First visual occurrence data for deep-sea cnidarians in the South-western Colombian Caribbean

Abstract Background Attention to the deep-sea environment has increased dramatically in the last decade due to the rising interest in natural resource exploitation. Although Colombia holds a large submerged territory, knowledge of the seabed and its biodiversity beyond 1,000 m depth is very limited. During 2015–2017, Anadarko Colombia Company (ACC) carried out hydrocarbon exploratory activities in the South-western Colombian Caribbean, at depths between 375 m and 2,565 m. New information Capitalising on available data resources from these activities, several cnidarian species were observed in ROV and towed camera surveys. We analysed over nine hours of video and 5,066 still images from these surveys, identifying organisms to the lowest possible taxonomic level. The images and associated data presented here correspond to 108 observations of deep-sea cnidarians, including seven new records for the Colombian Caribbean. Given the paucity of research and funding to explore the deep-sea in Colombia, the present dataset comprises the largest deep-sea Cnidaria imagery inventory to date for the Colombian Caribbean.


Introduction
Interest in the deep seas has increased dramatically in the last decade, due to the potential presence of natural resources (oil, gas, precious and rare metals, fishery resources, pharmaceuticals, biodiversity etc.). Likewise, there is an increasing concern about the health of this environment, which calls for the acquisition of baseline data (Mengerink et al. 2014). Although Colombia holds a large submerged territory and there has been extensive research on cnidarians from the continental margin and upper slope (Florez andSantodomingo 2010, Santodomingo et al. 2012), knowledge of the seabed below 1,000 m depth is sparse. All previous investigations addressing the characterisation of deep-water ecosystems below 1,000 m have relied on samples taken by box corer or benthic sleds (Polanco F et al. 2017), but no in-situ visual confirmations have been provided.
Visual observation of the seabed and its inhabitants is possible using vessels and submersible equipment that can be remote (ROV), autonomous (AUV) or manned. This is a non-invasive in-situ procedure that can reach depths and environments that are normally out of human reach (Huvenne et al. 2018). Therefore, visual data (photographs and videos) provide valuable information on the geological characteristics, physical structure and biological components of a benthic habitat (Pawson et al. 2015, Zawada et al. 2007). However, this sampling method is limited only to the identification of large organisms, i.e. megafauna, with no evasive behaviour (Zawada et al. 2007). Furthermore, the quality of the images is highly dependent on the equipment and experience and qualifications of the operator (Solem 2017). In addition, in instances such as the identification some soft-bodied cnidarians like anemones and cerianthids, image data provide limited information allowing the identification at high taxonomic ranks only (i.e. family and genus in some cases). Species-level identification for these particular cases requires a detailed description of internal and microanatomical characters (e.g. Carlgren 1949, England 1987, Häussermann 2004, Rodriguez and López-González 2013, all of which requires specimen sampling and examination. Video and still photographs are useful for documenting biological and ecological information, species identification, public outreach and scientific publications (Etnoyer et al. 2006). It provides a useful resource that can be made readily available to other scientists, as it was in this case. However, the use of images for organism identification may impose some technical limitations, in particular regarding the quality of the images. High-quality images depend on the camera used during the survey (digital quality, the angle of the setup and zoom), water transparency conditions, the lighting setup (type and angle of the lights) and the file format used to store the image data. Quality also depends on the skill of the engineer who operates the equipment and his/her expertise on the observed organisms. This will determine whether essential morphological features for accurate identification of the organisms (diagnostic characters) could be targeted for close-ups stills. Finally, for some groups of organisms (especially those relying on internal characters for identification, e.g. sea anemones), this type of data provides limited information for reliable identifications (even at high taxonomic levels) and detailed taxonomic descriptions, requiring sampling and morphological observations in order to provide species-level identifications. Nonetheless, such "first-pass" biodiversity surveys provide an invaluable source of information, especially from previously unexplored environments and constitute the foundation for further, more thorough initiatives.
Several types of surveys are essential for the deep-sea hydrocarbon exploratory activities in Colombia. Exploratory drilling activities normally include ROV video surveys before and after drilling. These surveys are critical to evaluate the seabed conditions around the well to minimise potential impacts to sensitive ecosystems (e.g. chemosynthetic communities, deep-sea coral reefs) or archaeological sites. These vehicles are also essential to carry out inspections of the subsea infrastructure to keep the integrity of the well (Gates 2016, Tena 2011. In addition to ROV, detailed characterisation of seabed features and fauna can be obtained from towed camera surveys if needed. Given the paucity of research and funding to explore the deep-sea in Colombia, the present dataset comprises the largest deep-sea Cnidaria imagery inventory to date for the deep sea of Colombian Caribbean. Cnidarians (jellyfish, corals, sea anemones, amongst others) are one of the most ancient invertebrate groups that keeps a simple body structure with two cell layers and a blind gut , Daly et al. 2007). They are found in all aquatic (marine and freshwater) environments, being more diverse in marine habitats. Although simple in body organisation, cnidarians have evolved as specialised carnivores, catching their prey aided by cnidocytes, which are specialised cells with stinging structures, a phylum-defining trait (Technau and Steele 2011). In oceanic waters, cnidarians can be found in nearly all ecosystems from shallow waters to abyssal depths and from polar regions to tropical latitudes. Benthic cnidarians, such as corals colonising both soft and hard substrates, are efficient suspension feeders that provide a tridimensional habitat in the deep-sea (Etnoyer and Morgan 2005), which sustains high numbers of associated invertebrates (Buhl-Mortensen and Mortensen 2005). Cnidarians, particularly anthozoans, are now recognised First visual occurrence data for deep-sea cnidarians in the South-western ... as major contributors of biogenic environments in the deep-sea (Hourigan et al. 2017). In addition, even the rarest types of pelagic or benthic cnidarians are also found in the deepsea (Osborn et al. 2007, Miranda et al. 2018. Given the low density of deep-sea fauna in general, numerous surveys are usually required to recognise common vs. rare species. Hydrocarbon drilling activities have provided some of the few opportunities to perform extensive observations in the deep-sea (Gates et al. 2017).
During 2015-2017, Anadarko Colombia Company (ACC), a subsidiary of Anadarko Petroleum Corporation, carried out hydrocarbon exploratory activities in the deep sea South-western Colombian Caribbean. The activities included both ROV and towed camera surveys at depths between 375 m and 2,565 m. Capitalising on the availability of the images obtained from these activities, several organisms from different phyla were spotted and identified to the lowest possible taxonomic status. This is the first of a series of datasets reporting visual confirmations on the occurrences of deep-sea organisms, in this case, 108 cnidarians.

Sampling description: ROV surveys
ROV video surveys were performed in a cross pattern. From a central point that could be a transponder (tool for positioning the drilling vessel) or the drilling location, surveys were executed in an 80 metre-long transect with a north trajectory, then an 80 metre-long transect with a south trajectory, an 80 metre-long transect with an east trajectory and finally an 80 metre-long transect with a west trajectory (Fig. 1). Video-transects included soft bottom images, where only one species of a pelagic cnidarian was encountered. Videos were taken before and after drilling for all the exploratory wells, or even during drilling for two of the wells.

Towed camera surveys
Survey areas of interest (i.e. Chamana, Chamai, Yaduli, Cana Norte and Cawa) were assessed using 25 towed camera transects. These transects registered seafloor features, during 3-hour surveys, taking still images every 20 seconds. Benthic and pelagic megafauna specimens were recorded in the images.
Quality control: Videos were analysed twice for the presence of cnidarians by two different experts. This methodology allowed us to ensure all cnidarians were registered. Still images were also analysed twice by the same expert, who did preliminary identification to phylum and class. All cnidarian images were then identified to the lowest possible taxonomic level. When in doubt about the identification, additional experts were contacted and the images were sent to them in jpg format for towed camera photographs and as a snapshot (also in jpg format) for ROV videos.
Given that the acquisition of the images used in this paper was carried out for other purposes and objectives, they do not have the best quality for species identification. Nevertheless, the images were useful, and represent the first visual confirmation of these deep-sea organisms for the Colombian Caribbean.
Step description: For ROV surveys, we analysed a total of 48 video transects (duration: 9 h 9 min 26 sec), looking for benthic and pelagic cnidarians. We took snapshots of each cnidarian, registered coordinates and depth and identified them to the lowest possible taxonomic level. On the other hand, for towed camera images, we analysed a total of 5,066 photographs, looking for benthic and pelagic cnidarians. For each cnidarian, coordinates and depth were registered and identified to the lowest possible taxonomic level. We cropped the photographs to include only the organism and performed image correction to reduce the bluish colour cast with the Auto Tone function in Adobe Photoshop CC 2018 (Fig. 2).
Using photographs and video footage, we highlighted the occurrences for Cnidaria here.

Taxonomic coverage
Description: The data presented here correspond to 108 occurrences of deep-sea cnidarians from the South-western Colombian Caribbean, spotted over soft bottoms at depths between 375 to 2,565 m. The dataset contains the original data of depth, geographical coordinates, date and hour of the event, for each image that registered a cnidarian. Additional information includes the methodology used to obtain the images (see Methods section), taxonomic identification to the lowest possible taxonomic level, the name of the expert who identified the organism and the number of individuals of the species per image. The dataset also contains an extension with links to the images supporting the occurrences. Based on the 101 occurrences from towed camera surveys, we registered eight new reports of cnidarians for the Colombian Caribbean. The new records for the area comprised two sea anemones, one zoanthid, one corallimorpharian, one octocoral, one black coral (Fig. 4) and two jellyfish species (Fig. 5).
The genus Corallimorphus has a total of six valid species ( (Briggs et al. 1996). This last study reports the presence of P. michaelsarsi, effectively expanding the distribution range for the species, currently limited to the East and Mid Atlantic Ocean (Molodtsova et al. 2008). Unfortunately, this report does not provide enough information regarding the identification methodology employed by the authors in order to determine this species identity (i.e. geographical proximity, detailed morphological description), thus we cannot evaluate the validity of this record.
Anemones, belonging to the genus Adamsia, are known for holding symbiotic relationships with hermit crabs, most commonly with the species Sympagurus pictus (Daly et al. 2004), but reports also exist for the species Parapagurus pilosimanus and the gastropods of the genus Oocorys (Ammons and Daly 2008). There are a total of 6 valid species (Fautin 2016) with A. obvolva being the only species reported in the region, specifically in several localities from the Gulf of Mexico (Daly et al. 2004, Ammons andDaly 2008). The bathymetric range of the specimens reported here (Table 1) is larger than the range reported for A. obvolva (300 to 800 m). The records, presented here, would correspond to an expansion of the species' geographic and bathymetric ranges; however, it was not possible to confirm its identity. Even the examination of diagnostic external anatomical characters, such as the presence of cinclides (small insertions occurring in the column of the anemone), requires detailed observation under the stereomicroscope.
Other hexacorallians, described as new reports for the Colombian Caribbean, include Bathypathes cf. patula and Epizoanthis cf. stellaris. Bathypathes patula is also a widespread, cosmopolitan species that has been described from the Pacific, the Atlantic, the Indian Ocean, the Gulf of México and Puerto Rico (Molodtsova 2006, Horowitz et al. 2018. It belongs to the Antipatharian genus Bathypathes Brook, 1889 (Anthozoa: a b Figure 5.
New reports of cnidarian medusae for the Colombian Caribbean.

a: Crossota millsae b: Voragonema pedunculata
Hexacorallia) which currently holds 19 species (Molodtsova and Opresko 2019). Bathypathes patula is characterised by a wide vertical distribution that ranges from 100 m to 5500 m in depth (Molodtsova 2006). On the other hand, the zoanthid genus Epizoanthus Gray, 1867 (Anthozoa: Hexacorallia) is a rich genus with 105 species and a global distribution Sinniger 2019, Kise andReimer 2016). Particularly, Epizoanthis stellaris Hertwig, 1888 has been reported to grow over the siliceous spicules of glass sponges (Beaulieu 2001), as also seen in this study.
The pennatulacean genus Umbellula comprises 13 species (Cordeiro et al. 2019) and belongs to the monogeneric family Umbellulidae (Anthozoa: Octocorallia). This genus has a widespread to nearly worldwide distribution and is the deep-sea pennatulacean genus with the widest vertical distribution reaching depths of 6,000 m (Williams 2011). The most common species is U. lindahi that has a cosmopolitan distribution in all of the world's oceans (Tyler et al. 1995).
Finally, the two hydromedusae from the Rhopalonematidae family, Crossota millsae and Voragonema pedunculata (previously know as Benthocodon pedunculata), are bathypelagic and often found in groups drifting above the bottom; this is the reason why they are also described as benthopelagic. These two species have been reported in the Western Atlantic (the Bahamas and Dry Tortugas) (Larson et al. 1991, Larson et al. 1992 and the Gulf of Mexico (Valentine and Benfield 2013  The status of the use of the scientificName as a label for a taxon