Environmental DNA illuminates the darkness of mesophotic assemblages of fishes from West Indian Ocean

Abstract

The advent of environmental DNA (eDNA) represents a pivotal advancement in probing mesophotic communities, offering a non-intrusive avenue for studying marine biodiversity at greater depths. Using this approach, we characterized and compared the mesophotic reef fish assemblages of two West Indian Ocean islands, Mayotte and La Réunion, which are characterized by different geo-morphological contexts. The sequences obtained were assigned taxonomically and grouped into molecular operational taxonomic units to perform richness and beta diversity analyses. The functional diversity of the assemblages was assessed using five traits, enabling each sequence to be assigned to a functional entity corresponding to a specific trait combination. On both islands, the fish assemblages exhibited a comparable level of taxonomic and functional richness, consisting mainly of piscivorous and planktivorous fishes. These assemblages were primarily composed of families such as Serranidae, consistent with expectations for the mesophotic zone. However, beta diversity analyses revealed that the two islands exhibited different taxonomic and functional structures. For example, La Réunion was characterized by a greater importance of the Muraenidae, while Mayotte displayed a higher representation of families strongly associated with coral reefs (e.g., Zanclidae or Malacanthidae). These results suggest that depth-related forcing constrains fish assemblages to some extent, but that differences in structure remains determined by other, more local factors, likely linked to the geo-morphological contexts of the islands and their habitats. This study also revealed that eDNA is a promising method for studying difficult-to-observe taxa, such as moray eels or lanternfish, and may also be relevant for monitoring species depth ranges. Overall, results highlighted the "local scale", "functionally integrative" and "temporally integrative" characteristics of eDNA for studying mesophotic reef fish assemblages. However, this study also highlights the limitations of reference DNA databases, pointing to future prospects for fully exploiting the potential of eDNA approaches in the mesophotic zones of the Indian Ocean.

Fig 1. Maps of environmental DNA samples (black circles).

(A) Depths of 68-89 meters around Mayotte (November to December 2021). (B) Depths of 80-107 meters around La Réunion (September 2020 to January 2021). Samples IDs are indicated in brackets. Bathymetric data were sourced from GEBCO and data.gouv.fr. The GEBCO gridded bathymetry data is placed in the public domain and was freely download from https://www.gebco.net/data_and_products/gridded_bathymetry_data/. Source: GEBCO Compilation Group (2024) GEBCO 2024 Grid (https://doi.org/10.5285/1c44ce99-0a0d-5f4f-e063-7086abc0ea0f). Figure created with QGIS Desktop 2.18.0 (https://qgis.org/download/).

Fig 2. Comparison of habitat characteristics across islands.

The plot depicts the first two dimensions of the non-metric multidimensional scaling (NMDS) based on Gower distances of environmental habitat variables for each island (n_Mayotte = 10 and n_Réunion = 8). Only environmental factors with significant correlation (p < 0.05) were fitted and plotted on the NMDS. Coral dominance: Coral was the primary component of the benthic community; Meso-community dominance: Sessile organisms other than coral dominant; Mud dominance: Muddy substrate was the main substrate; High slope: Steep reef slope; Low slope: Flat reef slope. Sample origins are color-coded: Mayotte in blue and La Réunion in beige.

Fig 3. Diversity analysis.

The boxplot in A) represents the distribution of the pairwise genetic distances between sequences of the customized MiFish sequence dataset, categorized by taxonomic comparison type. Class: between sequences from different classes; Order: between sequences within the same class but different orders; Family: between sequences within the same order but different families; Genus: between sequences within the same family but different genus; Species: between sequences within the same genus but different species; Intra-species: between sequences within the same species. The red dotted line corresponds to the divergence threshold defined for clustering ZOTUs. In B) curves correspond to the extrapolated rarefaction diversity of ZOTU clusters for each island (incidence-based data) with three orders: richness q = 0 (circle), Shannon diversity q = 1 (triangle) and Simpson diversity q = 2 (square). Horizontal lines represent the asymptotic estimation values for each diversity order.

Fig 4. Taxonomic composition of fish assemblages.

Barplots depict the cumulative occurrence of ZOTUs assigned to each family per sample. The font size of taxa names is proportional to their cumulative occurrence across the 18 samples. Informative taxa significantly associated with one of the islands are indicated by the color code corresponding to the island: Mayotte in blue and La Réunion in beige.

Fig 5. Functional trait composition of fish assemblages.

Each barplot represents the relative occurrence of trait values by island, based on the ZOTU table for the five functional traits analysed in this study. Barplots for Mayotte are indicated in blue, while those for La Réunion are in beige.

Fig 6. Taxonomic and functional structure of fish assemblages.

The graph represents the first two axes of non-metric multidimensional scaling (NMDS) of fish taxonomic assemblages using the Jaccard and Bray-Curtis indices in A) and B), respectively, and fish functional assemblages using the Jaccard and Bray-Curtis indices in C) and D). Sample vectors from island barycenter of each sample and island convex hull are displayed based on their island origin, with Mayotte in blue and La Réunion in beige. Informative taxa and functional entities (FEs) are plotted according to their NMDS scores, with island association indicated by the island color code. The goodness of fit of the two-dimension ordination are indicated for each NMDS by the 2D stress values.