SATELLITE INVESTIGATION OF DRIVERS OF PHYTOPLANKTON COMPOSITION IN A CONTINENTAL SHELF ECOSYSTEM

Phytoplankton size classes (PSCs) play an important role in ecological and biogeochemical processes in marine ecosystems. Studying the dynamics of phytoplankton groups and the environment can inform how we manage marine resources, including fisheries, which are of critical importance in the Northeast U.S. continental shelf ecosystem. This study aims to shed light on phytoplankton community structure distribution and the environmental drivers of their variability in the Northeast U.S. continental shelf from a satellite remote sensing perspective, which has a captivating ability to view the global ocean and its dynamics on large spatial scales and long-term temporal scales. 23 years (1997 to 2019) of imagery and reanalysis products were used to understand variability in the main drivers over time and applied an existing phytoplankton size classes algorithm to ocean color satellite products. The variables affecting the abundance and distribution of PSC in the Northeast U.S. continental shelf are diverse and intricately intertwined by season and region. Above all, chlorophyll-a concentration and euphotic depth are the most significant associations of all phytoplankton size class distribution in all seasons and regions, followed by salinity and mixed layer depth. Studying changes in such climate-forced phytoplankton community composition plays a crucial role in the management of energy and resources in the region from a fisheries perspective and in preparing for future climate change. iii ACKNOWLEDGMENTS I would like to express my respect and gratitude to Dr. Colleen B. Mouw, my mentor and advisor, who gave me a lot of advice, help, and support both inside and outside of my research during my master’s program. I also would like to thank Dr. Yeqiao Wang, Dr. Melissa Omand, and Dr. Kimberly Hyde for serving as my committee members and to thank Dr. Matthew Bertin for serving as my defense chair. I would like to acknowledge my lab manager, Audrey Ciochetto, for helping me technically with data processing and coding. I am also grateful to have my lab mates Virginie Sonnet, Rowan Cirivello, Vitul Agarwal, and Matt Guanci, for being a great team and always having constructive discussions and communications. Lastly, I would like to thank my parents, Dr. Byeong-seok Song and Moonhui Kim, and my sisters for sending me love and support from afar. This research was funded by the Established Program to Stimulate Competitive Research (EPSCoR), Rhode Island Consortium for Coastal Ecology Assessment Innovation & Modeling (RI C-AIM) of the National Science Foundation, and NOAA Joint Polar Satellite System (JPSS) Proving Ground and Risk Reduction program. I would also like to acknowledge several projects that have contributed to the advancement of ocean science by enabling satellite and reanalysis data available to public. This includes Ocean Color Climate Change Initiative (OC-CCI), Global Ocean Physics Reanalysis (GLORYS) from the Copernicus Marine Environment Monitoring Service, and the Advanced Very High Resolution Radiometer (AVHRR) Pathfinder. iv PREFACE This thesis manuscript has been formatted to meet the criteria set by the Graduate School of the University of Rhode Island. This work is intended to be submitted to Continental Shelf Research for publication.


INTRODUCTION
Phytoplankton form the base of the marine food web and are a crucial component in Earth's carbon cycle. Phytoplankton composition impacts the structure, function, and sustainability of the marine food web (Margalef, 1978;Mouw et al., 2017).
Phytoplankton community composition and biomass are highly variable in time and space, changing in response to both bottom-up (i.e., nutrient availability, environmental conditions) and top-down variables (i.e., grazing, mortality) (Houliez et al., 2021;Margalef, 1978;Longhurst, 1998;Cullen et al., 2002;Turner et al., 2021). Thus, it is critical to understand the dynamics of phytoplankton regarding both abundance and community structure for a more comprehensive understanding of biogeochemical and ecological roles, including their influence on the marine food web.
Since phytoplankton are composed of diverse taxonomic groups which are manifested as distinct morphology, size, and pigment composition, they can be detected with ocean color satellite radiometry (Mouw et al., 2017). There have been many studies that have retrieved information on phytoplankton composition using satellite estimates, these have included phytoplankton size classes (PSC), phytoplankton functional types (PFT), and particle size distribution (IOCCG, 2014). The majority have been PSC algorithms given that phytoplankton cell size is considered a fundamental trait that affects many important biological and ecological processes (Litchman and Klausmeier, 2008). with in situ PSC estimates derived from a regional HPLC pigment dataset.
It is important to study the ecosystem of the Northeast U.S. Continental Shelf using various approaches since the region is highly productive and provides ample energy and economic resources for the biome's species ranging from microbes to humans. In addition, the region contains diverse marine environments, providing essential habitat for breeding, spawning, and feeding of abundant marine life (Mills et al., 2013;Goode et al., 2019;Zang et al., 2021). The Northeast U.S. continental shelf is identified by diverse water masses (Townsend et al., 2006). The warm and saline northward-flowing Gulf Stream and the colder and fresher southward-flowing Labrador Current converge in the region, forming the source waters for the continental shelf (Loder et al., 1998;Fratantoni and Pickart, 2007;Greene et al., 2013;Richaud et al., 2016;Cai et al. 2021). The surface layers are dominated by shelf waters entering the Gulf of Maine from the north, while deeper layers are sourced by a mixture of slope waters from the north and south entering the Gulf of Maine through the Northeast Channel (Mountain and Manning, 1994;Mountain and Taylor, 1998;Mountain, 2012).
The shelf circulation and annual cycle of heating, along with influxes of freshwater from riverine sources, result in local variability of water properties (Cai et al., 2021).
Across the physically dynamic and productive Northeast U.S. Continental Shelf, phytoplankton respond and adapt to various physical, biological, and ecological conditions. Phytoplankton community composition and abundance in the region vary seasonally. In the Gulf of Maine and George's Bank region, diatoms dominate in a typical winter-spring bloom, and dinoflagellates, cryptophytes, and cyanobacteria become more prevalent during the summer (O'Reilly and Zetlin, 1998;Pan et al., 2011;Richaud et al., 2016;Turner et al., 2021). Also, one of the most obvious spatial patterns is the decrease in phytoplankton abundance from the coast to the shelf break (defined as the 100-m isobath  (Cai et al., 2021). While warming, acidification, and deoxygenation are occurring globally, these factors are playing out distinctively in the North Atlantic and the Northeast U.S. Continental Shelf within it (Pershing and Stamieszkin, 2020). These environmental changes are important because they have the potential to affect the broader marine food web and global biogeochemical cycles within the region (Falkowski et al., 1998;Henson et al., 2012;Barton et al., 2014).
Previous researchers have studied the relationship between the marine ecosystem and environmental conditions. Pershing and Stamieszkin (2020) (Barton et al., 2016;Pershing and Stamieszkin, 2020).
For example, the increased stratification supported elevated phytoplankton abundance during the fall and winter but less intense spring blooms (Greene and Pershing, 2007;Ji et al., 2007). This, in turn, affected the zooplankton and upper trophic levels (Pershing and Stamieszkin, 2020). A comprehensive understanding of the environmental dynamics and long-term changes of the Northeast U.S. Continental Shelf is imperative to understnading the impacts of climate-forced phytoplankton community composition changes and supporting ecosystem-based fisheries management.
Ocean remote sensing imagery has allowed the quantification of abundance and composition of the phytoplankton community and numerous physical parameters to be assessed on large spatial scales and long temporal scales. These products present an unprecedented view of ocean dynamics due to their ability to view the global ocean nearly every other day, while having the caveats of detecting only the surface of the ocean (Mascarenhas and Keck, 2018). Furthermore, the availability of phytoplankton composition (as size classes or functional types) from ocean color imagery is relatively recent and presents a more thorough surface view of spatial and temporal patterns than previously available from in situ observations.
There have been various assertions on the influence of various biological, chemical, and physical oceanic factors on the resulting phytoplankton community composition (Barton et al., 2014;Dutkiewicz et al., 2020;Raven et al., 2021). However, this long-term satellite-based study provides another perspective to understanding the connection between various environmental variables and phytoplankton composition in the Northeast U.S. continental shelf (Turner et al., 2021;NOAA Northeast Fisheries Science Center, 2022).
This study aims to shed light on studying the phytoplankton community structure distribution and the environmental drivers of their variability in the Northeast U.S. continental shelf from a satellite remote sensing perspective. A 23-year time series of regionally tuned phytoplankton size class imagery was compared with coincident physical, chemical, and biological observations to better understand the drivers that impact phytoplankton community variability. This study addresses the potential to use high temporal and spatial information of satellite imagery to revisit the question: What are the most important environmental drivers of phytoplankton community composition variability in the Northeast U.S. continental shelf? The question is resolved through the following goal: Investigate statistical relationships between long-term variability of phytoplankton composition and environmental parameters using remote sensing satellite imagery in the Northeast U.S. continental shelf.

Satellite Imagery and Reanalysis Products
Satellite products from September 1997 through December 2019 were derived from a variety of missions (Table 1)  Therefore, all two-dimensional directions of the ocean surface velocity were considered with the two variables ( ! , ! ) from GLORYS.

Data Processing
The final resolution of all satellite products for the data analysis was monthly composites with 9-km spatial resolution. The monthly imagery was further averaged into seasons. Spring included March, April, and May, summer included, June, July, and August, autumn included September, October, and November, and winter included December, January, and February. The standards of monthly and 9 km resolution were established based on the spatial and temporal scales of importance to broad-scale physical and biological features in the Northeast U.S. continental shelf. This spatial resolution provided a balance between capturing important oceanographic features and computational efficiency. When the original product was not available in a monthly 9km product, the product was re-gridded, and monthly composited.  The monthly OC-CCI ocean color products were spatially smoothed from an initial resolution of 4-km (0.04°) to 9-km (0.09°) using a maximum likelihood estimator for every two pixels, allocating the geometric mean inside of the merged pixels and averaged to monthly resolution (Mouw et al., 2019). The same approaches were applied to SST imagery. GLORYS products that include salinity, sea surface height, eastward velocity, and northward velocity were on the same projection and the pixels were the same distance apart as the OC-CCI data, but they were centered ½ a pixel width from the OC-CCI data, resulting in a spatial resolution approximatively 8.3-km (1 12 ⁄°).
Therefore, I matched the grid by smoothing the GLORYS data in a 2-by-2 pixel dimension and matched the value into the unified OC-CCI grid.
To further prepare the imagery datasets for statistical analysis, I tested the datasets for normal distribution, eliminated outliers, and filled data gaps. First, I performed a skewness test for each satellite imagery product to investigate if the dataset had a normal distribution. The datasets that were not normally distributed were lognormally distributed and log-transformed , and MLD). For quality control purposes, outliers larger than five standard deviations from the mean were eliminated and any temporal gaps of six months or less were filled using spline interpolation. Finally, the long-term trend and the monthly climatological cycle of each pixel were removed from the dataset since the purpose of this thesis was to determine the temporal and spatial variability of the dataset over a long time period. Figure 2 represents the average values of the pre-processed imagery datasets considered in this study.

Phytoplankton Community Composition Algorithm
Phytoplankton groups were derived from algorithms using the 9-km monthly preprocessed satellite products as inputs. In determining the best-performing PSC approach for the U.S. Northeast continental shelf, Turner et al. (2021)

Analysis
I applied the Theil-Sen approach (Barton et al., 2014) and the partial least squares regression (PLSR) (Wold et al., 2001) to assess the changes in environmental variables over time and to determine the relative importance of each environmental parameter to phytoplankton composition. The methods were the same as those used by Mouw et al. (2019). After the preprocessing steps described above, the data were divided into the sub-regions. To investigate the long-term environmental change trend in the study area, I first performed time series analysis using Theil-Sen approach. This method retrieved slopes as the median of the distribution of slopes between every pair of points in the dataset, resulting in Kendall rank correlation coefficient (Barton et al., 2014). The PLSR analysis uses tenfold cross-validation combining preprocessed and normalized predictor variables into principal components that are then regressed with phytoplankton composition, resulting in variable importance for the projection (VIP) scores (Wold et al., 2001). VIP scores represent the relative importance of each predictor to the response variable. The correlation coefficients were also calculated to investigate the magnitude and direction of the relationship between the predictor variables, environmental parameters in this study, and the response variables, PSCs in this study.
Finally, to increase the reliability of the calculated VIP scores and correlation coefficients, I performed leave-one-predictor-out validation (Martens & Martens, 2000).
This is an approach that first includes all the variables, then removes the predictor variables one-by-one and repeats the PLSR analysis to derive the results of average coefficients and VIP scores, and to create an error bar using the minimum and maximum of the resulting values.

Time Series Analysis
Changes in each of the environmental variables and PSC with time were investigated for each season. For the time series analysis, a linear trend was derived, defined as t using the Theil-Sen approach. A positive t means that the change of the variable has an increasing trend over time, and a negative value means that the variable has a decreasing trend. Overall, the increasing tendency was prevalent in all PSC in all seasons, particularly for microplankton ( Figure 4). This suggested the possibility that the marine environment of the region has become nutrient-rich, resulting in increased microphytoplankton (Siokou-Frangou et al., 2010). In the case of spring and winter, phytoplankton tended to increase in offshore areas and to increase in summer and fall in coastal areas over time. In summer, the increasing trend was strong along the path of the Gulf Stream (Figure 2g and h). Given phytoplankton size classes and Kd(490) were derived from chlorophyll concentration and euphotic depth is calculated from Kd (490), similar trends were found in chlorophyll-a concentration and euphotic depth ( Figure 5).
In addition, nano and microplankton's increasing tendencies were observed along the coast in summer and fall ( Figure 5).
The trend of each environmental variable is shown in Figure 5. The observed deepening trend in MLD along the MAB and WGOM regions in spring and especially in summer had similar trends with some other environmental variables. Chlorophyll-a concentration tended to increase most steeply in all seasons, especially in the spring and summer seasons. The intense deepening trend of MLD in summer was also reflected in the same pattern in the increasing trend of *+ . ! and ! became faster over time in the northern part of the study area and GB region, which could be interpreted as the marine physical activities in the area became more intense and dynamic, and the same pattern was projected in SSH trend as well. It also was observed that the tendency of salinity increased along with the flow of the Gulf Stream in summer. SST increased overall but was especially intense in the winter. This may indicate a warming tendency over time in the winter season.  Figure 5. Seasonal changes in environmental variables over time using long-term linear trend from the Theil-Sen approach. Furthermore, as expected, it was found that the relative importance score of [Chl] increased when the cell size increased.

VIP Scores
The variable showing the highest score after [Chl] was the euphotic depth ( *+ ), and the relative importance of *+ in each PSC by season and region is shown ( Figure   7). *+ had almost the same spatial and temporal pattern as [Chl] because [Chl] was used directly in the calculation of the Kd(490) product used to calculate *+ . As [Chl] increases, *+ becomes shallower. The most obvious result was that in all sizes of phytoplankton and all seasons, the impact of *+ variability on PSC increased with increasing latitude. This results due to the north-south gradient of daylight availability depending on the season. Also, the results showed that the impact of )* on PSC was a proportional relationship; as the size of the phytoplankton decreases, the relation with )* decreases. In winter, the overall VIP score was low, especially in MAB and WGOM regions.  for each size class, the VIP scores were not high compared to other environmental parameters, and some regions and seasons did not show statistically significant relationships (VIP score >0.5). Thus, it can be inferred that SST variability did not play a large role in driving the abundance of phytoplankton classes, even though the overall SST was important in improving PSC retrieval from satellite imagery (Turner et al., 2021). Overall, in spring and autumn, significant VIP scores were found in the MAB and WGOM and in the EGOM and GB in summer. This may indicate that SST variability played a large role in spring and autumn phytoplankton bloom in the North Atlantic, suggesting it could be used as an indicator of ocean stratification, which is one forcing of phytoplankton variability (Behrenfeld et al., 2006). In that regard, in summer, stratification dampened SST variability, possibly resulting in lower VIP scores.
SSH also did not show high impacts on phytoplankton except for the winter season ( Figure 9). There was no significant importance in spring and summer, and autumn for picophytoplankton (VIP score < 0.5). In autumn and winter, high VIP scores were shown in GOM and GB regions, affected by the dynamic nature of the circulation and tidal mixing, particularly due to winter storms.   The salinity VIP scores for seasonal PSC were shown in figure 10. Overall, the interrelationship between phytoplankton abundance and salinity variability in MAB and WGOM regions was high due to terrestrial, freshwater inputs that were likely also bringing nutrients needed for phytoplankton growth. In addition, the VIP score was high in the MAB every season due to the PSC variability responding to salinity variability across the region from the warm, saline Gulf Stream balanced by the freshwater inputs coming from land. Likewise, the reason why salinity in the GOM had a relatively low influence on PSC was that the relative outflow of fresh water from the St. Lawrence River and Penobscot River on the northern coast of the GOM was low in comparison to the volume of surface water in the region (Richaud et al., 2016). Thus, they did not introduce substantial variability to the overall salinity of the whole region. In spring and especially summer, the VIP scores were relatively lower compared to other seasons due to weaker variability of salinity from continuous freshwater input (Richaud et al., 2016), and therefore, other environmental parameters more strongly drove the variability of PSCs at that time.     Figure 12. Variable influence on projection (VIP) scores for eastward velocity in each season and phytoplankton size classes (PSC). VIP score under 0.5 represents insignificant. Figure 12 represented the seasonal relationship between PSC and ! variability.
Overall, The VIP score was higher in the nearshore areas (although not always above the significance threshold), likely due to the decadal average currents flow offshore near the coast (Roarty et al., 2020). In addition, it showed the highest values of VIP scores in the order of winter and then spring. This can be explained by considering the ! dynamics of the region were related to the wind (Li et al., 2014), and the strong winter and more variable wind generated the well-mixed upper ocean layer presenting higher VIP scores and relative importance of ! in winter. Moreover, high near-surface nutrient concentrations in the winter-spring transition in the region that cause spring bloom with increased chlorophyll concentrations (Townsend and Thomas, 2001)  represented as indicators of modes of nutrient delivery. In addition, in these seasons, it was found that cell size and relative importance score are inversely proportional. Figure 13. Variable influence on projection (VIP) scores for northward velocity in each season and phytoplankton size classes (PSC). VIP score under 0.5 represents insignificant.
In the seasonal relationship between PSC and + (Figure 13), the Gulf Stream governed the overall + in the Northeast U.S. continental shelf. This could be identified by the relatively high VIP score in MAB regions, and in winter and spring when the Gulf Stream is has been documented to be strong (Worthington, 1976). However, the strongest relative importance was found in the summer nearshore regions. This was not because of the Gulf Stream, since it weakens in summer, but because of the observed eddies that were more energetic during summer, when surface waters were less dense than bottom waters, compared to winter when the ocean was well mixed (Kirincich et al., 2022). Similar to the size pattern shown in + , the smaller the cell size in + , the higher the VIP score indicated the relative importance and influences. Table 2 summarizes the results of the VIP scores representing the relative importance of each oceanic variable to the PSC by season.  significant relationships in all size classes. S and MLD had the highest VIP scores after [Chl] and *+ , meaning they were the second most significant drivers of phytoplankton size class distribution for all seasons. The environment variables that showed a notable difference by cell size were + , and + . That is, the VIP scores of + , and + decreased as the cell size increased.
The results of which parameter variability affects phytoplankton by region by season were summarized as follows ( Figure 17). The influence of velocity ( + , + ) likely due to the decadal average currents flow offshore was prominent to phytoplankton community composition in the southern part of the study area (SMAB). In addition, the dominant influencing factor in the SMAB was salinity in spring and fall, but in spring phytoplankton community appeared to be more affected by wind-related factors' variability, such as + and MLD. In the case of fall, the variability of SSH, which in this region was heavily influenced by local bathymetry, was the main influencing factor of phytoplankton. In winter, current movement ( + , + ) and the resulting MLD had a large impact under the influence of strong winds. This often resulted in favorable conditions for phytoplankton growth, which was especially observed in smaller sized pico-and nanophytoplankton with higher VIP scores, as seen in figure 14 and 15. Furthermore, the regions within the area of influence of the warm and saline Gulf Stream were also affected by S and SST variabilities. SSH was a strong influencing factor in the GB and EGOM, particularly in fall and winter due to seasonal feature of winter storms that caused strong tidal mixing and circulation.   Figure 15. Same as Figure 14 but for nanophytoplankton.

DISCUSSION AND CONCLUSIONS
The question posed in this research was 'What are the most important environmental drivers of phytoplankton community composition variability in the Northeast U.S. continental shelf?'. The Northeast U.S. continental shelf is an environmentally dynamic, highly productive, and economically important region.
Therefore, it is important to understand the environment and the ecology of the region for human activities and marine life. I investigated the time series of a suite of environmental parameters and PSC to understand the long-term relationships and variability of phytoplankton size classes to environmental parameters that can be remotely sensed and the relative importance of each environmental variable to PSC through statistical analysis.
Through this study, I found that the variables that affect PSC varied regionally.
In the MAB (including both the NMAB and SMAB) and WGOM, many variables were important concurrently, including salinity, MLD, SST, and + and + . Salinity variability was associated with terrestrial freshwater input, the MLD was well-mixed (Cai et al., 2021), SST was particularly relevant to blooms in spring and autumn as an indicator of the ocean stratification (Behrenfeld et al., 2006), and + and + were important and influential due to decadal average current flows (Roarty et al., 2020), and transient summer eddies (Kirincich et al., 2022).
In the EGOM and GB, SSH was a notable influencer, especially due to bathymetric features, strong tidal activity in spring and winter, and also by being an indicator of upwelling and downwelling that drive nutrients availability (NOAA Northeast Fisheries Science Center, 2021). Salinity had the least impact in these regions.
)* showed an increasing influence on PSC with increasing latitude and decreasing influence in winter due to less sunlight and more clouds. Likewise, a deep MLD in winter was an important influencing factor for that season. Previous studies (Barton et al., 2016;Pershing and Stamieszkin, 2020) suggested that indirect effects of climate change besides warming (i.e., enhanced stratification) were likely the main driver of ecosystem changes in the North Atlantic, including the study area. Similarly, these studies further suggested enhanced stratification due to a well-mixed surface layer was likely the driver of ecosystem changes. During the fall and winter seasons, stratification continued to increase from summer, resulting in elevated phytoplankton abundance but less intense spring blooms (Mountain and Manning, 1994;Greene and Pershing, 2007;Ji et al., 2007).
The variables that had the most significant relationships on the variability of phytoplankton size classes were [Chl], )* , + , and + .
[Chl], and )* showed a proportional relationship with the size, and physical variables ( + , and + ) had an inverse relationship, indicating an increase in smaller cells when + , and + were slower.
Larger cells need greater mixing to keep them in suspension in the water column and vice versa. Taken altogether, synthetically, the variables associated with the abundance and distribution of PSC were [Chl], )* , S, and MLD, which were consistent with the previous studies (Mouw et al., 2019). Mouw et al. (2019) reported that in warm ocean regimes, which the Northeast U.S. continental shelf broadly falls within, [Chl] was more important than physical parameters that may be indicative of modes of nutrient delivery.
Nevertheless, the study also highlighted the importance of )* and MLD as influencing variables in the region's phytoplankton variability, which were representative of light availability and nutrient delivery from mixing, respectively.
Variables affecting the abundance and distribution of phytoplankton in the Northeast U.S. continental shelf were diverse and complexly intertwined. Moreover, since this study considered only environmental parameters acquired by satellites and reanalysis products, some important variables, such as nutrients, were not considered.
Also, from previous studies, it was found that the traditional "bottom-up" view that focuses solely on local changes in atmospheric forcing and ocean conditions may not be sufficient to understand how phytoplankton communities respond to changing climates, both in the historical record and in anthropogenic climate change scenarios (Barton et al., 2014). Dutkiewicz et al. (2020) also claimed the importance of considering complex trophic interactions (such as a decrease in the stocks of a predator's grazer) leading to an unexpected reshuffling of the phytoplankton community structure.
To thoroughly understand the dynamics and influencing factors of the phytoplankton of the Northeast U.S. continental shelf, various environmental variables such as nutrients or top-down factors should be included.
Satellites only observe the surface of the ocean corresponding to the first photic depth. This means that the approach of this research can lead to different results from those derived from using data directly acquired from the ocean or model results that consider the full water column. Furthermore, in this study, the spatiotemporal scale was reduced to 9 km to incorporate various satellite datasets available on different resolutions, and the data was further reduced to ecoregions. Thus, the results were indicative of large-scale, average patterns in these regions. This indicated that finer-scale studies are likely to differ from the results presented here. To comprehensively study phytoplankton community composition dynamics and ecology in the Northeast U.S. continental shelf, this study provided a satellite perspective. However, a variety of approaches, including observational data, empirical, and numerical models incorporating additional environmental variables, will provide additional insights.
This study investigated the broad-scale, long-term dynamics and potential influence factors of phytoplankton communities in the Northeast U.S. continental shelf region. Studying changes in such climate-forced phytoplankton community composition is important for regional to global carbon budgets, drives bottom-up processes throughout the pelagic food chain (Richardson and Schoeman, 2004), and plays a crucial role in the management of energy and resources in the region from a fisheries perspective, and in preparing for future climate change.