Pickle Bank seamount is a remote offshore seamount located in the Central Caribbean Sea, 78km north-north-west of Little Cayman and 120km south-west of Cuba ( Figure 1A ; 20.417°, -80.417°). The seamount is occasionally visited by recreational and commercial fishermen as a prime location for catching large pelagic species, yet there is no formal characterization of the fish communities. The seamount is approximately 5.2 square km and is characterized by a central lagoon, with large patch reef outcroppings rising to 14m in depth. These outcroppings transition to spur and groove reef structures, leading to a deep reef edge at roughly 35m depth before descending with near vertical walls to depths exceeding 1,000m ( Figures 1B, C ).

In situ visual fish surveys were conducted on Pickle Bank from 14 July 2024 to 17 July 2024 using closed-circuit rebreathers (Prism2, Hollis Rebreathers) by two experienced technical divers and professional scientists who have extensively published fish data from mesophotic depths in the Western Atlantic (Pinheiro et al., 2016; Andradi-Brown et al., 2017; Goodbody-Gringley et al., 2019, 2023; Le Gall et al., 2024). Due to the time required to reach the seamount each day (8-h round trip) and safety requirements of diving to 45m, a maximum of two dives were conducted each day, limiting the total number of surveys attainable within the logistical constraints of the expedition. A total of 17 fish transects were surveyed at two different depth ranges on the seamount, with shallow surveys (n=8) conducted at 25m, while deep surveys (n=9) were conducted at 45m. At shallow sites, replicate transects (30m x 2m) were haphazardly placed on reef spurs ( Figure 1C ) that resembled classical benthic assemblages associated with shallow coral reef habitats in the region (Manfrino et al., 2013). Deep transects were attached haphazardly to the benthos and laid horizontally along the reef wall to ensure a constant depth, with benthic assemblages resembling those described for nearshore MCEs in the Cayman Islands ( Figure 1B ; Slattery and Lesser, 2019; Carpenter et al., 2022; Le Gall et al., 2024). Along each transect, all fish encountered were identified to the species level and their total length visually estimated (Johnson et al., 2023; Le Gall et al., 2024). Fish total length was categorized into size classes (0–5 cm, 6–10 cm, 11–20 cm, 21–30 cm, 31–40 cm, and > 40 cm estimated to the nearest 10cm value) to allow for biomass calculations using the formula:

where W is the weight of the fish and L is the maximum length based on the size classes above. Values for a and b are species-specific constants based on empirical data for calculating fish biomass from size-weight relationships (Bohnsack and Harper, 1988; Torres, 1991; Froese and Pauly, 2010). These constants were obtained from Fish Base (Froese and Pauly, 2010), with values from congenic species used if data for a specific species were not available. Each fish species was subsequently grouped into the appropriate trophic guild based on groupings also derived from Fish Base (Froese and Pauly, 2010).

All data manipulation and statistical analyses were conducted using R 4.1.1 (R Core Team, 2023). Comparisons between depths for fish biomass and species richness of all fish were compared using ANOVAs, as data were normally distributed based on visual inspection of histograms, statistically confirmed with a Shapiro–Wilks test. Mann–Whitney U tests were used to compare the abundance and Shannon diversity of fish between the depths and for comparisons of biomass and abundance for all fish trophic guilds between the depths, as these data did not follow a normal distribution (Shapiro–Wilks, P<0.001). The Kruskal–Wallis test was used to compare the abundance and biomass of fish among trophic guilds as these data were also non-normally distributed (Shapiro–Wilks, P<0.001). Post hoc Dunn’s tests were used for pairwise comparisons of trophic guilds implemented using the FSA package (Ogle et al., 2017). To assess differences in the community composition of all fish species between the depths, and the community composition of trophic guilds, we used a PERMANOVA from the vegan package (Oksanen et al., 2020). Bray–Curtis dissimilarity matrices of the community were calculated from data normalized with a square-root transformation. Ordinations using non-metric multi-dimensional scaling (nMDS) were built also using the vegan package on the Bray–Curtis dissimilarity matrix (Oksanen et al., 2020).

Across all of our transects (n=17), we recorded a total 6,049 individual fish, representing 45 species. Between the two depth zones surveyed (25m vs. 45m), species richness ( Figure 2A ) was significantly higher at 25m (median = 19), compared to 45m (median = 14) (ANOVA, F=5.929, P=0.028). There was no difference in total fish biomass between the depths ( Figure 2B ). However, total fish abundance ( Figure 2C ) was significantly higher at 45m compared to 25m (Mann–Whitney U, W=63, P=0.008), while fish diversity ( Figure 2D ) was significantly lower at 45m compared to 25m (W=1, P<0.001). Additionally, there is a significant difference in the community composition between the mesophotic and shallow reef depths ( Figure 2E ; PERMANOVA, F=11.939, df=1, P<0.001).

Overall, we recorded 16 fish families on our transects, with the highest biomass at the shallow depth for the families Labridae, Scaridae, Carangidae, and Balistidae ( Figure 3A ). Comparatively, Carangidae and Labridae were the two dominant families of biomass at the mesophotic depth ( Figure 3B ). An overview of species contributions to biomass at the two depths and overall are shown in Figure 3C .

Regarding fish abundance, Labridae was the dominant fish family at the shallow depth, followed by Pomacentridae and Grammatidae ( Figure 4A ). These families were also dominant at the mesophotic depth, but Grammatidae showed the highest abundance, followed by Labridae and Pomacentridae ( Figure 4B ). Gramma melacara had the largest contribution out of all the species to overall fish abundance but was only recorded at the mesophotic depth. An overview of all species abundance contributions to each depth is shown in Figure 4C .

The biomass of reef fish on Pickle Bank was not significantly different across trophic guilds ( Figure 5A ), however, there were significant differences between the depths ( Figure 5B ). Macrocarnivores had significantly higher biomass at the mesophotic sites compared to the shallow sites (W=8, P=0.02), while planktivore biomass was significantly higher at the shallow depth compared to the mesophotic depth (W=63, P=0.008).

The shallow and mesophotic sites at Pickle Bank showed a significant difference in the abundance of fish ( Figure 5C ) across trophic guilds (χ² = 53.758, df=4, P<0.001). Pairwise comparisons of the difference between trophic guilds are shown in Table 1 .

There were significant differences in the abundances of trophic guilds between the shallow (depth of 25m) and mesophotic (depth of 45m) sites ( Figure 5D ), with higher herbivore abundance (W=64, P=0.008) and higher invertivore abundance (W=66, P=0.004) at the shallow depth compared to the mesophotic sites. Meanwhile, higher macrocarnivore abundance (W=8, P=0.022) and higher omnivore abundance (W=8, P=0.006) were recorded at the mesophotic sites compared to the shallow sites.

Fianlly, there was a significant difference in the trophic community composition between the depths ( Figure 5E ; PERMANOVA, F=8.894, df=1, P<0.001).

From our 17 transects, we found distinct differences in the fish community between the 25m and 45m depths. Notably, higher species richness and higher diversity of reef fish exist in the shallow water, generally congruent with the higher habitat complexity associated with shallow water reefs that drives fish diversity (Cornell and Karlson, 2000). However, higher fish abundance at greater depths on Pickle Bank is driven by large schools of species such as Gramma melacara, which were absent at the 25m depth. Overall, it appears that while Labridae, Grammatidae, and Pomacentridae dominate the fish communities at both depths ( Figure 3 ), the relative proportions of species in each family drove differences in the fish community.

Such partitioning is reinforced by differences in the biomass and abundance of different trophic groups. For example, higher biomass of typical macrocarniverous reef fish (e.g., species from Carangidae and Lutjanidae) was observed at 45m, congruent with the hypothesis that interspecific competition in shallow water leads to increased carnivore biomass at greater depths (Pinheiro et al., 2023; Grove et al., 2024; Heidmann et al., 2024). However, this hypothesis also applies to planktivorous fish, yet we found higher planktivore biomass at shallow depths. Higher planktivore biomass at shallow depths on Pickle Bank may be an artifact of the limited sample size from our study, which is supported by there being no difference between planktivore abundance between depths. Alternatively, it could be related to the hydrodynamic drivers of fish communities (Galbraith et al., 2023), which are unquantified on Pickle Bank. Herbivore and invertivore abundances are significantly greater at shallow depths on Pickle Bank compared to deeper sites, aligning with the species-energy hypothesis (Whittaker et al., 2001), where light-dependent resources that drive benthic energy availability become limited at greater depths (Pinheiro et al., 2023). Thus, the overall differences in the species and trophic communities between the depth found here appear to follow the prevailing hypotheses regarding fish communities across depth gradients.

Overall, the observed depth partitioning of the Pickle Bank fish community generally aligns with assembly rules that govern fish communities along the depth gradient (Pinheiro et al., 2023), with differences in community composition at 25–27m compared to 44–46m depths. These differences, while preliminary based on our small sample size of 17 transects, also likely exist because of convergent filtering of fish species with depth, where taxonomic groups and trophic strategies are similar at mesophotic depths because of the environmental conditions, placing less emphasis on regional and biogeographic drivers (Pinheiro et al., 2023). For example, our mesophotic sites show similar species compositions as nearby mesophotic locations, including Cuba (Cobián-Rojas et al., 2021), Grand Cayman (Le Gall et al., 2024), and more distant locations such as Puerto Rico (Bejarano et al., 2014), US virgin islands (Grove et al., 2024; Heidmann et al., 2024), Curacao (Pinheiro et al., 2016), and the Mesoamerican Barrier Reef (Andradi-Brown et al., 2016). Additionally, the shallow (25m) fish community appears similar to the adjacent shallow water reefs of Cuba (Navarro-Martínez et al., 2022) and the Cayman Islands (Johnson et al., 2024) based on species composition.

Given that our findings are provisionally congruent with the theories on depth-dependent rules associated with reef fish assemblages and the unexplored nature and the likely importance of Pickle Bank as a stepping stone, our findings incentivize further research to understand the ecological community and enact policy for protection. Pickle Bank likely maintains a healthy fishery given the larger biomass of targeted fishery species (mainly macrocarnivores), such as grouper, in comparison to the rest of the Cayman Islands (Stock et al., 2021). Larger biomass of targeted fishery species is likely associated with reduced harvesting pressure, highlighting the efficacy of reduced harvesting for sustaining fish stocks (Caldwell et al., 2024). Additionally, offshore seamounts provide a habitat for carnivorous reef fish and can act as magnets for predatory fish biomass (Cresswell et al., 2023; Weber et al., 2025), likely also explaining the larger grouper community at Pickle Bank compared to nearshore reefs of Little Cayman and Grand Cayman. Furthermore, Pickle Bank is likely a stepping stone for fish species, contributing to region-wide genetic connectivity, as overlap in species composition between nearshore reefs of Cuba and the Cayman Islands in shallow waters clearly exist (Navarro-Martínez et al., 2022; Johnson et al., 2023). Connectivity between mesophotic seamounts for reef-dwelling fish communities can transcend biogeographic realms, for example, between the Coral Sea, Southwest Pacific, and the Coral Triangle (Galbraith et al., 2024). Such connectivity via either larval dispersal or migrations of pelagic species in the Caribbean appears likely.

Additionally, the overall species richness of fish on Pickle Bank (45 species recorded) being similar to mesophotic reefs in the nearby island of Grand Cayman (48 species recorded), species per transect also being similar (Le Gall et al., 2024), may suggest these mesophotic depths are more sheltered from local anthropogenic stress even in proximity to high human population density. For example, shallow water reef fish are substantially impacted by local-scale activity associated with high human populations in Grand Cayman, such as the presence of mega cruise ships (Johnson et al., 2023). Yet, little difference in fish species richness at mesophotic depths between Pickle Bank and Grand Cayman may suggest the depth of MCEs buffer impacts of such local-scale activity. However, it is important to consider that further quantification of fish communities from both Pickle Bank and Grand Cayman mesophotic reefs are required to reach saturation for measuring species richness. One would expect that the combination of remoteness and depth may shelter fauna associated with Pickle Bank from anthropogenic activity, at least while Pickle Bank is not directly targeted.

For these reasons, ensuring Pickle Bank remains sheltered from unsustainable harvesting activities is likely crucial to avoid ecosystem collapse, given the small size of the seamount. Further characterization of the fish and benthic community will be beneficial for understanding ecological processes in this unique ecosystem and further inform effective management. Such ecological processes can be further understood with a holistic characterization of ecological communities, including genetic connectivity and the role of remote mesophotic reefs as a refuge during climate change (Smith et al., 2014; Bongaerts and Smith, 2019). The role of remote offshore seamounts as refuges during the current rapid period of global warming is especially pertinent given that remoteness is not protection from heatwaves for benthic organisms such as corals (Baumann et al., 2022; Johnson et al., 2022), and depth-dependent survival during thermal stress depends on a variety of factors (Smith et al., 2016; Rocha et al., 2018). Thus, acquiring more data on the fish community and characterizing the benthic community is vital for implementing effective policy to protect the unique habitat of Pickle Bank seamount.

Our work characterizing the fish community of Pickle Bank is the first step toward quantifying the community in this unique habitat. Quantifying mesophotic communities on offshore seamounts with in situ surveys is an incredible logistical challenge, but it is an important challenge to overcome given the insights society can gain from ecological, evolutionary, and biogeographical theories (Pinheiro et al., 2017, 2023; Lesser et al., 2018). Additionally, if we are to effectively protect the biological communities associated with offshore seamounts, we first need a holistic understanding of the community that exists (Lesser et al., 2018), followed by continued monitoring efforts to assess community changes as pressures from climate warming, mining, and overfishing inevitably reach untouched MCEs (Rocha et al., 2018; Rogers, 2018; Pinheiro et al., 2019). Therefore, while we show the differences in the communities between depths, further data to reach saturation points for species accumulation would greatly increase the ecological insights we can infer from the unique habitat of Pickle Bank seamount. Characterizing the benthic community is especially paramount to further understand the influence on the fish community. Such characterization becomes even more important given the threats posed to benthic coral habitats, including MCEs.