a1 Biozentrum Grindel and Zoological Museum, University of Hamburg, Martin-Luther-King-Platz 3, 20146 Hamburg, Germany
a2 British Antarctic Survey, NERC, High Cross, Madingley Road, Cambridge CB3 OET, UK
The remote South Sandwich arc is an archipelago of small volcanic islands and seamounts entirely surrounded by deep water and about 600 km away from the closest island, South Georgia. As some of the youngest islands (< 5 m.y.) in the Southern Ocean they are ideal for studying colonization processes of the seabed by benthic fauna, but are rarely investigated because of remoteness and extreme weather. The current study attempted to quantify the richness and abundance of the epibenthic macrofauna around the Southern Thule group by taking five epibenthic sledge samples along a depth transect including three shelf (one at 300 m and two at 500 m) and two slope stations (1000 and 1500 m). Our aim was to investigate higher taxon richness and community composition in an isolated Antarctic locality, since recent volcanic eruptions between 1964 and 1997. We examined patterns across all epibenthic macrofauna at phylum and class levels, and investigated trends in some model groups of crustaceans to order and family level. We found that abundance was highest in the shallowest sample and decreased with depth. Shelf samples (300 and 500 m) were dominated by molluscs and malacostracans while at the deeper stations (1000 and 1500 m) nematodes were the most abundant taxon. Surprisingly, the shallow shelf was dominated by animals with restricted dispersal abilities, such as direct developing brooders (malacostracans) or those with lecithotrophic larvae (bivalves of the genus Yoldiella, most bryozoan species). Despite Southern Thule's geological youth, recent eruptions, and its remoteness the shallow shelf was rich in higher taxa (phyla/classes) as well as orders and families of our model groups. Future work at higher taxonomic resolution (species level) should greatly increase understanding of how life has reached and established on these young and highly disturbed seabeds.
(Received July 19 2007)
(Accepted January 02 2008)
This publication is dedicated to the memory of Dr Helen R. Wilcock for whom to visit Antarctica was a lifelong dream.
List of Figures and Tables
Fig. 1. a. The position of the South Sandwich Islands in the Southern Ocean, and b. detail of South Sandwich archipelago.
Fig. 2. Bathymetry around Southern Thule archipelago, modified from Larter et al. (1998). ESR = East Scotia Ridge Segment, ST = South Sandwich forearc near Southern Thule, SST = South Sandwich Trench.
Table I. Epibenthic sledge stations taken during the BIOPEARL 1 cruise of RRS James Clark Ross off Southern Thule, South Sandwich Islands, Scotia Sea.
Table II. Abundance of phyla and classes, standardized value (to 1000 m) in brackets, percentage in italics.
Table III. Abundance of Crustacea to order level (Amphipoda, Tanaidacea, Isopoda, Cumacea, Mysidacea, Leptostraca). Standardized value (to 1000 m) in brackets, percentage in italics.
Table IV. Abundance of Isopoda at family level. Standardized value (to 1000 m) in brackets, percentage in italics.
Table V. Bryozoan species sampled during the BIOPEARL 1 cruise from the upper slope of Southern Thule.
Table VI. Depth range and new records (n) of South Sandwich Islands isopod families. Data from the BIOPEARL (BP), ANDEEP II (AII, Brandt, unpublished data) and LAMPOS (L, Arntz & Brey 2003) cruises and Brandt (B, 1991).
The Southern Ocean (SO) surrounding Antarctica has many remote islands and archipelagos of differing size, age and isolation. As such they form a natural laboratory for studies of colonization, biogeography, ecology and conservation. Terrestrial studies at locations such as Marion Island and South Georgia (Chown et al. 1998) have shown relatively low levels of indigenous biodiversity and high invasions by new species. Although the terrestrial fauna of many islands has been well described, the shores and immediate subtidal are poorly known (see Blankley & Grindley 1985) and the seabed, where most of the regional biodiversity occurs, has not been adequately studied. In general, the shelf around Antarctica is disproportionately rich for its area (Clarke & Johnston 2003) but is probably still in the process of recolonization since the last glacial maximum (Thatje et al. 2005). Small, remote islands provide a hard target for colonization for shelf fauna, especially when they are young and surrounded by abyssal depths (Arntz et al. 2006). The young (< 5 m.y.) (Baker 1990) South Sandwich Islands are of particular interest in that they are the only land at 55°–59°S in the Atlantic sector of the SO and arguably the only link between the sub-Antarctic and Antarctic islands. Within their short lifespan there have been many glacial cycles (Holdgate & Baker 1979) and thus challenges to colonization. It seems likely that seasonal sea ice reached and surrounded the South Sandwich Islands for much of the last glacial maximum (Gersonde et al. 2005), but it is unknown whether grounded ice advanced across their slopes down several hundred metre water depth at this time and erased the faunas.
In the last decade there has been much discussion over whether grounded ice masses of the last glacial maximum (more than 15 000 years ago) or previous glacial maxima completely bulldozed life off the Antarctic continental shelf (Anderson et al. 2002). There are a number of ways in which fauna could survive. For example, there may have been a degree of diachronism, i.e. grounded ice and ice shelves expand and retreat at different times during a glacial period, so not all shelf areas are covered at any one time (see Thatje et al. 2005). Brandt (1991) suggested that the South Shetland Islands and South Orkney Islands might have been refuges. Using a phylogenetic analysis, Sieg (1988) postulated that the tanaid shelf fauna was nearly wiped out during the Miocene and the recent fauna has recolonized high latitude southern shelves from the slope or deeper. Antarctic shelf fossil evidence suggests that some ostracod species inhabiting the shelf are also descended from deep-sea species but that most have derived from former littoral taxa (Hartmann 1990). Conversely, the deep-sea isopod families Serolidae, Antarcturidae and Acanthaspidiidae all seem to have evolved on the Antarctic shelves and migrated downslope (Brandt 1991). The current faunal assemblages on shelf, slope and in the deep sea apparently have been formed by a variety of routes and directions, probably involving many migrations, some coincident with major changes in glaciation. As a result, many Antarctic species show a remarkably high eurybathy (Brey et al. 1996), which may have aided survival advances during glacial periods. It is hypothesized that the recolonization from the past glacial period is still in progress and may continue for tens of thousands of years (Gutt 2006).
Like Deception Island (South Shetland Islands) the South Sandwich arc is seismically very active, and so colonization cycles must have also occurred on shorter, ecological timescales. Their shelf fauna would have been reduced by a series of volcanic eruptions such as 1962 (Protector Shoal seamount, Gass et al. 1963), between 1964 and 1997 (Southern Thule group, Leat et al. 2003), and most recently in 2001 (Patrick et al. 2005) and 2005 (at Montague Island, authors' personal observations). Within the South Sandwich forearc the only submarine caldera producing “significant pyroclastic deposits” is Southern Thule (Smellie et al. 1998, Lachlan-Cope et al. 2001). In the early 1960s pumice from eruptions there (Protector Shoal) reached New Zealand (Coombs & Landis 1966) and even Hawaii (Jokiel & Cox 2003). One of the most significant events in the geological history of Southern Thule was the formation of the Douglas Strait caldera, which probably took place within the last few thousands of years (J.L. Smellie personal communication 2007). This event probably wiped out major parts of the benthic fauna around Southern Thule and also drastically reduced the benthic animals on the slope due to subsequent mass burial (cf. Thatje et al. 2005). Following recent eruptions (1967–1970), recolonization at Deception Island has been well studied (e.g. Gallardo et al. 1977, Lovell & Trego 2003). In contrast to Southern Thule, though, Deception Island has old and multiple sources of colonization close by, e.g. the South Shetland Islands and north-western Antarctic Peninsula. Recolonization of shallow and shelf habitats in the SO is fairly well studied at a few localities (reviewed in Barnes & Conlan 2007) but poorly for remote, slope and abyssal areas. Certain attributes, such as having planktotrophic larvae, highly mobile adults, or transport by drifting material must greatly improve colonization prospects for small, remote or young areas.
On ecological and evolutionary time scales, Southern Thule and its surrounding seabed represent a highly dynamic and complex area (in terms of seismic and tectonic processes, debris flow deposition, sedimentation rates, primary production, see Vanhove et al. 2004, Patrick et al. 2005) and therefore an ideal locality to study epibenthos in the process of recolonization. We took three epibenthic samples at two shelf (300 and 500 m) and two samples at two slope (1000 and 1500 m) depths around Southern Thule. We hypothesized that levels of faunal abundance would be highest on the shallow slope but this shallowest depth would be poorer in higher taxa (due to still being in the process of recolonization). We expected highly dispersive taxa to be most represented on Southern Thule's slope and selected one model group, the Malacostraca, to investigate at lower taxonomic (order, family) levels.
During the BIOPEARL 1 (JR 144) cruise of the RRS James Clark Ross epibenthic samples were taken around the Southern Thule group (South Sandwich Islands, SO) (Fig. 1a & b) in the summer of 2005/2006. Samples of epibenthic macrofaunal richness and abundance were taken at five stations and four distinct depth horizons of the south-eastern entrance of the Southern Thule caldera comprising three shallow slope (~300 and 500 m) and two mid slope (1000 and 1500 m) stations (Table I).
Epibenthic sledge stations taken during the BIOPEARL 1 cruise of RRS James Clark Ross off Southern Thule, South Sandwich Islands, Scotia Sea.
Southern Thule is the southernmost tip of the South Sandwich Islands consisting of Thule, Bellingshausen and Cook islands. Bellingshausen and Thule represent stratovolcanoes with steep flanks (Smellie et al. 1998, Leat et al. 2003), which are still active. Its volcanic evolution can be regarded as typical for the entire South Sandwich arc (Smellie et al. 1998), whereas Bellingshausen Island and the Douglas Strait Caldera, a submarine caldera of 600 m depth (Smellie et al. 1998) between Thule and Cook islands may be the youngest features within the forearc (< 1 m.y., see Leat et al. 2003). New geological studies suggest that parts of the Southern Thule group, such as the Douglas Strait Caldera, might have evolved even more recently (< few kyr ago, J.L. Smellie personal communication 2007) during a major volcanic eruption. Of the South Sandwich Islands investigated to date (Zavodovski, Leskov, Visokoi and Candlemas islands), sediments dominated by recent volcanism extend down to 3000 m (British Antarctic Survey (BAS) unpublished data), thus suggesting shelf and slope fauna obliteration during eruptions. Due to its steep flanks, a massive downslope transport of volcanic sediment is also suggested for Southern Thule (C.-D. Hillenbrandt personal communication 2007).
The apparatus used was a modified epibenthic sledge (EBS, Brenke 2005). Additionally, one Agassiz trawl was undertaken at each of the four depth horizons for the collection of mega- and epibenthic macrofauna (reported separately, see Griffiths et al. 2008). Because of additional ship time, a replicate EBS sample was taken at 500 m depth. The EBS is equipped with two 500 µm nets (supra- and epibenthic net sampler) of 100 cm width and 33 cm height which both end up in a cod-end (300 µm). Both samplers are closed in the water column by box doors that open when the sledge reaches the seabed due to a fixed shovel. Each EBS was trawled along the seabed for 10 minutes at 1 knot. The trawling distance was calculated after equation (4) in Brenke (2005). Thereby the trawling distance varied between 487 m (at the shallowest depth) and 1002 m (at the deepest depth) (Table I).
On board, samples were immediately fixed in pre-cooled 96% ethanol and kept for at least two days in a -20°C freezer. Samples were partly sorted on board and this work was continued in the laboratory of the British Antarctic Survey, Cambridge (UK) and the Zoological Museum, University of Hamburg (Germany). Unfortunately it was not possible to include foraminifers or diatoms. From each sample, specimens were discriminated to phylum and then class level (following Barnes 1998). Samples of model groups, Malacostraca and Bryozoa, were further identified to high resolution (i.e. order- (Malacostraca), family- (Isopoda), and species level (Bryozoa)). Epi- and supranet samplers were treated as one sample following Brenke (2005). Supranet samples were used for comparison with other studies only, which also reported just supranet results (i.e. Linse et al. 2002, Lörz & Brandt 2003). Taxon abundances were standardized, because of differing trawl distances (Table I), to 1000 m for comparison following similar data treatments (e.g. Brenke 2005, Brandt et al. 2007, Tables II–IV). The abundance levels between stations were statistically tested using an Analysis of Variance (ANOVA).
Abundance of phyla and classes, standardized value (to 1000 m) in brackets, percentage in italics.
Abundance and richness levels differed considerably between Southern Thule samples. Abundance of epibenthic macrofauna decreased significantly (ANOVA associated with regression, F1 = 12.9, P = 0.037) with increasing depth, by more than an order of magnitude from the deep to the shallow slope depths (Table II). The differential abundance in the two 500 m replicates did, however, show that there can also be strong within-depth variability. The most numerous taxa were the nematodes, which dominated mid slope samples (500 to 1500 m), and molluscs, which dominated upper slope samples (Table II). Crustaceans and annelids were also highly abundant. These four phyla were the only ones represented at every depth. The abundance of molluscs was mainly bivalves, annelids were mainly polychaetes and crustaceans were mainly malacostracans and copepods. Thus, the abundance of epibenthic macrofauna increased across the deep slope to shallow slope, broadly changed in dominance but was mostly represented by similar phyla and classes. Although lack of replication makes trend analysis difficult, the data suggest that several phyla and classes show clines in abundance levels with depth. Nematodes proportionally increased with decreasing depth, whereas bivalves, gastropods, polychaetes, copepods and malacostracans all decreased in abundance from the shallowest to the deepest station (Table II). The abundance of nematodes on the shallow shelf was remarkably low (one order of magnitude lower than at the deeper stations, see Table II).
In comparison to patterns of abundance, higher taxon richness (phyla and classes) was also highest at the shallowest depth sampled. The number of phyla found in EBS samples increased from 1500 m to the upper slope (300 and 500 m, Table II). The two 500 m samples, unlike the high variability in abundance, both comprised nine phyla and 13 classes each (Table II). However, several phyla and classes were only found in only one of each replicate. Patterns of class richness mirrored those of phyla being poorest in the deep slope, but richest at 300 m depth (17 classes, Table II). More detailed investigation of one class, the Malacostraca, revealed some similarities and contrast with higher taxa in terms of abundance and richness.
In total, six orders of malacostracans were found within the samples and these were all most abundant on the upper slope (300 m) (Table III). The dominance of the malacostracans (17.1%) on the shallow shelf is caused by the high number of amphipods within the sample (12.1%) followed by the tanaids (2.7), isopods (1.7), cumaceans (0.4) and mysids (0.06) (Table III). At the remaining stations (500 to 1500 m depth) tanaids were the dominant malacostracan order displaying a decline in abundance with increasing depth. Isopods were present at all stations and (compared to amphipods) with a relatively low number of specimens at the shallowest station and with similar abundances at the remaining stations. Similar to the abundance pattern, richness decreases with increasing depth, but in contrast is lowest at the 1000 m station.
Abundance of Crustacea to order level (Amphipoda, Tanaidacea, Isopoda, Cumacea, Mysidacea, Leptostraca). Standardized value (to 1000 m) in brackets, percentage in italics.
Abundance and richness patterns of isopod families resembled those of higher taxa as the 10 families at the shallowest sample decreased towards 1500 m depth (Table IV). The two 500 m samples had the same number of specimens; yet, they were quite different in terms of richness (1 compared to 5 families, Table IV).
Abundance of Isopoda at family level. Standardized value (to 1000 m) in brackets, percentage in italics.
The number of taxa, which were brooders or had lecithotrophic larvae, was high in the shallowest sample. Most of the numerous bivalves belonged to the genus Yoldiella, which possesses a lecithotrophic pericalymma larva. Most of the gastropods were represented by the superfamily Rissoidea, which are all oviparous. The bryozoans present in our 300 m sample have lecithotrophic larvae, apart from Harpecia spinosissima (Calvet, 1904), which has planktotrophic larvae. All malacostracan orders are brooders and were most abundant at the shallow shelf.
Studies at very young islands, such as Southern Thule, have the potential to reveal much about colonization levels and processes in the SO. Being remote, young and entirely surrounded by deep water (on average 2300 m depth between islands, Baker 1990) makes their shelf and slope environments difficult to reach from other areas of similar depth. Generally, near-surface primary productivity is so much greater than at depth that we expected to find faunal abundances on the shelf to be much higher than those on the slope, which indeed the samples revealed. Episodic volcanic activity, including eruptions at Montague Island during the sample period (authors' personal observations), produced much iron-rich ash, which can stimulate particularly intense phytoplankton ‘blooms’ (Bay et al. 2004). The seabed around Southern Thule is characterized by many striking environmental and geographical features, which have strong influences on its current fauna. Usually, Antarctic shelf communities are highly disturbed on many different spatial and temporal scales (Barnes & Conlan 2007) but conditions around Southern Thule can be considered particularly extreme and complex. In addition to eruptions, the most severe geological factor is the mass burial of volcaniclastic detritus on the steep flanks of the Southern Thule slopes (cf. Howe et al. 2004), which are prone to gravitational instability. Frequency and magnitude of slumps and slides, debris flows and turbidity currents have probably a devastating effect on the benthic fauna on the slope (C.-D. Hillenbrand personal communication 2007). Even the limited sampling in the current study can provide important insights into abundance, richness and affinities of the fauna, as we know so little about it.
The epibenthic macrofaunal abundance levels we report here, at Southern Thule, fall within the range of those found elsewhere in the wider region of the Weddell Sea and Scotia Arc (Linse et al. 2002, Lörz & Brandt 2003). Given the severity of many environmental factors, e.g. several eruptions between 1964 and 1997 coupled with its isolation, this is perhaps surprising. Deception Island, which erupted less recently (between 1967 and 1970) and is closer to major sources of colonists (shelf fauna) than Southern Thule, still has much reduced megafaunal abundances compared with surrounding areas (Cranmer et al. 2003, but see Lovell & Trego 2003). Many environmental conditions on the slopes of active volcanoes could be considered unsuitable for organisms, due to temperature gradients, toxic sediments and sediment accumulation rates (Saiz-Salinas et al. 1998, Lovell & Trego 2003). Another young and isolated island in the region, Bouvetøya, would be ideal for comparison with our data from Southern Thule but firstly, levels of abundance are hard to estimate because of the paucity of samples taken there and secondly, studies used dissimilar equipment (Arntz et al. 2006).
The South Sandwich Islands are surrounded by abyssal and hadal depths (see Fig. 2). Shallower depths (600 m, see Smellie et al. 1998) only occur between the three islands of the Southern Thule group and between Candlemas and Vindication Island (24 m, Baker 1990). The shallowest depth of connectivity between Southern Thule and another island (only Bristol Island) is sufficiently deep, 1600 m (BAS unpublished data), reducing possibilities for along slope recolonization. Thus, the principal source of fauna for recolonization of shelf and upper slope habitats are deep or distant. The Antarctic Circumpolar Current and Weddell Sea gyre potentially carry larvae (and rafting adults) to the South Sandwich Islands and larval densities at another isolated Scotia Arc locality, Signy Island, have been argued to be high (Stanwell-Smith et al. 1999). Pumice from previous eruptions in the South Sandwich Islands has almost circumnavigated Antarctica and stranded on South Shetland Islands shores (Risso et al. 2002) indicating the potential for oceanic transport of long-lived planktotrophic larvae (Scheltema 1971). For example, such transport might explain the presence of one cheilostome bryozoan found in our samples, Harpecia spinosissima, which is one of the few to possess such larvae. The importance of abyssal and hadal fauna for recolonization is harder to assess. Peracarids, which all show direct development and, thus, must disperse as adults, were well represented at shelf depths in the samples (Table III), suggesting considerable upslope recolonization. Potential transport mechanisms are varied, including crawling, e.g. larger amphipods and isopods (such as isopod families Chaetilidae, Munnopsidae and Desmosomatidae), swimming (e.g. mysids), passively drifting (e.g. isopod families Munnidae and Paramunnidae) or travelling as parasites (Gnathiidae) (Brandt 1991) while how some sedentary taxa (e.g infaunal tanaids and cumaceans) get there is a mystery still (U. Mühlenhardt-Siegel personal communication 2007). Some tanaids can be quite resistant to increased temperature and are even able to survive anoxic conditions for a few days (J. Guerrero-Kommritz, personal communication 2007) and so perhaps a few are able to survive eruptions. Sieg's (1988) phylogenetic studies supported the idea of the current fauna being formed from a combination of species surviving in refuges, and others, which have recolonized the shelf from the depth. Species that migrated up the slope and shelf of the South Sandwich Islands are most likely to have done so from the west rather than the east as the latter is bordered by the deep South Sandwich Trench (SST, Fig. 2). It is also possible that benthic fauna can be resuspended and transported by currents. We consider that the considerable distances to the nearest upstream shelf area (>1000 km) combined with young age and small size of Southern Thule as a target make this unlikely.
Meaningful comparison of data between, and sometimes even within studies can be difficult. As well as variation with season, depth and habitat, Antarctic epibenthos is known to be highly patchy in space (e.g. Rehm et al. 2007, Kaiser et al. 2007) and can vary drastically between replicates (Gerdes et al. 1992). Despite being an excellent tool for sampling epibenthic biodiversity (Brenke 2005), samples have rarely been taken in the SO using EBS and abundances recorded in these to date have proved very patchy (Kaiser et al. 2007). In contrast, there is a wealth of information on abundance levels of shelf and slope mega- and macrobenthos from use of multicorers, boxcorers, Agassiz trawls and Rauschert dredge (e.g. Gerdes et al. 1992, Ramos 1999, Hilbig et al. 2006, Rehm et al. 2007). The type of fauna found does to an extent reflect differences in collection apparatus. This can be seen by comparison of our samples with those collected by Agassiz trawl at the same localities. Griffiths et al. (2008) found similar abundances of ophiuroids across our study depths (using Agassiz trawl) whilst we found them only on the shelf in EBS samples (Table II). Data from Vanhove et al. (2004) would also suggest that we undersampled some taxa, such as the nematodes. EBS is a semiquantitative apparatus and not ideal to measure total abundances or to collect animals as small as meiofauna. It is more appropriate for sampling epibenthic and vagile fauna rather than infauna (Hilbig 2004). Despite taxon bias, it seems that nematodes, molluscs, crustaceans and annelids really are highly abundant at Southern Thule. Our second hypothesis, that taxa with higher dispersal ability would be most represented at the shallowest station, was, therefore, not accepted. The most abundant taxa in our samples were either brooders (malacostracans), oviparous (gastropods) or possess lecithotrophic larvae (bivalves, most bryozoan species). Within such groups our findings were not striking different to those found typically in the SO, although isopods were poorly represented within the Malacostraca compared with elsewhere (see Linse et al. 2002, Lörz & Brandt 2003, Rehm et al. 2007). However, no general pattern has been established in the peracarids. In the Rehm et al. (2007) and Lörz & Brandt (2003) studies, amphipods were the most dominant taxon, followed by the isopods, tanaids and cumaceans, while in Linse et al. (2002) samples from the same depth transect (200 to 500 m) isopods were the most dominant. Rehm et al. (2007) stressed the high variability in peracarid composition between samples probably reflecting patchy distribution, different apparatus and seasonal effects. Notably, tanaids, though, were the most abundant malacostracans in our samples except in the shallowest (300 m) despite having the most restricted dispersal abilities amongst peracarids.
We also hypothesized that our geologically young and isolated locality would be low in richness, at least on the upper slope, as it would probably still be in the process of recolonization. Very few endemic species have been reported from the South Sandwich Islands to date (Brandt 1991, Brandt et al. 1999, Zelaya 2005). However, representatives of most higher taxa (phyla and classes) found elsewhere in the Scotia Arc have been found at Southern Thule (see Griffiths et al. 2008). In our model taxon, the Malacostraca, order and family levels were as well represented at Southern Thule (Tables III & IV) as at other Antarctic sites (Linse et al. 2002, Lörz & Brandt 2003). At species level, though, richness around the South Sandwich Islands appears to be highly impoverished (e.g. Brandt 1991: isopods, de la Cuadra & Garcia Gómez 2000: bryozoans, Zelaya 2005: bivalves). For example, in the Agassiz trawl samples at the same depths and vicinity the Echinoidea were represented by just a single deep sea species (Linse et al. 2008). That single echinoid species (Pourtalesia aurorae Kœhler, 1926), also illustrates the importance of the surrounding deep sea as a source of colonists for the South Sandwich Islands. In general, Antarctic benthos has an unusually high level of eurybathy (e.g. bivalves, gastropods, amphipods; Brey et al. 1996), suggesting there is great potential for upslope colonization. Antarctic polychaetes have such a high degree of eurybathy that Hilbig et al. (2006) found no significant difference between a shelf and deep-sea fauna. In the gastropods more than 30% of the species have a vertical distribution range of more than 500 m (14% more than 1000 m, 4% more than 2000 m). The buccinid Chlanidota signeyana Powell, 1951 and the philinid Philine alata Thiele, 1925 cover depth ranges of more than 5000 m (Schwabe et al. 2007). Antarctic bivalves show an even higher percentage of eurybathic species; 50% of the species cover a depth range of more than 500 m, 35% more than 1000 m and still 15% have depth ranges extending 2000 m. Members of the taxodont genus Yoldiella, the arcoid Limopsis marionensis (Smith, 1885) and the anomalodesmatan Cuspidaria tenella Smith, 1907 are examples for species with a wide bathymetric range (Linse 2004). No bryozoans have been reported from SO abyssal depths to date, yet can be abundant and diverse at slope depths (Barnes 2008). Even accounting for sampling differences it seems clear that the fauna changes drastically between slope and abyssal depths. Isopod species composition differs significantly between shelf and abyssal depths (Brandt et al. 2007). They do show strong eurybathy at family and generic levels (Table V, Schotte et al. 1995 onwards, Brandt et al. 2007) but the vertical distribution of most isopod species seems to be quite restricted, with a few notable exceptions (Glyptonotus antarcticus Eights, 1852: 10–634 m (Held & Wägele 2005); Ilyarachna antarctica Vanhoeffen, 1914: 252–7000 m; Leptanthura glacialis Hodgson, 1910: 50–5216 m; Neasellus kerguelensis Beddard, 1886: 220–1097 m (Schotte et al. 1995)). To date, though, very few comparative studies, such as the present one, have been carried out in the SO to investigate epibenthic macrofaunal assemblages or even on model taxa such as the isopods from the shelf to the deep sea. Thus, it is possible that species sampled on the shelf and the deep sea may be cryptic ones. Held & Wägele (2005) investigated the speciation in Glyptonotus antarcticus based on molecular analysis, found two haplotypes of this species in the eastern Weddell Sea and argued whether the coexistence might be due to different vertical distribution.
It is not straightforward to estimate expected levels of richness at Southern Thule. Even if it was well sampled (which it is not), interpretation of the richness found there would also not be easy. Besides being isolated and young, the South Sandwich Islands have only small areas at upper slope and shelf depths, with only restricted habitats. Furthermore they have erupted recently, so that along slope migrations would be difficult.
The South Sandwich Islands seem to be depauperate in gastropod and bivalve mollusc species (Zelaya 2005, Linse et al. 2006), which was interpreted as primarily due to their distance from South America and the Antarctic continental shelf. However, Zelaya (2005) conceded that the sediment, mainly consisting of pumice and lavas, was fairly unsuitable for bivalves. Low richness has also been reported for other taxa such as isopods (Brandt 1991, Brandt et al. 1999), bryozoans (Moyano 2005) and other groups. Yet, recent visits to Bouvetøya, a site also considered to be low in richness (with sampling intensities equivalent to the current study), have doubled or tripled the number of species known for some taxa, clearly showing sampling effort to be a major factor (Arntz et al. 2006) as may also be the case at Southern Thule. For example, three of the isopod families we found had not been previously documented in this region (Table VI). Like Southern Thule, many taxa are species poor at Deception Island, but as this is not isolated and last erupted decades before it, the impoverished species richness seems explained largely by still being in the process of recolonization and the suitability of local conditions (Cranmer et al. 2003, Lovell & Trego 2003, Barnes et al. 2008).
Depth range and new records (n) of South Sandwich Islands isopod families. Data from the BIOPEARL (BP), ANDEEP II (AII, Brandt, unpublished data) and LAMPOS (L, Arntz & Brey 2003) cruises and Brandt (B, 1991).
Many of the species known from the region tend to be widely distributed. Unsurprisingly, what we know of the fauna of the South Sandwich Islands to date suggests a strong affinity with those from other archipelagos of the Scotia Arc (see Tatian et al. 2005, Zelaya 2005, Linse et al. 2006), especially at the generic (e.g. see Moyano 2005, Zelaya 2005) or family level (Ramos-Espla et al. 2005). These might be explained by passively drifting organisms due to the west wind drift or Weddell Sea gyre especially for those with planktonic larvae; the dispersal of taxa with lecithotrophic larvae, though, is divisive still (Zelaya 2005). With the exception of Chaperiopsis galeata (Busk, 1854), all the bryozoans reported here from Southern Thule are typical shelf species also known at upstream localities such as the South Orkney Islands. As C. galeata is at its southernmost limit at Southern Thule, its presence there suggests that colonization from the shelves of Patagonia, South Georgia or other South Sandwich Islands is possible in ecological time despite distance and current directions.
The affinities can take the form of a cline of dissimilarity away from continent margins. For example, the species overlap between Magellan and Scotia arc isopods decreases with increasing distance from South America and at the South Sandwich Islands there are just five species in common discovered so far (Brandt et al. 1999). The archipelago represents the southernmost record for some species, e.g. the bryozoan Chaperiopsis galeata (Table V) in our samples and the northernmost limit of some Antarctic species, e.g. the echinoid Pourtalesia aurorae (Linse et al. 2008). Identification of Agassiz trawl and EBS samples from the BIOPEARL 1 cruise to genus and/or species level across taxa (so far only completed for echinoids, see Linse et al. 2008) should, on completion, provide a major step forward in assessment of faunal relationships between localities in this region.
The authors would like to thank the German Centre of Marine Biodiversity Research (DZMB) for the loan of Meta (EBS). We are grateful to Drs J. Smellie, P. Leat, G.D.F. ‘Buz’ Wilson, U. Mühlenhardt-Siegel, J. Guerrero-Kommritz, B. Stransky, P. Rehm for discussions on the manuscript. Peter Fretwell kindly provided the maps, H.J. Griffiths the AGT data. Thanks to L. Kramer, D. Münd and M. Schmiing for the help sorting the samples and to the officers and crew of the RRS James Clark Ross for the help on board. Finally we are particularly grateful to Dr C.-D. Hillenbrand for his valuable thoughts and comments on the manuscript throughout its development. Thank you to two anonymous referees for helpful comments. The financial support for S. Kaiser was provided by a grant of the German Science Foundation (DFG) under contract No Br 1121/ 26-2-3 and the Unistiftung, University of Hamburg. This study is a contribution to the SCAR EBA programme.