Restricted geographic distribution and low genetic diversity of the brooding sea urchin Abatus agassizii ( Spatangoidea : Schizasteridae ) in the South Shetland Islands : A bridgehead population before the spread to the northern Antarctic Peninsula ?

The glacial cycles of the Pleistocene have promoted the principal climatic changes of the Southern Ocean, and motivated scientifi c interest regarding the strategies developed by marine benthic invertebrates to tolerate and overcome the extension and contraction of the ice sheet on the Antarctic continental platform. A recent study of the bathymetric zonation and distribution of macro-invertebrates in a shallow subtidal area of Fildes Bay (King George Island, South Shetlands Islands, Antarctica) highlighted the presence of a large aggregation of the brooding sea urchin Abatus agassizii, whose geographic distribution is known only for localities south of the Antarctic convergence (Antarctic Peninsula and South Shetland and South Georgia Islands in the Scotia Arc). Its presence is atypical, given that these shallow populations should have been erased from the vicinity of the Antarctic Peninsula by the advances and retreats of the ice sheet, and the absence of a larval stage associated with brooding should limit re-colonization from northern Subantarctic areas. The aim of the study was to evaluate whether A. agassizii may have survived the glaciations in its narrow bathymetric range in the South Shetland Islands, or whether this population corresponds to a newcomer that re-colonized the area despite its low dispersal capacities. For this, we combined multidisciplinary approaches based on the geographical distribution of A. agassizii, its genetic diversity and its phylogenetic relationships with other species of the genus. In spite of an intensive sampling effort, the low occurrence of A. agassizii indicated that its distribution is very scarce along the Shetlands Islands and the Antarctic Peninsula, and seems to be restricted to protected and ice-free areas of Fildes Bay in King George Island. Moreover, this population presented very low genetic diversity associated with the signal of a recent demographic expansion. Finally, the reconstruction of the phylogenetic relationships among species of Abatus using mitochondrial COI sequences established the affi nity of the Antarctic A. agassizii with Subantarctic species. Based on these results we consider that the presence of this species in the Shetland Islands more likely corresponds to a recent re-colonization from Antarctic Islands located further north.


INTRODUCTION
The marine benthic fauna of the Antarctic continental shelf is considered to be the most isolated on the planet, due to the extreme environmental conditions and the surrounding barriers such as geographic distances, oceanic currents and deep waters (Lawver et al. 1992, Crame 1999, Clarke et al. 2005, Pfuhl & McCave 2005, Tripati et al. 2005).The current Antarctic biodiversity results from biogeographic and evolutionary processes driven by tectonic and climatic events since the Mesozoic (Clarke & Crame 1989, Crame 1999, Zachos et al. 2001, Brandt & Andreoli 2005, Pfuhl & McCave 2005, Scher & Martin 2006, Brandt et al. 2007, Rogers 2007, Aronson et al. 2007).
The Pleistocene glacial cycles were marked in Antarctica by the extension of the ice coverage over the continental shelf, and up to 100 km on the continental slope (Grobe & Mackensen 1992).Cer tainly, the glacial periods of the Cenozoic have attracted a lot a scientifi c attention regarding the strategies developed by the benthic fauna to tolerate these huge climatic changes in the Southern Ocean and the extension and contraction of the ice sheet on the Antarctic continental platform (Kennet 1977, Brey et al. 1996, Zachos et al. 2001, Convey et al. 2008).Three main hypotheses have been proposed to explain the survival during glacial maxima of benthic species that inhabit the Antarctic continental shelf.First, many benthic Antarctic taxa have extended bathymetric ranges compared to their relatives on other continental shelves.This allows movement in and out of deeper areas and could be an evolutionary adaptation (Brey et al. 1996), thus maintaining large population size and genetic diversity during glacial periods, as observed in the nudibranch Doris kerguelensis (Wilson et al. 2009) and in the pycnogonid Nymphon australe (Mahon et al. 2008, Arango et al. 2011).Second, according to the expansion-contraction model of Quaternary biogeography (Provan & Bennet 2008), it is possible that taxa with narrow bathymetric ranges, whose ecosystems were entirely disturbed during ice extension, could have endured the glacial maxima in the northern extremes of their distribution such as the Antarctic and Subantarctic islands of the Scotia Arc and other Subantarctic areas (González-Wevar et al. 2011, Hoffmann et al. 2011), where the shallow ecosystems may have been less impacted (Convey et al. 2008).According to this, current populations may represent recent re-colonization processes.The third hypothesis is the in situ persistence in Antarctic glacial refugia as occurred in terrestrial invertebrates including acarians, colembola and copepods (Convey & Stevens 2007, Rogers 2007, Convey et al. 2008) and green algae (De Wever et al. 2009).
The sea urchin genus Abatus (Spatangoidea: S c h i z a s t e r i d a e ) i n c l u d e s 1 1 b r o o d i n g species distributed in dif ferent provinces of the Southern Ocean (David et al. 2005).Two species are distributed in Subantarctic Provinces, A. cordatus (Verrill 1876) endemic to the Kerguelen Archipelago in the Indian Ocean, and A. cavernosus (Philippi 1845) in Patagonia, southern South America.The rest of the species (nine), A. agassizii (Mortensen 1910), A. beatriceae (Larrain 1986), A. bidens (Mortensen 1910), A. curvidens (Mortensen 1936), A. elongatus (Koehler 1908), A. ingens (Koehler 1926), A. nimrodi (Koehler 1911), A. philippii (Lovén 1871) and A. shackletoni (Koehler 1911) are distributed in the Antarctic continent (David et al. 2005).These species are deposit feeders that inhabit muddy to sandy de dispersión.A pesar de un intensivo esfuerzo de muestreo, la baja ocurrencia de A. agassizii indica que su distribución es muy escasa a lo largo de las islas Shetland del Sur y la Península Antártica y parece estar limitada a densas poblaciones someras en áreas protegidas y libres del hielo en la Bahía Fildes en la Isla Rey Jorge.Además, estas poblaciones presentan una muy baja diversidad genética asociada a una señal de una expansión demográfi ca reciente.Finalmente, la reconstrucción de las relaciones fi logenéticas entre las especies de Abatus, con secuencias mitocondriales del gen COI, establecieron la afi nidad de la especie antártica A. agassizii con especies subantárticas.De acuerdo a los resultados obtenidos, consideramos que la presencia de esta especie en las islas Shetlands del Sur probablemente corresponde a una reciente recolonización desde las islas antárticas situadas más al norte.
ecosystems (Ar naud 1974, Poulin & Féral 1995), and are found mainly in the upper part of the continental shelf (David et al. 2005).
In a recent study of the bathymetric zonation and distribution of macro-inver tebrates in a shallow subtidal zone of Fildes Bay (King George Island, South Shetlands Islands; Fig. 1), Palma et al. (2007) discovered a large aggregation of the brooding sea urchin Abatus agassizii between 3 and 8 meters depth.Such shallow distributions with high density have been described for the Subantarctic species A. cordatus (Poulin &Féral 1995, Ledoux et al. 2012) andA. cavernosus (Gil, pers. comm.), in sheltered areas of the intertidal and infralitoral zones.In contrast, Antarctic representatives including A. nimrodi, A. ingens and A. shackletoni are usually found at lower densities and below 20-25 meters depth (Chenuil et al. 2004, E Poulin & J-P Féral, unpublished data).Another remarkable characteristic of A. agassizii is the low number of occurrences repor ted, par ticularly around the Antarctic continent, where the species was found only once, in the eastern Weddell Sea, during the Polarstern Expedition Antarktis IX (OBIS).By contrast, this species has been reported recurrently from shallow subtidal areas from the South Georgia and South Shetland Islands.
Based on the above information, and in the context of the hypotheses previously advanced about the strategies developed by Antarctic species to overcome the glacial maxima, we provide here new information about geographic and bathymetric distribution, genetic diversity and phylogenetic relationships of A. agassizii.First we fur ther explore the extent of the distribution of A. agassizii along the South Shetlands Islands and the Antarctic Peninsula, compiling presence and absence records accumulated during several field trips and diving surveys between 2004 and 2011.Second, given that past demographic changes may leave a strong signal in the present genetic diversity, we evaluated the genetic diversity of the mitochondrial Cytochrome c Oxidase subunit I (COI) gene in A. agassizii in order to detect signals of recent population expansion.Third, we reconstructed phylogenetic relationships between Antarctic and Subantarctic Abatus species using the COI sequences.Finally, based on our results and other information available on distribution and life-history traits of Abatus species, we discuss whether a population of A. agassizii could have survived the glaciations in the South Shetland Islands within its narrow bathymetric range, or whether it may represent a newcomer that re-colonized the area despite its low dispersal capacities.

Sampling effort in the Antarctic Peninsula and surrounding islands
We compiled presence-absence records collected during eight successive fi eld trips at 57 geo-referenced locations between the latitudes 61-65°S (Table 1; Fig. 1), including the Shetland Islands, Palmer Archipelago, West Antarctic Peninsula and Red Island in the northeastern extremity of the Antarctic Peninsula.Exploration using scuba diving was made at different scales; fi rst at hundreds of kilometers to evaluate the overall distribution of the species along the northern Antarctic Peninsula and surrounding islands, and second at the scale of large bays (Admiralty and Fildes Bays).Surveyed areas were chosen based on typical environmental characteristics described for Abatus species (Poulin & Féral 1995, Palma et al. 2007); sheltered areas with medium to fine sand substrate and moderate to slight slope.Scuba diving samplings between 0 -15 m depth consisted of gently removing by hand the upper layer of sediment to detect directly the presence of Abatus individuals.

Analysis of genetic diversity
The genetic diversity of the A. agassizii populations from Antarctica was evaluated from COI fragments of 799 bp.We estimated the number of haplotypes (K), haplotype diversity (H), the number of polymorphic sites (S) and the mean number of pairwise differences (П) of each sequence set obtained for the three species using the program DNAsp v5 (Librado & Rozas 2009).The genealogical relationships among haplotypes were built using the Median Joining algorithm (Bandelt et al. 1999), implemented in the program Network v.4.1.5(Röhl 2002).To determine past demographic changes in A. agassizii, Tajima's D and Fu's F S tests were performed to evaluate the assumption of selective neutrality of mtDNA sequences and population mutation-drift equilibrium.Significant negative values for these tests are due to an excess of rare polymorphisms in a population, indicating either recent demographic expansion or positive selection.To evaluate the existence of a signal that this population has undergone a sudden population growth we constructed the mismatch distribution of the COI sequences, comparing it to the Poisson expectation for varying population sizes under a sudden expansion model (Slatkin & Hudson 1991, Rogers & Harpending 1992).For this, we estimated the raggedness statistic r (Harpending 1994) and evaluated its signifi cance by computer simulations using the coalescent.This statistic quantifi es the smoothness of the observed distribution of pairwise differences.Past population dynamics over time for A. agassizii were estimated using the Bayesian Skyline model in BEAST (Drummond et al. 2005;2012).The appropriate model of nucleotide substitution was GTR+G, determined using ModelTest ver.3.06 (Posada & Crandall 2001).Two independent runs of 2 x 10 8 iterations were performed; the model parameters were sampled every 1000 iterations.We used the mutation rate estimated by Lessios (2008) for sea urchins COI (3.5 % Ma-1).Demographic plots were visualized using Tracer v1.0.1 (Drummond & Rambaut 2007).

Phylogenetic relationships
The phylogenetic relationships among eight of the eleven existing Abatus species were reconstructed using mtDNA COI sequences: A. agassizii from Fildes Bay (King George Island), A. cordatus from Kerguelen Island, A. cavernosus from Argentinean Patagonia, A. elongatus from the Weddell Sea (Antarctic Peninsula), A. bidens, A. shackletoni, A. ingens and A. nimrodi from Adélie Land, East Antarctic.We rooted the tree with three outgroups: two belonging to the Order Spatangoidea: Paleopneustes cristatus (AJ639908) and Spatangus raschi (AJ639911.1),following the phylogeny reconstruction proposed by Smith et al. (2006), and one outgroup belonging to the Family Schizasteridae, Brachisternaster chesheri (JX000215).The size of the common fragment after alignment of all sequences was 690bp.Sequences were deposited in GenBank (JX000212-JX000219).Phylogenetic trees were constructed by Bayesian analysis using the program MrBayes v.3.1.2(Huelsenbeck & Ronquist 2003), with the substitution model GTR+G as indicated by the Modeltest program (Nylander 2004).This method estimates the posterior probability of each taxon being a monophyletic unit.All the Markov Chain Monte Carlo analyses (MCMC) were performed twice, using independent runs with random trees for a total of 7 x 10 6 generations.Trees were sampled every 500 generations and the fi rst 1000 trees were discarded as burn-in.The remaining trees were used to construct a consensus phylogram with the majority rule and posterior probability for the nodes.In addition we performed a Maximum Likelihood analysis using RaxML-VI v.7.0.4. (Stamatakis 2006), implemented in the Phylobench Phylogenetic Software Benchmarking Suite.Branch support was assessed with 100 non-parametric bootstrap replicates under the GAMMA model, and bootstrap values were plotted on the ML tree with the best likelihood.

Occurrence and absence of A. agassizii
No specimen of A. agassizii was found either along the west coast of the Antarctic Peninsula or on the east side (Fig. 1).Searches were also unsuccessful in the South Shetlands, except for King George Island.At this latter location, A. agassizii was absent from Admiralty Bay (Fig. 1A); it was found mainly in Fildes Bay, but also in King George Bay (only 1 specimen found after an extensive search, but not processed in this study) (Table 1).In Fildes Bay, A. agassizii was found in the southwestern part, outside the area af fected by the glacier.Ten local populations were identifi ed in the proximity of the Ardley Peninsula, Fildes Bay, King George Island (Fig. 1B).In several oppor tunities another irregular sea urchin, Trypilus abatoides (Schizasteridae) was found in syntopy with A. agassizii, but its occurrence never exceeded 5 % of the total abundance.

Genetic diversity
A low level of genetic diversity was observed in A. agassizii.Among 27 COI fragments of 799 bp, only four variable sites were found.One of the five haplotypes recovered was shared by 23 individuals (85.15 %), leading to extremely low values of haplotype diversity (H = 0.28) and mean number of pairwise nucleotide differences (П = 0.296).Negative and signifi cant values of Fu's F S and Tajima's D indicated an excess of low frequency haplotypes (-3.024  and -1.888, respectively), while the haplotype network exhibited a star-like structure (Fig. 3).The mismatch distribution showed a unimodal L-shaped pattern that fi t the expected Poisson distribution under the sudden expansion model (r = 0.27, P = 0.51, Fig. 3), suggesting the existence of a recent demographic expansion or a selective sweep.Similarly, the Bayesian Skyline plot revealed a very recent expansion that started around 10000 yr.Since this time, population effective size increased from around 400 individuals up to 40000 at the present (Fig. 4).

Phylogenetic relationships among Abatus species
Bayesian and Maximum Likelihood methods produced similar phylogenetic hypotheses in terms of branching and the robustness of some nodes.The tree topology suggests that Abatus species do not represent a monophyletic clade.In fact, the analysis revealed that Brachisternaster chesheri is the sister species of A. nimrodi.However, the rest of the species did form a monophyletic group with two welldif ferentiated clades.The first one includes A. agassizii (Antarctica) and the Subantarctic species A. cordatus and A. cavernosus, with a bootstrap suppor t of 75 % and posterior probability 0.96.The second clade includes the remaining Antarctic species with 95 % bootstrap support and 0.98 posterior probability (Fig. 2).Comparison of sequences among A. agassizii, A. cordatus and A. cavernosus revealed deep divergences among Antarctic, Patagonian and Kerguelenian species of Abatus.The average number of pair wise dif ferences between A. agassizii (Antarctica) and A. cavernosus (Kerguelen) was 28.15, and between A. agassizii (Antarctica) and A. cordatus (Patagonia) the average was 22.15.Finally, the average number of differences between Subantarctic A. cavernosus (Kerguelen) and A. cordatus (Patagonia) was 30.

DISCUSSION
The current distribution of a species is the result of a complex interplay between extrinsic factors such as isolation of landmasses, climatic conditions and niche availability, and intrinsic factors such as dispersal capacity and physiological tolerance (Leese et al. 2008).Although the presence of A. agassizii has been previously described in the South Shetlands Islands (OBIS 2007), the global occurrence data reported for A. agassizii are very scarce for a species accessible by Scuba diving.Our sampling ef for t allowed us to confirm that the presence of A. agassizii is sporadic along the South Shetland Islands and it is probably absent from the northern part of the Antarctic Peninsula.So far, A. agassizii populations are known to be common only in the eastern sector of Fildes Bay, in 10 sites located in Ardley Bay, around the Ardley Peninsula and the vicinity of the Great Wall Chinese Base.
In general, the presence of dense A. agassizii populations appears to be restricted by substrate type as well as wave exposure; it is found in patchy and dense aggregations at low Tendencia demográfi ca histórica del tamaño poblacional efectivo (Ne) construido con la aproximación bayesiana Skyline plot basado en haplotipos del citocromo Oxidasa subunidad I (COI) de A. agassizii.El eje x es el tiempo en unidades de años y el eje y es igual a N e t, el producto del tamaño poblacional efectivo (Ne) y la extensión de la generación en una escala logarítmica.Se muestran la estimación media (línea negra) y el 95 % de los límites más altos de densidad de probabilidad (HPD) (área gris).La línea discontinua gruesa representa el tiempo del ancestro común más reciente (tacmr) y la línea discontinua delgada representa el tiempo de la expansión en la especie.Sites visited and sampled in the Antarctic Peninsula and surrounding islands: geographic area, details, GPS coordinates of the locations, presence/absence of Abatus agassizii.
depth (1.5 to 8 meters), buried in a medium to fi ne sandy sediment.In this regard, Lopez & Levinton (1987) emphasized that the particle size composition of the substratum is usually a primar y factor determining the habitat of deposit feeders, infl uencing their respirator y and locomotive functions.These sites were all swell-sheltered and far from iced shores such as the Collins ice cap that infl uences the northeastern side of the bay.For example, the northernmost site where Abatus was found, near the Russian Tanks Bay (Table 1), is more exposed to swell action.In this locality, only few specimens have been found in deeper zones, generally around 10 m depth, where swell action is less severe.This shallow distribution contrasts with those repor ted for other Antarctic Abatus species such as A. nimrodi, A. ingens and A. shackletoni, that are generally found bellow 25-30 m depth (Chenuil et al. 2004, Chiantore et al. 2006).In this respect, the bathymetric range of A. agassizii is ver y similar to those described for the Subantarctic species A. cordatus and A. cavernosus.In these species, some dense aggregations (up to 64 and 140 individuals m -2 in Argentinean Patagonia and Kerguelen Islands, respectively) are found in shallow water at the bottom of sheltered bays in sediments that range from medium to fi ne sand (Poulin & Féral 1995, Gil et al. 2009).Probably due to all of these characteristics, current habitats along the Antarctic Peninsula are unsuitable for the settlement of A. agassizii populations.In the context of the current rapid climatic changes in the Antarctic Peninsula and the acceleration of ice sheet retreat, it is

Fig. 1 :
Fig. 1: Distribution of sites where the presence of Abatus agassizii was investigated by scuba diving.Top left corner: Antarctic Peninsula and surrounding Islands.Top right and bottom frames are zooms of King George Island and the South Shetlands.A: Admiralty Bay.B: Fildes Bay.Distribución de los sitios donde se investigó la presencia de Abatus agassizii mediante buceo.Cuadro izquierdo superior: Península Antártica e islas cercanas.Cuadros derecho superior y inferior son zoom de la Isla Rey Jorge, Shetland del Sur; A: Bahía Almirantazgo.B: Bahía Fildes.

Fig. 2 :
Fig. 2: Bayesian inference (BI) tree of the COI sequences obtained for the genus Abatus.The branch support values are from ML and BI analyses.

Fig. 3 :
Fig. 3: Mismatch distribution and network of COI haplotypes of A. agassizii.Haplotypes are represented by circles whose size indicates their frequency.bp: Base pairs.Distribución de Mismatch y red de haplotipos de COI para A. agassizii.Los haplotipos son representados por círculos cuyos tamaños dependen de la frecuencia.bp: pares de bases.

Fig. 4 :
Fig. 4: Historical demographic trends of the effective population size (Ne) constructed using the Bayesian skyline plot approach based on Cytochrome oxidase subunit I (COI) haplotypes of A. agassizii.The x-axis is the time in units of years and the y-axis is equal to N e t, the product of effective population size (Ne) and generation length in a log scale.The median estimate (black solid line) and 95 % highest probability density (HPD) limits (grey) are shown.The thick dashed line represents the time of the most recent ancestor (trcma) and the thin dashed line represents time for the expansion in the species.