The R.V. Kexue conducted a photographic survey of the Formosa Ridge during a research cruise in 2016 . Faxian, an ROV equipped with an L1000 imaging and laser profiling system which was designed and supplied by Cathx Ocean Ltd., was used to obtain all images. This system was equipped with a still camera, two LED lights with outputs of 7,000 and 28,000 lumens, and two green line lasers. All original images associated with geographic coordinates were captured at a rate of nine frames per second. The mosaic was pieced together using these coordinates and the same object identifier in adjacent images ( Figure 2 ) (Wang et al., 2019). PTGui software was used for the initial splice, and Adobe Photoshop was used for the subsequent image optimization. Then, for statistical convenience, the mosaic was divided uniformly by 200 x 200 tiles (19,516 small pictures with an area of about 1 m², except the blank). All visible macrofauna in the mosaic were counted and identified to the lowest taxonomic level possible. To compare the spatial distribution and density of macrofauna, the whole cold seep area was divided into a 10 m by 10 m grid. Geographic information and the density of macrofauna in each grid were used to perform cluster analysis to verify the similar occurrence of macrofauna. Meanwhile, a non-metric multidimensional scaling (NMDS) approach was used to compare the similarity between different groups. All these analyses were performed using Primer v.6. To calculate the mean body length in each grid, 20 to 50 individuals of the dominant species (G. platifrons and S. crosnieri) were chosen at random.

To compare the spatial difference and interannual variation of macrofauna (compared to the 2016 data, mainly from the south and north of Site F), TV grab operations (including GA, GB, and GC monitored twice at each station) were taken by the R.V. Kexue at Site F in 2020. All macrofauna collected by TV grab were counted. The Shannon–Weiner index was calculated to compare the species diversity at different stations. Meanwhile, environmental factors at each station were also identified. Benthic environmental variables (methane and dissolved oxygen) were measured by the Seabird SBE 43 dissolved oxygen sensor (DO) (Sea-Bird Electronics, Inc., Washington, USA) and the CONTROS HydroC^® CH₄ sensor (Kongsberg Gruppen, Norway) installed in the ROV tool sled (Cao et al., 2021). The manipulator arm performed every in situ detection as it approached the target sites. Pearson correlation analysis was used to identify correlations between environmental factors (methane and DO) and the dominant species.

The body length and weight (wet and dry) of G. platifrons and S. crosnieri were measured in 50 specimens, respectively, to model the relationship between length and weight. Dry weights were measured after at least 48 hours in a freeze dryer. For G. platifrons, the dry weight was calculated only after the shell was removed. For S. crosnieri, the shell length of the head and the whole-body length were all recorded. Because of the inconvenient measurement, the whole-body length appears worse in modeling. So, the shell lengths of the head were used as the body length of S. crosnieri. These data were used to model the following standard equation:

where W is the weight (g) and L is the body length (or the shell, for G. platifrons), measured with a vernier caliper, in mm.

To assess the metabolism in the G. flatirons community, the relationship between oxygen consumption and body weight of mytilids was determined according to this equation (Khripounoff et al., 2017):

where R is the oxygen consumption, and W is the dry body weight of G. platifrons. W is calculated by the equation given in Section 2.3. L is the mean body length of G. platifrons measured from the mosaic. The methane consumption of G. platifrons was calculated by the relationship in metabolic gas flux (per mol CH₄ consumed, the G. platifrons consumed ~1.2 mol O₂) (Kochevar et al., 1992). Meanwhile, the mean methane consumption rate (1.98 μmol/h) of S. crosnieri as estimated by Watsuji et al. (2014) at 12.0 MPa was used.

Methane concentrations at Site F varied quickly and were unstable, ranging from 237.25 ppm to 15,000 ppm with a mean of 4120.88 ppm ( Supplementary Table 1 ). The variation in concentration of dissolved oxygen was relatively smaller, ranging from 2.89 mg/L to 3.16 mg/L with a mean of 3.07 mg/L. The substrates at Site F contained hard (central authigenic carbonates) and soft (peripheral muddy) sea bottoms. Large mussel beds, lobster clusters, and shell debris were found on hard substrates formed by authigenic carbonates ( Figure 2 ). In addition to authigenic carbonates, there was a large, soft argillaceous environment, including reduced sediments and muddy sea bottoms.

In this survey, 41 species or categories, including Mollusca (12), Porifera (3), Annelida (9), Echinodermata (4), Arthropoda (9), Cnidaria (3), and Vertebrata (1) were found in the mosaic and TV grabs. There were 34 taxa identified as species or genera, and others were the higher taxa ( Supplementary Table 2 ). There were 17 categories collected or identified through this method, including Cnidaria (Anthomastus sp., Caryophyllia sp., and Alvinactis sp.), Mollusca (Conchocele bisecta, Solemya sp., Enigmaticolus inflatus sp. nov., cf. Abyssochrysos sp., and Chaetoderma sp.), Arthropoda (Lysianassoidea sp., and Pleurocryptella fimbriata), Annelida (Lepidonotopodium sp., Branchinotogluma sp., Lamellibrachia sp., Glycera sp., Nereis sp., and Capitellidae sp.), and Enteropneusta (cf. Balanoglossus) . Among them, there were 10 newly recorded species.

The quantity of macrofauna in the whole mosaic (20,000 m²) reached as high as 252,943 individuals. G. platifrons and S. crosnieri were the dominant species at Site F, with 215,339 and 37,422 individuals, respectively. Clustering analysis revealed a 77.5% similarity between the four groups (r = 0.995, P < 0.01) and was used to verify the similarity of macrofauna occurrence in each grid ( Figure 3A, B ). The densities (individuals/100 m²) of Groups 1 through 4 were as follows: Group 1, 3,584–17,806; Group 2, 1,075–3,108; Group 3, 106–815; and Group 4, 0–133. According to the spatial distribution of each grid at the cold seep in Site F, we discovered a pattern of patchy distribution of macrofauna in which the most abundant areas (Group 1) were concentrated in the center of the entire site , followed by Groups 2, 3, and 4 ( Figure 3C ).

The density of macrofauna at Site F was greater than that of the surrounding seabed, especially for dominant species. G. platifrons and S. crosnieri reached their highest densities of 560 and 396 individuals/m², respectively. Their density distributions in every grid are shown in Figure 4 . G. platifrons mostly covered a large area and had the highest population density in the north, while S. crosnieri was not widely distributed and had the highest population density in the south ( Figure 4 ). These locations, which had a dense population of the dominant species, possessed visibly active seepages (such as the release of strings of bubbles from the seabed). In contrast, areas without any sign of seepage, such as shell debris and exposed carbonate rocks, had a relatively lower density of the dominant species. In comparison with G. platifrons and S. crosnieri, the abundance of Munidopsis sp. at these locations was not high, with 861 individuals and a maximum density of 45 individuals/m². Other macrofauna densities in the mosaic were considerably lower than those of G. platifrons and S. crosnieri. Lithodes longispina, Ophidiidae sp., Benthesicymus sp., and Henricia sp. Populations reached 103, 55, 23, and 19 individuals/m², respectively. In the meantime, the dominant species collected by TV grab in 2020 were counted to calculate the species density ( Supplementary Table 1 ). TV grab collections showed that the highest densities of G. platifrons, Histampica sp., Bathyacmaea lactea, Phymorhynchus buccinoides, and Provanna glabra reached 629, 28, 56, 17, and 78 individuals/m², respectively. The grabs in three stations showed that G. platifrons in GA (356 individuals/m²) and GB (336 and 330 individuals/m²) were similar in density, while S. crosnieri in GB (29 and 129 individuals/m²) was higher than GA (25 individuals/m²). Compared with GA and GB, the densities of G. platifrons (289 and 629 individuals/m²) and S. crosnieri (225 and 271 individuals/m²) in GC were relatively high. The Shannon–Weiner index of GC (1.15) was the highest, followed by GB (0.66) and GA (0.49).

To assess the biomass of the dominant species, we calculated its wet and dry weight. For G. platifrons, flesh wet weight accounted for 20.78 ± 30.37% of total weight. Flesh dry weight accounted for 3.98 ± 5.08% of total weight and 19.07 ± 11.46% of flesh wet weight. Dry weight accounted for 32.82 ± 13.80% of wet weight in S. crosnieri. None of the data shows an obvious trend of change with the variation in body length. Meanwhile, the length-to- weight ratios of G. platifrons and S. crosnieri fit well ( Table 1 ). The mean body lengths of these two species calculated from the mosaic were 74.44 mm and 39.08 mm, respectively. Therefore, their respective mean dry weights in the mosaic were 2.71 g and 3.81 g. Based on the total number of the two species, the dry biomass of Site F reached 726.15 kg. Otherwise, the ratio of dry weight to wet weight was used to calculate the biomass (dry weight) of the three stations in 2020. The highest and lowest total mean biomass per station were 1.62 kg dry weight/m² (dw/m²) and 16.42 kg wet weight/m² (ww/m²) at GC and 0.43 kg dw/m² and 6.83 kg ww/m²at GB.

A patch distribution pattern of macrofauna was found at the Site F cold seep ( Figures 4 and 5 ). Almost all macrofauna, such as G. platifrons, S. crosnieri, Lithodes longispina, and most of the Ophidiidae sp., were found living on or around the hard substrates. Meanwhile, some common species (B. lactea, P. buccinoides, Alvinocaris longirostris, and several species of Polynodiae) at Site F were also found on hard substrates through the TV grab and the ROV camera. According to grab operations in different substrates, the macrofauna in soft substrates, even in reduced sediments (only four types of macrofauna were found), had relatively low diversity and density when compared to hard substrates. Munidopsis sp. was found on carbonate rocks and the exterior (mainly a mix of mussels and shells) of the G. platifrons –S. crosnieri community. A phenomenon we observed was that Munidopsis sp. and S. crosnieri did not exist in large numbers (>30 individuals/m²) at the same location. An abundance of S. crosnieri was found at Site F, but no significant numbers of Munidopsis sp. In addition, two species of sponges, Esperiopsis sp. and Stlocordyla sp., were abundant (the highest density reached 578 individuals/m²) and widely distributed in carbonate rocks.

The body length of the dominant species was measured to compare the age composition in different areas. The mean body lengths of G. platifrons and S. crosnieri we found in the mosaic in 2016 reached 35.89 to 94.15 mm and 24.26 to 55.35 mm, respectively. The central area was almost covered by the larger G. platifrons, while the larger S. crosnieri were abundant only in the southeast ( Figure 6 ). Obvious signs of seepage were found in these two locations. To compare the interannual variation in body length, samples were taken from three stations ( Supplementary Table 1 ). The average body length and weight in GA were higher as compared to those in the other locations. Individuals with a smaller body length (< 20 mm) and no signs of seepage were absent. Using the normal distribution of body length in 2016 and 2020, we analyzed the body length at these three stations ( Figure 7 ). The average body length of G. platifrons and S. crosnieri in GA was significantly increased. Their length ranges were noticeably narrower. In contrast, the mean body length of G. platifrons in GB and GC decreased dramatically, with the former dropping from 70–80 mm to 30–40 mm and the latter dropping from 80–90 mm to 30–40 mm. The proportion of medium and small body lengths in both areas increased significantly.

The concentration of methane showed a strong positive correlation with the density of G. platifrons (r = 0.895, P < 0.05) and S. crosnieri (r = 0.833, P = 0.08), while DO showed a strong negative correlation with G. platifrons (r = –0.908, P < 0.05) and S. crosnieri (r = –0.823, P = 0.08).

The oxygen consumption calculated from the 2.71 g dry weight of G. platifrons was 17.54 μmol/(g dry weight by h). G. platifrons consumed 14.62 μmol/(g dry weight by h) of CH₄. Therefore, with a dry weight of 2.71 g and a quantity of 215,339 individuals, the G. platifrons in the whole of Site F (2,000 m²) consumed 204.74 mol of methane per day. Methane consumption of S. crosnieri at the entire Site F was 1.78 mol/d by 37,422 individuals.

For the Site F cold seep, 41 species of macrofauna were found in our study. Compared with 42 chemosynthesis-based ecosystems in the northwest Pacific (range 2–38), including cold seeps in Sagami Bay, Nankai Trough, and so on, the species richness of Site F was high (Nakajima et al., 2014; Zhao et al., 2020). A previous study proved the strong influences among seepage activity and density, species evenness, and taxonomic composition of macrofauna (Levin et al., 2017). The relatively young carbonate rocks at Site F (Feng et al., 2018) suggested that past seep activity may have contributed to the relatively high species diversity. In addition to the previously reported macrofauna from the Formosa cold seep (Dong and Li, 2015; Li, 2017; Zhang et al., 2020; Zhao et al., 2020), we also found some new or newly recorded species. This indicates that species richness in this area remains under-sampled. It is worth noting that almost half of the species were found through TV grabs. Most of these were new or newly recorded species. Thus, further research on traditional physical collection could help in the discovery of possible new species.

Density-wise, Bivalvia, mainly G. platifrons, and Crustacea, particularly S. crosnieri, dominated community composition at Site F. The highest densities of these two species in the mosaic reached 560 and 396 individuals/m², respectively. Our research found a higher density of G. platifrons (629 individuals/m²) with our TV grab in 2020. This was more than double what Zhao et al. (2020) had previously reported. However, there were still a substantial number of individuals that had not been collected. Therefore, the density of G. platifrons and S. crosnieri at Site F has been greatly underestimated. The total mean biomass at Site F was low when compared with other cold seeps, including the Monterey Bay (0.7–2.05 kg ash-free dry mass/m²), the Gulf of Mexico (0.47–3.13 kg ash- free dry mass/m²), and the Peru Trench (30 kg ww/m²) (Olu et al., 1996; Bergquist et al., 2003; Barry et al., 2007). One cold seep dominated by vesicomyid at the Regab pockmark (0.77–2.03 kg dw/m²) (Decker et al., 2012) had the closest amount of biomass. However, because of the undersampling of the dominant species, the biomass at Site F was underestimated.

Site F had both hard (central authigenic carbonates) and soft (peripheral muddy) substrates. This pattern may be a result of the succession of cold seeps. When seepage begins in an area, microbial mats form and methanogenesis occurs, creating carbonates (Levin, 2005). Then, the fauna began to gather. As time goes on, the outermost mussel beds decompose into shell debris.

Almost all the macrofauna were found on or around the hard substrates (mainly the authigenic carbonates), and their relatively high density was always found at the locations where there were obvious signs of seepage. Authigenic carbonates appeared to be a major factor contributing to the higher macrofaunal diversity and abundance (Sen et al., 2019). This is mainly because these structures (e.g., cracks, outcrops) formed by authigenic carbonates can enhance habitat heterogeneity and might be used as the migration pathway for seepage. For example, small clusters of S. crosnieri can be found living in the cracks formed by authigenic carbonates. These cracks might be used as the dominant migration pathway for cold seep fluid transportation and provide sufficient energy and shelter for the overlying fauna. Some macrofauna, such as some top predators (L. longispina, Ophidiidae sp., Henricia sp.) and some small species (like Histampica sp. and some Gastropoda) were also observed in the vicinity of carbonates ( Figure 5 ) because the high biomass communities formed by G. platifrons and S. crosnieri could provide more food sources. Some studies also found that the prevalence of provannid snails, pyropeltid limpets, and polynoid polychaetes occurred on carbonates in common with another hard substrate vent, seep, and whale fall communities in the Pacific (Bergquist et al., 2007; Levin and Le Bris, 2015). L. longispina has been observed feeding on G. platifrons and possibly S. crosnieri (Li, 2017). Bathyacmaea lactea, a type of limpet, may consume biofilm on mussel shells (Chen et al., 2019). As for soft substrates, even in the reduced sediments, there was a lack of macrofauna. Only three species (Solemya sp., Glycera sp., and Chaetoderma sp.) were discovered in locations where oxidizing and reducing sediments were mixed. Sulfide in reduced sediments, a vital energy source, is important for the regulation of sulfide-dependent symbiont-bearing species (Cao et al., 2021). Despite the fact that it has been proven that animals can adapt to the toxic sulfide (Grieshaber and Volkel, 1998), relatively high concentrations may prevent biological adaptations. Otherwise, habitat suitability, geochemical differences, fluid flow, and seepage rates also had an effect on species distribution (Roy et al., 2007; Rybakova et al., 2013; Sen et al., 2018).

Overall, a distinct patchy distribution pattern of macrofauna was found in the mosaic, where the most abundant areas were clustered in the center of the entire cold seep area. Strong spatial differences in faunal composition and large aggregations characterized the megafauna pattern at the seep site (Astrom et al., 2018). Other cold seep ecosystems with patchy faunal distributions include the pockmark Regab, a cold seep near Baltimore Canyon, the northern Peruvian margin, and so on (Olu et al., 1996; Menot et al., 2010; Bourque et al., 2017). To clarify the spatial differences at Site F, we analyzed the macrofauna composition by comparing five stations in the south and north. Predictable successional processes were occurring within the G. platifrons –S. crosnieri communities. The density, biomass, and diversity of the associated macrofauna all decreased in the north. Narrowed body length ranges, a lack of juveniles ( Figure 7 ), and the disappearance of the cold seep vent may indicate that the northern region was in the final stages of the cold seep eruption and on the way to extinction. In the south, however, high densities and biomass of macrofauna were discovered. An obvious cold seep vent was found in this area in 2020, which may have contributed to a rise in local primary productivity. Some local macrofauna, such as B. lactea, P. glabra, and Histampica sp., were abundant in the south, especially in the southeast. Therefore, in 2020, the macrofauna in the north at Site F was declining, whereas the southern macrofauna was thriving .

The dominant species in the cold seep area had strong connections with methane and DO concentrations. In in situ environment detection and macrofauna collection were critical for better understanding these connections. However, the variation in these environmental factors, especially methane, was relatively different in the cold seep area (Cao et al., 2021). Indeed, a number of studies have shown that the concentration of methane can vary greatly in a short period of time, even by an order of magnitude (Davide et al., 2013; Bayrakci et al., 2014). This undoubtedly increased the difficulty of detection. Data from several stations were used to understand the interaction between environmental factors and dominant species. G. platifrons and S. crosnieri in central areas that had relatively high concentrations of methane were found to have comparatively higher densities ( Supplementary Table 1 ). G. platifrons and S. crosnieri obtained energy through symbiosis with methane-oxidizing bacteria and were limited to areas where chemical flux was sufficient for endosymbiont nutrition (Sibuet and Olu, 1998). This may also suggest that high methane concentrations have some positive effects on these two species. As for DO, we attributed the negative correlation to oxygen consumption by life processes. Other environmental factors, including temperature, pH, salinity, and depth, were also applied in correlation analysis and showed insignificant correlations. In addition, to better understand the biological-environmental relationships, more sampling sites and environmental factors (such as fluid flow, seepage rate, and so on) needed to be included in future investigations.

The consumption of methane (204.74 and 1.78 mol/d) at the entire Site F by the dominant species G. platifrons and S. crosnieri was estimated. The equation between oxygen consumption and body weight of mytilids was used to calculate the oxygen consumption of G. platifrons. The conditions for this equation were that the dry weight of bivalves should be 0.03- 4 g (Khripounoff et al., 2017). The dry weight of G. platifrons at Site F was predominantly in the range of 0.03–2.36 g. Meanwhile, the similar temperature of the habitat could also reduce the estimation error because temperature is the most important factor affecting oxygen consumption. Since it was assumed that mytilids were solely responsible for oxygen consumption , the oxygen consumption of G. platifrons may be overestimated. As for the relationship between methane and oxygen, the presence of related species (Gigantidas sp.) at Site F and the Gulf of Mexico added credibility to the estimation of methane consumption. For assessing the influence of dominant species on methane flux, the methane flux (60-1,598 mol/d) at the whole of Site F (20,000 m²) was estimated according to Mau et al. (2020). Methane consumption by the dominant species accounted for 12.93% of the maximum methane flux at Site F. Although the methane flux was underestimated (ignoring any methane that remained in bubbles), the consumption of methane by macrofauna on the seafloor cannot be ignored.