Hydrothermal vent fields discovered in the southern Gulf of California clarify role of habitat in augmenting regional diversity

Hydrothermal vent communities are distributed along mid-ocean spreading ridges as isolated patches. While distance is a key factor influencing connectivity among sites, habitat characteristics are also critical. The Pescadero Basin (PB) and Alarcón Rise (AR) vent fields, recently discovered in the southern Gulf of California, are bounded by previously known vent localities (e.g. Guaymas Basin and 21° N East Pacific Rise); yet, the newly discovered vents differ markedly in substrata and vent fluid attributes. Out of 116 macrofaunal species observed or collected, only three species are shared among all four vent fields, while 73 occur at only one locality. Foundation species at basalt-hosted sulfide chimneys on the AR differ from the functional equivalents inhabiting sediment-hosted carbonate chimneys in the PB, only 75 km away. The dominant species of symbiont-hosting tubeworms and clams, and peripheral suspension-feeding taxa, differ between the sites. Notably, the PB vents host a limited and specialized fauna in which 17 of 26 species are unknown at other regional vents and many are new species. Rare sightings and captured larvae of the ‘missing’ species revealed that dispersal limitation is not responsible for differences in community composition at the neighbouring vent localities. Instead, larval recruitment-limiting habitat suitability probably favours species differentially. As scenarios develop to design conservation strategies around mining of seafloor sulfide deposits, these results illustrate that models encompassing habitat characteristics are needed to predict metacommunity structure.

SKG, 0000-0002-9110-9591; RS-M, 0000-0002-9561-355X Hydrothermal vent communities are distributed along mid-ocean spreading ridges as isolated patches. While distance is a key factor influencing connectivity among sites, habitat characteristics are also critical. The Pescadero Basin (PB) and Alarcón Rise (AR) vent fields, recently discovered in the southern Gulf of California, are bounded by previously known vent localities (e.g. Guaymas Basin and 218 N East Pacific Rise); yet, the newly discovered vents differ markedly in substrata and vent fluid attributes. Out of 116 macrofaunal species observed or collected, only three species are shared among all four vent fields, while 73 occur at only one locality. Foundation species at basalt-hosted sulfide chimneys on the AR differ from the functional equivalents inhabiting sediment-hosted carbonate chimneys in the PB, only 75 km away. The dominant species of symbiont-hosting tubeworms and clams, and peripheral suspension-feeding taxa, differ between the sites. Notably, the PB vents host a limited and specialized fauna in which 17 of 26 species are unknown at other regional vents and many are new species. Rare sightings and captured larvae of the 'missing' species revealed that dispersal limitation is not responsible for differences in community composition at the neighbouring vent localities. Instead, larval recruitment-limiting habitat suitability probably favours species differentially. As scenarios develop to design conservation strategies around mining of seafloor sulfide deposits, these results illustrate that models encompassing habitat characteristics are needed to predict metacommunity structure.

Introduction
Exploration of the seafloor continues to uncover new habitats fostering chemosynthetic communities in diverse tectonic settings. Altogether, 285 hydrothermal vent fields are presently confirmed, with hundreds more inferred and predicted (see https://www.interridge.org; [1,2]). Reduced volcanic and biogenic compounds dissolved in vent effluents (primarily H 2 S, CH 4  physico-chemical gradients, a highly specialized and habitatrestricted fauna has evolved [3,4], with many taxonomic similarities across ocean basins and tectonic backdrops (e.g. [5,6]). Nevertheless, contemporary connectivity and historical evolutionary relationships among vent taxa, and with those at other chemosynthetic habitats, are complex and remain incompletely understood [7,8].
Seafloor hydrothermalism can host large massive sulfide deposits that form as mineral-laden hot water emerges at the seafloor [9,10]. As interest grows in mining the metals in these deposits, so also does the need to understand possible consequences for the associated ecosystems [11] and how alteration and/or removal of vent fields will affect persistence of a regional vent assemblage spread among small habitat islands [12,13]. Thus far, connectivity models assume that vent site proximity is the primary driver of recruitment success [14]. Deep-ocean circulation tends to be constrained to axial valleys that course along mid-ocean ridge (MOR) systems, thereby reducing the loss of animal larvae by off-axis or cross-axis currents, and promoting long-distance connectivity among vent fields [15,16]; larval behaviour and local source populations also enhance survivorship [17,18]. Thus, MOR vent species are often distributed over large distances creating high similarity (low beta diversity) among assemblages (e.g. 9850 0 N to 218 N East Pacific Rise (EPR), 1200 km; [19]). However, within assemblage, alpha diversity is also affected by factors such as dispersal barriers, age of the vent, ecological succession, substratum type and vent fluid chemistry [20,21].
The Gulf of California (GoC) formed as North America overrode the EPR, and now encompasses the northernmost segments of the EPR (figure 1). The Alarcón Rise (AR) is the northernmost segment of the EPR, before entering the GoC, with bare lava exposed. Further north in the GoC, deep extensional basins, typically filled with sediments, are separated by transform faults that accommodate shearing between the Pacific and North American plates. Here, high-temperature venting supports dense animal communities (e.g. Guaymas Basin; [22]). About 60 km north of the Guaymas vents, cold seeping fluids rich in hydrocarbons also host a chemosynthetic-based community [23]. Portail  these vent and seep assemblages to find extensive taxonomic overlap (85% of species shared), and suggest that vents and seeps support a continuum of communities and of food web complexity. By extension, we expect fauna at any vent localities in the GoC also to have high overlap with each other and with the Guaymas Basin. The opportunity to test this hypothesis arose with the discovery of neighbouring vent fields in the southern GoC, when autonomous underwater vehicle (AUV) surveys and remotely operated vehicle (ROV) dives located vents in the Pescadero Basin (PB) [26,27] and on the AR [28].
Our study examines the composition of faunal assemblages at these two localities in comparison with vents at 218 N EPR and at approximately 278 N in Guaymas Basin, and implications for understanding community connectivity.

(c) Larval detection
Seawater was sampled for biological particulate matter at the PB and AR hydrothermal vent fields, in and out of visible plumes, and over the PTF hydrocarbon seep. Roughly 60 l of water was collected using the ROV suction sampler operated at 20% power for 10 min from six different depths (up to approx. 50 m above chimneys). Most samples were acquired at a constant altitude from the seafloor (or top of chimney) as estimated by the ROV altimeter. Samples collected within the visible vent plume required the ROV to sometimes follow it as the current changed; however, these changes were relatively small and only varied by a few metres. Water was then filtered on 30-100 mm EMD Millipore TM nylon woven net filters, and prepared for Illumina high-throughput sequencing with a MiSeq sequencer (Illumina, Inc., San Diego, CA, USA) using the mitochondrial COI and ribosomal 28S primers indicated in the electronic supplementary material,

(d) Carbon and nitrogen isotope analysis
Specimens for measures of stable carbon and nitrogen isotopes were identified shipboard and frozen at 2808C. Tissues were extracted from large organisms, and in some cases, small individuals were pooled, to produce a minimum dry weight of 2 mg (dried at 608C  figure 2a-c), while Riftia pachyptila was rare. Oasisia displayed two discrete morphotypes, akin to Ridgeia tubeworms at northeast-Pacific vents [40,41], but mitochondrial cytochrome c oxidase (mtCOI) sequences were identical; within the Oasisia 'group I' clade of Hurtado et al. [42], and were greater than 99% similar to Oasisia from AR and EPR vents. A dorvilleid polychaete, Ophryotrocha cf. akessoni, was extremely abundant (figure 2c), in places covering the tubeworms entirely. Polynoid scaleworms included new species of Peinaleopolynoe (figure 2f ), Lepidonotopodium and Branchinotogluma, as well as Branchiplicatus cupreus. Also abundant were the amphinomid Archinome levinae (described from methane seeps and Guaymas Basin; [43]) and Nereis cf. sandersi. The only alvinellids were Paralvinella grasslei and Paralvinella n. sp. (G. Rouse 2017, unpublished data).
Anemones occurred among the tubeworms, including one abundant species with white and red morphotypes (figure 2e; genetic sequencing of two mitochondrial loci did not differentiate between these two colour morphotypes). Scattered carbonates and sediments with emerging fluids featured other anemones, some burrowing, zoanthids and small beds of the vesicomyid clam, Archivesica sp. 7, known from methane seeps along the Costa Rica and Peru margins ( [44,45]; figure 2d ). Here, a newly described xenacoelomorph, Xenoturbella profunda, was collected [46]. Other taxa are indicated in the electronic supplementary material, table S1, including six first time sightings in the GoC.
Notably absent from PB vents were numerous polychaete species observed elsewhere in the GoC and EPR localities, especially ampharetids, hesionids, serpulids and the alvinellid genus Alvinella (electronic supplementary material, table S1). The PB fauna is further unusual in the low gastropod diversity and scarcity of decapod crustaceans (electronic   While polychaetes again dominated the biomass on chimneys, there was greater taxonomic diversity at the AR vents. In vigorous venting, the alvinellids Alvinella pompejana (figure 2g,l ) and Alvinella caudata formed dense colonies. The large hesionid Hesiolyra bergi was common among Alvinella tubes, as were other hesionids, including Hesiospina vestimentifera and three new Hesiospina species. Large R. pachyptila tubeworms covered moderately venting surfaces on chimneys and adjacent diffuse vents emanating from basalts (table 1 and figure 2h-i). The Riftia mtCOI haplotypes were identical to those from Guaymas Basin through 328 S on the Pacific Antarctic Ridge [47]. In contrast with PB, Oasisia aff. alvinae was much less abundant (table 1). Alvinellids in the genus Paralvinella (Paralvinella palmiformis and P. grasslei) were present, along with the ampharetid Amphisamytha fauchaldi, known from the Guaymas Basin and methane seeps off Oregon and Costa Rica [48]. Nine species of scaleworm included three species each of Branchinotogluma and Lepidonotopodium. Many other species inhabited the Riftia tubes including very abundant limpets (Lepetodrilus elevatus, Lepetodrilus cristatus and Euleptopsis vitrea) and an undescribed species of Peltospira snail. Also common on AR chimneys were the predatory brachyuran crabs, Bythograea thermodron and Cyanograea praedator, and the zoarcid fish Thermarces cerberus.
On basalts around the chimneys, serpulid polychaetes were numerous, one species of which is known from vents further south (Laminatubus alvini; figure 2k), while two are new (Laminatubus and Protis n. spp.), but also found at Costa Rica methane seeps [49]. The galatheid squat lobsters, Munidopsis recta and Munidopsis lentigo, were frequently observed. Dense clusters of the giant clam, 'Calyptogena' magnifica (figure 2j), occupied venting cracks in basalts (unresolved genus assignment; [45]). Other taxa are indicated in the electronic supplementary material, table S1, including 23 range extensions of known species. Notably absent from AR vents were numerous other limpets (e.g. Neolepetopsis, Hirtopelta) and snails (e.g. Provanna) observed elsewhere in the GoC and EPR localities.

(c) The Pescadero transform fault seeps
The 2400 m deep hydrocarbon seep on the PTF lies at 23838.5 0 N/108823.6 0 W. While most of the uplifted sediment hills were surrounded by younger lava flows (e.g. [50]), only discrete lowtemperature seepage was observed along the transform in volcanic rubble and sediments. The seeps were dominated by scattered dense patches of siboglinid tubeworms, Escarpia spicata and Lamellibrachia barhami (electronic supplementary material, figure S2). Abundant polychaetes included the amphinomid Ar. levinae, the serpulid species, Laminatubus sp. (shared with the AR vents) and a putatively new species of Branchinotogluma polynoid. The vesicomyid clams Calyptogena costaricana and Archivesica. mt-V [44] were common. Several gastropod genera more typically associated with hydrothermal vents were observed, including Provanna ios, Paralepetopsis sp. and Neolepetopsis aff. gordensis. Munidopsis squat lobsters, which are typically abundant at eastern Pacific vents and seeps, were not observed. The d 13 C signatures of L. barhami and Es. spicata were slightly more depleted than those of siboglinids from the PB and AR vents (electronic supplementary material, table S6), while the single Calyptogena specimen was notably more depleted in both isotopes than Cal. magnifica at AR (figure 4).

(d) Taxonomic overlap among Gulf of California and East Pacific Rise vent fields
The closest known hot vents in the GoC are at Guaymas Basin 425 km to the north of PB and 218 N EPR 285 km to the south of AR ( figure 1 inset). In all, 116 macrofaunal taxa (greater than 1 cm size) were recognized as distinct species from the four main vent fields, many of which are undescribed (electronic supplementary material,

(e) Faunal abundance comparisons between vent fields
Video transects of active chimneys revealed markedly different patterns of macrofaunal abundance between PB and AR (table 1; electronic supplementary material, tables S4 and S5). NMDS comparison of animal densities showed complete separation of the two localities, with observed community membership from all 10 PB transects clustered closely together, to the exclusion of the 14, more dispersed, AR transects ( figure 3). A Shepard plot of NMDS scores revealed very low stress (less than 0.12) for a two-dimensional plot (data not shown). At PB vents, Oasisia aff. alvinae tubeworms enveloped the large chimneys and mounds in densities up to 2400 individuals m 22 (avg. ¼ 408; figure 2a,b). At AR vents, Oasisia density was much lower and large groups were infrequent. By contrast, giant R. pachyptila tubeworms dominated AR vents to a maximum of 730 individuals m 22 (figure 2h,i), but were relatively scarce at PB vents (table 1). On the seafloor at both AR and PB, vesicomyid clam abundances are fairly similar, although species differ (table 1; electronic supplementary material, table S1). The Alvinella species at AR are notable in their contribution to biomass in intense fluid flows on chimneys.
Among the grazers and small predators on chimneys at PB, the relative contribution of polynoids such as  (table 1), while nereids, crabs and fishes were mostly absent at PB. Cnidarian suspension feeders on the vent periphery were present at both localities, although species differed. At AR, but not PB, dense fields of serpulids (to 380 m 22 ; figure 2k)

Discussion
Notably different faunal communities colonize the two newly discovered neighbouring localities; only seven of 61 species are shared and 10 additional species are new to science, despite proximity to known vents. Computational models suggest that larval lifespans, distances between localities and ocean currents strongly influence the composition of vent communities in the western Pacific [51]. Geostrophic estimates show distinct two-way flow near bottom between PB and AR with a Pacific-ward flow to the east and inflows to the west [52]; thus, two-way larval exchange is possible, and the 75 km distance between PB and AR vent fields lies within the average dispersal distances for most worm, clam and crustacean larvae [18,53,54]. While the recent Lagrangian connectivity model of Montañ o-Cortés et al. [55] does not encompass the AR field, it illustrates particles crossing basin boundaries with some vertical component in four-and eight-week runs. Thus, the distinction of macrofauna between PB and AR locales suggests that community composition is related less to geographical proximity and larval supply than to habitat suitability. Biological interactions (e.g. priority effects, competitive interactions) and stochasticity in colonization can also strongly influence community structure [56,57]; however, the large differences between the PB and AR vent communities probably exceed the scale expected for these factors. Indeed, the size and extent of the deposits at both localities indicate a long-term stability in hydrothermalism sustaining community development; smaller Guaymas chimneys are around 4000 years old [58]. The approximately 1300 m depth differences may play a role in differentiating the fauna [59][60][61] because of hydrostatic pressure effects on animals [62], but the dominant PB vent species were not novel or unknown from other regional vents. Depth transcendence does occur; e.g. in the Mariana Backarc Basin, the same vent species occur from 1500 to 3600 m over a distance of 600 km [63]. In the absence of obvious biogeographic barriers, we hypothesize that other physical and chemical parameters control the structure of PB faunal communities. PB and AR vent fields differ in underlying substratum composition, with carbonate structures and rubble embedded in sediment versus sulfide chimneys and mounds on basalt, respectively. The work of Portail et al. [24], also in the GoC, determined that macrofaunal community composition was significantly influenced by substratum type, rather than fluid temperature or pH. Deep sediments at the PB vents probably limit the presence of 'Cal.' magnifica, serpulids and some anemones, all of which inhabit basalts or other hard substrata. Similarly, the paucity of gastropods, and possibly polynoids, may relate to limited substratum availability.
In their comparative study of six seep and four vent fields within a 60 km range in Guaymas Basin, Portail et al. [24] found high faunal similarity among localities; all vent families occurred at the seeps. However, at least seven of 16 animal families from PB are absent at Guaymas seeps or vents, revealing low similarity between these sedimented sites. Considering the additional dissimilarity between the PB and AR vents to the south (15 families here are not shared between locales), we suggest their hypothesis of 'continuity among deep-sea seep and vent ecosystems' ( [24], p. 5455) may not apply universally. The present study suggests that more substantial diversification and differentiation can result from variation in physical and chemical factors.
The PB fluids have notably elevated levels of H 2 , CH 4 and large hydrocarbons (M. Lilley 2016, personal communication), and pH. Fluids here, like the Guaymas Basin vents, emerge through thick sediments where hydrothermal alteration of sedimentary organic matter produces methane and hydrocarbons [64], resulting in conditions that differ greatly from the original end-member fluids [65]. Concentrations of reduced compounds, particularly methane, are hypothesized to influence variability in macrofaunal composition [24,66]. How these reduced gases directly affect the macrofauna is not known; however, biochemical alteration of vent fluids, including the accumulation of both hydrogen and methane, has implications for the structure and functioning of both the free-living microbial communities, as well as bacterial symbionts hosted by foundation fauna.
Stable isotope results suggest that organic carbon sources may differ between AR and PB vent fields (figure 4). In general, the d 13 C values of tested animals were more depleted at PB (from 237.9 to 213.1‰), compared with AR (234.2 to 29.4‰). The most depleted value was recorded from C. costaricana from PTF (239.5‰; electronic supplementary material, table S6). This implies that some of the dissolved inorganic carbon has a methane origin, perhaps a CO 2 intermediate. The highly depleted d 13 C values for some of the larger anemones also reveal this influence.
The deep carbonate-hosted vents at PB form a novel environment with a high alpha diversity in the spatially restricted area of the southern GoC; a majority of species are not recorded from elsewhere in the GoC. Moreover, the high habitat variability among vent localities from north EPR through the GoC promotes high beta diversity among communities over a relatively short distance. Larval DNA detection of taxa not present as adults suggests that dispersal among the southern GoC localities is possible. However, the PB site is biologically distinct from other regional vents; foundation species of tubeworms and clams differ, and major space colonists (i.e. serpulids, patellogastropods) are absent. Therefore, structuring factors for faunal composition and food-webs must include geological setting, vent fluid characteristics and substratum heterogeneity, all of which affect habitat suitability and limit recruitment of select species.
High variability of fauna and habitat conditions among spatially discreet communities creates challenges for managing rspb.royalsocietypublishing.org Proc. R. Soc. B 284: 20170817 conservation approaches in the ocean. Deep-sea mining of polymetallic sulfides will target large deposits that may support vent communities [12,67]. Managing consequences of such extraction includes the need to better understand connectivity patterns and conservation of regional deep-sea vent faunal networks (e.g. [51,68]). However, the framework to support management and conservation must include assessment of the habitat characteristics that determine current community composition, especially in the context of 'recovery' at a mine site that has undergone drastic habitat alteration and loss of species. The vent fields of the southern GoC provide an excellent opportunity to explore how habitat diversity influences biodiversity distribution.
Data accessibility. This article has no additional data. Authors' contributions. R.V. and D.Cl. wrote proposals that funded the expedition, assembled the expedition teams and were chief scientists during the expeditions. S.G., S.J., V.T., L.L., G.R. and D.L.S. were cruise participants in 2015, focused on biology. D.Ca., J.B.P., R.Z. and R.S.-M. were cruise participants in 2012 and 2015, focused on geology, and who collaborated to originally locate the sites. E.E. and L.A.S. were international collaborators specializing in ecology, trophic levels and isotope analysis. L.A.S. and D.L.S. conducted the stable isotope analyses. S.G., S.J., L.L. and G.R. carried out the molecular laboratory work and sequence alignments. J.P. constructed all of the maps. All authors participated in data analysis and helped draft the manuscript (especially S.G. and V.T.) and gave final approval for publication.