Deep-sea biodiversity at the extremes of the Salas y Gómez and Nazca ridges with implications for conservation

RESEARCH ARTICLE Deep-sea biodiversity at the extremes of the Salas y Go ´mez and Nazca ridges with implications for conservation Alan M. FriedlanderID1,2*, Whitney Goodell1,3, Jonatha GiddensID3, Erin E. Easton4,5, Daniel Wagner6 1 Pristine Seas, National Geographic Society, Washington, DC, United States of America, 2 Hawaiʿi Institute of Marine Biology, University of Hawaiʿi, Kāne‘ohe, Hawaiʿi, United States of America, 3 Exploration Technology Lab, National Geographic Society, Washington, DC, United States of America, 4 Ecology and Sustainable Management of Oceanic Islands, Universidad Cato´ lica del Norte, Coquimbo, Chile, 5 School of Earth, Environmental, and Marine Sciences, University of Texas Rio Grande Valley, Brownsville, Texas, United States of America, 6 Conservation International, Center for Oceans, Arlington, VA, United States of America * alan.friedlander@hawaii.edu Abstract The Salas y Go´ mez and Nazca ridges are underwater mountain chains that stretch across 2,900 km in the southeastern Pacific and are recognized for their high biodiversity value and unique ecological characteristics. Explorations of deep-water ecosystems have been limited in this region, and elsewhere globally. To characterize community composition of mesopho- tic and deep-sea demersal fauna at seamounts in the region, we conducted expeditions to Rapa Nui (RN) and Salas y Go´ mez (SyG) islands in 2011 and Desventuradas Islands in 2013. Remote autonomous baited-cameras were used to conduct stationary video surveys

between 150–1,850 m at RN/SyG (N = 20) and 75–2,363 m at Desventuradas (N = 27). Indi- vidual organisms were identified to the lowest possible taxonomic level and relative abun- dance was quantified with the maximum number of individuals per frame. Deployments were attributed with associated environmental variables (temperature, salinity, dissolved oxygen, nitrate, silicate, phosphate, chlorophyll-a, seamount age, and bathymetric position index [BPI]). We identified 55 unique invertebrate taxa and 66 unique fish taxa. Faunal com- munity structure was highly dissimilar between and within subregions both for invertebrate (p < 0.001) and fish taxa (p = 0.022). For fishes, dogfish sharks (Squalidae) accounted for the greatest dissimilarity between subregions (18.27%), with mean abundances of 2.26 ± 2.49 at Desventuradas, an order of magnitude greater than at RN/SyG (0.21 ± 0.54). Depth, seamount age, broad-scale BPI, and nitrate explained most of the variation in both inverte- brate (R2 = 0.475) and fish (R2 = 0.419) assemblages. Slightly more than half the deploy- ments at Desventuradas (N = 14) recorded vulnerable marine ecosystem taxa such as corals and sponges. Our study supports mounting evidence that the Salas y Go´ mez and Nazca ridges are areas of high biodiversity and high conservation value. While Chile and Peru have recently established or proposed marine protected areas in this region, the major- ity of these ridges lie outside of national jurisdictions and are under threat from overfishing,

plastic pollution, climate change, and potential deep-sea mining. Given its intrinsic value, PLOS ONE PLOS ONE | https://doi.org/10.1371/journal.pone.0253213 June 30, 2021 1 / 27 a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 OPEN ACCESS Citation: Friedlander AM, Goodell W, Giddens J, Easton EE, Wagner D (2021) Deep-sea biodiversity at the extremes of the Salas y Go´mez and Nazca ridges with implications for conservation. PLoS ONE 16(6): e0253213. https://doi.org/10.1371/ journal.pone.0253213 Editor: Jeffrey A. Eble, Florida Institute of Technology, UNITED STATES Received: February 1, 2021 Accepted: May 29, 2021 Published: June 30, 2021 Copyright: © 2021 Friedlander et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. Data Availability Statement: Our data are publicly available at Data Dryad: doi.org/10.5061/dryad. c866t1g6j. Funding: The expeditions were funded by the National Geographic Society and Pristine Seas donors with support to AMF. Conservation International, the Paul M. Angell Family Foundation, Tom and Currie Barron, and Alan Eustace provided additional support to DW, WG and JG. The Chilean Millennium Science Initiative Program grant #NC120030 provided support to EEE. The funders

this region should be comprehensively protected using the best available conservation mea- sures to ensure that the Salas y Go´ mez and Nazca ridges remain a globally unique biodiver- sity hotspot. Introduction As the largest habitat on Earth, the deep sea is remarkably underexplored and under-charac- terized [1, 2]. Consequently, many fundamental questions in deep-sea ecology remain unre- solved [3, 4]. Although knowledge of deep-sea community composition and ecosystem functioning has advanced rapidly in recent decades, both in terms of the scale of sampling [5, 6] and large-scale data syntheses [5, 7–10], fundamental ecological data for much of the deep sea is lacking [11]. For example, a recent three-year, deep-sea exploration campaign in the Pacific found that fewer than 20% of the taxa could be identified, highlighting our knowledge limits of the deep ocean [12]. Lack of fundamental ecological knowledge, including what lives in the deep sea and how biodiversity and community composition are distributed, hinders efforts to map current spe- cies distributions and predict future distributions in a changing climate [13, 14], information which is highly relevant to planning management and conservation efforts. For example, habi- tat heterogeneity is thought to drive global patterns of species diversity, but relationships between environmental drivers and biological responses are still uncertain [4]. Further, main- taining well-connected populations, communities, and ecosystems is critical to design effective

the movement of benthic and pelagic species and providing important feeding, resting, and spawning grounds [31–33]. The high variability of community composition within and among seamounts is related to environmental variation in these heterogeneous habitats [34, 35], pro- viding evidence that each seamount is unique. Seamounts are vulnerable to anthropogenic dis- turbance (e.g., trawling) [31, 36] and often do not recover from intensive fishing pressure [37– 39]; therefore, they are a high priority for conservation [27]. Due to the dense aggregations of seamounts that comprise the Salas y Go´mez and Nazca ridges in the southeastern Pacific, this region represents an important reservoir for global PLOS ONE Deep-sea biodiversity along the Salas y Go´ mez and Nazca ridges PLOS ONE | https://doi.org/10.1371/journal.pone.0253213 June 30, 2021 2 / 27 provided support in the form of salaries for authors [AMF, WG, JG, EEE, DW], but did not have any additional role in the study design, data collection and analysis, decision to publish, or preparation of the manuscript. The specific roles of these authors are articulated in the ‘author contributions’ section. Competing interests: The authors have declared that no competing interests exist.

marine biodiversity [27, 29, 40, 41] (Fig 1). Despite its significance, there have been limited explorations of this region, especially in the deep sea, a pattern that is also evident for isolated islands globally [42–44] and in Pacific deep waters [45, 46]. The few deep-sea explorations along the Salas y Go´mez Ridge have documented many species new to science (e.g., [47–49]). One campaign surveyed different seamounts across the ridge and found that the fauna on every seamount has a unique community composition, with nearly complete species turnover between opposite ends of the ridge [50]. Because of the high rate of species discoveries and uniqueness within and among seamounts, the marine fauna of this region likely contains numerous undiscovered species, providing an enormous opportunity for marine biodiversity research and high conservation value [51–53]. An expedition to Rapa Nui (i.e., Easter Island) and Salas y Go´mez Island was conducted by the National Geographic Society’s Pristine Seas in February–March 2011 to provide a baseline assessment of the marine ecosystem at 75–2,363 m (Fig 1, [54]). In February and March 2013, National Geographic Society’s Pristine Seas also led an expedition to the Desventuradas Fig 1. Deep-sea camera system deployments on the Salas y Go´mez and Nazca ridges. Dotted lines depict exclusive economic zones (EEZs). Red dots depict camera deployment sites. Bathymetric data source: GEBCO Compilation Group (2020) GEBCO 2020 Grid (doi:10.5285/a29c5465-b138-234de053-6c86abc040b9). https://doi.org/10.1371/journal.pone.0253213.g001 PLOS ONE

Deep-sea biodiversity along the Salas y Go´ mez and Nazca ridges PLOS ONE | https://doi.org/10.1371/journal.pone.0253213 June 30, 2021 3 / 27

Islands (Fig 1). Both expeditions resulted in the first systematic survey of these subregions, including its deep waters, using National Geographic Society Exploration Technology Lab’s deep-sea camera system [55]. These surveys provided a rare opportunity to visually examine the deep waters of the region at subregions that bookend the eastern and western portion of the Salas y Go´mez Ridge, as well as the western portion of the Nazca Ridge (Fig 1). The objectives of this research were to characterize biodiversity and community composi- tion of deep-sea demersal fauna at seamounts surrounding Rapa Nui, Salas y Go´mez Island, and the Desventuradas Islands to better understand deep-ocean biogeography and its environ- mental drivers in the region. We tested the hypotheses that biodiversity and faunal assemblage composition vary with location and environmental variables (e.g., depth, ocean chemistry, habitat complexity, geology). Our study represents the first characterization of deep-sea biodi- versity spanning the east and west ends of the Salas y Go´mez seamount chain, as well as the west end of the Nazca Ridge. By surveying remote deep-sea locations in a region of high con- servation value, we provide important insights into patterns in biodiversity, connectivity, and its drivers for deep-sea ecosystem science and conservation, thereby advancing the global knowledge base for effective conservation and sustainable use. Methods Ethics statement Data were collected by all authors in a collaborative effort. Non-invasive research was con-

ducted, which included photographs and videos as described in the methods below. The Republic of Chile granted all necessary permissions to conduct this research. No vertebrate sampling was conducted and therefore no approval was required by any Animal Care and Use Committee. Our data are publicly available at Data Dryad: doi.org/10.5061/dryad.c866t1g6j. Study region The Salas y Go´mez and Nazca ridges are two adjacent volcanic seamount chains in the south- eastern Pacific that stretch across more than 2,900 km of seafloor (Fig 1). The Nazca Ridge is adjacent to South America and stretches across approximately 1,100 km of seafloor between the eastern edge of the Salas y Go´mez Ridge and the subduction zone off the Peruvian coast (Fig 1). A small northeastern section of the Nazca Ridge is in Peruvian national waters, whereas the majority of this ridge lies in areas beyond national jurisdiction (ABNJ). Stretching across roughly 1,600 km between the Nazca Ridge and Rapa Nui, the central portion of the Salas y Go´mez Ridge lies within ABNJ. Both the western end around the islands of Rapa Nui and Salas y Go´mez and the eastern end around San Fe´lix and San Ambrosio (Desventuradas Islands) fall within Chile’s exclusive economic zone (EEZ) (Fig 1). The Easter Island Ecoregion consists of Rapa Nui, Salas y Go´mez, and the nearby sea- mounts [56]. Salas y Go´mez Island (SyG) is considered part of Polynesia, making it the east- ernmost landmass of Polynesia and the most southeastern coral reef in the Pacific. Rapa Nui

(RN) is considered one of the most isolated inhabited islands globally [57]. The island (total area of 166 km2) is located ~ 2,000 km east of Pitcairn Island, and ~3,700 km west of continen- tal Chile. SyG is a small, emergent rock 0.15 km2 in area and 770 m in total length, with the highest point 30 m above sea level [58]. It is located ~390 km east of RN, and 3,210 km west of the Chilean mainland. These two islands are a part of the Salas y Go´mez Ridge, and are the only places where the submarine mountain range rises above sea level [59, 60]. The several dozen seamounts in the Easter Island Ecoregion extend 2,232 km eastward where the Salas y Go´mez Ridge joins the Nazca Ridge. PLOS ONE Deep-sea biodiversity along the Salas y Go´ mez and Nazca ridges PLOS ONE | https://doi.org/10.1371/journal.pone.0253213 June 30, 2021 4 / 27

The Desventuradas Archipelago is located ~ 200 km southeast of the junction of the Salas y Go´mez and Nazca ridges, rising from a depth of 4,000 m at just over 850 km from the Chilean coast. This archipelago consists of the islands of San Ambrosio, with a surface area of 2.2 km2, San Fe´lix with a surface area of 1.4 km2, and the Gonza´lez and Roca Catedral islets (Saint Peterborough) (Fig 1). In total, the Chilean EEZ around the Desventuradas covers an area of 449,805 km2. Data collection. Remote autonomous benthic landers with baited camera systems [55, 61] were used to conduct stationary video surveys at RN/SyG and the Desventuradas (Fig 1). These systems include high-definition cameras (Sony Handycam HDR-XR520V 12 megapixel) encased in a borosilicate glass sphere that are rated to 11,000 m depth. The camera is fixed at a 45˚ declination from the horizontal plane and has a 35 mm equivalent focal length of 29.8 mm, which results in a horizontal 62.27˚ and a vertical 37.53˚ angular field-of-view. Lighting at depth was achieved through a high-intensity LED array directed using external reflectors, and lighting intensity was constant among all deployments. Depth gauging was conducted using an external pressure sensor. The anchor line extended 2 m from the seafloor to optimize the field of view to both capture large mobile scavenging megafauna and allow for identification of organisms. The deep-sea camera system is positively buoyant, resulting in an ascent rate of 0.5 ms−1, and weighted with a 22 kg external weight, resulting in a descent rate of 1.5 ms−1. The weight

is released by burn wire, activated using onboard battery voltage, and the camera system was located for recovery by communication of an onboard VHF transmitter and locating antennae, with backup location via communication with the ARGOS satellite system. Up to three deep-sea camera systems were simultaneously deployed between 22 February– 4 March, 2011 at 20 stations between 150–1,850 m depth (�X = 1,104.6 ± 433.6) along the island slopes of SyG and RN and the slopes of two seamounts west of SyG (Fig 1). At Desventuradas, 27 successful deployments were conducted between 10–26 February, 2013 at 75–2,363 m depth (�X = 755.8 ± 649.2) (Fig 1). Cameras were deployed with a minimum distance of 500 m from another deployment location, and each camera was baited with chopped frozen fish and deployed for 1 to 5 hours with an average footage per deployment of 112 min at Desventuradas and 129 min at RN/SyG. Deployments targeted geological features and specific depths based on available bathymetry. Sampling design aimed to represent a gradient of depths away from a feature (e.g., island, seamount), and thus deployments generally occurred in a line radiating out from the shallow depths of a feature to deeper waters. The substrata for each camera deployment were classified as hard or soft, based on video footage. Seafloor type was defined as soft if the seafloor observed during the deployment con- sisted of �80% sediment (sands and muds), and hard if <80% of the seafloor observed during the deployment was composed of sediment. Hard habitats included rock, cobbles, pebbles,

gravel, biogenic material (including live coral), coral rubble, and shell material. Biological data processing. Annotations were made from the video footage based on standardized taxonomic nomenclature according to the World Register of Marine Species (WoRMS). Video files were stored in Tator, which is a cloud-based collaborative video annota- tion platform developed by CVisionAI (http://cvisionai.com/). Frame grabs of representatives of each species were taken in Tator for archival and identification purposes. Individuals were identified to the lowest possible taxonomic level using the Hawaii Under- water Research Laboratory (HURL) database ([62]; http://www.soest.hawaii.edu/HURL/ HURLarchive/guide.php), the NOAA Office of Ocean Exploration Benthic Deepwater Animal identification Guide (https://oceanexplorer.noaa.gov/okeanos/animal_guide/animal_guide. html), FishBase [63], regional guides and reports of trawl data from the Salas y Go´mez and PLOS ONE Deep-sea biodiversity along the Salas y Go´ mez and Nazca ridges PLOS ONE | https://doi.org/10.1371/journal.pone.0253213 June 30, 2021 5 / 27

Nazca ridges [64–67], and input from regional taxonomic experts (Figs 2 and 3). The maxi- mum number of individuals of each species in a single video frame (MaxN), rather than a total tally per deployment, was recorded to ensure that individuals were not double counted [68]. Fig 2. Representative and unique taxa observed with deep-sea camera systems around Rapa Nui and Salas y Go´mez islands. (A) Hexanchus griseus (B) Hydrolagus melanophasma (C) Chromis mamatapara� (D) Parapristipomoides squamimaxillaris (E) Homolidae, Yaldwynopsis sp. (F) Geryonidae (G) Rexea sp. (H) Polymixia salagomeziensis�. �endemic species. https://doi.org/10.1371/journal.pone.0253213.g002 PLOS ONE Deep-sea biodiversity along the Salas y Go´ mez and Nazca ridges PLOS ONE | https://doi.org/10.1371/journal.pone.0253213 June 30, 2021 6 / 27

Taxa were classified as vulnerable marine ecosystem (VME) taxa based on recent studies [69, 70]. Geospatial data processing. To characterize samples by the physicochemical factors of a deployment location, each deployment was attributed with its associated ecological marine unit (EMU) [71]. EMUs are ecological divisions of the 3D marine environment, representing physically and chemically distinct water volumes, from sea surface to seafloor. The divisions are based on statistical clustering of six marine environmental parameters (temperature, Fig 3. Representative and unique taxa observed with deep-sea camera systems at Desventuradas Islands. (A) Nemadactylus gayi�, Scorpaena orgila�, Squalus sp. (B) Pterygotrigla picta (C) Hydrolagus sp. (D) Lithodidae (E) Antimora rostrata (F) Squalus sp. �endemic species. https://doi.org/10.1371/journal.pone.0253213.g003 PLOS ONE Deep-sea biodiversity along the Salas y Go´ mez and Nazca ridges PLOS ONE | https://doi.org/10.1371/journal.pone.0253213 June 30, 2021 7 / 27

salinity, dissolved oxygen, nitrate, phosphate, and silicate) (S1 Table). The environmental attri- bute data used in the statistical clustering are based on a set of objectively analyzed climatologi- cal fields of in situ long-term historical average data from the 2013 World Ocean Atlas version 2 (https://www.nodc.noaa.gov/OC5/woa13/) and represent the means of the decadal averages from a 57-year climatology of the six parameters. The database is a point-mesh grid that includes over 52 million points that depict the global ocean in x, y, and z dimensions. The data points were statistically clustered based on the six environmental parameters using a k-means clustering algorithm to identify environmentally distinct regions in the water column. The resulting 37 distinct EMUs partition the ocean environment into ecologically relevant regions and are a valuable tool in exploring drivers of biological distributions. This resulting EMU dataset has a spatial resolution of ¼˚ × ¼˚ (~27 km × 27 km at the equator) in the horizontal dimension and was accessed from ESRI EMU online (http://esri. maps.arcgis.com/home/group.html?id=6c78a5125d3244f38d1bc732ef0ee743). EMU numbers, and values for oceanographic variables derived from the World Ocean Atlas 2013 database (temperature, salinity, dissolved oxygen, %O2 saturation, nitrate, silicate, phosphate, chloro- phyll-a) were extracted from the EMU dataset using the Extract Multi Values to Points tool in the ArcGIS Spatial Analyst toolbox. EMUs and associated environmental variables vary across

horizontal space as well as with depth, represented by the EMU dataset as a 3D point grid. Deployments were attributed with the respective EMU at the given sample depth, thus captur- ing depth-specific variations in physicochemical factors. Values for particulate organic carbon (POC) flux to the seafloor were obtained from Lutz et al. [72] via Watling et al. [73]. Seamount data were obtained from the Global Seamount Database [74]. Distance to the nearest seamount and the number of seamounts near sample sites were derived for a 30 km radius around deployment locations using the Buffer tool in ArcGIS Pro. Age of the underly- ing crust [75] was extracted for the seamount nearest each deployment location, using the Extract Multi Values to Points tool, as an indication of seamount age. Bathymetric position index (BPI) is an index of bathymetric heterogeneity that highlights areas of topographic anomalies and is used to identify benthic features (e.g., ridges and valleys) [76]. It is a continu- ous metric of the elevation of a pixel relative to other pixels in a landscape [77]. BPI values at or near zero represent consistent bathyscapes, either flat areas or areas of constant slope, whereas sharp peaks or pits are represented by higher absolute values [77, 78]. BPI was calcu- lated in ArcMap 10.2 using the Benthic Terrain Modeler extension [79]. Input bathymetric data used was Global Multi-Resolution Topography (GMRT) [80] from the Marine Geosci- ence Data System (www.marine-geo.org), with spatial resolution of 3 arcseconds (~90 m). The

GMRT dataset, available through OpenTopography (https://opentopography.org/), is a compi- lation of edited multibeam sonar datasets collected worldwide. The data for our study region includes swaths of mapping data throughout the region, intersecting and surrounding deploy- ment locations. Broad-scale BPI was calculated using an inner radius of 25 cells and an outer radius of 50 cells. Fine-scale BPI was derived with inner and outer radii of 1 and 10 cells, respectively. Both broad- and fine-scale BPI have been shown to be associated with variations in marine faunal communities [76, 81], including those of the deep ocean [45, 82]. Statistical methods Due to the numerical dominance of Amphipoda in our samples, they were excluded from all analyses, except for the comparisons of number of taxa and analyzed separately. Total MaxN per deployment was calculated as the sum of MaxN for all taxa for that deployment. Species diversity was calculated from the Shannon-Weaver Diversity Index [83]:H0 ¼ PðpilnpiÞ, PLOS ONE Deep-sea biodiversity along the Salas y Go´ mez and Nazca ridges PLOS ONE | https://doi.org/10.1371/journal.pone.0253213 June 30, 2021 8 / 27

where pi is the proportion of all individuals counted that were of species i. Pielou’s evenness was calculated as: J = H´/ln(S), where S is the total number of species present. A principal components analysis was used to examine environmental variables between subregions. Environmental variables in this analysis included depth, temperature, salinity, dis- solved oxygen, %O2 saturation, nitrate, silicate, phosphate, chlorophyll-a, broad-scale BPI, fine-scale BPI, POC, crust age, nearest seamount, and number of seamounts within 30 km of the sample sites. Data were Box-Cox transformed, centered, and standardized prior to analysis. The absolute values of fine-scale and broad-scale BPI metrics were used since peaks and troughs both represent measures of habitat complexity. A comparison of environmental vari- ables between subregions and habitats was conducted using permutation-based multivariate analysis of variance (PERMANOVA) with a Euclidean distance matrix. Deployment depth was examined as a predictor variable in explaining the variability in invertebrate taxa richness, fish taxa richness, and fish MaxN between subregions and habitats (i.e., hard and soft) using generalized linear models (GLMs) with a Poisson distribution with a log-link function. The GLMs for fish diversity and evenness used a Gamma distribution with a log-link function. Comparison of Amphipoda abundance between subregions was tested using a Wilcoxon rank sum test. Comparisons of invertebrate and fish assemblage structure between

subregions and habitats was also performed using PERMANOVA with Bray-Curtis similarity matrices. Similarity percentages analysis (SIMPER) was used to examine differences in inverte- brate and fish assemblages between subregions using Bray-Curtis similarity analysis of hierar- chical agglomerative group average clustering. Relationships between environmental variables and invertebrate and fish assemblages were analyzed using distance-based linear models (DistLM) and redundancy analysis (dbRDA). Linear models are appropriate for these data because a preliminary detrended correspondence analysis showed short gradient lengths (<2 SD) [84]. The dbRDA introduces a series of explanatory (environmental) variables and resembles the model of multivariate multiple regression, allowing us to determine what linear combinations of these environmental vari- ables determine the gradients. These analyses were conducted using Bray-Curtis similarity matrices on MaxN Class or Order-level invertebrate data and fish family-level data. MaxN invertebrate data were ln(x+1) transformed and MaxN fish data were 4th-root transformed prior to analyses to conform to the assumptions of normality and homogeneity of variances [85]. Normality was tested using a Shapiro-Wilk W test (p < 0.05) while a Bartlett’s test (p < 0.05) was used to examine homogeneity of variance. We determined the model with envi- ronmental drivers that best explained variation in invertebrate and fish assemblages using a stepwise routine that employed 9999 permutations based on Akaike’s Information Criterion

with second order correction (AICc) selection criterion. Environmental variables that had high correlations (r > 0.95) were removed prior to analysis. Those removed were temperature, silicate, %O2 saturation, and phosphate. All statistical analyses were conducted using JMP Pro 15 and Primer 6.0. Results Environmental variables Of the 20 camera deployments conducted around RN/SyG, 40% (N = 8) were located on hard bottom habitat. In contrast, only 15% (n = 4) of the 27 deployments at Desventuradas were located on this habitat type. Deployment depth was not significantly different between subre- gions, habitats, or their interaction (all GLMs p > 0.05). Subregions were well separated in ordination space based on environmental variables (Fig 4). Environmental variables were significantly different between subregions (PERMANOVA PLOS ONE Deep-sea biodiversity along the Salas y Go´ mez and Nazca ridges PLOS ONE | https://doi.org/10.1371/journal.pone.0253213 June 30, 2021 9 / 27

pseudo-F1,43 = 5.456, p < 0.001) but not between habitats (PERMANOVA pseudo-F1,43 = 1.524, p = 0.188) or their interaction (PERMANOVA pseudo-F1,43 = 0.699, p = 0.552). PC1 accounted for 48.1% of the variation in environmental variables, with the major loadings being silicate (-0.349), nitrate (-0.342), phosphate (-0.340), apparent oxygen utilization (-0.340), and depth (-0.335). PC2 accounted for an additional 17.8% of the variation in environmental variables, with the major loadings being crust age (-0.504), chlorophyll-a (-0.503), number of seamounts within 30 km pf the sample sites (0.358), and dissolved O2 (0.329). Nitrate, phosphate, silicate, salinity, and depth were all highly correlated with one another and inversely related to tempera- ture. Subregions separated out along PC2 with sites at Desventuradas being older geologically with higher chlorophyll-a. Deployments at RN/SyG had more seamounts within 30 km of the sample sites and higher measures of habitat complexity (BPI) at both fine and broad scales. Invertebrate assemblages There were 56 invertebrate taxa from 5 phyla, 12 classes, 13 orders, and 22 infra-orders or fam- ilies identified on camera deployments (S2 Table). There were no significant differences in the Fig 4. Principal Components Analysis (PCA) of environmental variables at camera deployment sites. Data were Box-Cox transformed prior to analysis. Des. = Desventuradas. https://doi.org/10.1371/journal.pone.0253213.g004 PLOS ONE Deep-sea biodiversity along the Salas y Go´ mez and Nazca ridges

PLOS ONE | https://doi.org/10.1371/journal.pone.0253213 June 30, 2021 10 / 27

number of invertebrate taxa between subregions, habitats, or their interactions (all GLMs p > 0.05). Amphipoda were the most common and abundant taxa observed. They were pres- ent on 95% of the deployments at RN/SyG (�X = 68.50 ± 108.67) and 48% of the deployments at Desventuradas (�X= 11.89 ± 23.9). In addition, Amphipoda were 5.8 times more abundant at RN/SyG compared to Desventuradas (Wilcoxon = 2.273, p = 0.023). Invertebrate assemblage structure was significantly different between subregions (PERMA- NOVA pseudo-F1,43 = 12.848, p < 0.001) but not between habitats (PERMANOVA pseudo- F1,43 = 1.927, p = 0.096) or their interaction (PERMANOVA pseudo-F1,43 = 0.186, p = 0.951). The invertebrate assemblages were more similar among deployments at RN/SyG (45.98%) compared to Desventuradas (21.37%), with average dissimilarity between subregions equal to 85.52%. Decapoda accounted for the greatest dissimilarity between subregions (60.83%), with abundances at RN/SyG (�X= 22.40 ± 25.91) more than 15 times higher compared to Desventur- adas (�X = 1.48 ± 1.74). The dominant components of this order were shrimp from the families Benthesicymidae, Aristeidae, and Sergestidae. Anthozoa accounted for an additional 15.1% of the dissimilarity between subregions, with abundances at Desventuradas (�X = 5.26 ± 11.96) more than 20 times higher than at RN/SyG (�X= 0.25 ± 0.72). Vulnerable marine ecosystems Consistent with other studies, colonial Anthozoa and Porifera were classified as vulnerable marine ecosystem (VME) taxa ([79] and references therein). Within Anthozoa, there were sev-

eral VME taxa, including Pennatulacea, Antipatharia, Actiniaria, and Alcyonacea (Fig 5). Within Porifera, Hexactinellida and Demospongiae were classified as VME taxa. More than half the deployments at Desventuradas (n = 17, 63%) had VME taxa present, while only 3 (15%) of the deployments at RN/SyG had VME taxa present. Invertebrate-habitat relationships The best distance-based linear model solution between environmental variables and inverte- brate assemblage structure included crust age, nitrate, broad-scale BPI, and depth (AICc = 349.17, R2 = 0.475) (Table 1, Fig 6). Crust age accounted for 26.75% of the variation followed by nitrate (13.96%), broad-scale BPI (3.84%), and depth (2.95%). Deployments by subregions were well-separated in ordination space, and sites within Desventuradas had a greater spread and lower concordance compared to RN/SyG (Fig 7). dbRDA1 accounted for 70.35% of the fit- ted variation and 33.41% of the total variation, with crust age (ρ = -0.779) and nitrate (ρ = 0.559) being the major drivers along this axis (Table 1). dbRDA2 accounted for 19.75% of the fitted variation and 9.38% of the total variation, with all four variables contributing to this axis. Decapoda (ρ = 0.518), Echinodermata (ρ = -0.448), and Anthozoa (ρ = -0.407) were most highly correlated with dbRDA1. Euphausiacea (ρ = 0.612), Holothuroidea (ρ = 0.391), and Scy- phozoa (ρ = 0.385) were most closely correlated with dbRDA2. Fish assemblages There were 61 fish taxa from 17 orders and 39 families identified on camera deployments (S3

Table). Most families only had one or two representative taxa except for Etmopteridae, Macro- uridae, and Serranidae, which each had three taxa per family. There were no significant differ- ences in the taxonomic richness, diversity, or evenness between subregions, habitats, or their interactions (Fig 8, all GLMs p > 0.05). MaxN was significantly different between subregions (GLM X21,43 = 4.156, p = 0.042) and for the interaction between region and habitat (GLM X21,45 = 34.027, p < 0.001) but not between habitats alone (GLM X21,43 = 0.414, p = 0.520). PLOS ONE Deep-sea biodiversity along the Salas y Go´ mez and Nazca ridges PLOS ONE | https://doi.org/10.1371/journal.pone.0253213 June 30, 2021 11 / 27

MaxN was highest in Desventuradas soft bottom habitats and lowest in RN/SyG soft bottom habitats. Fish assemblage structure was significantly different between subregions (PERMANOVA pseudo-F1,43 = 2.482, p = 0.022) but not between habitats (PERMANOVA pseudo-F1,43 = 1.431, p = 0.184) or their interaction (PERMANOVA pseudo-F1,43 = 1.974, p = 0.059). The average dissimilarity between subregions was 91.01%. Squalidae accounted for the greatest dis- similarity between subregions (18.27%), with abundances at Desventuradas (�X= 2.26 ± 2.49) an order of magnitude greater than at RN/SyG (�X = 0.21 ± 0.54) (Table 2). Moridae accounted for an additional 7.98% of the dissimilarity between subregions, with abundances at Desven- turadas (�X= 0.63 ± 1.04) 42% higher than at RN/SyG (�X = 0.26 ± 0.56). Abundances of Fig 5. Vulnerable marine ecosystem indicator taxa, including deep-sea corals and sponges, observed on deep-sea camera deployments along the Salas y Go´mez and Nazca ridges. (A) Sala y Go´mez—552 m (B) Desventuradas– 1,714 m. https://doi.org/10.1371/journal.pone.0253213.g005 PLOS ONE Deep-sea biodiversity along the Salas y Go´ mez and Nazca ridges PLOS ONE | https://doi.org/10.1371/journal.pone.0253213 June 30, 2021 12 / 27

Macrouridae at RN/SyG (�X = 0.53 ± 1.02) were 43% times higher than at Desventuradas (�X= 0.37 ± 0.69) and accounted for an additional 7.31% of the dissimilarity between subregions. Deployments within subregions were highly dissimilar, with the average similarity among deployments at RN/SyG equal to 14.47% and the average similarity among deployments at Desventuradas equal to 15.99%. Fish-habitat relationships The best distance-based linear model solution between environmental variables and fish assemblage structure included depth, salinity, crust age, nitrate, and broad-scale BPI (AICc = Table 1. Distance-based linear model for invertebrate assemblages and environmental variables. Sequential tests Multiple partial correlations Variable AICc SS(trace) Pseudo-F P Prop. Cumul. dbRDA1 dbRDA2 Crust age 357.63 31603 16.432 0.001 0.267 0.267 -0.779 0.546 Nitrate 349.98 16493 10.359 0.001 0.140 0.407 0.559 0.558 BPI broad 349.23 4531 2.974 0.019 0.038 0.445 -0.036 0.425 Depth 349.17 3487 2.361 0.052 0.030 0.475 0.282 0.457 Multiple partial correlations are the relationships between dbRDA coordinate axes and orthonormal X variables. Prop. = proportion of variation explained. Cumul. = cumulative proportion variation explained. https://doi.org/10.1371/journal.pone.0253213.t001 Fig 6. Environmental variables at the deep-sea camera system deployment locations. (A) Crust age (younger ! older: yellow ! red) was extracted for the nearest

seamount to each deployment location. (B) Nitrate levels; bubbles depict nitrate levels at deployment location and depth (low ! high nitrate: small/light ! big/dark). (C) Absolute broad-scale bathymetric position index (BPI); bubbles depict absolute BPI values at deployment location (low ! high BPI: small/light ! big/dark), basemap depicts absolute BPI values of region (low !high: dark blue ! gray). (D) Depth of each deployment; bubbles depict depth at deployment location (small/light ! big/dark: shallow ! deep), basemap depicts bathymetry (deep ! shallow: dark blue ! white). Purple lines in Panel D are 500 m contour lines. Bathymetric data source: Global Multi-Resolution Topography (GMRT) [80, www.marine-geo.org]. https://doi.org/10.1371/journal.pone.0253213.g006 PLOS ONE Deep-sea biodiversity along the Salas y Go´ mez and Nazca ridges PLOS ONE | https://doi.org/10.1371/journal.pone.0253213 June 30, 2021 13 / 27

367.83, R2 = 0.419) (Table 3). Depth accounted for 22.68% of the variation followed by salinity (5.79%), crust age (4.83%), nitrate (4.58%), and broad-scale BPI (4.12%). Deployments by sub- regions were well-separated in ordination space and sites within Desventuradas had a greater spread and lower concordance compared to RN/SyG (Fig 9). dbRDA1 accounted for 59.04% of the fitted variation and 24.79% of the total variation. Depth (ρ = 0.947) was the major driver of dbRDA1. dbRDA2 accounted for 15.16% of the fitted variation and 6.37% of the total varia- tion, with the major drivers being crust age (ρ = -0.810), followed by nitrate (ρ = -0.385) and salinity (ρ = -0.384). Squalidae (ρ = -0.797), Triglidae (ρ = -0.627), Moridae (ρ = 0.519), and Synaphobranchidae (ρ = 0.509) were most highly correlated with dbRDA1. Gempylidae (ρ = 0.678), Sternoptychi- dae (ρ = 0.407), Polymixiidae (ρ = 0.350) and Moridae (ρ = 0.330) were most closely correlated with dbRDA2. The snake mackerel family (Gempylidae), primarily Rexea spp., was correlated with hard bottom sites at RN/SyG. Spiny dogfish sharks (Squalidae—Squalus spp.) and sea robins (Triglidae—Pterygotrigla picta) were most closely correlated with shallow sites at Fig 7. Distance-based redundancy analysis (dbRDA) ordination to investigate the relationship between the environmental variables and invertebrate assemblages. Bray-Curtis similarity matrix based on ln(x+1) transformed MaxN higher level invertebrate taxa. Symbol sizes are proportional to depth. https://doi.org/10.1371/journal.pone.0253213.g007 PLOS ONE

Deep-sea biodiversity along the Salas y Go´ mez and Nazca ridges PLOS ONE | https://doi.org/10.1371/journal.pone.0253213 June 30, 2021 14 / 27

Desventuradas, while morid cod (Moridae–primarily Antimora rostrata) were correlated with deeper sites at Desventuradas. Cutthroat eels (Synaphobranchidae–primarily cf. Synaphobran- chus affinis) were correlated with deeper sites at both RN/SyG and Desventuradas. Discussion Community assemblage and habitat relationships Faunal community structure based on our camera deployments was highly dissimilar between and within subregions. Our findings are consistent with previous studies that note that every seamount along the ridges may have a unique faunal composition [50, 65, 66]. Seamount sys- tems such as the Salas y Go´mez and Nazca ridges, which are geographically and hydrographi- cally isolated due to the influence of the region’s current systems, are likely to have developed highly endemic taxa and ecosystems [66, 86]. The Salas y Go´mez and Nazca ridges comprise over 110 seamounts, which represent roughly 41% of seamounts found in the southeastern Fig 8. Fish assemblage characteristics (taxa richness, MaxN, diversity, and evenness) at Desventuradas and RN/SyG on hard and soft-bottom habitats. Box plots showing median (black line), mean (red line), upper and lower quartiles, and 5th and 95th percentiles. https://doi.org/10.1371/journal.pone.0253213.g008 PLOS ONE Deep-sea biodiversity along the Salas y Go´ mez and Nazca ridges PLOS ONE | https://doi.org/10.1371/journal.pone.0253213 June 30, 2021 15 / 27

Pacific [41]. Given the high rate of discovery of new species, this region may well harbor many more undiscovered marine species [51, 52]. Depth, crust age, broad-scale BPI, and nitrate explained the most variation in both inverte- brate and fish assemblages on our camera deployments, and our findings are largely consistent with other studies [45, 82, 87]. Community composition on seamounts is depth stratified, reflecting environmental gradients that correlate with depth, such as temperature, oxygen con- centration, food availability, and pressure [32]. The high habitat heterogeneity on seamounts, quantified by BPI, combined with the biological effects of depth, increases local biodiversity by Table 2. SIMPER for fish families most responsible for the percent dissimilarities between subregions using Bray-Curtis similarity analysis of hierarchical agglom- erative group average clustering. Families Des RN/SyG Diss. Contrib% Cum% Squalidae 2.26 (2.49) 0.21 (0.54) 16.63 (0.89) 18.27 18.27 Moridae 0.63 (1.04) 0.26 (0.56) 7.27 (0.69) 7.98 26.25 Macrouridae 0.37 (0.69) 0.53 (1.02) 6.65 (0.68) 7.31 33.56 Halosauridae 0.41 (0.97) 0.37 (0.68) 6.25 (0.63) 6.87 40.43 Synaphobranchidae 0.30 (0.67) 0.47 (0.70) 6.18 (0.64) 6.80 47.22 Pentacerotidae 1.04 (2.9) - 5.65 (0.39) 6.21 53.43 Gempylidae - 0.58 (0.90) 5.57 (0.55) 6.12 59.55 Cheilodactylidae 0.85 (3.05) - 3.12 (0.27) 3.42 62.97 Carangidae 0.28 (1.53) 0.16 (0.97) 2.97 (0.32) 3.27 66.24 Etmopteridae 0.22 (0.42) - 2.50 (0.47) 2.75 68.99

Chimaeridae 0.07 (0.27) 0.16 (0.50) 2.45 (0.40) 2.69 71.68 Sebastidae 0.37 (1.18) - 2.33 (0.30) 2.56 74.24 Ophidiidae 0.30 (1.20) - 2.19 (0.26) 2.41 76.65 Serranidae 0.30 (1.35) 0.53 (2.29) 2.14 (0.35) 2.35 79.00 Labridae 0.15 (0.77) 0.16 (0.69) 1.58 (0.23) 1.74 80.73 Lutjanidae - 0.58 (2.29) 1.53 (0.31) 1.68 82.41 Triglidae 0.26 (0.53) - 1.45 (0.45) 1.59 84.00 Polymixiidae 0 (0) 0.16 (0.50) 1.39 (0.29) 1.52 85.53 Monocentridae 0.22 (0.64) - 1.21 (0.34) 1.32 86.85 Myctophidae 0.07 (0.27) 0.05 (0.23) 1.12 (0.35) 1.23 88.08 Alepocephalidae 0.07 (0.27) - 1.05 (0.24) 1.15 89.23 Hexanchidae - 0.11 (0.46) 0.98 (0.22) 1.08 90.31 Average dissimilarity = 91.08%. Values are average MaxN with one standard deviation of the mean in parentheses. DES = Desventuradas, Diss. = dissimilarity, Contrib % = percent contribution to dissimilarity, Cum% = cumulative percent contribution to dissimilarity. https://doi.org/10.1371/journal.pone.0253213.t002 Table 3. Distance-based linear model for fish assemblages and environmental variables. Sequential tests Multiple partial correlations Variable AICc SS(trace) Pseudo-F P Prop. Cumul. dbRDA1 dbRDA2 Depth 371.18 39264 12.903 0.001 0.227 0.227 0.947 -0.006 Salinity 369.90 10021 3.479 0.001 0.058 0.285 0.264 -0.384 Crust age 369.08 8370 3.044 0.002 0.048 0.333 -0.033 -0.81 Nitrate 368.34 7924 3.020 0.003 0.046 0.379 0.166 0.385 BPI broad 367.83 7136 2.842 0.007 0.041 0.420 0.076 0.217

Multiple partial correlations are the relationships between dbRDA coordinate axes and orthonormal X variables. Prop. = proportion of variation explained. Cumul. = cumulative proportion variation explained. https://doi.org/10.1371/journal.pone.0253213.t003 PLOS ONE Deep-sea biodiversity along the Salas y Go´ mez and Nazca ridges PLOS ONE | https://doi.org/10.1371/journal.pone.0253213 June 30, 2021 16 / 27

providing neighboring areas with different ecological niches, which helps explain the high dis- similarity of invertebrate assemblages within and between seamounts of a region [88]. In our study, crust age was highly correlated with region, as the seamounts become progressively older moving eastward along the ridges from 2 million years on the western portion of the chain to over 27 million years towards the northeastern end [60, 89]. An endemism and biodiversity hotspot The nearshore marine ecosystems around the emergent islands of RN/SyG and Desventuradas are known to have some of the highest marine endemism anywhere on Earth [29, 54, 67]. Recent genetic studies have confirmed that Rapa Nui is an endemism hotspot: a location for the emergence of small-range endemic fishes (likely resulting from founder-event speciation), a route of dispersion for larger-range endemics, and a stepping stone for the diversification of some groups [90]. Fig 9. Distance-based redundancy analysis (dbRDA) ordination to investigate the relationship between the environmental variables and fish assemblages. Bray-Curtis similarity matrix based on 4th root-transformed MaxN higher level fish taxa. Symbol sizes are proportional to depth. https://doi.org/10.1371/journal.pone.0253213.g009 PLOS ONE Deep-sea biodiversity along the Salas y Go´ mez and Nazca ridges PLOS ONE | https://doi.org/10.1371/journal.pone.0253213 June 30, 2021 17 / 27

Our study supports evidence that the biodiversity hotspot and role of ecosystem connectiv- ity extends beyond the shallow coastal waters, and into the deep sea surrounding RN/SyG and Desventuradas. For example, Chromis mamatapara is a recently described damselfish from around Rapa Nui and is believed to be endemic to the mesophotic ecosystems of Rapa Nui, Salas y Go´mez, and nearby seamounts. We recorded two individuals of Chromis mamatapara at Salas y Go´mez Island in 150 m (Fig 2). Several species are known to occur at both RN/SyG and Desventuradas islands, which highlights the connectivity between these emergent islands along the ridge. The moray eel Gymnothorax bathyphilus, which was previously known only from Rapa Nui and Desventuradas, was observed at Salas y Go´mez Island in 150 m. Similarly, Gay’s morwong (Nemadactylus gayi) is known only from the Desventuradas and Juan Ferna´n- dez islands and was observed at three sites around Desventuradas with up to 14 individuals in a single frame. The beardfish (Polymixia salagómeziensis) is endemic to the Salas y Go´mez Ridge and we recorded individuals at two locations near Salas y Go´mez Island between 640 and 644 m. Plectranthias nazcae is endemic to the Nazca Ridge and was recorded at Desven- turadas in 167 m. We observed several invertebrate species that are endemic to the region, including two crabs (Chaceon chilensis and Paromola rathbuni), one lobster (Jasus frontalis), and one sea urchin (Stereocidaris nascaensis). Collectively, these findings highlight that the

region is a global hotspot of endemic of species, and that the ridges are important for species connectivity. Threats Fishing. A number of commercially exploited fisheries species were observed during our study. These included two endemic decapods: the Juan Ferna´ndez lobster (Jasus frontalis), which was only observed at one station at Desventuradas in 156 m, and the Chilean golden crab (Chaceon chilensis), which was observed on 44% of the deployments around Desventura- das between 201 and 1,352 m. These species are currently harvested at sustainable levels by the small-scale artisanal fishers from the nearby Juan Ferna´ndez Archipelago [91, 92]. The recently created Nazca-Desventuradas Marine Park and Mar de Juan Ferna´ndez Marine Park were strongly promoted by the Juan Ferna´ndez community to preserve their traditional lobster fish- ing and to exclude industrial fishing [93]. In addition, the Juan Ferna´ndez trevally (Pseudocar- anx chilensis) and Gay’s morwong (Nemadactylus gayi) are both endemic to the Desventuradas and Juan Ferna´ndez archipelagos, where they are important bait for the lobster fishery [94]. However, these endemic species are not protected on the seamounts in adjacent ABNJ and are susceptible to overexploitation by industrial-scale fisheries. Immediate threats to the region are from large-scale foreign fishing fleets, primarily from China, who accounted for over 72% of the total fishing effort in this region between 2012 and 2020 [53]. Seamount habitats worldwide are often heavily impacted by fishing efforts as deep-

sea fish species aggregate and often spawn on seamounts and can be slow to recover from dis- turbance. For example, heavy fishing efforts in the 1960s to the 1980s on seamounts on the far northwestern Hawaiian Ridge and Emperor Seamount Chain resulted in the largest quantity of fish and invertebrate biomass removed from any documented seamount fishery in the world [38]. These fisheries collapsed by the early 1980s, and after 40 years of protection, the habitats are only showing signs of limited recovery [39]. Further, seamounts are often domi- nated by deep-sea corals and sponges that are generally slow growing and extremely long lived. For example, a black coral species taken at 450 m off O‘ahu, Hawai‘i is the oldest skele- tal-accreting marine organism known (4,265 yrs) and the oldest colonial organism yet found on Earth [95]. PLOS ONE Deep-sea biodiversity along the Salas y Go´ mez and Nazca ridges PLOS ONE | https://doi.org/10.1371/journal.pone.0253213 June 30, 2021 18 / 27