Characterization of deep-sea benthic invertebrate megafauna of the Galapagos Islands

The deep sea represents the largest and least explored biome on the planet. Despite the iconic status of the Galapagos Islands and being considered one of the most pristine locations on earth, the deep-sea benthic ecosystems of the archipelago are virtually unexplored in comparison to their shallow-water counterparts. In 2015, we embarked on a multi-disciplinary scientific expedition to conduct the first systematic characterization of deep-sea benthic invertebrate communities of the Galapagos, across a range of habitats. We explored seven sites to depths of over 3,300 m using a two-part Remotely Operated Vehicle (ROV) system aboard the E/V Nautilus, and collected 90 biological specimens that were preserved and sent to experts around the world for analysis. Of those, 30 taxa were determined to be undescribed and new to science, including members of five new genera (2 sponges and 3 cnidarians). We also systematically analysed image frame grabs from over 85 h of ROV footage to investigate patterns of species diversity and document the presence of a range of underwater communities between depths of 290 and 3,373 m, including cold-water coral communities, extensive glass sponge and octocoral gardens, and soft-sediment faunal communities. This characterization of Galapagos deep-sea benthic invertebrate megafauna across a range of ecosystems represents a first step to study future changes that may result from anthropogenic impacts to the planet’s climate and oceans, and informed the creation of fully protected deep-water areas in the Galapagos Marine Reserve that may help preserve these unique communities in our changing planet.

The deep sea represents the largest and least explored biome on the planet. Despite the iconic status of the Galapagos Islands and being considered one of the most pristine locations on earth, the deep-sea benthic ecosystems of the archipelago are virtually unexplored in comparison to their shallow-water counterparts. In 2015, we embarked on a multi-disciplinary scientific expedition to conduct the first systematic characterization of deep-sea benthic invertebrate communities of the Galapagos, across a range of habitats. We explored seven sites to depths of over 3,300 m using a two-part Remotely Operated Vehicle (ROV) system aboard the E/V Nautilus, and collected 90 biological specimens that were preserved and sent to experts around the world for analysis. Of those, 30 taxa were determined to be undescribed and new to science, including members of five new genera (2 sponges and 3 cnidarians). We also systematically analysed image frame grabs from over 85 h of ROV footage to investigate patterns of species diversity and document the presence of a range of underwater communities between depths of 290 and 3,373 m, including cold-water coral communities, extensive glass sponge and octocoral gardens, and soft-sediment faunal communities. this characterization of Galapagos deep-sea benthic invertebrate megafauna across a range of ecosystems represents a first step to study future changes that may result from anthropogenic impacts to the planet's climate and oceans, and informed the creation of fully protected deep-water areas in the Galapagos Marine Reserve that may help preserve these unique communities in our changing planet.
When Charles Darwin visited the Galapagos Islands aboard the HMS Beagle in 1835 1 , deep-sea exploration was in its infancy and the presence of life at depth was questioned 2,3 . The Galapagos served as inspiration for Darwin's breakthrough theory of evolution by means of natural selection 4 and ever since, the iconic islands have captivated the global scientific community. While the Galapagos terrestrial fauna and flora have been the subject of intense study [5][6][7] , and research on shallow (< 40 m) coastal marine ecosystems has been considerable over the past four decades [8][9][10][11][12] , only a limited number of expeditions have explored beyond SCUBA diving depths. Most of these deep-sea research efforts have followed on the historic discovery of hydrothermal vents on the Galapagos

Results and discussion
Throughout the ROV surveys we documented the presence of a range of underwater communities between depths of 290 and 3,373 m, including cold-water coral communities, extensive glass sponge and octocoral gardens, and soft-sediment faunal communities ( Fig. 1; Table 1). A total of 90 taxa from six phyla, 13 classes, 22 orders and 44 families were collected and identified to the lowest taxonomic level possible by experts. Of the specimens analysed, thirty were identified as species new to science. Eleven of these taxa were also identified as belonging to new undescribed genera ( Table 2).
Some of the new taxa discovered includes a giant solitary coral of the genus Bathyalcyon (Fig. 2c), the first known for the region; a new genus of bamboo corals of the family Isididae, with this specimen recorded at 2,923 m deep (Fig. 2d); a new genus of glass sponges of the family Farreidae, with some of the colonies encountered growing over 1 m in width (Fig. 2e); and a new demosponge genus in the family Phellodermidae, with an unusual 'kebab-like' growth form (Fig. 2f).
This study represents one of the most comprehensive surveys of deep-sea invertebrate megafauna across a varied range of habitats, depth gradients and geographical locations within the Galapagos Islands. To our knowledge, this study provides the first systematic description of deep-sea specimens of several phyla, including Porifera, Echinodermata, Arthropoda and Annelida for Galapagos and also the Tropical Eastern Pacific (TEP) 42 . We also provide additional deep-water octocoral (Anthozoa) discoveries, a group that has previously received by far the most scientific attention from deep Galapagos waters 24,25,42 . Heterogeneous environments were found on many of the seamounts explored, including the presence of multiple fragile habitats, such as glass sponge gardens, coral gardens and cold-water coral colonies, that are considered Vulnerable Marine Ecosystems (VMEs) by the United Nations General Assembly 43,44 .
Our analysis of ROV imagery, revealed that VME indicator species were present on all the dives conducted, although their presence varied considerably among locations. According to the criteria used in our image analysis, VME communities were present on three of the sites visited, including one located on the far north bioregion of the archipelago (Dive H1435) and two sites within the central-south-eastern bioregion (Dive H1443; Table 3). The presence of cold-water corals was restricted to depths shallower than 500 m, while coral and sponge gardens, dominated by hexactinellids, were documented to depths of around 1,000 m ( Table 3). The limited presence of VME communities on the remaining four sites sampled (Dives H1436; H1440; H1441; H1442) might be explained in part by the differences in the depths sampled, with ROV transects at these sites conducted at depths greater than 930 m, and the structural complexity of the seafloor. Dives H1436 and H1440, were conducted at seamounts with a much deeper, less topographically complex slope to the summit, while dives H1441 and H1442, were conducted on a predominantly sediment-covered seafloor, and only experienced a depth change of 66 m and 72 m respectively (Supplementary material I).
Our systematic analysis of ROV imagery documented a total of 70 benthic megafauna morphospecies (or morphospecies groups) across the locations surveyed (Supplementary Material II). The most speciose phyla observed were cnidarians (n = 31), echinoderms (n = 14), and arthropods (n = 9). Morphospecies of sponges, molluscs, annelida and tunicates were also observed ( Fig. 1). In order to compare richness across dive locations www.nature.com/scientificreports/ and depth, morphospecies richness was derived from the number of the species present in each 100 m depth band. Analyses revealed that Dive H1435 in the Far north region of the GMR displayed the highest values in species richness within the 700-1,400 m depth strata, with a peak of 37 different morphospecies recorded at 800 m (Fig. 3). The lowest species richness was recorded at the deeper locations, with a minimum richness of four morphospecies recorded at 1,400-1,700 m on Dive H1436, also in the far north region of the GMR (Fig. 3). The detailed species inventory from physical samples and spatial and depth related biodiversity patterns presented here provides one of the first comprehensive characterizations within the TEP, a marine ecoregion of South America where benthic marine biodiversity is presently poorly known and underestimated 45 . It also contributes to the global effort to publish deep-sea biodiversity inventories, a step critical to identify VME indicator species and communities in an effort to develop long-term conservation management plans for these areas on both national and regional scales. To reach these goals, there is a need to expand the taxonomic coverage of collections and compile this information from regional and global datasets 46 . In this study, we had the support and collaboration of taxonomic experts for each phylum from all over the world; nevertheless, taxonomy is still a major constraint on such studies as there is a general shortage of taxonomic expertise for marine biodiversity, particularly for the TEP 45 .
From the image analysis, we present morphospecies present on young volcanic formations (0.1-< 1 Ma) around the Darwin and Wolf lineament in the north of the GMR (H1435, H1436 and H1440), and one location relatively close (< 100 km) to the Galapagos Hotspot to the west of the archipelago 47 (H1443; Fig. 4). We only recorded major differences in morphospecies richness between shallow depth strata (200-700 m) from North and Southeast regions (Fig. 3), which could suggest the presence of different biogeographical areas within the Archipelago, as it has been proposed for shallow marine communities 39 . However, given the limited number of locations surveyed at each region, further data is required to determine whether this biogeographic division applies to deep-sea communities. In addition, given the movement of the Nazca plate east towards the South America mainland, the islands that are formed in the Galapagos hotspot travel east, subsiding to eventually become drowned islands 48 . Therefore, the islands and seamounts to the east of the archipelago represent much older geological formations, with some dated to be > 8 Ma 47 . It is thus likely that additional explorations to the east of the archipelago might result in the discovery of distinct populations and additional new taxa. Future deep-sea research directed at investigating the effect of environmental predictors on deep-sea biodiversity is also required to better understand these relationships at the Galapagos Islands.
Although deep-sea communities are rarely included into spatial-planning processes, the Ecuadorian government recently reviewed the zoning process for the Galapagos Marine Reserve and prioritized the protection of seamounts and open water environments 50 . The information presented here contributed to this spatial planning process and promoted the protection of specific areas based on the demonstration of the presence of VME indicator species and communities and unique taxa and behaviours 31,51,52 . In March 2016, a new zoning plan was proposed and fully protected areas were recommended across the reserve to protect seamounts and other deep-sea habitats, including the creation of a 40,000 km 2 marine sanctuary to protect the three seamounts Table 1. Summary of the ROV dive transects, sampling events, presence of vulnerable marine ecosystems and environmental parameters within the Galapagos Marine Reserve (GMR) during expedition NA064. ( † )value in brackets represents number of biological samples collected (*) value in brackets represents percentage of biological samples that are either undescribed or new to science from each dive location. ( + )Indicator species are those listed in the VME criteria under the methods section (octocoral, hexacorals, scleractinians, and erect sponges) but do not meet the required abundances. www.nature.com/scientificreports/ we surveyed around Darwin and Wolf islands, a hotspot for pelagic species 10,53,54 . However, due to continuous political pressure from the Galapagos artisanal fishing sector, this zoning plan is still under re-evaluation 55 . The discovery of these diverse deep-sea biological communities highlights yet again the ecological significance of the marine ecosystems of the Galapagos Islands. This deep-sea world that Darwin never saw represents a unique and pristine environment, particularly considering the historical absence of destructive human practices known to have catastrophic impacts upon fragile communities 56,57 , such as bottom trawling or deep-sea mining; and their current conservation status and level of protection granted by the Galapagos Marine Reserve. Under the present climate crisis, that will have major effects across the world's marine ecosystems 58 , including the deep sea 59-61 , the Galapagos Islands represent an ideal location from which to monitor future changes in this poorly studied and understood biome.

ROV surveys and sample collection. Exploration of the seafloor was carried out using the two-body
Remotely Operated Vehicle (ROV) system Argus and Hercules, each rated for 4 km water depth. Video and still images of the sites were acquired using Insite Pacific Zeus Plus HD color video cameras on both vehicles, each equipped with a 10 × mechanical zoom lens. Biological specimens were collected using the ROV manipulator arm and placed on sample boxes aboard Hercules ROV for recovery. www.nature.com/scientificreports/ www.nature.com/scientificreports/ We conducted a total of six exploratory dives to the north, west, and central part of the Galapagos Archipelago (Fig. 4, Table 1). In general terms, sampling transects began at the base of each feature explored and we conducted a general upslope transect, following sonar and visual features along the way. Dives H1435, H1436 and H1440 explored three conical seamounts around the most northern islands of the archipelago, Darwin and Wolf. Dives H1441 and H1442 explored the abyssal plain located to the west of Fernandina and adjacent to the Galapagos. The last dive, H1443, explored a shallower volcanic cone, with two horseshoe-shaped craters, located between Santiago and Isabela islands in the central part of the Archipelago (Fig. 4, Table 1).
During each dive, biological samples were collected opportunistically using the ROV's arms or suction sampler, with the aim of collecting a representative sample of specimens, but with a sampling bias towards large and conspicuous megafauna and their associates. Sampling involved the collection of individual organisms (or part of them) with their associated epibionts, as well as collecting rocks with organisms attached to them. When the ROV was back on deck, individual organisms were separated and photographed. Specimens were then sub-sampled  www.nature.com/scientificreports/ and preserved following standardized protocols 62 and in accordance with subsequent morphological and genetic analysis. Scleractinian corals were not included in the taxonomic analysis due to permit restrictions.
Morphological and genetic identification of collected specimens. Whole specimens or subsamples were sent to renowned experts on specific taxa for morphological and/or genetic identification. Most specimens were identified by morphological traits to the lowest taxonomic level. Additionally, some Annelida, Arthropoda, Mollusca and Echinodermata tissue sub-samples were sequenced for genetic characterization. For the Annelida the standard Folmer (M1-M6) region (~ 650 bp) of the mitochondrial cytochrome oxidase subunit 1 gene (mtDNA COI) was sequenced 63 . For the Echinodermata, a hybrid enrichment phylogenetic method was used to obtain sequences from 415 housekeeping genes (285 kbp) 64 . Accepted species names were verified using WoRMS (World Register of Marine Species, 2017) and WoRDSS (World Register of Deep-sea Species, 2017). Geographical distribution of identified taxa was investigated using the OBIS (Ocean Biogeographic Information System). Voucher specimens will be stored at the Charles Darwin Foundation Galapagos Biodiversity Collection. All taxonomic identifications and metadata have been formatted to DarwinCore standard 65 . Presence of vulnerable marine ecosystems and imagery analysis. Due to the volume of nonquantatitive data that had been recorded from the Hercules ROV during the NA064 expedition, it was decided to analyse a series of still images rather than the full video sequence for each dive. Video files were converted to image sequences, with one frame extracted every 5 s using QuicktimePro 7. As the speed of the ROV varied over the duration of the full transect, an image was selected every 10 m for analysis rather than a set time interval. To achieve this, the ROV USBL navigation was cleaned and smoothed in ArcGIS (version 10.3) using editing and cartography toolboxes and split into 10 m segments. A shapefile for the mid-point of each 10 m segment was then generated. Using the join and relate functions in ArcGIS, the 5-s image frame (co-registered with the smoothed USBL positional data) that was in closest proximity to the 10 m mid-point, was selected for further analysis. Video data recorded during sampling events, were removed from the transect data. Dives H1441 and H1442 that were conducted on deep abyssal plains were excluded from the morphospecies analysis given the homogeneity of the terrain surveyed.
Image analysis was conducted using the BIIGLE 2.0 software 66 by three analysts. To minimise bias, each analyst was randomly assigned a third of the images from each dive to analyse. To account for the variation in image quality (i.e. blue water, blurred images, variation in field of view) but to be able to obtain the maximum amount of viable data, all images were annotated using the following label trees; 1) image quality; 2) dominant www.nature.com/scientificreports/ and secondary substrate type; 3) presence of habitat-forming fauna (or VMEs) such as sponges, corals or bivalves; or large aggregations of benthic fauna such as brittlestars or featherstars within the field of view of the oblique high-definition camera and finally; 4) the presence of morphospecies (> 3 cm in size) identified to the lowest taxonomic level possible. Only images of sufficient quality were analysed. For the purpose of this study VME communities were defined as follows; 1) coral gardens were defined as the presence of > 5 colonies of octocorals and/or hexacorals (not including scleractinians) of the same or different morphospecies; 2) sponge gardens were defined as the presence of > 5 colonies of erect (not encrusting) of the same or different morphospecies and; 3) cold-water coral communities were defined as the presence of 3D habitat-forming scleractinian colonies. Scaling lasers set 10 cm were used to opportunistically to measure biota or geological formations. However, as these were not used consistently for the duration of the dive and owing to the oblique camera angle, densities of morphospecies have not been calculated. To ensure consistency between analysts, any conflicts in morphospecies identification or VME community identification were resolved using the "largo" re-evaluation function in BIIGLE. A total of 2077 individual images were analysed across the five ROV dives. A species matrix was created by enumerating taxonomically distinct organisms in each image as present or absent over each 100 m depth band. Given the uncertainties of identifying morphospecies from ROV imagery we adopted best practices in analyses to provide the best possible yet conservative identifications, and where appropriate, morphospecies have been grouped at higher taxonomic levels where identifications were uncertain. Indeterminate taxa that could not be confidently distinguished or identified to phylum were excluded from the analyses.

Data availability
Data cannot be shared publicly because of Galapagos National Park research permit conditions. A previous authorization from the Galapagos National Park Directorate is required for further use of this data. Access to data can be requested via this email address at the Galapagos National Park Directorate Applied Research Department: investigacion@galapagos.gob.ec.