Introduction
Classifying Antarctic and Southern Ocean flora and fauna into biogeographic regions has been a discussion point for decades (e.g. see Dell Reference Dell1972 and references therein). More recently, the Convention on the Conservation of Antarctic Marine Living Resources (CCAMLR), its scientific committee (SC-CAMLR) and the Committee for Environmental Protection (CEP) have begun developing bioregionalizations of the Southern Ocean. Bioregionalization is essentially a tool for biodiversity or conservation management that involves classifying Southern Ocean marine life into distinct biogeographic regions. The resultant maps are intended to be used to develop an ecologically representative system of marine protected areas (Grant et al. Reference Grant, Constable, Raymond and Doust2006). The question for managers is how well do these maps, usually based on easy to measure or model physico-chemical surrogate variables, capture the diversity and functioning of the seafloor communities we are trying to protect? The bioregionalization process is particularly difficult in subtidal marine systems (where direct observations of life are difficult), and even more so in the Antarctic, due to its remoteness, ice conditions and difficulty of access. Use of biological data in these classifications is generally impractical due to scarcity or poor geographical coverage of data (except perhaps in smaller subregions that have been better studied; e.g. McMurdo Sound, Terra Nova Bay, Prince Edward Islands).
The first Southern Ocean bioregionalization was based on a combination of datasets of mostly physical variables (e.g. bathymetry, sea surface temperature, nutrient concentrations, proportion of year < 15% sea ice concentration, satellite observed sea surface chlorophyll; Grant et al. Reference Grant, Constable, Raymond and Doust2006). A later separation into pelagic and benthic bioregionalizations has allowed use of variables that are more relevant to these different (but interlinked) systems (e.g. Grant et al. unpublished). While considerable advances have been and are being made in this process (e.g. Sharp unpublished), their fitness-for-purpose will ultimately be judged on how well the bioregionalization captures the relationships between the biological communities under consideration and their environment. This assessment is urgently needed in order to ensure the ecological relevance of variables used to generate the classifications. It is also particularly relevant for Ross Sea marine benthic ecosystems, for which such data are scarce.
Given the spatially discrete nature of benthic sampling in the Ross Sea, and the natural variability in community composition, an ability to clearly link communities and environmental features such as depth, seafloor sediment type, or latitude, would ensure production of meaningful bioregionalizations. For example, in a survey of megabenthos at 270–1173 m deep sites, Barry et al. (Reference Barry, Grebmeier, Smith and Dunbar2003) noted lower densities with increasing depth, a higher abundance of suspension feeding taxa in shallow waters and deposit-feeders at the deepest sites. Natural gradients in environmental conditions and productivity along the latitudinal range of the Ross Sea provide a useful framework in which to identify key physical variables and primary production pathways driving site-specific patterns in benthic community structure and function. With the exception of macroalgae (Miller & Pearse Reference Miller and Pearse1991, Wiencke & Clayton Reference Wiencke and Clayton2002), latitude per se in the Ross Sea is not generally a good descriptor of species diversity or community composition (e.g. Cummings et al. Reference Cummings, Thrush, Norkko, Andrew, Hewitt, Funnell and Schwarz2006, De Domenico et al. Reference De Domenico, Chiantore, Buongiovanni, Ferranti, Ghione, Thrush, Cummings, Hewitt, Kroeger and Cattaneo-Vietti2006, Schiaparelli et al. Reference Schiaparelli, Lörz and Cattaneo-Vietti2006), mostly due to the fact that the environmental variables that structure these communities are not monotonically related to latitude. For example, iceberg disturbance and very strong coastal currents are dominant factors north of Terra Nova Bay, while south of Terra Nova Bay the presence of fast ice and proximity to primary production originating in polynyas becomes increasingly important (Thrush et al. Reference Thrush, Dayton, Cattaneo-Vietti, Chiantore, Cummings, Andrew, Hawes, Kim, Kvitek and Schwarz2006a, Norkko et al. Reference Norkko, Thrush, Cummings, Gibbs, Andrew, Norkko and Schwarz2007). However, this picture is complicated by the fact that iceberg disturbance does occur in the south (e.g. McMurdo Sound), and there can be as much as 1–2 months difference in duration of open water in summer months within some regions (e.g. Terra Nova Bay). We also have little understanding of how benthic community structure and function are moderated by localized environmental or ecological factors.
Here we describe the macrofaunal assemblage composition of coastal Cape Adare, Cape Hallett, Cape Russell and Coulman Island from 100–500 m deep, and assess the extent to which their similarities and differences are related to measured differences in seafloor sediment characteristics and/or depth and location along the coast. Specifically, we test the hypothesis that there are strong monotonic relationships between benthic communities and depth, sediment type and the quality and quantity of food resources derived from primary production.
Methods
Sampling
Sampling was conducted off the Victoria Land coastline, at four locations spanning three degrees of latitude: Cape Adare (one transect), Cape Hallett (two transects), Coulman Island (one transect) and Cape Russell (one transect) (Fig. 1, Table I). One of the Cape Hallett transects was located close to the cape itself (hereafter ‘Cape Hallett Inside’), with the other further offshore (hereafter ‘Cape Hallett Outside’). At each location, five stations (stations 1 to 5) were arranged along a depth gradient, at nominal depths of 500, 400, 300, 200 and 100 m, respectively. At some locations, ice conditions were such that fewer than five stations could be sampled (Table I).
Fig. 1 Map of the north-western Ross Sea showing the locations sampled along the Victoria Land coast (left hand panel), and positions of sampling stations at each location (right hand panels). CH = Cape Hallett. Blue areas indicate large ice features.
Table I GPS location of grab samples collected from each location and station. Stations 1, 2, 3, 4 and 5 correspond to targeted depths of 500, 400, 300, 200 and 100 m, respectively; the actual depths are also given.
CA = Cape Adare, CH Out = Cape Hallett Outside, CH In = Cape Hallett Inside, CI = Coulman Island, CR = Cape Russell.
At each station, three Van Veen grab samples (60 litre volume, 0.2 m2 surface area) were collected. Where possible, three grabs were taken at each station, but in some cases only two grabs of sufficient quality were able to be collected, despite repeated attempts, due to the cobbled nature of the bottom. Grabs were subsampled by collecting a core (7 cm diameter) to quantify macrofaunal community composition, and surficial sediment scrapes to characterize sediments. Thus, in our study, one individual core-1 equates to 260 individuals m-2, assuming the spatial scale of heterogeneity from cores to m2 are consistent. Core samples were sieved (500 μm mesh) and preserved in 70% isopropyl alcohol. Sediment scrapes were homogenized prior to being subsampled for chlorophyll a (chl a), particle size and organic content analysis. The sediment samples were stored frozen until they could be analysed.
Sample processing and analysis
Macrofauna core samples were sorted and identified to the lowest taxonomic level practical. Sediments for particle size analysis were digested in 6% hydrogen peroxide to remove organic matter. A Galai particle analyser (Galai Cis – 100; Galai Productions Ltd., Midgal Haemek, Israel) was then used to determine percentage volumes for the gravel/shell hash, coarse, medium and fine sand, silt and clay fractions. The organic matter content of the sediment was measured as loss on ignition (LOI) by drying the sediment at 60°C to constant weight, followed by combustion at 400°C for 5.5 h. Chlorophyll a was extracted from freeze dried sediments by boiling in 90% ethanol. The extract was measured spectrophotometrically, and an acidification step was included to separate degradation products (phaeophytin) from chl a (Sartory Reference Sartory1982).
Numerical analysis
Variations in species composition and relative abundance of benthic fauna and flora within and between sites were determined using a combination of univariate (McCullagh & Nelder Reference McCullagh and Nelder1989) and multivariate analytical procedures (Legendre & Legendre Reference Legendre and Legendre1998, Warwick & Clarke Reference Warwick and Clarke2001). Total number of species, total number of individuals, Shannon-Wiener diversity (H'), species evenness (Pielou’s J) and species richness (Margalef) were calculated using PRIMER (Clarke & Gorely Reference Clarke and Gorely2001). Differences in these parameters between stations and locations were assessed with ANOVA (using the GLM procedure within SAS). Spearmans correlation coefficients were used to identify correlations between depth and a number of variables (i.e. number of individuals, number of taxa, feeding modes) using the CORR procedure within SAS (SAS Institute 1999). Relationships between phaeophytin concentrations and both number of individuals and number of taxa were similarly investigated. Similarities and differences in community composition at the different locations and stations were assessed using non-metric multidimensional scaling ordinations (MDS) of the untransformed and presence/absence transformed data using PRIMER. Canonical correspondence analysis (CCA, ter Braak Reference ter Braak1986, Reference ter Braak1987), using untransformed community data, was then used to identify the contribution of measured environmental factors (e.g. sediment characteristics and depth) to the observed patterns in community composition.
Results
Seafloor sediment characteristics
Sediment grain size composition differed between locations: Cape Hallett Inside and the two southernmost locations (Coulman Island and Cape Russell) had a high proportion of fine sediments, while Cape Adare and Cape Hallett Outside were predominantly coarse sand and gravel (Table II). The Coulman Island and Cape Russell sediments were the most heterogeneous (Table II). There was no consistent pattern in sediment grain size distribution with station depth across locations.
Table II Sediment grain size and organic content (measured as loss on ignition) at the five locations. Data presented are mean % (± standard error). Location abbreviations as for Table I.
The sediment organic content was < 3.5% at all stations and locations. Levels were most variable between stations at Cape Adare (average 0.81–3.43%), and most similar at the Cape Hallett Outside stations (average 2.08–2.69%) (Table II). There was no relationship with latitude or depth (Table II).
There was no gradient in sediment chl a or phaeophytin concentration with latitude or depth, although both were highest at the southernmost location, Cape Russell (Fig. 2). Not surprisingly, deep sites generally had very low levels of chl a (i.e. < 0.5 μg g-1 sediment; Fig. 2a). The exception was Cape Russell, where seafloor sediment at 200 and 300 m depth contained on average 1.5–3.3 μg chl a g-1 sediment (Fig. 2a). The amount of phaeophytin (a chl a degradation product) was comparatively high (i.e. 0.1–15.7 μg g-1 sediment, recorded at Cape Adare (500 m) and Cape Russell (200 m), respectively; Fig. 2b), with sediments at all stations having more phaeophytin than chl a (i.e. by 2 to > 10 times).
Fig. 2 Levels of a. chl a (mean ± standard error), and b. phaeophytin present in seafloor sediments at each station and location in February 2004. Location abbreviations as for Table I.
Macrofauna
At Cape Adare the macrobenthic assemblage was comprised mostly of bivalves and crustaceans (amphipods, isopods and ostracods), while at Cape Hallett Outside polychaetes were amongst the dominant taxa at all but the shallowest station (100 m, station 5) (Table III). In comparison, the dominant taxa at Cape Russell and Coulman Island were polychaetes, nematodes and oligochaetes. The most common taxa at the Cape Hallett Inside assemblages were a bivalve, crustaceans and polychaetes. The bivalve, Genaxinus debilis Thiele, 1912, the most abundant species at stations 2 and 3, did not feature amongst the dominant taxa at any other location (Table III).
Table IIIa Dominant macrofaunal taxa (number of individuals core-1, mean ± standard error) at each location and station. Location abbreviations as for Table I.
Table IIIb Feeding mode of the macrofauna found in this study. d = deposit feeder, det = detritus feeder, g = grazer, s = suspension feeder, p = predator, sc = scavenger, * could be either, but specific information not available for this species, misc = large group containing a variety of feeding modes.
Average numbers of macrofaunal taxa collected in each core ranged from 3–21 across all locations and stations (Table IV). The number of taxa was most variable between stations at Cape Adare (4.0–20.5 taxa core-1), and most similar between Cape Hallett Outside stations (2.5–8.0 taxa core-1). The Cape Hallett Outside and Coulman Island stations had the lowest diversity and the fewest number of individuals of all locations (< 11.3 individuals core-1; Table IV). The highest numbers of individuals and taxa were found at Cape Adare station 4 (69.0 individuals, 20.5 taxa core-1).
Table IV Diversity at the five locations. Data presented are mean ± standard error core-1. n = number of grab samples on which statistics are based. SW = Shannon-Wiener diversity index. Location abbreviations as for Table I.
Across all locations, the relationship between number of individuals and/or taxa and depth was weak (Spearmans r = -0.4233, P = 0.0904 for number of individuals; Spearmans r = -0.4140, P = 0.0985 for number of taxa). However, strong differences were noted within locations. For example, at Cape Adare, significantly more individuals and taxa were found at station 4 (200 m deep) than at stations 1, 2 and 5 (500, 400 and 100 m deep, respectively; number of individuals: P = 0.0458, number of species: P = 0.0508). Station 4 also had more individuals than station 3 (300 m deep; P = 0.0458). At Cape Hallett Outside, significantly more individuals were found at the shallowest station (100 m, station 5) than at the 500–300 m deep stations (stations 1, 2 and 3; P = 0.0542). At Cape Hallett inside (where only the 200–400 m deep stations were sampled) more individuals were found at station 4 (200 m) than at station 3 (300 m) (P = 0.0314). There were no significant differences noted for the number of taxa found at stations within either the Cape Hallett Outside or Cape Hallett Inside locations, and no differences in numbers of individuals or taxa at Coulman Island or Cape Russell (P > 0.05).
Interestingly, the pattern in number of taxa and number of individuals at the Cape Adare, Cape Hallett Inside and Cape Hallett Outside stations was very similar to the pattern noted in the sediment phaeophytin levels at these stations/locations (compare Figs 3 & 2b). There was a significant correlation between both number of individuals and number of taxa and phaeophytin concentration of the sediments across all locations (Spearmans r = 0.6564, P = 0.0042 for number of individuals; Spearmans r = 0.5565, P = 0.0203 for number of taxa). This correlation is being driven by the patterns at Cape Adare and Cape Hallett Outside in particular, where there were highly significant and very strong correlations between phaeophytin concentrations and numbers of individuals and taxa (P < 0.0001; r = 1.00 in all cases).
Fig. 3 Multidimensional scaling analysis ordination plot showing the similarities in macrofaunal assemblages within and between locations. Location abbreviations as for Table I.
Evenness was high at all locations/stations (i.e. 0.9–1.0; Table IV), indicating a lack of numerical dominance (a value of 1.0 indicates all taxa are equally abundant). The Shannon-Wiener diversity index was highest at sites with high numbers of taxa/individuals: Cape Adare stations 3 and 4, Cape Hallett Inside stations 2 and 4, and Cape Russell station 4 all had levels ≥ 2. This index is affected by rare taxa, and increases both with increasing numbers of taxa and a more even distribution of individuals amongst taxa.
Examination of the distribution of functional groups (i.e. feeding modes) of the common macrofaunal taxa revealed no clear relationship with depth, either across all locations, or within individual locations (P > 0.05 in all cases). In addition, closer examination of the composition of the Coulman Island and Cape Russell assemblages did not help elucidate potential reasons for the lack of correlation between number of species/individuals and phaeophytin levels at these locations noted above (e.g. absence of a particular functional group). Predator/scavengers and deposit feeders were common at all locations, while suspension feeders were more common at Cape Hallett Outside (Table III).
The relationship between macrofaunal assemblage composition at the different stations and locations are illustrated in two dimensions in Fig. 3. There is little similarity in macrofaunal assemblage composition across locations at stations of similar depth (Fig. 3a). Generally, the assemblages varied considerably within locations and there was overlap of the different locations. Cape Adare is the most distinct within the ordination space, indicating that the macrofaunal assemblages at the five Cape Adare stations are the most similar to one another. Cape Hallett Outside exhibits the highest variability in macrofaunal assemblage composition, while Cape Hallett Inside stations are similar to those from Coulman Island. In order to isolate the effect of changes in the densities of species from changes in species composition, an ordination was also conducted on the presence/absence transformed data (Fig. 3b). While this shows a similar pattern to that of the untransformed MDS, with the exception of one Cape Hallett Inside station 3 sample, there is more overlap and tighter clustering of the Cape Hallett Inside, Coulman Island and Cape Russell assemblages (Fig. 3b), indicating that these communities were comprised of similar taxa (see previous discussion of individual taxa).
Explaining variability in macrofaunal assemblage composition using environmental variables
The environmental variables measured during this study and initially included in the CCA model were sediment characteristics (grain size composition, organic content, chl a, phaeophytin, ratio of chl a to phaeophytin) and depth. Those most important in explaining the differences in assemblage composition were % fine sand and silt, the ratio of sediment chl a to phaeophytin, and depth (Fig. 4). However, the overall percentage of community variability explained is low (17.3%), indicating that the environmental factors included in the model are having only a weak influence on macrofaunal assemblage composition.
Fig. 4 Canonical correspondence analysis ordination plot showing the environmental variables important in explaining the macrofaunal assemblages at each location and station. Ratio = chl a:phaeophytin. Location abbreviations as for Table I.
Discussion
This paper has described the benthic macrofaunal assemblages at locations spanning three degrees of latitude along the coastal north-western Ross Sea shelf. Macrofaunal assemblages at Cape Adare were comprised mostly of bivalves and crustaceans, polychaetes were predominant at Cape Hallett Outside, polychaetes, nematodes and oligochaetes at Cape Russell and Coulman Island, and the Cape Hallett Inside assemblages comprised a mix of taxa found at the other locations. Average numbers of macrofaunal taxa collected in each core ranged from 3–21 across all locations and stations. Across all locations, the relationship between number of individuals and/or taxa and depth was weak.
To identify the use of seafloor sediment characteristics and/or depth measures in bioregionalizations, we also assessed the extent to which assemblage compositions are related to measured differences in these factors. Although our analyses identified % fine sand and silt, the ratio of sediment chl a to phaeophytin, and depth as important explanatory variables, in combination they explained only 17.3% of the between-location differences in assemblage composition. Consequently, these variables are clearly not strong determinants of macrofaunal assemblage structure. In addition, there was no clear pattern in assemblage structure with location along this coastline, indicating that latitude per se is not a useful measure of community variability and change. For reasons outlined in the Introduction, and similar to studies documented in Table V, the latter finding is not surprising. This lack of a clear link between these variables and assemblage composition highlights the need for caution when creating bioregionalizations based on such variables. This adds impetus to the need to gather data more appropriate for capturing seafloor diversity.
Table V Summary of Ross Sea studies of distribution of benthic assemblages and important explanatory environmental variables identified. ROAVERRS: 270–1173 m deep, stations arrayed from Cape Hallett (CH) south to southern McMurdo Sound, and from the Victoria Land coast to c. 165°E (Barry et al. Reference Barry, Grebmeier, Smith and Dunbar2003). RV Italica: Cape Adare (CA), Cape Hallett, Coulman Island (CI), Cape Russell (CR), 100–500 m (this paper). RV Tangaroa: Cape Adare to Cape Hallett, plus the Balleny Islands, targeting three depth strata (50–750 m; e.g. Kröger & Rowden Reference Kröger and Rowden2008).
Barry et al. (Reference Barry, Grebmeier, Smith and Dunbar2003) found that the Ross Sea megabenthos (sampled by video) was dominated by suspension feeders (87%), with a smaller proportion of deposit feeders and predators. These authors also noted that suspension feeders were more abundant in shallow waters (360 ± 105 m, mean ± SD), deposit-feeding taxa increased at the deepest sites, and overall faunal densities decreased with depth. Arntz (Reference Arntz1999) also noted that suspension feeders dominated the Ross Sea benthos. This depth-related pattern was not apparent for the macrofauna sampled by grab in our study (although the depths covered were not as wide ranging as those of Barry et al.), with a variety of functional groups found at all locations (Table III).
In coastal waters, one consistency that seems to be emerging is the high degree of heterogeneity that benthic communities exhibit (e.g. Hewitt et al. Reference Hewitt, Thrush, Legendre, Funnell, Ellis and Morrison2004, Thrush et al. Reference Thrush, Dayton, Cattaneo-Vietti, Chiantore, Cummings, Andrew, Hawes, Kim, Kvitek and Schwarz2006a). This is probably a reflection of local disturbance regimes (see below discussion of icebergs), larval supply, hydrodynamics, and the importance of ecosystem engineers along with changes in the relative proportions of primary food sources. We measured sediment organic content and pigment concentrations as an indication of primary food sources (i.e. sinking of material through the water column and in situ production). Sediment organic content was low, and showed no relationship with latitude or depth. Values of 0.5 to 1.5% C were noted for 270–1173 m deep stations in the south-western Ross Sea by Barry et al. (Reference Barry, Grebmeier, Smith and Dunbar2003) and, although these values are not directly comparable to our organic content (which was measured as loss on ignition), they are also considered low. Sediment chl a and phaeophytin concentrations were variable and highest at the southernmost location, Cape Russell. All sediments contained more phaeophytin than chl a, indicating a stronger benthic-pelagic coupling in this area (Povero et al. Reference Povero, Castellano, Ruggieri, Monticelli, Saggiomo, Chiantore, Guidetti and Cattaneo-Vietti2006). Pusceddu et al. (Reference Pusceddu, Dell’Anno and Fabiano2000) sampled sediment pigments from 12–127 m in Adélie Cove, Terra Nova Bay, and noted an increase with depth to 76 m and, as in this study, a predominance of phaeophytin amongst the sedimentary pigments. Interestingly, we found a significant correlation between both number of individuals and number of taxa and sediment phaeophytin concentration across all locations. This pattern may be due to the fact that both phaeophytin and macrofauna are time-integrated measures, and there is thus a better matching of macrofauna with phaeophytin than with chl a, and suggests that the distribution of the benthos in these locations reflects their response to time-averaged periods of productivity.
A number of factors not measured in this study have probably influenced the structure and function of assemblages and habitats. These include: strong near shore currents (particularly noticeable at Cape Hallett and Cape Adare) and bottom topography, rapid changes in sea ice conditions in the summer (e.g. over several hours in some locations), and iceberg disturbance. The latter in particular probably contributes to the variability between stations at some of our sampling locations, as local disturbance by ice and the subsequent recolonization creates a patchy pattern on the seafloor with epifaunal organisms exhibiting different life forms dominating in different stages of succession (Gutt Reference Gutt2001, Teixido et al. Reference Teixido, Garrabou, Gutt and Arntz2004). Studies have noted differences in macrofaunal biomass and taxa richness (Lenihan & Oliver Reference Lenihan and Oliver1995, Gerdes et al. Reference Gerdes, Hilbig and Montiel2003) and feeding types (Conlan et al. Reference Conlan, Lenihan, Kvitek and Oliver1998) in relation to the degree of iceberg disturbance. While there was no clear pattern in distribution of functional groups in our study that would indicate that one station was more or less impacted by icebergs, video and multibeam imagery collected on the RV Italica and Tangaroa voyages demonstrate that iceberg scouring is a significant source of topographic variation to the seafloor in our sampling regions. Dr R. Kvitek estimated iceberg disturbance had occurred in at least 40% of the seafloor off Cape Hallett (Thrush et al. Reference Thrush, Dayton, Cattaneo-Vietti, Chiantore, Cummings, Andrew, Hawes, Kim, Kvitek and Schwarz2006a). Kröger & Rowden (Reference Kröger and Rowden2008) reported considerably lower disturbance in the Cape Adare/Cape Hallett regions, and differences in disturbance rates with depth. Information on the distribution and track of icebergs and the spatial and temporal scales of disturbance in the Ross Sea, in tandem with observations of seafloor disturbance/community recovery dynamics, is needed to help interpret patterns in seafloor diversity. Icebergs can also affect phytoplankton production and distribution (e.g. Schwarz & Schodlok Reference Schwarz and Schodlok2009) and thus benthic-pelagic coupling.
Good understanding of the ecological processes that underpin biodiversity in the Ross Sea’s coastal regions (continental and islands) is vital in realistically assessing the threats to this environment (e.g. Agardy Reference Agardy2005). While there has been a long and productive history of taxonomic research on the benthos of the Ross Sea (e.g. Borchgrevink Reference Borchgrevink1901, Bullivant Reference Bullivant1967a, Reference Bullivant1967b, Dell Reference Dell1972), our understanding of the structure and function of these communities remains poor. Benthic communities have not been sampled in many areas and our understanding of functional relationships and their spatial and temporal variability is even more constrained. This limited knowledge is recognized as a major barrier to developing and testing bioregionalization (e.g. Lombard et al. Reference Lombard, Reyers, Schonegevel, Cooper, Smith-Adao, Nel, Froneman, Ansorge, Bester, Tosh, Strauss, Akkers, Gon, Leslie and Chown2007) and more broadly in predicting the consequences of change (e.g. removal of top predators, climate variability, tourism).
Importantly, we show that latitude and depth are not good predictors of community composition for north-western Ross Sea shelf macrofaunal communities. This provides insight into the role of changes in ecosystem structure and function directly associated with latitude along the Victoria Land coast (a Key Question within the Latitudinal Gradient Project). Several studies of Ross Sea fauna have looked for consistencies in assemblage composition with factors such as depth or latitude, but generally have not found evidence of simple patterns; rather their distributions are explained by sediment type, hydrodynamic conditions and iceberg disturbance (although these studies seldom measured the latter) (see summary in Table V). These findings emphasize the complexity of these marine systems, and the role of other environmental and biotic factors in governing abundance and distribution patterns. The low correlation found in this study between the infaunal community and the (largely physical) habitat characteristics measured highlights the need for caution when choosing marine protected areas (MPAs) based only on such variables.
Most researchers involved in marine bioregionalization development acknowledge the limited availability of benthic biological data (Sharp et al. Reference Sharp, Parker, Pinkerton, Breen, Cummings, Dunn, Grant, Hanchet, Keys, Lockhart, Lyver, O’Driscoll, Williams and Wilson2010). This is unfortunate as seafloor habitats represent a major component of the biodiversity we are trying to protect. As technological improvements allow ecologists to better sample seafloor habitats and communities, there is growing recognition of the essence of spatial variability (Levin & Dayton Reference Levin and Dayton2009). Often this important heterogeneity is generated by the presence and actions of organisms directly modifying habitats or generating important bio-physical or bio-geochemical feedback processes (e.g. Chiantore et al. Reference Chiantore, Cattaneo-Vietti, Albertelli, Misic and Fabiano1998, Van de Koppel et al. Reference Van de Koppel, Rietkerk, Dankers and Herman2005, Thrush et al. Reference Thrush, Gray, Hewitt and Ugland2006b, Van Nes et al. Reference Van Nes, Amaro, Scheffer and Duineveld2007). Given their longevity and the potential of many Antarctic benthic species to create three-dimensional habitats we may except such features to be especially important in the Antarctic (Arntz et al. Reference Arntz, Gutt and Klages1997, Thrush et al. Reference Thrush, Dayton, Cattaneo-Vietti, Chiantore, Cummings, Andrew, Hawes, Kim, Kvitek and Schwarz2006a). The implementation of scale-dependent hierarchal approaches to bioregionalization recognizes that variability within scales can be functionally important (Legendre Reference Legendre1993), but cross-scale interactions and the maintenance of ecological connectivity across landscapes is also important for conservation (e.g. Zajac et al. Reference Zajac, Lewis, Poppe, Twichell, Vozarik and DiGiacomo-Cohen2003). While the challenge of improving bioregionalization processes via new data must be met we must also consider how to better incorporate dynamic process, such as variation in sea ice extent, thickness and snow cover. Also important is the need to incorporate biogenic habitat complexity into protection strategies and to define and understand the processes that contribute to seafloor biodiversity.
Acknowledgements
We thank the expedition leader Roberto Meloni, Programma Nazionale Di Recherché in Antartide (PNRA) staff on board RV Italica (especially Nicola la Notte), and the crew for being always helpful and cheerful, and for midnight pasta. We also thank our colleagues on the cruise for stimulating discussions and company. Greig Funnell, Jane Halliday and Ron Ovenden helped with sample processing. Geoff Read is thanked for polychaete identifications, Arne Pallentin for Fig. 1, and two anonymous reviewers for comments on the manuscript. Finally, thanks to Antarctica New Zealand and PNRA for their excellent logistical support, and the Ministry of Fisheries BioRoss and NIWA for funding.
