxonomic study. Three or more samples of each species were processed for stable isotope analysis. The gills and feet of B. manusensis; the tentacles, foreparts, and trunk of A. ivanovi; and the muscles of the gastropods and crustaceans M. lauensis and R. vandoverae were dissected. Muscle tissue was selected for nutritional analysis, and other tissues were used for comparative analysis. After separation, the tissue samples were rinsed with deionized water to remove any residual seawater and frozen at -80˚C.
POM samples for isotope analysis were collected from three water depths (a surface layer at 0 m, a mid-water layer at 800 m, and a demersal layer at 1637 m) using Niskin water bottles mounted on the ROV above the sampling site at Site Dive #32. Approximately 2 L of water was passed through a 200 μm sieve to exclude large prey items and then filtered through GF/F filters of 0.70 μm pore size that had been pretreated at 450˚C for 4 h. The filters were frozen at -80˚C after filtration.
In the laboratory, all faunal samples for stable isotope analysis were freeze-dried and homogenized in an agate mortar. One milligram of tissue was placed in tin capsules for carbon and nitrogen isotope analyses. Particulate organic matter samples were separated from filters and a subset was acidified to remove inorganic carbon and measure the δ 13 C signature of organic carbon only. Acidification was carried out by adding drops of 0.1 M HCl until effervescence ceased. The sample was then dried at 60˚C under a fume extractor to evaporate the acid. To prevent the loss of dissolved organic matter, samples were not rinsed [39].

Stable isotope analysis
The isotopic compositions were analyzed using an elemental analyzer (Flash EA 1112Ht, Thermo Fisher Scientific, Inc., San Diego, CA, USA) coupled with an isotope-ratio mass spectrometer (Finnigan Delta V Advantage, Thermo Fisher Scientific, Inc.). Stable isotope ratios are expressed in δ (‰) notation with respect to Pee Dee Belemnite (PDB) for δ 13 C and atmospheric N 2 for δ 15 N where X is either 13 C or 15 N, R sample is the 13 C/ 12 C or 15 N/ 14 N isotope ratio in the sample and R standard is the 13 C/ 12 C or 15 N/ 14 N isotope ratio for the standard. An internal standard (glycine) was run for every twelve samples. Measurement precision was 0.1 ‰ and 0.2 ‰ for δ 13 C and δ 15 N values, respectively.

Statistical analysis
The normal distribution of data was confirmed using the one-sample Kolmogorov-Smirnov Test. The homogeneity of variance was assessed using Levene's test. One-way ANOVA was performed to assess the significance of differences in δ 13 C and δ 15 N values across different species and different tissues when the data fits normal distribution and homogenous variance. A Kruskal-Wallis H-test was used when the data did not approach normality or homogeneity of variance. Multiple parametric comparisons were performed using the S-N-K multiple range test. A Mann-Whitney U-test was used when the data lacks normal distribution or homogeneous variance. SPSS Version 19 was used for all statistical analyses.

Stable isotope ratios in individual tissues and POM
The full suite of results is presented in Table 2. The δ 15 N of all taxa and POM were plotted against the corresponding δ 13 C (Fig 3). The δ 13 C value of POM in the surface water layer was the highest ( The gastropod I. nautilei and two crustaceans were collected at Site Dive #32. The gastropod was found in the middle of the animal clusters and on the stacks of shrimps. The ratios of I. nautilei were approximately -28.83 ‰ for δ 13 C and 5.03 ‰ for δ 15 N. The ratios of the two crustaceans M. lauensis and R. vandoverae, were higher than those found on gastropods. R. vandoverae had the highest ratios, which were approximately -10.57 ‰ for δ 13 C and 8.64 ‰ for δ 15 N.
For the taxa at Site Dive #33, the mussels B. manusensis and commensal worms Branchipolynoe spp. had lower δ 13 C values, and the mussels had the lowest δ 15 N values. The crustaceans M. lauensis had the highest δ 13 C and δ 15 N values, while other animals showed intermediate signals.
The values of δ 13 C and δ 15 N differed significantly between the gill and foot tissues of the mussels (p < 0.05). The mussels displayed little range, from -33.12 ‰ to -32.41 ‰ for δ 13 C and from 2.32 ‰ to 3.22 ‰ for δ 15 N in gills and from -31.96 ‰ to -31.57 ‰ for δ 13 C and from 2.19 ‰ to 3.79 ‰ for δ 15 N in foot tissue. The commensal scale worms were found in the mantle cavity of the mussels. The δ 13 C values did not differ between the worms and the mussel gill tissues, but a significant difference was present between the worms and the foot tissues of the hosts (p < 0.05). The values of δ 15 N in scale worm tissues were, on average, 3.6 ‰ higher than the values in the foot tissues of the host. For the vestimentiferans A. ivanovi, the δ 13 C values of all three parts were approximately -22 ‰ and notably high compared to the values of the mussels and commensal scale worms. In δ 15 N, however, the values in the trunk were especially low among the three parts, which were similar to the values of the mussels' foot. The values in the other two parts, tentacles and foreparts, were in the range of δ 15 N in mussels and the commensal scale worms. Although A. ivanovi vestimentiferans were collected in association with the mussels, no statistical comparison was possible because of the limited number of specimens.
The gastropod P. nassariaeformis and two crustaceans were collected at Site Dive #36. The ratios of P. nassariaeformis were approximately -22.30 ‰ for δ 13 C and 4.44 ‰ for δ 15 N. R. vandoverae showed the highest ratios of δ 13 C. The ratios of δ 15 N were similar between these two crustaceans.
M. lauensis crustaceans were collected from all three sampling sites and R. vandoverae crustaceans were collected at Site Dives #32 and #36. Both species showed higher values of δ 13 C and δ 15 N, with significant differences in these measures between fields (p < 0.05). The δ 13 C and δ 15 N values of M. lauensis ranged from -23.73 ‰ to -16.16 ‰ and from 4.55 ‰ to 10.02 ‰, respectively. The values of δ 15 N were highest at Site Dive #36, followed by Site Dive #32 and Site Dive #33. The δ 13 C and δ 15 N values for R. vandoverae ranged from -14.93 ‰ to -9.11 ‰ and from 8.04 ‰ to 9.85 ‰, respectively. The values of δ 13 C were higher at Site Dive #32, followed by Site Dive #36. The values of δ 15 N were higher at Dive #36, followed by Site Dive #32.

Potential carbon sources and POM
Generally, methane and carbon dioxide are the two common carbon sources at hydrothermal vents. The carbon sources of vent species could be inferred according to the carbon isotopic ratios in tissues. Carbon isotopic ratios of animal tissues were lower or equal to -40 ‰ when they used methane-based energy sources. When animals used heterotrophic and sulfur-based energy sources, the carbon ratios of tissues were above -40 ‰ [33]. In addition, the dual symbiosis of thiotrophic and methanotrophic bacteria occurred in many Bathymodiolus species, whose isotopic values were related to the proportion of different bacteria present [47,48]. Moreover, methane oxidation occurs less frequently in the basic carbon fixation process, although it is energetically comparable to sulfide in oxidation potential [35,49]. Because metazoans are unable to bind and store methane, species with endosymbiotic methanotrophic bacteria may thrive only in environments with stable and elevated methane levels [35]. Our data, which were all above -40 ‰ for δ 13 C in tissues, suggest that the main carbon source for all sampled animals in our study was carbon dioxide, which is consistent with previous findings showing that sulfide oxidation was the primary energy acquisition pathway at hydrothermal vents [50,51]. Essentially, two distinct carbon isotopic signals existed at the vents: one group with a composition of approximately -10 ‰ to -15 ‰, which usually included shrimps and vestimentiferans, and a second group with an isotopic signal of -20 ‰ to -30 ‰, which usually included mussels [25,52,53]. This strongly suggests that the use of carbon fixation pathways by chemoautotrophic bacteria differs between the two groups [52,54]. Heavier values can be explained by the operation of the reverse tricarboxylic acid (rTCA) cycle, with a fractionation of 2 ‰ to 14 ‰ [55][56][57]. The lighter values can be explained by the operation of the Calvin-Benson-Bassham (CBB) cycle. This pathway includes two types according to different forms of RuBisCO. Here, the fractionation by bacteria using RuBisCO form I ranges from 22 ‰ to 33 ‰ [53,58,59], whereas the fractionation by RuBisCO form II ranges from 18 ‰ to 22 ‰ [53]. Compared with the two carbon fixation pathways, fractionation via the rTCA cycle contributes very little, resulting in a higher δ 13 C value in chemoautotrophic bacteria. In contrast, the δ 13 C value is lower in bacteria using the CBB cycle.
Among the isotopic values of the POM in the three water layers in our study, the δ 13 C values of the POM at the surface layer were high, while the δ 15 N values were low, reflecting the abundance of phytoplankton. Photosynthetic primary production in phytoplankton takes place using carbon dioxide as a carbon source in the photic zone, placing phytoplankton at the bottom of the food web with the lowest δ 15 N values compared to other organisms [26,27]. The δ 15 N values of POM at 800 m were the highest, likely due to the presence of detritus or remains because of the higher δ 15 N of animals at higher trophic levels [27]. The δ 15 N values of the deep sea at 1637 m decreased compared to the middle layer, possibly due to a mixture of materials from the hydrothermal vent fields and detritic fall from the upper water layers. Because POM was mainly made up of chemoautotrophic bacteria which had lowest δ 15 N values as primary producers at vent ecosystems, the values may be situated between the values of chemoautotrophic bacteria and detritic fall.

Mussels and commensal scale worms
Mussels are common in hydrothermal fields and always thrive on the bacterial mats or rocks in overflow areas [42,60,61]. Although mussels' nutrition is primarily obtained through the productivity of endosymbionts [9,13], POM may also be a component of their diets due to their filtration ability [62,63]. In our samples, the isotopic signatures of foot tissues can be regarded as reflective of individual nutritional conditions over a long time period, while the values obtained from gill tissues more likely represent a mixture of individuals' and endosymbionts' values due to the high number of bacteria hosted there [13]. Our findings demonstrated that the average carbon value was -31.83 ‰ in foot tissues, which was in the range of organic carbon derived from seawater CO 2 fixed via sulfide oxidation (-42 ‰ to -30 ‰) in mussels and differed significantly from mussels harboring exclusively methanotrophic bacteria endosymbionts [32,33]. It is possible that the nutritional sources for B. manusensis are primarily from sulfur-based endosymbionts. Because mussels rely on endosymbionts, and the carbon isotopic fractionation between the contiguous levels was small [26], we can infer that the carbon signature of endosymbionts approximates the signature in gills containing numerous bacteria, whose average carbon value was -33 ‰.
The δ 13 C values of mussels in our study were in the range of the light carbon isotopic group, suggesting that the pathway of the symbiotic bacteria in mussels is primarily via the CBB cycle with RuBisCO form I as the CO 2 -fixing enzyme [25]. This further indicated that the δ 13 C value of dissolved inorganic carbon (DIC) could be calculated according to the average carbon value of gills, which was -33 ‰, and the fractionation by bacteria using RuBisCO form I which ranged from 22 ‰ to 33 ‰ in the local environment. The δ 13 C values of DIC may range from -11 ‰ to 0 ‰, varying between individual mussels. Additionally, some undigested food debris was observed in the gut contents during the dissection, which could indicate filterfeeding. However, the carbon isotopic values of the foot tissue were closer to those of gills rather than the POM values. It could be suggested that even if there was an input of filter-feeding, it was low.
We observed that the δ 13 C isotopic composition of the commensal worm Branchipolynoe spp. was very similar to that of its host mussels, but its δ 15 N values were higher, from 3.6 ‰ to 4.3 ‰ than those of foot tissues in the mussels. These were in the range of accepted δ 15 N enrichment per trophic level (3 to 4 ‰) in agreement with previous studies [25,39,64,65]. This suggests that the worms' food is sourced mainly from its host.

Crustaceans
Crustaceans were collected in considerable numbers at three sites in the cracks and rock surfaces. Vent crustaceans have been recorded as bacterial grazers, episymbiont hosts, scavengers or detritus feeders, or even predators [22,24,[66][67][68]. In addition, some authors have suggested that crustaceans subsist largely on a bacterial diet; unlike mussels, crustaceans primarily feed on episymbiotic bacteria and bacterial mats [16,39,[66][67][68]. For alvinocaridid shrimps, nutrition could be transferred from bacteria to the host through soluble bacterial products that occur by permeation across the gill chamber integument directly, or to a significant degree, via the digestive tract [69,70]. However, much higher fractions of photosynthetic carbon may occur in some vent fields [69]. For crabs in the genus Munidopsis, most were not reliant upon nutrition from bacterial endosymbionts, but on a mixed diet based on sediment bacteria, metazoans associated with bacterial mats, or other animals including polychaetes, limpets, protozoans and crab larvae [25,71,72].
Our study found δ 13 C values ranging from -23.73 ‰ to -9.11 ‰, with R. vandoverae showing values from -14.93 ‰ to -9.11 ‰ and M. lauensis showing lower values from -23.73 ‰ to -16.16 ‰, which were both in the range of values found in previous studies about related species [18,25,[73][74][75]. Compared to mussels, the values of crustaceans showed a much larger range, resulting from the omnivorous characteristics. According to the highest δ 15 N values for the crustaceans, compared to other animals at the same site, bacteria might be one of the food sources, but not the only one. Some sources with higher trophic levels than bacteria, such as POM and animals, should also be food sources. Among the three sites we studied, a mass of crustaceans was found at Site Dive #32 and Site Dive #36, whereas few mussels or vestimentiferans were found. As such, crustaceans from these two sites were not thought to feed on mussels or vestimentiferans. In addition, the δ 13 C values of R. vandoverae, which were only collected at Site Dive #32 and Site Dive #36, were isolated from other animals or potential food sources and were in the range of the heavy-value group, suggesting that bacteria with an rTCA pathway are the main nutritional source for the shrimps. In addition, the δ 13 C values of M. lauensis varied widely between individuals and was in the middle of the two groups, which may indicate that the carbon was fixed with Rubisco form II enzyme in the CBB cycle or in a mixture of the two cycles by the bacteria they consumed at these two sites.
At Site Dive #32, the two stable isotope values of the two crustaceans were significantly different, with the signatures of M. lauensis low and highly individual, whereas R. vandoverae showed relatively weak differences between individuals. The δ 15 N values of M. lauensis ranged from 4.55 ‰ to 10.02 ‰, which might suggest that they were more omnivorous. At Site Dive #36, the δ 13 C values of M. lauensis were still low, but the δ 15 N values were similar to those of R. vandoverae. They were at the same trophic level, but the food sources were different. The δ 13 C values of M. lauensis were closer to those of the gastropods, which might suggest that a proportion of P. nassariaeformis is the food source or that there is similar source between them. At Site Dive #33, the isotopic values of M. lauensis were similar to the same species from another two sites, which indicated similar food sources. In general, M. lauensis were more omnivorous. The high δ 15 N values indicated that they fed not only chemoautotrophic bacteria but also POM and animals.

Gastropods
We collected two gastropod species in limited quantities. Compared with the crustaceans collected from Site Dive #32, the carbon and nitrogen isotope values of I. nautilei were low, which indicated a lower trophic level. On the basis of previous research showing symbiotic relationships of bacteria in gastropods, I. nautilei derived its nutrition from intracellular gill symbionts that can oxidize both sulfide and thiosulfate to fuel autotrophy [40,[76][77][78]. Methanotrophic bacteria may also be present, but in low abundance. The lower δ 15 N values in our results support that the source is from chemoautotrophic bacterial endosymbionts. The average δ 13 C value was -28.83 ‰, which suggested that sulfide-oxidizing bacteria were the dominant contributor. We suggested that the bacteria were likely utilizing the CBB cycle for carbon sequestration.
The carbon isotopic values of P. nassariaeformis from Site Dive #36 were similar to those of M. lauensis, whereas the nitrogen isotopic values were significantly lower at approximately 5 ‰. The genus Provanna is a characteristic gastropod of chemosynthetic faunas. Some of them may harbor symbiotic bacteria, but these species mainly feed on filamentous bacteria or detrital organic material [24,79,80]. From our results, the lower δ 15 N values indicated that they were at a low trophic level. Unlike another gastropod I. nautilei, the δ 13 C values of P. nassariaeformis were higher at approximately 6 ‰, which may indicate that their food sources were different.

Vestimentiferans
Vestimentiferans were observed only at Site Dive #33, and there were only two individuals available for dissection. These are local species with food sources from endosymbiotic bacteria [23,81]. The chemoautotrophic bacterial endosymbionts were mainly sulfur-oxidizing symbionts [16,52]. Compared three parts of the vestimentiferans, there were no differences in δ 13 C values but significant differences in δ 15 N values. The signals in the trunk tissues were lowest. The δ 15 N values were lower than those of the crustaceans and higher than those of the mussels, which indicates a low trophic level. At Site Dive #33, the carbon fixation pathway of the symbiotic bacteria in vestimentiferans should be not the same as the mussels, resulting from a 10 ‰ higher δ 13 C value. The δ 13 C values of A. ivanovi were around -22 ‰, which were not in the heavier-value group and different from several previous results. This may resulted from the different types of symbiotic bacteria. Assuming that the vestimentiferans wholly rely on symbiotic bacteria, the different isotope signatures suggest that the symbionts primarily used RuBisCO form II enzyme in the CBB cycle to fix inorganic carbon, or that several symbionts are present, each using different fixation pathways.

Conclusions
Our study encompassed several dominant species collected from each of three sites. Although local conditions were different at each site, nutritional relationships and trophic structures were similar for the same species across locations. However, for the two species collected in more than one site (R. vandoverae and M. lauensis), there were significant differences in the isotopic signatures between sites. First, chemosynthetic bacteria were the major primary source of carbon in this ecosystem. Second, the dominant species, including shrimps, galatheids, mussels and gastropods, at these hydrothermal vents were not apparently selective in their food choices, which may be due to the limited range of food sources. As primary consumers, mussels mainly obtained energy from symbiotic bacteria that used RuBisCO form I for the CBB cycle. Mussels also obtained some energy by filter-feeding, but in considerably lower amounts. Symbiotic polychaetes consumed mussels and symbiotic bacteria as prey. Like mussels, gastropods relied mainly on symbiotic bacteria for energy with some supplementary consumption of detritus. Tube worms obtained energy from symbiotic bacteria, but their bacterial species were different from those of the mussels. Finally, crustaceans were found to be omnivorous on chemosynthetic bacteria, detritus, or even animals.
Since most of the trophic structure relied on chemosynthetic bacteria and the bacteria relied on reduced compounds from the vent emissions, any disruption of vent emission by mining activities could have major consequences on the trophic structure and finally on vent ecosystem functioning. Moreover, our study suggested that there might be food relationship between animals, i.e., mussels and commensal scale worms. In addition, other members of the vent community must be considered in order to obtain a more comprehensive picture of the vent trophic network. Therefore, to evaluate the potential impacts of deep sea mining, it is necessary to study not only the direct destruction of the organisms attached to the mineral rocks but also the cascade effects on the food web of the whole ecosystem. Furthermore, the dense communities attached to rocks formed several microenvironments and presented high biodiversity. Although the relationship of interdependence among organisms in the microenvironment is not clear, there was no doubt that mining had effects on these microenvironments. Before exploiting deep-sea resources, further systematic investigations concerning the protection of deep-sea ecosystems are deeded.
Supporting information S1