The study area comprised the northeastern Brazilian coast, from Rio Grande do Norte to Alagoas states (5°–9°S), and the seamounts and oceanic islands of the Fernando de Noronha Ridge, including the Rocas Atoll (3°52′S, 33°49′W) and the Fernando de Noronha Archipelago (3°50′S, 32°25′W) ( Figure 1 ). The main oceanographic physico-chemical features of the region were recently described by Assunção et al. (2020); Costa da Silva et al. (2021), and Dossa et al. (2021). Overall, the SWTA is considered oligotrophic. However, locally the banks and islands act as topographic obstacles to currents, driving subsurface enriched waters to the surface (Travassos et al., 1999; Tchamabi et al., 2017; Costa da Silva et al., 2021; Silva et al., 2022). This process increases primary production and enhances the mass and energy fluxes throughout the food web (Travassos et al., 1999; Tchamabi et al., 2017).

Data and specimens were collected during the Acoustics along the BRAzilian COaSt (ABRACOS) surveys, carried out from 29 August to 21 September 2015 (AB1) and from 9 April to 10 May 2017 (AB2), aboard the French RV Antea (Bertrand, 2015; Bertrand, 2017). Temperature profiles were collected using a CTDO SeaBird911+. Mesopelagic fishes were collected day and night at 80 trawl stations by using mesopelagic (AB1; body mesh 30 mm, cod-end mesh 4 mm, size of the net mouth: 16.6 x 8.4 m; Bertrand, 2015) and micronekton (AB2; body mesh 40 mm, cod-end mesh 10 mm, size of the net mouth: 24 x 24 m; Bertrand, 2017) nets ( Figure 1 ; Supplementary Material 1 and 2 ). Targeted depth ranged from 10 to 1,113 m and was defined by the presence of acoustic scattered layers or patches detected by a Simrad EK60 (Kongsberg Simrad AS) split-beam scientific echosounder, operating at 38, 70, 120, and 200 kHz. Except for the layers 200–300 and 700–800 at night, where no aggregation of organisms were observed through acoustics, all depth strata were sampled at least once ( Supplementary Material 1 ). The net geometry was monitored using SCANMAR sensors, to give headline height, depth, and distance of wings and doors to ensure the net was fishing correctly. Based on SCANMAR the estimated opening area of the micronekton trawl was 120 m². For the mesopelagic trawl, however, the opening resembled an ellipse of 65 m². As the trawl was not fitted with an opening and closing mechanism, the collection of specimens during the lowering or hoisting of the net was reduced as much as possible by decreasing ship velocity and increasing winch speed. At the target depths, trawling activity lasted for about 30 minutes at 2–3 kt. Therefore, collection of specimens most likely occurred at target depths, which are indicated as capture depths in the species accounts.

Specimens were sorted to the lowest taxonomic level and frozen or, in the case of rare species or taxonomic uncertainty, fixed in 4% formalin and then preserved in a 70% alcohol solution (Eduardo et al., 2020a). In the laboratory, specimens were identified, measured (nearest 0.1 cm of standard length, SL), and weighed (nearest 0.01 g of total weight, TW). Excluding a few specimens of the Stomiidae, Sternoptychidae, and Myctophidae used for biological analyses (Eduardo et al., 2020a; Eduardo et al., 2020b; Eduardo et al., 2021), all specimens were deposited in the NPM – Fish Collection of the “Instituto de Biodiversidade e Sustentabilidade, Universidade Federal do Rio de Janeiro” (NUPEM/UFRJ). Taxonomic classification follows Nelson et al. (2016), with exceptions noted in Villarins et al. (2022) for the Stomiiformes.

We first computed a randomised species accumulation curve to assess whether the fish community was exhaustively sampled with the gears employed (Gotelli and Colwell, 2001). This enables calculating a mean number of species for a given number of samples within a 95% confidence interval. The Chao1 index, which extrapolates the total expected number of species in the area for a given sampling gear, was subsequently calculated (Magurran, 2004).

Other aspects of the biodiversity were assessed based on the sample-size-based rarefaction and extrapolation sampling curves, calculated for the species richness, Shannon diversity, and Simpson dominance, the three most widely used species diversity indexes (Magurran, 2004). For that, we used Hill’s numbers, which integrate species richness and relative abundance to propose a more intuitive and statistically rigorous alternative to calculate diversity measures (Chao et al., 2014). Statistical significance was evaluated based on the confidence interval overlapping of the curves.

Sample-size-based rarefaction and extrapolation sampling curves (Hsieh et al., 2016) were also constructed to test for differences in diversity indexes when considering depth strata (epipelagic 0–200 m; upper mesopelagic 200–500 m; lower mesopelagic 500–1000 m) and the diel period (day and night). As the sampling strategy employed in the AB2 expedition was much more efficient in collecting mesopelagic fishes (see Discussion), comparisons using diversity indexes were only made for this survey. Statistical analyses and the calculation of diversity indices were performed using the software R version 4.0.3 through the packages “iNext” (Hsieh et al., 2016) and “vegan” (Oksanen et al., 2017). Fish larvae and species traditionally classified as epipelagic were excluded from the species list, and they were not considered for the diversity assessments. Specimens identified at the genus level only (small-sized and/or damaged specimens), which might represent more than one species, were also excluded from the analyses ( Supplementary Material 3 ).

Considering our two surveys, 7,119 specimens of mesopelagic fishes, representing 200 species in 130 genera, 56 families, and 22 orders of the Teleostei and one shark (Isistius brasiliensis: Dalatiidae, Squaliformes), were collected and identified ( Table 1 ). The species accumulation curve was steep, indicating that more species would be recorded with additional sampling using the same gears ( Figure 2 ). Indeed, richness estimators indicated that about 100 (50%) additional mesopelagic species of fishes are expected to occur in the area ( Figure 2 ). Additionally, 759 specimens representing about 40 fish taxa were sampled. However, they could not be identified to species level given their small size and/or poor condition. As it was not possible to determine whether these specimens belong to species not listed in Table 1 , they were placed in a separate list to ensure a more robust assessment of species diversity ( Supplementary Material 3 ).

Ranges of standard length (SL) and wet weight for all species collected on the two surveys are provided in Table 1 . Overall, a wide size range was sampled, from 4 mm (Diretmoides pauciradiatus) to 1,880 mm SL (Eumecichthys fiski, Lophotidae). However, 90% of the specimens measured between 30 and 200 mm SL ( Supplementary Material 4 )

The five orders with the highest number of species were the Stomiiformes (at least 62 species, four families), Myctophiformes (35 spp., two families), Aulopiformes (18 spp., seven families), Beryciformes (16 spp., three families), and Lophiiformes (12 spp., seven families), accounting for 70% of the total number of species recorded on the two surveys. Thirteen orders included less than five species each. Considering families, the most representative were the Stomiidae (38 spp.), Myctophidae (34 spp.), Melamphaidae (11 spp.), Sternoptychidae (10 spp.), and Gonostomatidae (7 spp.) ( Figure 3 ). Half of the families (28) were represented by a single species.

In terms of abundance, the most representative families when considering the two surveys were the Myctophidae (Myctophiformes; 36%), Sternoptychidae (Stomiiformes; 26%), Gonostomatidae (Stomiiformes; 16%), Stomiidae (Stomiiformes; 8%), and Melamphaidae (Beryciformes; 2%) ( Figure 3 ). These families together accounted for 88% of all specimens collected. The remaining families represented individually no more than 2% of the total number of specimens collected. At the alpha taxonomic level, the following taxa represented almost 50% of all specimens collected: Sternoptyx diaphana (14%), Cyclothone spp. (11%; see Discussion), Diaphus brachycephalus (6%), Argyropelecus affinis (6%), Chauliodus sloani (5%), Lampanyctus nobilis (4%), and Diaphus perspicillatus (4%). About 126 species were represented by five specimens or less, of which 62 were represented by a single specimen.

Considering biomass, the most representative families were the Myctophidae (24%), Stomiidae (21%), Setarchidae (Scorpaeniformes, 11%), Sternoptychidae (10%), and Melamphaidae (7%) ( Figure 3 ). These families together accounted for 73% of the biomass of all fishes collected. The remaining families individually accounted for less than 4% of the total weight. At the specific level, the following species represented 42% of the biomass: Ectreposebastes imus (11%), Chauliodus sloani (9%), Borostomias elucens (6%), Eumecichthys fiski (6%, a single specimen), Sternoptyx diaphana (4%), Melamphaes polylepis (3%), and Argyropelecus affinis (3%).

Based on the two campaigns, 60 species (29%) were recorded in a wide longitudinal distribution ( Table 1 ). In contrast, 133 species (64%) were collected only in a few localities, with 116 being restricted to the Fernando de Noronha Ridge area, which aggregates most specimens collected ( Table 1 ). Considering depth and period, the highest diversity, abundance, and biomass were found between depths of 700 and 1,000 m during the day ( Figure 4 ). At night, the highest number of species was recorded at lower mesopelagic depths (500–1,000 m). However, much larger values of number of species, abundance, and biomass were detected in shallow waters (0–200 m), likely reflecting the ascent in the water column of several species at night. At least 50 species seem to have a wide range of depth distribution and tolerance to variations in water temperature (up to 800 m and 25°C; e.g., species of Sternoptychidae and Myctophidae). However, 66 species seem to be restricted to deeper (> 600 m) and colder waters (< 6° C) regardless of the time period (e.g., Lophiiformes and Beryciformes; Table 1 ).

Significant differences in biodiversity indexes (calculated only for ABS2, see methodology) were found when considering diel periods and depth. Higher values of richness and diversity were found in lower mesopelagic waters and during the daytime. However, dominance values were significantly higher in epipelagic waters and also during daytime ( Figure 5 ). Detailed values for the calculated indexes are given in the Supplementary Material 5 .

Based on our two campaigns, at least 201 species of mesopelagic fishes occur in the SWTA. Results also indicate that about 100 additional species could have been collected if sampling efforts were increased. The taxonomically diverse pool of mesopelagic species recorded in our surveys also reveals a vast array of diversity not only in terms of the number of species but also in terms of size, anatomy, and behaviour. In a recent global biogeographic classification of the mesopelagic zone (Sutton et al., 2017), the Tropical and western Equatorial Atlantic, which is the larger area encompassing the SWTA, was not considered a region particularly diverse in terms of mesopelagic fishes. However, the mesopelagic species richness revealed by our two campaigns is higher than those reported for other parts of the world, such as the Mediterranean (25 spp.; Olivar et al., 2012), central Equatorial Pacific (113 spp.; Barnett, 1984), southwestern Indian Ocean (121 spp.; Cherel et al., 2020), eastern Equatorial Atlantic (132 spp.; Olivar et al., 2017), and South China Sea (169 spp.; Wang et al., 2019). The species richness of mesopelagic fishes in the SWTA is actually more similar to that reported for the North Pacific (228 spp.; Barnett, 1984) and the Gulf of Mexico (approximately 300 spp.; Sutton et al., 2020), which are considered as comprising some of the most speciose deep-sea ichthyofaunas of the world (Sutton et al., 2017). Major factors driving deep-sea biodiversity, such as climate, seabed structure, water masses, and phylogenetic history, are likely responsible for the variation in species richness of different parts of the world. However, an asymmetry in collecting effort is certainly affecting the values recorded so far. In the Gulf of Mexico, a much higher sampling effort has been deployed to assess the deep-sea diversity compared with most regions of the world, with several expeditions conducted only in the last decade (Sutton et al., 2020). That situation is in striking contrast with the SWTA, where just a handful of expeditions to assess deep-sea diversity have been conducted in the last centuries.

The relatively high number of mesopelagic species of fishes recorded in our two campaigns is likely related to the diversity of habitats and the high variability of oceanographic processes of the SWTA. Despite being located in an oligotrophic portion of the ocean, this region is also characterized by the presence of underwater canyons, oceanic islands, and several seamounts that interact with local currents and enhance marine productivity (Travassos et al., 1999; Tchamabi et al., 2017; Costa da Silva et al., 2021). As an example, small uplifting processes have been reported along the shelf-break and oceanic islands of the region (Travassos et al., 1999; MMA, 2006; Tchamabi et al., 2017; Silva et al., 2022), a situation that has been directly associated with the occurrence of hotspots of fish biodiversity (Hazin, 1993; Eduardo et al., 2018; Eduardo et al., 2020a). Distinct biogeographic provinces, with different thermodynamic features, current systems and water mass properties, are also present in the SWTA (Bourlès et al., 1999; Assunção et al., 2020; Costa da Silva et al., 2021; Dossa et al., 2021; Tosetto et al., 2021). This results in a high complexity of habitats and oceanographic conditions that likely contribute to higher levels of species diversity (Levin et al., 2001).

The highest levels of richness and diversity (considering only specimens collected during AB2, see methodology) were found at lower mesopelagic depths (500–1,000 m), with several species collected only at these depths (e.g., species of the Beryciformes and Lophiiformes). Interestingly, many of these species are considered bathypelagic and/or benthopelagic (Priede, 2017; Melo et al., 2020). The collection of those species in mesopelagic waters is likely related to the presence of seamounts and oceanic islands. In addition to being related to an increase in habitat complexity, seamounts may increase the occurrence of pelagic and benthic predators that actively seek these areas to hunt for prey trapped by flow-topographic processes (Cascão et al., 2019). For instance, in the Azorean seamounts plateau, the micronekton community is dominated by non- or weakly migratory benthopelagic fishes (Cascão et al., 2019). In summary, our results also seem to indicate that seamounts play a significant role in the biodiversity structuring and ecology of mesopelagic fishes in the SWTA.

The two surveys conducted during this study resulted in different patterns of species richness. For example, 17 species were exclusively recorded in AB1 (mesopelagic trawl), whereas 136 species were recorded only in AB2 (micronekton trawl). The two campaigns were conducted in different seasons. However, since the study area is located in a tropical region, few oceanographic differences were noted in the mesopelagic zone (for further info refer to Assunção et al., 2020; Costa da Silva et al., 2021; Dossa et al., 2021). Therefore, the significant disparity in species richness between the two expeditions is clearly related to differences in sampling strategies. The use of multiple sampling gears is vital to maximizing the representation of fish diversity (Magurran, 2004), especially in the deep-sea. However, the sampling strategy used in AB2, which included the use of larger gear, with greater mesh sizes, deeper hauls, and broader sampling area, resulted in the collection of a higher number of specimens of different species in a broader size range ( Supplementary Material 2 and 3 ).

In terms of taxonomic composition, five families of the Teleostei accounted for 52% of the species richness, 88% of the specimens, and 66% of the total biomass collected on the two surveys: the Stomiidae (38 spp., 8% of the specimens, 21% of the biomass), Myctophidae (34 spp., 36%, 24%), Melamphaidae (11 spp., 2%, 7%), Sternoptychidae (9 spp., 26%, 10%), and Gonostomatidae (7 spp., 16%, 4%). These families, therefore, seem to be the most represented in the mesopelagic fish fauna of the SWTA. The dominance of these families in mesopelagic waters has also been reported in other parts of the world (e.g., Gjøsaeter and Kawaguchi, 1980; Olivar et al., 2017; Wang et al., 2019; Cherel et al., 2020). A strong pattern of dominance was also observed within these families, with few species accounting for 50% of the total number of specimens: Sternoptyx diaphana (14%), Cyclothone spp. (11%), Argyropelecus affinis (6%), Diaphus brachycephalus (6%), Chauliodus sloani (5%), Diaphus perspicillatus (4%), and Lampanyctus nobilis (4%). The pattern of dominance at the species level detected in the SWTA was, however, distinct from other parts of the world. In the Eastern Tropical Atlantic, for instance, the lanternfishes B. suborbitale, C. warmingii, and H. macrochir were dominant (Olivar et al., 2017), whereas these same species were considered rare in our study. The viperfish C. sloani is usually globally recorded in low abundances (e.g., Olivar et al., 2017; Wang et al., 2019; Cherel et al., 2020), whereas the species is among the most relevant mesopelagic species in the SWTA considering the abundance and total weight (Eduardo et al., 2020c). These differences in the pattern of dominance at the species level in different parts of the world are likely associated with different sampling strategies employed and differences in oceanographic and biogeographic features (e.g., seabed structure, water masses, and hydrographic fronts), which are major factors driving the structure and composition of mesopelagic assemblages (Hulley and Krefft, 1985; Olivar et al., 2017; Cascão et al., 2019). Cyclotone is another seemingly abundant genus of mesopelagic fish in the SWTA (Olivar et al., 2017). Eight species of the genus were reported for the SWTA: C. acclinidens, C. alba, C. braueri, C. microdon, C. obscura, C. pallida, C. parapallida, and C. pseudopallida (Villarins et al., 2022) The sampling gears employed in the study, however, seemed to be only partially adequate to collect specimens of the genus. In several trawls we observed onboard that a substantial number of specimens of Cyclothone escaped back into the sea during the hoisting of the net. Additionally, given their poor condition of preservation, specimens of the genus could not be identified at species level. Therefore, the abundance of species of Cyclothone presented here is underestimated.

Among the 201 species of mesopelagic fishes recorded during the ABRACOS expeditions, 50 (25%) represent new records for Brazilian waters, all of which have been dealt with in a series of recent papers ( Table 1 ). In addition to these new records, eight species (five Eustomias, one Melanostomias, one Melamphaes, and one Poromitra) are potentially new and will be formally described later. Several species recorded here are also rare worldwide, and their occurrence in the SWTA adds new information on their global distribution. For instance, three specimens of Platyberyx paucus and one of Platyberyx pietschi were collected during the AB2. Before these records, only four specimens of P. paucus were known, from the central North Pacific and western Central Atlantic. Platyberyx pietschi, in turn, was known from just two specimens collected in the western Central Atlantic, one specimen collected in the central Pacific, and another from the western South Pacific (Stevenson and Kenaley, 2013; Mincarone et al., 2019). Other species considered rare worldwide that were collected in the ABRACOS expeditions are Aulotrachichthys argyrophanus, Rhynchohyalus natalensis, Eumecichthys fiski, Macrouroides inflaticeps, Pseudoscopelus cordilluminatus, Melamphaes leprus, and Gigantactis watermani (Pimentel et al., 2020; Afonso et al., 2021; Mincarone et al., 2021b; Mincarone et al., 2022).

The high number of new records made during the ABRACOS expeditions reflects not only the high diversity of the SWTA, but also the overall lack of scientific information on deep-sea diversity in the region, as noted previously (e.g., Reis et al., 2016; Mincarone et al., 2022). The United Nations Decade of Ocean Science roadmap recognizes the deep-sea as a frontier of science and discovery (Ryabinin et al., 2019). There is an unequal capacity to conduct science among nations, with developing economies facing substantial barriers to participating in deep-sea research. Consequently, the least-studied parts of the deep-sea are located off the least economically developed countries (Howell et al., 2020). These biases are highlighted by the fact that a French research institution financed the surveys described here, and that those expeditions are among the very few that have addressed the mesopelagic ichthyofauna of Brazil. To achieve sustainability, we need a well-known and predictable ocean. Only by thinking globally and strengthening international cooperation we will develop an ocean research that corrects asymmetry in funding and knowledge among countries, meeting the crucial need for a more encompassing deep-sea knowledge aimed at the conservation and sustainable use of its unique habitats.