Biodiversity and biogeography of Antarctic and sub-Antarctic mollusca

24 Pages·2006·1.18 MB·English

ARTICLE IN PRESS
Deep-SeaResearchII53(2006)985–1008
www.elsevier.com/locate/dsr2
Biodiversity and biogeography of Antarctic and
sub-Antarctic mollusca
(cid:1)
Katrin Linse , Huw J. Griffiths, David K.A. Barnes, Andrew Clarke
BritishAntarcticSurvey,NaturalEnvironmentalResearchCouncil,HighCross,MadingleyRoad,Cambridge,CB3OET,UK
Received2November2004;accepted3May2006
Availableonline11July2006
Abstract
For many decades molluscan data have been critical to the establishment of the concept of a global-scale increase in
species richness from the poles to the equator. Low polar diversity is key to this latitudinal cline in diversity. Here we
investigaterichnesspatternsinthetwolargestclassesofmolluscsatbothlocalandregionalscalesthroughouttheSouthern
Ocean. We show that biodiversity is very patchy in the Southern Ocean (at the 1000-km scale) and test the validity of
historicalbiogeographicsub-regionsandprovinces.Weusedmultivariateanalysisofbiodiversitypatternsatspecies,genus
and family levels to define richness hotspots within the Southern Ocean and transition areas. This process identified the
followingdistinctsub-regionsintheSouthernOcean:AntarcticPeninsula,WeddellSea,EastAntarctic—DronningMaud
Land,EastAntarctic—Enderby Land,EastAntarctic—Wilkes Land,RossSea,andtheindependentScotiaarc andsub
Antarctic islands. Patterns of endemism were very different between the bivalves and gastropods. On the basis of
distributionalrangesandradiationcentresofevolutionarilysuccessfulfamiliesandgenerawedefinethreebiogeographic
provinces in the Southern Ocean: (1) the continental high Antarctic province excluding the Antarctic Peninsula, (2) the
Scotia Sea province including the Antarctic Peninsula, and (3) the sub Antarctic province comprising the islands in the
vicinity ofthe Antarctic Circumpolar Current.
r2006Elsevier Ltd. Allrights reserved.
Keywords:Biogeography;Biodiversity;Endemism;Mollusca;Antarctica
1. Introduction seem predominantly explained bydistancefrom the
equatorial region or a correlate of this (Gaston and
Clinesofbiodiversityfromthetropicstothepoles Spicer, 2004). In the marine environment, however,
have long been considered to be one of the most robust data have taken much longer to gather and
fundamental patterns in spatial organisation of are still lacking for many major groups. Present
organisms.Althoughthereareclearlychangesalong data indicate a strong poleward decline in diversity
gradients other than latitude, patterns in terrestrial for northern hemisphere marine taxa (Clarke and
mammals,reptiles,birds,insectsandvascularplants Lidgard, 2000; Crame, 2000; Roy et al., 1996; Roy
et al., 1998; Stehli et al., 1967).
(cid:1) One of the earliest taxa to be used to investigate
Correspondingauthor.Tel.:[phone];
trendsinmarinebiodiversityhasbeentheMollusca,
fax:[phone].
E-mailaddress:[email protected](K.Linse). or more specifically the bivalves and gastropods,
0967-0645/$-seefrontmatterr2006ElsevierLtd.Allrightsreserved.
doi:10.1016/j.dsr2.2006.05.003
ARTICLE IN PRESS
986 K.Linseetal./Deep-SeaResearchII53(2006)985–1008
and data from these remain important today. 1995).Theseanalysespartitiontheglobaloceaninto
However,recentevidencefrombothoftheseclasses four major ‘biomes’, Trades, Westerlies, Polar and
hasshownthatpatternsinthesouthernhemisphere Coastal,andeachoftheseisfurthersubdividedinto
are quite different from those in the north (Valdo- atotalof50provinces.Ofthese,twooccursouthof
vinos et al., 2003) and that longitudinal patterns of the Polar Front (PF). The Antarctic Province lies
richness are just as strong as those with latitude between the PF at about 501S and the Antarctic
(Crame, 2000). Until recently a major difficulty in Divergence at 60–651S; it thus comprises the south-
theestablishmentof truly global patterns ofmarine ern branch of the powerful Antarctic Circumpolar
diversity has been the paucity of data for high Current (ACC) (Deacon, 1982; Orsi et al., 1995).
southernlatitudes,thatis,theundersamplingofthe South of the Antarctic Divergence lies the Austral
Southern Ocean and southernmost Atlantic, Indian Polar province, which comprises essentially the
andPacificoceans.Inthepastdecadeorso,marine seasonally ice-covered seas surrounding the coast
sampling intensity in these regions has started to of Antarctica. Immediately to the north of the PF
approach the levels elsewhere in the world, and our lies the Subantarctic Water ring province, part of
view of global patterns of marine diversity is the Antarctic Westerlies Biome, and including the
changing. The Southern Ocean constitutes (cid:1)8% of northern branch of the ACC (Deacon, 1982; Orsi
the world’s ocean surface area, (cid:1)2% of the global et al., 1995; Heywood and King, 2002). The
coastline, and (cid:1)11% of global continental shelf. oceanographic context is important to understand-
Recent sampling has revealed considerably higher ing the evolution of the Antarctic fauna: for
southern polar diversity than in the northern polar exampletheACC(theWestWindDriftintheearly
region (Arntz et al., 1994; Brey et al., 1994; Gutt biogeographic literature) is highly influential to the
et al., 2000), and many taxa have higher than links between the Magellanic region and the
expected richness relative to ocean area. Pycnogo- Subantarctic island faunas around the margins of
nids, polychaetes, ascidians, amphipods and the Southern Ocean (Barnes and De Grave, 2001;
bryozoans are all particularly speciose around Helmuth et al., 1994; O’Hara, 1998). Furthermore,
Antarctica (17.5%, 12.2%, 9.4%, 8.3% and 8.1% Antarctica’s long geographic and oceanographic
of global species, respectively) (Clarke and John- isolation have been critical to the evolution of the
ston, 2003). Though o1% of world species of high levels of endemism that characterise many
gastropods occur in southern polar waters, richness marine taxa.
is higher than reported from most of subtropical The early studies of the Southern Ocean marine
and temperate South America (west coast) (Valdo- fauna defined subregions based on several ideas.
vinos et al., 2003). Regan (1914), Nybeling (1947) and Powell (1951)
Taxa tend to form characteristic groupings by focussed on the distribution of model taxa, such as
region,andthescienceofbiogeographyhavegrown fish or molluscs. Earlier still, Waite (1916) and
to identify and characterise such groupings and the Norman(1937)usedtaxondistributionincombina-
environmental conditions that cause them. As with tion with geological data, and later Ekman (1935,
biodiversity studies, research and analysis of pat- 1953) relied on oceanographic evidence. Common
terns in biogeography historically have advanced to all these approaches was latitudinal division into
faster in the terrestrial than in the marine realm, a Subantarctic/low Antarctic zone, and an Antarc-
particularly in botany. For example, plant assem- tic/high Antarctic zone. In terms of longitude, the
blageshavebeengroupedbyclimaticfactorssuchas major divisions proposed were either into East and
precipitation, temperature and humidity to form West Antarctica, or into four quadrants. Later
‘zones’. In the sea, broad biogeographic regions studies (Holdgate, 1960; Knox, 1960; Hedgpeth,
have been erected by similarities in the known 1969) divided the sub-Antarctic zone (north of the
faunal groupings (Briggs, 1974; Dell, 1972; Hedg- PF, in those days termed the Antarctic Conver-
peth, 1969). As with large-scale patterns in biodi- gence) into the Magellan region, the Tristan da
versity, one of the main taxa used to elucidate such Cunha/Gough Island Archipelago and the Kergue-
similarity were the molluscs. A recent classification lenian region. They further divided the Antarctic
of marine biogeography has been based on a zone into a continental and South Georgia district.
combination of water-column physics and ocean- TheschemeproposedbyHedgpeth(1969,1970)was
colour data from Coastal Zone Colour Scanner widelyacceptedalthoughDell(1972)arguedagainst
(CZCS)imagery(Longhurst,1995;Longhurstetal., fine distinctions in the Antarctic region and
ARTICLE IN PRESS
K.Linseetal./Deep-SeaResearchII53(2006)985–1008 987
discussed the problems of fitting isolated islands (1972). To test these hypotheses we have investi-
into regional patterns. Since these seminal studies gated patterns of southern polar biodiversity in the
our knowledge on the distributions and diversity of shelled gastropod and bivalve molluscs from the
the Antarctic marine fauna has increased through local, through regional to global scale.
intensive sampling in recent years (Arntz et al.,
1997; Clarke and Johnston, 2003), but the biogeo- 2. Material and methods
graphic patterns of Southern Ocean molluscs have
not been re-examined. 2.1. Geographic areas
With the recent increase in biodiversity and
biogeographic studies, there has been a marked We investigated data availability for all major
increase in the amount of distributional data for areas within the Southern Ocean (waters south of
Southern Ocean marine organisms, and this has the PF), the sub-Antarctic islands, the Magellan
lead to the traditional faunal/floral lists becoming region of South America and the southern New
too large and unmanageable. Relational databases Zealandislands.WithintheSouthernOcean12sub-
have begun to take the place of the faunal/floral regions were defined based on multivariate analysis
catalogues in biodiversity and biogeographical of the recent diversity records in shelled gastropods
studies (Budd et al., 2001; Hill et al., 2000; and bivalves (Clarke et al., under review). For our
Rosenberg, 1993; Peat, 1998; Zhang et al., 2000). biogeographicanalysisweusedthese12sub-regions
The advantage of relational databases is that they theWeddellSea,EastAntarctica—DronningMaud
can store and link distributional data with the Land, East Antarctic—Enderby Land, East Ant-
information of abundance, ecology and environ- arctic—Wilkes Land, Ross Sea, West Antarctic
ment that are needed to analyse and explain (EightsCoast,BellingshausenandAmundsenSeas),
patterns in biodiversity and to test ecological and Antarctic Peninsula, South Shetland Islands, South
biogeographic hypotheses (Grassle and Stocks, Orkney Islands, South Sandwich Islands, Bouvet
1999). Relational databases can handle vast Island, and South Georgia, as well as Heard and
amounts of data and complicated queries relatively Peter I islands. The neighbouring areas included in
quickly, and deliver raw data matrixes for multi- our recent study were the Prince Edward, Crozet,
variate analyses. Several such databases have been Chatham, Auckland, Kerguelen archipelagos,
created,includingmarinemolluscandatabasesfrom Campbell Island and the Magellan region, subdi-
different geographic regions (Griffiths et al., 2003; vided into the Falkland Islands, Tierra del Fuego,
Rosenberg, 1993) and the database of Indo-Pacific Strait of Magellan, Atlantic and Pacific coasts of
Marine Molluscs housed by the Acadamy of the Magellan region. The ocean areas of the
Natural Sciences, Philadelphia). Linking relational neighbouring areas and sub-regions within Antarc-
databases to a Geographical Information System tica were written into polygons, which defined the
(GIS) creates a powerful tool to analyse large-scale geographic range of each area using GIS (ArcGIS
biogeographic patterns (e.g., Markwick, 2002; 8.2) (Fig. 1A).
Markwick and Lupia, 2002; Griffiths et al., 2003;
Myers and Worm, 2003; OBIS: www.iobis.org; 2.2. Data source and definitions
Census of Marine Life: www.coml.org).
Here we analyse one of the most comprehensive BiogeographicdatafromtheSouthernOceanand
regionaldatabasesyetassembledforanytaxon, the neighbouring areas were regularly updated in
Southern Ocean Molluscan Database (SOMBASE: SOMBASE (Griffiths et al., 2003). For compar-
Griffiths et al., 2003), to investigate richness and ability of data between areas, and to minimise the
biogeographictrendsintworelatedmolluscantaxa, sample effort bias, only distribution records for
gastropods and bivalves. We hypothesise that their shelfspeciesofshelledgastropodsandbivalveswere
biogeographicpatternsinAntarcticaaredetermined included (Fig. 1B, C). We defined ‘shelf’ species as
primarilybyhistoricalfactorsandthatthecompar- those occurring in waters between 0 and 1000m
ison of biogeographic patterns between the ecologi- (ClarkeandJohnston,2003).Thereforespecies,that
cally different taxa gastropods and bivalves will are recorded only from depths deeper than 1000m
show similar trends in biodiversity and biogeogra- were excluded from the analyses presented here.
phy and that biogeographic regions will be in ToassesstheAntarcticspeciesrichnessinrelation
accordancewith those ofHedgpeth(1969) and Dell toglobalrichnesspatternsinshelledgastropodsand
ARTICLE IN PRESS
988 K.Linseetal./Deep-SeaResearchII53(2006)985–1008
Fig.1. MapsofSouthernOceanandsurroundingwithneighbouringcontinentsoverlaidwithPolarFront:(A)Biogeographicareas,(B)
shelledgastropodrecordsand,(C)Bivalverecords.
ARTICLE IN PRESS
K.Linseetal./Deep-SeaResearchII53(2006)985–1008 989
bivalves, we assembled species richness per zone 2.3.2. Diversity centres
data using published diversity records for gastro- Numbers of species per genus and number of
pods (Spencer and Willan, 1995; Roy et al., 1998; species per family were compared between study
Higo et al., 1999; Bouchet et al., 2002; Valdovinos areas to investigate any general patterns. Only
etal.,2003;CLEMAMdatabase,2005;OBISIndo- species-rich families and genera, or those judged to
Pacific Molluscan Database,) and bivalves (Crame, have radiated significantly within the Southern
2000). Ocean, were included in this analysis. The shelled
gastropod families we selected were: Buccinidae,
Turridae, Trochidae, Naticidae, Rissoidae, Murici-
2.3. Data analysis
dae, Cerithiidae, Eatoniellidae, Diaphanidae, Cy-
clostrematidae, and Cancellariidae. The genera we
Raw species-distribution data for the Southern
selected were: Prosipho, Pareuthria, Chlanidota,
Ocean and neighbouring areas were analysed by
Trophon, Margarella, Falsimargarita, Eatoniella,
probing SOMBASE in selected fields, such as
Diaphana, and Toledonia. Likewise the bivalve
taxonomic level, defined depth range, or defined
families we selected were: Nuculanidae, Sareptidae,
area polygons, then using the query results for GIS
Philobryidae, Limopsidae, Limidae, Pectinidae,
or multivariate analysis in either ArcGIS (ESRI,
Montacutidae and Cuspidariidae; and the bivalve
2002) or PRIMER 5 (Clarke and Warwick, 2001).
genera we selected were Nucula, Yoldiella, Philo-
The methodologies used for the different analysis
brya, Lissarca, Adacnarca, Limopsis, Limatula,
are described below.
Mysella, and Cuspidaria.
2.3.1. Taxon distribution patterns 2.3.3. Estimating endemism
For the analysis of distribution patterns at Wedefinedendemictaxaassumingthattaxawere
different taxonomic levels (family, genus, species), not found beyond the limits of our database
taxon counts for each of the pre-defined areas were (SOMBASE). Thus we designated a taxon as
donebyusingaspatialjoinbetweenthedatapoints endemic if it only occurred in one of our pre-
and the geographic polygons, which defined the defined areas. The identifications of taxa that were
extent of each of the areas under investigation. The onlyfoundinoneareawerebasedonthetaxonper
database created included a table containing the area richness lists produced for the taxon distribu-
speciesdata,samplesites,andtheareainwhichthe tionpatternsanalysis(seeSection2.3.1).Anytaxon
siteswerelocated.Toavoidcountingasingletaxon found in only one area was then assumed to be
more than once per area, a second query was endemic. This method uses the entire database, not
createdtoreducethelist tothefirstoccurrence ofa justtheselected areas,toensurethatoccurrencesin
taxoninanarea.Oncethissecondquerywasextant, regions outside the polygons are not ignored. The
a third query (taxa per area query) was entered to level of endemism per area was then displayed as
count the number of taxa per area. This analysis taxonnumbersandalsoasapercentageofthetotal
was done six times to assess the taxon arearichness taxon number per area.
for gastropod families, genera and species, and
bivalve families, genera and species separately. 2.3.4. Biogeographic relationships
To analyse the number of shelf taxa only, the We used the software package PRIMER 5
steps in the analysis were very similar to that just (Clarke and Warwick, 2001) to analyse biogeo-
described. On the initial query, only species known graphic relationships amongst shelled gastropods
from less than 1000m depth (the Antarctic con- and bivalves in our study areas. The analysis was
tinental shelf is unusually deep Clarke and John- based on the distributions of shelf species only, but
ston, 2003) were selected by adding the pre-defined on all records of these species, including those
depth range to the table. This was done by using records from deeper than 1000m depths. The
bathymetric data for the entire study area, the species-per-area lists produced by the taxa-per-area
Southern Ocean and neighbouring areas, and using query (see Section 2.3.1) were used as the data
a location based GIS selection to produce a list of matrix for PRIMER 5. Faunal similarity between
species found in areas less than or equal to 1000m areas was measured by quantitative Bray-Curtis
depth.Theanalysiswasrunsixtimesforeachofthe similarities (Bray and Curtis, 1957) of non-trans-
different taxonomic units studied. formed binary (presence-absence) data. Non-metric
ARTICLE IN PRESS
990 K.Linseetal./Deep-SeaResearchII53(2006)985–1008
multidimensional scaling (nMDS) and cluster ana- Molluscantaxonrichnessdifferedbetweenshelfand
lysis were applied to resemblance data to display alldepthsattheSouthSandwichIslands,theSouth
faunal similarities in two-dimensions. The data Shetland Islands, and in the Weddell Sea. Typically
matrixes for shelled gastropods and bivalves were larger(shelf)areas,e.g.,theWeddellSea,RossSea,
analysed both separately and pooled. Tierra del Fuego and East Antarctic–Wilkes Land,
were richest, followed by the larger islands (Falk-
3. Results land Islands, South Georgia, and the Kerguelen
Islands). The small and most isolated islands, such
3.1. Southern Ocean gastropod and bivalve richness as Bouvet Island, the Prince Edward Islands, and
Campbell Island, exhibited the lowest richness. The
From our 27 study areas in the Southern Ocean three richest areas of our study areas were South
and adjacent regions approximately 895 shelled Georgia (199 species/107 genera/62 families) and
gastropod and 379 bivalve species are known to those on the margins of west and east Antarctica:
date.Ourresultshighlighttheverypoorknowledge theWeddellSeashelf(279/119/65)andtheRossSea
of the Amundsen Sea fauna, and also that certain (192/111/57).
other areas have been so poorly sampled (e.g., the The Southern Ocean areas had higher species per
Bellingshausen Sea) that the diversity values are of family(median2.7:1)andspeciespergenus(median
littlemeaning.Weexaminedshelf(0–1000mdepth) 1.66:1) ratios than areas north of the Southern
patterns, but taxon lists in most areas (e.g., South Ocean(medians1.88:1and1.27:1respectively).The
OrkneyIslands,StraitofMagellanortheRossSea) WeddellSeaandTierradelFuego,forexample,had
differedlittlebetweenshelfandalldepths(Table1). similar numbers of genera and families, but the
Table1
Molluscantaxonnumbers(Shelledgastropods&bivalves)perarea
Area Species Genera Families
Shelf Alldepths Shelf Alldepths Shelf Alldepths
AucklandIsland 123 123 85 85 50 50
BellingshausenSea 11 11 9 9 7 7
EightsCoast 6 6 3 3 3 3
BouvetIsland 36 36 31 31 24 24
CampbellIsland 47 47 43 43 32 32
ChathamIslands 5 5 5 5 5 5
CrozetIslands 74 74 43 43 29 29
EastAnt.EnderbyLand 110 110 69 69 44 44
EastAnt.WilkesLand 164 167 92 95 54 55
EastAnt.Dron.MaudLand 105 105 63 63 34 34
FalklandIslands 126 126 82 82 51 51
HeardIsland 15 15 12 12 9 9
KerguelenIslands 129 129 78 78 49 49
MagellanStrait 116 116 89 89 54 54
NewZealand(SouthIsland) 70 72 64 66 40 42
Peninsula 130 131 80 81 47 47
PeterIIsland 3 3 3 3 2 2
PrinceEdwardIslands 52 52 44 44 37 37
SouthGeorgia 199 206 107 112 62 65
SouthOrkneyIslands 112 112 73 73 46 46
SouthSandwichIslands 40 60 33 48 24 34
SouthShetlandIslands 133 160 78 90 49 56
SouthernArgentina 26 26 24 24 20 20
SouthernChile 125 125 96 96 62 62
RossSea 192 194 111 113 57 58
TierradelFuego 178 178 107 107 63 63
WeddellSea 279 301 119 126 65 69
ARTICLE IN PRESS
K.Linseetal./Deep-SeaResearchII53(2006)985–1008 991
Weddell Sea was much richer in species. The had higher ratios of gastropod to bivalve taxa
dominant trends in taxon richness in the pooled (median2:1species)thanareastothenorth(median
shelf dataset were also apparent in individual 1.46:1).
analyses for shelled gastropods (Appendix A) and There were, however, distinct differences in the
bivalves(AppendixB). Foralmostevery study area richness patterns both between the shelled gastro-
and at all taxonomic levels gastropods were much podandbivalves,andalsowithinthesetwogroups,
richer than bivalves. Southern Ocean study areas at different taxonomic levels (Fig. 2). Shelled
(A) (B)
0-21 0-7
22-42 8-14
43-63 15-21
64-84 22-28
85-105 29-35
106-125 36-42
126-146 43-49
147-167 50-56
168-188 57-63
189-209 64-70
Gastropod Species Bivalve Species
(C) (D)
0-8 0-6
9-17 7-11
18-25 12-17
26-33 18-22
34-42 23-28
43-50 29-33
51-58 34-39
59-66 40-44
67-75 45-50
76-83 51-55
Gastropod Genera Bivalve Genera
(E) (F)
0-4 0-4
5-8 5-7
9-12 8-11
13-16 12-14
17-20 15-18
21-24 19-22
25-28 23-25
29-32 26-29
33-36 30-32
37-40 33-36
Gastropod Families Bivalve Families
Fig.2. Hotspotsoftaxonomicrichnessinpre-definedareasoftheSouthernOceanandneighbouringareas;(A)gastropodspecies,(B)
bivalvespecies,(C)gastropodgenera,(D)bivalvegenera,(E)gastropodfamilies,(F)bivalvefamilies.
ARTICLE IN PRESS
992 K.Linseetal./Deep-SeaResearchII53(2006)985–1008
gastropods had higher richness levels in high but high at generic and familial levels in shelled
Antarctic areas than bivalves, and two areas of gastropods, whereas species richness was higher
East Antarctica were impoverished in bivalve taxa than generic or familial levels in bivalves (Fig. 2).
butmid-richnesslevelsingastropods(Fig.2).Areas
identified as having the highest taxonomic diversity 3.2. Centres of diversity in the Southern Ocean
(‘hotspots’) differed depending on the taxonomic
level of the analysis. For example South Georgia Analysisofspeciesrichnessinfamiliesandgenera
patterns of richness were moderate at species level of shelled gastropods and bivalves revealed that
Fig.3. Distributioncentresofhigh-latitudesouthernmarinemolluscs.Areasofhightolowrichnessoffamilies/genera.Thespecifictaxa
areCyclostrematidae,Buccinidae,Sareptiidae,Turridae,Diaphanidae,NaticidaeandCuspidariidae,andToledoniaandCuspidaria(A),
Rissoidae,Trochidae,Philobryidae,MuricidaeandEatoniellidae,andPareuthria,EatoniellaandTrophon(B),LimidaeandMontacutidae
and Mysella, Limatula, Yoldiella, Prosipho, and Falsimargarita (C), Pectinidae, Nuculidae, Volutidae, Mytilidae, Gaimardiidae or
Condylocardiidae (D), and Limopsidae, Cancellariidae, and Cerithiidae, Philobrya and Limopsis (E). The shading represents levels of
speciesrichnessfromhightolow.
ARTICLE IN PRESS
K.Linseetal./Deep-SeaResearchII53(2006)985–1008 993
mostwereencompassedbyjustfiveoverallpatterns Table2
(Fig.3).ThefamiliesCyclostrematidae,Buccinidae, Molluscanendemictaxa(Shelledgastropods&bivalves)perarea
Sareptiidae, Turridae, Diaphanidae, Naticidae and
Area Species Genera Families
Cuspidariidae, and the genera Toledonia and
Cuspidaria had the Weddell and Ross Seas as N % N % N %
centres of taxon richness (Fig. 3A), with richness
BellingshausenSea 1 9.1 0 0.0 0 0.0
decreasing towards the Scotia arc, Antarctic Penin-
EightsCoast 0 0.0 0 0.0 0 0.0
sula region, and East Antarctica, and lowest in the
BouvetIsland 13 36.1 0 0.0 0 0.0
sub-Antarctic, Magellanic and other areas. A CrozetIslands 2 2.7 0 0.0 0 0.0
second pattern apparent was a richness centre EastAnt.EnderbyLand 5 4.5 0 0.0 0 0.0
spanning the Weddell Sea to Magellanic areas, EastAnt.WilkesLand 17 10.4 0 0.0 0 0.0
EastAnt.Dron.MaudLand 20 19.0 0 0.0 0 0.0
through the Scotia arc (Fig. 3B). The families
FalklandIslands 23 18.3 0 0.0 0 0.0
Rissoidae, Trochidae, Philobryidae, Muricidae,
HeardIsland 0 0.0 0 0.0 0 0.0
and Eatoniellidae, and the genera Pareuthria, KerguelenIslands 21 16.3 0 0.0 0 0.0
Eatoniella and Trophon all exhibited this pattern. MagellanStrait 21 18.1 0 0.0 0 0.0
ThethirdpatternwashighSouthernOceanrichness Peninsula 7 5.4 2 2.5 0 0.0
PeterIIsland 0 0.0 0 0.0 0 0.0
coupled with low richness in the sub-Antarctic and
PrinceEdwardIslands 9 17.3 0 0.0 0 0.0
other northern areas (Fig. 3C). The families
SouthGeorgia 65 32.7 2 1.7 0 0.0
Limidae and Montacutidae, and the genera Mysel- SouthOrkneyIslands 22 19.6 3 4.1 0 0.0
la,Limatula,Yoldiella,Prosipho,andFalsimargarita SouthSandwichIslands 2 5.0 0 0.0 0 0.0
showed this type of distribution. Conversely, the SouthShetlandIslands 6 4.5 0 0.0 0 0.0
RossSea 22 11.5 1 1.0 0 0.0
third distribution type was a high richness north of
TierradelFuego 41 23.0 0 0.0 0 0.0
thePFandlowrichnesssouthofit.Thiswasshown
WeddellSea 55 19.7 2 1.7 0 0.0
by the families Pectinidae, Nuculidae, Volutidae, SumSouthernOcean 464 75.2 27 13.2 0 0.0
Mytilidae, Gaimardiidae and Condylocardiidae
(Fig. 3D). Finally, the families Limopsidae, Can-
cellariidae, and Cerithiidae, and the genera Philo-
brya and Limopsis had a centre of richness in the (Anderssonia, Antistreptus, Eatoniopsis, Kaitoa,
Weddell Sea (Fig. 5E). Munditia, Pickenia and Tropidomarga), which only
occurred in one area of the Southern Ocean, did so
3.3. Endemism along the Scotia arc/Antarctic Peninsula range.
Dickdellia and Intortia were only found in the
The species level of Antarctic endemism for both Weddell Sea, and Murdochella only recorded in the
shelledgastropodsandbivalveswithintheSouthern Ross Sea.
Oceanwasapproximately75%(Table2).Atgeneric There were far more endemic species of shelled
and familial levels this was reduced to 13.2% and gastropodsthanbivalves(Fig.4,AppendixC).Only
0%,respectively.Comparisonofregionalendemism in the Strait of Magellan and Tierra del Fuego was
within and outside the Southern Ocean showed no the number of endemic bivalve species higher than
obvious patterns, though with the exception of that of gastropods. Just two bivalve genera,
South Georgia, areas within the Southern Ocean Adamussium and Ptychocardia, were endemic to
had low rates of species endemism (4.5–19.7%). the Southern Ocean, but neither were endemic to
Species level endemism in areas outside the PF any of the individual study areas. In contrast to
varied between 2.7% (Crozet Islands) and 23% bivalve patterns, the Weddell Sea and South
(Tierra del Fuego). The two highest levels of Georgia were the two hotspots of endemism in
endemism amongst our study areas were at South gastropods (Fig. 4A, C).
GeorgiaandBouvetislands;apartfromlatitudeand
being within the PF, these two areas have little in 3.4. Analysis of zoogeographic relationships
common. South Georgia is a large, old island
whereas Bouvet is a small, young one. At genus Multivariate analysis of the 27 study areas at the
level none of our study areas north of the PF had threeselectedtaxonomiclevelsrevealedthreemajor
endemics and only five areas within the Southern biogeographic regions at high southern latitudes.
Ocean did (Table 2). Seven of the ten genera, These were the Magellan region, Southern Ocean,
ARTICLE IN PRESS
994 K.Linseetal./Deep-SeaResearchII53(2006)985–1008
(A) (B)
0-6 0-2
7-12 3-4
13-17 5-7
18-23 8-9
24-29 10-11
30-35 12-13
36-41 14-15
42-46 16-18
47-52 19-20
53-58 21-22
Gastropod Species Bivalve Species
(C) (D)
0-1 0-1
2-3 2
4 3
5 4
6-7 5-6
8 7
9 8
10 9
11-12 10
13 11
Gastropod Genera Bivalve Genera
Fig.4. Patternsofendemisminhigh-latitudesouthernmarinemolluscswithtaxonandtaxonlevel.Thetaxaare,(A)gastropodspecies,
(B)bivalvespecies,(C)gastropodgenera,(D)bivalvegenera.Thecolouredscalebarrepresentsnumbersofendemictaxa.
and New Zealand. This pattern emerged both with similarity this cluster separated into three Magella-
the pooled data and with separate analyses for nic groups, Southern Chile, Falkland Islands, and
shelled gastropod and bivalve data (Figs. 4 and 5). Strait of Magellan/Tierra del Fuego, and two
Pooled and separate analyses also suggested that Antarctic ones, one of which was East Antarctica
there were five main outliers: Bellingshausen Sea, and the other was the Antarctic Peninsula, Scotia
Eights Coast, Chatham Islands, Heard Island and arc islands, Kerguelen, Crozet, and Prince Edward
Peter I Island, all of which are poorly sampled Islands. At genus level, three major clusters
locations. At the family level the regions split at separated at 40% similarity, centred around New
(cid:1)60% similarity level, whereas at generic level Zealand, South America and Antarctica. Bouvet,
divergence occurred at (cid:1)45% similarity. At the Heard and the South Sandwich Islands clustered
species level the split occurred at between 20 and separately. Within the Antarctic generic cluster
25%, depending on whether the shelled gastropod threesub-groupsformedseparatingintoEast,West
and bivalve data were considered separately or (includingtheScotiaarcislands),andsub-Antarctic
combined (Fig. 4, Table 3). islands (Kerguelen, Crozet, and Prince Edward
Separate analyses of shelled gastropod data Islands). In contrast, at species level these sub-
showed similar area relationships to analyses of Antarctic islands had formed a separate cluster
pooled data at species and genus level, but not at alongside those of New Zealand, South American,
family level (Table 3). A cluster at the family level and Antarctic clusters.
with New Zealand affinities split at (cid:1)40% from a Separate analyses of the bivalve data showed
large cluster combining most other areas. At 65% patterns similar to the pooled data, with two