An Investigation of the Fisheries Ecosystem Dynamics in Rhode Island's Nearshore Waters

This research was developed in response to recent interest in offshore wind energy development and the ongoing need for ecosystem-based spatial management planning in Rhode Island’s nearshore waters. Despite heavy use and close proximity to a number of marine science institutions, Rhode Island and Block Island Sounds have been neglected in terms of scientific research, resulting in a poor understanding of the fisheries ecosystem in this area. This research aimed to address this knowledge gap by assessing the biogeography, trophic dynamics and habitat associations of the fish and invertebrate communities in this region. Specifically, the goals of this research were to: 1) Evaluate the fine-scale spatial structure of the demersal fish and invertebrate community, 2) Assess the dietary guild structure and the flow of energy through the fisheries food web, and 3) Investigate the relationship between the fish community and benthic habitat. Otter trawls and beam trawls were used to sample fish and invertebrates throughout Rhode Island and Block Island Sounds from 2009 to 2012. Field work was conducted in collaboration with two commercial fishing vessels, the F/V Darana R and F/V Mister G, and the Northeast Area Monitoring and Assessment Program. During otter trawl surveys, stomach and white tissue samples were taken from 25 species for analysis of diet composition and nitrogen and carbon stable isotope signatures. A combination of site-specific water column profiles, high resolution acoustic surveys, and seafloor video surveys were used for habitat characterization. Regionally-grouped abundance, biomass, diversity, and size spectra were used to assess spatial patterns in the aggregate fish community, and nonparametric hierarchical cluster analysis was used to determine species assemblages. Analyses revealed coherent gradients in fish community biomass, diversity and species composition extending from inshore to offshore waters, as well as patterns related to the differing bathymetry of Rhode Island and Block Island Sounds. Species assemblages were characterized by a combination of piscivores (silver hake, summer flounder), benthivores (American lobster, black sea bass, little skate, scup) and planktivores (sea scallop), and exhibited geographic patterns that were persistent from year to year, yet variable by season. Such distributions reflect the cross-shelf migration of fish and invertebrate species in the spring and fall, highlighting the importance of considering seasonal fish behavior when planning construction schedules for offshore development projects. Stomach content analysis was used to define trophic structure according to dietary guilds, while nitrogen and carbon stable isotopes were used to determine the trophic position of fish and invertebrate species and to assess the relative importance of benthic and pelagic production in supporting the fisheries food web. Results suggest that the fisheries food chain in Rhode Island and Block Island Sounds consists of four trophic levels and six distinct dietary guilds (planktivores, benthivores, shrimp and amphipod eaters, crab eaters, small fish and shrimp eaters, piscivores). Inter-species isotopic and dietary overlap within guilds was high, suggesting that resource partitioning plays a major role in structuring the fish community in this region. Furthermore, carbon isotopes indicate that most fish are supported by pelagic phytoplankton, although there is evidence that benthic production also plays a role, particularly for obligate benthivores such as skates. Multivariate analysis of otter and beam trawl catch data and acoustic, videographic, and oceanographic benthic habitat parameters suggest that the fish communities in Rhode Island and Block Island Sounds are structured by both permanent (i.e. depth, habitat type) and transient (i.e. bottom water temperature) habitat characteristics. As such, otter trawl and beam trawl species assemblages can be explained by a suite of seafloor and oceanographic habitat parameters, including mean depth, surface and bottom water temperature, standard deviation of benthic surface roughness, minor grain size, mean slope, and surface salinity. Furthermore, spatial patterns in diet composition indicate habitat-specific feeding by demersal fish species, such as winter flounder and silver hake. Feeding on benthic prey is, therefore, an important link between demersal fish assemblages and their habitats in this region. The results of this work not only provide valuable insight into fisheries ecosystem dynamics in a temperate nearshore environment, but will also inform spatial management plans for Rhode Island and Block Island Sounds. Furthermore, the methods for this study are consistent with European guidelines for assessing the impacts of offshore wind turbines on the marine environment and could provide a baseline for measuring the cumulative effects of offshore development projects within Rhode Island and Block Island Sounds.

Common and scientific names, size ranges, isotope and stomach sample sizes, total number of prey types, and Levins standardized niche breadth for each predator species included in dietary guild and stable isotope analyses.   proportional contribution of each prey category to species-specific diet as derived from wet weight measurements. Table 2.5. Stomach contents of silver hake, smooth dogfish, spiny dogfish, striped bass, summer flounder, weakfish, winter flounder, winter skate, and yellowtail flounder. Values represent proportional contribution of each prey category to species-specific diet as derived from wet weight measurements. Table 3.1. Sources, resolutions, and coverage of all habitat parameters for otter trawls and beam trawls. Habitat variables marked with an asterisk (*) were retained in the otter trawl BIOENV analysis. Habitat variables marked with a cross ( †) were retained in the beam trawl BIOENV analysis. Habitat variables marked with a superscript c (ᶜ) are categorical and were used in univariate Analysis of Variance (ANOVA) and multivariate Analysis of Similarity (ANOSIM). Table 3.2. Categorical and numerical habitat types and corresponding grain sizes. Photos of each habitat type are provided in Figure 3.3. Table 3.3. P-values from Analysis of Variance (ANOVA) models testing for the effects of categorical habitat variables (depth strata, major habitat type, minor habitat type, and number of habitat types) on otter trawl and beam trawl fish community abundance, biomass, and species diversity. Bold text signifies a significant result (p<0.05).
xii Table 3.4. Adjusted R-squared and p values of linear regressions between continuous habitat variables and log transformed otter trawl fish community abundance (number per km 2 ), biomass (kg per km 2 ), and diversity (Shannon Weiner's H). Table 3.5. Summary statistics of stepwise multiple linear regression models that were used to assess the cumulative impact of 24 habitat parameters on otter trawl and beam trawl abundance, biomass, and species diversity. Optimal regression models were selected using Akaike's Information Criterion corrected for small sample bias (AICc).         Island Sound and Block Island Sound. Symbols represent dietary guilds and subgroups as identified by (primary prey): Light green diamonds = planktivores (pelagic zooplankton), Dark purple triangles = benthic omnivores (small crustaceans and worms), Dark green diamond = planktivore (benthopelagic zooplankton), Light purple triangles = benthivores (gammarid amphipods), Grey square = piscivore (small fish and crustaceans), Blue squares = piscivores (fish), Orange circles = benthopelagic omnivore (shrimp and fish), and Red circles = crustacean eaters (decapod crabs and shrimp).  Mean proportional composition (by wet weight) of major prey items for: a) planktivores and Atlantic herring, b) benthivores and benthic omnivores, c) crustacean eaters and silver hake, and d) piscivores and weakfish. The planktivore guild includes butterfish, American shad, and alewife. The benthic omnivore group includes scup, winter flounder, little skate, and winter skate. The benthivore group includes yellowtail flounder and haddock. The crustacean eater guild includes smooth dogfish and black sea bass. The piscivore guild includes bluefish, striped bass, monkfish, summer flounder, and spiny dogfish.              xvi similar habitats appearing close together. Each point represents the habitat features at one otter trawl station. Symbols represent habitat group, as defined by BIOENV and LINKTREE analysis. Dashed contours represent species assemblage groups as defined by CLUSTER analysis. Analysis of similarity indicates that there are significant differences in habitat characteristics between otter trawl species assemblage groups (R=0.475, p=0.001).   Table 3.6.  Table  3.6.  Table  3.6. Multidimensional scaling plot depicting the pattern in beam trawl fish and invertebrate species composition in Rhode Island and Block Island Sounds, with similar species compositions close together. Each point represents the species composition of one beam trawl. Symbols represent major habitat type (pink circles = pebble, dark blue inverted triangles = coarse sand, green triangles = medium sand, red diamonds = fine sand, and light blue squares = mud). Analysis of similarity indicates that beam trawl fish community composition is significantly different between major habitat types (R=0.229, p=0.023).
Figure 3.14. Multidimensional scaling plot depicting the pattern in beam trawl fish and invertebrate species composition in Rhode Island and Block Island Sounds, with similar species compositions close together. Each point represents the species composition of one beam trawl. Symbols represent number of habitat types (pink circles = 1, green triangles = 2, light blue squares = 3, dark blue inverted triangles = 4, red diamonds = 5). Analysis of similarity indicates that beam trawl fish community composition is significantly different at sites with different numbers of habitat types (R=0.223, p=0.015). xvii otter trawl station. Symbols represent habitat group, as defined by BIOENV and LINKTREE analysis. Dashed contours represent species assemblage groups as defined by CLUSTER analysis. Analysis of similarity indicated that there were significant differences in habitat characteristics between beam trawl species assemblage groups (R=0.506, p=0.001). on protected species, habitat, and non-target species (Crowder & Norse 2008). In addition, EBFM recognizes other ocean use sectors, such as mineral/energy extraction, tourism, recreation, and transport, and involves stakeholders in the fisheries management process (Pikitch et al. 2004). The EBFM process is designed to be more transparent than traditional fisheries management, so as to reduce stakeholder frustrations and ensure management accountability. While considering the human dimension of the fisheries ecosystem is not a new concept (the goal has been to manage for maximum yield/profit for many years), the cooperation of stakeholders from assessment to application is novel to EBFM .
Two key components of EBFM that are central to my dissertation research are the trophic dynamics and habitat requirements of the fish community. Traditionally, the fisheries management system has focused on single-species assessments and policies, with little acknowledgement of species and ecosystem interactions ). Conversely, EBFM has a distinct multispecies focus for assessments and policies, which over time, will progress to an ecosystem focus, incorporating, not only species interactions, but also climate and habitat (Johnson & Welch 2009).
Dietary guild analysis and stable isotope analysis are two common approaches for assessing the trophic structure in a fisheries ecosystem, with each technique providing a unique ecological perspective , Wilson 1999. Fish stomach content analysis, upon which dietary guild analysis is based, provides a direct measure of predator consumption (Hyslop 1980). A unique and powerful attribute of stomach content and dietary guild analysis is their utility in identifying specific trophic linkages (i.e. predator-prey relationships), which is critical for developing multispecies models and, thus, an ecosystem based approach to fisheries management (Fogarty 2013). Stomach content analysis, however, does not take into account temporal variation in predator diets, as stomach contents represent a snapshot of fish feeding behavior. Furthermore, stomach content analysis is often ineffective for planktivorous species, due to the size and digestive state of prey.
Nonetheless, stomach content and dietary guild analysis are valuable approaches to assessing the trophic structure of fisheries ecosystems, as the resulting classification of species into functional groups, assessment of resource partitioning, and identification of competitive interactions enables the progression of multispecies models and ecosystem-based fisheries management (Auster & Link 2009.
Stable isotope analysis is also a powerful tool for assessing trophic dynamics in fisheries ecosystems, with nitrogen stable isotopes (δ 15 N) indicating the timeintegrated feeding histories and trophic positions of consumer species, and carbon stable isotopes (δ 13 C) revealing the relative importance of different basal resources in supporting fish production , Post 2002, Mackenzie et al. 2011. Stable isotopes are assimilated in fish tissue over weeks to months, and thus reflect the time-integrated feeding history of consumer species (Peterson & Fry 1987). As such, in contrast to gut content analysis, stable isotopes are an effective means to assess temporal variability in fish feeding behaviors, which arguably, is equally important as identifying specific predator-prey relationships. Given the temporal integration of stable isotopes, δ 15 N and δ 13 C often reflect feeding behavior in different locations and ecosystems (i.e. estuaries v. continental shelf). This discrimination can be a useful tool for describing the movement patterns and habitat use of fish species, but can also confound analytical interpretation if baseline isotopic signatures are not known for different locations (Abrantes & Barnett 2011, Mackenzie et al. 2011, Dixon et al. 2015. Overall, δ 15 N and δ 13 C analysis is a useful tool for developing an ecosystem-based approach to management, as it identifies species that act as direct links to basal resources as well as species that share trophic roles. With regards to the habitat, traditional and ecosystem-based management tools for characterizing and protecting fish habitat have been similar (marine protected areas and rotating closures) . As habitat mapping capabilities have improved over the last 10 years, however, EBFM has begun to consider habitat in a more process-oriented manner ). More specifically, EBFM has begun to consider the role that habitat plays in not only the distribution of marine species, but also the productivity of the ecosystem (Erikkson et al. 2006).

A final and key component of EBFM that is particularly pertinent in Rhode
Island's nearshore waters is that EBFM identifies and incorporates other ocean use sectors from the outset . Perhaps the best example of this is marine spatial planning ). The purpose of marine spatial planning is to minimize conflicts between competing ocean uses and preserve ecosystem services by allocating the spatial and temporal distribution of human activities in marine areas (Beck et al. 2009. Many ocean uses impact benthic habitat (mineral/energy extraction, fishing, dredge disposal), and thus, it is an essential consideration in the marine spatial planning process .
Areas that are targeted by fishing often have particular environmental and biological conditions that contribute to the productivity of an ecosystem (e.g. George's Bank, Cox's Ledge) (Jennings et al. 2009). Benthic habitat maps, particularly those that incorporate oceanographic features and benthic biota, are essential to understanding the relationship between fishing effort, fish production, and ecosystem services , Freidlander & Brown 2003). An example of such an application is Cordell Bank, where a combination of habitat maps and submersible surveys has led to the development of closed areas to protect sensitive rockfish (Sebastes spp.) habitat (Iampietro et al. 2008, Anderson et al. 2009. A similar case is evident in the Gulf of Alaska, where the designation of closed areas for the protection of young halibut, Hippoglossus stenolepis, has been based on the distribution of benthic habitat maps (Witherell et al. 2000). And in our own backyard, the designation of essential fish habitat on Stellwagen Bank has been based on benthic habitat maps (Auster et al. 2001). Thus, fine-scale fish biogeography characterizations and biologically relevant habitat maps are essential data for the development of effective marine spatial plans. Overall, making tradeoffs between habitat protection, fisheries extraction, and other ocean uses will become increasingly important as our continental shelves get more crowded with offshore energy ventures .
A common motivator of marine spatial planning worldwide has been the development of offshore wind energy (Jay 2010). An example of this is in Danish waters, where thousands of offshore turbines are operational ). Without a thorough understanding of the fine-scale distribution and significance of benthic habitat to fish and other benthic biota, however, the sustainability of offshore wind farm development is debatable (Punt et al. 2009).
Ideally, managers direct developers to construct turbines in areas that will have the least negative impact on particular species or communities (e.g. scallop beds, cerianthid anemone aggregations), but the data required to make such recommendations are often lacking. Thus, my dissertation research aims to address this need in the first area in US territorial waters being planned for offshore wind energy development, Rhode Island and Block Island Sounds.
Rhode Island Sound and Block Island Sound separate the estuaries of Narragansett Bay and Long Island Sound from the outer continental shelf. As such, they provide important linkages between nearshore and offshore processes, including nutrient fluxes, larval transport, and migration of the adult stages of resource species, such as the American lobster (Homarus americanus) and winter flounder (Pseudopleuronectes americanus) (Costa-Pierce 2010). Furthermore, Rhode Island and Block Island Sound support a variety of commercial and recreational fishing activities, such as scallop dredging, otter trawling, long-lining and gill-netting, producing over $60 million in seafood landings every year (Hasbrouk et al. 2011).
Despite their heavy use and close proximity to a number of marine science institutions, Rhode Island and Block Island Sounds are neglected in terms of scientific research, resulting is a poor understanding of the distribution and dynamics of the fisheries resources in this area. My dissertation research seeks to fill this data gap through cooperative research and interdisciplinary collaborations.
Studies to support the management of Rhode Island's offshore waters have become a priority since 2000, when interest in developing artificial reefs, aquaculture sites, and offshore wind turbines emerged in this region. Combined with traditional fisheries and existing dredge-disposal sites, these multiple uses require integrated spatial management planning to site activities in appropriate habitats that will minimize, to the extent possible, the cumulative impacts on resident species and the ecological and economic services derived from this near-shore region (Crowder & Norse 2008 Although a general understanding of the ecology of Rhode Island Sound and Block Island Sound exists, there is a lack of site-specific data to guide spatial management planning (Mahon et al. 1998, Costa-Pierce 2010. My dissertation research aims to address this need by conducting comprehensive sampling of the demersal fish and invertebrate community and their associated habitats in Rhode Island Sound and Block Island Sound. Specifically, the goals of my doctoral research are to: 1) Evaluate the fine-scale spatial structure of the demersal fish and invertebrate community, 2) Assess the dietary guild structure and the flow of energy through the fisheries food web, and 3) Investigate the relationship between fish species assemblages and benthic habitat.
The results of this work will not only provide valuable insight into fisheries ecosystem dynamics in a temperate coastal environment, but will also guide spatial management plans for Rhode Island and Block Island Sounds. The products of this research will be immediately available to state and federal management agencies to help guide the sustainable location of renewable energy structures within the Rhode Island's nearshore waters. Furthermore, the methods for this study are consistent with European guidelines for assessing the impacts of offshore wind turbines on the marine environment and could provide a baseline for measuring the cumulative effects of offshore development projects within Rhode Island and Block Island Sounds (CEFAS 2004, BSH 2007. In the end, the incorporation of this research into marine spatial planning efforts will help to conserve and protect the ecological resiliency of Rhode Island's coastal waters and the variety of uses they support.

CHAPTER 2
A manuscript published in Estuarine, Coastal, and Shelf Science, May 2014

Abstract:
The abundance, biomass, diversity, and species composition of the demersal fish and invertebrate community in Rhode Island Sound and Block Island Sound, an area slated for offshore renewable energy development, were evaluated for spatial and seasonal structure. We conducted 58 otter trawls and 51 beam trawls in the spring, summer and fall of 2009-2012, and incorporated additional data from 88 otter trawls conducted by the Northeast Area Monitoring and Assessment Program. We used regionally-grouped abundance, biomass, diversity, and size spectra to assess spatial patterns in the aggregate fish community, and hierarchical cluster analysis to evaluate trends in species assemblages. Our analyses revealed coherent gradients in fish community biomass, diversity and species composition extending from inshore to offshore waters, as well as patterns related to the differing bathymetry of Rhode Island geographic patterns that are persistent from year to year, yet variable by season. Such distributions reflect the cross-shelf migration of fish and invertebrate species in the spring and fall, highlighting the importance of considering seasonal fish behavior when planning construction schedules for offshore development projects. The fine spatial scale (10s of km) of this research makes it especially useful for local marine spatial planning efforts by identifying local-scale patterns in fish community structure that will enable future assessment of the ecological impacts of offshore development. As such, this knowledge of the spatial and temporal structure of the demersal fish community in Rhode Island and Block Island Sounds will help to guide the placement of offshore structures so as to preserve the ecological and economic value of the area.

Introduction:
An ecosystem-based approach to management is essential to attain systemwide sustainability and to ensure the continued availability of marine resources that humans want and need . Designing an effective ecosystem-based management plan requires a comprehensive understanding of the distributions, population structures, interactions and trends of local fish and invertebrate species. Such detailed information, however, is rarely available even in the most well-studied ecosystems (Cury et al. 2005).
Recent interest in offshore energy development combined with the ongoing need to assess the status of overfished groundfish species has focused attention on ecosystem-based spatial management planning in Rhode Island's offshore waters.
The broad-scale (100s of km) distribution of fish species in this area is well-known from standardized trawl surveys (Gabriel 1992, Jordaan et al. 2010. However, spatial management is often implemented at smaller scales , requiring knowledge of fish distributions and fish-habitat associations at 10-km scales (Smith et al. 2013). Thus, when developing spatial management plans, targeted fisheries surveys should be employed to fully assess fine-scale fish community dynamics and potential ecological impacts of offshore energy development.
Rhode Island Sound and Block Island Sound separate the estuaries of Narragansett Bay and Long Island Sound from the outer continental shelf (Figure 1.1).
As such, they provide important linkages between near-shore and offshore processes, including nutrient fluxes, larval transport, and migration of the adult stages of resource species, such as the American lobster (Homarus americanus) and winter flounder (Pseudopleuronectes americanus) (Costa-Pierce 2010). Furthermore, Rhode Island Sound and Block Island Sound support a variety of commercial and recreational fishing activities, such as scallop dredging, otter trawling, long-lining and gill-netting, producing over $60 million in seafood landings in Rhode Island every year (Smythe & Beutel 2010. Despite their heavy use and close proximity to a number of marine science institutions, Rhode Island and Block Island Sounds have been neglected in terms of scientific research, resulting in a poor understanding of the distribution and dynamics of the fisheries resources in this area. Studies to support the management of Rhode Island's offshore waters have become a priority since 2000, when interest in developing artificial reefs, aquaculture sites, and offshore wind turbines emerged in this region. Combined with traditional fisheries and existing dredge-disposal sites, these multiple uses require integrated spatial management planning to site activities in appropriate habitats that will minimize, to the extent possible, the cumulative impacts on resident species and the ecological and economic services derived from this near-shore region (Crowder & Norse 2008 Sound exists, there is a lack of site-specific data to guide spatial management planning (Mahon et al. 1998, Costa-Pierce 2010. Compounding the challenge, this spatial planning process is being conducted against a background of changing coastal climate (Nixon et al. 2009). As a result, historical baseline data may no longer represent current conditions (Collie et al. 2008).
We aimed to address these challenges by conducting comprehensive sampling of the demersal fish and invertebrate community in Rhode Island Sound and Block Island Sound. In particular, we sought to: 1) Evaluate the spatial structure of the demersal fish community in Rhode Island and Block Island Sounds, and 2) Determine whether intra-or inter-annual variations in the composition of these communities exist. With this information, we will then begin to assess the potential impacts of offshore development and climate change in Rhode Island's offshore waters (Punt et al. 2009, BSH 2013.

Study Area
The study area, encompassing Rhode Island Sound and Block Island Sound, is located on the inner continental shelf in the northwest Atlantic (Figure 1.1). This area is seasonally dynamic, with sea surface temperatures ranging from 2°C in the winter to 25°C in the summer, and primary production ranging from 59 mg C m -2 d -1 in the winter to 1738 mg C m -2 d -1 in the spring (Nixon et al. 2010, Ullman & Codiga 2010).
There are three major bathymetric features in Rhode Island Sound and Block Island Sound: 1) Block Island, a 25 km 2 island that lies in the center of Block Island Sound, 2) Cox's Ledge, an expansive rocky shelf in southeast Rhode Island Sound, and 3) Southwest Ledge, an abrupt rocky shoal southwest of Block Island (Figure 1.1

Field Methods
We used otter trawls and beam trawls to sample the demersal fish and invertebrate communities throughout Rhode Island and Block Island Sounds in spring, summer and fall 2009-2012. This dual-gear sampling approach was employed to attain a more holistic assessment of the macrofaunal communities in our study area than could be achieved with one gear type alone. The distinct sampling efficiencies of the two types of gear were recognized at the beginning of the project, and thus otter trawl and beam trawl data were treated separately, then explored in a complimentary manner. Given the limitations of individual sampling gears (i.e. otter trawl, fixed gear, beam trawl), multi-gear approaches are frequently used to achieve more complete evaluations of coastal ecosystems (Franco et al. 2012).
As such, we used otter trawls to sample soft-bottom habitats (sand, mud, clay) and beam trawls to sample fish and invertebrate populations in areas that were too rough for otter trawls (gravel, cobble, moraine). We selected otter trawl stations to achieve maximum coverage of the study area and beam trawl stations to target hard bottom habitats. We also conducted beam trawls at 10 of the otter trawl stations to provide direct gear comparisons. A total of 58 otter trawls were Otter trawls were performed aboard the 90' F/V Darana R using the sampling gear and vessel crew of the Northeast Area Monitoring and Assessment Program (NEAMAP) (http://www.vims.edu/fisheries/neamap). Each tow was conducted with a 400 x 12-cm, three-bridle, four-seam otter trawl, coupled with a pair of Thyboron, Type IV 66" trawl doors. The cod-end was made of double 12-cm stretch mesh (knot to knot) with a 2.43 cm knotless nylon liner. All tows were 20 minutes in duration with a target tow speed of 1.5 m s -1 , resulting in a tow distance of approximately 1.8 km.
Beam trawls were conducted on the 55' F/V Mister G, using a 3-m beam trawl, with cod-end mesh equivalent to that of the NEAMAP otter trawl. All tows were 20 minutes in duration with a target tow speed of 2.0 m s -1 , resulting in a tow distance of approximately 2.4 km.
After each trawl, we recorded aggregate wet weights (kg), counts, and individual length measurements (Fish: Fork length, Squid: Mantle length, Lobster: Carapace length, Crab: Carapace width) for all species collected (Table 1.1). We measured temperature, salinity, and dissolved oxygen profiles at each trawl station and recorded weather conditions and sea-state.
Data from additional otter trawls conducted independently by NEAMAP were later incorporated into the data set to increase sampling coverage in inshore waters, which were sparsely sampled by our field work (Figure 1 The NEAMAP survey targets the coastal zone, and thus all NEAMAP trawls were conducted between 6 and 27 meters water depth. The sampling gear and catchprocessing protocol used by NEAMAP are identical to that of our work, allowing NEAMAP data to be appended without transformation.

Statistical Methods -Univariate Analyses
We accounted for the different gear configurations and catchabilities of beam rough scad (Trachurus lathami), round herring (Etrumeus teres), and round scad (Decapterus punctatus). Standardized catch data were used to calculate aggregate fish community abundance, biomass and diversity at each trawl site. We used Shannon-Weiner's H as a diversity index because it is sensitive to changes in rare species . While insufficient field calibrations prevented full integration of otter trawl and beam trawl data, results were interpreted simultaneously to provide a comprehensive evaluation of the aggregate demersal fish and invertebrate community.
Prior to analysis, all data were tested for normality and homogeneity of variance. Data were log transformed before analysis to achieve a normal distribution.

Univariate Analyses -Seasonality & Geography
We used 2-way analysis of variance (ANOVA) models and posthoc pairwise comparisons to test for the effect of season (spring, summer, fall, winter) and trawl type (otter, beam) on fish community abundance, biomass, and diversity. ANOVAs were followed by multiple comparison tests. To facilitate spatial analysis, we combined site-specific abundance, biomass, diversity, and length frequency into six We used 2-way ANOVA models to test for the effects of geographic position (region, zone, subsection) on total fish community abundance, biomass, and diversity. Data were tested for normality and homogeneity of variance. Data were log transformed before analysis to achieve a normal distribution. Tukey Honest Significant Difference tests (Tukey HSD) were used to make pairwise comparisons between subsections, as well as to assess broader-scale spatial patterns in fish community abundance, biomass, and diversity between inshore (IW, IE), nearshore (NW, NE) and offshore zones (OW, OE), as well as eastern (IE, NE, OE) and western regions (IW, NW, OW).

Univariate Analyses -Size Spectra
We constructed aggregate length-frequencies for each trawl site to assess trends in overall community structure, using length data from all fish and invertebrates that were measured. Length frequencies were generated by pooling across species and plotting logarithmic frequency against geometric length class (Warwick 1984). These specifications reduced noise in the length-frequency distributions and facilitated ecological interpretation (White et al. 2007).

Statistical Methods -Multivariate Analyses
In contrast to the univariate analyses described above, all demersal species were included in multivariate analyses to fully resolve spatial patterns in species composition. We used the Plymouth Routines In Multivariate Ecological Research (PRIMER), version 6.0, for all multivariate analyses (Clark & Gorley 2006). Speciesspecific fish abundance data from each site were fourth-root transformed to reduce the influence of highly abundant species and a Bray-Curtis similarity index was used to measure the similarity in fish community composition between sites. The Bray-Curtis measure is widely used and has properties that are desirable for ecological studies, such as complementarity, localization, and dependence on totals . A multi-dimensional scaling plot (MDS plot) was derived from the similarity matrix to ordinate the sites in two dimensions such that the relative distances apart of all points are in the same rank order as the dissimilarities of the study sites (Kruskal & Wish 1978). Accordingly, points that are close together represent sites that have very similar species assemblages and points that are far apart represent sites that have highly dissimilar species assemblages. We used MDS plots to visualize between-site similarity in fish community compositions.

Multivariate Analyses -Seasonality & Geography
We performed an analysis of similarity (ANOSIM) on the Bray-Curtis similarity matrix of species-specific fish abundance using season (spring, summer, fall, winter) as a factor. ANOSIM tests the null hypothesis that there are no differences between groups of samples (the fish abundance Bray-Curtis similarity matrix) when examined in the context of an a priori factor (season) (Clarke 1993). An R value of 0 indicates there are no differences between groups, while an R value greater than 0 reflects the degree of the differences. The test was permuted 999 times to generate a significance level (p<0.05 used here).
Permutational multivariate analysis of variance (PERMANOVA, Anderson et al. 2008) was used to test for geographic differences in species composition. For these tests the factors were zone (inshore, nearshore, offshore) and east-west region, corresponding to Rhode Island Sound and Block Island Sound. Permutations of the residuals (9999) were used to test main effects of zone and region and their interactions. Pair-wise contrasts were made between zones.

Multivariate Analyses -Species Assemblage Analysis
We used hierarchical clustering analysis with a group-average linking algorithm to divide trawl sites into groups based on the similarity of fish community composition. The cluster analysis was carried out with the SIMPROF routine, which determines statistically significant station clusters within an a priori ungrouped set of stations (Clarke 1993). We used the SIMPER function to determine the group of species that characterized each species assemblage group.

Results:
A total of 101 fish and invertebrate species were caught during otter trawl and beam trawl surveys, of which 25 species were consistently prevalent (

Univariate Analyses Univariate Analyses -Seasonality & Geography
Both season and trawl type had a significant effect on fish community abundance, biomass and diversity (ANOVA p<0.05). Thus, otter trawl and beam trawl data were considered separately for the remainder of the analyses. Furthermore, spring otter trawls and winter beam trawls were excluded from analysis due to low sample sizes and limited temporal and spatial coverage. As such, the following results We identified both regional (East-West) and zonal (Inshore-Nearshore-Offshore) patterns in demersal fish and invertebrate community abundance, biomass, and diversity throughout Rhode Island and Block Island Sounds (Figure 1.2). Spatial trends in fish community abundance were primarily regional, with higher fish abundance in the western region around Block Island (otter trawl: p=0.03; beam trawl: p=0.08). Zonal trends in fish community abundance were not significant. Fish community biomass, however, exhibited a distinct gradient from inshore to offshore, with the greatest fish biomass in the offshore zone (otter trawl: p=0.004). This zonal trend was most pronounced in Block Island Sound, but was persistent throughout the study area. Fish community diversity exhibited similar spatial patterns as biomass, with the highest diversity in the offshore zone (otter trawl: p<0.001; beam trawl: p=0.01). Pairwise comparison of subsections further identified two areas of particularly high biodiversity: 1) North of Cox's ledge (NE) and, 2) South of Block Island (OW) (otter trawl: p=0.003; beam trawl: p=0.04; Figures 1 and 2).

Univariate Analyses -Size Spectra
Considered together, beam trawls and otter trawls sampled a broad size spectrum of the demersal fish and invertebrate community in Rhode Island and Block Island Sounds (Figure 1.3). The beam trawl captured a higher number of smaller individuals and the otter trawl captured more larger individuals with good overlap at intermediate lengths. Small individuals (<20 cm) were more prevalent in the western region of the study area, whereas larger individuals (>60 cm) were more abundant offshore. These spatial patterns in length frequencies reflect the presence of ultraabundant species, such as spiny dogfish and Cancer spp. crabs.

Multivariate Analyses
Despite catch data standardization by area towed, we found that otter trawls and beam trawls caught different species assemblages (ANOSIM: R=0.925, p=0.001).
For this reason, we conducted separate multivariate analyses for otter trawl and beam trawl catch data. Permutational MANOVA revealed significant differences in fish species composition by zone (otter trawl: p<0.001; beam trawl: p=0.001). The demersal fish assemblage offshore was more distinct than those in nearshore and inshore zones.

Multivariate Analyses -Seasonality & Geography
There were also significant interactions between zone and region. For the otter trawl data, the inshore-offshore gradient was stronger in Rhode Island Sound than Block Island Sound. For the beam trawl data, the inshore-offshore gradient was more pronounced in Block Island Sound, because there were few shallow beam trawl stations in Rhode Island Sound.

Multivariate Analyses -Otter Trawl Species Assemblages
Hierarchical cluster analysis of the species abundance data from each otter trawl revealed five species assemblage groups in Rhode Island and Block Island Sounds (Figure 1.4). Of the 105 sites sampled, the majority (80 of 105) were characterized by scup and summer flounder. Of the remaining sites, 17 were characterized by spiny dogfish and sea scallops, three were characterized by silver hake and summer flounder, two were characterized by black sea bass and sea scallops, and two were characterized by silver hake and American lobster. One otter trawl site had a unique fish community structure, reflecting an overall low abundance and diversity. Ledge. More specifically, we found higher densities of scup, summer flounder, skates (Leucoraja spp.), and lobster inshore and around Block Island, and higher densities of spiny dogfish and sea scallops offshore (Figure 1.5). Many sites sampled in different years fell into the same cluster, which indicates that the species composition at these sites is stable from year to year.

Multivariate Analyses -Beam Trawl Species Assemblages
Cluster analysis of the species abundance data from each beam trawl revealed six distinct species assemblage groups in Rhode Island and Block Island Sounds (Figure 1.6). Of the 38 sites sampled, 14 were characterized by Leucoraja spp. skates and Cancer spp. crabs, nine were characterized by sea scallops and sand dollars, six were characterized by sea scallops and sea stars, five were characterized by silver hake and American lobster, and three were characterized by yellowtail flounder and sea scallops. One beam trawl site, located just east of Montauk Point, supports a particularly unique demersal community (Figure 1.6 & 7). Many species caught were unique to that site, such as white sea cucumbers (Pentamera sp.), shortbrowed mud shrimp (Callianassa atlantica), mantis shrimp (Squilla empusa), and clearnose skate (Raja eglanteria).
When examined spatially, the clusters indicate geographic grouping of species assemblages (Figure 1.7). For example, species assemblages characterized by sea scallops, sand dollars, and yellowtail flounder are associated with flat, sandy seafloor found offshore, while species assemblages characterized by silver hake and lobster are associated with shallow, irregular seafloor found north of Block Island ( Figure   1.7). Overall, we found higher densities of skates (Leucoraja spp.), crabs (Cancer spp.), and lobster inshore and around Block Island and higher densities of sea scallops, yellowtail flounder, sea stars, and sand dollars offshore.

Discussion:
This study suggests that the spatial structure of the demersal fish and invertebrate community in Rhode Island and Block Island Sounds is persistent from year to year, yet distinct by season. We found pronounced gradients in fish community biomass, diversity and species composition extending from inshore to offshore waters, as well as patterns related to the differing bathymetry of Block Island Sound and Rhode Island Sound. Cluster analysis revealed geographically distinct species assemblages, which appear to be shaped by a combination of physical, oceanographic and biological factors.  (Deegan 1993, Oviatt 2004, Scopel et al. 2009, Wuenschel et al. 2009). In ecosystems such as this, where sub-annual climactic cues determine species distributions, it is essential to incorporate seasonal dynamics in spatial management plans so as to account for potential impacts to all life stages and species present throughout the year.
The geographic patterns in fish community abundance, biomass, diversity, and species assemblage within Rhode Island and Block Island Sounds may be influenced by a variety of factors, including primary production, water depth and benthic habitat , Bosman et al. 2011, Planque et al. 2011.
Spatial patterns of demersal fish community abundance are often related to trends in primary production (Iverson 1990), which preliminary studies have found to be higher in Block Island Sound than in Rhode Island Sound during summer months (Nixon et al. 2010). If the typical bottom-up ecological model is followed, this pattern in primary production would lead to increased fish abundance in Block Island Sound, which we document here . As such, this study provides initial evidence for the coupling of chlorophyll and fish production in Rhode Island's coastal waters. Ongoing studies that directly link primary production to fish community dynamics, however, are crucial to understanding the strength of bottom-up forcing in Rhode Island and Block Island Sounds (Friedland et al. 2012).
This mechanism is particularly important to understand prior to offshore development, as the distribution and quantity of primary production may be altered by new ocean uses (Lindeboom et al. 2011, Chassot et al. 2007).
In addition to the bottom-up effects of regional-scale spatial variability in primary production, the megafaunal community in Rhode Island and Block Island Sounds may also be influenced by top-down predation pressure, operating at finer scales. Top-down control posits that consumer species structure the ecological community via predation, such that an increase in predator populations (i.e. spiny dogfish, summer flounder, black sea bass) leads to a decrease in prey abundance (i.e. scup, butterfish, lobster, crabs) . In Rhode Island and Block Island Sounds, offshore development will likely alter benthic habitat, which may enhance predator populations, and thus impact demersal fish and invertebrate community structure (Boehlert & Gill 2010). Furthermore, previous work has shown that, even in the absence of habitat alteration, predation pressure influences the distribution and recruitment patterns of various species that inhabit the study area (Levin et al. 1997, Lough 2010. Thus, local-scale predation pressure may play an important role in structuring the demersal species assemblage in Rhode Island and Block Island Sounds. However, further research is needed to fully understand the interacting effects of bottom-up and top-down trophic forces in this dynamic area. The affinity of demersal fish assemblages for specific depth ranges has been observed in a variety of ecosystems ).
Relationships between fish community biomass and water depth are also apparent in other bottom trawl surveys conducted in this area (Bohaboy et al. 2010). While water depth was a significant determinant of fish community composition within Rhode Island and Block Island Sounds, our work suggests that fish species assemblages are also shaped by the physical features of the surrounding seafloor and the proximity to hard-bottom habitat (i.e. Cox's Ledge, Southwest Ledge, Block Island). Thus, depthbased ecosystem classifications that have been widely used in marine spatial planning may, in themselves, not be sufficient for Rhode Island and Block Island

Sounds (Douvere & Ehler 2009).
A general paradigm about marine benthic communities is that, as bottom complexity increases from smooth sand and mud to rock and cobble, ecological complexity and species diversity increase . The presumed relationship is that the more heterogeneous the habitat, the more species it can support because more niches are available . This pattern appears to hold true in Rhode Island's offshore waters, where the more complex seafloor (i.e. more habitat diversity) around Block Island and Cox's Ledge supports more diverse fish communities than the less complex seafloor found inshore (LaFrance et al. 2010). However, a detailed analysis that couples site-specific benthic habitat parameters and demersal fish community metrics is needed to fully understand the fish-habitat relationship in Rhode Island and Block Island Sounds (Anderson et al. 2009(Anderson et al. , 2013. This relationship is particularly important to understand in order to site offshore development activities in appropriate habitats that will minimize the impacts on resident species ).
A core challenge of developing an ecosystem-based spatial management plan is selecting species or species-groups to serve as ecological indicators . In systems such as Rhode Island and Block Island Sounds, where a wide variety of species constitute the fish community, ecological indicator species should represent all functional groups present (i.e. piscivore, benthivore, planktivore). In this way, management plans will be sensitive to: 1) species that structure the ecological community via predation (piscivores), 2) species that are most sensitive to changes in the physical features of the seafloor (benthivores), and 3) species that respond rapidly to changes in primary production (planktivores) , Lindeboom et al. 2011. Thus, we propose the following indicator species for Rhode Island and Block Island Sounds: summer flounder, silver hake, black sea bass, American lobster, and sea scallop. These species were selected based upon the aforementioned criteria as well as to their significance in structuring the aggregate fish community and otter trawl and beam trawl species assemblage groups. Carefully selecting indicator species to track ecosystem change, as outlined above, provides essential insight into the structure and function of complex fisheries food webs in highly dynamics areas, such as Rhode Island and Block Island Sounds.
Many spatial management plans suffer from a lack of information at an appropriate spatial scale ). The spatial coverage and sampling densities of federal trawl surveys, such as the Northeast Fisheries Science Center's Bottom Trawl Survey (~1 station per 687 km 2 ), and inshore trawl surveys, such as NEAMAP (~1 station per 130 km 2 ), are insufficient for assessing the small-scale patterns in fish and invertebrate communities that is necessary for local marine spatial planning (Stauffer 2004, Bonzek et al. 2011). The sampling density of the work presented here (~1 station per 20 km 2 ), however, enables the identification of fine-scale spatial trends in demersal fish assemblages, thus providing a scientific foundation for spatial management planning for Rhode Island and Block Island Sounds. Furthermore, the methods for this study are consistent with European guidelines for assessing the impacts of offshore wind turbines on the marine environment, and as such provide a baseline for measuring the cumulative effects of offshore development projects within Rhode Island and Block Island Sounds (CEFAS 2004, BSH 2013. Thus, the incorporation of this research into marine spatial planning efforts will help to conserve and protect the ecological resiliency of Rhode Island's coastal waters and the variety of uses it supports .
Ultimately, this work provides a novel description of the spatial structure of the demersal fish and invertebrate community in Rhode Island and Block Island sounds, serving as a microcosm for similar fish biogeography studies along the Atlantic coast of North America and other continental shelves around the world.
Furthermore, the species assemblage characterization established by this work provides a baseline against which to measure the impacts of imminent climate change in the highly dynamic southern New England region. Moreover, by sampling areas slated for offshore development as well as suitable control sites, this research facilitates future efforts to understand the severity and extent of the ecological impacts from offshore wind farm development. The spatial scale (10s of km) of this work makes it particularly useful for spatial management planning, as ~10 km is likely to be the minimum scale for development activities and their associated management, as well as the smallest scale at which we can detect differences in habitat use by demersal fish (Jay 2010). Thus, our approach serves as a model for other fisheries surveys that aim to inform marine spatial planning in nearshore ecosystems.

Abstract:
In this study, we used a combination of dietary guild analysis and nitrogen (δ 15 N) and carbon (δ 13 C) stable isotope analysis to assess the trophic structure of the fish community in Rhode Island and Block Island Sounds, an area slated for offshore wind energy development. Between 2009 and 2011, stomach and tissue samples were taken from 20 fish and invertebrate species for analysis of diet composition and δ 15 N and δ 13 C signatures. Stomach content analysis was used to define trophic structure according to dietary guilds, while δ 15 N and δ 13 C stable isotopes were used to determine the trophic position of fish and invertebrate species and the relative importance of benthic and pelagic production in supporting the marine food web.
The food chain in Rhode Island and Block Island Sounds consists of approximately four trophic levels. Within these trophic levels, the fish community is divided into distinct dietary guilds, including planktivores, benthivores, crustacean-eaters, and piscivores. Within these guilds, inter-species isotopic and dietary overlap is high, suggesting that resource partitioning or competitive interactions play a major role in structuring the fish community of this area. Carbon isotopes indicate that most fish are supported by pelagic phytoplankton, although there is evidence that benthic production also plays a role, particularly for obligate benthivores such as skates (Leucoraja spp.). This type of analysis is useful for developing an ecosystem-based approach to management, as it identifies species that act as direct links to basal resources as well as species groups that share trophic roles.

Introduction:
Globally, fisheries scientists and managers have asserted the need for an ecosystem-based approach to fisheries management to better account for the interactions among commercially harvested species and their prey, predators, and habitat , Fogarty 2013. Development of ecosystem-based fisheries management, however, requires a thorough understanding of the trophic structure of the fisheries ecosystem of interest (Latour et al. 2003, Smith et al. 2007, Gilliland & Laffoly 2008. Such knowledge can be challenging to ascertain and apply, particularly in biologically and oceanographically complex ecosystems such as the northwest Atlantic continental shelf , Fogarty & Rose 2013. To address this challenge, methods such as dietary guild analysis and stable isotope analysis have been used to simplify the structure and function of highly complex ecosystems and examine the flow of energy through food webs , Wilson 1999, Metcalf et al. 2008. We sought to apply these approaches to a nearshore Northwest Atlantic fisheries ecosystem, where recent interest in offshore energy development has focused attention on ecosystem-based spatial management planning (RI SAMP 2010, . Dietary guild analysis is a common approach for assessing the trophic structure of fisheries ecosystems , Pasquaud et al. 2008, Reum & Essington 2008. By definition, a guild is "a group of species that exploit the same class of environmental resources in a similar way, and thus overlap significantly in their niche requirements" (Root 1967). As such, dietary guild analysis can be used to identify group of species with similar functional roles, and assess resource partitioning and competitive interactions within an ecosystem . Theoretically, fisheries ecosystems are more stable when within-guild functional redundancy is high, as ecosystem function is maintained despite fluctuations in the abundance of individual guild members (Bell et al. 2014). The classification of species into dietary guilds enables the progression of ecosystembased fisheries management, as species are assessed as functional groups, rather than individual species (Auster & Link 2009).
Nitrogen and carbon stable isotope analysis are also valuable tools in understanding the trophic structure of fisheries ecosystems (Peterson & Fry 1987, Layman et al. 2007). Specifically, nitrogen stable isotopes (δ 15 N) describe the time-integrated feeding history of a consumer and can be used to identify the trophic position of a species. The δ 15 N content of a consumer's tissue is enriched approximately 3.4‰ relative to that of its diet due to trophic fractionation, thus reflecting the species' role in the marine food chain (Post 2002). Carbon stable isotopes (δ 13 C) are used to investigate the relative importance of different basal resources in supporting fish production , Post 2002, Mackenzie et al. 2011. Boundary layer effects lead to differential uptake of 13 C by pelagic phytoplankton and benthic macroalgae, such that the average δ 13 C of pelagic phytoplankton is -22‰, while the average δ 13 C of benthic macroalgae is -17‰ (Peterson & Fry 1987, France 1995. This disparity in the δ 13 C of benthic and pelagic carbon sources is reflected in marine consumers, with benthic-feeding animals enriched approximately 5‰ compared to pelagic-feeding animals . In this way, the δ 13 C of resident fish species reflect the initial carbon sources to the food web, thus allowing for the differentiation between pelagic and benthic food webs.
Although previous work has assessed the trophic structure of fish communities in the northwest Atlantic, the transitional seas of Rhode Island Sound and Block Island Sound have not been adequately sampled by routine state and federal surveys , Jordaan et al. 2010 In this study, we used a combination of fish stomach content and stable isotope analyses to assess the dietary guild structure and flow of energy through the fisheries food web in Rhode Island and Block Island Sounds. More specifically, we aimed to determine the relative importance of benthic and pelagic production in supporting the fisheries food web, whether species within the same dietary guilds maintain consistent trophic positions, and whether silver hake, scup, or winter flounder exhibit spatial patterns in foraging behaviors. These analyses are useful for developing an ecosystem-based approach to management, as they identify species that act as direct links to basal resources as well as species groups that share trophic roles.

Methods:
We assessed the diet compositions and trophic interactions of 20 fish species using stomach content analysis and stable isotope analysis (

Dietary Guild & Niche Breadth Analysis:
Stomach content analysis was used to define dietary guilds, which represent functionally similar species within the fish community. For highly abundant species, a random sub-sample of five fish per target species per station was selected for diet analysis. For less abundant species (< five individuals per station), all specimens were used for diet analysis. Fish stomachs were extracted immediately after capture and preserved in Normalin, a non-toxic preservative. In the laboratory, the contents of preserved stomachs were extracted and the total weight (mg wet weight) measured with an analytical balance (Bowman et al. 2000). All recovered prey items were identified to the lowest practical taxon with the aid of stereomicroscopes, and their contribution to overall diet was measured as percent of total stomach content weight (Hyslop 1980).
Data from 20 predator species and 1,762 stomach samples were used in the dietary guild analysis (Table 2.1). Stomach samples from an additional five species were collected and processed, but the sample sizes were not sufficient for inclusion in the guild analysis (>10 stomachs). Prey items were grouped based on dietary prevalence (by weight) and digestive state (fresh, partially digested, or well digested).
Abundant prey items were grouped at lower taxonomic levels, while less abundant items were grouped at higher levels. The resulting prey classification consisted of 47 categories (Table 2.2).
A cluster sampling design was used to calculate the contribution of each prey type to the diet of individual predator species (Buckel et al. 1999). The mean proportional contribution of a prey type by weight was calculated using the following formula for each predator species.
where is the proportional contribution of prey type k to the diet of a given predator species weighted by the number of that predator species caught at each station, is the proportional contribution of prey type k to the diet of a given predator species pooled over predator samples at station s, is the number of a given predator species captured in a trawl at station s, is the total weight of all prey for a given predator species from station s, and is the weight of prey type k for a given predator species at station s. Levins (1968) standardized index of niche breadth was used to assess the dietary specialization of each predator species (Colwell & Futuyma 1971, Hulbert 1978, as follows. where, (Bi) is the standardized index of niche breadth for predator species i, is the proportional contribution of prey type k to the diet of predator species i ( for predator i), and is the total number of prey categories consumed by predator species i (Table 2.1). ranges between 0 and 1, with a value of zero indicating maximum dietary specialization (i.e. a single prey type comprising a predator's diet) and 1 indicating nondiscrimination among prey (i.e. each prey type contributes the same proportion to a predator's diet).
The Schoener (1970) similarity index was used to assess the dietary overlap, , between predator category pairs : where is the dietary overlap between predator i and predator j, is the mean proportional weight of prey k in predator i ( for predator i), and is the mean proportional weight of prey k in predator j ( for predator j). The statistical software package PRIMER 6.0 was used to create a resemblance matrix containing the dietary similarity ( ) of each predator pair.
Hierarchical clustering was used to group species into dietary guilds based on similarity of diet composition. The cluster analysis was carried out with the SIMPROF (similarity profiling) routine, which defines statistically significant groups among samples (Clarke & Gorley 2006). A dendrogram was derived from the cluster analysis to visualize the dietary similarities and dissimilarities between species and the resulting dietary guilds. Finally, a multi-dimensional scaling plot was derived from the dietary resemblance matrix to ordinate species-specific diet compositions in two dimensions, such that the relative distance between points represents the degree of dietary similarity between species (Kruskal & Wish 1978). A SIMPER (similarity percentages) analysis was further used to identify prey types that primarily account for the differences between dietary guilds.

Stable Isotope Analysis:
In addition to stomach content analysis, we used nitrogen (δ 15 N) and carbon  Sub-samples of fish tissue (~1 mg dry weight) were analyzed for nitrogen and stable isotopes at the Boston University Stable Isotope Laboratory with an automated continuous-flow isotope ratio mass spectrometer (Preston & Owens 1983). Isotopic ratios of 15 N/ 14 N and 13 C/ 12 C are expressed in delta notation (δ) as the relative per mil Atlantic sea scallop makes this species a suitable benchmark organism for nitrogen stable isotope analysis (Naidu 1991, Black et al. 1993). The δ 15 N of sea scallops sampled for this study (7.3‰), further confirmed its primary consumer trophic positioning in the food web, as it was approximately 3.4‰ lower than the zooplanktivorous species sampled.
Trophic fractionation of carbon was assumed to be 0.5‰ δ 13 C, and was accounted for using the following formula for each species (Deniro & Epsten 1977, Post 2002, McCutcheon et al. 2003. where, 13 is the carbon isotopic signature corrected for trophic fractionation, 13 is the raw carbon isotopic signature, is species-specific trophic position derived from δ 15 N, 1 is the difference between the trophic level of the benchmark species (Atlantic sea scallop) and the base of the food web (phytoplankton), and 0.5 is the rate of trophic fractionation of carbon (Post 2002).
Isotopic turnover rates of nitrogen and carbon are directly related to growth rate, with faster growing animals exhibiting shorter turnover rates ). For marine fish species, previous studies have found that stable isotope signatures in white muscle tissue have isotopic turnover rates ranging from a few months to over a year , MacNeil et al. 2006 and δC 13 between guilds. Tukey's post-hoc multiple comparison tests were used to assess pair-wise differences in δN 15 and δC 13 between dietary guilds.

Spatial and Annual Analysis:
Spatial analysis of fish diet composition and stable isotope signatures were conducted for silver hake, scup, and winter flounder. These species had sufficient stomach and isotope sample coverage from across the study area to enable spatial analysis (Smith 2009, Table 2.1). Species-specific diet and stable isotope data were divided into four regions for spatial analysis, based on their proximity to shore (Inshore, Offshore) and location within Rhode Island Sound and Block Island Sound (RIS, BIS). This delineation resulted in four regions: Inshore RIS, Offshore RIS, Inshore BIS, and Offshore BIS.
A multivariate Analysis of Similarity (ANOSIM) was used to test for differences in fish diet between regions (Inshore RIS, Offshore RIS, Inshore BIS, Offshore BIS) and years (2009,2010,2011). Multidimensional scaling plots were used to visualize the results of regional and annual ANOSIMS. Site-specific diet compositions were also projected in ArcGIS and used to visualize spatial patterns in species-specific diet composition.
Bivariate plots of species-specific δ 15 N and δ 13 C were used to visualize patterns in trophic position and basal carbon sources by region and year. ANOVA models and post-hoc Tukey Honest Significant Difference (Tukey HSD) tests were used to test for regional and annual differences in mean species-specific δ 15 N and δ 13 C values.

Dietary Guilds & Niche Breadth:
Niche breadth (Bi) ranged from 0.02 to 0.52, with 17 out of 20 species having niche breadths less than 0.3 (Table 2.1). Alewife, smooth dogfish, and weakfish exhibited the most specialized feeding behavior, with niche breadths of 0.02, 0.07, and 0.08, respectively ( The planktivore guild consisted of American shad, alewife, and butterfish and exhibited 55.4% dietary similarity (SIMPROF: π = 1.85, p = 0.536, Figure 2.3). The diets of these species were characterized by high proportions of unidentified animal remains, which likely represent well-digested zooplankton (Table 2.4, Figure 2.4). The dietary composition of Atlantic herring, a known planktivore, was significantly different than American shad, alewife, and butterfish, due to higher abundances of gammarid amphipods and cumaceans   Table 2.4, Figure   2.4).
The benthivore guild was split into two groups, based upon prey diversity. The first benthivore group consisted of scup, winter flounder, winter skate, and little skate, and exhibited 52.4% dietary similarity (SIMPROF: π = 1.96, p = 0.185, Figure   2.3). These species fed upon a wide variety of prey, representing 46 of the 47 prey categories used in this study (Tables 2.4 & 2.5). The most common prey types were amphipods, polychaete worms, and unidentified animal remains (Figure 2.4). The second benthivore group consisted of yellowtail flounder and haddock, and exhibited 75.5% dietary similarity (SIMPROF: π = 2.80, p = 0.530). These species fed primarily on gammarid amphipods, which accounted for 51% of these species' diets (Tables 2.4 & 2.5).
The crustacean-eater guild consisted of black sea bass and smooth dogfish, and exhibited 41.3% similarity (SIMPROF: π = 2.77, p = 0.824, Figure 2.3). Crabs accounted for 54% of the diets of these species, while shrimp accounted for nearly 10% ( Figure   2.4, Tables 2.4 & 2.5). Silver hake were also consumers of decapod crustaceans, but their diets were characterized by higher prevalence of shrimp (67%) and fish (12%), and thus were classified as a unique group (Table 2.5).

Stable Isotopes:
Nitrogen stable isotope analysis indicates that the species sampled for this work represent two major trophic groups in the fisheries food web, secondary consumers (i.e. foragers) and tertiary consumers (i.e. predators) (Figure 2.5).
Contrary to the dietary guild classification, spiny dogfish was found to exhibit the lowest trophic position of all species sampled (TP = 3.30), suggesting that stomach contents did not accurately classify the trophic role of this species. This result is likely due to the consumption of ctenophores, which are difficult to assess via stomach content analysis . The highest trophic position in the fisheries food was occupied by striped bass (TP = 4.42).
Within dietary guilds, inter-species isotopic overlap was high (Figure 2.5). The planktivore, benthivore, and crustacean-eater guilds all fell within trophic level 3, while the piscivore guild fell into trophic level 4. The planktivore and benthivore guilds exhibited the lowest trophic positions (3.60 ± 0.15 and 3.57 ± 0.13, respectively), while the crustacean-eater and piscivore guilds maintained the highest trophic positions (3.78 ± 0.17 and 4.06 ± 0.52, respectively). While an initial ANOVA analysis suggested that dietary guilds maintained distinct trophic positions (ANOVA p=0.031), post-hoc pairwise tests indicated that only the piscivore guild was significantly unique.
Carbon isotopic analysis indicates that most fish in Rhode Island and Block Island Sounds are supported by pelagic phytoplankton, with an aggregate carbon isotopic signature of -19.37 (± 0.13). There is also evidence, however, that benthic production plays a role, particularly for obligate benthivores, such as skates and flatfish (little skate δ 13 C = -17.63; yellowtail flounder δ 13 C = -18.98; Figure 2.5).
Dietary guilds did not exhibit unique δ 13 C signatures (ANOVA p=0.199), but the foraging strategies of planktivores and benthivores were apparent.

Spatial and Annual Analysis:
ANOSIM results suggest that silver hake and winter flounder exhibit spatial  Silver hake exhibited significant spatial patterns in isotopic composition, characterized by an inshore-offshore gradient, with higher δ 15 N and less negative δ 13 C signatures inshore and lower δ 15 N and more negative δ 13 C signature offshore (ANOVA: δ 13 C p<0.001; δ 15 N p<0.001, Figure 2.9). These results suggest that silver hake feeds higher in the food chain and derives more energy from benthic production in inshore waters. Silver hake also exhibited interannual trends in δ 13 C

Discussion:
This work highlights the complex interactions of the fish community on the northwest Atlantic continental shelf and provides details on the trophic structure of the nearshore fish community in Rhode Island and Block Island Sounds. The dietary guilds classified by this research are consistent with previous studies conducted in the region . The spatial scale of this work, however, provides a unique perspective on the trophic structure of the fisheries ecosystem that is applicable to local-scale management efforts (Langton et al. 1995, Moore & Sowles 2010, RISAMP 2010. On a regional scale in the northwest Atlantic, competition for food resources is typically not an important factor in structuring the fish community, as prey resources are consistently abundant and diverse (Auster & Link 2009).
Furthermore, many species exhibit opportunistic feeding behavior and are able to switch prey resources as they are available . In areas where preferred habitats constrict species distributions or in ecosystems where highly competitive species are increasing, however, competition for food may become limiting (Warwick 1984, Vinagre et al. 2014. Thus, given the diverse habitats Despite unique dietary guilds and foraging strategies, carbon isotope analysis indicates that most of the fish species in Rhode Island and Block Island Sounds rely on pelagic phytoplankton as a basal energy resource . There is, however, evidence that benthic production also plays a role, particularly for obligate benthivores such as skates and flounder (Vander & Vadeboncoeur 2002). Such isotopic analyses are useful for developing an ecosystem-based approach to management, as they identify species that act as direct links to basal resources as well as species groups that share trophic roles (Marasco et al. 2007, Crowder & Norse 2008.
While the results from our work are generally consistent with previous, largescale studies, a few discrepancies are worthy of discussion (Smith & Link 2000. First, a number of planktivore species, including American shad and alewife, exhibit higher δ 15 N signatures (as high as 13.30‰) than would be expected for species of their feeding ecology , Trenkel et al. 2014. By comparison, silver hake, which primarily feeds upon fish and shrimp and should thus exhibit a higher δ 15 N than planktivores, has a lower δ 15 N signature (12.98‰) and trophic position. The enriched signals exhibited by American shad and alewife may reflect their anadromous life histories or recent estuarine foraging, as δ 15 N increases markedly in coastal environments were the enriched δ 15 N signal of human sewage and other terrestrial nutrient sources is persistent throughout the food web (Cabana & Rasmussen 1996, McKinney et al. 2010. In general, the isotopic signatures of migratory species often reflect feeding in multiple areas and habitats (Clément et al. 2014, Dixon et al. 2015. As such, one must interpret the isotopic signatures of highly mobile species, which integrate multiple months of feeding behavior, with care , Abrantes & Barnett 2011. Ontogenetic diet shifts have been shown to be an important for a number of the species sampled as part of this work, including silver hake, bluefish, and spiny dogfish (Buckel 1999. Unfortunately, the sampling carried out for this study was insufficient for a robust comparison of species trophic roles between size classes. Preliminary analyses, however, indicate that there are no differences in the diet or isotopic signatures between different size classes for most of the species sampled. There was, however, a trend toward increasing δ 15 N and δ 13 C signatures with increasing spiny dogfish size (data not shown).
The One explanation for the dietary guild-trophic position dichotomy is that spiny dogfish exhibits net feeding behavior (i.e. rapid consumption of fish while the trawl is being hauled back), which could lead to an overestimation of the proportion of fish in spiny dogfish diet and, thus, a misclassification into the piscivore guild . Another confounding detail of the trophic structure of spiny dogfish is the highly negative carbon isotopic signature (-23.0 ‰ δ 13 C), which is more negative than the pelagic phytoplankton in this region (-20 ‰ δ 13 C, EPA unpublished data). The low δ 13 C signature of spiny dogfish may be a result of feeding offshore or in deep waters, where the planktonic community is super-depleted in δ 13 C, but further investigation is needed to fully understand his unique trophic process (Ostrom et al. 1997).
In a spatial context, this work suggests that silver hake and winter flounder exhibit distinct patterns in their feeding behaviors, while scup does not. More specifically, isotopic analyses indicate that silver hake occupies a higher trophic position (i.e. feeds higher in the food chain) and derives more energy from benthic production in inshore waters. The dietary patterns of silver hake corroborate these findings, with small fish dominating silver hake's diet inshore and shrimp dominating silver hake's diet offshore. These results suggest that the reduced depth and estuarine outflow in Rhode Island and Block Island Sound's inshore waters concentrate small fish prey and increase benthic-pelagic coupling. Spatial patterns in winter flounder diet also appear to be associated with prey availability, as winter flounder diet is dominated by amphipods in Block Island Sound, where amphipod tube mats are abundant ). In the case of scup, a lack of spatial dietary pattern could be a result of its narrow niche breadth, which may prevent this species from readily switching prey despite spatially distinct availability. It is important to consider that factors such as offshore wind energy development will likely shift prey availability and distribution in this region.
In conclusion, application of trophic structure analyses, such as those presented here, to the development of ecosystem-based fisheries management will help to preserve the balance between trophic components and maintain a productive fisheries ecosystem. A specific application of this work is to the modelling of species as functional groups , Latour et al. 2003. Furthermore, it is particularly important to consider this type of work in the management of species that have specific habitat requirements and highly specialized diets, such as yellowtail flounder and black sea bass. Management of migratory predators, such as striped bass, on the other hand, may not require consideration of prey availability as they are able to integrate resources regionally. The results of this work not only provide valuable insight into fisheries ecosystem dynamics in a temperate coastal environment, but also inform spatial management plans for Rhode Island and Block Island Sounds (RI SAMP 2010). Furthermore, the methods for this study are consistent with European guidelines for assessing the impacts of offshore wind turbines on the marine environment and could provide a baseline for measuring the effects on local-scale trophic dynamics from offshore development projects (BSH 2013). Future work will focus on developing an understanding of the seasonal trends in trophic structure and the impacts of planned offshore wind energy development.            Sounds, we hope to guide the location of future ocean uses so as to preserve the ecological and economic value of the area.

Introduction:
The physical and oceanographic characteristics of benthic habitat affect fish community structure in a variety of marine ecosystems , Gratewick & Spite 2005, Anderson et al. 2009). For example, Hawaiian coral reef fish communities exhibit distinct relationships with the rugosity and depth of benthic habitat, while groundfish on George's Bank in the northwest Atlantic exhibit seasonally distinct relationships to bottom water temperature and depth  Historically, the distribution, scale, and structure of fish habitat in marine ecosystems have been difficult to assess due to limited seafloor survey techniques . Recent technological developments, such as interferometric sonar systems and autonomous underwater vehicles, however, have begun to address this challenge (ICES 2007, Brown et al. 2011. In comparison to traditional techniques, habitat assessments that utilize interferometric data, in addition to seafloor imagery and oceanographic conditions, typically produce a more holistic, and more biologically-meaningful, characterization of the seafloor , Brown et al. 2012. Despite advances in seafloor survey techniques, however, it still remains difficult to assess the link between benthic habitat and fish community structure . One factor contributing to this difficulty is the range of spatial scales at which organisms may be associated with their environment , Anderson et al. 2009). As such, the most effective method for combining biological and habitat data are still under debate (Brown 2011). Here we will apply a non-parametric, multivariate approach to linking fish community structure to benthic habitat at a scale relevant for local spatial management efforts.
Marine spatial planning is typically considered an "ecosystem-based" approach to management . By definition, ecosystem-based approaches consider not only species interactions and climate, but also benthic habitat. Thus, marine spatial planning, ecosystem-based fisheries management, and fish habitat characterization go hand in hand ). More specifically, integrated spatial management planning requires activities to be sited in appropriate habitats that will minimize, to the extent possible, the cumulative impacts on resident species and the ecological and economic services derived from this nearshore region (Beck et al. 2009). To achieve this, however, a thorough understanding of the spatial distribution of benthic habitats and their linkages to fish distribution and production is required . Understanding habitat requirements and distributions is especially important for vulnerable or overfished species, whose rebuilding programs could include large area closures if other management tools are unsuccessful ).
This project addresses the general challenge of developing an ecosystembased approach to marine spatial planning in Rhode Island's nearshore waters, but there is a lack of site-specific data to guide spatial management planning . Compounding the challenge, the spatial planning process is being conducted against a background of changing coastal climate. As a result, historical baseline data may no longer represent current conditions. Studies to support the management of Rhode Island's nearshore waters have become a priority since 2000, when new uses, such as offshore wind energy, aquaculture, and sand extraction were proposed in this region.
Understanding the spatial distribution of benthic habitats and the relationship to the fish community is essential in developing effective spatial management practices. As such, the primary objective of this project was to obtain site-specific data about the benthic habitats and the fish communities in Rhode Island's nearshore waters. To do this, we mapped and classified benthic habitats using interferometric sonar, seafloor video, and oceanographic sampling, and assessed fish community structure using otter trawls and beam trawls. In the end, this project sought to develop a better understanding of the fish-habitat relationships in the nearshore Northwest Atlantic ecosystem of Rhode Island and Block Island Sounds so as to guide spatial management plans and advance the field of fish-habitat research.

Fish Community Assessment
Otter trawls and beam trawls were used to obtain habitat-specific fish and The resolution of the Northeast Atlantic CRM is 90 meters. The full Northeast Atlantic CRM was clipped to the extent of the study area and converted to GRID format for application in this study (Figure 3.1).
A suite of benthic habitat parameters was derived from the backscatter and bathymetry data for each of the acoustically mapped trawl stations (Table 3.1). The minimum, maximum, mean, and standard deviation of depth and slope were calculated from the bathymetry grid for each trawl site using the Spatial Analyst and Raster Processing toolboxes in ArcInfo 10.3 ).
These metrics were calculated at 90 meter resolution within a 5 meter wide buffer around each otter trawl track and a 3 meter buffer around each beam trawl track.
The minimum, maximum, mean and standard deviation of backscatter were also derived from the side-scan mosaics in ArcInfo 10.3. These metrics were calculated at 2 meter resolution within a 5 meter wide buffer around each otter trawl track and a 3 meter buffer around each beam trawl track.
In addition, a map of benthic surface roughness was used to characterize the habitat complexity at each trawl site (RI SAMP 2010, Figure 3.2). The benthic surface roughness layer represents the standard deviation of the slope within a 1000 meter radius calculated at 100 meter pixel resolution. The mean, minimum, maximum, and standard deviation of the surface roughness was calculated for each of the trawls using the Raster Processing toolbox in ArcInfo 10.3 (Table 3.1). These metrics were calculated at 100 meter resolution within a 5 meter wide buffer around each otter trawl track and a 3 meter buffer around each beam trawl track.

Seafloor Video Surveys
The benthic habitat types present at each trawl site were investigated using seafloor video surveys. The video survey system is comprised of a Microvideo AM301 underwater video camera, mounted on a stainless steel video sled with two Pro-V8 LED lights for illumination. Two lasers, fixed 8 inches apart, provide scale for habitat features and enable measurement of epifaunal species. At each trawl station, the video sled drifted for 10 minutes, with the camera collecting continuous video footage. The target camera altitude was 1 meter, giving a field of view of approximately 1 m 2 . The objective was to obtain at least 20 clear and useable photos for quantitative analysis from each station.
Bottom photos were analyzed with a point-count program written in Matlab that was revised for this work (Lengyel et al. 2009). Data extracted from each photo include the major and minor sediment types, the percent cover of colonial epifauna, and the frequencies of free-living animals. Epifaunal coverage and megafaunal occurrence data were excluded from these analyses due to their rarity in seafloor videos. Major and minor sediment types were recorded on a scale consistent with Wentworth grain size and were converted to numerical values for quantitative analysis (Table 3.2, Figure 3.3). Major and minor sediment types were defined as the sediment types covering ≥75% and ≤25% of the seafloor, respectively. The major grain size at each trawl station was calculated by taking the mean of the numerical major sediment type from the 20 seafloor photos at each station. The same routine was followed for classification of site-specific minor grain size. The total number of habitat types observed at each trawl site was used for categorical analysis and interpreted as a measure of habitat heterogeneity.

Oceanographic Sampling
Oceanographic data were collected at each otter trawl station using a Yellow Springs Instruments (YSI) multiparameter probe that recorded surface and bottom temperature (°C), salinity (ppt), and dissolved oxygen (mg/L). Due to intermittent equipment malfunctions, full oceanographic data were available for only 36 of the 44 otter trawl stations (Table 3.1). Surface temperature (°C), salinity (ppt), and dissolved oxygen (mg/L) were also recorded at each beam trawl station using a YSI multiparameter probe. A Sonotronics Depth and Temperature Logger (DTL) was used to record bottom water temperature (°C) at each beam trawl station. Again, due equipment malfunctions full oceanographic data were available for only 35 of the 38 beam trawl stations (Table 3.1).

Assessing Benthic Habitat and Fish Community Relationships
A suite of 24 continuous and four categorical site-specific habitat parameters, derived from bathymetry, slope, benthic surface roughness, backscatter, videographic, and oceanographic data, were combined with fish community metrics derived from trawl surveys to test for relationships between habitat characteristics and fish and invertebrate abundance, biomass, diversity, and species assemblage structure (Table 3.1). Beam trawl and otter trawl data were analyzed separately due to differences in gear selectivity .

Univariate Analyses
For univariate fish-habitat analyses, aggregate fish community abundance and biomass were standardized by the area swept (otter trawl area swept = 0.022 -0.031 km 2 ; beam trawl area swept = 0.0066 -0.0076 km 2 ) and log transformed to achieve a normal distribution. Shannon-Wiener's H was used as a diversity index because it is sensitive to changes in rare species .
Relationships between continuous habitat parameters and univariate fish community metrics were assessed with linear regression analysis in R. It was hypothesized that fish abundance, biomass, and diversity would be positively correlated with measures of bottom complexity (i.e. benthic surface roughness, slope, standard deviation of backscatter, major sediment type) .
It was also hypothesized that fish diversity would be positively correlated with depth ).
Stepwise multiple linear regression models were used to assess the cumulative effects of 24 habitat parameters on aggregate fish community abundance, biomass, and species diversity (Table 3.1). Akaike's Information Criterion corrected for small sample bias (AICc) was used to evaluate and select the optimal regression model (Burnham & Anderson 2002).

Multivariate Analyses
For multivariate fish community response variables (e.g. species composition in trawls), associations with habitat parameters were tested using nonparametric techniques in the software package PRIMER-E . These analyses aimed to assess which habitat parameters are most important in structuring the fish assemblages in Rhode Island Sound and Block Island Sound.
Prior to fish-habitat analysis, species-specific fish abundance data from each trawl site were fourth-root transformed to reduce the influence of highly abundant species (Clark & Green 1988). A Bray-Curtis similarity index was used to assess the similarity in fish community composition between sites and a hierarchical clustering analysis with a group-average linking algorithm was used to divide trawl sites into species assemblage groups based on the similarity of fish community composition (Clark & Gorley 2006). The cluster analysis was carried out with the SIMPROF routine, which determines statistically significant station clusters within an a priori ungrouped set of stations (Clarke 1993).
To test of the effect of categorical habitat parameters on fish community composition, analysis of similarity (ANOSIM) was performed on the fish community Bray-Curtis similarity matrix using depth strata, major habitat type, minor habitat type, and number of habitat types as factors. ANOSIM tests the null hypothesis that there are no differences in fish species assemblage between groups of samples when examined in the context of an a priori factor (depth strata, major and minor habitat type, number of habitats) (Clarke & Gorley 2006). An R value of 0 indicates there are no differences in species assemblages between factor groups, while an R value greater than 0 reflects the degree of the differences. The test is permuted 999 times to generate a significance level.
Prior to multivariate analysis of continuous habitat variables, a Draftsman plot, consisting of pairwise scatterplots, was created to assess the correlation between habitat variables. Variables that were highly correlated (r > 0.85), and therefore redundant, were eliminated from further analysis (see Table 3.1; variables marked with an asterisk or cross were retained). Habitat variables were then normalized to correct for differences in units, and a Euclidean distance resemblance matrix was created to assess the multivariate habitat similarity between sites. A multi-dimensional scaling plot (MDS plot) was derived from the habitat resemblance matrix to ordinate the sites in two dimensions. The MDS plot was used to visualize between-site similarity in habitat and to compare the environmental patterns to that of the fish community.
The relationship between the non-correlated habitat parameters and fish community composition was examined using the BIOENV procedure, which identifies a subset of habitat parameters that best explain fish community composition (Clarke & Gorley 2006). More specifically, the BIOENV approach analyzes the extent to which a suite of habitat variables match the species assemblage data by searching for high rank correlations between variables in the two matrices (the habitat Euclidean distance matrix and the fish community Bray-Curtis similarity matrix). Thus, the BIOENV procedure identifies combinations of benthic habitat parameters that result in the highest Spearman rank correlation with the fish community similarity matrix. A maximum of five variables was permitted in the output. Single parameter runs were also conducted to assess the significance of individual habitat parameters to fish community structure. The BIOENV procedure was permuted 999 times in order to evaluate the level of significance of the results.
The group of five benthic habitat parameters found to best explain fish community structure were then subjected to the LINKTREE procedure to classify the stations according to patterns in the selected habitat parameters. The LINKTREE routine groups the fish community samples (stations) by successive binary division using the habitat parameters as drivers and maximizing the ANOSIM R value at each division (Clarke & Gorley 2006). The ANOSIM R was constrained to be greater than 0.300 and the minimum group size was set at three. Each resulting class contains a group of fish community samples (stations), classified by quantitative thresholds of habitat parameters. An ANOSIM was performed on the habitat groups defined by the LINKTREE analysis to test whether there are significant (p > 0.05) differences in fish assemblages among habitat groups. ANOSIM was also used to test for differences in habitat characteristics between species assemblage groups.

Benthic Habitat and Fish Community Integration
Otter Trawls -Univariate Analyses ANOVA models testing for the effect of categorical habitat variables (depth stratum, major habitat type, minor habitat type, and number of habitat types) on aggregate fish community metrics (abundance, biomass, and diversity) were largely insignificant, with the exception of the effect of depth strata on species diversity (Table 3.3). Tukey HSD tests revealed a significantly higher species diversity in depth strata 5 than in depth stratum 3 and 4 (p=0.007). Thus otter trawl sites in deeper water were characterized by higher species diversity than otter trawl sites in shallower water.
Regressions between continuous habitat parameters and otter trawl fish community metrics also revealed a relationship between depth and species diversity, such that species diversity increased with deeper minimum, maximum, and mean water depth (Table 3. 4,Figures 3.4,3.5,& 3.6). In addition, species diversity exhibited significant proportional relationships with backscatter (minimum and mean) and bottom salinity (Table 3.4, Figure 3.6). Bottom dissolved oxygen and bottom temperature, on the other hand, were negatively related to fish community diversity. In term of fish community abundance, surface and bottom salinity were significant predictor variables, such that fish abundance decreased in more saline water (Adj. R 2 =0.209, p=0.003) (Table 3.4, Figure 3.4). Finally, fish community biomass was negatively related to bottom water temperature (Adj. R 2 =0.100, p=0.043) ( Table 3.4, Figure 3.5). The remaining regressions were not significant (Adj.  In a spatial context, the "Shallow" habitat group is primarily located around Block Island, where it exhibits significant overlap with the "Scup and Summer Flounder" species assemblage group (Figure 3.9). The "Deep and Smooth" habitat group, on the other hand is located in the offshore extent of Rhode Island Sound, where the "Spiny Dogfish and Sea Scallop" species assemblage group dominates.
Finally, the "Deep and Rough" habitat group is located in the deep waters surrounding Cox's Ledge, which are primarily occupied by the "Silver Hake and Lobster" species assemblage (Figure 3.9).

Beam Trawls -Univariate Analyses
ANOVA models testing for the effect of categorical habitat variables (depth stratum, major habitat type, minor habitat type, and number of habitat types) on aggregate fish community metrics (abundance, biomass, and diversity) were largely insignificant, with the exception of the effect of depth strata on species abundance (Table 3.3). The result of the depth strata ANOVA, however, is unreliable, due to the disparity in beam trawl sample size between strata (two beam trawls in stratum 3, nine beam trawls in stratum 4, and 27 beam trawls in stratum 5).
Regressions between benthic habitat parameters and fish community metrics revealed a proportional relationship between depth and fish community abundance, such that fish abundance increased with deeper minimum, maximum, and mean water depth (Table 3.6, Figure 3.10). Beam trawl species diversity, on the other hand, was negatively related to water depth, with lower species diversity in deeper waters (Table 3.6, Figure 3.12). Fish community diversity also exhibited a significant proportional relationship with bottom water temperature, whereas fish community abundance and biomass exhibited inverse relationships with bottom water temperature (Table 3.6, Figures 3.10, 3.11 & 3.12). In addition, fish community abundance and biomass were significantly influenced by surface salinity, such that fish abundance and biomass were higher in more saline water ( The LINKTREE analysis divided the beam trawl sites into three groups based on thresholds of mean water depth and minor grain size: 1) Shallow, 2) Deep and Coarse Grained, 3) Deep and Fine Grained (Figure 3.15). The "Deep" habitat group was characterized by water depths greater than 39 meters. The "Shallow and Coarse" habitat group was characterized by water depths less than 35 meters and minor grain size between 3.9 and 5.8 (shell debris or pebble). The "Shallow and Fine" habitat group was characterized by water depths less than 38 meters and minor grain size between 8.45 and 9 (fine sand or mud).
These habitat groups exhibit similar patterns as the species assemblage groups defined by cluster analysis, but they do not fully explain the fish community structure observed via beam trawl sampling (Figure 3.15). Thus, there are likely additional habitat characteristics that were not incorporated in this analysis that influence the structure of beam trawl species assemblages. There are, however, significant differences in species assemblages between habitat groups (R=0.582, p=0.001). More specifically, the "Deep" habitat group was primarily occupied by fish communities with high abundances of sea scallops, sand dollars, and sea stars, where as the "Shallow and Coarse" habitat group was dominated by fish communities with high abundances of skates and cancer crabs, and the "Shallow and Fine" habitat group was characterized by fish communities with higher abundances of silver hake and American lobster. There are also significant differences in the habitat characteristics that are associated with each beam trawl species assemblage group, although between-assemblage habitat differences are not as consistent or pronounced as with otter trawl assemblages (ANOSIM: R=0.275, p=0.021).
In a spatial context, the "Deep" habitat group is primarily located along the southeastern flank of Cox's Ledge, where the "Sea Scallop and Sand Dollar" and "Sea Scallop and Sea Star" species assemblages dominate (Figure 3.16). The "Shallow and Fine" habitat group, on the other hand, is located in the inshore extent of Rhode Island Sound, where it exhibits significant overlap with the "Silver Hake and Lobster" species assemblage group. Finally, the "Shallow and Coarse" habitat group is located around the southern end of Block Island, an area primarily occupied by the "Skates and Cancer Crabs" species assemblage.

Discussion:
The fisheries ecosystem of Rhode Island and Block Island Sounds is composed of many environmental factors, including water depth, water temperature, and benthic habitat heterogeneity. Understanding the relationship between these factors and the fish and invertebrate community is central to the protection of important habitats and the maintenance of ecosystem stability in the face of new ocean uses.
Thus, the work presented here represents fundamental progress towards ecologically sound spatial management decisions and the general advancement of ecosystem based fisheries management in Rhode Island's nearshore waters (RISAMP 2010, Fogarty 2013 While the preference of fish and invertebrate communities for specific depth ranges has been observed in a variety of ecosystems, such strong, system-wide patterns were previously undocumented in Rhode Island's nearshore waters ). In terms of otter trawl species diversity, depth related trends may be driven by the tendency of inshore waters to intensify the interactions between bentho-pelagic species, as the water column is truncated and benthicpelagic coupling is enhanced. Thus, inshore fish communities are more likely to be dominated by a few abundant species (scup, skates, silver hake), therefore reducing the diversity of the fish community (Scharf et al. 2000 A general paradigm about marine benthic communities is that as bottom roughness increases from smooth mud and sand to cobble and boulder ecological complexity and species diversity increase . The presumed relationship is that the more heterogeneous the habitat, the more species it can support because more niches are available (Guegan & Oberdorff 2000. This pattern appears to hold true in Rhode Island's nearshore waters, such that areas with higher backscatter intensities and, thus, coarser sediments, support more diverse fish and invertebrate communities . From a multivariate perspective, three measures of habitat roughness were found to be influential in structuring fish and invertebrate species assemblages in Rhode Island and Block Island Sounds, with the standard deviation of surface roughness and mean slope important in shaping otter trawl assemblages and minor grain size important in shaping beam trawl assemblages. Such measures of seafloor roughness, however, did not wholly explain the patterns observed in fish and invertebrate assemblages, and thus, must be considered in combination with other habitat parameters, such as oceanographic conditions and water depth.
By nature, the benthos is an intricate system, characterized by a collection of distinct environmental parameters. Relationships between such habitat parameters and fish communities has been well documented in coral reefs and seagrass beds, but the work presented here is novel to the temperate, nearshore environment of Rhode Island and Block Island Sounds (Ault & Johnson 1998, Christensen et al. 2003 is not defined by one distinctive parameter, but rather a combination of seafloor and oceanographic features.
Consistencies between the habitat groups and demersal fish assemblages identified in this study further suggest that the fish community in Rhode Island and Block Island Sound is shaped by the physical environment. Since most habitat features are relatively static and most fish and invertebrates are mobile, the fish community is likely shaped by the environment and not vice versa. One ecological mechanism that may account for this fish-habitat association is the interaction of predators and prey

Discussion:
As is outlined in the manuscripts of this dissertation, there are many factors, both biotic and abiotic, that influence the structure and function of the demersal fish and invertebrate community in Rhode Island and Block Island Sounds. It is logistically infeasible, however, to assess each and every one of these factors in a short term research project such as this. Thus, I hope to use this concluding section to consider additional ecological and environmental factors that may influence the fisheries ecosystem dynamics in Rhode Island's nearshore waters, as well as to discuss the theoretical and practical implications of this work.
In terms of spatial structure of the fish and invertebrate community, there are many factors that could play a role that were not addressed explicitly by this work.
For example, the schooling behavior of certain fish species may influence the structure and spatial distribution of the fish community. Previous work has shown that large aggregations of prey attract schools of predators, which, in turn, shape the fish community through top-down control .
Evidence of this phenomenon in Rhode Island and Block Island Sounds is apparent in the diet analysis and spatial distribution of the spiny dogfish, Squalus acanthias, and longfin inshore squid, Doryteuthis pealei (chapter 3, Gerry 2008). Spiny dogfish are opportunistic feeders and are known to exhibit schooling behavior, therefore, dominating the assemblage and size of the fish community when they are present (chapter 2). In Rhode Island and Block Island Sounds, spiny dogfish, along with summer flounder and winter skate are key predators of longfin squid, a common schooling species (chapter 3). This suggests that squid inhabit both the benthic and pelagic realm in Rhode Island and Block Island sounds and, therefore, attract bottom feeders (e.g. summer flounder, winter skates) as well as semi-pelagic feeders (e.g. spiny dogfish, striped bass). Thus, the predator-prey interactions and schooling behaviors of dogfish and squid appear to play an important role in the fisheries ecosystem dynamics of Rhode Island's nearshore waters. Techniques, such as midwater trawls or acoustic surveys, would be best suited for testing this hypothesis .
Scup, Stenotomus chrysops, are similar to dogfish in their schooling behaviors . Scup, however, are smaller and more benthivorous in their feeding regime and, therefore, tend to school in areas with aggregations of small benthic prey, such as amphipod tube mats  are key prey items for many species (chapter 3, Smith & Link 2010). As the development of new ocean uses proceeds, it will be important to protect such unique benthic habitats and the food resources they provide so as to sustain vulnerable groundfish species and maintain overall ecosystem balance.
With respect to the interplay between species assemblages and trophic structure, the results of this work suggest that both bottom-up and top-down trophic cascades play a role, as otter trawl and beam trawl species assemblages were characterized by a wide array of species, including predators (spiny dogfish, summer flounder, silver hake), planktivores (sea scallop), detritivores (American lobster, Cancer crabs), and omnivores (scup, skates, winter flounder) .
In the context of bottom-up trophic mechanics, planktivore species would be the first fishes to respond to changes in primary productivity, with predator populations changing in response to availability of their food source (planktivore species) ). Conversely, top-down trophic cascades are based on the theory that predators structure the ecological community via predation, such that an increase in predator populations (dogfish, bluefish, striped bass) leads to a decrease in prey species abundance (squid, herring, butterfish) . Top predators usually take many years to reach maturity and may commit substantial parental investment to each offspring (ovovivipary or vivipary). Thus, even small changes in the number of spawning adults in predator populations can have long-term impacts on fish community structure, including prey resources. This process is exemplified by the initial decline of the northern cod, Gadus morhua, population and the subsequent increase in its primary prey species, crabs and lobster ). Thus, when attempting to predict the effects of development and exploitation on the fish and invertebrates community in Rhode Island's nearshore waters, it is essential to consider such trophic cascades, as impacts to specific species will likely propagate throughout the food web.
The mobility of most fish and invertebrate species is a factor that must be considered when discussing spatial patterns in species assemblages, trophic structure, and habitat use in temperate marine environments such as Rhode Island and Block Island Sounds, particularly at the fine spatial scale of this research. The mobility of fishes allows them to move between ecosystems and habitats at will, thus obscuring spatial patterns in diet and isotopic signatures and reducing the measurability of habitat associations ).
However, some fish exhibit strong site fidelity or habitat preferences, which can improve our ability to detect fine scale trophic structure and habitat use . In Rhode Island and Block Island Sounds, such a phenomenon is evident in the persistent isotopic spatial patterns of winter flounder and black sea bass, species known for site fidelity, versus the absence of spatial structure in the isotopic signatures of highly mobile herring and scup (chapter 3, . Similarly, the amenability of sessile or slow-moving species (which are more strongly associated with specific locations) to fish-habitat research is also evident in Rhode Island and Block Island Sounds, where species assemblages characterized by sea scallops, skates, crabs, and lobster (less mobile species) exhibit persistent habitat associations (chapter 4).
Another factor that potentially impacts the structure and function of demersal fish communities is ontogenetic shifts in diet. Although we did not achieve large enough sample sizes to statistically assess ontogenetic patterns of the fish species in Rhode Island and Block Island Sounds, exploratory analyses suggest that a number of species exhibited size-based shifts in diet and isotopic signatures. Spiny dogfish presents one of the best examples of this phenomena, as young spiny dogfish exhibit planktivorous feeding behavior, where as adults exhibit more piscivorous foraging strategies . These tendencies were evident in the elevated δ 15 N and trophic positions of larger spiny dogfish in Rhode Island and Block Island Sounds. Bluefish also exhibited enriched δ 15 N at larger sizes, again reflecting a shift towards piscivory around age 1 . For most species, however, our otter trawl surveys did not effectively capture a wide variety of size classes, which limited our ability to fully assess ontogenetic shifts in diet and isotopic signatures.
Size-based patterns in habitat use may also influence the structure of the demersal fish community in Rhode Island's nearshore waters. Red hake provides a good example of this, as it exhibits a symbiotic relationship with scallops during early juvenile stages and a preference for sandy habitat as adults . The methodologies employed for this research, however, are insufficient to assess red hake's size-based habitat use in Rhode Island's nearshore waters. The American lobster is also known to exhibit ontogenetic patterns in habitat use, but given the low catch efficiency of lobster in otter and beam trawls, additional trap-based sampling programs would be needed to fully assess this relationship in Rhode Island and Block Island Sound.
The fish-habitat relationships established by this work provide a useful step towards the delineation of Essential Fish Habitat (EFH) in Rhode Island and Block Island Sounds . Essential Fish Habitat is defined as the environment(s) required for the successful spawning, feeding, recruitment, and growth to maturity of fished species and their prey . EFH refers to both abiotic and biotic habitat features, and is inclusive of both water-column and seafloor environments. Thus, essential fish habitat may include: spawning grounds, migration corridors, nursery grounds, foraging grounds, and theoretically, larval conduits. A common approach to determining EFH for a given species is to identify the distribution patterns of each life stage throughout the year, and to classify the habitat in areas where high densities of individuals are found. While my dissertation research deviated from this classical design, its identification of spatial patterns in species assemblages and habitat use are certainly applicable to EFH delineation in Rhode Island and Block Island Sounds. More specifically, the results of this work suggest that the deep waters surrounding Cox's Ledge are important in supporting economically valuable species, such as sea scallops and lobsters. Furthermore, the area immediately south and east of Block Island exhibits marked habitat heterogeneity, and thus, is likely an important environment for the early life stages of many fish and invertebrate species. The fish-habitat relationships established by this work are particularly timely as a series of closed areas have been proposed in Rhode Island Sound with the purpose of protecting essential fish habitat. In order to substantiate the classification of EFH in Rhode Island and Block Island Sounds, however, further research is needed to establish the functional relationships between individual fish species and location-specific habitat features and verify their persistence over time. In addition to application in marine reserve and closed area planning, the delineation of EFH is also key to the general advancement of ecosystem-based fisheries management .
Interestingly, the same areas in Rhode Island Sound that have been proposed as EFH closed areas have also been leased for development of a large-scale (200+ turbine) offshore wind energy facility. Considered theoretically, offshore wind energy development could have a number of impacts on the fisheries ecosystem in Rhode Island and Block Island Sounds, including but not limited to: habitat alteration via scouring, sedimentation, and construction of turbine support structures, shifts in surface and subsurface currents around and within turbine fields, changes in pelagic and benthic productivity and the associated trophic cascades, and modification of foraging behaviors and migration patterns due to electromagnetic fields. With respect to direct impacts on fish and invertebrate communities, sedimentation could smother sessile species (i.e. sea scallops), scouring could create inhospitable environments surrounding turbines, alteration of surface and subsurface currents could advect larvae to unsuitable habitats, reduced productivity could limit food availability (or vice versa), and EMF around cables could obstruct inshore-offshore migrations (i.e. lobster) or attract elasmobranch predators to false food sources within the windfarm field and along the cable route to shore. Furthermore, turbine construction would introduce large structures into the relatively low relief seafloor of Rhode Island Sound, providing high relief habitat for some species and eliminating essential low relief habitat for other species. From an ocean-use context, windfarms are often closed to fishing and can act as de-facto marine reserves, reducing fishing mortality and potentially increasing fish biomass. Thus, with the true ecological repercussions of offshore wind energy development yet to be seen, research such as this is essential to begin to understand, predict and mitigate impacts to fisheries ecosystem dynamics in areas slated for wind energy development. Overall, as the designation of essential fish habitat and/or the development of offshore wind energy facilities proceeds in Rhode Island and Block Island Sounds, this research will play a critical role in the development of new ocean use policies and the advance of ecosystem-based fisheries management.