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

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