Food web structure and bioregions in the Scotia Sea: A seasonal synthesis

Abstract

Bioregionalisation, the partitioning of large ecosystems into functionally distinct sub-units, facilitates ecosystem modelling, management and conservation. A variety of schemes have been used to partition the Southern Ocean, based variously on frontal positions, sea ice, productivity, water depth and nutrient concentrations. We have tested the utility and robustness of ecosystem bioregionalisation for the Scotia Sea, by classifying spring, summer and autumn stations on the basis of nutrient concentrations, phytoplankton taxa, meso- and macrozooplankton, fish catches and acoustic data. Despite sampling across different seasons and years, at different spatial scales and taxonomic resolutions, cluster analysis indicated basically consistent spatial divisions across this wide range of trophic levels. Stations could be classified into two main groups, lying broadly to the north and south of the Southern Antarctic Circumpolar Current Front (SACCF). In some aspects the 2 station groups were similar, with both having variable but often high phytoplankton biomass as well as similar biomass of fish. However, the colder water southern group, most of which was covered by seasonal sea ice, had a fundamentally different food web structure to that in the northern Scotia Sea. The cold water community had a depleted, cold-adapted fauna characterised by low zooplankton biomass, Euphausia superba and the fish Electrona antarctica and Gymnoscopelus braueri. In contrast the northern group was richer with higher mesozooplankton biomass and a fauna of warmer or more cosmopolitan species such as Themisto gaudichaudii, Euphausia triacantha and the fish Protomyctophum bolini, Kreffticthys anderssoni and Gymnoscopelus fraseri. The position of the food web transition, broadly consistent with the position of the SACCF, supports a recent circumpolar-scale bioregionalisation. However, there is little evidence that this relatively weak frontal transition represents a significant barrier either here or elsewhere in the Southern Ocean. We suggest that broader-scale factors, namely temperature and possibly the extent of the seasonal ice-zone, within which most of the southern stations lay, were more likely to influence biological zonation.

Introduction

Partitioning the Southern Ocean into discrete regional units, based on ecological and physical properties, helps in understanding, modelling and ultimately managing the great biological heterogeneity present (Grant et al., 2006). Such partitioning can help define functionally similar or different sub-systems or foodweb types and their extent, which is also necessary to understand, monitor and conserve marine biodiversity. Consideration of baseline data from these regions also provides a measure against which future change can be determined.

Because of its great size and biological complexity, regionalisation in the Southern Ocean is clearly scale related. At the global scale it has been described as a single functional province in the context of Large Marine Ecosystems (Sherman and Duda, 1999) and with increasing complexity as comprising ice-free open water, seasonal pack-ice and shelf (permanent pack-ice) regions (Hempel, 1985). Tréguer and Jacques (1992) defined 4 functional units based on phytoplankton and nutrient dynamics encompassing the Polar Front Zone (PFZ), the Permanently Open Zone (POOZ), Seasonal Ice Zone (SIZ) and the Coastal and Continental Shelf Zone (CCSZ). Longhurst (1998) suggested a modification of this scheme based on ocean colour data and knowledge of the response of planktonic algae to physical forcing, that has 4 annular provinces extending from the Sub-Tropical Convergence (STC) separated by intervening frontal zones. The circumpolar W–E flow of the Antarctic Circumpolar Current (ACC) and the corresponding distribution of animals (Baker, 1954) tends to mean that community transitions are more apparent when travelling from N to S rather than E to W. More recently the Commission for the Conservation of Antarctic Marine Living Resources (CCAMLR) concluded that a primary regionalisation of the Southern Ocean comprising 14 sub-units could be achieved using just 4 properties namely, surface temperature, water depth, silicate and nitrate (Grant et al., 2006, SC-CAMLR-XXVI, 2007). These sub-units were again largely annular, reflecting current flow of the water masses involved, particularly the ACC and physical discontinuities within it. By incorporating two additional datasets, sea-ice duration and remotely sensed surface chlorophyll, the number of sub-units was increased to 40 (Grant et al., 2006). However all of the variables used for these various bioregionalisations were selected because of the availability of circumpolar data rather than a priori knowledge that they structure ecosystems. The contrasting zonations they produce are thus heavily dependent on the input data used. Grant et al. (2006) also concluded that further work to refine the primary regionalisation should focus on the inclusion of biological data, particularly at regional scales.

Foodwebs are not measurable using satellites alone and ship-based studies are therefore an essential and complementary part of the inclusion of biological data in such ecosystem partitioning. The question of how biological data can be used to enhance what are essentially physical regionalisations, was the focus of a second workshop organised by CCAMLR in 2007 (SC-CAMLR-XXVI, 2007/Annex 9, para 97). Among the methods used to extrapolate often sparse point biological data to the circumpolar domain were various modelling approaches such as boosted regression tree analysis, (Pinkerton et al., 2010) generalised dissimilarity modelling (GDM) (Koubbi et al., 2011) and species habitat modelling. Potential problems associated with these approaches include sparse data availability, sampling bias and extrapolation outside of the range of the data both in geographic and environmental space. Here we used a classification method, clustering physical data and point biological samples to test how well the physically derived clusters distinguished between different biological properties in this synthesis of seasonal food web structure across the Scotia Sea. We have investigated the utility of physically based partitions by assessing how various classes of biota, ranging from phytoplankton to mesopelagic fish, were distributed across the region on a seasonal basis and whether patterns evident at the base of the food-chain could be traced through to higher trophic levels.

The Scotia Sea is one of the better studied parts of the Southern Ocean and was the focus of the Discovery Investigations in the early and mid part of the 20th century (Kemp, 1929) as well as subsequent international expeditions and surveys (e.g. BIOMASS, El-Sayed, 1994; CCAMLR, 2000, Watkins et al., 2004). Its food-web structure has recently been reviewed by Murphy et al. (2007), although largely from a krill centred perspective. It lies downstream of Drake Passage in the Atlantic sector of the Southern Ocean, is some 750 km wide and is bounded on 3 sides by the Scotia Arc (Fig. 1) Seasonal sea-ice of variable extent covers the southern part during winter.

The Scotia Sea is a dynamic region, where following constriction at Drake Passage, mixing and overturning associated with interactions with bathymetry and wind stress significantly affects the circumpolar flow (Naveira Garabato et al., 2004). Embedded within it lie the Southern Antarctic Circumpolar Current Front (SACCF) and the Southern Boundary of the ACC (SB-ACC) (Brandon et al., 2004, Orsi et al., 1995). The former passes through the Scotia Sea before looping north-westwards around South Georgia, whereas the latter lies to the south and exits the Scotia Sea around the South Sandwich Islands.

Compared to the rest of the ACC the Scotia Sea is a region of high biological production which is generally associated with the interaction of the Antarctic Circumpolar Current (ACC) with bottom topography, particularly the shallow shelf regions of the Scotia Arc. This contrasts with the High Nutrient Low Chlorophyll (HNLC) conditions in the central Drake Passage, where waters from the deep basin of the land-remote SE Pacific enter the Scotia Sea (Whitehouse et al., 2012, Ardelan et al., 2010, Korb and Whitehouse, 2004, Korb et al., 2010, Park et al., 2010). Physical mixing and upwelling, typically in shelf regions, promotes the supply of limiting nutrients, particularly iron, to surface waters, which in turn stimulates phytoplankton production (Blain et al., 2007, Korb et al., 2008, Pollard et al., 2009). Marginal islands hold large concentrations of land based marine predators (Murphy et al., 2007) which are augmented by whale stocks in summer (Reilly et al., 2004) and which exist alongside commercial fisheries (Agnew, 2004). The significance of this region within the Southern Ocean can be gauged from the observations that phytoplankton biomass is much greater in the Scotia Sea and APF region between 10° and 60°W than in other pelagic Antarctic waters (Holm Hansen et al., 2004) and that the Atlantic sector of the Southern Ocean from 0° to 90°W, encompassing the Scotia Sea, contains 70% of the total Southern Ocean krill stock (Atkinson et al., 2004, Atkinson et al., 2008).