MORPHOLOGICAL AND GENETIC ANALYSES SUGGEST THAT SOUTHERN AFRICAN CROWN CRABS, HYMENOSOMA ORBICULARE, REPRESENT FIVE DISTINCT SPECIES BY M. T. EDKINS1 ), P. R. TESKE2 ), I. PAPADOPOULOS2,3 ) and C. L. GRIFFITHS1,4 ) 1 ) Department of Zoology, University of Cape Town, Private Bag Rondebosch, 7701 Cape Town,

South Africa 2 ) Molecular Ecology and Systematics Group, Botany Dept., Rhodes University, 6140

Grahamstown, South Africa 3 ) Zoology Department, Nelson Mandela Metropolitan University, P.O. Box 77000, Port Elizabeth

6031, South Africa

ABSTRACT The crown crab, Hymenosoma orbiculare, occurs all along the coast of southern Africa and displays great morphological variation across this range. To determine whether the species comprises several distinct taxonomic units, H. orbiculare were collected from 18 estuaries and lagoons between Walvis Bay in Namibia and Kosi Bay in north-eastern South Africa. Open ocean individuals were also obtained from False Bay in south-western South Africa. Morphological and genetic (using mitochondrial DNA) comparisons were carried out between individuals from all locations. Five monophyletic clusters were identified on the basis of genetic data, each confined to specific portions of the distribution range. Morphological data supported the distinctness of each of these clusters. The typical H. orbiculare, characterized by large size (maximum carapace width 28 mm) and absence of the characteristic ornamentations of other morphotypes, occurred in estuaries and lagoons all along the west and south coasts. False Bay deep-water individuals were of two forms. The first small, granulose, and setose morph appears to represent the previously synonymized species, H. geometricum, and is distinguished by a long rostrum, a raised setaceous gastric region on the carapace, and post-branchial projections. The second False Bay form represents a new, undescribed species, identified primarily by genetic characters. Estuarine forms from the southeast and east coast formed two more clusters, distinguished by their small size (carapace width <10 mm), relatively large eyes, long walking legs, projections on the abdomen, and small anterior spikes on the coxae of the legs. Specimens from south-eastern sites differed from more northern samples by their longer second walking legs and darker colour. We propose that South African crown crabs in fact represent five distinct species: the true H. orbiculare, the form previously described as H. geometricum (to be re-established as a valid species), and three new species that remain to be formally described.

4 ) e-mail: [email protected]

© Koninklijke Brill NV, Leiden, 2007 Also available online: www.brill.nl/cr

Crustaceana 80 (6): 667-683

668

M. T. EDKINS ET AL.

ZUSAMMENFASSUNG Die Kronenkrabbe, Hymenosoma orbiculare, kommt überall entlang der Küste des südlichen Afrikas vor und zeichnet sich in diesem Verbreitungsgebiet durch beträchtliche morphologische Variation aus. Um festzustellen, ob es sich bei der Art um verschiedene taxonomische Einheiten handeln könnte, wurden Proben von H. orbiculare in 18 Flussmündungen und Lagunen zwischen Walvis Bay in Namibia und Kosi Bay im nordöstlichen Südafrika gesammelt sowie in tieferem Wasser der False Bay im Südwesten von Südafrika. Die Tiere der verschiedenen Fundstellen wurden sowohl morphologisch als auch genetisch (mitochondriale DNS) miteinander verglichen. Mit Hilfe der genetischen Analyse konnten fünf monophyletische Einheiten identifiziert werden, jede davon beschränkt auf einen bestimmten Bereich des Verbreitungsgebietes der Art. Dieses Resultat wird durch morphologische Unterschiede unterstützt. Die eigentliche H. orbiculare wurde in Flussmündungen und Lagunen der West- und Südküste gefunden. Sie lässt sich von anderen Morphotypen durch Grösse (maximale Carapaxweite 28 mm) und Fehlen einiger für andere Morphotypen typischer Ornamentierungen unterscheiden. Zwei Morphotypen wurden in tieferem Wasser der False Bay gefunden. Der erste davon ist kleiner, granulöser und beborsteter und scheint die früher einmal synonymisierte Art H. geometricum zu repräsentieren. Diese unterscheidet sich von anderen Kronenkrabben der False Bay durch ein längeres Rostrum, eine beborstete Vorwölbung der gastrischen Region des Carapax und post-branchiale Vorsprünge. Der zweite Morphotyp aus der False Bay stellt eine neue, noch unbeschriebene Art dar, deren Eigenständigkeit hauptsächlich auf genetischen Daten beruht. Die in den Flussmündungen der Südost- und Ostküste vorkommenden Kronenkrabben repräsentieren zwei weitere Morphotypen, die sich durch ihre geringe Körpergrösse (Carapaxweite <10 mm), die relative Grösse der Augen, die Länge der Schreitbeine, durch Vorsprünge auf dem Pleon und kleine anteriore Dornen auf den Coxae der Beine auszeichnen. Individuen der Südostküste unterscheiden sich von weiter nördlich vorkommenden Krabben durch ihr längeres zweites Beinpaar und ihre dunklere Färbung. Wir sind der Meinung, dass es sich bei den südafrikanischen Kronenkrabben um fünf eigenständige Arten handelt: die eigentliche Art H. orbiculare, die Art, die vormalig als H. geometricum beschrieben wurde (und als eigene Art wieder etabliert werden sollte), sowie drei weitere Arten, die noch beschrieben werden müssen.

INTRODUCTION

Hymenosoma orbiculare Desmarest, 1825 (Decapoda, Brachyura, Hymenosomatidae) is a well-known and common crab in southern African estuaries and lagoons, as well as sheltered marine habitats up to 80 m depth (Barnard, 1950). An unusual population also persists in the relict coastal freshwater Lake Sibaya (Allanson et al., 1966). The recorded distribution extends from Walvis Bay and Lüderitz in Namibia to Inhaca Island in Mozambique (Broekhuysen, 1955; Dornelas et al., 2003), and even Zanzibar (Barnard, 1950). The crabs live in soft sediments, spending the day buried and emerging mainly by night to forage for food (Hill & Forbes, 1979). H. orbiculare is one of more than 90 species within 16 genera in the family Hymenosomatidae (Lucas, 1980; Ng & Chuang, 1996). Hymenosomatidae are among the smallest brachyuran crabs, their carapace widths ranging from <2 to 26 mm. The smallest species, Elamenopsis minima Lucas & Davie, 1982,

STATUS OF HYMENOSOMA ORBICULARE AUCT.

669

and Halicarcinus orientalis Sakai 1932, achieve only about 1.7 mm carapace width (Chuang & Ng, 1994). Because the crabs are small, and dull and cryptic in colour, they have generally been poorly studied. Reviews of the family have been published from New Zealand (Melrose, 1975), Australia (Lucas, 1980), and Southeast Asia (Chuang & Ng, 1994). In South Africa three species have been recorded, Hymenosoma orbiculare, Neorhynchoplax bovis (Stimpson, 1858) (previously known as Rhynchoplax bovis, or Elamenopsis bovis, cf. Lucas, 1980), and Elamena mathaei (Desmarest, 1825) (cf. Barnard, 1950). The genus Hymenosoma was established for H. orbiculare by Desmarest (1825) and currently includes two other species, H. depressum Jaquinot, 1846, and H. hodkini Lucas, 1980, from New Zealand and eastern Australia, respectively (Lucas, 1980). Although H. orbiculare is currently the only recognized species in the genus Hymenosoma found in southern Africa, its great range of morphological variation has let to suspicion that several species are in reality represented (Hill & Forbes, 1979; Lucas, 1980). A more granulose deep-water form of H. orbicluare was, in fact, originally described as H. geometricum by Stimpson in 1858 (cf. Stebbing, 1905), but later treated as a junior synonym of H. orbiculare by Stebbing (1914), on the grounds that it was within the variability of that species. H. geometricum was said to differ from H. orbiculare in its smaller size (carapace width 6-7 mm), longer rostrum, which exceeds the eye-stalks, and more granular and setose appearance (Barnard, 1950). Other characteristics of H. geometricum included two large callosities at the base of the rostrum, large and pointy hepatic projections, and a raised setiferous median boss in the gastric region (Barnard, 1950, 1955). Authors working on H. orbiculare from southern Africa’s east coast have also questioned the systematic position of their specimens (Hill & Forbes, 1979; Dornelas et al., 2003), with east coast individuals appearing to be smaller, lighter, less setose, and less granular. Hill & Forbes (1979) also noticed unusual abdominal projections on east coast individuals. In this study, we conducted a systematic analysis of H. orbiculare from sites throughout its range and used both morphological and mtDNA characters to test the hypothesis that several species are in fact represented.

METHODS AND MATERIALS

Collections An extensive series of Hymenosoma orbiculare samples was collected during 2005, ranging from Walvis Bay in Namibia to Kosi Bay in north-eastern South Africa (fig. 1). Specimens were collected by wading in estuaries, lakes, or sheltered

670

M. T. EDKINS ET AL.

Fig. 1. Map of Southern Africa indicating the sites sampled.

marine environments, and sieving the soft bottom sediment with a D-frame handnet. Deep-water samples from False Bay were collected by hand while SCUBA diving (5 m deep individuals), or with a small dredge towed behind a speedboat for several hundred meters (20 m deep individuals). Three additional preserved samples from deeper water (20 m) in False Bay were obtained from the Iziko Museum, Cape Town. As an outgroup for genetic analyses, the hymenosomatid crab Neorhynchoplax bovis was collected in the Msimbazi estuary on the east coast of South Africa. All specimens were preserved in 96% alcohol and have been subsequently deposited at the Iziko Museum. Genetic analysis Muscle tissue was obtained either from one of the chelae, or the base of the legs. Genomic DNA was isolated following the Chelex® extraction protocol (Walsh et al., 1991) and a portion of the mitochondrial cytochrome oxidase c subunit I gene (mtDNA COI) amplified using the polymerase chain reaction (PCR). Primers were designed to anneal to regions that were relatively conserved among published COI sequences of various decapods. Sequences of forward and reverse primers were CrustCOIF 5
-TCA ACA AAT CAY AAA GAY ATT GG-3
and DecapCOIR 5
-AAT TAA AAT RTA WAC TTC TGG-3
, respectively (Teske et al., 2006). Each 50 µl PCR reaction contained 5 µl of 10 × NH4 reaction buffer (Bioline), 6 mM of MgCl2 , 0.16 mM of each dNTP (Bioline), 3 pmol/µl of each primer, 0.2 µl of BIOTAQTM DNA Polymerase (5 units/µl, Bioline), and 2 µl of DNA extracts. Samples that were difficult to amplify additionally included 0.2-0.5 µl of 10 mg/ml Bovine Serum Albumin solution. The PCR profile comprised an initial denaturation step (3 min. at 94◦ C), 35 cycles of denaturation (30 s at

STATUS OF HYMENOSOMA ORBICULARE AUCT.

671

94◦ C), annealing (45 s at 45-50◦ C), and extension (90 s at 72◦ C), and a final extension step (10 min. at 72◦ C). PCR products were purified with the Wizard® SV Gel and PCR Clean-Up System (Promega), cycle sequenced, both in the forward and reverse direction, using BigDye® Terminator v3.1 Cycle Sequencing Kit (Applied Biosystems), and visualized on an ABI 3100 genetic analyser. Partial COI sequences 598 nucleotides in length were obtained. The neighbour-joining method (Saitou & Nei, 1987) was used to reconstruct phylogenetic relationships among haplotypes. The most appropriate model of sequence evolution was determined using the Akaike Information Criterion (Akaike, 1973) as implemented in the program MODELTEST version 3.06 (Posada & Crandall, 1998), and was incorporated into phylogenetic reconstructions performed using PAUP* version 4.0b10 (Swofford, 2002). The neighbour-joining tree was rooted with a COI sequence of N. bovis. Nodal support was obtained from 1000 bootstrap pseudo-replications. Morphological analysis To determine whether mitochondrial DNA lineages recovered by means of genetic analyses could also be distinguished morphologically, a number of morphological characters were chosen for analysis, following Barnard (1950), Melrose (1975), Hill & Forbes (1979), and Lucas (1980), and by closely examining the individuals. In total, 29 characters were successfully polarized (table I). Quantitative characters were gap-coded by plotting histograms of all quantitative characters, and coding the character states according to ‘identifiable gaps’ (Conlan, 1988; Stewart & Griffiths, 1995). Variations in the quantitative measures were controlled for sex dependence by running t-tests for the character values. Quantitative measurements included carapace width, carapace length, eye width, rostrum length, first and second walking leg length, first and second leg dactylus length, and cheliped propodus width, all measured in mm (fig. 2A). Although measurements of the merus and ischium of the third maxilliped were also taken, these did not polarize effectively, and were discarded from the analysis. Furthermore, the relative carapace length to carapace width did not yield any histogram gaps. Measurements were taken on individuals fixed in 96% alcohol, using a binocular microscope fitted with a graticule, or with Vernier callipers for longer parts. Leg lengths were determined by adding the measurements of each segment. Log-transformed quantitative characters were analysed using Principal Components Analysis (PCA), and the patterns recovered compared to those of the genetic analyses. Furthermore, discriminant analyses were carried out separately for all pairs of lineages identified using the genetic analyses, in order to identify characters useful to distinguish these morphologically. Forward stepwise procedures were used to select a minimum set

672

M. T. EDKINS ET AL.

TABLE I The 29 characters used for morphological analysis of individuals of Hymenosoma orbiculare Desmarest, 1825, sampled from the southern African coastline. Characters apply to both male and female individuals. Asterisks indicate morphometric characters No. Character 1 2 3 4 5 6

7 8 9 10 11 12 13

14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30

General Colour Epistome Carapace Carapace length* Carapace width* Carapace setae Granularity Eye Eye width* Walking legs 1st leg legth* 2nd leg length* 1st dactylus length* 2nd dactylus length* Denticles on dactyli Setae on legs Rostrum Rostral length* No. of rostral lobes Rostral appearance relative to eyes Size of callosities at base of rostrum Cheliped Propodus width* Inner margins of finger and thumb Gap between finger and thumb Setae on dactylus and propodus of cheliped Angles (projections) Infra-orbital tooth No. of post-ocular teeth Post-ocular tooth appearance relative to the eyes Hepatic tooth 2nd hepatic tooth Setaceous median boss in gastric region Post-branchial projection Coxae with anterior spike Abdominal projections

State (0) light; (1) mixed; (2) dark (0) absent; (1) present

(0) none; (1) a few; (2) many (0) only on carapace borders; (1) as (0) with some on gastric region; (2) on most of carapace

(0) absent; (1) present (0) a few on hind margins; (1) most hind margins covered; (2) on hind and fore margins

(0) 1; (1) 3 (0) shorter; (1) equal; (2) longer (0) small; (1) large

(0) serrated towards the ends; (1) same as (0) with large tooth mid-way along length; (2) not serrated, sometimes spatulate (0) no/small gap; (1) obvious gap (0) none; (1) only on ends; (2) all over (0) absent; (1) present (0) 1; (2) 2 (0) shorter; (1) equal; (2) longer (0) absent; (1) present, small; (2) present, tall and pointy (0) absent; (1) present, small; (2) present, tall and pointy (0) absent; (1) present (0) absent; (1) present (0) absent; (1) present (0) absent; (1) present

STATUS OF HYMENOSOMA ORBICULARE AUCT.

673

Fig. 2. Diagrams of Hymenosoma orbiculare Desmarest, 1825, indicating some of the important morphological features assessed. A, typical specimen indicating some measurements taken for the morphological analysis; B, depiction of a museum specimen collected by Barnard (1955) from False Bay (20 m) indicating some of the projections noted; C, abdomen variation in east coast individuals collected from Lake Sibaya by Hill & Forbes (1979).

of characters, beyond which the inclusion of additional characters would not improve separation of groups. A number of qualitative measures were also incorporated in the analysis, which included both categorized and presence/absence data (table I). Colour was noted

674

M. T. EDKINS ET AL.

as light, mixed, or dark (fig. 3), and the presence or absence of an epistome recorded. The setoseness and granularity of the carapace was assessed (fig. 2B). The setoseness of the walking legs was determined by averaging setae densities on each leg segment (fig. 2A). The presence or absence of denticles on the dactyli was noted. The number of rostral lobes was recorded, the rostral projection relative to the eyes was qualified, and the size of the callosities at the base of the rostrum assessed (fig. 2C). Also noted were the presence or absence of infra-orbital teeth, the number of post-ocular teeth, the size of the hepatic projections, the presence or absence of post-branchial projections, the presence or absence of anterior spikes on the coxae of the walking legs, and the presence or absence of abdominal projections (fig. 2B, C). Abdominal projections were considered to be present if a “pointy tail” was seen on females, or if two flattened posterior projections were observed on males. These measures were taken for 2-10 individuals from each location, depending on the size of the collected sample. Most samples contained both males and females, and both were included in the analysis, with a selection preference for larger individuals. Ordination plots of the qualitative data were constructed by means of multidimensional scaling (MDS). This method was considered particularly suitable for data that were on discontinuous or arbitrary scales, and for that reason was particularly suitable for categorized data (McCune & Mefford, 1997).

RESULTS

Genetic analysis All sequences generated were submitted to GenBank (accession numbers DQ351389-DQ351426 and EF198477-EF198485). Five major clusters were evident in the genetic analysis, all of which were monophyletic and strongly supported by bootstrap values of 99. On average, sequences among clusters differed by 17%, with the largest difference found between clusters C and E (21%) and the smallest difference between clusters B and C (14%). In comparison, mean differences within clusters were low (0.3%; range: 0.07% [cluster D] − 0.54% [cluster A]). The first cluster comprised large-bodied estuarine individuals from the west and south coasts, as well as marine specimens from Langebaan Lagoon and from 5 m depth in False Bay (Clade A, fig. 4). Deep-water individuals from False Bay appeared as two distinct, monophyletic clusters (Clades B and C, fig. 4), and estuarine individuals from the southeast coast and east coast formed two further monophyletic clusters (Clades D and E, fig. 4). We are confident that the sequences generated are of mitochondrial origin, rather than amplicons of a nuclear pseudogene (e.g., Williams & Knowlton, 2001; Nguyen et al., 2002) because no double

STATUS OF HYMENOSOMA ORBICULARE AUCT.

675

Fig. 3. Photographs of the five proposed South African Hymenosoma species. A, Hymenosoma orbiculare Desmarest, 1825 representative of estuarine and shallow marine habitats all along southern Africa’s west and south coast, from Walvis Bay to Gqunube Estuary; B, Hymenosoma geometricum Stimpson, 1858 from 20 m depth in False Bay; C, unnamed Hymenosoma species also from 20 m depth in False Bay; D, unnamed Hymenosoma species recorded along South Africa’s south-eastern coast, from Qolora to Mngazi estuaries; E, unnamed Hymenosoma species, representative of the eastern species, from Port St. Johns northwards to Mozambique.

676

M. T. EDKINS ET AL.

STATUS OF HYMENOSOMA ORBICULARE AUCT.

677

bands were amplified, no multiple electrophenogram peaks of equal intensity were identified in the same positions, the majority of variable sites in the ingroup sequences (81%) were in third character positions, and there were no shifts of the reading frame or stop codons. Morphological analysis The first principal component (PC1) accounted for 93% of the total variance in the quantitative data-set, and an ordination plot constructed using PC1 and PC2 recovered Clades A, B and C as distinct groups, whereas Clades D and E (smallbodied estuarine individuals from the southeast and east coast) overlapped strongly (not shown). The weights of PC1 were all positive and of similar magnitude, suggesting that it represents size and is thus not representative with regard to shape. Components PC2 and PC3 together explained 65% of the remaining variation. An ordination plot of these components recovered Clade B as a distinct cluster, and with the exception of Clade C, there was strong overlap between the remaining lineages identified using the genetic data (fig. 5). Ordination plots of the remaining principal components failed to recover any distinct groups (not shown). Discriminant analyses for pairwise comparisons of clades provided 96-100% correct identifications, and Wilk’s lambda (a measure of the level of discrimination, with 0 being perfect discrimination and 1 being no discrimination) ranged from <0.001 to 0.41 (table II). An MDS ordination plot of the first two dimensions of qualitative morphological data recovered the five clades as clusters that showed little overlap (fig. 6). The clades were, however, not recovered as clusters when the other dimensions were plotted against each other (not shown). Morphological description of clusters Individuals comprising Cluster A represent the typical H. orbiculare as described by Desmarest (1825). They have a maximum carapace width of 28 mm, are dark in appearance, have some setae and granules on their carapace, and have setose hind-margins to their walking legs, but tend to lack a number of the projections recorded on individuals from other clusters (fig. 3A). Fig. 4. Neighbour-joining phylogram constructed from HKY+I+G (Hasegawa et al., 1985) distances among COI sequences of southern African Hymenosoma specimens. The model incorporated an assumed proportion of invariable sites of 0.55 and a gamma distribution parameter of 0.72 as selected by MODELTEST. Cluster support from 1000 bootstrap replications (>50%) is shown above some branches. The tree was rooted using a COI sequence of Neorhynchoplax bovis (Stimpson, 1858).

678

M. T. EDKINS ET AL.

Fig. 5. Principal Component Analysis based on the correlation matrix of 9 log-transformed quantitative morphological characters of Hymenosoma orbiculare Desmarest, 1825. Letters A-E and geometrical symbols indicate clusters identified by means of the genetic analysis.

TABLE II Disciminant analysis for pairwise comparisons of Hymenosoma cf. orbiculare Desmarest, 1825 clusters A-E. Percent correct classification and Wilk’s lambda are given for a function of a reduced set of variables selected by a forward stepwise procedure. Partial Wilk’s lambda values are given for each quantitative character included in the analysis. The two variables that contribute most to distinguish between two clades are shown in boldface Pairwise comparison

A-B

A-C

A-D

A-E

B-C

B-D

B-E

C-D

C-E

D-E

% correctly classified Wilk’s lambda Partial Wilk’s lambda Carapace width Carapace length Eye width Rostrum length 1st leg length 2nd leg length 1st dactylus length 2nd dactylus length Propodus width

100 0.14

100 0.41

100 0.11

96 0.25

100 <0.001

100 0.04

100 0.01

100 0.05

100 0.13

100 0.15

0.80

0.99

0.66

0.61 0.34

0.58 0.69

0.48 0.94

0.97 0.97 0.68 0.85

0.93 0.44 0.45

0.83 0.84 0.66 0.64

0.96 0.97 0.61 0.84

0.80 0.75 0.64 0.92

0.80 0.66 0.76

0.72

0.97 0.67 0.78 0.94 0.84

0.95 0.98 0.96 0.86 0.87 0.84

0.01 0.01 0.60 0.03 0.42 0.29 0.08 0.15

0.31 0.90

0.12 0.90

0.70 0.77 0.99 0.96

STATUS OF HYMENOSOMA ORBICULARE AUCT.

679

Fig. 6. Multidimensional scaling ordination plot of the first two dimensions constructed from 21 qualitative morphological characters of Hymenosoma orbiculare Desmarest, 1825. Letters A-E and geometrical symbols indicate clusters identified by means of the genetic analysis.

Deep-water individuals from False Bay (Clades B and C) formed two clusters. The distinctness of Clade B was well supported using both genetic and morphological methods. This cluster was represented by pale individuals that were small in size (<10 mm) and very granular and setose in appearance. Their relative rostral lengths were far longer than those of any other specimens, and they had characteristic, large callosities at the base of their eyes. The gastric regions of their carapaces were raised into a setaceous median boss, and they had large, visible, conical postbranchial projections. This phenotype appears to represent the morph previously described as H. geometricum by Stimpson (1858) and Stebbing (1905). Clade C was morphologically more similar to Clade A, but these specimens were lighter and more granular than the typical H. orbiculare. Clades D and E represent estuarine individuals from the southeast and east coasts, respectively, and appear to represent two new, undescribed species of Hymenosoma. Both were comparatively small, with maximum carapace widths less than 10 mm, a characteristic only shared by H. geometricum. They also had very smooth carapaces, with granular beads only on the rim of the carapace, and few setae. Both had large eyes in relation to carapace size. This does not appear to be a result of their small size, since the equally small H. geometricum had relative eye widths similar to those of H. orbiculare. The walking legs of clades

680

M. T. EDKINS ET AL.

D and E were also least setose of any of the clades identified. When viewed from below, small anterior spikes were visible on the coxae of the walking legs (fig. 2C). Although the relative size of the chelipeds differed little from those in other clades, they were often less setose, and some of the larger males had an unusual tooth on the inner margin of the dactyl. Another unusual characteristic was the presence of abdominal projections, males having two flattened posterior projections and females a slight point on the third abdominal somite (fig. 2C). Walking legs of individuals from both of these clusters were relatively longer (>1.1 times the carapace width) than in H. orbiculare. The length of the second leg was identified as an important character to distinguish between Clades D and E (table II). The second leg of the south-eastern crabs was always more than 1.5 times longer than the first, whereas the eastern lineage had the second leg 1.3-1.5 times as long as the first (table I). Another morphological difference between the two lineages was that in eastern individuals the post-ocular tooth always reached beyond the eye, whereas this was not always so with south-eastern individuals. Eastern individuals were also generally lighter than the south-east coast specimens, individuals from Kosi Bay lakes appearing almost translucent (fig. 3).

DISCUSSION

By genetically and morphologically analysing Hymenosoma orbiculare samples from around southern Africa, we conclude that five distinct species are in fact represented. The original H. orbiculare (fig. 3A), as described by Desmarest (1825), ranges across all west and south coast lagoons and estuaries, from Walvis Bay to the Gqunube Estuary near East London, and extends into the shallow marine environment to a depth of about 5 m. A smaller, more setose and granulose form was recorded from deeper marine habitats in False Bay. We collected a few specimens from 20 m depth in False Bay and supplemented these with museum specimens borrowed from Iziko Museum, Cape Town. Based on genetic analyses these individuals appear to represent a valid species characterized by the morphological features mentioned by Barnard (1950), in particular the long rostrum and the unusual number of abdominal projections. This form should be resurrected as H. geometricum, based on Stimpson’s (1858) original description (Stebbing 1905). The other marine deeper-water (20 m) specimens collected in False Bay (fig. 3C) appear to represent another new species, labelled Species C. Although it clustered half-way between the first two species in morphological analyses, it was

STATUS OF HYMENOSOMA ORBICULARE AUCT.

681

not strongly distinct from the type form, and more samples need to be analysed to fully confirm its status. Interestingly, when we were dredging in False Bay, this morph was found in samples containing a few red bait, Pyura stolonifera (Heller, 1878) (Ascidiacea Pleurogona: Stolibranchiata: Pyuridae), while the H. geometricum morph was often either picked off P. stolonifera or found in dredge samples that contained this large ascidian, suggesting that P. stolonifera represents a habitat for H. geometicum, but not for the other deep-water species. The east coast samples comprise two new and very distinct species (fig. 3D, E). The geographical boundary between H. orbiculare and the first undescribed southeast coast species, labelled Species D, corresponds to a well-documented zoogeographic boundary separating the Warm Temperate South Coast and the Subtropical East Coast Provinces (Emanuel et al., 1992). Species D ranges from the Qolora to the Mngazi estuaries, and is then replaced by the other east coast species, Species E (fig. 3E). This inhabits the South African east coast north of Mzimvubu Estuary to Lake Sibaya and Kosi Bay, and probably extends further north, possibly even to Zanzibar (Barnard, 1950) and Madagascar (Ng & Chuang, 1996). The small individuals noted at Inhaca Island, Mozambique (Macnae & Kalk, 1958; Dornelas et al., 2003) probably also belong to this species. The fact that east coast individuals represent undescribed species would explain why Hill & Forbes (1979) described the zoeae from Lake Sibaya and the surroundings as having longer rostral and carapacial spines than typical of H. orbiculare. The two eastern species can be distinguished from each other by their colour, the northern species (Species E) being lighter, and by their relative leg lengths. The second leg of the south-eastern crabs (Species D) is always more than 1.5 times the length of the first, whereas in Species E the second leg is 1.3-1.5 times the length of the first. Both eastern lineages also have longer legs, relative to the carapace, than H. orbiculare. Although adult individuals of the two eastern species are superficially similar to immature H. orbiculare, the presence of gravid females make them readily identifiable as being mature. These proposed new species will be formally described elsewhere.

ACKNOWLEDGEMENTS

Our thanks to those colleagues who collected samples on our behalf, especially Bronwyn Currie (National Marine Information and Research Centre, Swakopmund, Namibia) and Brent Newman (University of Zululand); also to members of our dive team, especially Georgina Jones and Peter Southwood. Loan of material from the Iziko Museums (South African Museum) is gratefully appreciated. We also thank Henning Winker and an anonymous reviewer for comments on an

682

M. T. EDKINS ET AL.

earlier version of the manuscript. Financial support for this study was provided through a grant to C. L. Griffiths from the Sea and the Coast 2 Programme of the South African National Research Foundation. M. T. Edkins was supported by a University of Cape Town Council Honours Scholarship and Research Scholarship, and P. R. Teske’s postdoctoral research at Rhodes University was supported by the Claude Harris Leon Foundation.

REFERENCES A KAIKE , H., 1973. Information theory as an extension of the maximum likelihood principle. In: B. N. P ETROV & F. C SAKI (eds.), Second International Symposium on Information Theory: 267-281. Akademiai Kiado, Budapest. A LLANSON , B. R., B. J. H ILL , R. E. B OLTT & V. S CHULTZ , 1966. An estuarine fauna in a freshwater lake in South Africa. Nature, London, 209: 532-533. BARNARD , K. H., 1950. Descriptive catalogue of South African decapod Crustacea. Ann. South African Mus., 38: 1-836. — —, 1955. Additions to the fauna-list of South African Crustacea and Pycnogonida. Ann. South African Mus., 43: 1-107. B ROEKHUYSEN , G. J., 1955. The breeding and growth of Hymenosoma orbiculare Desm. (Crustacea, Brachyura). Ann. South African Mus., 41: 313-343. C HUANG , C. T. N. & P. K. L. N G , 1994. The ecology and biology of Southeast Asian false spider crabs (Crustacea: Brachyura: Hymenosomatidae). Hydrobiologia, 285: 85-92. C ONLAN , K. E., 1988. Phenetic and cladistic methods applied to a small genus (Corophioidea: Ischyroceridae: Microjassa) and a larger outgroup. Crustaceana, (Suppl.) 13: 143-166. D ESMAREST, A. J., 1825. Considérations générales sur la classe des Crustacés, et descriptions des espèces de ces animaux, qui vivent dans la mer, sur les côtes, ou dans les eaux douces de la France: i-xix, 1-446, IV, pls. 1-56. (Paris, Strasbourg). D ORNELAS , M., J. PAULA & A. M ACIA , 2003. The larval development of Hymenosoma orbiculare Desmarest, 1825 (Crustacea: Decapoda: Brachyura: Hymenosomatidae). Journ. nat. Hist., London, 37: 2579-2597. E MANUEL , B. P., R. H. B USTAMANTE , G. M. B RANCH , S. E EKHOUT & F. J. O DENDAAL , 1992. Zoogeographic and functional approach to the selection of marine reserves on the west coast of South Africa. South African Journ. mar. Sci., 12: 341-354. H ASEGAWA , M., K. K ISHINO & T. YANO , 1985. Dating the human-ape splitting by a molecular clock of mitochondrial DNA. Journ. mol. Evol., 22: 160-174. H ILL , B. J. & A. T. F ORBES , 1979. Biology of Hymenosoma orbiculare Desm. in Lake Sibaya. South African Journ. Zool., 14: 75-79. L UCAS , J. S., 1980. Spider crabs of the family Hymenosomatidae (Crustacea: Brachyura) with particular reference to Australian species: systematics and biology. Rec. Australian Mus., 33: 148-247. M ACNAE , W. & M. K ALK , 1958. A natural history of Inhaca Island, Mozambique: i-iv, 1-163, pls. 1-11. (Witwatersrand University Press, Johannesburg). M C C UNE , B. & M. J. M EFFORD , 1997. Multivariate analysis of ecological data. Version 3.03. (Gleneden Beach, Oregon). M ELROSE , M. J., 1975. The marine fauna of New Zealand: family Hymenosomatidae (Crustacea, Decapoda, Brachyura). Mem. New Zealand oceanogr. Inst., 34: 1-123. N GUYEN , T. T. T., N. P. M URPHY & C. M. AUSTRIN , 2002. Amplification of multiple copies of mitochondrial cytochrome b gene fragments in the Australian freshwater crayfish, Cherax destructor Clark (Parastacidae: Decapoda). Anim. Genetics, 33: 304-308.

STATUS OF HYMENOSOMA ORBICULARE AUCT.

683

N G , P. K. L. & C. T. N. C HUANG , 1996. The Hymenosomatidae (Crustacea: Brachyura) from Southeast Asia, with notes on other species. Raffles Bull. Zool., 3: 1-82. P OSADA , D. & K. A. C RANDALL , 1998. MODELTEST: testing the model of DNA substitution. Bioinformatics, 14: 817-818. S AITOU , N. & M. N EI , 1987. The neighbour-joining method: a new method for reconstructing phylogenetic trees. Mol. Biol. Evol., 4: 406-425. S NEATH , P. H. A. & R. R. S OKAL , 1973. Numerical Taxonomy: 1-573. (Freeman, San Francisco, California). S TEBBING , T. R. R., 1905. South African Crustacea. Part III. Mar. Invest. South Africa, 4: 21-123. — —, 1914. Stalk-eyed Crustaeca Malacostraca of the Scottish National Antarctic Expedition. Trans. Roy. Soc. Edinburgh., 50: 253-307. S TIMPSON , W., 1858. Prodromus descriptionis animalium evertebratorum, quae in Expeditione ad Oceanum Pacificum Septentrionalem, e Republica Federata missa, Cadwaladaro Ringgold et Johanne Rodgers ducibus, observavit et descripsit. V. Crustacea Ocypodoidea. Proc. Acad. nat. Sci. Philadelphia, 10: 93-110. S TEWART, B. A. & C. L. G RIFFITHS , 1995. Revision of the family Paramelitidae (Crustacea, Amphipoda) from South African fresh waters. Ann. South African Mus., 104: 181-247. S WOFFORD , D. L., 2002. PAUP* — Phylogenetic Analysis Using Parsimony (* and other methods), version 4.0b10. (Sinauer Associates, Sunderland, Massachusetts). T ESKE , P. R., C. D. M CQUAID , P. W. F RONEMAN & N. P. BARKER , 2006. Impacts of marine biogeographic boundaries on phylogeographic patterns of three South African estuarine crustaceans. Mar. Ecol. Progr. Ser., 314: 283-293. WALSH , P. S., D. A. M ETZGER & R. H IGUCHI , 1991. Chelex 100 as a medium for simple extraction of DNA for PCR-based typing from forensic material. Biotechniques, 10: 506-513. W ILLIAMS , S. T. & N. K NOWLTON , 2001. Mitochondrial pseudogenes are pervasive and often insidious in the snapping shrimp genus Alpheus. Mol. Biol. Evol., 18: 1484-1493.

First received 25 September 2006. Final version accepted 21 February 2007.