Composition of decapod crustacean assemblages in beds of Pinctada imbricata and Arca zebra (Mollusca: Bivalvia) in Cubagua Island, Venezuela: Effect of bed density ; Composición de ensambles de crustáceos decápodos en bancos de Pinctada imbricata y Arca zebra (Mollusca: Bivalvia) en la Isla de Cu

1 Departamento de Ciencias, Unidad de Cursos Básicos, Núcleo de Nueva Esparta, Universidad de Oriente, Venezuela. E-mail: ivanhernavila@yahoo.com 2 Grupo de Investigación en Carcinología, Universidad de Oriente, Venezuela. 3 Escuela de Ciencias Aplicadas del Mar, Núcleo de Nueva Esparta, Universidad de Oriente, Venezuela. 4 Maestría en Ciencias Marinas y Costeras, Universidad Nacional de Costa Rica. 5 Instituto Nacional de Investigaciones Agrícolas, Venezuela.


INTRODUCTION
Bivalve aggregations constitute a microhabitat for a wide variety of organisms in intertidal, subtidal and deep-water marine benthic habitats (Tsuchiya and Nishihira 1985, 1986, Thiel and Ulrich 2002, Turnipseed et al. 2004, Galkin and Goroslavka 2008).The aggregations increase the spatial heterogeneity of the benthic environment and provide shelter and food for a diverse assemblage of organisms (Jacobi 1987, Jones et al. 1997, Seed 1996).Local diversity, population dynamics, food webs and nutrient cycling could be affected by the presence of bivalve beds, thus increasing the abundance and diversity of associated fauna (Tsuchiya and Nishihira 1985, Thiel and Ulrich 2002, Gutiérrez et al. 2003, Carranza et al. 2008, 2009).
In the southeastern Caribbean, conspicuous beds of the turkey wing (Arca zebra) and the Atlantic pearl oyster (Pinctada imbricata) are present in shallowwater marine environments.The Atlantic pearl oyster has been exploited for pearl production since the 15th century (Mackenzie et al. 2003), and it supports a small artisanal fishery mainly for local food consumption.However, the turkey wing supports a locally important artisanal fishery with a volume ranging from 35000 to 40000 tm yr -1 , depending on the demands of the canning industry (Arias et al. 2002).Both species have a wide distribution and form beds with associated benthic fauna (Prieto et al. 2001b, Díaz and Liñero 2003, Liset et al. 2009).
The fauna associated with beds of different bivalve species has been studied over a large spatial gradient (Thiel and Ulrich 2002), comparing different sizes of aggregations of bivalves (Tsuchiya and Nishihira 1985), incorporating other factors such as age, size and density of patches (Tsuchiya andNishihira 1986, Borthagaray andCarranza 2007), and comparing different beds of different bivalve species in different locations (Turnipseed et al. 2004) and at different depths ( Van Dover and Trask 2000).However, studies of the differences between assemblages associated with patches of different bivalve species in the same area are scarce and generally deal with temperate waters.The role of bivalve aggregations in tropical benthic ecosystems is poorly understood.
Many publications have shown the importance of topographical complexity in the structure of benthic fauna (Bourget et al. 1994, Chapman and Underwood 1994, Pech et al. 2001).Richness, abundance, and diversity in benthic communities are positively related to complexity (Archambault and Bourget 1996).In bivalve beds the increase in density or size is positively correlated with topographical complexity (Gutierrez et al. 2003, Borthagary andCarranza 2007).Increase in density of bivalve beds could offer more crevices and substratum for the associated fauna, diminishing inter-and intraspecific competition for space and bearing a positive relationship with abundance, species number and diversity.In the present study, we tested the hypothesis that abundance, diversity and structure of decapod crustacean aggregations living in a bivalve bed are associated with the density of the bed, by comparing different densities of beds of two species of bivalves at a single tropical locality (Cubagua Island, Venezuela).

Area of Study
Cubagua is a semi-arid island of 22 km 2 situated in eastern Venezuela (10°47'-10º51'N; 64°8'-64°14'W).It is strongly influenced by coastal upwelling and is characterized by a shallow continental shelf with sandy bottoms of mainly coarse sediments (Cervigón 2005).The upwelling process occurs mainly during the first months of the year (January-May) and consists of the seasonal intrusion of water with low temperatures and oxygen levels, and high nutrient concentrations.The surface water temperature ranges between 22°C and 26°C and exhibits continuous high productivity (Gómez 1996).The Cubagua littoral zone is composed mainly of shallow waters (0-10 m deep) occupied by marginal reefs, rocky shore, Thalassia testudinum beds, Arca zebra and Pinctada imbricata beds and sandy areas.In the southeastern region of the island, patches of A. zebra and P. imbricata are found 200 to 500 m from the coast.Previous studies on the decapod fauna of Cubagua Island have focused on an inventory of the species and their taxonomic aspects (see Hernández-Ávila et al. 2007).

Sampling and data analysis
Decapod crustacean assemblages were compared between different bivalve beds and between the different densities of each bed.Bivalve beds were evaluated at two levels (P.imbricata and A. zebra bed), and density was considered at three discrete levels of bivalve abundance composing the bed (1, 60-100; 2, 100-200; 3, >200 ind m -2 ), corresponding to a two-way factorial design with seven replicates for each combination.Samples were collected between September 2005 and May 2006 during periods of non-upwelling (Sep-Nov) and upwelling (Dec-May).Due to logistics, it was not possible to evaluate temporal variation as an additional factor.To exclude potential seasonal effects in this design, randomizing the replicates between the periods was considered.Although randomization could generate additional patterns in experimental design (Hurlbert 1984), there was no pattern detected between collection periods (χ 2 rxc test =9.17,df=5, P=0.12).
Samples were collected within a 0.5×0.5 m plot by scuba diving at depths of 5 to 12 m.The plots were covered with the mouth of a plastic bag and removed from the bottom into the plastic bag avoiding escape of motile fauna.All benthic components associated with bivalve beds (including the bivalves) were collected for evaluation in the laboratory.The bivalves forming the bed were counted to determine their density and the dominance of the species forming the bed.Because there were few cases with low density, plots characterized by mixed beds (with a similar ratio of P. imbricata and A. zebra) were omitted.Decapod crustaceans were separated and fixed for later counting and identifica-tion.Additional qualitative samples were collected in both types of bed by scuba diving to collect species that were not sampled by the quantitative sampling techniques.
These indicators were contrasted using a two-way orthogonal permutational analysis of variance (PER-MANOVA) (Anderson 2001, McArdle andAnderson 2001) to test the null hypothesis of no differences in univariate descriptors (total abundance, species number, evenness, Shannon diversity, taxonomic diversity and taxonomic distinctness in each case) associated with the factors "density", "bivalve bed", and of their interaction.The pair-wise t statistic was used as an a posteriori analysis for detecting groups between levels of densities.
To test the null hypothesis of no differences in structure of decapod assemblages in the study design described above, a similarity matrix was constructed using the Bray-Curtis coefficient and tested with a PERMANOVA.A dummy species was added to each sample to avoid indeterminacy of the Bray-Curtis values between two samples without species (Clarke et al. 2006).The number of permutations of residuals used to determine the statistical significance of each term was 999 under a reduced model (Anderson 2001).A multivariate dispersion test (PERMDISP) was used to determine whether a significant source of variation was related to difference in multivariate dispersions or difference in centroid position.A SIMPER analysis was performed to identify species associated with differences between densities (Clarke 1993).

RESULTS
In the quantitative samples, 35 species belonging to 18 families (Table 1) were identified; two species could not be identified because the specimens were damaged (one hippolytid and one caridean, family not identifiable).The families represented by the largest number of species were Majidae (8 species) and Alpheidae (five species), whereas other families were represented Differences in univariate descriptors of decapod assemblages associated with the species of each bed were not detected.Richness, abundance and taxonomic distinctness exhibited significant differences related to bivalve density.However, differences associated with bivalve density were not detected for Shannon diversity, evenness, or taxonomic diversity (Table 2), which showed low values (mean ± se H'=1.08±0.15,J'=0.50±0.06,∆=29.89±4.22)because of the large ratio of M. forceps in most of the samples and the low frequency of many species.
The multivariate analysis (PERMANOVA) showed differences between the species composition of decapod crustaceans related to bed density, but no differ-ences between bivalve beds of different species (Table 4).The differences detected were not associated with differences in multivariate dispersion (PERMDISP, F=0.132, P=0.936).The assemblages in beds with medium and high density were distinct from those in the lower density beds (t=2.76,P=0.016; and t=2.92,P=0.006 respectively) (Fig. 2A).According to the SIMPER analysis, the species most related to the differences between groups were Mithraculus forceps (mean contribution of dissimilarity 47.6%), Pilumnus caribaeus (8.55%), Cuapetes americanus (6.67%), and Petrolisthes galathinus (5.24%); the mean individual contributions of the other species of the dissimilarity were less than 5%.The abundance of the former species increased in the assemblages on beds with medium and high density (Fig. 2B-E).

DISCUSSION
We found similarities in decapod crustacean assemblages between beds of A. zebra and P. imbricata, which could be serving as alternative habitats.Most of the common decapod crustacean species associated with bivalves have been reported from other habitats in the region.M. forceps has a wide distribution in the Caribbean and is frequent in most of the shallow-water substrates of northeastern Venezuela (Rodríguez 1980, Hernández et al. 2000, Hernández-Reyes et al. 2001, Hernández-Ávila et al. 2007).Additionally, P. caribaeus, P. herbstii, P. lherminieri, P. galathinus and C. americanus have been reported from various locations in Venezuela and are associated with coral, rubble and Thalassia beds (Rodríguez 1980, Carmona-Suárez and Conde 1996, Hernández-Ávila et al. 2007).Some infrequent species have recently been reported in the region (Hernández-Ávila 2004, Hernández-Ávila and Campos 2006, Hernández-Ávila et al. 2007).Thus, these bivalve aggregations represent a potential habitat for some common shallow-water decapod crustaceans and for some uncommon species.However, no species have been detected living in exclusive association with beds of A. zebra or P. imbricata, respectively.At Cubagua Island, 60% of the decapod species found in bivalve beds have also been collected in subtidal rocky biotopes, 42.9% in coral patches and 28.6% in Thalassia beds.
Various publications have identified the role of oyster reefs as a potential habitat or as a refuge for decapods (Ruiz et al. 1993, Dittel et al. 1995, Eggleston et al. 1998, Posey et al. 1999).The physical structure of oyster reefs serves as substratum for the recruitment  of grass shrimps and blue crab megalopae in laboratory and field experiments (Eggleston et al. 1998, Welch et al. 1997).Laboratory habitat experiments suggest that shrimps actively select oyster habitats in response to the presence of fish predators (Posey et al. 1999).Our results suggest that bivalve aggregations at Cubagua Island provide an additional habitat for decapod crustaceans living in other shallow habitats such as Thalassia beds, corals, and rocky environments.Different assemblages of decapod species were associated with beds with contrasting bivalve densities.Beds with a low density exhibited lower levels of rich-ness, abundance and taxonomic distinctness of decapod crustaceans than beds with 100 or more ind m -2 .Since the relative importance of the common species of the assemblages was similar in beds with different levels of density, differences observed in decapod composition were due to the increased decapod abundance and to the incorporation of species that were not recorded in low-density beds.The bivalve aggregations in highdensity beds generate a more complex topography.Spatially complex surfaces tend to support richer faunal communities, presumably because they provide a greater number of crevices (Abele 1974, Ricciardi et  al. 1997).Aggregated mussels have an abundance of interstitial spaces that may serve as a refuge from disturbances and predation in mobile macrofauna (Gosselin andChia 1995, Borthagaray andCarranza 2007).
The increase in the number of crevices in high-density beds could also decrease the competition for space between members of a decapod assemblage.Moreover, the increase in the surface for attachment and the control of transport of particles could be provided by bivalve aggregations and associated with bed density (Crooks andKhim 1999, Gutiérrez et al. 2003).Hernández-Ávila et al. (2007) recorded 14 species of decapod crustaceans in aggregations of Arca zebra along the coastal margin of Cubagua Island, M. forceps being the most common species found.Only six species found in this previous study were collected in the present study.Differences between the coastal margin and the depth investigated in the previous study (0-100 m off the coast and at 2-5 m depth) and those investigated in the present study (300-500 m off the coast and at 5-12 m depth) could have been responsible for the differences in the species sampled.With respect to assemblages of decapods associated with Pinctada imbricata, only a few sporadic records of decapods associated with these beds exist.Spatial variation, patterns of aggregation of other taxa, and dynamics of both bivalve beds and associated fauna could provide a better understanding of this habitat.
Bivalve beds of P. imbricata and A. zebra are exploited by artisanal fisheries, though A. zebra is processed industrially and subjected to wider commercialization than P. imbricata.In the study of the biology of these two species a serious effort has been made to develop conditions for their culture and to regulate fisheries (Jiménez et al. 2000, Prieto et al. 2001a, Lodeiros et al. 2002, Marcano et al. 2005).The associated fauna reported previously for both species (Prieto et al. 2001b, Díaz and Liñero 2003, Liset et al. 2009) and the effect of bed density in decapod assemblages detected in the present study suggest the importance of controlling bivalve fisheries for the maintenance of benthic communities.The local bivalve fisheries operate with small trawls (about 0.9 m mouth width) that remove most of the benthic components from the beds.Although the large megafaunal bycatch is returned during fishing, decreasing density of beds could result in less abundance, richness and complexity of decapod crustacean assemblages associated with the beds.

Table 1 .
-Mean abundance (ind 0.25 m -2 ) of decapod crustaceans collected in oyster beds and turkey wing beds at three densities.

Table 4 .
-PERMANOVA of decapod assemblages related to species bed and density.Italics denote significant differences.