Marine Biology

, Volume 160, Issue 4, pp 821–841

Seeing stars: a molecular and morphological investigation into the evolutionary history of Odontasteridae (Asteroidea) with description of a new species from the Galapagos Islands


    • Department of Biological SciencesAuburn University
    • Department of Biological SciencesUniversity of West Florida
  • Kenneth M. Halanych
    • Department of Biological SciencesAuburn University
Original Paper

DOI: 10.1007/s00227-012-2136-x

Cite this article as:
Janosik, A.M. & Halanych, K.M. Mar Biol (2013) 160: 821. doi:10.1007/s00227-012-2136-x


Odontasteridae (Asteroidea: Echinodermata) (Verrill in Am J Sci, 1899) is placed within Valvatida, a derived assemblage of sea stars. Odontasterids are found in the Southern, Atlantic, and Pacific Oceans and are concentrated in high southern latitudes. To date, the phylogenetic and evolutionary history of Odontasteridae as a whole has not been rigorously examined. We conducted molecular and morphological phylogenetic analyses of Odontasteridae to assess the interrelationships among and within recognized genera. We used mitochondrial 16S and cytochrome c oxidase subunit I molecular markers and 29 external morphological characters in an attempt to reconstruct the evolutionary history of the group. Generally, our results indicate that traditionally used external skeletal characters are not representative of phylogenetic history of Odontasteridae. We can conclude that species present in high latitudes in the Southern Hemisphere (i.e., Southern Ocean) are the most derived taxa. Additionally, mtDNA data suggest unrecognized lineages of odontasterids are present in high southern latitudes. A new species Odontaster cynthiae sp. nov. is described from the Galapagos Islands.


Odontasteridae Verrill, 1899 is within the largest and most taxonomically diverse group of Asteroidea, Valvatacea (sensu Blake 1987). Odontasterids are a clade of sea stars typically found in lower shelf and upper bathyal regions, though some have been collected in the tidal and intertidal zones. Found in the Southern, Atlantic, and Pacific Oceans (Fig. 1), Odontasteridae is characterized by two series of equal, opposite, and usually conspicuous marginal plates without intermarginal channels, two rows of suckered tube feet, and triangular mouths (Clark and Downey 1992). Odontasterid sea stars also characteristically have five rays, are most notably distinguishable by hyaline-tipped recurved spines surrounding the mouth (Fig. 2), and range from stellate to almost pentagonal in shape.
Fig. 1

Worldwide distribution of Odontasteridae. Dots indicate select known collection localities
Fig. 2

Characteristic hyaline-tipped recurved spines surrounding the mouth. a Single recurved spine at jaw apex, b double recurved spines at jaw apex, c spines of O. cynthiae. The scale bar represents 5 mm

Odontasterid sea stars are exclusively benthic as adults and play an important role in marine ecosystems and food chains. For example, Odontaster validus has been labeled a keystone species in the Southern Ocean, capable of applying considerable influence on the environment through predation. It can be very abundant in shallow waters of high productivity where O. validus consumes a variety of organisms (McClintock et al. 1988).

The taxonomic history of Odontasteridae has been complicated throughout the past two hundred years. Originally, Odontasteridae members were known as Gnathasteridae Perrier 1894. The name Odontasteridae was first erected by Verrill in 1899, and as a result, Gnathasteridae was invalidated when Gnathaster Sladen, 1889 was synonymized with Odontaster Verrill, 1880. Fisher (1940) and Clark and Downey (1992) have been the primary authorities in providing species descriptions and sorting out Odontasteridae taxonomy. As currently recognized, there are six genera (Acodontaster, Diabocilla, Diplodontias, Eurygonias, Hoplaster, and Odontaster) and twenty-seven accepted species (Table 1). Acodontaster Verrill, 1899 is composed of five species, A. capitatus, A. conspicuus, A. elongatus, A. hodgsoni, and A. marginatus (Fig. 3a–d); Gnathaster elongatus Sladen 1889 is the type species and is now regarded as synonym of A. elongatus. This genus is found mainly in the Antarctic, with one “subspecies” A. elongatus granuliferus, extending into the Atlantic. Diabocilla contains only one species, Diabocilla clarki McKnight, 2006, and is possibly a synonym of Hoplaster Perrier in Milne-Edwards, 1882 (H. kupe and H. spinosus: Fig. 3f) (unpublished data; C. Mah pers. comm.). Hoplaster spinosus Perrier 1882 is the type species. Despite the fact that Odontasteridae is characteristically known for hyaline-tipped glassy teeth, both the deep-sea D. clarki and Hoplaster lack this character. Diplodontias Fisher, 1908, comprising D. dilatatus, D. miliaris, D. robustus, and D. singularis (Fig. 4a–d), was previously referred to as Asterodon Perrier, 1891 and is uniquely characterized by a pair of large hyaline-tipped teeth at each jaw (Fig. 2b). Pentagonaster dilatatus Perrer (1875) by monotypy in Goniodon Perrier 1894 is the type for Dipladontias. Diplodontias is revived from the synonymy of Asterodon Perrier, to which it was referred by Fell (1953) since both Asterodon and Goniodon are invalid junior synonyms. Eurygonias is monospecific; E. hyalacanthus Farquhar, 1913 (Fig. 3e) has a single recurved spine on each oral plate and is endemic to New Zealand. Finally, Odontaster Verrill, 1880, comprised of 14 species, is the most conspicuous genus and is found in the Atlantic Ocean to the Gulf of Mexico, the northeastern Pacific Ocean, and Southern Ocean (for depth and specific localities: see Table 1). This genus includes O. aucklandensis, O. australis, O. benhami, O. crassus, O. hispidus, O. mediterraneus, O. meridionalis, O. pearsei, O. penicillatus, O. rosagemmae, O. robustus, O. roseus, O. setosus, and O. validus (Figs. 5, 6, 7). Odontaster hispidus Verrill, 1880 is described as the type species.
Table 1

List of Odontasteridae species used in this study along with description reference, distribution, depth, and catalog numbers





Museum/catalog number




Clark (1962)

Bellingshausen to Ross Sea

193–647 m

USNM E53545, E53473


Fisher (1940)

Adelie Land, Graham Land, South Georgia

46–647 m

USNM E53236


Fisher (1940)

Sub-Antarctic, Palmer Archipelago, Patagonia, Falkland Islands

91–600, 50–336 m

USNM E13681, 1082875


Clark (1962)

Western Antarctic, South Georgia

4–457 m

CASIZ 174674; USNM E43860


Clarke and Johnston (2003)

Graham Land and Queen Mary Land

250–291 m

USNM 1082938




McKnight (2006)

known only from near Chatham Rise, central New Zealand, on the hills Diabolical and Zombie

890–970 m





Clark and McKnight (2001)

Cook Strait southwards to Snares Island, New Zealand

0–70 m

NIWA 43632, 43646; USNM E9986


Clark and McKnight (2001)

East coast of South Island, Kiakoura to Foveaux Strait, New Zealand

0–101 m

NIWA 43636; USNM E10145


Clark and McKnight (2001)

Auckland Islands, South of New Zealand




Clark and Downey (1992)

Mar del Plata, N. Argentina to Tierra del Fuego, also from Chile

0–84 m

USNM 1084439,




Clark and McKnight (2001)

east coast of New Zealand from Cook Strait south to Snares Island

0–7 m

NIWA 43624, 43622, 43618




Clark and McKnight (2001)

West of North Island, New Zealand, Fairway Trough, Bellona Gap, Lord Howe Rise

2,000–2,417 m

NIWA 15439, 43647, 43631


Clark and Downey (1992)

Azores, west Ireland north up the Rockall Trough and far to the south off Cape Town

1,795–3,310 m





Clark and McKnight (2001)

Chatham Rise, Campbell Plateau, Bounty Platform, New Zealand

55–353 m

NIWA 43626, 43629, 43637, 43688, 31216, 43640


Clark and Downey 1992

west coast of South Africa

243–366 m



Clark and McKnight (2001)

Hawke Bay southwards to S. New Zealand, Chatham Islands; New South Wales, Australia

0–549 m, in Australia 468–549 m

NIWA 43627, 43634, 43619, 43633, 43630, 28106, 43644; USNM E09755


Fisher (1911)

Monterey Bay to San Diego, California

80–500 m

CASIZ 113242; USNM 31828


Pawson and Ahearn (2000)

Galapagos Islands

105–925 m

USNM E51299, E51298; CASIZ 115202


Clark and Downey (1992)

George’s Band, NE of Cape Cod to Florida Strait

50–1,160 m

USNM E26326


Clark and Downey (1992)

Porcupine Seabight; SW of Ireland, Bay of Biscay; Mediterranean

414–1,800 m

USNM 030212


Fisher (1940)

Antarctic, circumpolar; north to South Georgia, Marion Island, Kerguelen

0–646 m

NIWA 43639USNM 1104652, E53413, 1091163


Janosik and Halanych (2010)

Antarctic Peninsula

132 m

USNM 1127022


Clark and Downey (1992)

around Cape Horn, Argentina, south to the Falkland-Magellan region; Chile

8–350 m

NIWA 43628; USNM E47752, 1082945, 1104651, 1084431


Clark and McKnight (2001)

east of Chatham Island and off the east coast of North Island

445–1,190 m

NIWA 43623, 43635, 43621


Janosik and Halanych (2010)

Antarctic Peninsula

132 m

USNM 1127023


Clark and Downey (1992)

S. of Cape Cod to Florida; northern Gulf of Mexico

160–675 m

USNM E12910, E37332,


Clark and Downey (1992)

from off Martha’s Vineyard to Carolina coast

100–739 m

USNM 1017559, 1017562


Fisher (1940)

Antarctic, circumpolar, north to South Georgia, Bouvet Island

0–653 m

NIWA 43620, 43625, 27912, 27928; USNM E13408
Fig. 3

Plate I: aboral view. aA. marginatus, bA. elongatus, cA. capitatus, dA. conspicuus, eE. hyalacanthus, fH. kupe
Fig. 4

Plate II: aboral view of Diplodontias species. aD. dilatatus, bD. singularis, cD. robustus, dD. miliaris
Fig. 5

Plate III: aboral view of Odontaster species. aO. benhami, bO. crassus, cO. hispidus, dO. aucklandensis, eO. cynthiae, fO. meridionalis
Fig. 6

Plate IV: aboral view of Odontaster species. aO. setosus, bO. validus, cO. robustus, dO. rosagemmae, eO. penicillatus, fO. roseus
Fig. 7

Bayesian inference topology for combined 16S and COI sequence data. Details of analysis are provided in text. Number next to node indicates Bayesian posterior probabilities. Color indicates species locality, see legend

To date, the phylogenetic and evolutionary history within Odontasteridae has not been rigorously examined. Here, we use a combination of external morphological characters and molecular markers, 16S ribosomal DNA (16S) and cytochrome c oxidase subunit I (COI), to investigate their evolutionary history, providing insight into speciation and biogeographical patterns that may have shaped the evolution of Odontasteridae. In particular, we were interested in understanding whether this group originated and radiated from the Southern Ocean to more northern latitudes or vice versa.


Phylogenetic relationships within Odontasteridae were examined using 16S and COI molecular data. Then to further elucidate the evolutionary trends, morphological characters were mapped onto the molecular topology.

Specimen collection

Specimens were obtained from the Division of Echinoderms, Smithsonian Institution National Museum of Natural History (USNM) in Washington, DC, the Department of Invertebrate Zoology, California Academy of Sciences (CASIZ), San Francisco, California, and the National Institute of Water and Atmospheric Research (NIWA), New Zealand (Table 1). Most specimens were dried. Antarctic species were collected during two five-week research cruises aboard the R/V Laurence M. Gould in November/December of 2004 and May/June of 2006. Images of D. clarki were provided by NIWA.

Molecular data

Molecular methods follow Janosik et al. (2011). DNA extraction of specimens was performed by using DNeasy® Tissue Kit (Qiagen). Two mitochondrial DNA markers (16S and COI) were utilized to estimate the evolutionary history of Odontasteridae. Specifically, a 508-bp region of the mitochondrial 16S gene was amplified using the 16SarL and 16SbrH primers and protocols of Palumbi et al. (1991). For the same individuals, a 627-bp region of the COI gene was amplified using primers designed to work with Odontaster COI-Ast 22F (5′-TTYTCNACNAAACAYAAGGA-3′) and COI-Ast722R (5′-GGRTGNCCRAARAAYCARAA-3′) (Janosik et al. 2011). Amplified products were purified with either a Qiagen QIAquick® Gel Extraction Kit (Qiagen Inc.) or a Montage PCR Filter Units (Millipore) according to the manufacturer’s directions. Purified products were then sequenced bidirectionally on a Beckman CEQ 8000 Genetic Analysis System (Beckman Coulter). Sequences were edited and aligned using Sequencher 4.6 (Gene Codes Corporation) and Bioedit v.7.0.8 (Hall 1999). COI sequences were translated according to the echinoderm mitochondrial DNA code to aid in proofreading. Genbank accession number is listed in Table 2.
Table 2

List of outgroup species used in this study along with description reference and GenBank accession numbers



GenBank accession numbers

B. loripes

Clarke and Johnston (2003)


C. moorei

Bell (1894)


C. papposus

Clark and Downey (1992)

COI-AF217383, 16S-DQ2917113

L. foliolata

Fisher (1911)

16S-EU072952, COI-HM473924

M. aequalis

Fisher (1911)

16S-EU072953, COI-HM542271

S. stimpsoni

Verrill (1909)

COI-217382, 16S-DQ297084

Based on current understandings of sea star relationships (Blake 1987; Mah and Foltz 2011), Bathybiaster loripes, Chaetaster moorei, Crossaster papposus, Luidia foliolata, Mediaster aequalis, and Solaster stimpsoni were selected for the outgroup, and these sequences were downloaded from GenBank ( (Table 2).

Models of nucleotide substitution were selected with AIC (GTR + G was calculated separately for the 16S and COI data sets) in MrModeltest ver. 2.2 (Nylander 2004). Separate analyses were performed on 16S and COI data sets followed by an analysis of the concatenated data set. A Bayesian approach was used to infer phylogeny using MrBayes 3.1.2 (Ronquist and Huelsenbeck 2003). Posterior probabilities were obtained by a Markov chain Monte Carlo (MCMC) algorithm, which uses two sets of one cold and three heated chains. Samples of trees and parameters were drawn every 100 steps from a total of 2 × 106 MCMC generations. The first 4,000 trees were discarded as the burn-in (based on convergence of likelihood values), and the remaining trees were used to compute a consensus tree.

Morphological data

Characters consist of external skeletal features and variation in accessory structures and spines. For each species, multiple specimens were examined with the unaided eye and by stereomicroscope. Morphological characters were scored from published descriptions, photos, and/or museum specimens. Terminology follows Lambert (2000) and Clark and Downey (1992). Table 1 provides a list of species, references containing descriptions, and museum numbers employed herein. Additionally, Odontaster specimens collected from around the Galápagos Islands were also included in morphological character analyses. Pawson and Ahearn (2000) published a report of the echinoderms collected from submersible dives using the Johnson-Sea-Link, including a new Odontaster species, which does not corroborate the descriptions of known species. We scored morphological characters of these specimens to quantify previously unrecognized biodiversity, but were unable to extract usable genomic DNA.

A data matrix consisting of 29 characters and 28 in-group taxa was constructed in NEXUS data editor 5.0 (Page 2001) (Table 3 in Appendix). Nine characters were scored as binary and 19 were coded as unordered multistate. Morphological characters were mapped onto the recovered molecular tree to distinguish the important external characters useful for phylogenetic analysis. Character transformations were evaluated and mapped onto the molecular tree using a parsimony approach to show all most parsimonious states at each node using Mesquite ver. 2.74 (Maddison and Maddison 2010). First, the morphological character matrix was imported and followed by the combined 16S and COI Bayesian inference. Mesquite applies stochastic models of character state change and can explicitly accommodate uncertainty in ancestral states. Characters were mapped only for species present in the molecular tree.


Analyses of the concatenated 16S and COI molecular data sets consisted of 1,135 bp and provided greater resolution than topologies based on individual genes. Analyses based on individual genes (16S and COI) are shown in Appendices 2 and 3. Discussion will focus on the concatenated gene topology (Fig. 7). The recovered topology of the combined 16S and COI data sets using the GTR + G model had marginal support across the tree. Nonetheless, a few key nodes had strong posterior probabilities, and they will be the focus of the discussion. For example, the node defining a monophyletic Odontasteridae had a posterior probability of 0.97 and monophyly of Diplodontias and Odontaster are well supported (1.00, 0.97, respectively). Acodontaster is not a recovered monophyletic as A. elongatus is a recovered sister to Odontaster species, and Hoplaster kupe is recovered with the majority of Acodontaster species. Eurygonias hyalacanthus is a recovered basal to Acodontaster and Odontaster species. Diplodontias is recovered as monophyletic and as the most basal clade within Odontasteridae.

Morphological characters of Odontasteridae were mapped onto the Bayesian topology from the combined molecular markers (Appendix 4). Mapping was conducted using a parsimony criterion with the accelerated transformation in Mesquite ver. 2.74 (Maddison and Maddison 2010). Because we are interested in understanding which characters are phylogenetically informative, characters with a Consistency Index of 0.50 or greater are shown in Fig. 8. Such characters include the following: recurved spine(s) on oral plates (character 1), abactinal plate shape (character 2), number of spines per abactinal plate (character 3), presence or absence of glassy granules on abactinal plates (character 5), abactinal spine shape (character 6), distribution of papulae on abactinal surface (character 7), distinctness of marginal plate border (character 8), length of spines on inferomarginal plates (character 14), and presence or absence of glassy granules on actinal plates (character 19).
Fig. 8

Morphological characters with Consistency Index of 0.50 or greater mapped onto Bayesian inference tree. Characters include the following: recurved spine(s) on oral plates (character 1), abactinal plate shape (character 2), number of spines per abactinal plate (character 3), presence or absence of glassy granules on abactinal plates (character 5), abactinal spine shape (character 6), distribution of papulae on abactinal surface (character 7), distinctness of marginal plate border (character 8), length of spines on inferomarginal plates (character 14), and presence or absence of glassy granules on actinal plates (character 19). Analysis details are provided in text. Bars indicate where a change has occurred, and squares indicate character reversals or loss of a character. Asterisks indicate character states that were equally parsimonious at a node

Monophyly of Odontasteridae is supported by recurved spine(s) on oral plates (character 1). Reversals were evident for four characters (Appendix 4: characters 1, 6, 7, 14). For example, shape of spines on inferomarginal and superomarginal plates (characters 12 and 14) is often used to distinguish between genera, but in terms of assessing phylogenetic relationships, these characters may not be useful due to repeated convergent evolution. Specific character changes, reversals, and losses are described in character descriptions below.

The Galapagos Odontaster specimens (3 individuals) are morphologically distinguishable from all other known Odontaster species. Specifically, these Odontaster specimens can be recognized by a longer, prominent spine in the middle of each abactinal plate (character 6). No other known Odontaster species possess such a spine. Thus, Galapagos specimens represent distinct biodiversity and warrant full species status as Odontaster cynthiae (see description below).

Description of taxa

  • Family ODONTASTERIDAE Verrill, 1899

  • Genus Odontaster Verrill, 1880

  • Odontaster cynthiae nov. sp. (Fig. 5e)

Material examined. Holotype.

Equatorial Pacific Ocean; Galapagos Islands, Darwin Island, 01°42′N, 092°00′W, 348–435 meters in depth, specimen wet (alcohol) R = 2.3 mm, r = 1.4 mm, USNM E51298, July 18, 1998, collected by D.L. Pawson and J. McCosker. Paratypes: USNM E51299 (two individuals R = 1.8 cm, r = 1.2 cm; R = 2.0 cm, r = 1.3 cm), Equatorial Pacific Ocean; Galapagos Islands, Darwin Island, 01°42′N, 092°00′W, 348–435 meters in depth, specimen wet (alcohol); CAS 115,202 (one individual R = 1.9 cm, r = 1.2 cm), North Pacific Ocean; Galapagos Islands, Isla Espanola, 01°22.20′S, 089°49.20′W, 353.5 meters in depth, specimen wet (alcohol).


The descriptor cynthiae is named in honor of Cynthia Anne Gust Ahearn, Museum Specialist, Department of Invertebrate Zoology, USNM, whose contributions greatly enriched echinoderm biology.


Arms 5. R = 2.03 mm; r = 1.30 mm; R/r = 1.56. Body form almost pentagonal; rays not distinguishable. Interradial margin slightly incurved. Overall, body laterally flat. Arm tips tilting slightly up. Characteristic recurved glassy-tip spine on each oral plate. All plates of this sea star are decorated with spines.


Abactinal plates small and almost flat. Abactinal plates hexagonal to rounded in shape and smaller in interradial regions. Radial regions swollen with interradial regions depressed. Each abactinal plate covered with 5–8 rough and slender spines, with one longer prominent spine in the center of the plate. All spines on abactinal plates taper from a thicker base to the tip, forming a point. Pedicellariae present on abactinal surface mostly at border between abactinal plates and marginal plates. Pedicellariae straight, with two to four hooks. Madreporite rounded surrounded by abactinal plates with spines.

Fourteen marginal plates present interradially (arm tip to tip). Marginal plates form a strong rounded boarder with abactinal plates, but a smooth transition from marginals to actinal plates. Marginal plates are larger compared to abactinal and actinal plates. Plates homogenous in size in interradial region and slightly smaller at the tip of the arm. Superomarginal plate surface covered in rough short spines with one longer spine at the edge of the plate. Pedicellariae lining edges of superomarginal plate with a few randomly scattered on plates. Inferomarginal plates similar in size to superomarginals. Plate surface covered in rough spines with two prominent rows of 3–5 longer spines at the edge of the inferomarginal plate and superomarginal plate. Rows of medium spines also present before and after the row of prominent longer spines. No pedicellariae present on inferomarginal plates.

Actinal plates arranged to form 3 complete chevrons with plates shaped polygonal to rectangular. Actinal plates toward arm tips are small compared to plates found interradially. Plates small and flat with spines for armature. All spines rough in appearance and taper, being more slender at tip compared to base. Spines per plate ranges from 6 to 9, with spines arranged in a circular fashion around one longer prominent spine situated in the center of the plate. Pedicellariae present on actinal plates positioned parallel to the adambulacral plate series. One pedicellariae per plate, with two to four jaws. A few randomly scattered pedicellariae present on actinal plates. Furrow, sub-ambulacral regions crowded. Three to four rough, slender furrow spines with approximately 8 spines on each adambulacral. Oral plates covered with spines, with one dominating enlarged glassy-tipped recurved spine on each plate. Armature on each plate consists of four spines lining each side of the enlarged recurved spine. Oral plates triangular in shape. Ambulacral furrow shallow, tube feet biserial. Wet preserved specimens, colored off-white.


North Pacific Ocean, Galapagos Islands, Darwin Island; Isla Espanola, 349–436 meters in depth.


This species is distinguishable by one longer, prominent spine in the middle of each abactinal plate (character 6). The distribution of this species appears isolated and specimens have only been collected from around the Galapagos Islands. At the time of description, only formalized samples were available and attempts at DNA characterization failed. Additionally, based on morphological characters, O. cynthiae is most likely closely related to O. crassus.



Phylogenetic reconstruction based on mitochondrial genes (Fig. 7) suggests that a monophyletic Odontasteridae consists of three main clades. These clades roughly correspond to the three more speciose genera: Acodontaster, Diplodontias, and Odontaster. Whereas the monophyly of Odonataster (PP = 0.97) and Diplodontias (PP = 1.00) are both well supported by Bayesian inference, support for a monophyletic Acodontaster is lacking. Moreover, Diplodontias is recovered as the basal-most clade (PP = 0.97) and Odontaster as a derived clade within Odontasteridae. Eurygonias hyalacanthus is recovered sister to Acodontaster and Odontaster species. Hoplaster appears to be related to the main Acodontaster clade, which is consistent with depth distribution.

Monophyly of Acodontaster is called into question by placement of Acodontaster elongatus as a sister to Odontaster clade, though support for this placement is generally weak. Interestingly, A. elongatus has been previously referred to Odontaster cremeus Ludwig, 1903. Fisher (1940) considered Odontaster cremeus a synonym of A. elongatus. Clark (1962) argued that O. cremeus was in fact a valid species, but according to Jangoux and Massin (1986), the O. cremeus type has been lost and thus made O. cremeus a junior synonym of A. elongatus. Based on our inspections herein, A. elongatus superficially looks more similar to Odontaster rather than Acodontaster. Specifically, A. elongatus has prominent marginal plates, a compressed flattened body, rather than the typical puffy Acodontaster appearance, and a body shape outline that is more similar to Odontaster species. Given the placement of A. elongatus as sister to Odontaster, Ludwig’s original designation deserves reevaluation. Thus, taxonomic revision is likely necessary for A. elongatus, definitive revision should wait until the phylogenetic position of A. elongatus is better supported, and additional samples O. cremeus and A. elongatus have been studied. With further investigation and possible location of the Odontaster cremeus type, taxonomic revisions are likely necessary for A. elongatus.

Biodiversity (based on molecular data and morphological characters) within Odontasteridae is greater than previously thought, particularly in the Southern Ocean (see Janosik and Halanych 2010). Given previous hypotheses about Antarctic invertebrate endemicity and distributions spanning the Drake Passage (Dell 1972; Dayton et al. 1974; Arntz et al. 1994; Shaw et al. 2004; Hunter and Halanych 2008; Thornhill et al. 2008), we included individuals of Acodontaster capitatus and Odontaster meridonalis collected from both the north and south sides of the Drake Passage in the Southern Ocean (i.e., South American waters and Antarctic Peninsular waters). Molecular data show individuals from South American waters are different from individuals from Antarctic waters; a common trend in the Southern Ocean (reviewed in Janosik and Halanych 2010). Further phylogeographic analyses and specimens are needed, especially for these species to determine whether species diversity has been under recognized.

Morphological and taxonomic implications

Examination of the morphological characters in light of the molecular topology suggests some characters, that is, number of recurved spine(s) (character 1); shape, texture, number of abactinal spines and plates (characters 2, 3, 6), and presence or absence of glassy granules (character 6), are taxonomically useful when determining genera, species boundaries, or unrecognized biodiversity in Odontasteridae. Overall, the number of, or lack of, glassy recurved spines at the oral apex (character 1) appears to be an informative character when defining species of Odontasteridae and when distinguishing between genera (CI = 1.0, Fig. 2a–b). Specifically, Acodontaster, Eurygonias, and Odontaster species all possess a glassy recurved spine on each set of oral plates, equaling a total of five per individual (Fig. 2a). Diplodontias species have two glassy recurved spines on each set of oral plates, for a total of ten spines at the jaw apex (Fig. 2b). Even though the number of recurved spines at the jaw apex is a strong generic-level diagnostic character (Clark and Downey 1992), Diabocilla and Hoplaster both lack oral spines at the jaw apex. However, Diabocilla and Hoplaster maintain the other Odontasteridae taxonomically informative characters, such as abactinal plate shape (character 2), number of spines per abactinal plate (character 3), and marginal plate border (character 8). Whether the last common ancestor of Odontasteridae possessed one or two recurved glassy spines, are both equally parsimonious hypotheses (Fig. 8). Additionally, a single recurved glassy spine at the jaw apex (Acodontaster, Odontaster) is a more derived character, while no spine(s) at the jaw apex appears to have been lost at least once (Hoplaster).

Other taxonomically informative characters include the following: paxillate or tabulate abactinal plates (character 2, CI = 0.67), number of abactinal spines per plate (character 3, CI = 0.5), and abactinal spine shape (character 6, CI = 0.71). For example, Odontaster species have paxillate abactinal plates (character 2), while Eurygonias hyalacanthus has highly paxillate plates, both more derived character states than tabulate plates which are seen in most Acodontaster and Diplodontias individuals. Likewise, more basal genera have more abactinal spines per plate, while more derived genera have fewer. Abactinal spine shape (character 6, CI = 0.71) and presence or absence of glassy granules on the abactinal plates (character 5, CI = 0.50) are helpful in distinguishing species. Distribution of papulae on the abactinal surface (character 7, CI = 0.67) is also informative when distinguishing between ancestral and more derived genera. Overall, several characters on the abactinal plates appear to be important and informative for taxonomy and species designation, but the majority of external morphological characters are not as informative in reconstructing the evolutionary history of Odontasteridae.

Broadly, presence or absence of certain characters can be used to unite members within Acodontaster, Diplodontias, or Hoplaster, but not within Odontaster. Particularly, Odontaster seems to be the most variable and problematic genus, because finding morphological characters that unite this genus and/or distinguish it from other members are lacking. As a general rule, Odontaster species have paxilliform abactinal plates (character 2). The great diversity of character variation within Odontaster may be due to the fact that there are 15 currently recognized species in this genus. Other genera with Odontasteridae are far less speciose (Acodontaster = 5, Diabocilla = 1, Diplodontias = 4, Hoplaster = 2, Eurygonias = 1). Thus, given that Odontaster has a near worldwide distribution and is collected at a variety of depths, it is likely occupying different ecological niches possibly resulting in unique character evolution (Fisher 1940; Clark and Downey 1992; Clark and McKnight 2001). This variable and speciose genus may also be a result of human-induced bias when defining and assigning species. Thus, although many morphological characters have been useful for taxonomic designations, several of the external skeletal characters typically used in asteroid taxonomy are evolutionarily plastic. Because of this situation, more effort is need to understand which morphological features are appropriate for use in phylogenetic inference versus taxonomic designation and identification. In addition, more thorough sampling of each species is necessary to determine the synonymy.

Unfortunately, we were not able to gain usable DNA from all Odontasteridae samples. Thus, outstanding questions still remain, including placement of O. cynthiae and D. clarki. As discussed below, some morphological characters are more representative of seastar phylogeny, as judged by the mtDNA tree, than others. By focusing on characters that show limited homoplasy, we are able to hypothesize that O. cynthiae is closely related to a derived Odontaster clade circumscribed by O. mediterraneus and O. hispidus. Similarly, D. clarki is likely associated with Hoplaster. McKnight (2006) states that D. clarki differs by having abactinal plates barely elevated and both abactinal and marginal plates that are covered with tubercles (granules) rather than spines (characters 2, 3). Characters uniting Diabocilla and Hoplaster include the following: missing recurved glassy spines at the jaw apex (character 1), clavate-shaped abactinal spines (character 6), and the marginal plates form an even border with the abactinal plates (character 8). Character differences between D. clarki and Hoplaster include the following: more spines per abactinal plate (character 3), fewer spines per actinal plate (character 18), and papulae on the abactinal surface (character 7), although these characters are often variable within a genus.

Biogeographical implications

Southern Ocean organisms are thought to have originated by introduction and subsequent diversification into Antarctic regions from adjacent regions or by origins in Antarctic waters with diversification into surrounding regions. Diversity within Odontasteridae is highest in high southern latitudes. Our analyses show species occurring strictly in Antarctic waters (Acodontaster spp. and Odontaster spp. shown in blue in Fig. 7) are among the most derived members, suggesting that patterns of diversification occurred into or within Antarctic regions. This scenario is plausible given that several odontasterids occur in areas adjacent to the Southern Ocean (i.e., southern tips of South America, and South Africa, New Zealand, Australia). Whereas this conclusion is consistent with other studies (e.g., Brandt 1992; Crame 1993; Gebruk 1994; Briggs 2003; Pawlowski et al. 2007; Brandt et al. 2007) that suggest the Southern Ocean to be a center of origin for deep-sea organisms, it contrasts with Strungell et al.’s (2008) findings that deep-sea octopods radiated out of the Antarctic during periods of diversification. Presumably, the presence of cold water is the common factor in both deep and polar waters that allow the organisms to survive, but patterns of diversification in the deep Southern Ocean may need to be treated on a taxon-by-taxon basis.


Dr. Christopher Mah from the Smithsonian Institution National Museum of Natural History kindly helped AMJ while visiting to study the Odontasteridae collection. Special thanks go to Cynthia Ahearn from the USNM for lending and coordinating specimens used in this study. Loans were also provided by the California Academy of Sciences and the National Institute of Water and Atmospheric Research for loan of specimens. We are grateful for the help and logistical support by the crew and participants of the 2004 and 2006 Antarctic cruise aboard the R/V Laurence M. Gould. This project was supported by National Science Foundation grants (OPP-9910164, OPP-0338087, and OPP-0338218, ANT-1043745) to K.M Halanych and R.S. Scheltema. This is AU Marine Biology Program contribution #95 and Molette Lab contribution #9.

Copyright information

© Springer-Verlag Berlin Heidelberg 2012