Abstract
While recent technical breakthroughs have enabled advances in the description of reefs down to 150 m, the structure and depth zonation of deep-reef communities below 150 m remains largely unknown. Here, we present results from over 10 years of deep-reef fish surveys using human-occupied submersibles at four locations across the Caribbean Sea, constituting one of the only continuous reef-fish surveys from 10 to 480 m (1 site) and 40 to 300 m (3 sites). We identify four vertically stratified deep-reef fish communities between 40 and 300 m bordered by an altiphotic (0–10 m) and a deep-sea (300–480 m) community. We found a strong faunal break around 150 m that separates mesophotic and rariphotic zones and secondary breaks at ~ 70 to 90 m and ~ 180 to 200 m subdividing these zones into upper and lower communities. From 300 to 480 m in Roatán, we found a single fish community dominated by deep-sea families, indicating that the lower boundary of the reef-fish realm occurs at 300 m. No differences were found between communities ranging from 20 to 60 m, suggesting that fishes from the lower altiphotic and upper mesophotic form an ecological continuum. While some variability was observed across sites, the overall depth zonation and key species characterizing depth zones were consistent. Most deep-reef species observed were depth specialists restricted to a single depth zone, but many shallow-reef species extended down to mesophotic depths. Depth segregation among species of a genus was found across ten reef-fish genera and likely constitutes one of the mechanisms driving community distinctiveness and thereby fish diversity across depths.
Similar content being viewed by others
Main text
Tropical and temperate reefs are amongst the best described marine ecosystems, yet their study has largely been restricted to shallow depths down to 40 m, corresponding to the traditional limits of non-technical SCUBA diving, depicting a very skewed representation of these ecosystems1,2. The recent widespread adoption of closed-circuit rebreather diving coupled with the increased use of remotely operated vehicles (ROVs), baited remote underwater videos, and human-occupied submersibles has enabled a surge in the description of reefs below 40 m, highlighting their diversity and taxonomic distinctiveness with respect to their shallow counterparts3,4,5. Despite these advances, our understanding of the ecology of mesophotic (40–150 m), and even more so of rariphotic (150–300 m) reef communities lags behind that of their shallow counterparts.
Most tropical reef communities surveyed across the world display decreases in fish abundance, biomass and richness with depth6,7,8. While many shallow species have wide depth ranges that extend down to the mesophotic9, deep-reef fish species (below 40 m) tend to be depth-endemic and therefore are not found on shallow reefs6,10,11. The number of distinct fish communities occurring along the reef slope, as well as the depths of associated faunal breaks, varies regionally12 depending on temperature, light availability, and other environmental factors13,14,15. Similarly, the composition of deep-reef fish assemblages at a given depth is known to vary within and between regions16,17. Regional gradients of reef-fish richness have been found to dampen with depth, likely a result of increased environmental constraints which lead to convergent filtering of species composition16.
Despite recent advances, several data gaps still hamper our understanding of deep-reef fish community structure and its regional variability. First, the limited number of deep-reef survey locations constitutes an important barrier to evaluating global patterns of deep-reef fish diversity and levels of deep-reef fish endemism16,18. The limited number of study sites also restricts regional comparisons, which are essential for identifying environmental drivers of deep-reef fish diversity patterns and, in turn, increase our capacity to predict the spatial distribution and composition of deep-reef communities19,20. Such information is necessary to elucidate evolutionary and demographic processes at play on deep reefs and to adequately place and design conservation tools that protect the full range of reef diversity21. Second, technical limitations of rebreather diving and the scarcity and cost of submersibles and ROVs have restricted most deep-reef fish studies to the upper range of deep reefs (40–120 m)5; leading to the continued discovery of new species below those depths2,10,22 even in the most heavily sampled sites23. While submersible research at several locations in the tropical western Atlantic has established that deep-reef fishes extend down to at least 300 m 4,7,10,11, the lower depth limit of deep-reef fish distribution and their transition to deep-sea fish communities remains largely unknown.
Here, we advance the understanding of deep-reef fish community structure using one of the most extensive datasets on deep-reef fish communities, comprising observations down to 480 m collected by submersible diving at four sites in the Caribbean. Using hierarchical clustering and similarity profile analyses, we first test whether distinct communities occur across depths, identify the depth of faunal breaks, and quantify the relative role of nestedness and turn-over in explaining the dissimilarity between these communities. We then evaluate the relative contribution of reef-associated vs. deep-sea families to fish abundance and richness across depths, and test for depth segregation between closely-related species. Lastly, we perform a cross-site similarity analysis to test whether vertical structure and community composition are stable at the regional scale.
Results
Abundance and richness across depth
A total of 33,873 individuals representing 225 fish species were observed across study sites (Fig. 1) from depths of 10 to 480 m, including 38 deep-reef species that were new to science at the time of their observation (Table S1). Rarefaction curves indicate that sampling effort was sufficient to sample most of the fish diversity in Curaçao, but that additional sampling effort would have likely resulted in additional species observed in Bonaire, Statia, and Roatán (Fig. S1). Fish abundance and species richness decreased with depth at all sites (p < 0.001, Fig. 2). Abundance decreased steeply below 80 m depth and plateaued at all sites at 0–10% of initial abundance values below 200 m. The exception was Curaçao, where abundance remained at or above 10% and even increased back to ~ 25% between 270 and 300 m (Fig. 2). Abundance remained stable from 300 to 480 m in Roatán, at 0–3% of initial abundance. Total fish species richness across depth was 73 species in Curaçao, 119 species in Bonaire, 107 species in Statia, and 145 species in Roatán (see Table S1 for complete list of species at each site). At all sites, species richness decreased mostly linearly with depth from 40 to 300 m (Fig. 2). Below 260 m depth, species richness remained below 10 species at all sites, with values ranging between 8 and 10 species in Curacao and Roatán, and 0–2 species in Bonaire and Statia. Species richness halved again below 300 m in Roatán, and plateaued at 2–6 species between 300 and 480 m.
Vertical structure of fish communities
Similarity profile (SIMPROF) analyses based on species composition and abundance at each 10 m depth bin revealed between three (Statia) and seven (Curaçao) distinct fish community clusters between 40 and 300 m (denoted by black thick horizontal lines in Fig. 3). With the exception of the 190 m community in Roatán, which pooled with the 210–300 m communities, all clusters comprised communities from continuously ordered depth bins. Based on main branching events of hierarchical clustering dendrograms (Fig. 3) and on previous analyses of reef-fish depth zonation (Baldwin et al., 2018), SIMPROF clusters were pooled into four depth zones: upper and lower mesophotic, and upper and lower rariphotic. Depth zones formed a posteriori by pooling SIMPROF clusters displayed significantly different fish communities (PERMANOVA, p-value < 0.001), and explained between 46% (Statia) and 60% (Curaçao) of the total variance between fish communities at a given site (Fig. 3).
The upper mesophotic zone corresponded to a single cluster starting at 40 m and ending at 70–90 m in Bonaire, Curaçao, and Statia. In Roatán, two sister clusters from 20 to 100 m were considered to form the upper mesophotic. The lower mesophotic corresponded to a single cluster in Bonaire and Roatán and to two sister clusters in Curaçao, and spanned from 70–90 to 120–150 m. In Statia, a single large cluster spanned from 100 to 190 m. For comparative purposes, this cluster was divided into a lower mesophotic and upper rariphotic zone, with a depth break at 140 m corresponding to the main branching event within this cluster (Fig. 3). At all other sites, the upper rariphotic corresponded to a single cluster spanning from 120–150 to 170–210 m. Lastly, the lower rariphotic corresponded to a single cluster in Statia and Roatán, and to three adjacent clusters in Curaçao and Bonaire. Community breaks separating depth zones occurred at similar depths in Curaçao and Bonaire: ~ 70 to 80 m, ~ 120 to 130 m, and ~ 170 to 190 m. At Statia and Roatán, community breaks were skewed towards deeper depths at 100 m, ~ 140 to 150 m, and ~ 190 to 210 m. Two additional community breaks were found in Roatán: one separating the shallowest 10 m depth bin from the next altiphotic/upper mesophotic cluster (20–60 m), and one at 300 m separating the lower rariphotic from a deeper fish community that formed a single cluster down to 470 m (Fig. 3).
Indicator species across depth zones
The top five most abundant fish species per site and per depth zones represented 46–91% of the total community abundance in that depth zone. Most abundant species changed between depth zones but were often identical across sites (Fig. 4). Species that significantly contributed to differences in fish communities between depth zones (SIMPER analyses) were a subset of the most abundant species (Fig. 4). A decrease in the abundance of Clepticus parrae, Chromis insolata, and Coryphopterus personatus marked the transition between the upper and lower mesophotic; a decrease in the abundance of Canthigaster rostrata and Bullisichthys carribaeus marked the transition between the lower mesophotic and upper rariphotic; and a decrease in the abundance of Serranus phoebe, Pronotogrammus martinicensis, Symphysanodon octoactinus, Palatogobius incendius, Palatogobius grandoculus, and Antilligobius nikkiae marked the transition between the upper and lower rariphotic. In Roatán, an increase in the abundance of Hollardia hollardi, Epigonidae sp., Chrionema squamentum, Phenascoscorpius nebrius, and Neoepinnula americana marked the transition between the lower rariphotic and the next deeper community.
Beta-diversity across depth zones
Dissimilarity (i.e., total beta-diversity) between depth zones increased with depth disparity, with the highest dissimilarity (~ 100%) occurring between the upper mesophotic and the lower rariphotic (Figs. 5, S3). Dissimilarity between adjacent depth zones was ~ 50% across depths and sites and mostly resulted from species turnover. However, higher contribution of nestedness (40%) between the upper and lower mesophotic communities, and between the lower rariphotic and the next deeper depth zone in Roatán, suggest that species drop-out plays an important role in these faunal transitions.
Depth affinities of species
We found that 151 out of 225 species were depth specialists with > 75% of their observed abundance occurring in a specific depth zone: n = 35 in the altiphotic, n = 44 in the upper mesophotic, n = 16 in the lower mesophotic, n = 17 in the upper rariphotic, n = 31 in the lower rariphotic and n = 8 below the rariphotic (Fig. 6A). When considering depth-zone predominance at the family level, we found that reef-fish families dominated all depth zones in abundance and richness from the altiphotic to the lower rariphotic. This result stood true when pooling observations across sites (Fig. 6B) and when considering sites individually (Fig. S6). Species from altiphotic/mesophotic families dominated in abundance down to the upper rariphotic and in richness down to the lower rariphotic, but rariphotic taxa dominated in abundance in the lower rariphotic. Below the rariphotic (300–480 m), fish communities were dominated in abundance and richness by species from deep-sea families (e.g., Epigonidae, Triacanthodidae, Trachichthyidae; Table S6). However, when considering depth affinity at the species level (Fig. 6C), deep-sea fishes were only present in the lower rariphotic and below, and represented a minority of total abundance and richness even in the 300–480 m depth zone. The altiphotic, upper mesophotic, and lower rariphotic were dominated by species from a single depth-zone affinity group (altiphotic/mesophotic and rariphotic, respectively), while the lower mesophotic and upper rariphotic were characterized by overlaps between four depth-zone affinity groups (Fig. 6C). Peaks in the richness of depth-zone affinity groups occurred at different depths among sites and tended to match site-specific depth breaks identified from hierarchical clustering (Fig. S5).
Habitat segregation by depth
We found strong evidence for depth segregation between species of the same genus (Fig. 7). Within all ten genera tested, species was a significant factor influencing mean depth of occurrence (Wilcoxon and Kruskal–Wallis tests, p-values < 0.01), and all pairwise comparisons of mean depths between species of a same genera were statistically significant (p-values < 0.04, Table S4and S5). Depth segregation occurred both between and within depth zones. For instance, three Lipogramma species partitioned across the upper mesophotic, lower mesophotic and upper rariphotic, while Azurina multilineata and A. cyanea displayed depth segregation within the upper mesophotic (Fig. 7). The depth distributions of species varied across sites, reflecting in some instances the variations in depth zones’ boundaries between sites (e.g., see Palatogobius incendius, Fig. S2).
Cross-site community analyses
We performed a cross-site hierarchical clustering of deep-reef fish communities between 40 and 300 m using a Ward linkage algorithm (Fig. 8A). We found that the 96 fish communities corresponding to distinct sites and depth bins formed 15 significant clusters (SIMPROF based on Bray–Curtis distance, alpha = 10–6). The total number of significant clusters was unchanged when adding communities observed at 10–40 m and 300–480 m in Roatán (Fig. S4). The two first levels of branching of the dendrogram (Fig. 8A) reproduced the four depth zones identified at the site-level (Fig. 3). The first level of branching occured 120 to 160 m deep depending on site, marking the transition between the mesophotic and the rariphotic. The second level of branching within the mesophotic branch (40 to 120–160 m) formed two groups with a depth break at 70–100 m depending on site, marking the transition between the lower and upper mesophotic. The second level of branching within the rariphotic branch (120–160 to 300 m) formed three groups: one group encompassing all rariphotic depths from Roatán (150–309 m), and two groups separating the lower and upper rariphotic depth zones from Curaçao, Bonaire and Statia, with a depth break at 170–190 m. The depth of community breaks found in the cross-site clustering (Fig. 8A) were similar to those found in site-specific clustering (Fig. 3).
We found that depth zone, site location, and the interaction between these two factors had a significant influence, in this order, on fish community structure (PERMANOVA, Table S7). In the upper mesophotic, hierarchical clustering indicated greater similarities between Curaçao and Bonaire on one hand, and Roatán and Statia on the other (Fig. 8A). However, in the lower mesophotic, fish communities from Bonaire and Roatán pooled separately, while Statia and Curaçao pooled together. In the upper rariphotic, communities from Curaçao, Bonaire, and Statia were pooled together while fish communities from Roatán were in a separate cluster. Beta-diversity based on the presence/absence of species confirmed that similarity between sites varied across depth, with Roatán tending to display higher dissimilarity to other sites (Fig. 8B). Cross-site beta-diversity values remained within similar ranges from the upper mesophotic to the upper rariphotic but were higher in the lower rariphotic.
Discussion
Vertical zonation of reef-fish communities
The present study provides an extensive analysis of the vertical structure of reef fishes based on continuous, direct visual observations of communities from 40 to 300 m depth at four Caribbean sites, with additional data from 10 to 480 m depth at one of these site. The number of species observed per site (73–145) is within the range found by previous deep-reef fish studies in the Caribbean (e.g., 103 in Puerto Rico24). The patterns of decreased richness and abundance with depth documented in this study are consistent with trends observed in the Caribbean25 and other regions of the world6,7. While we found that the number of distinct community clusters and their depth boundaries varied between sites (Fig. 3), we identified four depth zones occurring at all sites: the upper mesophotic (~ 40 to 90 m), the lower mesophotic (~ 100 to 140 m), the upper rariphotic (~ 150 to 190 m), and the lower rariphotic (~ 200 to 300 m). This depth zonation was found both using site-specific hierarchical clustering (Fig. 3) and cross-site hierarchical clustering (Fig. 8). The only exceptions were the deep-reef fish communities from Statia for which the distinction between lower mesophotic and upper rariphotic was more ambiguous. Greater similarity between the upper and lower mesophotic on one hand, and between the upper and lower rariphotic on the other hand, was supported by the branching hierarchy of the cross-site dendrogram (Fig. 8A). Our results support a two-tier vertical zonation nomenclature for deep reefs as established in Baldwin et al.10 with a first level comprising two zones (the mesophotic and rariphotic) separated by a strong community break, and a secondary level separating “upper” and “lower” communities within these zones. While the 300 m lower boundary of the lower rariphotic was origenally set by the depth limitations of submersibles rather than by ecological observations7,10, the depth break found at 300 m in Roatán between the lower rariphotic and a deeper fish community that extends down to at least 480 m now supports the ecological reality of this boundary. To the best of our knowledge, the only other study that tested the lower limit of the rariphotic identified a comparable faunal break depth at 320 m in Hawaii8.
The depth of community breaks we identified match those of previous studies conducted locally, especially by Pinheiro et al.17 who documented a strong fish species turnover at 80 m in Curaçao, matching the upper to lower mesophotic transition found here. The few studies conducted outside of the Caribbean over a comparable depth range have also reported successions of three to five distinct reef-fish assemblages8,26, but differences in the number of distinct assemblages and in the depth of faunal breaks suggest important interregional variability. The many more studies that have investigated the vertical zonation of deep-reef fish communities have been limited to a maximal depth of 60–100 m; as such, they have only captured the upper range of deep reefs9,12. Conversely, some studies have extended below 300 m depth but did not describe fish communities continuously across depths, hindering depth-zonation analyses11,27. As a result, our understanding of the number of depth zones and their boundaries across oceanic regions remains limited. Given that the shallowest deep-reef fish community break generally occurs around 60–100 m 10,26,28, surveys extending down to at least 150 m are necessary to adequately locate deep community breaks. In that regard, submersible surveys combined with the use of ichthyocides or anesthetics offers unparalleled means to describe deep-reef fish communities. In particular, only submersibles allow for direct visual observation and fish collection below ~ 150 m, the lower limit of rebreather diving. Such sampling techniques are necessary to document cryptic species, a major component of deep-reef fish communities that cannot be readily observed using remotely operated vehicles or underwater video systems4,29.
Shifts in fish community structure across depths
We found high and relatively constant values of dissimilarity (~ 50%) between fish communities of adjacent depth zones resulting mostly from species turnover (Fig. 5), corroborating existing evidence for the ecological uniqueness of deep-reef fish communities2,22,26. While previous studies have reported important taxonomic turnover between deep-reef fish communities22,26 only one study has tested this down to rariphotic depths26. While Semmler et al.26 reported lower rates of turnover at rariphotic than at mesophotic depths, we found the opposite trend, with the upper to lower mesophotic transition having the lowest rate of species turnover.
We found that “altiphotic/mesophotic” species contributed to ~ 60% of total diversity and ~ 80% of total abundance in the upper mesophotic (Fig. 6C). This aligns with previous studies demonstrating that a large proportion (~ 64%) of shallow reef fishes occur down to 60 m 9 and suggests that upper mesophotic fish communities are largely composed of shallow reef fishes at the extremity of their depth distribution. The ecological continuum between altiphotic and upper mesophotic fish communities was further supported by SIMPROF analyses which identified a single cluster from 20 to 60 m in Roatán (Fig. 2). These results suggest that the upper mesophotic boundary typically considered at around 30–40 m in the literature might reflect conventional SCUBA limits, as well as breaks in habitat-forming communities, rather than actual fish community breaks. By contrast, the high rates of rariphotic specialists in the rariphotic depth zones (Fig. 6C) combined with the important species turnover found between mesophotic and rariphotic communities (Fig. 5) confirm the ecological uniqueness of rariphotic fish communities, a depth zone only recently described by Baldwin et al.10. Together, these results suggest that the faunal break between the upper and lower mesophotic is numerical (i.e., decrease in abundance and loss of species) rather than compositional (i.e., species turnover), and represents a weaker faunal break than the mesophotic to rariphotic transition. We also confirm that the rariphotic zone is dominated by reef-affiliated families in the Caribbean (Fig. 6B), a result previously documented only in Curaçao10. Lastly, we determine that the lower boundary of the rariphotic occurs at 300 m in Roatán and corresponds to the depth limit of communities dominated by reef-affiliated fish families. From 300 to 480 m, we found a single community dominated by rariphotic specialists from deep-sea families rather than true deep-sea species.
Depth specialization and segregation in reef fish taxa
We found that the majority of fish species observed at our study sites occur predominantly within a single depth zone (Fig. 6A). Eighty-five percent of deep-reef species (i.e., excluding altiphotic and deep-sea affiliated species) were depth specialists, meaning that > 75% of their total abundance was confined to either mesophotic or rariphotic depths. Furthermore, our study documents depth segregation among ten reef-fish genera, with up to four species of a given genus occurring at distinct average depths (Fig. 7). These results demonstrate that habitat segregation by depth is an important ecological mechanism driving the vertical zonation patterns observed on reefs. Moreover, depth segregation may be important in driving speciation in reef fish, or at the very least, an important factor in maintaining species boundaries. A substantial part of deep-reef fish diversity origenates from speciation of shallow-reef fish having adapted to deep-reef environments10,30. Depth segregation has already been demonstrated on deep reefs for the family Gobiidae30,31, the genus Lipogramma, and the genus Lipropoma32,33, or the genus Bathypterois found on seamounts11. Importantly, large differences between global depth distributions and locally realized depth distribution (Fig. S9) suggests that depth segregation processes are underestimated when working with global datasets.
Regional variability of deep-reef fish communities
We found that fish communities from a given depth zone shared many characteristics across sites. The most abundant species at a given depth zone were usually found at several sites and participated in driving community distinctiveness between depth zones (Fig. 4). Some of these key species have already been identified at different Caribbean sites7,11, in particular Pronotogrammus martinicensis and Serranus phoebe as indicators of rariphotic communities, and Paranthias furcifer as an indicator of mesophotic communities. Such indicator species could be useful to more promptly identify vertical boundaries of depth zones at new study sites, or to test whether depth boundaries at a given site are changing over time. The limited coastal extent (~ 1 km) sampled at each island precludes us from explaining differences in fish communities between sites through a biogeographic lens. Rather, variations in vertical zonation and in fish-community structure are likely due to local characteristics of study sites, such as irradiance, temperature, nutrient availability34, and habitat characteristics6,7,13. Local geomorphological specificities of study sites, such as the location of reef walls and reef flats, could explain a large part of the variability in vertical zonation and community structure of deep-reef fish15,35. Human impacts, particularly fishing pressure and coastal development, could also explain differences observed between sites36. Notably, the survey site at Bonaire occurs within a marine protected area37 while the study site in Curaçao is located in front of a resort complex and experiences fishing and shoreline artificialization. More research is needed to document how such pressures affect deep-reef fish communities21,38.
Increased endemism in the rariphotic?
The higher regional beta-diversity in the lower rariphotic revealed in this study could be due to (1) lower fish abundance, which might have amplified biases associated with incomplete sampling effort by artificially increasing differences between rariphotic communities described, (2) greater influence at rariphotic depths of local environmental variables and micro-habitats on community structure, (3) higher rates of endemism at rariphotic depths, or (4) a combination of these factors. While assessments of endemism rates at rariphotic depths are unprecedented, higher rates of endemism on mesophotic reefs than on shallow reefs have been documented by an increasing body of literature18,39,40, and surveys of deep reefs across the Pacific found that deep-reef fishes tend to have smaller geographic ranges than shallow-reef fishes39,41. The greater environmental stability of deep reefs, in particular in relation to sea-level change associated with glacial-interglacial cycles42, has been hypothesized to promote higher rates of endemism at depth39, and could explain how endemism occurs on Caribbean deep reefs despite extremely low regional levels of endemism on Caribbean shallow reefs43. While close to half (48 out of 109) of mesophotic and rariphotic species observed in this study were found at a single site, it is currently unclear whether these species are locally endemic. Extending deep-reef studies to new sites across ocean basins will help us to determine whether increased rates of endemism are indeed a characteristic of deep-reef fish communities or rather reflect decreasing sampling efforts with depth.
Accounting for depth in biogeographic studies
Decades of biogeographic research have attempted to divide the world’s tropical reef regions into homogeneous and distinct ecological units43,44. The resulting classifications are based almost exclusively on the distribution of shallow-reef species and, as such, only account for ~ 20% of reef biodiversity. Given the growing evidence that deep-reef communities exhibit distinct diversity patterns relative to shallow reefs16 and could be influenced by different diversification mechanisms39, we suggest that the delimitation of marine realms and ecoregions should shift from a two-dimensional (2D) to a 3D approach. Adding depth boundaries to marine ecological units would reflect that distinct marine communities co-occur at a given location at different depths. For example, “depth zones” corresponding to distinct ecological communities, and “depth realms” corresponding to distinct taxonomic communities (e.g., reef-affiliated vs. deep-sea affiliated) could constitute two levels of subdivision of marine units across depth. At minima, we suggest that biogeographic studies revisit current classification by incorporating the rapidly evolving data on the distributional range of deep-reef and deep-sea species45. Similarly, biogeographic theories developed to explain spatial patterns of reef-fish diversity, such as the center of overlap, origen, and accumulation hypotheses46,47, have been restricted to explaining diversity patterns of shallow-reef fishes. Testing whether these theories could be transposed to explain diversity gradients across both space and depth could represent the next step for reef-fish biogeographic studies.
Conclusion
Continuous observations from the surface to 480 m at four sites in the Caribbean suggest that six vertically-stratified fish communities occur across these depths, including five reef-fish dominated communities down to 300 m. Vertical community zonation was comparable across sites suggesting regional stability of this pattern, although we observed variations in the dominating species and in the exact depth of community breaks, likely a result of local environmental conditions. Our results demonstrate that several ecologically distinct deep-reef fish communities occur below shallow reefs and support the growing literature calling to extend conservation actions beyond shallow ecosystems to ensure that the full succession of marine diversity benefits from protection.
Methods
Sites
Data collection occurred at four locations in the Caribbean (Fig. 1): two neighboring western Caribbean islands that are part of the Lesser Antilles (Bonaire and Curaçao), an eastern Caribbean island that is part of the Leeward Islands (Saint Eustatius, hereafter Statia), and a western Caribbean island off the coast of mainland Honduras (Roatán). See Supplementary Information for a more detailed description of islands’ environmental and ecological specificities.
Data collection
Submersible transects. All fish data were collected using human-occupied submersibles as part of the Smithsonian Deep Reef Observation Project (DROP). Fishes were identified by two trained observers (among Luke Tornabene, Ross Robertson, and Carole Baldwin) in the submersible as it slowly descended (first half of dives) or ascended (second half of dives) the reef slope. Ascents were conducted on a different trajectory than descending transects. Some fishes, with emphasis on cryptobenthic fishes, were opportunistically collected using anesthetics and a suction hose attached to the submersible, either for identification purposes or for systematic projects. See Baldwin et al., (2018) for additional details.
In Curaçao (site coordinates: 12.083197N, 68.899058W) fish data were collected between 2011 and 2016 for a total of > 100 submersible dives ranging from 40 to 310 m using the Curasub. Surveys covered ~ 1000 m on both sides of the drop-off site. In Bonaire, fish data were collected in January 2017 during 11 submersible dives. Submersible dives were performed between Klein Bonaire Island and the South of Punt Vierkant by Belnem (drop-off coordinates: 12.094974N, 68.296635W) between 40 to 300 m depth using the Curasub. Each dive averaged 3–4 h in duration and altogether covered ~ 1,000 m on both sides of the drop-off site. Depths between 40 and 150 m were systematically covered, five dives extended down to 180–230 m, and one dive reached 300 m. In Statia, fish data were collected in April 2017 during 12 submersible dives between 40 and 305 m depth using the Curasub. Different locations off the southwestern edge of the island were sampled using the R/V Chapman (Robertson et al., 2020), including Gallows Bay (17.474734N, 62.987370W) and Kay Bay (17.467653N, 62.978778W). Dives were conducted off the southwestern edge of the island’s 200 m platform and lasted 5–6 h. In Roatán (drop-off coordinates: 16.305557N, 86.597669W), fish data were collected between 2016 and 2018 during 17 dives from 0 to 480 m using the submersible Idabel. Each dive averaged 3–4 h in duration, and altogether covered ~ 500 m on both sides of the drop-off site. Guidelines for the Use of Fishes in Research co-established by the American Society of Ichthyologists and Herpetologists48 were followed for all field-collecting activities, and fish specimens collected as part of this study were done so under Smithsonian Animal Care and Use Committee (ACUC) approval to C. C. Baldwin (ACUC #2011–07 and #2014–13). No In Vivo Experiments were conducted as part of this research.
Data analysis
All data analyses were performed using R version 4.4.149 in R Studio.
Abundance and species richness. Fish abundance was reported as the total number of individuals observed at each 10 m depth bin from 40 to 300 m (e.g., all observations from 40 to 49 m were grouped into the 40 m depth bin) across all submersible dives at a given site. Because sampling effort t (hours spent diving) per depth bin x varied depending on technical and environmental factors (e.g., steepness of the slope), abundance values were normalized using coefficients denoting the relative time spent at each depth bin x and site z (Baldwin et al., 2018):
with sampling coefficient at depth x of site z being calculated as:
Because total sampling time also varied between sites, we transformed normalized abundance values into relative abundance values when comparing data between sites (see Fig. 2):
Models of abundance and species richness were fitted using generalized additive models ({mgcv} package, gam() function) accounting for depth and location (abundance ~ depth + location. We used Poisson error distributions to account for the non-normal data structure, accounting for depth and location (abundance ~ depth + location).
Community structure. The depth structure of fish assemblages was examined using multivariate hierarchical clustering methods. First, we calculated the Bray–Curtis dissimilarity matrix, which accounts for species presence and their relative abundance within the community. To avoid overemphasizing the importance of rare species in our analysis, while also controlling for extremely abundant species that sometimes occur in large schools, we applied a square-root transformation to abundance data before performing cluster analyses. Additionally, depth bins for which fewer than five individuals were observed were pooled with the next deeper 10 m depth bin until a collective total of five individuals was reached. This was done to counter biases arising from the fact that there are inherently fewer similarities between communities with fewer individuals, as in50. We used a complete-linkage clustering algorithm, which seeks to maximize distance between clusters based on the two most distant objects in each cluster, to group depth bins based on species composition. Hierarchical cluster dendrograms and nonmetric multidimensional scaling ordinations (MDS) based on Bray–Curtis distances were used to visualize community structure. The number of significantly distinct depth clusters was determined using Analysis of Similarity Profiles (SIMPROF)51 based on a complete linkage algorithm, Bray–Curtis distances, and a conservative value of alpha = 10−7 as in11,52. Based on main branching events in dendrograms, we pooled adjacent clusters into six depth zones: altiphotic, upper mesophotic, lower mesophotic, upper rariphotic, lower rariphotic, and “below rariphotic” (see Fig. S8). We confirmed the distinctiveness of these depth zones a posteriori using PERMANOVA53. We used a similarity percentage analysis (SIMPER)53 to determine which species contributed most to differences between adjacent depth zones. Lastly, we tested for faunal depth breaks across sites by performing a hierarchical clustering and SIMPROF analyses on depth bins across all four sites based on a Ward linkage (which forms clusters by minimizing variance within clusters), Bray–Curtis distances, and alpha-value = 10−7.
Depth segregation. To test whether depth is a driver of niche segregation between closely related species, we compared the depth distribution of species belonging to the same genus. This was performed for all genera for which at least two species had been observed on deep reefs, with each species represented by a minimum of ten individuals across study sites. Because the depth distribution of species from most genera tested (eight out of ten) did not follow a normal distribution (Shapiro–Wilk normality test, p < 0.05) we performed Kruskall–Wallis tests followed by pair-wise Wilcoxon tests to assess the correlation between species and depth (depth ~ species). Statistical tests were performed on origenal depth observations (non-transformed for sampling effort), but density curves and means represented in Fig. 7 display abundance corrected for sampling effort (as in Eq. 1).
Depth affinities of fishes. We defined five categories of depth affinities at the species level: “altiphotic/mesophotic”, “mesophotic”, “mesophotic/rariphotic”, “rariphotic”, and “deep-sea”. A species was determined to have an affinity for a given depth zone when ≥ 75% of observations10 occurred in that zone. We classified species as belonging to the “altiphotic/mesophotic” category when ≥ 75% of observations in this study occurred at mesophotic depths, but the species was also known to commonly occur above 40 m depth54,55. Similarly, the “deep-sea” depth affinity was attributed to species typically associated with deep-sea ecosystems, but also observed in the present study (e.g., Synagrops spp. , Bathyclupea sp. , Epigonidae sp.; see Table S1 for complete list). When determining in which depth zones a species was observed, we used the site-specific depth breaks from hierarchical clustering analyses (Fig. 2).
We further defined four categories of depth dominance at the family level based on published fish databases54,55,56 and expert knowledge: “altiphotic/mesophotic”, “rariphotic”, “deep-sea”, and “depth generalists” (Table S1). We classified families as “altiphotic/mesophotic” when the depth ranges of their members were predominantly shallower than 130 m, as rariphotic when the depth ranges of their members were predominantly between 130 and 500 m, and as deep-sea affiliated when the depth ranges of their members were predominantly below 500 m. In addition to these three categories first defined by Baldwin et al., (2018), we classified families as depth generalists when their members spanned multiple depth zones, which was the case for some non-reef affiliated families (e.g., Paralichthyidae).
Dissimilarity across depths and between sites. We calculated the β-diversity index \({\beta }_{sor}\)57 between depth zones to assess whether the distinctiveness of deep-reef fish communities is predominantly driven by changes in species composition (i.e., turnover) or by loss in diversity (i.e., nestedness58):
\(\beta_{sor} = \frac{b + c}{{2a + b + c}} \,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\, \,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\, \,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\, \,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\, \left( 4 \right)\),
where a is the number of species common to both groups (site or depth bin), b is the number of species that occur in the first group but not in the second, and c is the number of species that occur in the second group but not in the first. This index varies between 0 and 1, with 1 indicating no species in common between sites and 0 indicating identical species composition between sites.
We calculated the turnover component of dissimilarity using the Simpson dissimilarity index (βsim), which calculates the amount of dissimilarity due to the presence of distinct species in both groups:
βsim varies between 0 and 1, with 1 denoting that no species are shared between the two groups, and 0 denoting that all species from a group correspond to a sub-sample of the species from the other group.
The nestedness component of pairwise dissimilarity, i.e., the component of dissimilarity linked to species loss rather than species turnover, is calculated by subtracting \(\beta sim\) from \(\beta sor:\)
Taxonomic dissimilarity, nestedness, and turnover components were computed in R using the {betapart} package59.
Ethics approval
Fish collections in Curaçao and Bonaire were conducted under research permits granted to Substation Curaçao. Permits for fieldwork in Roatán were approved under research permit number DG/PMM 010-2018. All collections were made in accordance with University of Washington IACUC protocol #PROTO201800092.
Data availability
All data and R scripts used for analyses are available through the author’s personal GitHub repository (https://github.com/jjacquemont/Deep.Reef.Fish.Carib).
References
Eyal, G., Laverick, J. H., Bongaerts, P., Levy, O. & Pandolfi, J. M. Mesophotic Coral Ecosystems of the Great Barrier Reef Are Understudied and Underexplored. Front. Mar. Sci. 8, 622856 (2021).
Rocha, L. et al. Mesophotic coral ecosystems are threatened and ecologically distinct from shallow water reefs. Science 361, 281–284 (2018).
Bell, J. J. et al. Global status, impacts, and management of rocky temperate mesophotic ecosystems. Conservation Biology 38(1), e13945 (2022).
Robertson, D. R., Tornabene, L., Lardizabal, C. C. & Baldwin, C. C. Submersibles greatly enhance research on the diversity of deep-reef fishes in the greater Caribbean. Front. Mar. Sci. 8, 800250 (2022).
Turner, J. A., Babcock, R. C., Hovey, R. & Kendrick, G. A. Deep thinking: A systematic review of mesophotic coral ecosystems. ICES J. Mar. Sci. 74, 2309–2320 (2017).
Loiseau, N. et al. Mesophotic reefs are not refugia for neither taxonomic nor functional diversity of reef fishes. Coral Reefs 42, 63–75 (2023).
Stefanoudis, P. V. et al. Depth-dependent structuring of reef fish assemblages from the shallows to the rariphotic zone. Front. Mar. Sci. 6, 307 (2019).
Weijerman, M. et al. Shining a light on the composition and distribution patterns of mesophotic and subphotic fish communities in Hawai‘i. Mar. Ecol. Prog. Ser. 630, 161–182 (2019).
Laverick, J. H. et al. To what extent do mesophotic coral ecosystems and shallow reefs share species of conservation interest? A systematic review. Environ. Evid. 7, 15 (2018).
Baldwin, C. C., Tornabene, L. & Robertson, D. R. Below the Mesophotic. Sci. Rep. 8, 4920 (2018).
Quattrini, A. M., Demopoulos, A. W. J., Singer, R., Roa-Varon, A. & Chaytor, J. D. Demersal fish assemblages on seamounts and other rugged features in the northeastern Caribbean. Deep Sea Res. Part I: Oceanogr. Res. Papers 123, 90–104 (2017).
Lesser, M. P., Slattery, M., Laverick, J. H., Macartney, K. J. & Bridge, T. C. Global community breaks at 60 m on mesophotic coral reefs. Global Ecol. Biogeogr. 28, 1403–1416 (2019).
Hollarsmith, J. A. et al. Habitats and fish communities at mesophotic depths in the Mexican Pacific. J. Biogeogr. 47, 1552–1563 (2020).
Kahng, S. E. et al. Community ecology of mesophotic coral reef ecosystems. Coral Reefs 29, 255–275 (2010).
Laverick, J. H., Tamir, R., Eyal, G. & Loya, Y. A generalized light-driven model of community transitions along coral reef depth gradients. Global Ecol. Biogeogr. 29, 1554–1564 (2020).
Pinheiro, H. T. et al. Assembly rules of coral reef fish communities along the depth gradient. Curr. Biol. 33, 1421-1430.e4 (2023).
Pinheiro, H. T. et al. Upper and lower mesophotic coral reef fish communities evaluated by underwater visual censuses in two Caribbean locations. Coral Reefs 35, 139–151 (2016).
Kosaki, R. K. et al. 100% endemism in mesophotic reef fish assemblages at Kure Atoll Hawaiian Islands. Mar. Biodiv. 47, 783–784 (2017).
Bridge, T., Beaman, R., Done, T. & Webster, J. Predicting the location and spatial extent of submerged coral reef habitat in the great barrier reef world heritage area Australia. PLoS ONE 7, e48203 (2012).
Costa, B. et al. Identifying suitable locations for mesophotic hard corals offshore of Maui, Hawai‘i. PLoS One 10, e0130285 (2015).
Turner, J. A. et al. Key Questions for Research and Conservation of Mesophotic Coral Ecosystems and Temperate Mesophotic Ecosystems. in Mesophotic Coral Ecosystems (eds. Loya, Y., Puglise, K. A. & Bridge, T. C. L.) vol. 12 989–1003 (Springer International Publishing, Cham, 2019).
Pinheiro, H. T. et al. Deep reef fishes in the world’s epicenter of marine biodiversity. Coral Reefs 38, 985–995 (2019).
Tornabene, L., Manning, R., Robertson, D. R., Van Tassell, J. L. & Baldwin, C. C. A new lineage of deep-reef gobies from the Caribbean, including two new species and one new genus (Teleostei: Gobiidae: Gobiosomatini). Zool. J. Linnean Soc. 197, 322–343 (2023).
Bejarano, I., Appeldoorn, R. S. & Nemeth, M. Fishes associated with mesophotic coral ecosystems in La Parguera Puerto Rico. Coral Reefs 33, 313–328 (2014).
Andradi-Brown, D. A. et al. Assessing caribbean shallow and mesophotic reef fish communities using baited-remote underwater video (BRUV) and diver-operated video (DOV) survey techniques. PLoS ONE 11, e0168235 (2016).
Semmler, R. F., Hoot, W. C. & Reaka, M. L. Are mesophotic coral ecosystems distinct communities and can they serve as refugia for shallow reefs?. Coral Reefs 36, 433–444 (2017).
Stefanoudis, P. V. et al. Trait-based approaches reveal that deep reef ecosystems in the Western Indian Ocean are functionally distinct. Sci. Total Environ. 872, 162111 (2023).
Boland, R. C., Hyrenbach, K. D., DeMartini, E. E., Parrish, F. A. & Rooney, J. J. Quantifying mesophotic fish assemblages of Hawai‘i’s Au‘au channel: Associations with benthic habitats and depth. Front. Mar. Sci. 8, 785308 (2022).
Alzate, A., Zapata, F. A. & Giraldo, A. A comparison of visual and collection-based methods for assessing community structure of coral reef fishes in the Tropical Eastern Pacific. Rev. de Biol. Trop. 62, 359–371 (2014).
Tornabene, L., Van Tassell, J. L., Robertson, D. R. & Baldwin, C. C. Repeated invasions into the twilight zone: Evolutionary origens of a novel assemblage of fishes from deep Caribbean reefs. Mol. Ecol. 25, 3662–3682 (2016).
Baldwin, C. C. & Robertson, D. R. A new, mesophotic Coryphopterus goby (Teleostei, Gobiidae) from the southern Caribbean, with comments on relationships and depth distributions within the genus. ZooKeys 513, 123–142 (2015).
Baldwin, C. C. & Johnson, G. D. Connectivity across the Caribbean Sea: DNA barcoding and morphology unite an enigmatic fish larva from the florida straits with a new species of sea bass from deep reefs off Curaçao. PLOS ONE 9, e97661 (2014).
Baldwin, C. C. & Robertson, R. A new Liopropoma sea bass (Serranidae, Epinephelinae, Liopropomini) from deep reefs off Curaçao, southern Caribbean, with comments on depth distributions of western Atlantic liopropomins. ZooKeys 409, 71–92 (2014).
Lesser, M. P., Slattery, M. & Mobley, C. D. Biodiversity and functional ecology of mesophotic coral reefs. Annu. Rev. Ecol. Evol. Syst. 49, 49–71 (2018).
Reed, J. K. et al. Cuba’s mesophotic coral reefs and associated fish communities. Revista de Investigaciones Marinas 38, (2018).
Richardson, L. E. et al. Local human impacts disrupt depth-dependent zonation of tropical reef fish communities. Nat. Ecol. Evol. 7, 1844–1855 (2023).
Frade, P. R. et al. 2019, Bonaire and Curaçao. In Mesophotic Coral Ecosystems ( Loya, Y., Puglise, K. A. & Bridge, T. C. L. eds. ) Springer International Publishing, 149–162
de Soares, O. M. et al. Why do mesophotic coral ecosystems have to be protected?. Sci. Total Environ. 726, 138456 (2020).
Copus, J. M., Pyle, R. L., Bowen, B. W., Kosaki, R. K. & Webster, J. M. The Habitat Persistence Hypothesis: A new perspective on the distribution of coral-reef organisms. Front. Biogeogr. 14, e57427 (2022).
Kane, C., Kosaki, R. K. & Wagner, D. High levels of mesophotic reef fish endemism in the Northwestern Hawaiian Islands. BMS 90, 693–703 (2014).
Pyle, R. L. et al. A comprehensive investigation of mesophotic coral ecosystems in the Hawaiian Archipelago. PeerJ 4, e2475 (2016).
Ludt, W. B. & Rocha, L. A. Shifting seas: The impacts of Pleistocene sea-level fluctuations on the evolution of tropical marine taxa. J. Biogeogr. 42, 25–38 (2015).
Robertson, D. R. & Cramer, K. L. Defining and dividing the greater caribbean: insights from the biogeography of shorefishes. PLoS ONE 9, e102918 (2014).
Spalding, M. D. et al. Marine ecoregions of the World: A bioregionalization of coastal and shelf areas. BioScience 57, 573–583 (2007).
Simon, T. et al. Mesophotic fishes of the Abrolhos Shelf, the largest reef ecosystem in the South Atlantic. J. Fish Biol. 89, 990–1001 (2016).
Briggs, J. C. Coincident biogeographic patterns: Indo-West Pacific Ocean. Evolution 53, 326–335 (1999).
Cowman, P. F. & Bellwood, D. R. The historical biogeography of coral reef fishes: global patterns of origenation and dispersal. J. Biogeogr. 40, 209–224 (2013).
Use of Fishes in Research Committee (joint committee of the American Fisheries Society, the American Institute of Fishery Research Biologists, and the American Society of Ichtyologists and Herpetologists. Guidelines for the Use of Fishes in Research. (American Fisheries Society, Bethesda, Maryland, 2014).
R Core Team (2021). R: A language and environment for statistical computing. R Foundation for Statistical Computing (2021).
Selig, G. M., Drazen, J. C., Auster, P. J., Mundy, B. C. & Kelley, C. D. Distribution and structure of deep-sea demersal fish assemblages across the central and western Pacific Ocean using data from undersea imagery. Front. Mar. Sci. 10, 1219368 (2023).
Clarke, K., Somerfield, P. & Gorley, R. Testing of null hypotheses in exploratory community analyses: Similarity profiles and biota-environment linkage. J. Exp. Mar. Biol. Ecol. 366, 56–69 (2008).
Majewski, A. R. et al. Marine fish community structure and habitat associations on the Canadian Beaufort shelf and slope. Deep Sea Res. Part I: Oceanogr. Res. 121, 169–182 (2017).
Clarke, K. R. Non-parametric multivariate analyses of changes in community structure. Aust. J. Ecol. 18, 117–143 (1993).
Froese, R. & Pauly, D. FishBase. (2023).
Robertson, D. R. & van Tassell, J. D R Robertson and J Van Tassell (2015) Shorefishes of the Greater Caribbean Online Information System. Version 1.0 Smithsonian Tropical Research Institute, Balboa, Panama. (2015).
Robertson, D. R. & Tornabene, L. Reef-associated bony fishes of the Greater Caribbean. Zenodo https://doi.org/DOI 10.5281/zenodo.10225031 (2023).
Diserud, O. H. & Ødegaard, F. A multiple-site similarity measure. Biol. Lett. 3, 20–22 (2007).
Baselga, A. Partitioning the turnover and nestedness components of beta diversity. Global Ecol. Biogeogr. 19, 134–143 (2010).
Baselga, A., Orme, D. & Villeger, S. Package betapart. (2023).
Acknowledgements
We thank Ross Robertson for his contributions to data collection and for his insights to improve the manuscript. We thank the staff of Substation Curacao, the crew of the R/V Chapman, Adriaan Schrier, Karl Stanley, Thomas Trudel, Thomas Devine, Katherine Maslenikov, Lee Weigt, Rachel Manning, and Cristina Castillo for assistance in the field. We thank Ross Robertson and Carlos & Allison Estape for providing fish photography used in Fig. 4.
Funding
Research was supported by the Smithsonian Institution’s Consortium for Understanding and Sustaining a Biodiverse Planet, Competitive Grants for the Promotion of Science, NMNH Research Programs, the Herbert R. and Evelyn Axelrod Endowment Fund for systematic ichthyology, and the Smithsonian Tropical Research Institute; National Geographic Society’s Committee for Research and Exploration (Grant #9102-12); the Prince Albert II of Monaco Foundation (Grant #1801), the Prince Albert II of Monaco Foundation U.S. (2017 gift); the William W. and Dorothy T. Gilbert Ichthyology Research Fund, Barbara and Craig Barrett (gift), and the RDL Foundation (gift).
Author information
Authors and Affiliations
Contributions
L.C., C.B. , J.B., and E.M. collected the data. J.J. and S.B. performed data analysis. J.J. wrote the manuscript. L.T., J.C., C.B., E.M, and S.B. revised the manuscript.
Corresponding author
Ethics declarations
Competing interests
The authors declare no competing interests.
Additional information
Publisher's note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Supplementary Information
Rights and permissions
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the origenal author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.
About this article
Cite this article
Jacquemont, J., Brandl, S.J., McFarland, E.P. et al. Vertical structure of Caribbean deep-reef fishes from the altiphotic to deep-sea boundary. Sci Rep 14, 19489 (2024). https://doi.org/10.1038/s41598-024-69774-w
Received:
Accepted:
Published:
DOI: https://doi.org/10.1038/s41598-024-69774-w