Bivalve Diversity on the Continental Shelf and Deep Sea of the Perdido Fold Belt, Northwest Gulf of Mexico, Mexico

: Mollusk diversity in coastal areas of the Gulf of Mexico (GOM) has been studied extensively, but this is not the case for deep-water habitats. We present the ﬁrst quantitative characterization of mollusks in shallow and deep waters of the Perdido Fold Belt. The data came from two research cruises completed in 2017. Sediment samples were collected from 56 sites using a 0.25-m 2 box corer. We tested hypotheses about spatial patterns of α , β , and γ -diversity of bivalves in two water-depth zones, the continental shelf (43–200 m) and bathyal zone (375–3563 m). A total of 301 bivalves belonging to 39 species were identiﬁed. The two zones display similar levels of γ -diversity, but host different bivalve assemblages. In general, α -diversity was higher on the continental shelf, whereas β -diversity was higher in the bathyal zone. These patterns can be explained by the higher input of carbon (energy) to the near-coast shelf zone, as well as by the greater topographic complexity of habitats in the bathyal zone. These results enabled us to propose redirection of sampling efforts for environmental characterization from continental zones to the deep-water zone, especially in the context of environmental assessments during oil and gas exploration and production.


Introduction
The deep sea extends beyond the depth of the photic zone and the edge of the continental shelf, which lies at about 200 m depth [1,2]. Nomenclature for bathymetric zones in the deep sea is not consistent and can differ depending on the taxonomic group being studied. Although there is some disagreement regarding the terminology applied to these marine zones and their bathymetric limits, we follow the system proposed by [3]. The perception that deep-sea communities are "desert-like" and display low taxonomic diversity has changed in the last few decades [4]. It is now known that these deep-water environments are rich in species, comparable to coastal marine habitats [2,[5][6][7], and that the deep sea is highly complex and governed by interactions among multiple environmental factors [8,9]. Only a small proportion of deep-sea realm, however, has been studied, because its exploration is technically difficult and expensive [10,11]. Given the traditional perceptions about differences in shallow-and deep-water marine environments, and the lack of knowledge about the diversity of bivalve mollusk in deep-water habitats of the Mexican part of the PFB, this study had two objectives: (1) to document the bivalve mollusk species in the PFB; and (2) to assess spatial patterns of diversity (α, β and γ) of bivalve mollusks in the PFB region, with respect to water depth, specifically from the continental shelf to the bathyal zone.

Materials and Methods
The study area included the southern Perdido Fold Belt, a section of the Tamaulipas shelf, and the Rio Bravo slope [40]. The PFB is a frontier petroleum exploration province in the GOM, located off the east coast of Mexico [41], with an area of ~272,300 km 2 [10] and water depths between 2000 and 3600 m [42,43] (Figure 1). The northern sector has been well studied with geophysical surveys [32], whereas the southerly sector, in Mexican waters, is poorly known [32]. The PFB is part of the allochthonous salt sheets of the Sigsbee salt nappe [33]. Sediment input comes from the Hudson and Rio Grande embayments, located to the north and west, respectively [32,44,45].

Sample Collection
Bivalve comprise part of the benthic macrofauna and their shells, along with some live animals, were collected from soft-bottom sediments at water depths between 43 and 3563 m, and assigned to two water-depth zones: (1) the continental shelf samples (43-200 m) and (2) bathyal zone (375-3563 m) ( Figure 1, Table 1). Samples were collected from the Research Vessel "Justo Sierra" (Universidad Nacional Autónoma de Mexico) during oceanographic cruises Perdido 3 in June 2017 (29 sites), and Perdido 4 in September 2017 (27 sites) (Table 1). At each sampling site, sediment was collected using a 0.25-m 2 Hessler Sandia MK-III box corer. Site, latitude, longitude, water depth, zone, and cruise information for each sample were recorded (Table 1). Each sediment sample was subsampled for: bacteria, chemical contaminants, metals, sediment texture, and benthic fauna community (three subsamples). Subsequently, sediment samples were sieved through 500-and 300-μm mesh sizes. Retained shells were placed in labeled plastic bags and examined later in the laboratory.

Materials and Methods
The study area included the southern Perdido Fold Belt, a section of the Tamaulipas shelf, and the Rio Bravo slope [40]. The PFB is a frontier petroleum exploration province in the GOM, located off the east coast of Mexico [41], with an area of~272,300 km 2 [10] and water depths between 2000 and 3600 m [42,43] (Figure 1). The northern sector has been well studied with geophysical surveys [32], whereas the southerly sector, in Mexican waters, is poorly known [32]. The PFB is part of the allochthonous salt sheets of the Sigsbee salt nappe [33]. Sediment input comes from the Hudson and Rio Grande embayments, located to the north and west, respectively [32,44,45].

Sample Collection
Bivalve comprise part of the benthic macrofauna and their shells, along with some live animals, were collected from soft-bottom sediments at water depths between 43 and 3563 m, and assigned to two water-depth zones: (1) the continental shelf samples (43-200 m) and (2) bathyal zone (375-3563 m) ( Figure 1, Table 1). Samples were collected from the Research Vessel "Justo Sierra" (Universidad Nacional Autónoma de Mexico) during oceanographic cruises Perdido 3 in June 2017 (29 sites), and Perdido 4 in September 2017 (27 sites) ( Table 1). At each sampling site, sediment was collected using a 0.25-m 2 Hessler Sandia MK-III box corer. Site, latitude, longitude, water depth, zone, and cruise information for each sample were recorded (Table 1). Each sediment sample was subsampled for: bacteria, chemical contaminants, metals, sediment texture, and benthic fauna community (three subsamples). Subsequently, sediment samples were sieved through 500-and 300-µm mesh sizes. Retained shells were placed in labeled plastic bags and examined later in the laboratory. Shells were identified to species level, following standard guides [38,[46][47][48][49]. For each species or morphotype, it was noted whether the individual was alive (soft part intact) or dead (empty shell), and the degree of shell damage was evaluated to assess whether the shell had been subject to post-mortem transport. Shell damage can also be used as a surrogate measure for time of death. We assumed that a well-preserved shell (i.e., no signs of abrasion or fractures) indicated that the organism had lived in the area recently. Worn and broken shells suggested transport from elsewhere, or that the organism may have lived in the area, but died long ago. We compared the observed assemblages with an exhaustive inventory of Bivalvia species from the GOM that had information on depth range and habitat preferences [39]. The list, however, did not distinguish between provenance of live and dead shells in the records, and provided only general location data for species presence/absence information, without specific geographic coordinates, thus precluding site-specific comparisons. Nomenclature for our sampled species was assigned according to [50] and The World Register of Marine Species (WoRMS) [51]. Specimens were deposited in the collection "Moluscos de la Peninsula de Yucatán" (CMPY), UMDI-Sisal, UNAM, Yucatan, Mexico and in the collection of the laboratory "Biodiversidad Marina y Cambio Climático "BIOMARCCA", ECOSUR-Campeche, Mexico.

Data Analysis
Spatial patterns of bivalve diversity were described using the following metrics: (1) α-diversity, as the mean number of species per site; (2) β-diversity, as the variation in species composition among sites within the study area [52,53], and (3) γ-diversity, as the expected number of species in each zone (i.e., shelf and bathyal). Specifically, γ-diversity was estimated using two incidence-based, non-parametric estimators: Chao2 and 1storder Jackknife (Jack1) [54], mainly because these estimators are less biased than other approaches [55,56]. β-diversity was estimated using the Sørensen coefficient because it (i) excludes joint absences; (ii) allows direct comparisons across studies [57]; and (iii) can be partitioned to identify two opposing processes in β-diversity patterns: species replacement and species loss [58]. We tested the null hypothesis that variability in α-diversity within each of the shelf and bathyal zones was the same, using Levene's Test [59]. Then, we tested the null hypothesis that α-diversity was equal between the zones, using a two-way analysis of variance (ANOVA) based on permutations [60], using zones and cruises as orthogonal fixed factors.
Because we lacked sampling replicates at each site, the sampling sites within depth zones were treated as replicates, implying that the spatial variability at the scale of sites is indistinguishable from the sampling error. Differences in γ-diversity across zones of the PFB was tested using the Welch test, a t-test adjusted for heterogeneous variances [59]. This test was applied to each of the estimators of species γ-diversity (i.e., Chao2 and Jack1). Differences in β-diversity between the two zones (i.e., continental shelf and bathyal) was tested using a multivariate dispersion test (PERMDISP), using the site × site matrix of Sørensen dissimilarities [53]. In this case, the null hypothesis was homogeneity in the multivariate dispersions among sites in each of the two zones. Finally, differences in species composition between zones and cruises were tested using Permutational Multivariate Analysis of Variance (PERMANOVA) [61]. Of particular interest was to test the interaction zones × cruises, having as the null hypothesis that difference in species composition across zones was constant in both cruises. The p-values for the F-ratio in the ANOVA, the F-ratio of the univariate and multivariate Levene's tests, as well as for the pseudo-F-ratio of PERMANOVA were obtained by 9999 permutations of residuals under the reduced model. Additionally, the matrix of Sørensen dissimilarities was partitioned into two separate components, spatial turnover (i.e., species loss) and nestedness of species (i.e., species replacement), using the approach of [58]. To identify groups of common species across zones, hierarchical cluster analysis of species was performed based on Whittaker's Index of Association. Then, a shade plot, i.e., heat map, was used to represent those groups of species according to the zone in which they appear [62].
To evaluate the potential structuring of the spatial processes, a multivariate autocorrelation test based on permutations (9999) was applied using the geographic distances between sites (using the Euclidean distance to the geographic coordinates) against the Sørensen dissimilarities matrix [63]. The null hypothesis was that there is no relationship among Sørensen dissimilarities and the spatial proximity of sites. Statistical analyses were performed with the software Primer v.7 & PERMANOVA+ and with R [64], using the packages vegan [65] and betapart [66].

Patterns of α-and γ-Diversity
For samples collected during both cruises, the bivalve α-diversity was consistently significantly higher at shelf sites than at bathyal sites (Table 3) (Table 4A, two-way ANOVA, factor zones, p < 0.05). Species richness ranged from 1-7 spp/site (sd = 1.5 spp/site) in the shelf sites, and from 1-5 spp/site in the bathyal sites (sd = 1.1 spp/site), but there was no statistically significant difference between sites within a zone (Levene test, Fzones = 1.90, p > 0.05). There was also no statistical difference when γ-diversity between zones was considered (Welch test, Chao2 = 1.87 with df = 19.82 and Jack1 = 0.10 with df = 17.00, both p > 0.05).

Patterns of β-Diversity
Multivariate dispersion was consistently higher on the bathyal than on the continental shelf (PERMDSIP, Fzones = 28.17, p > 0.05, Table 3, Figure 6). This figure illustrates the higher variability in species composition for sites from the bathyal zone (i.e., greater MVD), but the opposite on the continental shelf, i.e., greater similarity among sites.

Patterns of β-Diversity
Multivariate dispersion was consistently higher on the bathyal than on the continental shelf (PERMDSIP, Fzones = 28.17, p > 0.05, Table 3, Figure 6). This figure illustrates the higher variability in species composition for sites from the bathyal zone (i.e., greater MVD), but the opposite on the continental shelf, i.e., greater similarity among sites.  In general, the partition of the multi-site Sørensen coefficient (0.965) indicated that most β-diversity corresponds to replacement of species (0.947), rather than a loss of taxa (0.018), with increasing water depth (Figure 7). Species composition showed significant differences between zones (two-way PERMANOVA, factor Zonation, p < 0.05, Table 4B) and cruises (two-way PERMANOVA, factor Cruise, p < 0.05, Table 4B). The two factors were shown to be independent (interaction FCruise × Zonation= 1.22, p > 0.05), but the effect of zone size (interpreted from the √ CV) was considerably higher than that of cruise. There was no relationship between species composition and spatial distance among sites ( = 0.1, p > 0.05). In general, the partition of the multi-site Sørensen coefficient (0.965) indicated that most β-diversity corresponds to replacement of species (0.947), rather than a loss of taxa (0.018), with increasing water depth (Figure 7). Species composition showed significant differences between zones (two-way PERMANOVA, factor Zonation, p < 0.05, Table 4B) and cruises (two-way PERMANOVA, factor Cruise, p < 0.05, Table 4B). The two factors were shown to be independent (interaction FCruise × Zonation= 1.22, p > 0.05), but the effect of zone size (interpreted from the √CV) was considerably higher than that of cruise. There was no relationship between species composition and spatial distance among sites (ρ = 0.1, p > 0.05).

Faunal Composition
This study provides new information about bivalve species richness off the coast of Tamaulipas, Mexico, and yielded the first bivalve species list for the deeper zones of the PFB, complementing the earlier study of benthic macrofauna in the PFB [40]. Live bivalves were extremely rare, with only a single live specimen for each species of Bathyarca glomerula, Macoploma extenuata, and Orthoyoldia solenoides, collected on both the continental shelf and bathyal.
Species found on the shelf of the PFB had taxonomic affinities with those from the shallow coastal zone of Tamaulipas. Specifically, six species (Nucula proxima, Saccella acuta, S. concentrica, Limopsis cristata, Macoploma tageliformis, and Caryocorbula contracta) were reported from this region by Correa-Sandoval and Rodríguez-Castro [67]. We also collected Orthoyoldia solenoides in this study, which has been found in the GOM from Tampico to Matamoros [68,69]. Species Saccella acuta, Limopsis sulcata, and Poromya rostrate have been reported from deeper waters of the Campeche Bank [70]. Caryocorbula contracta

Faunal Composition
This study provides new information about bivalve species richness off the coast of Tamaulipas, Mexico, and yielded the first bivalve species list for the deeper zones of the PFB, complementing the earlier study of benthic macrofauna in the PFB [40]. Live bivalves were extremely rare, with only a single live specimen for each species of Bathyarca glomerula, Macoploma extenuata, and Orthoyoldia solenoides, collected on both the continental shelf and bathyal.
Species found on the shelf of the PFB had taxonomic affinities with those from the shallow coastal zone of Tamaulipas. Specifically, six species (Nucula proxima, Saccella acuta, S. concentrica, Limopsis cristata, Macoploma tageliformis, and Caryocorbula contracta) were reported from this region by Correa-Sandoval and Rodríguez-Castro [67]. We also collected Orthoyoldia solenoides in this study, which has been found in the GOM from Tampico to Matamoros [68,69]. Species Saccella acuta, Limopsis sulcata, and Poromya rostrate have been reported from deeper waters of the Campeche Bank [70]. Caryocorbula contracta and Saccella acuta were the species with the highest number of specimens collected, 57 and 52, respectively, especially at sites near the coast, adjacent to the Laguna Madre. Eight genera (Nucula, Malletia, Neilonella, Tindaria, Bathyarca, Limopsis, Astarte, and Cuspidaria) found in the PFB were reported by Wei et al. [2,37] from the northern GOM.
Renewed interest in exploitation of deep-sea ecosystems (e.g., oil extraction) has created a race against time for researchers to inventory deep-sea fauna and explore the spatial distribution of species, the processes that sustain them, and the use of benthic taxa as potential indicators of oil spills. Some families of bivalves found in this study have been registered as apparently sensitive (Nuculanidae), cosmopolitan (Nuculidae), and tolerant (Thyasiridae) of deep-sea hydrocarbon blowouts (e.g., Deepwater Horizon oil spill) [71]. This study constitutes an initial effort to document bivalve mollusk diversity on the PFB before it is changed by human activities.

Insights from Death Assemblages
There has been considerable discussion about the representativeness of death assemblages as surrogates for regional biodiversity of mollusks [22,70,71]. Most of the shells in this study were in excellent condition, with morphologic details well preserved. Only three species showed signs of abrasion (Figures 2-5), suggesting that shells of bivalves sampled in the PFB represent modern mollusk diversity in the region.
Previous studies [22,27,40,72] also found very low numbers of live bivalve mollusks in these types of habitats, made no distinction between live and dead specimens, and covered a huge area [39,73,74]. Simply because its enormous size, there is a lack of detailed information about the geographical distribution of bivalves across important regions of the Mexican part of the PFB. Results presented here fill such information gaps, with data on the diversity and composition of bivalve communities derived from death assemblages. More taxonomic research and more sampling are needed to explore patterns at finer spatial and bathymetric scales. The study of Wei et al. [2] provides a good example. They collected 271 box-corer samples from 51 stations, which yielded 8055 live bivalve specimens from the deep northern GOM.

Spatial Patterns of Species Diversity and Potential Drivers of Variation
The γ-diversity is similar between the continental shelf and bathyal zones, but it seems that the processes that maintain such species richness differ between the two regions. Our results suggest that over the water depth range on the continental shelf, γ-diversity apparently is generated by a combination of high α-diversity and low β-diversity. On the other hand, deeper-water habitats are characterized by low α-diversity and high βdiversity. Such patterns might reflect spatial differences in nutrient input to the benthic zone from surface production, and topographic differences between the two zones. On the continental shelf, there is high primary production as a consequence of nutrient availability, whereas the bathyal zone is characterized by low rates of organic carbon flux [12,75,76]. Furthermore, there are marked differences in the geomorphology of the two zones, with the continental shelf possessing a largely flat, mud bottom, and the bathyal having a broad range of water depths with diverse habitat types of different sizes and degrees of separation/connectivity. Relatively high primary production might provide enough energy to sustain bivalve assemblages on the continental shelf, with a low rate of species replacement. This suggests that species interactions such as predation and competition may be important factors that shape these communities. For example, bivalves are usually the third-most abundant group of macrofauna [2,40]. With their numerical dominance, bivalves are a major food source for asteroids and decapod crustaceans [2].
In contrast to the shelf, nutrient supply might be lower in the bathyal zone, but habitat heterogeneity may facilitate replacement of species with different niche breadth [77], which could explain Bivalvia assemblages from deep-water sites with γ-diversity equivalent to that estimated for the continental shelf. Similar patterns have been described for seamounts [78], and interpreted in terms of variation in α-diversity related to water depth [77,79,80]. We suggest that in our study region, the processes that sustain species replacement within each water-depth range differ, but result in similar γ-diversity values for the two zones of the PFB. β-diversity in the deep sea has been defined as the spatial replacement of species along depth gradients [81]. This process has been associated with sporadic nutrient pulses that arrive from the continental shelf and that accumulate in the rifts, fissures, channels, or folds of the PFB [8].

Conclusions
This study provided a taxonomic and ecological baseline for bivalve community composition and diversity in the PFB, and contributed to a better understanding of the complex ecological patterns in the Gulf of Mexico. We demonstrated that death assemblages can help elucidate the diversity of bivalve species on the continental shelf and bathyal zone of the PFB, collected along a gradient of water depth during two cruises. α-diversity was significantly higher at continental shelf sites than at bathyal sites, a pattern that has been documented in other deep-sea areas and linked to declining rain of organic inputs with greater water depth. The continental shelf has comparatively low β-diversity, indicating that sites in the region are similar in term of bivalve diversity. Species replacement contributed more than turnover in shaping the dissimilarity of the bivalve death assemblages. However, the opposite has been found for living organisms in the northern Gulf of Mexico. More investigation would not only improve the resolution of this investigation, but would also resolve uncertainty about the relationship between live and dead bivalve assemblages. We recommended that future studies reduce the number of sampling sites in shallower waters and direct greater sampling effort to deep-water areas. Overall proper assessment of the bivalve fauna in the area will probably involve a much larger sampling effort, and perhaps the use of different sampling techniques, though this will present challenges in terms of cost and time investment. It is necessary to standardize the definition of bathymetric zones to facilitate comparisons among similar studies from sites around the world.

Data Availability Statement:
The data presented in this study are available on request from the corresponding author.
are very grateful to Editor Mario Pei. We also thank the three anonymous reviewers for their helpful comments and suggestions.        Overall geographic range. The species has been reported from North Carolina to Texas, Cuba, and Brazil, at depths of 0-120 m, benthic and infaunal [39].

Conflicts of Interest
Family Myidae Lamarck, 1809 Sphenia sp. Overall geographic range. The species has been reported from Massachusetts to Florida Keys, Texas, Cuba, West Indies, and Brazil, at depths of 0-14 m, benthic and infaunal [39]. Overall geographic range. Previously considered a synonym of Corbula disparilis d'Orbigny, 1853, Corbula limatula Conrad, 1846, and Varicorbula disparilis d'Orbigny, 1853, the species has been reported in North Carolina to Texas, Gulf of Mexico offshore banks, Cuba, West Indies, and Brazil, at depths of 0-549 m, benthic and infaunal [39].
Corbula sp. Overall geographic range. The species has been reported from Florida Straits to Florida Keys, North Gulf of Mexico, Cuba, Antilles, and Central America, at depths of 0-101 m, benthic and infaunal [39].