Xenograpsidae, N. K. Ng, Davie, Schubart & P. K. L. Ng, 2007

GUINOT, DANIÈLE, TAVARES, MARCOS & CASTRO, PETER, 2013, Significance of the sexual openings and supplementary structures on the phylogeny of brachyuran crabs (Crustacea, Decapoda, Brachyura), with new nomina for higher-ranked podotreme taxa, Zootaxa 3665 (1), pp. 1-414 : 145-211

Family Xenograpsidae

In this family represented by only one genus, Xenograpsus Takeda & Kurata, 1977 ( Takeda & Kurata 1977), the male gonopore is sternal, separated from the coxa by expanded episternite 7, which forms the lateral margin of the gonopore (N.K. Ng, Huang & Ho 2000: fig. 3B; N.K. Ng, Davie, Schubart & Ng 2007: 246, fig. 3H, 4A, X. testudinatus ; see also Jeng et al. 2004). Molecular analysis of the mitogenome of X. testudinatus , characterised by little mitochondrial genetic variation, showed that it was in accordance with its morphological systematics ( Ki et al. 2009).

Freshwater crabs families

The true freshwater crabs (see Ng 2004), those exclusively freshwater, currently known by about 1,300 species from all zoogeographical regions (except Antarctica), consist of several families, the number of which varying according to authors. The classification of Ng, Guinot & Davie (2008: 67, 159, 173, 187) is followed here for the most part: (1) Gecarcinucoidea Rathbun, 1904, with Gecarcinucidae Rathbun, 1904 , and Parathelphusidae Alcock, 1910 (included in Gecarcinucidae by Klaus et al. 2009); (2) Potamoidea Ortmann, 1896, with Potamidae Ortmann, 1896 (Potamiscinae Bott, 1970; Potaminae Ortmann, 1896), Potamonautidae Bott, 1970 (Hydrothelphusinae Bott, 1955 = Deckeniinae Ortmann, 1896; Potamonautinae Bott, 1970 = Platythelphusinae Colosi 1920); (3) Pseudothelphusoidea Ortmann, 1893, with Pseudothelphusidae Ortmann, 1893 (Epilobocerinae Smalley, 1964; Pseudothelphusinae Ortmann, 1893), and (4) Trichodactyloidea H. Milne Edwards, 1853 ( Trichodactylidae ). Various changes have been proposed since ( Klaus et al. 2009; Ng, Guinot & Davie 2010). Yeo et al. (2008), who concluded that a reasonable consensus was far to be reached for the phylogeny and classification of the freshwater families, recognised eight exclusively freshwater families: Deckeniidae , Gecarcinucidae , Platythelphusidae , Potamidae , Potamonautidae , Pseudothelphusidae , Parathelphusidae , and Trichodactylidae (see below).

All freshwater crabs are true heterotremes. The male gonopore is always exclusively coxal. Although sometimes offering a particularly striking pattern, the penis seems to have practically never been figured in detail, the rare existing illustrations only showing the location of the gonopore and/or the penial groove in which the penis may lie (e.g., Rodríguez 1992, see below; Cumberlidge 1999: fig. 7A), without indicating its precise emergence from the P5 coxa. The G1 of freshwater crabs is generally the focus of attention by taxonomists.

A trend to a coxo-sternal condition is recognised in Trichodactylidae , a variously developed penial groove being located along sternite 8 (see below), but a real penial groove seems absent in other freshwater families, even if the trace of such a groove has been exceptionally mentioned, the Potamonautidae completely lacking a groove ( Rodríguez 1992: 17, 23; Ng & Rodríguez 1995: 639; von Sternberg et al. 1999: 33).

The trichodactyloid penis consists of a thick, cylindrical, often calcified tube that prolongs into an elongated, soft tube ending in an elongated papilla, but may be completely soft, as in Rotundovaldivia latidens and Trichodactylus dentatus . The trichodactyloid penial groove shows various degrees of development ( Rodríguez 1992: 18, 23, 25, figs. 1C, E, 11B, 12, 13; see also Magalhães 2003a: fig. 87c). A groove only as a shallow depression as in Sylviocarcinus H. Milne Edwards, 1853 , represents the “rudimentary condition” ( Rodríguez 1992: 23, figs. 13A–C, 27H). A penis that is rather long, soft, with only a narrow calcified basal portion or no calcified portion, and lying in a faint groove, located either between the posterior margin of episternite 7 and the anterior margin of sternite 8 or along sternite 8, is present in Epithelphusa chiapensis , Fredius reflexifrons , F. stenolobus , Hypobolocera rathbuni , Kingsleya latifrons , Neopseudothelphusa simony , Raddaus bocourti , and Tehuana veracruzana . Sternite 8 may be raised and project over the penis in other genera, this projection becoming strong enough to form a “sternal lobe”, as in Dilocarcinus H. Milne Edwards, 1853 , which shows a thick posterior sternal lobe ( Fig. 31E ; Rodríguez 1992: fig. 13I, J). Episternite 7 may also expand posteriorly as a spur, so a penial groove is present. The sternite 8 may actually form two lobes, one anterior and one posterior, the penis being inserted between them. The episternal spur, together with the sternal lobe, partially covers the penial groove in the most derived trichodactylid condition, as in Trichodactylus Latreille, 1828 , Avotrichodactylus Pretzmann, 1965 , and Mikrotrichodactylus Pretzmann, 1965 ( Rodríguez 1992: figs. 11B, 12). Although the protection of the penis by thoracic sternites 7 and 8 varies in complexity in trichodactylids, the coxo-sternal condition is never much derived since the penis is within the groove being never completely covered and concealed.

Dissections of specimens belonging to a large number of species has revealed in some freshwater families the presence of a condylar protection of the penis, i.e., a penis enclosed within a variously enlarged, thick, and elongated coxo-sternal condyle with penis emerging from its extremity (see Modalities of penis protection: Condylar protection). The coxo-sternal condyle articulates on the sternum under the penis by an articulation point that is not dorsally visible. This disposition has been clearly figured, although not described, in the parathelphusid Sayamia sexpunctata by Rodríguez (1992: fig. 11C, as Somanniathelphusa sexpunctata (Lanchester, 1906) . Freshwater families showing a condylar penial protection include: Gecarcinucidae , exemplified by Barythelphusa jacquemontii ; Parathelphusidae , exemplified by Sendleria genuitei and Sayamia germaini ( Fig. 31B ); Potamidae ( Potamon ibericum and Pupamon prabang ; and Potamonautidae ( Deckenia imitatrix , D. mitis , Hydrothelphusa agilis , and H. madagascariensis ). The pattern is similar to that of other heteroteme families having a condylar protection (see Modalities of penis protection: Condylar protection).

The non-condylar protection of Trichodactylidae and Pseudothelphusidae was figured by Rodríguez (1992: fig. 11B, G). The Trichodactylidae is a first notable exception of a condylar protection of the penis in freshwater families. The coxo-sternal condyle may give at first sight the impression of a complete condylar protection. The trichodactylid coxo-sternal condyle may show, just anteriorly to its articulation on sternite 8, a short expansion from which emerges the penis that is proximally enclosed in a thick, calcified sheath resembling the condyle. This perhaps explains why, in the figure of Trichodactylus fluviatilis Latreille, 1828 , by H. Milne Edwards (1836–1844, Atlas , pl. 15, fig. 2d, as T. quadratus Latreille ), the long penis apparently emerging from the P5 coxo-sternal condyle without no visible coxal articulation on the sternum gives the false impression of an emergence from the condylar extremity. The trichodactylid penial pattern actually consists of a generally short coxo-sternal condyle that is obliquely directed and has a largely exposed superior border from which emerges the thick sheath enclosing penis. The condyle shows a normal, dorsally discernible articulation on sternite 8 that forms a salient, dorsally visible prominence (instead of being ventrally articulated and concealed under the penis). It has been found in all trichodactylid taxa that were examined: Trichodactylus fluviatilis , T. petropolitanus , and T. kensleyi (see Rodríguez 1992: fig. 12A, B, D, respectively), T. dentatus (see Rodríguez 1992: fig. 1E), Avotrichodactylus constrictus , Dilocarcinus pagei pagei ( Fig. 31E ), Goyazana castelnaui , Poppiana argentiniana , Rotundovaldivia latidens , Valdivia serrata and Zilchiopsis collastinensis (see Appendix I, Trichodactylidae ). The taxonomy of Dilocarcinus H. Milne Edwards, 1853 , Poppiana Bott 1969 , and Rotundovaldivia Pretzmann, 1968 , was revised by Magalhães & Türkay (2008a, b).

The Pseudothelphusidae is the second exception of a condylar protection in freshwater crabs, the P5 coxosternal condyle articulating on sternite 8 as figured in Rodriguezus garmani (Rathbun, 1898) by Rodríguez (1992: fig. 11G, as Eudaniela garmani ) and verified in many genera and species (see Appendix I, Pseudothelphusidae ). The pseudothelphusid male gonopore is located near the dorsal margin of the coxa, which forms a short expansion from which the penis exits. Thus, as in Trichodactylidae , the penis does not emerge from the extremity of the coxosternal condyle, and the articulation point of the condyle on the sternal gynglyme is clearly visible dorsally.

It is hypothesised that the male gonopore opens at the condylar extremity in all the non-trichodactylid and non-pseudothelphusid families (See Modalities of penis protection: Condylar protection) and, thus, the condylar protection of the penis could be a synapomorphy of the non-trichodactylid and non-pseudothelphusid freshwater crabs. The view of Klaus et al. (2006: 211; 2009: 509, 523; see also Yeo et al. 2008) of a single clade for all the Old World (Africa, Madagascar, Europe, Asia and Australasia) freshwater crabs, the superfamily “Potamoidea” forming potentially a monophyletic group, is supported by the character highlighted here, the condylar protection of the penis. Condylar penial protection is absent in the New World ( Mexico, Central- and South-America; São Sebastião I., South Atlantic; see Mossolin & Mantelatto 2008) Trichodactyloidea and Pseudothelphusoidea. The precise significance of the condylar protection remains unknown. A careful review of all the freshwater taxa is, however, necessary to test the consistency of the penial disposition as indicated here.

There are various hypotheses for the origin, diversification and phylogeny of freshwater crabs, being referred to as polyphyletic, archaic, and phylogenetic schools (von Sternberg et al. 1999). Authors such as Bott (1955) and Preztmann (1973) did not assume a common ancestry for the families of freshwater crabs by considering that they originated from several and diverse marine ancestors (polyphyly), and the similarities thus result from convergence. When vicariance is postulated as the key mechanism, the breakup of Gondwana being the key process, crabs share a Gondwanan origin that results in the present distribution pattern. A monophyletic origin of all families of freshwater crabs, with the exception of Trichodactylidae , was suggested by Colosi (1921) and referred to as the “archaic population hypothesis” (von Sternberg et al. 1999; Rodríguez & Magalhães 2005). According to von Sternberg et al. (1999), the Pseudothelphusidae and all the palaeotropical freshwater crabs families form a monophyletic clade, widely distributed in the southern Tethys Sea during the Cretaceous, that is sister group of Thoracotremata (grapsid ancestor) and could have given rise to the modern families after independent diversification into freshwater environments. The present global distribution pattern considered as the result of the colonisation of the tropical continental margins from a common, ancestral thoracotreme marine group cannot be subscribed here. Such a reversion from a thoracotreme to a heterotreme is most unlikely and, moreover, the thoracotreme disposition was not yet present during the Cretaceous. In the hypothesis proposing the monophyly, sister group relationships for Pseudothelphusidae + Gecarcinucioidea and Potamoidea + Trichodactylidae have been suggested ( Rodríguez 1986; Ng & Rodríguez 1995; Ng et al. 1995; Yeo et Ng 1999; see also Bănărescu 1990). A consensus seems to have been reached with the suggested mixture of “vicariance hypothesis” ( Klaus et al. 2006: 212) and “dispersal hypothesis” ( Yeo et al. 2008: 283). According to Klaus et al. (2011: 451) morphological and molecular studies point towards a monophyletic origin of the primary freshwater crabs, probably excluding the Trichodactylidae .

The Potamidae was supported monophyletic by morphological characters (Yeo 2000) and molecular data ( Shih et al. 2009), and with two major lineages, the mostly allopatric Potaminae and Potamiscinae. Their discrete distribution in Europe and Asia was hypothesised to be the result of vicariance due to the collision of the Indian tectonic plate with the Asian continent, and the orogeny, causing the separation of the two subfamilies around 22.8 Ma ( Shih et al. 2009: 703; see also Lohman et al. 2011).

The thoracic axial system of Trichodactylidae underscores the monophyly of the family (von Sternberg & Cumberlidge 2003). The Trichodactylidae , positioned as sister to Carcinus (von Sternberg & Cumberlidge 2001b: 31) , was suggested as the sister taxon to a portunoid subclade ( Rodríguez 1992; von Sternberg & Cumberlidge 2003: 24; see also Daniels et al. 2006).

New insights

The high diversity of Brachyura has led to a large number of phylogenetic hypotheses. At least two of the three higher-ranked taxa proposed by Guinot (1977a, b), Heterotremata and Thoracotremata, are widely used. A number of molecular analyses have shown that Podotremata is paraphyletic, implying that only some, but not all, of the descendants are from a common ancestor (see Hennig 1966; Nelson 1971; Guinot 1979c; Wiley 1981; Goujet & Tassy 1997; Ridley 2004; Ebach & Williams 2004; Williams & Ebach 2007a, b), or even polyphyletic, i.e., formed of separate lineages from diverse origins having convergently evolved similar characteristics. We here argue for a monophyletic Podotremata (see below). Moreover, recent studies of fossil Gymnopleura , one of the most questionable podotreme groups, bring to light the evolution of these podotremes in clarifying the relationships of different lineages ( Van Bakel, Guinot, Artal, Fraaije & Jagt 2012). A well-argued, modern classification with accurate diagnostic characters has yet to be proposed for Brachyura . There is strong support, morphological as well as genetic, for a monophyletic Eubrachyura and a monophyletic Thoracotremata (see Monophyletic Eubrachyura; Monophyletic Thoracotremata). Despite increased complexity due to the presence of sternal gonopores in the Hymenosomatoidea (see Modalities of penis protection: “Sternitreme” protection; Position of the Hymenosomatoidea within the Brachyura ), all brachyuran crabs may be considered either podotremes, heterotremes (some with a “sternitreme” disposition), or thoracotremes. The Podotremata remains, however, a more debated taxon, and relationships between the suprageneric taxa of Eubrachyura remain poorly resolved.

Monophyletic Podotremata

The Recent Podotremata, with 95 extant genera and 383 species, represent a small percentage of all Recent Brachyura , namely about 13% of the genera and 18% of the species (updated from Ng, Guinot & Davie 2008). Despite this relatively small number in the living fauna, podotremes show a considerable diversity in terms of their morphology, larval development, genetics, behaviour, ecology, and biogeography. The Podotremata contains a considerably higher number of fossil taxa, indicating the group was once much more abundant and diverse. Many became extinct across geological time. In total, the Podotremata actually consists of more than 30 families if all the true podotreme fossil taxa assigned by Schweitzer et al. (2010: 57–79) to their “ Dromiacea ” plus those incorrectly included by them in the Eubrachyura are counted. On the basis of the widespread fossil record, the extant podotremes ( Table 5) may be regarded as the surviving clades of a once richer fauna, now in decline.

The four major extant podotreme clades recognised for a long by the first two authors ( Guinot & Tavares 2001) are established here as subsections with the following nomina: Dynomeniformia nom. nov. (formerly Dromiacea ), Homoliformia Karasawa, Schweitzer & Feldmann, 2011 new status, Cyclodorippiformia nom. nov., and Gymnopleura Bourne, 1922 ( Tables 1, 5; see Appendix II), all with a fossil record (see below The four subsections of Podotremata; Extinct Podotremata). Table 5 proposes the high-ranked classification of Brachyura with emphasis on Podotremata, for which the number of taxa is indicated for the extant superfamilies, families and subfamilies; while Table 6 treats the classification of the subsection Gymnopleura including the high-ranked taxa (†Palaeocorystoidea, Raninoidea ), with the number of known taxa indicated only for extant Raninoidea .

The previous arrangement of Brachyura by Guinot (1977a, b, 1979a), revised by Guinot & Tavares (2001), proposed three major taxa: Podotremata, Heterotremata, and Thoracotremata, with Podotremata as basal. Guinot (1977a; see also Guinot et al. 1994: table 7; Guinot & Richer de Forges 1995: fig. 6) first divided the Podotremata into two subsections: Dromiacea (see Appendix II. Subsection Dynomeniformia) containing two superfamilies: Homolodromioidea and Dromioidea ; and Archaeobrachyura Guinot, 1977, which only comprised three superfamilies: Homoloidea (now attributed to H. Milne Edwards, 1837, see Appendix II. Subsection Homoliformia), Raninoidea (see Appendix II. Subsection Gymnopleura ) and Tymoloidea Alcock, 1896 = Cyclodorippoidea (see Appendix II. Subsection Cyclodorippiformia). Guinot & Tavares (2001: 507, 531, 532, fig. 15) later proposed the exclusion of Homoloidea from the Archaeobrachyura, a subsection thus only comprising Cyclodorippiformia and Gymnopleura (i.e., Archaeobrachyura emend. or redefined by Guinot & Tavares 2003). Guinot & Tavares (2003: 45) subsequently proposed a separate subsection, Homolidea (now Homoliformia). The classification of Števčić (2005) incorporates the same clades; his taxon “ Dromiacea ” is equivalent to Podotremata of Guinot (1977), including a similar diagnosis.

The Homolodromioidea , a suprafamilial rank proposed by Guinot (1993a: 1228), is known from the Jurassic and comprises several extinct families, evidence of the high diversity of the homolodromioid clade. It is represented in the living fauna by only two worldwide genera and has been considered the most ancestral brachyuran clade (e.g., Förster et al. 1985). Homolodromioids seem to have retained the ancestral appearance as they display many plesiomorphic features, which include the endoskeletal interlacement; the vestigial pleopods on male abdominal somites 3–5; the absence of a linea along which the carapace breaks open during ecdysis and its replacement on the branchiostegite by a poorly calcified area; the abdominal pleura; the cephalic organisation (see Cephalic condensation). It is true that the moult of homolodromioids has not yet been described to our knowledge ( Ng, Guinot & Davie 2008: 15), but is very likely that a local resorption of the branchiostegite permits exuviation.

Section Podotremata Guinot, 1977

Subsection Dynomeniformia nom. nov.

Superfamily Dromioidea De Haan, 1833 (46 genera; about 150 species)

Family Dromiidae De Haan, 1833 (41 genera; 128 species)

Subfamily Dromiinae De Haan, 1833 (37 genera; about 114 species)

Subfamily Hypoconchinae Guinot & Tavares, 2003 (1 genus; 6 species)

Subfamily Sphaerodromiinae Guinot & Tavares, 2003 (3 genera; 8 species)

Subfamily †Basinotopinae Karasawa, Schweitzer & Feldmann, 2011

Subfamily †Goniodromitinae Beurlen, 1932

Family Dynomenidae Ortmann, 1892 (5 genera; 22 species)

Subfamily Acanthodromiinae Guinot, 2008 (1 genus; 2 species)

Subfamily Dynomeninae Ortmann, 1892 (2 genera; 10 species)

Subfamily Metadynomeninae Guinot, 2008 (1 genus; 4 species)

Subfamily Paradynomeninae Guinot, 2008 (1 genus; 6 species)

Subfamily † Graptocarcininae Van Bakel, Guinot, Corral & Artal, 2012 (2 genera)

Superfamily †Glaessneropsoidea Patrulius, 1959

Family † Glaessneropsidae Patrulius, 1959

Family † Lecythocaridae Schweitzer & Feldmann, 2009

Family † Longodromitidae Schweitzer & Feldmann, 2009

Family † Nodoprosopidae Schweitzer & Feldmann, 2009

Family † Viaiidae Artal, Van Bakel, Fraaije, Jagt & Klompmaker, 2012

Superfamily Homolodromioidea Alcock, 1900 (2 genera; 24 species)

Family Homolodromiidae Alcock, 1900 (2 genera; 24 species)

Family † Prosopidae von Meyer, 1860

? Family † Bucculentidae Schweitzer & Feldmann, 2009

? Family † Tanidromitidae Schweitzer & Feldmann, 2008

Incertae sedis. † Diaulacidae (or †Diaulacinae) Wright & Collins, 1972; † Torynommatidae Glaessner, 1980 ;? † Konidromitidae Schweitzer & Feldmann, 2010 ; † Xandarocarcinidae Karasawa, Schweitzer & Feldmann, 2011 Subsection Homoliformia Karasawa, Schweitzer & Feldmann, 2011 new status

Superfamily Homoloidea H. Milne Edwards, 1837 (see Appendix II) (17 genera; 74 species)

Family Homolidae H. Milne Edwards, 1837 (see Appendix II) (14 genera; 67 species)

Family Latreilliidae Stimpson, 1858 (2 genera; 7 species)

Family Poupiniidae Guinot, 1991 (1 genus; 1 species)

? Family † Mithracitidae Števčić, 2005

Subsection Cyclodorippiformia nom. nov.

Superfamily Cyclodorippoidea Ortmann, 1892 (18 genera; 91 species)

Family Cyclodorippidae Ortmann, 1892 (10 genera; 49 species)

Subfamily Cyclodorippinae Ortmann, 1892 (7 genera; 38 species)

Subfamily Xeinostomatinae Tavares, 1992 (3 genera; 11 species)

Family Cymonomidae Bouvier, 1897 (5 genera; 38 species)

Family Phyllotymolinidae Tavares, 1998 (3 genera; 4 species)

Subsection Gymnopleura Bourne, 1922

Superfamily †Palaeocorystoidea Lǒrenthey in Lǒrenthey & Beurlen 1929

Superfamily Raninoidea De Haan, 1839 (12 genera; 46 species) (see Table 6)

Incertae sedis. †Etyoidea Guinot & Tavares, 2001 (Family † Etyidae Guinot & Tavares, 2001 ), † Dakoticancroidea Rathbun, 1917 (Families † Dakoticancridae Rathbun, 1917 , and † Ibericancridae Artal, Guinot, Van Bakel & Castillo, 2008 )

Section Eubrachyura Saint Laurent, 1980

Subsection Heterotremata Guinot, 1977 (see Tables 3, 4)

Subsection Thoracotremata Guinot, 1977 (see Table 7)

Although retaining a plesiomorphic organisation, the Dromioidea (uropods generally modified in dorsal plates; specialisation of the uropods as locking structures to hold the abdomen in Dromiidae ; endoskeletal fusion; cephalic condensation incomplete; long spermathecal tubes in Dromiinae ) is more derived than Homolodromioidea (uropods as ventral lobes; male pleopod formula always complete; endoskeleton unfused with connection by interdigitations; cephalic condensation lacking or weak; spermathecal tube short). The Hypoconchinae , undeniably a peculiar dromiid sublineage, and mainly the Sphaerodromiinae , which both share several plesiomorphies with Dynomenidae and Homolodromiidae , were assumed as more early branched off within Dromiidae ( Guinot & Tavares 2003: 108) . The elevation of Sphaerodromiinae to a family rank by Schweitzer & Feldmann (2010c), made in the absence of a coherent taxonomic frame, is not supported by any putative familial level of generality, i.e., in comparison with Dromiidae , Dynomenidae , and Homolodromiidae , and is justified only at a less inclusive subfamilial rank, i.e., in comparison with Dromiinae and Hypoconchinae (see Guinot & Tavares 2003, key p. 120). A new hierarchical rank cannot be a subjective category, even for the benefit of integrating a large number of fossil taxa, its introduction requiring a phylogenetic frame in conformity to the level of generality of characters in related groups. A family Sphaerodromiidae , as advocated by Karasawa, Schweitzer & Feldmann (2011: 539, 543) and used by Schweitzer & Feldmann (2012), is not accepted here. Also unacceptable is a familial rank for the † Basinotopidae Karasawa, Schweitzer & Feldmann, 2011 , which actually warrants only a subfamilial status within the Dromioidea , i.e., †Basinotopinae new status (see below; Table 5; Appendix II; †Basinotopinae new status).

The Homoliformia is a distinct basal clade with an unfused skeleton (only endoskeletal interdigitations, see Drach 1959, 1961; Pichod-Viale 1966; Secretan 1982, 1998) and in which a new type of abdominal-locking mechanism (homoloid press-button system) has evolved. Larval evidence suggests the Homoliformia as probably the most basal podotreme ( Williamson 1976).

Superfamily †Palaeocorystoidea Lǒrenthey in Lǒrenthey & Beurlen 1929 (see Van Bakel, Guinot, Artal, Fraaije &

Jagt 2012), extinct

Family † Camarocarcinidae Feldmann, Li & Schweitzer, 2008

Family † Cenomanocarcinidae Guinot, Vega & Van Bakel, 2008

Family † Necrocarcinidae Förster, 1968 (see Van Bakel, Guinot, Artal, Fraaije & Jagt 2012)

Family † Orithopsidae Schweitzer, Feldmann, Fam, Hessin, Hetrick, Nyborg & Ross, 2003

Family † Palaeocorystidae Lǒrenthey in Lǒrenthey & Beurlen 1929

*Superfamily Raninoidea De Haan, 1839 (12 genera; 46 species), fossil and extant

Family Lyreididae Guinot, 1993 (2 genera; 6 species)

Subfamily Lyreidinae Guinot, 1993 (see Van Bakel, Guinot, Artal, Fraaije & Jagt 2012)

Subfamily † Marylyreidinae Van Bakel, Guinot, Artal, Fraaije & Jagt, 2012

Family Raninidae De Haan, 1839 (see Van Bakel, Guinot, Artal, Fraaije & Jagt 2012) (10 genera; 40 species)

Subfamily Cyrtorhininae Guinot, 1993 (1 genus; 2 species)

Subfamily Notopodinae Serène & Umali, 1972 (4 genera; 17 species)

Subamily Ranininae De Haan, 1839 (1 genus; 1 species)

Subfamily Raninoidinae Lǒrenthey in Lǒrenthey & Beurlen 1929 (4 genera; 17 species)

Subfamily Symethinae Goeke, 1981 (1 genus; 3 species)

The Gymnopleura ( Table 6) is a very diverse major clade, with the living fauna consisting of only 46 species in contrast with the 196 exclusively fossil species quoted by De Grave et al. (2009: table 1, as Raninoida). The number of 37 genera (fossils included) that can be deduced from the list of Schweitzer et al. (2010) is too low since several palaeocorystoids († Necrocarcinidae Förster, 1968 , and † Orithopsidae Schweitzer, Feldmann, Fam, Hessin, Hetrick, Nyborg & Ross, 2003 ) have been included in the Eubrachyura and thus were not counted. There are approximately 25 palaeocorystoid genera, all extinct ( Van Bakel, Guinot, Artal, Fraaije & Jagt 2012; see also Luque et al. 2012).

One major event in the evolution of Podotremata is the appearance of a paired spermatheca, often with a displacement of its aperture in an anterior direction ( Figs. 2 , 7 ; Gordon 1950, 1963; Hartnoll 1979; Tavares & Secretan 1993; Guinot & Tavares 2001, 2003; Tavares & Franco 2004; Guinot & Quenette 2005). The basic anatomical organisation of the spermatheca is constant throughout podotremes. It is always derived from two adjacent thoracic somites, a split between the plates of the intersegmental phragma 7/8, one derived from sternite 8 and the second from sternite 7. The paired podotreme spermatheca always involves the same two somites ( Gordon 1950; Hartnoll 1975, 1979; Mori 1986a; Tavares & Secretan 1993; Guinot & Quenette 2005). This is a general rule that so far has no exceptions, in spite of several modalities in shape, size, and in the exact location of the spermathecal apertures on the sternal surface ( Guinot & Tavares 2001: fig. 10; Garassino 2009: pls. 1–16). The development in an anterior direction of sternal suture 7/8, leading to the forward displacement of the spermatheca, is shared synapomorphically by all the Dromiinae and by some cyclodorippines such as Neocorycodus stimpsoni ( Tavares 1996a: fig. 23D). A shorter spermathecal tube has a more restricted level of generality ( Nelson 1978: 339; Wiley 1981: 126) and is shared by only two subfamilies of Dromiidae (Sphaerodromiinae and Hypoconchinae) and by Dynomenidae .

The proximity to medial axis of both spermathecae in most Raninoidea is assumed to be a derived condition, and not an unpaired structure as in Nephrops ( Secretan-Rey 2002: 102, figs. 19, 25). The spermatheca pattern varies in Raninoidea : small spermathecal apertures facing each other on opposite sides of a depression (‘sunken pit’) of thoracic sternite 7, and separated by the median line ( Lyreididae ); small spermathecal apertures at the bottom of an elongated, deep depression on the anterior part of thoracic sternite 7 ( Ranininae De Haan, 1839, and Notopodinae ); small spermathecal apertures lying at the bottom of a deep, pit-like depression ( Raninoidinae ); small, contiguous spermathecal apertures recessed in a depression ( Cyrtorhininae ); and large, widely separated spermathecal apertures not recessed, horizontally oriented, and overhung by two calcified hoods ( Symethinae ).

The peculiar raninoid spermatheca is the result of the strong modifications related to the complete body linked to burying (see Concealment behaviour: Burying in the Raninoidea ), the same major transformation that, for instance, has led to the exposure of several thoracic pleurites (gymopleurity) ( Fig. 38B, C ) in contrast to the normally covered pleurites of †Palaeocorystoidea ( Fig. 38A ) (see below). Hartnoll (1979: 82, fig. 5) tried to postulate a pathway by which “the dromiid spermatheca could have been evolved into the raninoid one” to explain the evolution of the raninoid spermatheca beyond the condition found in the other Podotremata. There is, however, no evidence for a sequence by which these raninoid apertures on sternite 7, lying in the endosternite 7/8 as in all other podotremes, could be moved to sternite 6 and become the vulvae of Eubrachyura. Moreover, the vulvae (sternal gonopores) are homologous with the appendicular gonopores (as in raninoids) but not with the spermathecae. Dissection of Ranina ranina has shown that suture 7/8 is partially internalised and is long, continuing “within” the median plate. Through narrowing and infolding of the posterior sternites, the spermathecal apertures became “trapped” within sternite 7 and thus do not represent secondarily acquired openings ( Van Bakel, Guinot, Jagt & Fraaije 2012: 160, fig. 60). In this regard, the raninoid spermathecae is somewhat reminiscent of the spermathecal tubes of Dromioidea , the apertures of which are at the end of extended sutures 7/8 and have even acquired an anterior location in Dromiinae , supposedly to improve the efficiency of fertilisation ( Tavares & Franco 2004). The spermathecal apertures of the Raninoidea are the original apertures that have become displaced.

The presence of a spermatheca at the extremity of suture 7/8 is visible in the basal podotreme, † Basinotopus tricornis Collins & Jakobsen, 2004 , a remarkably preserved crab from the Eocene of Denmark (see Collins & Jakobsen 2004: 69, fig. 3, pl. 2, figs. 1–7). † Basinotopu s M’Coy, 1849 (type species: † Inachus lamarckii Desmarest, 1822 ) was traditionally assigned to Dromiidae ( M’Coy 1849; Bell 1858; Collins 2003: 83; Collins & Jakobsen 2004; Beschin et al. 2005: 15, pl. 2, fig. 9), then referred to Dynomenidae ( De Grave et al. 2009: 27; Schweitzer et al. 2010: 65). In the paratype female of † B. tricornis from Denmark (MGUH 26776, cast examined by the first author) there is a clearly visible gonopore on the P3 coxa, the thoracic sternite 7 laterally forms a rim that overhangs sternite 8, and the short sternal suture 7/8 bears a small distal spermathecal aperture concealed below the rim, thus located at the level of the female gonopore ( Fig. 37 ). The possibility that † B. tricornis is an immature female dromiine, with not yet advanced sutures 7/8 (see Tavares & Franco 2004), is improbable because of the size of the specimen and the well formed gonopore on the P3 coxa ( Collins & Jakobsen 2004: pl. 2, fig. 1b). The dromioid nature of † B. lamarckii ( Desmarest, 1822) can be indicated by the condition of the sternal sutures 7/ 8, the small spermathecal apertures, the presence of dorsal uropods, and the morphology of the last pereopods ( Bouchard 2000: fig. 21A). As in † B. lamarckii , the dorsal location and reduction of both P4 and P 5 in † B. tricornis (P4 and P5 partially present, with thin, elongated meri) reject the possibility of being a dynomenid representative (short sternal sutures 7/8 and only P5 reduced; see Guinot 2008). Both † B. lamarckii (male) and † B. tricornis (female) show a similar thoracic sternum, with well-developed episternites 4 and 5, and a sternite 4 with a markedly concave anterior margin. That † Basinotopu s could belong in Homolodromiidae on account of similar short sternal sutures 7/8 (see Guinot 1995; Guinot & Tavares 2003; Ng & Naruse 2007b; Schweitzer & Feldmann 2008) is rejected since the uropods show as ventral lobes in extant homolodromiids, instead of dorsal uropods as in † B. lamarckii . It should be noted, however, that the body of homolodromioids is often markedly elongated and prolonged by a rostrum, as in † Basinotopus , although with a distinctive shape in † Basinotopus and homolodromiids. † Basinotopus tricornis is not a typical Dromiinae , a subfamily where female sternal sutures 7/8 are long and the spermathecal apertures usually extend beyond the gonopores on the P3 coxae. The two other dromiid subfamilies, Hypoconchinae and Sphaerodromiinae , have short female sternal sutures 7/8 and spermathecal apertures that are positioned posteriorly ( Guinot & Tavares 2003). But † Basinotopu s should not be included in Hypoconchinae , obviously with different carapaces ( Guinot & Tavares 2003; see also Hendrickx 1977; Melo & Campos 1999) and many other distinctive characters such as abdomen shape. An assignment to Sphaerodromiinae , known by only four extant species (see figs. in McLay & Crosnier 1991; McLay 1991, 1993; Crosnier 1994; Cleva et al. 2007; see also Van Bakel, Artal, Jagt & Fraaije 2009), is not supported by the carapace’s shape and marked areolation. The inclusion of † Basinotopus in the sphaerodromiine tribe Frodromiini Števčić, 2005 , known by only one extant genus and species, Frodromia atypica , is rejected here. The status of † Basinotopus has been partially solved thanks to the establishment of the † Basinotopidae Karasawa, Schweitzer & Feldmann, 2011 ( Karasawa, Schweitzer & Feldmann 2011: 539), in the proximity of their “ Sphaerodromiidae ” (see above), a familial ranks we disagree with. A subfamilial rank within the Dromiidae , †Basinotopinae new status, is proposed here (see Table 5; Appendix II. †Basinotopinae new status), besides Sphaerodromiinae , Dromiinae and Hypoconchinae . A new diagnosis is under investigation (unpublished).

Additional evidence for a paired spermatheca similarly linked to thoracic sternal suture 7/8, shared synapomorphically by extinct as well as extant Podotremeta, has been found in several other fossil podotreme taxa: †Etyoidea ( Guinot & Tavares 2001: figs. 2, 3, 10J, as Etyidae ), † Dakoticancridae Rathbun, 1917 ( Guinot 1993a: figs. 7, 8; Guinot & Tavares 2001: fig. 10H; Artal et al. 2008: fig. 3D), † Ibericancridae Artal, Guinot, Van Bakel & Castillo, 2008 ( Artal et al. 2008: 17), † Cenomanocarcinidae ( Guinot, Vega & Van Bakel 2008: 719, Addenda; Van Bakel, Guinot, Artal, Fraaije & Jagt 2012: fig. 58), and † Palaeocorystidae ( Van Bakel, Guinot, Artal, Fraaije & Jagt 2012: figs. 57, 59).

The Podotremata displays several modalities of penis protection: a penial tube in the Dromiidae (Dromiinae and Hypoconchinae); a coxal protection in the Dynomenidae , Homolodromiidae , and Sphaerodromiinae ; a coxal protection combined with abdominal protection in the Raninoidea and Cyclodorippoidea ; a coxal protection thanks to a sternal setose ridge in the Homoloidea (See Modalities of penis protection: Penial tube; Coxal protection; Coxal protection combined with abdominal protection; Sternal setose ridge).

The transformation of the biramous uropods into calcified dorsal plates, a synapomorphy of Dromiidae ( Guinot & Tavares 2003: table 1) and Dynomenidae ( McLay 1999; Guinot 2008: fig. 5), is likely to represent a specialised, functionally adapted structure. The presence of uropods (as dorsal plates or lobes) and a similar internal organisation of the spermatheca are robust synapomorphies of the Dromiidae-Dynomenidae-Homolodromiidae lineage forming the subsection Dynomeniformia (see Appendix II. Subsection Dynomeniformia). The modification of the biramous uropods into ventral lobes, without apparent functionality, occurs in Homolodromioidea , Hypoconchinae , and only in a few taxa of Dromiinae ( Guinot & Bouchard 1998: table 3; Guinot & Tavares 2003: 110, table 1; Poore 2004: 303, 306, fig. 87d, e; Guinot & Quenette 2005). In contrast with the other podotremes ( Dromioidea , Cyclodorippiformia, Gymnopleura ) characterised by skeletal connections by fusion as in Eubrachyura, both Homolodromioidea and Homoliformia share an axial skeleton with connections by interdigitation, that is, the plesiomorphic condition (see Evolution of the axial skeleton in the Podotremata). These two groups show, however, a number of differences including the spermatheca (chamber, bulb, and spermathecal tube in Homolodromioidea ; chamber directly opening to the exterior and with membranous areas in Homoliformia, see Garassino 2009), uropods (as ventral lobes in Homolodromioidea , as sockets in Homoliformia), and abdominal-holding system (holding by the appendages in Homolodromoidea; holding by the appendages and/or homoloid press-button system in Homoliformia). The skeleton of the Dromiidae-Dynomenidae lineage with connections by fusion has been derived independently from that of Eubrachyura (see Evolution of the axial skeleton in the Podotremata). The sella turcica as redefined here as the structure formed by the endosternal intertagmal phragma that connects the tagma/thorax and the tagma/abdomen to thoracic interosternite 7/8 (“brachyuran sella turcica”) is present in all podotremes, with diverse modalities, however (see Axial skeleton; Evolution of the axial skeleton in the Podotremata). The Dynomenidae displays carcinisation in one of its four subfamilies, Dynomeninae (Guinot 2008; Artal et al. 2008), as does Latreilliidae in Homoliformia, both Dynomeninae and Latreilliidae showing an enlargement of the thoracic sternum (see Carcinisation and its outcomes).

The distinction of four dynomenid subfamilies proposed by Guinot (2008) was essentially based on extant genera, taking into account sternal and abdominal characters and, tentatively, the dorsal grooves of the carapace. These four sublineages recognised in living faunas must, at least partially, be distinguished in the rich fossil record of Dynomenidae , which dates from the Jurassic. These four taxa were, however, deleted from a list of extant and living genera ( De Grave et al. 2009: 8) for the reason that “the additions of these subfamilies would result in too large a number of unplaced fossil genera”, and similarly were “not recognized” in a systematic list of fosssil taxa (Schweitzer et al. 2010: 65). A listing of taxa should be free from an a priori selective treatment, all the more so since other fossil taxa in this list are cited as “unplaced at subfamily level” and are incertae sedis. In principle and deontologically, insufficient knowledge does not justify the suppression of taxonomic names from a list, the aim of which warrants an exhaustive treatment. The incertae sedis taxa provide partial and inconsistent information due to the poor proportion of the known structures, the insufficiency of character states, and intrusion by extensive homoplasies ( Bird 2007). Placement of incertae sedis taxa should be based on hypotheses of homology. It is therefore normal to choose to label taxa incertae sedis instead of guessing as to their placement and to list fossil taxa (as well as Recent taxa) as incertae sedis until the main characters are identified and a consensus as to how they relate to other taxa is found. Furthermore, it was rather easy to assign a subfamilial rank to some extinct dynomenid genera, and Guinot (2008: 1, 21) clearly stated that two genera, † Kierionopsis Davidson, 1966 , and † Kromtitis Müller, 1984 (at least some of their members), seemed obviously conform to Paradynomeninae Guinot, 2008 . † Kierionopsis , included in Dromiidae by Armstrong et al. (2009: 748, fig. 3.6–3.8) and Schweitzer et al. (2010: 64), was transferred to Dynomenidae by De Grave et al. (2009: 27). † Kromtitis , considered in close proximity of Paradynomen e Sakai, 1963, by Beschin et al. (2007: 26, 27), has been referred to Paradynomeninae by Collins (2010: 13, 14), whereas † K. koberiformis Beschin, Busulini, De Angeli & Tessier, 2007 , because of its close resemblance to Paradynomene was included in Dynomenidae by Tessier et al. (2011: 216). It is normal that the “short soft setae which accentuate unevenness of carapace forming transverse troughs” ( McLay 1999: 516) in the extant Metadynomeninae , as Metadynomene tanensis (see Ng & McLay 2010: fig. 4B, C) or M. tuamotu Ng & McLay, 2010 ( Ng & McLay 2010: fig. 4A; Poupin 2010: fig. p. 5), cannot be found in fossils. The arguments by Schweitzer & Feldmann (2009a: 358) that these troughs correspond, according to Guinot (2008), to “just depressions in the pilosity of hairs” is a misinterpretation. The undulations of the tomentum covering the carapace may also correspond to the shape of the underlying grooves, as shown in some hairy extant dromiids.

It is comprehensible, perhaps inevitable that, for the present, the late Jurassic (Tithonian) † Cyclothyreus Remeš, 1895 “cannot be accommodated within any of the existing dynomenid subfamilies or for that matter, within any existing family within the Dromioidea ” ( Schweitzer & Feldmann 2009a: 357). Assignment of † Cyclothyreus to † Goniodromitidae Beurlen, 1932 , within Homolodromioidea ( De Grave et al. 2009: 28; Schweitzer et al. 2010: 58) and, contradictorily, to Dynomenidae sensu lato within Dromioidea ( Schweitzer & Feldmann 2009a: 357, 359), reveals a probable confusion between all these primitive groups. On the other hand, the † Goniodromitidae as diagnosed by Schweitzer & Feldmann (2008) is clearly paraphyletic. A clear demarcation between the two families Dynomenidae and † Goniodromitidae , even with a key only using the known dorsal features, was unfortunately not provided by these authors. It is true that several transversely ovate, as well as smooth or grooved, dynomenid carapaces cannot be reliably assigned for the moment, the features being for now insufficiently informative and sometimes even misleading. The difficulty in finding a dynomenid subfamily to accommodate † Cyclothyreus may have several reasons: lack of data (in particular of ventral configuration), misinterpretation, or inadequate understanding of the organisation of earliest (Jurassic) Dynomeniformia. Nevertheless, due to the well-preserved ventral morphology of a new species of † Graptocarcinus Roemer, 1887 , from the Upper Cretaceous of northern Spain, a new subfamily † Graptocarcininae Van Bakel, Guinot, Corral & Artal, 2012 , has been established in Dynomenidae , encompassing the Middle-Upper Cretaceous † Graptocarcinus (as type genus) and the Jurassic † Cyclothyreus (see Van Bakel, Guinot, Corral & Artal 2012; see also Subfamily † Graptocarcininae ). Conclusions based on Jurassic crabs must be thus cautious and made only after careful consideration, particularly because their incomplete preservation lacks substantial information, even if the earliest fossils do provide the sole basis for our understanding of the early brachyuran organisation. However that may be, integration of fossil taxa in studies to test the status of Podotremata remains among the most fascinating current challenges.

All Podotremata share some types of concealment, either by carrying behaviour (Dynomeniformia: Fig. 53A, B ; Homoliformia: Fig. 53B, C ), carrying behaviour combined with burying (Cyclodorippiformia: Fig. 53E ), or only by burying ( Gymnopleura ) (see Concealment behaviour). Wicksten’s (1986a: 368) statement that “except for the Dorippidae [ Palicidae must be added], all families that carry are podotremes” could be updated by saying: “except for Dynomenidae , all podotremes carry or bury”. Carrying behaviour has presumably been lost in Dynomenidae (at least in the extant forms), which, despite the peculiar P5 and behaviour, is close to Dromiidae by several morphological characters of adults (presence of uropods, spermathecal organisation, axial skeleton architecture) and larvae ( Rice 1981c). Some derived Dromiidae and some Cyclodorippiformia tend to lose carrying behaviour (see Concealment behaviour: Carrying behaviour). We agree with Wicksten (1986a: 367) that the evidence from behaviour conflicts with the placement of Dromiidae in the Anomura , the elaborate dromiid carrying behaviour being absent in other Decapoda . The combination of burying and carrying performed by both cyclodorippoids and dorippoids raises some questions on the affinities between the two groups, if the behaviour is the result of convergence or by common ancestry ( Wicksten 1986a: 368).

It is noteworthy to observe that basal crabs appear to depict a “human face” on the carapace dorsal surface with conspicuous markings that correspond to the marking points of internal muscle attachments (sigilla). The “eyes” of the human face are due to the oblique V-shaped insertion of the attractor epimeralis muscle (i.e., the tergopleural muscle, connecting the carapace to the upper surface of internal pleurites), at the level of the branchiocardiac groove ( André 1939; Abrahamczik-Scanzoni 1942: 354, figs. 50, 63; Renaud 1977: 577, fig. 4; Glaessner 1933: 184, fig. 3; 1960: 36, fig. 16; 1969: R408, fig. 224). The H-shaped depression is formed by the attractor epimeralis muscle apodeme, the posterior gastric muscle scar, the median part of the cervical groove, and the branchiocardiac groove (M.D. Crane 1981: 5, fig. 2). The transformation of the cylindrical cephalothorax of the long-bodied decapods (e.g., Homarus ), which have a much elongated attractor epimeralis muscle, to a shortened conical shape in Brachyura leads to the forward shifting of the posterior end of this muscle insertion. The internal portions of the V-shaped muscle attachment, which form conspicuous lateral gastrocardiac markings, remain after the loss of the branchiocardiac groove. The consequences of this transformation are more marked on the carapaces of the primitive brachyurans, in which, moreover, the limit between the cephalic arch (arceau céphalique of H. Milne Edwards 1851: 10) and the scapular arch (arceau scapulaire of H. Milne Edwards) is generally deeply defined. This is the cervical groove, which may be recognised by the median gastric pits marking the attachment of the gastric muscles. Muscles tend to divide into fiber bundles that make particular markings on the carapace surface, varying in position relative to the cervical groove. Fossil crabs with such marks are, e.g., † Basinotopus lamarckii ( Desmarest, 1822) ( M’Coy 1849: fig. a, footnote p. 168; Bell 1858: pl. 5, figs. 1, 2) (see Appendix II. †Basinotopinae Karasawa, Schweitzer & Feldmann, 2011 new status) and † Dromiopsis elegans Reuss, 1859 ( Reuss 1859: pl. 4, fig. 1), a dynomenid according to Schweitzer et al. (2010: 65). “A V-shaped mass of wedgeshaped pits” scars the region around the gastric pits in † Palaeocarpilius aquilinus Collins & Morris, 1973 (Collins & Morris 1973: 286, pl. 30, fig. 1). A “H-shaped depression” is well marked in the hexapodid † Goniocypoda quaylei Crane, 1981 ( Crane 1981: 12, 19, fig. 7). The presence of a deep V-shaped groove on the carapace (which deserves a detailed comparative study) is a conservative character and may be considered a relict.

A carapace with a “human face” exists in extinct families, for instance in the † Dakoticancridae and in the podotreme representatives of † Diaulacidae Wright & Collins, 1972 , and † Torynommatidae Glaessner, 1980 (see Extinct Podotremata). Such a “face” exists in some fossil eubrachyurans, e.g., † Archaeopus vancouverensis ( Bishop 1986b: fig. 10C, D) and † A. mexicanus ( Schweitzer et al. 2002: fig. 17). A “human face” is a feature present in living podotremes ( Dromiinae , Homolidae , some cyclodorippoids) as well as in basal heterotremes, such as Dorippoidea (see Wang 1927; Neuville 1938; Holthuis & Manning 1990: 76, frontispiece), the inachoidid Paulita (see Guinot 2012), and Corystes (Corystidae) .

Is the Podotremata monophyletic? The monophyly of Podotremata has been seriously questioned. It is one the most important issues in brachyuran systematics, not only for elucidating the early brachyuran differentiation, but also for understanding the evolution of Brachyura . The debate is linked to two questions; whether the monophyletic Eubrachyura is nested within the podotremes or whether Podotremata and Eubrachyura form reciprocally monophyletic clades.

The views on brachyuran relationships, often contradictory, are mirrored in various recent classification schemes ( Martin & Davis 2001; Števčić 2005; Ng, Guinot & Davie 2008; De Grave et al. 2009; Schweitzer et al. 2010). The two podotreme superfamilies Cyclodorippoidea and Raninoidea were included in Dromiacea , thus considered to be basal, in the classification of living and fossil crabs by De Grave et al. (2009), but in Eubrachyura in the list of fossil species by Schweitzer et al. (2010) a year later. Schweitzer et al. (2010) did not find sufficient morphological evidence in the brachyuran fossil record and preferred to follow the molecular data and views of Ahyong et al. (2007: fig. 1) arguing for paraphyletic Podotremata (see also Ahyong et al. 2011: 186). Recent investigations on the earliest brachyuran fossils, based on structures other than the dorsal carapace (e.g., the thoracic sternum, gonopores, spermatheca, abdominal holding devices), have nevertheless shown the close relationships between all the podotreme lineages (see below and Van Bakel, Guinot, Artal, Fraaije & Jagt 2012).

Views against the monophyletic status of Podotremata can be grouped into three lines of arguments: (1) the Podotremata was originally based by Guinot (1977a, b) on a symplesiomorphy (e.g., Rice 1980; von Sternberg & Cumberlidge 2001a); (2) the spermatheca defined by Tavares & Secretan (1993; see Tavares 2003; Tavares & Franco 2004; Guinot & Bouchard 1998; Guinot & Tavares 2001, 2003; Guinot & Quenette 2005; Guinot, Vega & Van Bakel 2008; Van Bakel, Guinot, Artal, Fraaije & Jagt 2012) is unsuitable for phylogenetic purposes (e.g., Ahyong & O’Meally 2004; Ng, Shih, Tan, Ahyong & Ho 2009; Scholtz & McLay 2009); (3) lack of evidence from both molecular and morphological data other than the spermatheca (e.g., Spears et al. 1993; Brösing 2002, 2008; Brösing et al. 2002, 2007; Ahyong et al. 2007; Ahyong, Naruse, Tan & Ng 2009; Chu et al. 2009b). The analytical methods for inferring phylogeny, and therefore for grouping taxa, that have been used may have employed the same terms, but with a considerable difference in the philosophical background they assume. For instance, the reconstruction of the brachyuran phylogenetic “tree” of Scholtz & McLay (2009: 419) was obtained “by hand and brain following a Hennigian approach”. That is to say, use of a priori polarisation (a priori distinction between plesiomorphic and apomorphic states), character quality (a priori weighting), and a priori elimination of characters when ambiguous or contradicting the tree. As a result, there are no competing phylogenetic hypotheses because the analysis frequently ends up with only one tree. Unfortunately, the most parsimonious cladogram for ten taxa, for instance, has to be searched among 2,027,025 possibilities ( Felsenstein 1978).

It is no longer in dispute that a natural classification should contain only monophyletic groups, whose members (including the same common ancestor and all its descendants) are bound together by common ancestral relationships that they do not share with any other taxa. Monophyletic groups, or clades, are discovered by finding synapomorphies. After homology (conjunction, similarity, congruence) was equated with synapomorphy ( Patterson 1982), homologies were perceived as hypotheses that need to be tested and, perhaps, falsified. The initial hypothesis ( Guinot 1977a, b) that the shared possession of both appendicular male and female gonopores is a synapomorphy for the Podotremata proved to be inaccurate because that character has been used at the wrong level of generality (Podotremata instead of Malacostraca). It is still a homology but at a much higher (inclusive) hierarchical rank. It is plesiomorphic at the Podotremata rank. The discovery of a synapomorphy for the Podotremata, such as the paired spermatheca ( Tavares & Secretan 1993), gave support to the monophyletic status of the group. It should be obvious from this that homologies are continuously being tested, and that Podotremata should not be dismissed as a monophyletic group only because it was originally recognised by a symplesiomorphy. Conversely, the Podotremata should be abandoned if it cannot be justified by at least one synapomorphy. It is not important when a synapomorphy is added.

Ahyong & O’Meally (2004) addressed the phylogeny of the reptant decapods by simultaneous analysis of three molecular loci as well as morphology. Whereas the paraphyly of Podotremata was suggested, they cautioned that “greater taxonomic sampling is required to adequately test the status of Podotremata”. Their data matrix included 105 morphological characters. Relevant to our discussion is character 26 (seminal receptacle: “medial”, “absent”, or “paired”). “Paired seminal receptacles” is considered to be a synapomorphy of Brachyura , whereas “in other reptants, the seminal receptacle lies on the sternal midline”. Ahyong & O’Meally (2004: 691) avoided using the term “paired spermatheca” because of “its specific reference to the podotreme seminal receptacles”, and, as a result, two very different notions were confused, namely, the sternal vulvae and the spermathecae, introducing a false homology. Equating sternal vulvae and spermatheca induced them to consider that all brachyurans have the same kind of seminal receptacle and, therefore, regarding the paired seminal receptacles as synapomorphic for brachyurans.

In contrast, Scholtz & McLay (2009: 418) acknowledged that the spermatheca “is restricted to podotrematan representatives”. Surprisingly, however, they considered that “it suffers from a problematic polarization because nothing comparable exists in other reptant groups”. Moreover, unpaired spermathecae are well documented, for instance in Nephropidae ( Tavares & Secretan 1993; Secretan 2002; Ahyong & O’Meally 2004) and Cambaridae Hobbs, 1942 ( Hartnoll 1979; Rod & Babcock 2003; Ahyong & O’Meally 2004). Scholtz & McLay (2009: 418) stated that the “seminal receptacle [sternal vulvae] and spermathecae may not be homologous structures, so the derivation of one from the other (see Hartnoll 1979) is difficult”. We fully agree. Indeed, two states that co-occur in the same individual, spermatheca and female gonopore whether coxal or sternal, cannot be homologous ( Freudenstein 2005). Because the appendicular and sternal female gonopores are homologous, they can never co-occur in the same individual. Our argumentation has always been based precisely on the fact that both the female vulva and the spermatheca cannot be homologous: the vulva opening on sternite 6 as a skeletal invagination (and not a split between two adjacent phragmae) and connected to the gonad through the oviduct; the spermatheca being derived from the split of the intersegmental phragma 7/8 and never connected to the gonad. Scholtz & McLay (2009: 431) argued just the opposite: “the eubrachyuran condition [sternal vulvae] might be derived from that found in podotrematan groups [spermatheca]”. Actually, this conjecture was originally introduced by Hartnoll (1979: 82), who placed that possibility in serious doubt. In Hartnoll's own words (the italics are ours):

“It is certainly difficult to postulate a method whereby spermathecae initially lying within endosternite 7/8, as in the Raninidae [just like in any other Podotremata], came to open on sternite 6, as in higher Brachyura . It is possible that if endosternite 6/7 was reduced, endosternite 7/8 could extend even further forwards than in Raninidae , and fuse with sternite 6 instead of sternite 7. The spermathecae could then acquire new openings on the sixth sternite, and would be in a position to 'capture' the oviducts that pass through that segment. A subsequent reduction in dorsal flexion of the posterior thorax could conceal the origin of the spermathecae from endosternitre 7/8. Unfortunately there is no evidence for this sequence, and in many of the higher Brachyura the endosternite 6/7 is well developed, as are median apodemes 6 and 7. It is possible though that the extensive development of these structures is not a primitive character in the higher Brachyura , for they do show extensive reduction in the extant lower Brachyura , but that may be merely a specialized feature. It is perhaps most likely that the raninids are not on the direct line leading to the higher Brachyura , and that their spermathecal structure is not relevant to the origin of that of the higher forms ”.

Relevant to this discussion (though not mentioned by Scholtz & McLay 2009), is Hartnoll’s (1979: 82) advocacy of a paired spermathecae within endosternite 7/8 and opening on sternal suture 7/8 as “a common spermathecal structure for all early Brachyura ”, i.e., the Podotremata. The spermatheca is understood as having been derived early at some point in the stem, in the evolution of Brachyura , and then retained with a similar groundplan and only subsidiary modifications by all podotremes.

Another argument is the presence of “normal” spermathecae located at the extremities of sutures 7/8, separate and not cryptic in †Palaeocorystoidea ( Fig. 39C, D ), which presents more plesiomorphies than its sister group Raninoidea , as evidenced by the evolutionary series that links palaeocorystoids, all extinct, to the living raninoids (see below and Van Bakel, Guinot, Artal, Fraaije & Jagt 2012).

Ng, Shih, Tan, Ahyong & Ho (2009: 16, fig. 5) considered, without an explanation, that the “internalization of the spermatheca is therefore an innovation in the stem brachyuran [sic], but not a synapomorphy supporting Podotremata”, only “a feature retained by successive podotreme clades and then lost with derivation of the eubrachyuran synapomorphies”. Recognition of a synapomorphy, in the sense of secondary homology ( De Pinna 1991; Wägele 1996; Brower & de Pinna 2012), actually goes well beyond our willingness to accept it or not. The belief of derivation of one group to the next (i.e., one taxon can give rise to another), sometimes giving rise to more than one descendant is “acknowledged to be beyond proper scientific investigation” ( Kitching et al. 1998: 168).

Phylogenetic reconstruction orders synapomorphies into a nested hierarchy by choosing the arrangement of taxa that accounts for the distribution of characters requiring the least number of changes or steps, whether it be a gain or loss. Hennig (1966) stressed the importance of accessing the holomorphology (total form) of the semaphoront (“character bearer”), i.e., “any specimen from a given population and of a given sex, stage and age, that bears an indefinite number of characters” ( Dubois 2006a: 256). McLaughlin et al. (2004: 184) used the term semaphoront to refer to the various stages in the life history of an organism, “to indicate the assemblage of individuals of a monophyletic group at an identifiable and comparable period in their life cycles”. The comparisons should involve the holomorphology of comparable semaphoronts. A given character (or character complex) is usually informative only at specific levels of generality (a particular node in the cladogram), and the analysis may show that adult males, for instance, differ but juveniles do not. This is why the deliberate use of restricted morphological sources (as opposed to holomorphology) for phylogenetic inference, such as the foregut-ossicle system (Brösing 2002, 2008; Brösing et al. 2007), imposes tremendous limitations. Under such conditions, one should not be surprised to experience difficulties to recover monophyly for a number of group, i.e., holophyletic groups ( Ashlock 1971). The same obviously applies to molecular data.

Scholtz & McLay (2009: 421) claimed that “the paired spermathecal openings lead into an unpaired atrium” in the Raninidae by quoting Gordon (1963: 51), who had erroneously interpreted the raninoid spermatheca as unpaired. They were therefore not able to recognise the more significant synapomorphy of the Podotremata, a paired spermatheca, clearly acknowledged by Hartnoll (1979), Tavares & Secretan (1993), Guinot (1993), Guinot & Quenette (2005), Ng, Shih, Tan, Ahyong & Ho (2009), and eventually by McLay & López Greco (2011).

The classification of Števčić (2005: 15, 135) reflects for the most part the information provided by the first two authors (e.g., Guinot 1995; Guinot & Tavares 2000, 2001, 2003), but he used “section Dromiacea ” in place of section Podotremata, while recognising the same superfamilies ( Cyclodorippoidea , Dromioidea , Homolodromioidea , Raninoidea ) and families. Some authors (e.g., Dawson & Yaldwyn 2000; Dawson 2002: 3; Števčić 2005) were reluctant to accept the demise of Podotremata and the reallocation of some podotreme families to Eubrachyura as in the classifications of Martin & Davis (2001) or Schweitzer et al. (2010).

Claims that Podotremata is paraphyletic (Ahyong et al. 2007; Brösing et al. 2007; Ng, Shih, Tan, Ahyong & Ho 2009; Scholtz and McLay 2009; Bracken et al. 2009; De Grave et al. 2009) or even polyphyletic ( Spears et al. 1993) miss the point that there is only evidence for monophyly (synapomorphy). Paraphyly is not a discovery, and “Paraphyly and polyphyly are rhetorical statements—explaining why something is not there is not informative and does not contribute to our knowledge” ( Ebach & Williams 2004: 116, see also Williams & Ebach 2007a, b). The hypothesis of paraphyly only arises due to the lack of evidence of monophyly.

In the cladistic analysis based on small subunit nuclear ribosomal RNA sequences by Ahyong et al. (2007) Podotremata was paraphyletic, Cyclodorippoidea being identified as sister group to Eubrachyura. As a result, Ahyong et al. (2007: 584) proposed three podotreme clades, the “sections Dromiacea , Raninoida and Cyclodorippoida alongside [a non-podotreme] section Eubrachyura”. They admitted, however, that the inclusion of these three taxa within Eubrachyura could be “counterproductive” and rendered the eubrachyuran clade “meaningless with respect to the degree of structural organisation of the heterotreme-thoracotreme assemblage” (Ahyong et al. 2007: 584). A similar scheme, illustrated by a cladogram, was adopted by Ng, Shih, Tan, Ahyong & Ho (2009: 16–18, fig. 5) and Ahyong, Naruse, Tan & Ng (2009: 28) in a study of the podotreme crabs from Taiwan, and by De Grave et al. (2009: 5, 7, 26) in the classification of living and fossil Decapoda . The proposed suppression of Podotremata (Ahyong et al. 2007; Ahyong, Naruse, Tan & Ng 2009; Ng, Shih, Tan, Ahyong & Ho 2009: 16, fig. 5; see also De Grave et al. 2009: 5, 7, 8, table 1; Schweitzer et al. 2010: 57) is unjustifiable, particularly the avoidance of the paired spermatheca evidently shared by all podotremes. The conclusion of Ahyong et al. (2007) was reached mainly by mapping a morphological trait (paired spermatheca) on a molecular tree. That tree recovery was based on molecular evidence only, and the double spermatheca trait was mapped onto the tree afterwards, for concordance or discordance only, and played no role in tree recovery. As the morphological trait (double spermatheca) was discordant from the “true tree” (the molecular tree), it was dismissed as a synapomorphy supporting a monophyletic group (Podotremata). One often overlooked tenet is that “[m]orphological synapomorphies and homoplasies can only be discovered by morphological and combined analyses” ( Assis & Rieppel 2011: 94). In a study of the seminal receptacles and sperm storage in Brachyura McLay & López Greco (2011: 400) admitted that the argument in favour of the paraphyly of Podotremata as in Ahyong et al. (2007) and Scholtz and McLay (2009) resulted “in a number of difficulties” and contemplated the eventuality of a monophyletic Podotremata.

A section Podotremata comprising four subsections (Dynomeniformia, Cyclodorippiformia, Gymnopleura and, in addition, Homoliformia) is proposed herein, alongside a section Eubrachyura (see Classification and nomenclatural ranks; Appendix II; Tables 1, 5, 6). The nomen of the subsection Raninoidia, preliminarily diagnosed by Guinot, Vega & Van Bakel (2008: 681, 712), recognised by Osso-Morales et al. (2011: 249, 250) and recently highlighted by Van Bakel, Guinot, Artal, Fraaije & Jagt 2012), is replaced here by the nomen Gymnopleura (see Appendix II).

We consider untenable the scheme proposed by Brösing et al. (2007: 27) alluding to a “reverse evolution of the gonopores towards a coxal position” in raninids and cyclodorippoids or “a convergence of the sternal position of the female gonopores” (vulvae) in all families of Eubrachyura.

The recent first application of two combined nuclear protein-coding genes ( Tsang et al. 2008: 366, fig. 2, table 1) recovered three brachyuran clades, with monophyletic Podotremata as basal and including Raninoidea . This monophyly was, however, weakly supported: the tree was based on only one of the two genes, and the taxonomic sampling was limited. The Cyclodorippoidea was not included in this analysis, and only one member of the diverse Raninoidea was sequenced, with Ranina ranina consequently found related to the latreilliid Eplumula phalangium .

Chu et al. (2009b: 95, fig. 3), using only two protein-coding genes to sequence five species of podotremes and 30 of eubrachyurans and dismissing strong morphological evidence, concluded to the paraphyly of Podotremata; conversely, the monophyly of Podotremata seemed to be recovered when these five podotremes were analysed along with other reptantian groups ( Chu et al. 2009b: fig. 1).

The cladistic analysis of Karasawa, Schweitzer & Feldmann (2011), which attempted to include exclusively extinct (20 podotreme families) and extant brachyuran taxa (14 families, 12 of which podotremes, and two eubrachyurans), was unfortunately based on a data matrix with a great number of unknown character states (since information on ventral characters is lacking in many fossil crabs) and on oversimplifications. Moreover, some statements are approximate, even erroneous, in their character matrix, as e.g.: “most podotremes with a narrow thoracic sternum, but some homoloids, some dakoticancrids, all cyclodorippoids, and eubrachyurans have a relatively wide thoracic sternum” [a widened thoracic sternum characterises † Dakoticancridae ; basal Eubrachyura, such as Atelecyclidae , Cancridae , Corystidae , Thiidae , have a narrow thoracic thoracic sternum]; spermatheca “united” in Symethidae Goeke, 1981 , and Raninidae , “paired” in other podotremes [a paired spermatheca is present in all podotremes, even when the openings are close to each other]; penial tube of the P 5 male coxa present within only dromiids [a character until now never observable in fossils; a penial tube is also present in Dorippoidea ]; “most brachyurans lack pleopods on the pleonal segments, while the homolodromiids, most dromiids, sphaerodromiids, and dynomenids have the pleopods on pleonal segments 3-5 in males” [many Dromiidae , see Guinot & Tavares 2003, table 1, and all Brachyura lack pleopods on male somites 3–5] ( Karasawa, Schweitzer & Feldmann 2011: 527, 528, 529, table 2). Such a cladistic analysis in order to test if wether or not the podotreme spermatheca is a synapomorphy by examining phylogeny brings doubt on its conclusions.

On the other hand, several proposals of Karasawa, Schweitzer & Feldmann (2011) are not supported by recently published ( Van Bakel, Guinot, Artal, Fraaije & Jagt 2012) palaeontological data or ongoing work (Van Bakel et al. unpublished). Future studies may provide data for the understanding of optimal results. Karasawa, Schweitzer & Feldmann 2011: table 1) rectify, however, some affiliations proposed in the list of De Grave et al. (2009: 31) (e.g., † Necrocarcinidae and † Orithopsidae included in Dorippoidea ), and in the list of Schweitzer et al. (2010: 70, 78, 80, 81) where several groups of podotremes (e.g., Raninoidea , including † Palaeocorystinae , and Cyclodorippoidea ) and eubrachyurans were mixed together (see Guinot, Vega & Van Bakel 2008; Van Bakel, Guinot, Artal, Fraaije & Jagt 2012).

Since molecular analyses (nuclear and mitochondrial rDNA, two nuclear protein-coding genes) produced contrasting results, future applications using more genes and more taxa are welcome. The monophyly of Podotremata is assuredly disputable when the morphological features of the extant taxa (the wide thoracic sternum of Cyclodorippiformia for example) are considered. A high support, however, is the fact that a paired spermatheca, consistently with apertures at the extremities of sternal sutures 7/8 and found across a wide range of taxonomic ranks in extant families, has been discovered in fossil taxa, even in extinct ones from the Cretaceous: † Cenomanocarcinidae , † Dakoticancridae , † Ibericancridae , †Palaecorystoidea, and †Etyoidea.

Schweitzer Hopkins et al. (1999: 79) previously assigned † Xanthosia Bell, 1863 , and † Etyus Leach in Mantell, 1822, to Xanthidae “for several reasons”. In a subsequent revision, Schweitzer, Feldmann, Franṭescu & Klompmaker (2012: 129, 130, 138, 144, fig. 2.3, 2.4) recognised the presence of spermathecae in † Etyidae and thus the podotreme nature of the family, but suggested that the spermathecal apertures were not always positioned along sternal suture 7/ 8 in †Etyoidea. They considered that, whereas the aperture was truly located at the extremity of suture 7/ 8 in † Etyus , the aperture was “positioned entirely on sternite 7” and “apparently not as part of sternal suture 7/8” in a newly described etyoid genus († Steorrosia Schweitzer, Feldmann, Fran ṭescu & Klompmaker, 2012). It was also uncertain whether the aperture was located at the axial end of sternite 7 or showed as an “enlargement of suture 7/8” in another genus († Sharnia Schweitzer, Feldmann, Fran ṭescu & Klompmaker, 2012). The two published figures of the spermatheca in † Steorrosia simply show that the large aperture substantially encroaches on sternite 7 but remains, however, positioned at the extremity of suture 7/8. It is in fact less than clear. In the photograph of † Sterrosia pawpawensis ( Schweitzer Hopkins, Salva & Feldmann, 1999), in which the two spermathecal apertures are dissimilar, the left aperture clearly opens at the end of the suture 7/8 (Schweitzer, Feldmann, Franṭescu & Klompmaker 2012: fig. 2.3). Another incomprehensible and puzzling fact is that the configuration of † Steorrosia , suspected to be distinct from that of other etyids by these authors (2012), is the same as in † Etyus martini ( Guinot & Tavares 2001: figs. 2, 3, 10), in which the spermathecal aperture at the distal end of suture 7/8 is in most part situated on sternite 7. In other words we do not identify any remarkable difference between † Steorrosia and † Etyus (all the more that the spermathecal aperture seems relatively more elongeate in † E. martini than in † Steorrosia ). Besides, the discussion of Schweitzer, Feldmann, Franṭescu & Klompmaker (2012: 153, 154) on the spermathecae that “extend into a tube” is only relevant for Dromioidea , in particular for Dromiidae , in which most often a long spermathecal tube is present (and also for Homolodromiidae and Dynomenidae , both with a shorter tube), but does not apply to other podotremes ( Cyclodorippoidea , Raninoidea ). The structure of the spermatheca is well known in Homolidae thanks to Gordon (1950) (see also Guinot and Quenette 2005; Garassino 2009), who showed that the spermatheca communicates directly with the exterior by an aperture surrounded or lined by membranous areas that partly belong to sternite 7. The spermathecal configuration of Raninidae as understood thanks to the observations of Hartnoll (1979) and dissections by Van Bakel, Guinot, Artal, Fraaije & Jagt 2012 (160, fig. 61; “the spermathecal apertures become ‘trapped’ within sternite 7”, see above) appears to be quite distinct from that of † Steorrosia . The spermatheca (at least its aperture) of †Etyoidea needs more thorough studies before concluding that it is not formed by two adjacent somites (as in all other Podotremata), namely the endosternal phragma 7/8.

The thoracic sternum of males, not laterally visible (because it is covered by the abdomen), is only anteriorly exposed in †Palaecorystoidea and †Etyoidea, whereas a narrow thoracic sternum is exposed between the coxae and the abdomen in † Ibericancridae , and a wide sternum is exposed in † Dakoticancridae and Cyclodorippiformia. A sharper definition of the spermatheca instead of other types of seminal receptacles allowed for the recognition of the paired spermatheca as a feature exclusive of the Podotremata.

The taxon Podotremata was not used in the lists of fossil Brachyura of De Grave et al. (2009), Karasawa, Schweitzer & Feldmann (2009), and Schweitzer et al. (2010) as these authors argued that it was “artificial” or paraphyletic (see below). These lists show major discrepancies of opinion concerning the recognition and separation of the lineages, the assignment of several podotreme taxa and the constituent genera of different families, so that the number of fossil podotreme genera and species cannot be actually evaluated according to the lists. In any event, podotremes (a terminology now largely used) do exist and cannot be mixed with nonpodotremes. One of these obvious contradictions is the case of † Necrocarcinidae and † Orithopsidae , podotremes ( Tables 4, 5; Guinot, Vega & Van Bakel 2008; Van Bakel, Guinot, Artal, Fraaije & Jagt 2012) that were merged by Schweitzer, Feldmann, Fam, Hessin, Hetricks, Nyborg & Ross (2003) within Dorippoidea , which are heterotremes.

The opposite instance is the inclusion of † Binkhorstia Noetling, 1881 (heterotreme) in Cyclodorippoidea (podotreme) (see Monophyletic Podotremata: Extinct Podotremata, discussion on † Binkhorstia ). The consideration of two separate taxa, Dromiacea and Podotremata, by Armstrong et al. (2009: 747, 748), is a misinterpretation. The merging of Homolidae and Latreilliidae with Dromiacea (Ahyong et al. 2007: 582, fig. 4; Ahyong, Naruse, Tan & Ng 2009; Ng, Shih, Tan, Ahyong & Ho 2009: fig. 5; De Grave et al. 2009; Schweitzer et al. 2010) by not taking into account the distinctive morphology of these two basal podotreme clades, is a typical example of the failure to give due recognition to fundamental morphological traits such as the axial skeleton, the thoracic sternal pattern, and spermatheca ( Drach 1950, 1959, 1971; Secretan 1983, 2002; Guinot & Quenette 2005; Garassino 2009).

A few remarks on Archaeobrachyura Guinot, 1977, are deemed necessary. The Archaeobrachyura was established to group the podotremes with uropods not showing as ventral lobes or dorsal plates, i.e., Homoloidea , Cyclodorippoidea , and Raninoidea ( Guinot 1977b: 1051) . The group was recognised as such by Bowman & Abele (1982: 23), Förster & Mundlos (1982: 156), Schram (1986: 307), Tirmizi & Kazmi (1991: table 1), Collins & Wienberg Rasmussen (1992: 16), Konishi et al. (1995), Campos (1996: 163), Ingle (1996: 41, 47, 72; 1997: 17), Fransen et al. (1997: 81), Jamieson & Tudge (2000: 51), Anger (2001: 31), Chen & Sun (2002: 133), and Yang et al. (2008: 763). As the sockets of Homoloidea (as those of other Brachyura ) were later interpreted as modified uropods by Guinot & Bouchard (1998), Guinot & Tavares (2001: 531, 532) thus restricted the Archaeobrachyura to the two superfamilies Cyclodorippoidea and Raninoidea , the sockets being retained, however, in the raninoid living family Lyreididae . The recent discovery in †Palaeocorystoidea (sister to Raninoidea ) of an abdominal-locking mechanism in the form of a “double peg” on sternite 5 (central or lateral; sometimes located on a short projection) in addition to the presence of a socket in lyreidids (corresponding to a hook with two teeth on sternite 5) could be an indication that †Palaeocorystoidea belongs in Archaeobrachyura. The Archaeobrachyura emend. applies to both †Palaeocorystoidea and Raninoidea , i.e., the subsection Gymnopleura (see Appendix II; Subsection Gymnopleura : Raninoidea and †Palaeocorystoidea). It is clear that a similar locking-abdominal mechanism is shared by palaeocorystoids and raninoids, remaining present in the living and fossil lyreidids but lost in all other raninoid living families, which are modified, being more specialised. The oxystome condition, if not an homoplasy at this level, could be considered a synapomorphy of Archaeobrachyura emend., whereas both P3 and P4 with modified distal articles is a synapomorphy of Raninoidea + †Palaeocorystoidea ( Fig. 41 ).

The modification of uropods into sockets for abdominal-locking mechanism in Homoliformia, Raninoidea ( Lyreididae only), and †Palaeocorystoidea, versus lost in Cyclodorippiformia, is shared homoplasically with Eubrachyura. The position of the sternal complementary part of the abdominal-locking mechanism, i.e., a button matching a socket, is found on the sternite 5 in Gymnopleura and Eubrachyura. The only exception is the Homoliformia, where the button involves sternite 4. This location can be regarded as a synapomorphy at the homoloid level, homoloids retaining a long male abdomen inserted between the two mxp3, placing the abdominal somite 6 on the anterior part of the thoracic sternum, thus at sternite 4 level. The male abdomen is shorter in most Dromioidea . Guinot & Tavares (2001: fig. 16) argued in favour of the monophyly of the Podotremata but did not address the relationship between “ Dromiacea ”, Homoloidea , and Archaeobrachyura (Cyclodorippiformia and Gymnopleura ) due to a paucity of synapomorphies.

The simple and unique structure of the eubrachyuran press-button mechanism as well as the double peg and hook are structures that are invariably located on sternite 5. The socket in Raninoidea (Lyreididae) as well as in †Palaeocorystoidea, which is shared with Homoliformia and Eubrachyura, could be explained by assuming independent appearances of the socket (homoplasy). The socket was lost in Cyclodorippiformia. This structure may have originated three times, as proposed in the cladogram presented here ( Fig. 41 ): twice (for Homoliformia and Gymnopleura , respectively) in (monophyletic) Podotremata and once in Eubrachyura (see below).

The Archaeobrachyura redefined by Guinot & Tavares (2001), comprising the Raninoidea and Cyclodorippoidea , has been recognised by some neontologists (e.g., Dawson & Yaldwyn 2000: 49; Poore 2004: 290, 291). The subsection Raninoida of Martin & Davis (2001: 74) is synonym of Archaeobrachyura redefined by Guinot & Tavares (2001). The taxon Archaeobrachyura was rejected by Ahyong et al. (2007: 582) because of the lack of a synapomorphy.

Raninoidea and Cyclodorippoidea share several traits such as the shape of suture 4/5 and modified mouthparts ( Tavares 1992b) that could be a similar convergence to burying. Spermatozoal ultrastructure supports, however, Raninoidea and Cyclodorippoidea as a monophyletic clade, “unnamed” by Jamieson & Tudge (2000: 48, 51; see also Jamieson et al. 1994a, b) and corresponding in fact to Archaeobrachyura redefined by Guinot & Tavares (2001) (= Archaeobrachyura emend.).

The four subsections of Podotremata. The pattern of podotreme monophyly assumed here recognises the section Podotremata, comprising four main clades as subsections (Dynomeniformia, Homoliformia, Cyclodorippiformia, Gymnopleura ) alongside the section Eubrachyura ( Tables 1, 5, 6; see Appendix II). It is beyond the scope of this study discussing the interrelationships among the podotremes in detail. Podotreme interrelationships remain largely unknown, thus the cladogram presented here ( Fig. 41 ) is tentative (see the other possible schemes, below).

Subsections Dynomeniformia, Homoliformia, and Cyclodorippiformia. The subsection Dynomeniformia is restricted to families Dromiidae , Dynomenidae and Homolodromiidae , and cannot be expanded to receive the homoliform clade. Homoloidea has long been associated with “ Dromiacea ” (H. Milne Edwards 1837a; De Haan 1839; Bouvier 1896), and many authors continue to associate the two ( Gordon 1950; Balss 1957; Glaessner 1969; Hartnoll 1975; Števčić 1981; Bishop 1986b; Martin & Davis 2001; Dawson 2002; Ahyong et al. 2007; Ahyong, Naruse, Tan & Ng 2009; Ng, Shih, Tan, Ahyong & Ho 2009; De Grave et al. 2009; Schweitzer et al. 2010; Ahyong et al. 2011). It is suggested here that the three extant homoloid families ( Homolidae , Latreilliidae , Poupiniidae ) must be grouped in subsection Homoliformia (see Appendix II), which is basal in Podotremata as is Dynomeniformia. Morphological, spermatological, larval data, and palaeontological data support the proposal that Homoliformia forms a monophyletic group, whereas no characters lend support to the hypothesis of monophyletic “ Dromiacea ” grouping Dromioidea + Homoloidea . The homoloid families should be removed from Dynomeniformia as already done by Guinot & Tavares (2001: fig. 16) (as Homoloidea and Dromiacea , respectively). All the morphological characters support a homoliform clade: a dorsal, paired linea homolica on the carapace of Homolidae , a lateral paired linea only at the level of the long “neck” in Latreilliidae , and absence of any lineae in Poupiniidae , the two latter families having a carapace that does not envelops the body ( Guinot 1991; Guinot & Richer de Forges 1995); cephalic region ( Pichod-Viale 1966; see Cephalic condensation); axial skeleton ( Drach 1950, 1971; Guinot 1979a; Secretan 1983, 1998, 2002; Guinot & Quenette 2005); thoracic sternum; spermatheca ( Guinot & Quenette 2005; Garassino 2009); G1 and G2 ( Garassino 2009; Naruse & Richer de Forges 2010); general shape and large size of the male abdomen; absence of uropods as well as dorsal plates and ventral lobes; additional locking of the abdomen by the “homoloid press-button” system (prominence on sternite 4; presence of a socket, homologous to the uropod ( Guinot & Tavares 2001; Guinot & Bouchard 1998; Bouchard 2000); and the subchelate/chelate ending of the reduced, mobile P5 ( Fig. 53C ), except for Poupiniidae , which has a unique P5 pattern ( Guinot & Richer de Forges 1981 b, 1995; Guinot 1991). The monotypic tribe Latreillopsini Števčić, 2011 , established for Latreillopsis Henderson, 1888 ( Henderson 1888: 21) is not retained here because the diagnosis is not based on differentiating characters and therefore the new taxon is not justified.

Scholtz & McLay (2009) argued for a separate status of Homoloidea . The spermatozoal ultrastructure of Homoliformia (at least of Homolidae ) shows two convincing autapomorphies, demonstrating a monophyletic entity ( Jamieson et al. 1993b; Jamieson 1994; Guinot et al. 1994; Jamieson et al. 1995; Jamieson & Tudge 2000). The sperm of Latreilliidae is unique, hypothesised as the most plesiomorphic among podotremes or as “secondary simplified” ( Jamieson & Tudge 2000: 48). The monophyly of the assemblage Homolodromiidae-Dromiidae- Dynomenidae (Dynomeniformia) is supported by a “distinctive dromiacean spermatozoal ground plan” ( Jamieson & Tudge 2000: 40; see also Jamieson et al. 1993a, 1994c; Guinot et al. 1994, 1998).

Larval features also suggest that Dromiidae and Homolidae should be clearly separated ( Williamson 1965). Rice (1980: 293; see also Rice 1981a, b; Rice & Provenzano 1970) concluded that Homoloidea diverged from the primitive brachyuran line in an early stage and recognised a “close alliance” between the homolids and the raninids.

A particular synapomorphy of Homoliformia should be taken into account: suture 6/7 forms a wide, low, continuous arch enclosing sternite 8 and laterally overlapping episternite 7. It corresponds to a marked invagination of phragma 6/7, its median part internally showing as a tubular ridge with weak lumen ( Gordon 1950: 232, figs. 13, 16, 18, 20, 21, 22A; Hartnoll 1975: figs. 2A, 7A; Guinot & Bouchard 1998: fig. 9C; Guinot & Quenette 2005: figs. 19, 20; Garassino 2009: figs. 1, 2). Sternite 8 is folded twice: in addition to its dorsal folding, sternite 8 doubles up with regard to the median axis of the thorax. This external longitudinal line that medially divides sternite 8 corresponds to the median junction of the two symmetrical parts of sternite 8. These two parts are in contact, either along its entirety (e.g., Homola Leach, 1815 , Paromola Wood-Mason in Wood-Mason & Alcock 1891, Moloha Barnard, 1947 ) or partially (e.g., Homolomannia Ihle, 1912 ). The abdomen is inserted in the resulting space, the sterno-abdominal notch, when sternite 8 is completely separated into two parts (e.g., Latreillia Roux, 1830 , Eplumula Williams, 1982 ; see Castro et al. 2003). There is only a very short longitudinal bipartition of sternite 8 in Poupiniidae , resulting in a deep sterno-abdominal notch. A similar sterno-abdominal notch may occur in Homolodromiodea ( Guinot & Richer de Forges 1995: figs. 16C, 32c, d, 33A). Scholtz & Richter (1995: 316, fig. 21E) interpreted this incomplete fusion of both lateral parts of the posterior sternites as plesiomorphic, being derived from the divided sternum of the Fractosternalia Scholtz & Richter, 1995, a large group “including the Astacida, the Thalassinida and the Meiura”. Another synapomorphy of Homoloidea is a penis lodged in an excavation limited by a setose ridge of sternite 8 ( Fig. 12 ; see Modalities of penis protection: Sternal setose ridge).

The subsection Dynomeniformia as well as Homoliformia pro parte ( Homolidae and Poupiniidae but not Latreilliidae ) share a previously neglected character referred to as “sterno-coxal depression” by Guinot (1995) and Guinot & Bouchard (1998). Sterno-coxal depressions are laterally excavated at variable levels on each side of the thoracic sternum, allowing the pereopod coxae to slide inside them when moving to hold the abdomen against the ventral surface (Guinot 2008: 4, 14, 15, fig. 4A). Sterno-coxal depressions characterise virtually all podotremes that use their appendages to hold the abdomen: they probably provide additional help in the forward movements of the pereopods, a locomotion in a forward direction being found in basal brachyurans (see Locomotion). Sternocoxal depressions are well developed and are completely filled by the moveable coxae in Homolodromiidae (P1– P4 level), in Dynomenidae (P2–P4 level), and in Dromiidae (P2–P3 level) in which the coxae perfectly slide inside these depressions ( Fig. 2 ). They are deeper and proportionally larger in Homolodromiidae , allowing the legs to move forwards. In Homolidae and Poupiniidae the coxae move only on the external surface of the sterno-coxal depressions (limited by sternal sutures at P2–P4 level), and their internal portions form, at the surface of the sternal plate, quadrangular and depressed areas that are lateral to the flat floor of the sterno-abdominal depression. Sternocoxal depressions are absent in Latreilliidae (Homoliformia) , Cyclodorippiformia, and Gymnopleura , all lacking an appendicular holding system of the abdomen.

The Homoliformia, which first appeared during the Jurassic and were well established during the Cretaceous ( Wright & Collins 1972; Collins & Wienberg Rasmussen 1992; Collins 1998; Van Bakel, Guinot, Jagt & Fraaije 2012), encompasses three Recent families ( Homolidae , Latreilliidae , and Poupiniidae ). A linea homolica is well visible in many fossil homolids, such as the Cretaceous † Homolopsis mendryki Bishop, 1982 ( Bishop 1982: 221, fig. 4), † H. williamsi Bishop, 1992 ( Bishop 1992: 57, figs. 1, 2), † Eohomola adelphina Collins & Wienberg Rasmussen, 1992 ( Collins & Wienberg Rasmussen 1992: 16, fig. 8A–C), † Zygastrocarcinus cardsmithi Bishop, 1986 ( Bishop 1986c: 1099, fig. 2). The intact linea homolica observed in † Homoliformis vagus Collins, Schulz & Jakobsen, 2005 , a homolid from the early Eocene associated with driftwood, indicates the specimen is a corpse rather than a moult (Collins et al. 2005: 17). The holotype of † Hoplitocarcinus gibbosus (Schlüter, 1879) is known by “the median portion of carapace depressed along the lineae homolicae between partially preserved sidewalls” ( Collins et al. 2000: 122). When the fossil carapace is disarticulated at the level of the two lineae, only the interlineal portion of carapace is preserved, e.g., in † Homola vanzoi Beschin, De Angeli & Zorzin, 2009 ( Beschin et al. 2009: 66, fig. 5, pl. 2, fig. 1a, b), from the early Eocene (Ypresian), a species close to the living Homola barbata ( Fabricius, 1793) . It is the presence of a linea homolica in Jurassic “ Prosopidae ” von Meyer, 1860, such as † Laeviprosopon Glaessner, 1933 , and † Tithonohomola Glaessner, 1933 , which led Wehner (1988: 119, fig. 132) to include these genera in Homolidae .

† Mithracites vectensis Gould, 1859 , from the early Cretaceous (Lower Aptian), known by a thoracic sternum with sterno-coxal depressions and a “homoloid button” on sternite 4, gives evidence of homoloid affinities as has been suggested by Bouchard (2000: fig. 27B) and Guinot & Tavares (2001: figs. 17, 18), a way followed by all subsequent authors. A linea homolica was neither described nor figured for example by Withers (1951: 181) and Wright & Collins (1972: 40), and was supposed to be absent by Guinot & Tavares (2001: 535) when they included † Mithracites Gould, 1859 , in Homoloidea . The establishment of † Mithracitidae Števčić, 2005 , within the Homoloidea by Števčić (2005: 23) was not based on the absence of the linea homolica, and Števčić’ s diagnosis does not make evident the synapomorphies of the family. Schweitzer & Feldmann (2011a: 1) did not include this character in their diagnosis, which, like that of Števčić (2005), is not based on differentiating features. In contrast, Karasawa, Schweitzer & Feldmann (2011: 547: fig. 2), who fully recognised † Mithracites within Homoloidea , used “linea homolica absent” as a character for † Mithracitidae . The strength of the linea homolica varies greatly among extant Homolidae , being sometimes very thin, even practically indistinct ( Guinot & Richer de Forges 1995: 298), and its trace could have been simply lost during fossilisation in the case of † Mithracites as in other fossil homolids found without a discernible linea homolica in the taphonomic process ( Bishop 1986a, b; Van Bakel, Guinot, Jagt & Fraaije 2012).

The lateral portions of homolids may be absent because of the disarticulation at the level of the two dorsal lineae ( Crawford 2008), allowing the recognition of interlineal and extralineal portions of carapace in some fossils; the latter may be not preserved, as in † Homola vanzoi ( Beschin et al. 2009: fig. 5), see above. Such a linea is necessary for exuviation in Homolidae , in which the carapace envelops the body, whereas it is absent in Poupiniidae , in which the carapace is only as a cap on the body and may be easily shed at ecdysis without the implication of a linea. The carapace is similarly not ventrally expanded in Latreilliidae , and a linea for the dehiscence is only present on each side at the level of the elongated, narrow “neck”. Guinot & Tavares (2003: 535, fig. 19) remarked that a rim was clearly visible on lateral margin of the carapace of a specimen of † M. vectensis , the indication that the carapace perhaps does not fold ventrally to envelop the cephalothorax. If true, this explains the absence of a linea homolica in † Mithracitidae . This taxon was not previously recognised by De Grave et al. (2009: 28) and Schweitzer et al. (2010: 61), who included † Mithracites among the incertae sedis taxa of Homolodromioidea . The possible presence of a subdorsal P 4 in † Mithracites , as acknowledged by Schweitzer & Feldmann (2011a) and Karasawa, Schweitzer & Feldmann (2011), cannot support its inclusion in Homoloidea , diagnosed by a non-reduced P4. Recent new findings indicate a normal P4 and a reduced, obliquely oriented P 5 in † Mithracites (see Van Bakel, Guinot, Jagt & Fraaije 2012). Moreover, the assignment of Homoloidea to the section Dromiacea by Schweitzer & Feldmann (2011a) is not justified (see Appendix II).

† Gastrodorus von Meyer, 1864 , placed in Homoloidea by Glaessner (1969), Feldmann & Schweitzer (2009), De Grave et al. (2009), and Schweitzer et al. (2010), is not a brachyuran crab and, on the basis of strong arguments, was transferred to the Anomura in its own family, † Gastrodoridae Van Bakel, Fraaije, Jagt & Artal, 2008 ( Van Bakel et al. 2008) and given a full superfamily status by Klompmaker et al. (2011a). The superfamily is known from the Jurassic and early Cretaceous.

Due to their wide sternal plate and true sterno-abdominal cavity, the three families of Cyclodorippoidea are grouped in a separate subsection, Cyclodorippiformia ( Tables 1, 5; Appendix II. Subsection Cyclodorippiformia). This group is a key to a better understanding of the Podotremata, but the familial interrelationships remain partially unresolved (Ahyong et al. 2007). In spite of their small size, generally with cryptic habits (see Concealment behaviour), they begin to be better known as a result of recent data ( Campos 1997; Ho et al. 2008; Ahyong & Brown 2003b; Ahyong 2008; Ahyong & Ng 2009b, 2011), all showing a richer and more diverse clade than expected. Several fossil taxa known only by the carapace have been assigned to the Cyclodorippidae and Cymonomidae ( Conkle et al. 2006; Schweitzer et al. 2010: 78), but a reassessment of some taxa could be possible.

Subsection Gymnopleura : Raninoidea and †Palaeocorystoidea. The Recent Raninoidea , currently assigned to a single family, Raninidae (see Dawson & Yaldwyn 1994; Ng, Guinot & Davie 2008; Ahyong, Naruse, Tan & Ng 2009), comprises approximately 12 genera and 46 species ( Ng, Guinot & Davie 2008, updated).

A unique organisation, the exposure of pleurites 5–7 at P2–P4 levels, already briefly mentioned by H. Milne Edwards (1853: 62) in Ranina , is the “gymnopleurity” ( Fig. 38B, C ), at the origin of the nomen Gymnopleura given by Bourne (1922b) and used by numerous authors ( Glaessner 1929; Lôrenthey in Lôrenthey & Beurlen 1929; Beurlen 1930; Beurlen & Glaessner 1931: 68; Rathbun, 1937: iii, 6; Gurney 1942: 272; Garth 1946: 344; Richardson & Krefft 1949: 69; Barnard 1950: 9, 396; Lebour 1959: 130; Gordon 1963: 56; 1966: 353; Monod 1956: 23, 35, 47; Waterman & Chace 1960: 25; Tyndale-Biscoe & George 1962: 65, 89; Roberts 1962: 185; Bennett 1964: 23; Sakai 1965: 1; 1976: iv, 3, 47; Forest & Guinot 1966: 42; Pichod-Viale 1966: 1266; Hartnoll 1968a: 281; 1975: 658; Rodrigues da Costa 1970: 33; Takeda & Miyake 1970: 193; Serène & Umali 1972: 24; Fielding & Haley 1976: 131; Hartnoll 1968a: 280, 281; 1975: 658; 1979: 75, 76; Goeke 1980: 145; 1981: 971; Werding & Müller 1990: 209; Dai & Yang 1991: 41; Dai & Xu 1991: 1; Chen & Xu 1991: 50; Spears et al. 1993: 458; Watabe 2007: 54, 55, table 4.2). See Appendix II. Gymnopleura .

The Notopterygia Latreille, 1831 ( Latreille 1831: 366, 368; Bate 1888: table p. 4; Stebbing 1908: 17; 1922: 108; Bourne 1922b: 55, footnote) was established as a tribe within the family-group “Macrouri” for crustaceans with particular legs ending in “fins” and arranged in two rows, the posterior being dorsal (“ tous les pieds … à la fois terminés en nageoire et disposés sur deux rangs, les deux ou quatre postérieurs étant dorsaux ”). The tribe Notopterygia included all the Raninoidea , which have modified P2–P5 for burying and swimming ( Van Bakel, Guinot, Artal, Fraaije & Jagt 2012). The nomen is not available in the class-series nomina (see Appendix II. Subsection Gymnopleura ).

Recent important findings on fossil raninoid-like crabs support our views of podotreme evolution, in particular the crucial issue, the Gymnopleura . All raninoids are gymnopleures by definition (subsection Gymnopleura , see Appendix II), and their fossil representatives, known from the Cretaceous to Recent, prove to conform to the extant ones ( Fig. 38B, C ). In contrast, the † Palaeocorystidae does not show exposed pleurites, thus it is not gymnopleure ( Fig. 38A ; Guinot, Vega & Van Bakel 2008: 713) and therefore cannot be kept in Raninoidea . A suprafamilial rank, †Palaeocorystoidea, is justified ( Van Bakel, Guinot, Artal, Fraaije & Jagt 2012). The oldest known occurrence among †Palaeocorystoidea is † Necrocarcinidae , from the early Cretaceous (Hauterivian). † Palaeocorystidae and † Cenomanocarcinidae , known from the Albian, are supposed to have become extinct in the Upper Cretaceous (Lower Maastrichtian); only † Orithopsidae extended into the Oligocene. The †Palaeocorystoidea, presently with five families, 25 genera and 94 species (when taxa described by Luque et al. 2012, are added), shows morphological differences, notably in the thoracic sternum, spermathecae, and male abdomen. It also exhibits a unique cuticle microstructure, a feature with functional-morphological significance ( Haj & Feldmann 2002; Waugh & Feldmann 2003; Waugh et al. 2009). The strong dichotomy between the extinct †Palaeocorystoidea and the Raninoidea was not previously taken into account by palaeontologists (e.g., Glaessner 1969: R498; Van Straelen 1936: 39), and the recent classifications ( Tucker 1998: 321, 364; De Grave et al. 2009: 29; Schweitzer et al. 2010: 76) assigned only a subfamilial rank, † Palaeocorystinae Lôrenthey in Lôrenthey & Beurlen 1929. Finally, Karasawa, Schweitzer & Feldmann (2011: 30) agreed with the views of Guinot, Vega & Van Bakel (2008) in recognising a distinct status, † Palaeocorystidae , but they disagreed with the separate subsection Raninoidia recognised by Guinot, Vega & Van Bakel (2008). The inclusion of the extinct †Palaeocorystoidea and the Recent Raninoidea in the Raninoidia has been proposed by Van Bakel, Guinot, Jagt & Fraaije (2012) (see Tables 1, 5, 6; Appendix II. Subsection Gymnopleura ). †Palaeocorystoidea is sister to Raninoidea but has retained more plesiomorphies.

The synapomorphies of Gymnopleura are as follows:

(1) Modifications of at least the P3 and P4 for burying and swimming. This feature, verified in two extinct palaeocorystoid families, † Cenomanocarcinidae ( Guinot, Vega & Van Bakel 2008: figs. 2, 4, 6A) and † Palaeocorystidae ( Wright & Collins 1972: figs. 11a, b), also characterises all living Raninoidea . Such modified P3 and P4 are unique to Podotremata. According to Hartnoll (1971a: 35), the changes in Raninoidea (flattening and widening of distal articles) evolved on all legs (P2–P5) as burying adaptations, with an additional role for swimming by the P2 and P3 (see Ahyong, Naruse, Tan & Ng 2009: figs. 99, 100, 102, 104, 106, 108, 110, 112, 114, 116). Although Homola swims to escape from predators, there are no obvious natatory adaptations in the P2–P4 ( Hartnoll 1970: 588, 591, fig. 3, pl. 15; also 1971: 35). A modification of the P2 and P3 is rarely found in Eubrachyura, only present in Matutidae and Orithyiidae , the Matutidae being characterised by an active swimming and burying.

(2) Modifications of the buccal frame and mouthparts (oxystome condition) as morphological adaptations for burying ( Fig. 42D ; Bourne 1922b; Van Bakel, Guinot, Artal, Fraaije & Jagt 2012: 141).

(3) Double locking system of the abdomen. Presence of two close teeth (“double peg”) on sternite 5 was verified in several genera and species of three families of the †Palaeocorystoidea (extinct), i.e., † Cenomanocarcinidae ( Guinot, Vega & Van Bakel 2008: 684, 694, 710; Van Bakel, Guinot, Jagt & Fraaije 2012: fig. 40C–F), † Orithopsidae ( Van Bakel, Guinot, Jagt & Fraaije 2012: fig. 40A, B), and † Palaeocorystidae ( Fig. 40F–I ; Van Bakel, Guinot, Jagt & Fraaije 2012: fig. 39A–D). These two close teeth, which are placed on sternite 5 practically in line with the body axis, are supposed to be the sternal elements of a locking system. Their existence in the most ancient palaeocorystids, dated as Early Albian, Aptian and even Hauterivian ( Yazdi et al. 2009), demonstrates that abdominal holding was an early requirement for a brachyuran crab. The double peg is situated on a short hook-like projection in an extinct lyreidid subfamily, † Marylyreidinae Van Bakel, Guinot, Artal, Fraaije & Jagt, 2012 , from the Upper Albian-Lower Cenomanian, which is nevertheless a true Raninoidea (exposed sternites 5–7, thus gymnopleurity). The † Marylyreidinae does not show a thoracic sternum/pterygostome junction (it is possible, however, that “normal” Milne Edwards openings were present; the mxp3 coxa being flabelliform) in contrast to all other Raninoidea and thus is basal in the superfamily Raninoidea , representing the “missing link” between †Palaeocorystoidea and Raninoidea . In living Lyreididae the abdomen is tightly held in place thanks to two teeth at the extremity of a developed hook (episternite 5) matching a long socket on abdominal somite 6, thus secured in a flexed position ( Bourne 1922b: pl. 4, fig. 4; Hartnoll 1975: 672; Guinot 1993b: figs. 4, 6; Guinot & Bouchard 1998: 639, 681, fig. 11A–C; Van Bakel, Guinot, Artal, Fraaije & Jagt 2012: fig. 36; see below). The paired structure on each sternite 5 is a condition that is shared by all †Palaeocorystoidea and the Lyreididae and corresponds to a preserved plesiomorphy. The locking mechanism is lost in all the Recent Raninoidea , which have a short and unfastened abdomen, in contrast to Lyreididae . It remains unknown to what extent the lyreidid socket differs from the typical socket corresponding to the eubrachyuran press-button.

The origin of the “double peg” ( Fig. 40F–I ), found in several extinct palaocorystoid families as early as the Lower Cretaceous (Hauterivian), remains perplexing. The double peg is assumed here to be homologous to the two teeth of the lyreidid hook and homoplasious to the simple button of the eubrachyuran press-button mechanism, all these structures being invariably located on sternite 5. As such it engages itself into a socket in which it is tightly held. It is tentatively proposed that the double-peg structure was initially double because it could have matched a biramous uropod, therefore requiring two corresponding thoracic elements to be efficient. This double peg could be regarded as the coapted structure to match the uropod at the first stage of its modification into a socket. This putative “double-socket” type has, however, not yet been observed in extinct species. The exposure of such a socket needs a complete fossil male abdomen and the removal of the abdomen from the rest of the body to examine its ventral surface at the level of the sixth somite. The double peg matched a large socket (modified uropod), somewhat similar to that found in extant lyreidids ( Guinot 1993b: fig. 7; Van Bakel, Guinot, Artal, Fraaije & Jagt 2012: fig. 36C, D). The first known appearance of the double peg in Hauterivian crabs is assumed to have been anterior to the double peg, which is situated on a hook-like projection in † Marylyreidinae from the Lower Cenomanian. The disposition in fossil crabs, which provides information on the sequence of evolutionary changes, may reveal the type of locking mechanism in the earliest brachyurans, at least in part because all of its elements remain imperfectly known. Multiple locking teeth (three to four prominences), located on the steep walls of the sterno-abdominal cavity and likely interacting with a socket, have been described in fossil podotremes. Such is the case of the “dakoticancrid holding” that involves several teeth and which is present in † Dakoticancridae and † Ibericancridae ( Guinot 1993a: 1231, fig. 8; Van Bakel, Guinot, Artal, Fraaije & Jagt 2012; see also Artal et al. 2008). A locking device consisting of two teeth in the form of a double peg is known in several families of †Palaeocorystoidea: † Cenomanocarcinidae (Guinot et al. 2008: 694, 710; Van Bakel, Guinot, Artal, Fraaije & Jagt 2012: figs. 40C–F, 41), † Palaeocorystidae and † Orithopsidae ( Van Bakel, Guinot, Artal, Fraaije & Jagt 2012: figs. 37, 39, 40A, B).

The abdominal-locking mechanism exhibited by the extant inachoidid Paulita consists of two or several granules (roughly similar or with only one blunt, only slightly more salient granule) present between sutures 4/5 and 5/6, the abdominal edge being conspicuously thickened at the level of abdominal somites 4–6 and ending at abdominal somite 6 level (somite 6 fused to telson, thus pleotelson) in the form of a socket that is, however, not deeply excavated. The unique case of Paulita , a genus that displays several plesiomorphic features ( Guinot 2012), permits hypothesising on the perplexing origin of the eubrachyuran press-button, typically represented by a single prominence ( Guinot & Bouchard 1998). It is possible that several granules were initially present at the level of somite 6 that could have been able to grip the abdominal margin just at the somite preceding the moveable telson. One of these granules may have been the most effective in matching the abdominal surface at the first stage of the uropod modification into a socket, resulting in a unique structure to hold the abdomen. Paulita is the only known brachyuran with such a disposition. Several locking prominences, in the form of “swellings” at the level of sternites 6, 7 and matching the deeply concave margins of abdominal somites 4 and 5, exist in Libinia spinosa , but this case is different because the prominences occur together with a strong button on sternite 5 ( Tavares & Santana 2011: fig. 1C, D).

Several hypotheses based on morphological characters have been proposed to explain the phylogeny of raninoids. According to Števčić (1973: 631), “the raninids started their evolution from highly developed crabs” and “their subsequent evolutionary pathway was regressive in both a morphological and ecological sense”; Števčić (1995: 33) placed the Raninoidea “at the end of the dromiacean hierarchical system”. According to Hartnoll (1979: 75), raninoids were “the most advanced of the primitive Brachyura ”. The Archaeobrachyura, first established to include Homoloidea , Raninoidea , and Cyclodorippoidea , was later restricted to Raninoidea and Cyclodorippoidea ( Guinot & Tavares 2001: 531, 532). Ahyong et al. (2007: 582, 583) apparently did not recognise the uropod-socket homology and stated that the “absence of uropods unites raninoids and cyclodorippoids with Eubrachyura”. Most raninoids and all cyclodorippoids actually lack any type of vestigial uropods (dorsal plate, ventral lobe, socket) whereas eubrachyurans have a socket. The presence of a socket retained in Lyreididae (see below), shared by Homoliformia and Eubrachyura, rends the taxon Archaeobrachyura problematic, except if the socket is assumed to have appeared independently (see cladogram, Fig. 41 ). The oxystome condition may be a synapomorphy of Archaeobrachyura redefined by Guinot & Tavares (2001), unless homoplasy.

Martin & Davis (2001: 74, 75), based on molecular analysis, proposed a scheme in which Raninoidea and Cyclodorippoidea form subsection “Raninoida” within Eubrachyura (despite the absence of vulvae on thoracic sternite 6) besides the two subsections Heterotremata and Thoracotremata. Following Martin & Davis (2001) classification, some palaeontologists ( Feldmann 2003; Collins in Collins et al. 2003; Collins & Jakobsen 2004; De Angeli & Beschin 2001; De Angeli & Garassino 2006a: 280; Amato et al. 2008: table 1; Schweitzer et al. 2010: 70, 78) have regarded raninoids and/or cyclodorippoids as eubrachyurans. According to Rice (1981a: 290, fig. 2A; 1981b: 1007, fig. 3; see also Williamson 1965; Wear & Fielder 1985), apomorphic larval characters are shared by Raninoidea and Eubrachyura. No spermatozoal synapomorphies support a sister-group relationship of Raninoidea and Eubrachyura, whereas, in contrast, some sperm similarities are shared by Lyreidus brevifrons and other podotremes ( Jamieson et al. 1994b: 239, 248). According to the cladistic analysis of nuclear ribosomal RNA sequences by Ahyong et al. (2007: 582, figs. 2, 3), “the position of the raninoid clade is somewhat unexpected in occupying the ‘intermediate’ position in the podotreme grade”. A section “Raninoida” was recognised as a separate section by these authors (see Table 1; Appendix II). The inclusion of a member of Lyreididae in future molecular analyses should certainly be informative.

It is out of the scope of this study to discuss the interrelationships within the Recent Raninoidea , traditionally grouped in the single family Raninidae and subdivided into six subfamilies ( Cyrtorhininae Guinot, 1993 , Lyreidinae Guinot, 1993 , Notopodinae , Ranininae , Raninoidinae , and Symethinae ) (see Ng, Guinot & Davie 2008; De Grave et al. 2009: 28). The Lyreidinae is here provisionally raised to a familial rank, distinct from all Raninidae , which consists of five subfamilies: Cyrtorhininae , Notopodinae , Ranininae , Raninoidinae , and Symethinae . These subfamilies are obviously not equivalent, in particular Cyrtorhininae and Symethinae , which could deserve familial status. The interrelationships of extant Raninoidea will be determined in a revision being prepared by the second author. Some remarks on Lyreididae , Symethinae, and Cyrtorhininae are presented below.

Family Lyreididae . The Lyreididae , initially established at a subfamilial rank for two Recent genera, Lyreidus De Haan, 1841 , and Lysirude Goeke, 1986 , possesses a gymnopleure condition ( Fig. 38B ), hence its logical assignment to superfamily Raninoidea instead of †Palaeocorystoidea ( Fig. 38A ). The sperm ultrastructure of Lyreidus clearly shows several plesiomorphic features, similar to those of dromioids and homoloids, and supports the basal position of Lyreididae within the Raninoidea ( Jamieson et al. 1994b). Larval characters also distinguish Lyreidus from the other Raninoidea ( Williamson 1965; Rice & Provenzano 1970; Rice 1980, 1981b; Wear & Fielder 1985). Lyreidids have an abdominal-locking mechanism, unique in Brachyura , which is absent in other extant raninoids. It consists of a pair of strong, elongated projections from sternite 5 (“pterygoid processes” of Bourne 1922b: 69, pl. 4, fig. 4) that firmly fit into a pair of deep sockets (“small aliform processes” of Bourne 1922b) in the extended latero-posterior angles of abdominal somite 6 ( Fig. 40A–C, E ; Hartnoll 1975: 672; Guinot 1993b: figs. 4, 6; Guinot & Bouchard 1998: 639, 681, fig. 11A–C; Van Bakel, Guinot, Artal, Fraaije & Jagt 2012: fig. 36). The projection's tip, recurved and distally hooked, bears an additional subdistal hook so that the locking structure is paired. The complete structure, which acts as a safety catch, may remain functional in some females, even when ovigerous ( Guinot 1993b: fig. 7). The projections are always present although less developed as the females grow, and the two hooks become blunt, vestigial, or even absent, so that the system is no longer effective ( Fig. 40D ; Guinot & Bouchard 1998: fig. 11D; Bouchard 2000: figs. 29, 40, 41, 42A–C). This important feature, ignored as part of sexual dimorphism in Raninidae by Feldmann & Schweitzer (2007: 39, 46), should be added to the diagnosis of Lyreididae . When compared to other raninoids, the lyreidid thoracic sternites 5 and 6 are wider, sternites 1–7 are somewhat in the same plane, the male abdomen is longer, inserted between the pereopods, and fixed. The Lyreididae is the only extant raninoid family in which the gonopods are protected in a depression, instead of protruding from the short abdomen as in other raninoid families. Pleurites 5–7 are exposed as a flat plate continuing from the edge of the branchiostegite, and the thin P5 is not conformed to tightly fit the carapace border ( Fig. 38B ). This strongly contrasts with the disposition of Ranininae , for example, where the excavated plate formed by the exposed pleurites 5–7 shows as an excavated area, overhung by the prominent edge of the branchiostegite and by the modified P5 strictly apposed along the carapace border ( Fig. 38C ), the assemblage acting as a roof for water passage ( Bourne 1922b). The gymnopleure configuration of extant Lyreididae is present in the fossil representatives of the subfamily Lyreidinae but absent in the † Marylyreidinae that is older and extinct ( Van Bakel, Guinot, Artal, Fraaije & Jagt 2012: fig. 25). The lyreidid thoracic sternum is more plesiomorphic than that of other raninoids, in being not particularly narrow, less modified, and there is a sterno-abdominal depression that completely receives the male abdomen. The length and shape of the abdomen, which is entirely folded and applied against the sternal plate, allow locking, a condition that is absent in other raninoid families, which have a narrower, even posteriorly linear thoracic sternum and a shorter, unfolded abdomen, modifications that are evidence of a high degree of specialisation.

The Lyreididae is represented by many fossil species (e.g., Collins & Wienberg Rasmussen 1992; Feldmann 1992; Tucker 1998; Van Bakel, Guinot, Artal, Fraaije & Jagt 2012), sometimes with a visible projection of sternite 5 (episternite 5), as observed by Bouchard (2000: fig. 32A; unpublished data) in the Eocene † Rogueus orri Berglund & Feldmann, 1989 . Some specimens of † Lyreidus antarcticus Feldmann & Zinsmeister, 1984 , have been found fossilised with an intact abdomen attached to the carapace ( Feldmann & Wilson 1988: 481, fig. 9.9).

Subfamily Symethinae . A familial status for Symethinae Goeke, 1981 ( Goeke 1981: 972; onomatophore or type genus: Symethis ), was first suggested by Guinot (1993b: 1330), followed by Tucker (1998: 321, 359), Martin & Davis (2001: 74), Waugh et al. (2009: 35), and Schweitzer et al. (2010: 77), but not acknowledged by Davie (2002: 485), Feldmann & Schweitzer (2007: 50), Ng, Guinot & Davie (2008: 43), Ahyong, Naruse, Tan & Ng (2009: 135), and De Grave et al. (2009: 29). The molecular data of Ahyong et al. (2007: 583) provided little evidence to warrant a separate family, but these authors nevertheless remarked the “unusual synapomorphies” present in symethines. Symethis seems to require a familial rank, Symethidae , within Raninoidea with respect to the following characters: (1) shape and arrangement of the eyestalks (small), antennules (small, concealed), and antennae (massive, with distal articles reduced or lost); (2) rostrum long; (3) thoracic sternum very narrow at the level of the P2–P4 condyles; (4) thoracic sternum/pterygostome junction narrow; (5) sternum/pleurites connections (“epaulettes” of Bourne 1922b: 59) wide between P1–P2 and P2–P3 ( Goeke 1981; Guinot 1993b: 1330, fig. 3); (6) exposure of pleurites 5–7, not as an excavated plate, but formation of a region overhung by the hollow branchiostegite, the P5 fitting the branchiostegal margins, and the complete assemblage providing a roof for water passage; (7) non-recessed spermathecae, widely separated, horizontal, large, and overhung by calcified sternal hoods ( Goeke 1981: 976, fig. 2A; Davie 1989a: fig. 1c; Guinot & Quenette 2005: fig. 24D); (8) chelipeds with swollen palm and long fingers, without the usual raninoid shape; (9) G1 laterally compressed; (10) G2 long, with elongated, spatulate apical process (Gomes-Corrêa 1970: 10, pls. 5–7, figs. 61, 62; Goeke 1981: fig. 3A; Davie 1989a: fig. 1b); (11) P2–P5 markedly modified and with sickle-shaped dactyli for digging; (12) eroded dorsal surface of carapace, with a characteristic ornamentation and body cuticle microstructure with fungiform nodes, rather similar to that of extinct †Palaeocorystoidea ( Waugh et al. 2009: 35, figs. 9.7–9).

Subfamily Cyrtorhininae . The subfamilial rank proposed for Cyrtorhina Monod, 1956 , by Guinot (1993b: 1325, 1330, as Cyrtorhinae) and followed by Tucker (1998: 322, figs. 21, 22), Davie (2002: 485), Ng, Guinot & Davie (2008: 43), Waugh et al. (2009: 35), De Grave et al. (2009: 28) was not acknowledged by Martin & Davis (2001: 74) and Schweitzer et al. (2010: 71). Despite the broadly ovate body of Cyrtorhina that contrasts with the narrow, elongated carapace of Symethis , the Cyrtorhininae and Symethinae share several characters, in particular the smooth, non-spinose inferior border of the P1 propodus and the straight fixed finger. The Cyrtorhininae is distinguished by a peculiar sternite 3 that does not form a “crown” with sternites 1 and 2, by a narrow sternite 4 (enlarged in other Raninoidea ) so the P1 coxae are close to each other, and by a narrow, recessed thoracic sternum/ pterygostome junction. The Cyrtorhininae and Symethinae appear to be closely related as has been suggested by Serène & Umali (1972: 49) and Guinot (1993b), a resemblance judged to be superficial by Goeke (1980: 976). Their respective status and position among the Raninoidaea is presently being studied by the second author.

Raninoidea and †Palaeocorystoidea. The hypothesis that Raninoidea and †Palaeocorystoidea are putative sister groups is corroborated by the transformation series shown by fossil and Recent taxa. The carapace of †Palaoecorystoidea is in contact with the coxae of the pereopods ( Fig. 38A ) as in most Brachyura . In contrast, pleurites 5–7 are exposed in Raninoidea (gymnopleurity), as a flat plate in the more basal ( Lyreididae , Fig. 38B ), excavated in the more derived ( Raninidae Ranininae, Fig. 38C ). The †Palaoecorystoidea shows a “normal” (i.e., relatively wide) thoracic sternum, a “normal” (i.e., long) male abdomen, and “normal” (i.e., separated and not recessed) spermathecae. Such a “normal” sternal and abdominal condition also characterises the † Cenomanocarcinidae , a Cretaceous extinct family assigned to the †Palaeocorystoidea ( Guinot, Vega & Van Bakel 2008; Van Bakel, Guinot, Artal, Fraaije & Jagt 2012). In addition, a paired prominence was found on sternite 5 of a specimen of † Cenomanocarcinus Van Straelen, 1936 , which is assumed to lock the abdomen and is reminiscent of the two teeth on the sternite 5 projection of the Lyreididae . Moreover, in the same † Cenomanocarcinus specimen two spermathecae were located at the extremities of the sutures 7/8, normally interspaced, and not recessed. Because it is not gymnopleure, the † Cenomanocarcinidae cannot be included in Raninoidea and is assigned to the non-gymnopleure †Palaeocorystoidea ( Fig. 38A ), the earliest Gymnopleura , known from the Lower Cretaceous (see Tables 5, 6; Van Straelen 1923a). An inclusion in Raninoidia was adopted by Collins (2010: 15) for several new species of † Cenomanocarcinidae and † Necrocarcinidae from Nigeria.

The sternal organisation of the †Palaeocorystoidea is now better understood. In † Palaeocorystidae , † Notopocorystes stokesii (Mantell, 1844) (Lower Albian to Cenomanian) was studied by the first author with J.-M. Bouchard ( Bouchard 2000). The well preserved ventral surface of the numerous specimens examined had shown a rather wide thoracic sternum, a rather wide and long male abdomen with an elongated somite 6, an exposed lateral ridge on sternite 5 ending close to the P2 coxo-sternal condyle: the two faint tubercles (only visible when the abdomen was not in place) were not recognised as functional ( Bouchard 2000: 128, figs. 30 A, B, 42D, and unpublished sketches; see also Bell 1863: 15, pl. 3, figs. 5, 9; Wright & Collins 1972: 76, pl. 14, figs. 1, 2; Wright & Collins 1972: 76, pl. 14, fig. 6, for † N. serotinus ). The two tubercles of † N. stokesii and the two tubercles on a sternite 5 prominence of † Cenomanocarcinidae (see Guinot, Vega & Van Bakel 2008: 684, 685, 694, 710) may be now considered homologous to the lyreidid hook projection (episternite 5) ending in two teeth. The presence and particular morphology of a corresponding socket must be verified in the palaeocorystoid material to establish this homology. It is here assumed that † Palaeocorystidae , † Cenomanocarcinidae (†Palaeocorystoidea, i.e., nongymnopleure, exclusively fossil) and Lyreididae ( Raninoidea , i.e., gymnopleure, fossil and Recent) share a roughly similar abdominal-locking system. Although the relationships between the abdomen and sternum remain partially unclear in some “raninoid” fossils, the paired structure on sternite 5 (“double peg”) present in several genera helps in understanding the phylogeny of these podotremes, in particular the derivation of the specialised, more derived Raninoidea from the typically podotreme †Palaeocorystoidea ( Van Bakel, Guinot, Artal, Fraaije & Jagt 2012; see also Luque et al. 2012). Palaeocorystoids have a relatively long and male abdomen, filling the complete space between the pereopods, with only the anterior portion of sternite 4 exposed; the spermatheca is at the extremity of suture 7/8; a thoracic sternum/pterygostome junction is absent; Milne Edwards openings are present; some species may have also a short body (Collins & Breton 2008). The gymnopleurity ( Fig. 38B, C ) and the junction of the thoracic sternum with the pterygostome ( Fig. 42E ) clearly distinguish Recent Raninoidea . The sternum/ pterygostome junction, clearly visible in fossil raninoids but not taken into consideration by most authors (e.g., Förster & Mundlos 1982: figs. 10, 11, pl. 2, fig. 7b), characterises all the Raninoidea except † Marylyreidinae , which is sister to Lyreidinae .

The hypothesis that † Araripecarcinus ferreirai Martins-Neto, 1987 , only known by the very small, damaged holotype found associated with a fish in a limestone concretion from the Brazilian Cretaceous (Santana Formation of northeast Brazil, Romualdo Member, Araripe Basin, Lower-Middle Albian), belongs to Podotremata, specifically to Gymnopleura , was first suggested by Guinot & Breton (2006: 612), a view adopted by Karasawa, Schweitzer & Feldmann (2008: 93; 2011: 55) and Luque et al. (2012: 416). Indeed, † A. ferreirai exhibits enlarged articles on several pereopods, in particular a much reduced P5 with a cylindrical merus and an oval, flattened, presumably natatory propodus (Martins-Neto 1987: 408, figs. 1, 2), typical characters of Gymnopleura . Martins- Neto (2005: table 7), however, referred the taxon to Portunidae . † Araripecarcinus Martins-Neto, 1987 , probably belongs in †Palaecorystoidea, in an incertae sedis family ( Van Bakel, Guinot, Artal, Fraaije & Jagt 2012). Maisey & Carvalho (1995: 9, 10, 16) have discovered larval fossil carapaces among the stomach contents of fishes of the Araripe Basin, which provide further evidence that the environment was marine to brackish and confirm the presence of brachyurans in the Romualdo Member of the Santana Formation (see Saraiva et al. 2009: 71), only represented until 1995 by the holotype of † Araripecarcinus ferreirai . The fossil “protozoea” larvae of Maisey & Carvalho (1995: 9, fig. 4), which were estimated the first recorded instance of a fossil brachyuran larva and also the oldest known (Albian), have a long dorsal spine (partially broken) and a shorter rostral spine.

The first Gymnopleura were typical podotremes, and the particular morphology of the highly specialised living representatives is the result of adaptations for burying by digging, removing and raking the sediment, and using mechanisms that facilitate penetration in the sediment. This is clearly demonstrated by the evolutionary series seen in the fossil record. Modern raninoids accumulated a number of synapomorphies that are absent in their sister group (palaeocorystoids): narrowing of the thoracic sternum (keel-like for the most part), presence of a sternum/pterygostome junction (leading to the absence of the Milne Edwards inhalant openings and the development of respiratory adaptations), change in the orientation and shape of the pereopods, shortening and unfolding of the abdomen, loss of the abdominal-locking mechanism (except in Lyreididae ), recessing and closing up of the spermathecae, and development of gymnopleurity. An extreme result of this narrowing and folding of the body (a complete distortion) is the V-roof shaped carapace with two steep slopes found for instance in Cosmonotus Adams & White in White 1848 ( Tavares 2006: fig. 1 A, B). The frequent suggestion of close relationship between the raninoid genera and Eubrachyura and the assertion that “Raninoida is the sister group to Cyclodorippoida plus Eubrachyura” ( Ng, Shih, Tan, Ahyong & Ho 2009: 16, fig. 5) are challenged by the transformation series in the long, complex evolutionary history of Gymnopleura that is revealed by recent discoveries in the fossil record ( Van Bakel, Guinot, Artal, Fraaije & Jagt 2012; Luque et al. 2012).

The socket on abdominal somite 6, derived from the uropod, had an early appearance in brachyuran evolution. A “homoloid button” has been found on sternite 4 in Homoliformia from as early as in the Lower Cretaceous, e.g., in † Mithracites vectensis (? † Mithracitidae ), from the Aptian, and in the homolid † Homolopsis edwardsii Bell, 1863 , from the Albian ( Bouchard 2000: fig. 27). The presence of its corresponding abdominal structure, a socket, is expected in these fossils. The locking system in the primitive Gymnopleura , such as † Cenomanocarcinidae , is prefigurative of the eubrachyuran press-button system since the two parts (sternal structure and socket) come facing together for attachment, in contrast to Dromiidae where the two structures are side by side, with the involvement of the appendages and absence of a socket.

The Raninoidea and some Cyclodorippoidea share a similar disposition of afferent respiratory openings and sternite 4. An oxystome condition is completely developed in Cyclodorippidae , but Phyllotymolinidae and Cymonomidae has each a particular respiratory disposition ( Bouvier 1899; Tavares 1991, 1998). Correlatively, a sternum/pterygostome junction occurs in a part of Cyclodorippoidea , only in Cyclodorippidae (no junction in Phyllotymolinidae and Cymonomidae ), leading to the loss of the Milne Edwards openings, which are replaced by pterygostomial channels (Bouvier 1999; A. Milne-Edwards & Bouvier 1902; Ihle 1916). In all Raninoidea a thoracic sternum/pterygostome junction (by a sternite 4 extension) is present, varying from narrow ( Cyrtorhininae , Symethinae ) to wide as in Ranina ranina ( Fig. 42A ). An extension of several episternites is shared by the Cyclodorippoidea (thoracic sternum/branchiostegite junction between P1 and P2, P2 and P3, variable between P3 and P4) and the Raninoidea (sternum/pleurites 5-7 junction between P1 and P2, P2 and P3, and P3 and P4).

Spermatozoal analysis support a sister-group relationship of Cyclodorippoidea and Raninoidea , each with convincing and unique synapomorphies, the association of both being upheld, thus suggesting a monophyletic clade ( Jamieson et al. 1995: 277, fig. 1A; Jamieson & Tudge 2000: 48; see also Jamieson 1989; Jamieson et al. 1994a, 1994b), corresponding to the Archaeobrachyura redefined by Guinot & Tavares (2001).

Extinct Podotremata. The major extinct taxa that can be confidently included in the Podotremata are (see Table 5): Homolodromioidea , comprising family † Prosopidae von Meyer, 1860 (von Meyer 1860), and other possible families such as † Bucculentidae Schweitzer & Feldmann, 2009 , and † Tanidromitidae Schweitzer & Feldmann, 2008 (see Schweitzer & Feldmann 2009b; Hyžný et al. 2011), all known only from carapaces and thus based on partial information; †Etyoidea (see Superfamily †Etyoidea below); †Dakoticancroidea, comprising † Dakoticancridae and † Ibericancridae ( Artal et al. 2008; see Superfamily †Dakoticancroidea below); †Palaeocorystoidea, comprising five families († Palaeocorystidae , † Cenomanocarcinidae , † Necrocarcinidae , † Orithopsidae , and † Camarocarcinidae Feldmann, Li & Schweitzer, 2008 ) and constituting with the Raninoidea the subsection Gymnopleura ( Table 6; see above); † Torynommatidae , at least its type genus † Torynomma Woods, 1953 , pro parte (see Other extinct putative Podotremata below); and †Glaessneropsoidea Patrulius, 1959, comprising five families known from dorsal carapaces and thus uncertain († Glaessneropsidae Patrulius, 1959 , † Lecythocaridae Schweitzer & Feldmann, 2009 , † Longodromitidae Schweitzer & Feldmann, 2009 , † Nodoprosopidae Schweitzer & Feldmann, 2009 , and † Konidromitidae Schweitzer & Feldmann, 2010 ). The status of †Goniodromitinae Beurlen, 1932 ( Beurlen 1932) is uncertain.

† Binkhorstia , previously thought to be a podotreme, is eubrachyuran and deserves a family or subfamily rank, not described here. † Goniochele Bell, 1858 , is not a podotreme (See Monophyletic Heterotremata: Additional fossil Heterotremata).

Subfamily † Graptocarcininae . The Dynomenidae accommodates the extinct subfamily † Graptocarcininae Van Bakel, Guinot, Corral & Artal, 2012 , which groups † Graptocarcinus Roemer, 1887 (type species by monotypy: † G. texanus Roemer, 1887 ), from the Middle-Upper Cretaceous, and † Cyclothyreus Remeš, 1895 (type species by monotypy: † C. strambergensis Remeš, 1895 ), from the Late Jurassic, Tithonian) ( Van Bakel, Guinot, Corral & Artal 2012; see also Remeš 1895). A recently described species of † Graptocarcinus , with preserved ventral parts, shows a plesiomorphic configuration of the abdomen, i.e., completely filling in length the sternoabdominal depression, and an abdominal-holding system that is similar to that of the Acanthodromiinae Guinot, 2008 , and a P5 conspicuously reduced, carried subdorsally in oblique position ( Van Bakel, Guinot, Corral & Artal 2012: figs. 2, 3).

Superfamily †Etyoidea. Some fossil podotreme families need a high-ranked assignment. This is notably the case of † Etyidae Guinot & Tavares, 2001 (incorrectly spelled “ Etyiidae ” by De Grave et al. 2009: 27, although based on † Etyus ). The †Etyoidea shares with Dynomeniformia an abdomen that is entirely inserted laterally between the pereopods, but the absence of dorsal uropods as well as others characters (see Guinot & Tavares 2001) clearly distinguish †Etyoidea from Dynomeniformia. Karasawa, Ohara & Kato (2008: 106) erroneously stated that † Etyidae had been referred to the Archaeobrachyura by Guinot & Tavares (2001) but this is not correct: these authors left the family without subordination, treating it as an incertae sedis member of Brachyura . The traits that allowed De Grave et al. (2009: 26, 27, as † Etyiidae ) to include † Etyidae in their Dromioidea besides Dromiidae , Dynomenidae , and Homolidae remain unknown, the putative characters (apart from the spermatheca) shared by † Etyidae and these families not being identified. Armstrong et al. (2009: 750) and Breton (2010: 25, figs. 4J–L, 5A, B) subordinated †Etyoidea to the section Podotremata. We cannot agree to the inclusion of † Etyidae in Cyclodorippoidea as done by Števčić (2005: 25). Karasawa et al. (2009: 80) considered † Etyidae one of the 10 major clades of their paraphyletic Podotremata, and the family was attributed to a new section, †Etyoida Karasawa, Schweitzer & Feldmann, 2011, by Karasawa, Schweitzer & Feldmann (2011: 548; see also Schweitzer, Feldmann, Franṭescu & Klompmaker 2012).

† Caloxanthus A. Milne-Edwards, 1864 (type species: † Caloxanthus formosus A. Milne-Edwards, 1864), included in Carpiliidae by Wright & Collins (1972: 103), assigned with reservation to † Diaulacidae or, if not, “to an undescribed family within the Podotremata” by Guinot & Breton (2006: 609), then placed in Dynomenidae ( Collins & Breton 2011; Breton & Collins 2011), was transferred to † Etyidae by Armstrong et al. (2009) (see Klompmaker et al. 2011b; Charbonnier et al. 2012). The †Etyoidea appeared in the very Early Cretaceous (Valanginian) and then diversified throughout the Cretaceous to become extinct by the Paleocene (Danian) ( Breton 2010; Klompmaker et al. 2011b).

Superfamily †Dakoticancroidea. The Cretaceous † Dakoticancridae and † Ibericancridae (grouped in †Dakoticancroidea, see Artal et al. 2008) differ from Dynomeniformia in having, among other characters, a wider thoracic sternum, a more defined sterno-abdominal cavity, and, presumably, a socket on the abdominal somite 6 corresponding to a modified uropod. Inclusion of †Dakoticancroidea in the large group referred to as “ Dromiacea ” ( De Grave et al. 2009: 27; Schweitzer et al. 2010: 57) is not supported at all by morphological features, except the presence of a paired spermatheca. In fact, a diagnosis of “ Dromiacea ” that includes Dromioidea , †Dakoticancroidea, and Homoloidea as in De Grave et al. (2009: 26–28) and Schweitzer et al. (2010: 57–70) rests on unsettled data, the characters shared by these taxa (except coxal gonopores in both sexes and paired spermatheca) remaining to be identified. With the addition of the †Eocarcinoidea Withers, 1932 ( De Grave et al. 2009: 27; Schweitzer et al. 2010: 57) the monophyly of such “ Dromiacea ” is clearly untenable. †Dakoticancroidea being sister to Cyclodorippoidea + Eubrachyura according to Karasawa, Schweitzer & Feldmann (2011: 555), as preliminarily suggested in a cladogram by Guinot & Tavares (20001: fig. 16), is a hypothesis that needs further examination.

Other extinct putative Podotremata. The podotreme nature of some extinct families remains questionable because the most reliable characters (gonopores, ventral surface, thoracic sternum/male abdomen relationships) that may allow a confident assignment are still lacking.

The † Diaulacidae Wright & Collins, 1972 , is likely a podotreme family, but without any supporting evidence, the “ventral” characters remaining unknown for now, and in particular the presence, or absence, of dorsal uropods. † Diaulax Bell, 1863 , originally considered eubrachyuran in Cancridae ( Bell 1863: 6; see Van Straelen 1928b: 618), subsequently referred to Podotremata within “ Dromiacea ” ( Carter 1898: 19) and included in Dynomenidae ( Wright & Wright 1950; Glaessner 1969; Schweitzer, Feldmann, Fam, Hessin, Hetrick, Nyborg & Ross 2003; see also Schweitzer & Feldmann 2005: 38), ultimately became the onomatophore (type genus) of a separate family ( Wright & Collins 1972; Bishop 1983 a, 1986b). The † Diaulacidae , with only the P5 being dorsal in † Diaulax carteriana Bell, 1863 ( Bell 1863: 7) , was resuscitated by Guinot & Tavares (2001: 539). This initiative was followed by Števčić (2005: 18), De Grave et al. (2009: 26), and Schweitzer et al. (2010: 66), Karasawa, Schweitzer & Feldmann (2011: 539), who all include † Diaulacidae in Dromioidea , although the available material (see Karasawa, Schweitzer & Feldmann 2011: fig. 5D) does not provide significant information about the presence of uropods. Guinot (2008: 20) briefly compared the characters of † D. carteriana Bell, 1863 , the type species (questionably †Diaulacinae Wright & Collins, 1972), with those of Metadynomeninae Guinot, 2008 . The relationships of † Diaulacidae with †Etyoidea should be perhaps established if evidence is found for the absence of dorsal uropods ( Van Bakel, Guinot, Artal, Fraaije & Jagt 2012; Van Bakel, Guinot, Corral & Artal 2012). † Diaulax originated during the Late Jurassic according to Glaessner (1931; 1969; see also Wright & Wright 1950; Wright & Collins 1972; Fraaye 1996c), but this early “puzzling” record must be removed, the genus being Upper Cretaceous (Cenomanian) according to Wright (1997).

For a discussion of the familial status of † Mithracitidae , which belongs in Homoliformia, see Van Bakel, Guinot, Jagt & Fraaije 2012; The four subsections of Podotremata; Subsections Dynomeniformia, Homoliformia, and Cyclodorippiformia; Appendix II).

The nomenclature of † Dromilites is quite problematic, as stated by Quayle & Collins (1981). The nomen Dromilite , established by H. Milne Edwards (1837b: 255; see also 1837c: 115) for a fossil crab from the Isle of Sheppey, England, was not accompanied by a species name and thus is a nomen nudum. At the same time H. Milne Edwards (1837a: 179) created with a short description the binomen † Dromia bucklandii for a fossil crab from Sheppey, Enlgand. Reuss (1845: 15) assigned to “ Dromilites Milne Edwards ” two species: † D. pustulosus , which was later transferred by Reuss himslef (1859) to a new genus † Polycnemidium Reuss, 1859 , as its type species; and a second genus, † Brachyurites rugosus von Schlotheim, 1820 , similarly later transferred by Reuss (1859) to a new genus † Dromiopsis Reuss, 1859 , as its type species. Bell (1858: vii, 30, 31, pl. 6; see also M’Coy 1849: 167; Collins & Jakobsen 1995) used the nomen † Dromilites bucklandii for crabs from the London Clay. † D. bucklandii was designated as the type species of † Dromilites by Glaessner (1969 R: 487), an act that was commonly accepted. Schweitzer & Feldmann (2010c: 422, erroneously as D. bucklandi ) proposed to retain the binomen † Dromilites bucklandii with the authorship of H. Milne Edwards, 1837 (1837b) (see also Schweitzer & Feldmann 2012), although †”Dromilite” H. Milne Edwards (1837a) is a nomen nudum and the two species “ Dromilites Milne Edwards ” of Reuss (1859) belong in other genera. However that may be, † Dromilites as listed by Schweitzer et al. (2010: 64) is likely paraphyletic. If † D. bucklandii is a podotreme, other members are eubrachyurans, such as † D. americana Rathbun, 1935 ( Rathbun 1935: 79, pl. 17, figs. 1–16), which was referred to the goneplacid † Tehuacana Stenzel, 1944 , by Armstrong et al. (2009) (see also Vega et al. 2008; and below). † Dromiopsis Reuss, 1858 , as listed by Schweitzer et al. (2010: 65), could also be paraphyletic, including several podotreme taxa, as well as dromiines, sphareodromiines and, perhaps, dynomenids.

The Cretaceous † Torynommatidae Glaessner, 1980 , constitutes one of the ten major clades of podotreme crabs listed by Karasawa et al. (2009: 80), contrary to its previous inclusion in Cyclodorippoidea by Glaessner (1980: 181), De Grave et al. (2009: 29), and Schweitzer et al. (2010: 78). Such an assignment to the cyclodorippoids would be possibly correct but only for some representatives. The family was ranked as a new podotreme section, †Torynommoida Karasawa, Schweitzer & Feldmann, 2011 ( Karasawa, Schweitzer & Feldmann 2011, as † Torynommidae ). The † Torynommatidae (see also Tucker et al. 1987; Fraaye & Collins 1996; Guinot & Tavares 2001; Collins 2003; Van Bakel, Jagt, Fraaije & Coole 2003) is clearly in need of a reappraisal. Its status must be ideally based on the location of the sexual female openings that unambiguously determine a podotreme or eubrachyuran nature. Although these orifices are generally not visible in fossils, the important podotreme-eubrachyuran dichotomy may be inferred from the carapace (often with a “human face”) and firmly based on the thoracic sternum and male abdomen. These traits are more often preserved than assumed in the fossils and available after the proper preparation of the fossil specimens, but they are sometimes not correctly interpreted.

The original spelling for the family “ Torynommidae ” by Glaessner (1980: 180) was corrected to “ Torynommatidae ” by Collins et al. (1995: 200) and Collins (2003: 85) based on the third edition of the Code (Art. 29(a), 29(b)(i)). So the nomen “ Torynommatidae ” was in usage for a time (see Karasawa et al. 2009: 80; De Grave et al. 2009: 29; Schweitzer et al. 2010: 78). But Schweitzer & Feldmann (2011b: 249), arguing that their “reading of the International Code of Zoological Nomenclature did not reveal any need to have done so”, retained usage of the original “ Torynommidae ” (B. van Bakel, pers. comm. 2012). Consequently, the spellings “ Torynommidae ”, “Torynommoidea” and “Torynommoida” were used by Karasawa, Schweitzer & Feldmann (2011: 523, 532, 534, 548, plates 1, 2), in our view a misleading interpretation of the Code. We actually agree with the correction by Collins et al. (1995: 200) in referring to the similar Articles of the Code in usage today (1999; Art. 29): “A familygroup name is formed by adding to the stem of the name of the type genus (Art. 29.3) or to the entire name of the type genus a suffix as specified in Article 29.2”. As “ Torynomma ” (etymology unknown) does not seem a Greek or Latin word, the Latinised suffix -IDAE for the family name must be added to the entire name and not “by deleting the case ending of the appropriate genitive singular”. Consequently, the familial nomen mut be formed based on the entire nomen “ Torynomma ”, hence “ Torynommatidae ”. Another case is the nomen “ Xeinostoma ”, based on the Greek “strange” (xeino) and “mouth” (stoma), for the cyclodorippid taxon Xeinostoma Stebbing, 1920 ( Stebbing 1920: 243) , which was first used for “ Xeinostominae ” (see Tavares 1992b), then corrected to “ Xeinostomatinae ” (see Tavares 1994a) following a recommendation by L.B. Holthuis (in litt. 1993).

As listed by Schweitzer et al. (2010: 78, 79; with various gender endings) the † Torynommatidae is certainly paraphyletic. † Torynomma Woods, 1953 , was listed with the type species † T. quadrata Woods, 1953 ( Woods 1953: 54, fig. 3, pl. 2, figs. 6, 7; Glaessner 1980: fig. 8), † T. dentatum Glaessner 1980 ( Glaessner 1980: 181, figs. 10, 11, 20); † T. australis Feldmann, Tshudy & Thomson, 1993 (Feldmann et al. 1993: 34, figs. 27, 28; Feldmann 2003: fig. 1.5), † T. flemingi Glaessner, 1980 ( Glaessner 1980: 181, fig. 9; Feldmann 1993: 206, figs. 10, 11), † T. planata Feldmann, 1993 (Feldmann 1993: 206, figs. 12, 13), and † T. densa Bishop, 1983 ( Bishop 1983a: 53) , only known by a chela.

† Torynomma quadrata , described with the female gonopore located on the P3 coxa, is a podotreme. As † T. quadrata is the type species of † Torynomma , itself onomatophore (type genus) of the † Torynommatidae , on the basis of this feature (on condition, however, that it was correctly identified), the family could be referred to the Podotremata. But the podotreme nature of the other species referred to † Torynommatidae is still doubtful, and some could be eubrachyurans, perhaps retroplumoids. † T. quadrata was described with a P5 “very slender, attached close to mid-line of carapace”, and two portions of the supposedly reduced last pereopod(s) are actually visible in the published photographs ( Woods 1953: 55, fig. 3a, pl. 2, fig. 6; Glaessner 1980: fig. 8; see also Glaessner 1969: R493). The subgenus † Paratorynomma Glaessner, 1980 , created for † T. dentata Glaessner, 1980 ( Glaessner 1980: 181, figs. 10, 11, 20), although not listed in De Grave et al. (2009) or in Schweitzer et al. (2010), could be an available generic nomen for the species allied to † T. dentata , if it reveals to be distinct. The ventral characters of other genera currently assigned to † Torynommatidae , such as † Dioratiopus Woods, 1953 (see Bishop 1985), will need to be known by their ventral characters before their precise classification can be ascertained.

† Hillius Bishop, 1983 (type species: † H. youngi Bishop, 1983 ), from the Middle Cretaceous (Lower Albian), first regarded as a dorippoid ( Bishop 1983a; De Grave et al. 2009; Schweitzer et al. 2010), thus as a eubrachyuran, has been subsequently considered a podotreme crabs and referred to Cyclodorippoidea , and even Cyclodorippinae ( Schweitzer & Feldmann 2011a: 5; Karasawa, Schweitzer & Feldmann 2011: 557), in which it was regarded to represent the oldest known member (see Fossil Dorippoidea ). In contrast, Van Bakel, Guinot, Artal, Fraaije & Jagt (2012: 205) assigned (with reservation) † H. youngi to † Orithopsis Carter, 1872 , thus considered † Hillius as a possible member of † Orithopsidae in the superfamily †Palaeocorystoidea.

Proposal for a tentative cladogram. The cladogram proposed for podotremes by Guinot & Tavares (2001: fig. 16) is here updated to include the following synapomorphies ( Fig. 41 ): homoliform press-button system formed by a single sternal structure on sternite 4 and by uropod modified as a socket (Homoliformia: living families and extinct † Mithracites ); “gymnopleure” press-button system formed by uropod modified into a socket and sternal structure on sternite 5 with two prominences (“double peg” in † Cenomanocarcinidae and † Palaeocorystidae or “hook projection” with two teeth in Lyreididae ; socket lost in the other Raninoidea , thus abdominal locking lost); eubrachyuran press-button system with a single sternal structure on sternite 5 (Eubrachyura); and uropod absent (Cyclodorippiformia only). This scheme supposes several different apparitions of the socket [(5), (9) and (12) in Fig. 41 ] but locking structures were actually present in †Etyoidea and †Dakoticancroidea (extinct) (see above under these names). The locking structure that is double in Gymnopleura (13), simple in Homoliformia (5) and Eubrachyura (9) was perhaps multiple in †Etyoidea and †Dakoticancroidea. The corresponding abdominal element (socket) is not known in the fossil taxa but it is supposed to have been large, as in the extant Lyreididae and in basal Eubrachyura such as Palicidae ( Fig. 50H ). The accumulating data on the ventral morphology of extinct crabs clearly show that the fixation of the abdomen was an early feature of the Brachyura . Other phylogenetic trees may be envisaged. One such alternative phylogeny is considering the brachyuran uropod, which is never biramous but always present, being transformed into a socket in all Brachyura after the branching of Dynomeniformia. The loss of the socket in Cyclodorippiformia and most Raninoidea is probably linked to the shortening of the abdomen, with, as a result, new locking systems evolving in Cyclodorippiformia and the abdomen being used for digging in Raninoidea .

Such a cladogram must be slightly modified if the socket embodies a unique development in Podotremata, i.e., being a synapomorphic character for Homoliformia, Gymnopleura (socket retained in Lyreididae , lost in most part of Gymnopleura ) and Cyclodorippiformia (socket lost), and if it secondarily appeared in Eubrachyura. This scheme leads to a Podotremata consisting of only two large groups, Dynomeniformia and Archaeobrachyura defined by Guinot (1977), the latter consisting of Homoliformia + Cyclodorippiformia + Gymnopleura . The departure of the branch Homoliformia in this cladogram must be at the base of the branch Cyclodorippiformia + Gymnopleura . Another tree, one which implies a unique development of the abdominal socket at the base of Brachyura when the biramous uropod gave way to a socket instead of a uropod as a ventral lobe or dorsal plate (Dynomeniformia), is not retained. This scheme represents a complete reorganisation of the brachyuran phylogeny. The monophyly of Brachyura is, however, supported by a uropod that is never biramous but always present (see Monophyletic Brachyura ) and the presence of a sella turcica.

A dorsal uropod is not lost in all eubrachyurans. There is a uropod in the more basal representatives of Hymenosomatoidea , as in Odiomaris Ng & & Richer de Forges, 1996 ( Fig. 29C–E ; Guinot 2011a) and Amarinus Lucas, 1980 , but it is excavated on the ventral surface of abdominal somite 6 thus transformed into a socket, instead of being lateral to the locking coxal prominence as in the Dromiidae . The hymenosomatoid system functions like the typical eubrachyuran press-button for abdominal locking. The Hymenosomatoidea , in which the dorsal uropod unambiguously bears a socket, would show that the uropodal transformation could have taken place very early (see Monophyletic Heterotremata: Superfamily Hymenosomatoidea ; Position of the Hymenosomatoidae within the Brachyura ; Affinities between Dorippoidea and Hymenosomatoidea ; Affinities between Dorippoidea and Hymenosomatoidea ).

Monophyletic Eubrachyura

The Eubrachyura shares the following synapomorphies ( Fig. 41 ): (1) female gonopores sternal (vulvae), on thoracic sternite 6 ( Figs. 3–6 , 42C , 48 ), a condition that is unique among Decapoda (Podotremata with female gonopores on P3 coxae and a paired spermatheca at the extremities of sternal sutures 7/8; Figs. 2 , 7 ; Guinot & Tavares 2001: 522, fig. 10; 2003: 47, 120); (2) uropod modified into a socket for the locking of the abdomen (Podotremata with uropod as ventral lobe, dorsal plate, sometimes socket, rarely lost); (3) abdominal locking of press-button type (Podotremata with various types of abdominal holding; see Guinot & Bouchard 1998); (4) exposure of a portion of the thoracic sternum on each side of the abdomen in adult males and immature females (unexposed thoracic sternum, at least laterally, in basal Podotremata); (5) sutures 4/5–7/8 crossing the sternal plate, either complete or medially interrupted ( Guinot 1979a), but - if medially interrupted - never as lateral or indistinct as in Podotremata (see Axial skeleton; Thoracic sternum); (6) presence of a well-defined sterno-abdominal cavity (sterno-abdominal depression in basal Podotremata and in Gymnopleura ; only a small, posterior cavity in Cyclodorippiformia) (see Sterno-abdominal depression and sterno-abdominal cavity); (7) tubular G1 with two foramina (a variously folded G1, with a single foramen in Podotremata); (8) presence of “eubrachyuran sella turcica” and of a junction plate (see Axial skeleton; Evolution of the axial skeleton in the Podotremata; Evolution of the axial skeleton in the Eubrachyura); (9) chelipeds generally with heterochely and heterodonty, sexually dimorphic, often with conspicuous handedness in males (generally homomorphic, not sexually dimorphic in Podotremata); (10) scaphognatite of the first zoea, different in podotreme and eubrachyuran families, and a particular type in Majoidea ( Van Dover et al. 1982: 48, 50, figs. 1, 2).

Assignment of the vulva to its appropriate level of generality (a particular node in the cladogram) (see Wiley 1981: 126) at the base of Eubrachyura ( Guinot 1977a, b, 1978a, 1979a; Guinot et al. 1994: fig. 7; Jamieson et al. 1995: fig. 1; Guinot & Tavares 2001: fig. 16; Dixon et al. 2003: 947; Ahyong et al. 2007: fig. 4), has led to the recognition of a monophyletic Eubrachyura. As already discussed, the perforation of the sternum by the oviduct to form a vulva is a major and unique event in the evolutionary history of the Decapoda . The formation of the vulva seems to have preceded the development of a sterno-abdominal cavity. Indeed, vulvae are present in crabs devoid of a true cavity or with an undefined cavity, being either medially located but not covered by the short abdomen as in Corystes ( Fig. 3B ) and Pseudocorystes ( Fig. 3D ), or placed laterally and exteriorly to the cavity as in Telmessus ( Fig. 3C ) and Bellia ( Fig. 3E, F ) ( Guinot & Bouchard 1998: figs. 12B, E, 14C).

Martin & Davis (2001: 74) included in the Eubrachyura their subsection “Raninoida” grouping Cyclodorippoidea and Raninoidea . In this conception the widely-agreed synapomorphy of the Eubrachyura (the vulvae) can no longer be considered as such because the Raninoida have coxal female gonopores (instead of sternal vulvae) and share paired spermatheca. Besides, none of the possible additional synapomorphies for the Eubrachyura Saint Laurent, 1980 (see above) can be used to support Eubrachyura of Martin & Davis (2001). The Eubrachyura, as conceived by Martin & Davis (2001), share no synapomorphies. Yet, Felder et al. (2009) and Schweitzer et al. (2010) followed the same classification.

Ahyong et al. (2007: 584) similarly disagreed with the monophyly of Podotremata but separated, however, all podotreme crabs from the Eubrachyura, which they recognised as having another degree of structural organisation. The question at stake is whether Cyclodorippiformia and Gymnopleura (which formerly constituted the podotreme subsection Archaeobrachyura) are more related to Eubrachyura than to basal Podotremata (i.e., Dynomeniformia and Homoliformia). In other words, whether or not the paired spermatheca (shared by all Podotremata) evolved once or more than once.

All known members of the Eubrachyura share an abdominal-locking mechanism that is unique in Decapoda ( Guinot & Bouchard 1998) . Crucial in the evolution of this mechanism was the dramatic transformation of the uropod into a socket ( Pérez 1928a, b, 1929). The uropod-socket homology, adopted by Hartnoll (1975: 16), Guinot (1979a), Guinot & Richer de Forges (1997), Guinot & Tavares (2001, 2003), and Ahyong & O'Meally (2004), corresponds to a major transformation. The uropod may also be highly modified in Podotremata ( Guinot & Tavares 2003: 110, table 1), as a salient, mobile dorsal plate involved in abdominal holding in Dromiidae , where it always acts together with an appendicular structure. In Dynomenidae (where the abdomen is only loosely applied on the sternum and restricted in its lateral movements, see McLay 1999), one of the subfamilies, the Dynomeninae , shows a sternal prominence lying side by side with the uropod that is not transformed into a socket, and therefore lacks a press-button system (Guinot 2008). In Homoliformia, where there is a true socket, the mechanism corresponds to a press-button system, but the sternal prominence is located on sternite 4 (“homoloid press-button” system).

Information on larvae could provide strong support to the uropod/socket homology. The presence of biramous uropods with both strongly setose endopods and exopods in all the megalopae of Homolidae (less setose in the more derived Latreilliidae ) ( Rice 1981b: 1006, 1009, fig. 2a, c, g, h)

Dorsal plates homologous to uropods are present in the basal Hymenosomatoidea ( Odiomarinae Guinot, 2011 ), being used as abdominal-locking systems ( Fig. 29C–E ; see Guinot 2011a, b; Monophyletic Heterotremata: Superfamily Hymenosomatoidea ; Position of the Hymenosomatoidea within the Brachyura ).

The eubrachyuran G1 has no equivalent in the other Decapoda . In contrast to the podotreme G1, which is partially opened or at least incompletely folded and has only one large aperture (foramen), the eubrachyuran G1 is fully tubular, closed, and has two basal foramina for both the penis and G2, the latter may be also laterally inserted. Tubulation of the G1 is often far from being complete among podotremes, being incompletely wrapped around, and there is only one basal foramen for the introduction of both the penis and G2.

The Brachyura , together with the Anomura , show a wide diversity in the morphology of their eyes and optical designs ( Porter & Cronin 2009: 192, fig. 4, table 1). The Dynomeniformia and Homoliformia share plesiomorphically with many reptant Decapoda compound eyes with square facets that use reflecting superposition optics, leading Fincham (1980: 741, table 1) and Cronin (1986: 13, table 1) to remove these two superfamilies from Brachyura and include them in Anomura . Because of a similar reflecting superposition, considered the plesiomorphic condition for the Decapoda , was also found in Latreillia elegans (Latreilliidae) and Ranina ranina (Raninidae) , Gaten (1998: 228, fig. 4, table1) concluded that all Podotremata have a similar eye type. In contrast, hexagonal facets are found in Eubrachyura, which have subsequently lost reflecting optics and use parabolic superposition. Such apposition optics in some Brachyura was considered by Richter (2002: 521) a secondary appearance corresponding to a “reverse” evolution. Such an appostion eye that characterises the hymenosomatoid Amarinus lacustris (see Meyer-Rochow & Reid 1994, as Halicarcinus lacustris ) is here interpreted, on the contrary, as a plesiomorphic character (see Position of the Hymenosomatoidea within the Brachyura ). It is noteworthy that Scholtz & McLay (2009: 426, fig. 10B–D) showed the existence of square facets in Dynomene pilumnoides (Dynomenidae) , as in the other basal Podotremata, and of hexagonal facets in Lyreidus tridentatus (Raninoidea) and Krangalangia spinosa (Cyclodorippidae) as in Eubrachyura, which contradicts Gaten's hypothesis. Compound eyes of fossil specimens supposed close to the oldest records of raninoids have been described with unique features including extremely large eyes relative to extant raninids, one fossil showing a facet type not observed in extant species (Luque et al. 2011). Re-examination of more representatives of podotremes and eubrachyurans should provide further insights into the evolution of the brachyuran eye.

All podotremes conceal themselves thanks to a burying or carrying behaviour (see Concealment behaviour: Carrying behaviour; Locomotion). The change from shallow, soft bottoms to an increased diversity of hard substrates with holes and other microhabitats as refuges during the Late Cretaceous (Maastrichtian), together with the appearance of new communities, predators, and food sources, has been determinant in brachyuran evolution (Fraaije 1996 a, b, 2003).

Monophyletic Thoracotremata

A monophyletic Thoracotremata such as in Ng, Guinot & Davie (2008) ( Table 7) has been supported as monophyletic by molecular data ( Tsang et al. 2008; Chu et al. 2009b; Wetzer et al. 2009; Palacios-Theil et al. 2009) and is largely accepted at the present time. Sperm morphology merits, however, inclusion of Thoracotremata, as a monophyletic assemblage, within Heterotremata ( Jamieson et al. 1995; Jamieson & Tudge 2000), the Heterotremata then being a synonym of Eubrachyura.

Clearly, the thoracotreme condition occurs in crabs having a widened posterior thoracic sternum with all sutures being interrupted (see Axial skeleton; Thoracic sternum; Carcinisation and its outcomes: Evolution of the axial skeleton in the Eubrachyura). The ejaculatory duct has no link to the P5 coxa in the thoracotreme disposition. Instead of perforating the P5 coxa and having a long penis to cover the distance between the coxa and the G1, the ejaculatory duct perforates the thoracic sternite 8 and emerges close to the G1. Thus, the penis, being never far from the G1, is never long in thoracotremes. The thoracotreme male gonopore occupies, however, various locations on sternite 8: either close ( Fig. 23A ) or more or less distant from the P5 coxa ( Figs. 1D , 23C–F ), and either close to suture 7/8 ( Figs. 1D , 23A, C–F ) or distant from both P5 coxa and sternal suture 7/8 ( Figs. 1E , 23B , 26 ) (see Modalities of penis protection; Sternal thoracotreme protection).

The penis emerges as a variously sclerotised structure, usually with a basal sheath and a soft, distal papilla that is concealed under the abdomen ( Figs. 26 , 33 ). In Varuna the papilla invaginates into sternite 8 wall, the sclerotised sheath acting as an operculum ( Fig. 36 ). Compared to the heterotremes, the thoracotremes are rather conservative regarding the modalities of the penis protection (see Modalities of penis protection). There is nevertheless variation in the distance between the sternal gonopore and the P5 coxa, even within the same family. In the gecarcinids Cardisoma ( Fig. 23A ) and Discoplax , for instance, the gonopore opens on sternite 8 very close to the P5 coxa, whereas in Gecarcinus the male gonopore opens far from the appendage. As a result, some authors (e.g., Türkay 1983b; Hendrickx 1998; Rodríguez 1992; Magalhães & Türkay 1996; von Sternberg et al. 1999; von Sternberg & Cumberlidge 1998, 2001a, b) questioned the validity of the Thoracotremata by regarding the heterotreme (male coxal gonopore) and thoracotreme conditions as part of a continuum connected by transitional forms. Furthermore, proximity between the penis and the P5 coxa led to uncertainties as to whether or not the ejaculatory duct passes through the P5 coxa before opening on thoracic sternite 8. As previously discussed, the penis of thoracotremes protrudes from thoracic sternite 8 and the ejaculatory duct opens to the exterior without a detour by the P5 coxa, regardless its proximity to the P5 coxa ( Figs. 26 , 33 , 35 ). Following H. Milne Edwards (1834, 1837a), Bouvier (1940: 274, figs. 167C, 176) erroneously considered the panopeid ( Rhithropanopeus ) as well as the goneplacid ( Goneplax ) organisations (with the penis sometimes within a depression) catometope but “intermediate”, a “passage” between the coxal and the sternal conditions.

A robust synapomorphy of Thoracotremata is a G2 that is always short, opposed to various patterns observed in the Heterotremata, varying from long, to intermediate length, to short.

A reduction of the last thoracic leg (P5) has actually never evolved among the thoracotreme families, a reduced P5 being only present in brachyuran families with coxal male gonopores ( Dorippoidea , Hexapodoidea , Palicoidea , Retroplumoidea ).

The Palicoidea , which is characterised by a coxo-sternal condition, cannot be considered a thoracotreme as in the classification of Martin & Davis (2001: 75) recognising 11 thoracotreme families grouped in three superfamilies (see Superfamily Palicoidea ; Affinities between Palicoidea , Retroplumoidea , and Hexapodoidea ).

There have been doubts about the position of Hexapodoidea , consisting of only one family, on account of the putative suppression of both P5 and sternite 8. Hexapodids, first included in Pinnotheroidea , subsequently in Goneplacoidea and ultimately in Xanthoidea (see Male gonopores among selected taxa of Eubrachyura: Family Hexapodidae ), were questionably included in Thoracotremata ( Guinot 1978a) and later transferred to Heterotremata ( Saint Laurent 1989; Guinot & Richer de Forges 1997; Guinot & Bouchard 1998). As shown here, hexapodoids are clearly heterotremes, with vestigial P 5 in the males. This contradicts the hypothesis of von Sternberg & Cumberlidge (2001a: 332, 333) that “hexapodids arose from within the Thoracotremata with the reduction of the last sternite altering the topographical relationship of the seminal ducts” and are “a thoracotreme clade, albeit a highly derived assemblage”. The alternative hypothesis that Hexapodidae may be “thoracotremes derived from heterotremes” (von Sternberg & Cumberlidge 2001a: 332) is no longer supported (see Affinities between Palicoidea , Retroplumoidea , and Hexapodoidea ).

Aphanodactylidae Ahyong & Ng, 2009

Asthenognathidae Stimpson, 1858

Camptandriidae Stimpson, 1858

? Cryptochiridae Paul'son, 1875

Dotillidae Stimpson, 1858

Gecarcinidae MacLeay, 1838

Glyptograpsidae Schubart, Cuesta & Felder, 2002

Grapsidae MacLeay, 1838

Heloeciidae H. Milne Edwards, 1852

Macrophthalmidae Dana, 1851

Mictyridae Dana, 1851

Ocypodidae Rafinesque, 1815

Percnidae Števčić, 2005

? Pinnotheridae De Haan, 1833

Plagusiidae Dana, 1851

Sesarmidae Dana, 1851

Ucididae Števčić, 2005

Varunidae H. Milne Edwards, 1853

Xenograpsidae N.K. Ng, Davie, Schubart & P.K.L. Ng, 2007

Xenophthalmidae Stimpson, 1858

The recent establishment of the two families Brankocleistostomidae Števčić, 2011 , and Lazarocleistostomidae Števčić, 2011 ( Števčić 2011) , listed as incertae sedis by Ahyong et al. (2011: 189), was regarded as unjustified by Ng (2012), whereas Garthopilumnidae Števčić, 2005 , also listed as incertae sedis by Ahyong et al. (2011: 189), was regarded as an unavailable name by Ng, Guinot & Davie (2008: 139, 144).

The controversial Hymenosomatoidea was similarly removed from Thoracotremata by Guinot & Richer de Forges (1997) and Guinot & Bouchard (1998), an action followed by Martin & Davis (2001: 52), Števčić (2005), Naruse, Ng & Guinot (2008), Ng, Guinot & Davie (2008: 10). In the new interpretation proposed by Guinot (2011a, b) sternal male gonopores have evolved homoplasically in the Hymenosomatoidea and in Thoracotremata. The supposition that sternal male gonopores occur synapomorphically in the Thoracotremata and are found nowhere else among the remaining decapod crustaceans is therefore no longer supported (See Monophyletic Heterotremata: Superfamily Hymenosomatoidea ; Position of the Hymenosomatoidea within the Brachyura ; Affinities between Dorippoidea and Hymenosomatoidea ).

It is necessary in this context to examine the case of Pinnotheroidea and Cryptochiroidea , both with sternal male gonopores and thus generally assigned to Thoracotremata, although assigned to Heterotremata by Guinot & Richer de Forges (1997: 496, table 1), Guinot & Bouchard (1998: 654), and Martin & Davis (2001: 39). Števčić (2005: 115, 116, 120–132) followed this arrangement and, consequently, recognised only six thoracotreme families (with numerous subfamilies and tribes) grouped in three superfamilies. Ng, Guinot & Davie (2008: 30, 212–254), in contrast, listed as thoracotremes 16 families and four superfamilies, including Pinnotheroidea and Cryptochiroidea . A possible misinterpretation of the origin and evolution of these two groups of symbiotic crabs is cautiously envisaged here, and a tentative placement at the base of the Heterotremata is briefly discussed (see Position of the Cryptochiroidea and Pinnotheroidea within the Brachyura ). Molecular analyses, support, however, Thoracotremata (including Pinnotheroidea and Cryptochiroidea ) as monophyletic. An apparently unpublished thesis ( Anonymous 2010) does not support their status as superfamilies and suggests their affiliation with Ocypodoidea and Grapsoidea, respectively.

Guinot & Richer de Forges (1997) remarked that Thoracotremata, when Cryptochiridae and Pinnotheroidea are excluded, are for the most part limited to littoral, amphibious, and terrestrial crabs. Paulay & Starmer (2011) suggested, if the Thoracotremata is indeed monophyletic, that the thoracotreme taxa essentially inhabited “safe places” such as the intertidal, terrestrial and freshwater habitats, deep water, and the confinement within hosts in symbioses (as in Cryptochiridae and Pinnotheroidea ) rather than in the more congested and competitive shallow subtidal. They advanced the hypothesis that the diversification of thoracotremes has been facilitated by their adaptability to novel habitats avoiding the more intense competition of the shallow subtidal. Thoracotremes such as members of the Plagusiidae and Percinidae, however, do extend their vertical distribution to the shallow subtidal, as in the case of at least seven species of the plagusiid Euchirograpsus , one of which has been recorded as deep as 430 m (N.K. Ng & Martin 2010: table 1).

The thoracotreme families Camptandriidae , Gecarcinidae , Glyptograpsidae , Heloeciidae , Percnidae , Plagusiidae , Sesarmidae , Ucididae , and Varunidae are reviewed here in detail. The monophyly of Thoracotremata does not imply that these families are monophyletic. The configuration of the male gonopore and the penial region in these families should nevertheless be useful for comparative purposes.

Phylogenetic relationships within the Thoracotremata are far from clear. Data from larval stages reveal divergent characteristics suggesting the paraphyly of higher taxa. As an example, larval characters were used to transfer Ilyograpsus Barnard, 1955 , from Grapsoididea to Ocypodoidea ( Fukuda 1978; Cuesta et al. 1997; Cuesta 1999; Flores et al. 2003), the genus being subsequently raised to a subfamilial rank, the Ilyograpsinae Števčić, 2005, within Macrophthalmidae ( Ng, Guinot & Davie 2008: 237, 238; see also Sawada et al. 2005; Komai & Wada 2008). Nuclear and mitochondrial sequences ( Tsang et al. 2009: 81) partially support the monophyly of the ocypodoid and grapsoid families (both polyphyletic) but suggest “merging ocypodoid and grapsoid families under a single taxon within the Thoracotremata”, leaving aside longstanding morphological evidence.

Indeterminate growth (i.e., female moulting continuing indefinitely after maturity) is accompanied in all Thoracotremata by hard-female mating, an adaptive strategy for both sexes. This characteristic, however, is not a thoracotreme synapomorphy since fully aquatic heterotreme crabs with indeterminate growth have also evolved hard-female mating ( Hartnoll 2000).

The absence of thoracotremes in the Cretaceous fauna and the appearance of many families in the Eocene were documented by Schweitzer & Feldmann (2005: 26, 40).

Many thoracotremes have colonised terrestrial environments ( Burggren & McMahon 1988; Greenaway 1999; see also Cannicci et al. 2011) but no true, obligate freshwater crabs belong to the Thoracotremata (see Freshwater crabs families). The assignment by Schram (1986: 308) of nine freshwater families (including Trichodactylidae ) to Thoracotremata and two ( Deckeniidae and Potamidae ) to Heterotremata, has not been generally accepted by modern workers. Martin & Davis (2001: 76) kept all freshwater families (eight in total) in the Heterotremata. Likewise, Števčić (2005: 72–78, 88–89) included all freshwater crabs in Heterotremata, with many subfamilies and tribes grouped in two superfamilies, the Potamoidea and Trichodactyloidea for Trichodactylidae . Ng, Guinot & Davie (2008: 27–29) similarly considered all freshwater crabs families as heterotremes, with six families grouped in four superfamilies, the Trichodactyloidea ( Trichodactylidae ) at the same rank as the three others. The non-thoracotreme condition of all freshwater crabs is supported herein.

Monophyletic Heterotremata

Heterotremata was actually based by Guinot (1977a, 1979a) on the combination of two symplesiomorphies, sternal female gonopore and coxal male gonopore, thus with a double position (segmental-sternal and appendicular). Correct assignment of the coxal male gonopore and sternal female gonopore (vulva) at the base of Decapoda and Eubrachyura, respectively, left no synapomorphies to support the Heterotremata ( Fig. 41 , Tables 3, 4). The Heterotremata nevertheless should be regarded as the sister group to Thoracotremata, at least for now, as proposed by Saint Laurent (1980; see also von Sternberg & Cumberlidge 2001a: 334). Molecular analysis using nuclear protein-coding genes support Heterotremata and Thoracotremata being reciprocally monophyletic ( Tsang et al. 2008; Chu et al. 2009a, b). The Heterotremata was resolved with a rather good support in analyses based on arginine kinase sequences ( Mahon & Neigel 2008). The Heterotremata is a also a group recognised in paleontological papers (see De Grave et al. 2009; Schweitzer et al. 2010). The current conceptual basis for Heterotremata, although typological (see also Sternberg & Cumberlidge, 2001a), is useful in providing a nomenclatural framework. Whether the heterotreme (as a typological distinction) will also prove to be a monophyletic group (Heterotremata) is an open question. Based on the spermtozoal ultrastructure, Jamieson (1991, 1994) and Jamieson et al. (1995) have argued for inclusion of Thoracotremata (as a monophyletic assemblage) within the Heterotremata sensu lato, the latter being synonym with Eubrachyura ( Jamieson & Tudge 2000: 51).

The term “heterotreme” is herein used as an adjective to refer to eubrachyuran crabs with both vulvae and coxal male gonopores. Heterotremes cannot be considered “more derived than many other brachyurans” as stated by Schweitzer & Feldmann (2010b: 167, 168), an assertion that should be withdrawn if heterotremes and thoracotremes prove to be sister taxa.

Numerous opinions have been expressed as to the taxonomic position of the brachyuran families. Actually, higher-ranked relationships among the heterotreme crabs remain largely unknown. Whereas a number of heterotreme superfamilies have been recognised, the monophyletic status of most of them is far from clear. Martin & Davis (2001: 74, 75) recognised 16 superfamilies of extant heterotremes grouping 49 families. These superfamilies roughly corresponded to the traditional classification based on morphological data and in many cases, despite all efforts, do not reflect monophyletic groups. Števčić (2005) added another 25 extant heterotreme superfamilies to the list of Martin & Davis (2001) and, without much explanation or strong evidence of monophyly, recognised 44 superfamilies (20 new) for 97 suprageneric taxa (tribes, subfamilies, and families). Many are monotypic groups resulting in redundant taxa. In a more conservative approach and in order to reach a consensus, Ng, Guinot & Davie (2008: 26–30) recognised 28 heterotreme superfamilies embracing 66 families. Here, following a rigorous approach based on comparative morphology involving dissections of extensive material belonging to selected families, 66 heterotreme families (excluding freshwater crabs families) are recognised ( Table 3; see also Ahyong et al. 2011). Only the superfamilies for which a monophyletic status is not reasonably doubted are discussed below.

Superfamily Cancroidea . The taxonomy of Cancroidea has been quite unstable. The inclusion of the monotypic Atelecyclidae (Atelecyclus) in Cancroidea , an idea suggested by Hong & Ingle (1987), was pointed out by Guinot, De Angeli & Garassino (2008: 25, 29) and followed by Ng, Guinot & Davie (2008: 51). Based on two molecular phylogenies, Schubart & Reuschel (2009: 543, figs. 1–3) recovered the same Cancroidea restricted to two families, Cancridae and Atelecyclidae . All other families and genera traditionally assigned to Cancroidea by Martin & Davis (2001: 74), De Grave et al. (2009: 30), and Schweitzer et al. (2010: 100) must be removed. Trichopeltarion A. Milne-Edwards, 1880, and allied genera were excluded from Atelecyclidae and Cancroidea ( Guinot, Vega & Van Bakel 2008), and referred to a new family and superfamily, Trichopeltariidae Tavares & Cleva, 2010 , and Trichopeltarioidea ( Tavares & Cleva 2010).

The evolution of Cancer , using morphological, mitochondrial sequence, palaeontological, and biogeographical information, was given by Harrison & Crespi (1999) but a careful re-evaluation has resulted in the separation of the genus into several genera, some of which new, distributed in two subfamilies: Cancrinae Latreille 1802 , with living and fossil representatives, and (extinct) † Lobocarcininae Reuss, 1857 ( Schweitzer & Feldmann 2000c; see also Feldmann et al. 2006; Fraaije et al. 2010). Cancer is “one of the ‘oldest’ names in carcinology”, and an up-to-date classification of cancrids has been provided by Schram & Ng (2012).

The Pirimelidae , considered a cancroid by Ng, Guinot & Davie (2008), is here assigned to the Portunoidea (see also Tavares & Cleva 2010: table 1). The Thiidae (for the date of publication of the type genus Thia Leach, 1816 , see Ng & Low 2010: 39) is also assigned to Portunoidea based on morphological features, a relationship confirmed by Schubart & Reuschel (2009: table 4), who recognised the family as close to Carcinidae and Polybiidae .

Superfamily Carpilioidea . A monotypic superfamily, including the only living representative Carpilius , was proposed by Karasawa & Schweitzer (2006: 42, fig. 3), a scheme followed by Ng, Guinot & Davie (2008: 54). A diagnostic character of the family is the short and robust cheliped that shows a fusion of the basis-ischium with the merus and a direct articulation of the coxa with the merus, a feature present in Carpilius (Recent and fossil) and at least † Palaeocarpilius A. Milne-Edwards, 1862, and † Ocalina Rathbun, 1929 ( Guinot 1968b: 167, fig. 9, pl. 1; 1968c: 323; Vega et al. 2010: pl. 1, fig. 7; see also Laughlin et al. 1983; Clark et al. 2005). The monophyly of Carpilioidea Ortmann, 1893 , is questionable when the three extinct families † Paleoxanthopsidae Schweitzer, 2003 , † Tumidocarcinidae Schweitzer, 2005 , and † Zanthopsidae Vía Boada, 1959 , are added to Carpiliidae ( Karasawa & Schweitzer 2006: 42; De Grave et al. 2009: 30; Schweitzer et al. 2010: 114).

Placement of Liagore De Haan, 1833 , within Carpiliidae ( Serène 1968: 76; Sakai 1976: 388) is not supported by the morphology of its genital region. The gonopore and the proximal portion of the penis of Liagore are not visible, being covered by the wide sternite 7, which largely extends over the anterior margin of the coxa; moreover, a small portion of sternite 8 is exposed ( Guinot 1971b: 1096, fig. 5), which is not the case in Carpilius . Liagore was formerly included in Xanthinae MacLeay, 1838 ( Ng, Guinot & Davie 2008: 203; see also Guinot 1978a: 267; Davie 2002: 135; Ng & Naruse 2007a), but this subfamilial assignment needs a reappraisal. Molecular data ( Wetzer et al. 2003: 418) support a distinct and monophyletic Carpiliidae , excluding Liagor e. The diagnosis of Liagorinae Števčić, 2005, which can accommodate not only Liagore but also Demania Laurie, 1906 ( Lai et al. 2011), must be revised.

Superfamliy Cheiragonoidea . Ng, Guinot & Davie (2008: 55) and De Grave et al. (2009: 31) removed the family Cheiragonidae (see Števčić 1988) from Corystoidea to which it had been referred ( Bouvier 1942) and from Cancroidea ( Martin & Davis 2001: 74; Števčić 2005: 32; Vega et al. 2008: 56; Schweitzer et al. 2010: 103), and raised it to the superfamilial rank, Cheiragonoidea Ortmann, 1893 . Indeed, the Cheiragonidae is one of the rare eubrachyuran families in which vulvae open outside the sterno-abdominal cavity, being exposed on the external sides of sternite 6 and thus close to the P3 coxae. The vulvae are even located in a strong notch that cuts the external margin of abdominal somite 6 in Telmessus cheiragonus ( Fig. 3C ; Guinot & Bouchard 1998: fig. 13C; see Female sternal gonopores, or vulvae). It is not known whether the exposure of the vulvae in both Cheiragonidae (vulvae located on each side of the abdominal somite 6) and Corystidae (vulvae medially located well beyond the telson as a result of the shortening of the abdomen linked to burying behaviour; Fig. 3A, B ) evolved independently.

In Cheiragonidae , exemplified here by Telmessus ( Figs. 3C ), the thoracic sternum and axial skeleton are regularly compartmented as in Corystidae ( Figs. 3A, B , 56A ). In contrast to Corystes where sutures 2/3, 4/5–7/8 are complete, sutures 4/5 and 5/6 are incomplete in T. cheiragonus , although with short interruption points, and sutures 5/6 and 7/8 are complete. This could be sternal pattern 3 ( Fig. 56C ) except that the median line extends along sternites 6–8 as in sternal pattern 2 ( Fig. 56B ). The median line is actually discontinuous, showing as a deep invagination at sternite 6 level, but continuous along sternites 7, 8. The corresponding median plate is well-defined on each of sternites 6–8, high, becoming abruptly oriented vertically at suture 6/7 level; the thick, elevated phragmae 4/5–7/8 medially lean on each side of the median plate; the sella turcica is particularly long.

Larval characters suggest the separation of Cheiragonidae and Corystidae , both originating from a primitive stock. The combination of unique zoeal characters of Erimacrus and Telmessus , indicating “that they may have abbreviated a longer ancestral series of zoeal stages”, does not obviously ally them with any known group ( Rice 1980: 336, 358, fig. 47; 1981a: 292, 293; Sasaki & Mihara 1993: 511). See Superfamily Corystoidea below.

† Montezumella Rathbun, 1930 (type species: † M. tubulata Rathbun, 1930 ; from the Eocene of Baja California, México) was believed to be an atelecyclid [which included Erimacrus in previous times ( Rathbun 1930b: 4, pl. 2, fig. 3; Vía Boada 1970: 12, figs. 1, 2; Beschin et al. 1994: pl. 6, fig. 4; De Angeli 1995: 14, fig. 3.1, pl. 1, fig. 7, pl. 2, fig. 1)] and was subsequently assigned to Cheiragonidae ( Portell & Collins 2002: 595, fig. 2; Beschin et al. 2002: 16, fig. 11, pl. 3, fig. 1a, b; Schweitzer & Salva 2000; Salva & Feldmann 2001, with reservation; Collins & Donovan 2005: 3, pl. 1; Schweitzer et al. 2010: 103; Busulini et al. 2012: 55). The narrow male thoracic sternum preserved in † M. amenosi Vía Boada, 1959 ( Vía Boada 1970: 12, fig. 2) is close to the cheiragonid disposition, but the median line that is figured on sternite 4 does not conform to the configuration of Cheiragonidae that is without a median line on sternite 4. It is probable that the thoracic sternum of females, when known in † Montezumella , will be not that of a cheiragonid (A. De Angeli, pers. comm.). The carapace posterolateral margins of the different species assigned to † Montezumella seem devoid of teeth, whereas they bear several teeth in the two extant cheiragonid genera Erimacrus and Telmessus . It will be particularly worthy to know the ventral surface and especially the vulvae of † Karasawaia Vega, Nyborg, Coutiño & Hernández-Monzón, 2008 , placed in Cheiragonidae by Vega et al. (2008: 56, fig. 3, pl. 2, fig. 13) and De Grave et al. (2009: 31) (see Female sternal gonopores, or vulvae: Family Cheiragonidae ). The recent finding of vulvae located as usual inside the sterno-abdominal cavity and, as such, completely covered by the abdominal somites (instead of exposed in Cheiragonidae ), and the presence of other differentiating characters in † Montezumella and close fossil genera support the establishment of a separate family for these taxa (Osso et al. 2013).

Superfamily Corystoidea . The Corystoidea , originally a large group ( Bouvier 1942), is now restricted to the family Corystidae (see Anonymous 2009, Opinion 2238), with only three certain Recent genera, Corystes , Jonas , and Gomeza ( Guinot & Bouchard 1998: 644; Ng, Guinot & Davie 2008: 56; De Grave et al. 2009: 31), perhaps also Pseudocorystes (Guinot et al. manuscript). The sperm ultrastructure of C. cassivelaunus may be regarded plesiomorphic, the perforated operculum being an unusual condition for the Heterotremata and known only in podotremes and in eubrachyuran cancroids and majoids ( Jamieson et al. 1997).

In Corystes both the thoracic sternum, which corresponds to sternal pattern 1 (i.e., with complete sternal sutures 2/3, 4/5–7/8; Figs. 3A, B , 56A ), and the skeleton are peculiar. The anterior sternites are well delimited; each of sternites 5 – 8 is individualised, separated by a strong demarcation, and a median depression being present at each interruption point of sutures 4/5–7/8; the median line is long, partially extending on sternite 4; the median plate is similarly long, prolonging on sternite 4, becoming progressively and regularly higher from sternites 5–8; the axial skeleton ( Gordon 1966: 350, figs. 5, 6) is regularly compartmented, with the thick and high phragmae medially leaning on each side of the median plate. The narrowing of the body of Corystes is a burying adaptation

The Corystoidea and Cheiragonoidea are heterotreme families that show the following plesiomorphic features: thoracic sternum rather narrow, horizontally metamerised and longitudinally grooved by a median line; anterior sternites forming a well-developed triangle; first abdominal somites in dorsal position; vulvae exposed; G2 long ( Guinot 1979a: 88, 136, 178, fig. 65 E, F, pl. 25, figs. 1–3; Guinot & Bouchard 1998: 646, fig. 13C, D; Tavares & Cleva 2010: table 1). Cheiragonoids differ, however, by the location of their extra-abdominal vulvae (see Monophyletic Heterotremata: Superfamily Cheiragonoidea , above; Female sternal gonopores, or vulvae) and by several features of thoracic sternum, median line, and axial skeleton including the median plate.

Larvae of C. cassivelaunus (see Ingle & Rice 1971) show some similarities with those of the Cancridae although their zoeae differ in several important respects. The presence of numerous natatory setae on the maxillipeds in its later stages are shared with cheiragonoid zoeae, perhaps not a significant trait according to Rice (1980: 333) (See Superfamily Cheiragonoidea above).

According to Schubart & Reuschel (2009: 543, 546, figs. 1–3, table 4), based on two molecular phylogenies, the Cheiragonidae and Corystidae clustered together without an absolute support, appearing to hold a basal and unrelated position to all other Brachyura , and should constitute sister families in a “potentially monophyletic assemblage”, preferentially named Corystoidea . The two groups are almost certainly relatively basal in the Heterotremata.

Some fossil genera known from the fossil record show strong similarities with the extant genus Corystes and, although the ventral parts remain unknown, their assignment to the Corystidae is almost certain. † Micromithrax holsatica Noetling, 1881 , from the Lower to the Middle Miocene, shows a combination of morphological features of Corystes , Jonas , and Gomeza , and we agree with its placement in Corystidae as did Van Bakel, Artal, Fraaije & Jagt (2009). The Miocene † Gomezinus Collins in Collins, Lee & Noad 2003 (2003: 211, pl. 4, fig. 2), with nine lateral teeth, appears to be a member of Corystidae close to Jonas and Gomeza . † Corystites Müller, 1984 , with C. latifrons (Lôrenthey in Lôrenthey & Beurlen 1929), from the Miocene of Hungary (Lôrenthey in Lôrenthey & Beurlen 1929) as type species, and C. vicetinus De Angeli, Garassino & Ceccon, 2010 (De Angeli et al. 2010: 158, fig. 9), from the early Oligocene of Vicenza, are also typical corystoids. The supposed Middle Jurassic (Upper Bathonian) stratigraphic age of † Hebertides jurassica Guinot, De Angeli & Garassino, 2007 ( Guinot et al. 2007b) showed to be erroneous, the analysis of the bryozoans attached to the matrix indicating a Miocene age ( Taylor et al. 2012).

Superfamily Dorippoidea . Dorippe and Ethusa had been grouped together under Dorippidae since H. Milne Edwards (1837a: 151, as Dorippiens ). Even as new genera have been erected, this grouping has never been questioned. The differences between Dorippinae and Ethusinae (see key in Castro 2005: 505) are of such a high hierarchical rank that these two subfamilies warrant their separation as two families, Dorippidae and Ethusidae . Despite the fact that Ethusinae was elevated to a familial rank, it is justified to recognise a monophyletic status to the superfamily Dorippoidea receiving the two families ( Ng, Guinot & Davie 2008: 59).

Evidence for the monophyly of the Dorippoidea comes from several putative synapomorphies: (1) similar arrangement of the cephalic appendages; (2) respiratory system with the pre-chelipedal afferent branchial openings receiving the developed, calcified mxp3 coxa, but being distinctive in the two families: Dorippidae with ovate, elongated Milne Edwards openings that are separated from the chelipeds due to the fusion of the sternite 3 with pterygostome (thoracic sternum/pterygostome junction) ( Fig. 42C; H . Milne Edwards, Atlas , pl. 20, fig. 12; Ihle 1916: fig. 45); Ethusidae without sternum/pterygostome junction, the Milne Edwards openings being contiguous to the chelipeds, which is the most common arrangement in Eubrachyura ( Fig. 42A ); (3) closure of the endostomial gutter by the mxp1, their calcified endopods forming the floor of the gutter, the efferent openings being at the extremity of the forwards produced endostome; (4) presence of sternal extensions (lateral outgrowths of the sternites) joining the thoracic sternum to the carapace at the sutures 4/5–6/7 levels (each time formed by two consecutive sternites, thus being double, also joined to the corresponding pleurites) and inserted between P1 and P2 as well as between P2 and P3, and between P3 and P4 respectively, so that their arthrodial cavities are encircled by these sclerites ( Figs. 42A, C , 46A, B , 47A, B ; Guinot 1979a: 103); these extensions being much narrower in Ethusidae ( Fig. 42A ) than in Dorippidae ( Fig. 42C ); (5) thoracic sternites 7 and 8 perpendicular to the preceding sternites; (6) the P4, P5 arthrodial cavities not aligned to the preceding arthrodial cavities; (7) P4 and P5 strongly reduced, dorsally positioned over the carapace, and subchelate or hook-like for carrying behaviour; (8) first and second abdominal somites plus the half portion of the third somite not folded under cephalothorax; (9) G1 with a long protopodite; (10) presence of a long penial tube; (10) sperm ultrastructure ( Jamieson & Tudge 2000: 52, fig. 11B, C). The dorippid press-button is located in a marked curve of suture 5/6, a character already present in the first larval stage ( Quintana 1987: fig. 3D), whereas the ethusid button is located “at edge of thoracic sternite 5” and more or less close to the straight suture 5/6 ( Castro 2005).

The male genital region of the Dorippoidea shows a transformation series of penis protection found nowhere else in Eubrachyura, with several modalities in Dorippidae ( Figs. 15–19 ) and a more uniform pattern in Ethusidae ( Figs. 20–22 ). In Dorippidae ( Fig. 42C ) and Ethusidae , mature females preserve the well-developed and functional components of the abdominal-holding system (press-buttons and sockets), permitting an efficient locking.

The relative position of the male gonopore and the lateral foramen of the G1 is a major factor in determining the length of the penis. In Dorippidae the male gonopore and the lateral foramen of the G1 are close to one another in relation to the longitudinal body axis but are situated far from each other in relation to the horizontal body axis. As a result, the dorippid penis, at least its distal half, is vertically oriented, crossing from sternite 8 into sternite 7, to ensure contact between the gonopore and the G1 ( Figs. 15–19 ). In contrast, in Ethusidae ( Figs. 20–22 ), and as a response to the widening of the body, the male gonopore and the G1 lateral foramen are far from one another in relation to the longitudinal body axis but situated close to one another in relation to the horizontal body axis. The ethusid penis is not vertically oriented, never at an angle, being always situated along sternite 8 inside a gutter formed by a partial invagination of the anterior margin of sternite 8, close to its junction with sternite 7, as in Figure 14.

A molecular analysis based on the sequence data from 16S rRNA gene of five species supported the recognition of the Dorippidae as a monophyletic family with two main clades ( Fan et al. 2004). A more complete phylogenetic tree inferred from three mitochondrial genes (16S rRNA, 12S rRNA, and cytochrome c oxidase subunit I) by Sin et al. (2009: fig. 2) also supported the Dorippidae and Ethusidae as monophyletic families within the Dorippoidea , and recognised several dorippid clades. These clades concur with the groupings based on morphological features, including carapace, antennae, G1, and male abdomen (Guinot & Lai, study in progress).

Rice (1980: 318, 319, fig. 22) concluded that Dorippoidea , based on its unique combination of zoeal characters, has had “an evolutionary history distinct from that of all the brachyuran families”. The “degree of advancement” of the dorippid zoeae, seemingly similar to that of thoracotreme families and shared by Inachinae MacLeay, 1838 ( Rice 1980: 317, 319), corroborates our view they belong to an ancestral heterotreme clade (see Position of the Dorippoidea within the Brachyura ), as also is the inachid clade. Based on zoeal characters, Martin & Truesdale (1989: 215, 216, as Ethusinae and Dorippinae ) referred to the “naturalness” of the Dorippoidea . The larval characters of Ethusidae , although different from those of the Dorippidae ( Aikawa 1937; Gilet 1952; Kurata 1964; Terada 1981; Quintana 1987), do not contradict its inclusion in the same group as Dorippidae , the features that unite the two families distinguishing them from all other Brachyura ( Martin & Truesdale 1989; Paula 1987a, b; 1991: table 1, as Ethusinae and Dorippinae ). Our knowledge of the larval and postlarval development of Dorippidae was enhanced by Quintana (1987), who demonstrated the presence in the megalopa of carrying behaviour with objects held dorsally by the fully formed subchelate endings of P4 and P5. The existence of several morphologically distinctive dorippid lineages may be recognised already in the early juvenile stages.

The spermatozoal ultrastructure of Dorippoidea ( Jamieson & Tudge 1990: 347–354; 2000: 52; Jamieson 1994: 390–391; Jamieson et al. 1995: 278) demonstrates that it is the least modified member of the Heterotremata (see Position of the Dorippoidea within the Brachyura ).

It is noteworthy that dorippids and ethusids share with hexapodids, retroplumids, and palicids the reduction of their P5. The two dorippoid families are, with Palicidae , the only families using carrying behaviour (see Superfamily Orithyioidea below; Affinities between Palicoidea , Retroplumoidea , and Hexapodoidea ; Affinities between Dorippoidea and Hymenosomatoidea ; Concealment behaviour: Carrying behaviour). Some pinnotheroid crabs (e.g., Parapinnixa Holmes, 1895 , Alarconia Glassell, 1938 , Glassella Campos & Wicksten, 1997 , Austinixa Heard & Manning, 1997, Pinnixa , at least pro parte) have noticeably reduced P5.

The coexistence in Dorippoidea of many plesiomorphies, such as the disposition of eyes and cephalic appendages, the absence of orbits (see Cephalic condensation), the dorsal position of the first abdominal somites (therefore with an elongated G1 protopodite), the posterior thoracic curvature (P4 and P5 with arthrodial cavities that are not aligned with preceding ones), and of the more derived coxo-sternal condition is indicative of a long evolutionary history. The Dorippoidea , with their numerous character states, may serve as a model for the understanding of the coxo-sternal condition among heterotremes (see Coxo-sternal condition; Modalities of penis protection: Coxo-sternal penial tube; Position of the Dorippoidea within the Brachyura ; Affinities between Dorippoidea and Hymenosomatoidea ). The broadening of the thoracic sternum, at least posteriorly, of heterotremes has caused changes in the condition of the genital region, leading to a longer ejaculatory duct, and modified penis and male gonopores (see Carcinisation and its outcomes).

Fossil Dorippoidea . The dorippoids, known by several certain records, perhaps branched-off as early as the Cretaceous, are among the basal heterotreme crabs, their long evolutionary history being attested by the morphology of extant members (see Position of the Dorippoidea within the Brachyura ). The identity of the fossil genera must be re-evaluated in taking into account the reappraisal of the living dorippids by Holthuis & Manning (1990). The difficulty in only trusting on the carapace is exemplified by the case of † Hillius , from the Middle Cretaceous (Lower Albian). It was described in Dorippinae because of a “striking resemblance in carapace shape to Dorippe ” ( Bishop 1983a: 46, fig. 8C, pl. 1, figs. 8–11) and assigned to Dorippidae by Schweitzer et al. (2010: 79) and De Grave et al. (2009: 31), therefore considered to be eubrachyuran. † Hillius was subsequently regarded as a podotreme by Schweitzer & Feldmann (2011a: 5), stressing relationships with Cyclodorippoidea , and by Karasawa, Schweitzer & Feldmann (2011: 557) who subscribed to the view that † Hillius was the oldest known cyclodorippoid. Van Bakel, Guinot, Artal, Fraaije & Jagt (2012: 205) also regarded † Hillius as a podotreme but assigned it with reservation to † Orithopsidae in the superfamily †Palaeocorystoidea (see Other extinct putative Podotremata).

Recent genera and species have fossil representatives dating back from the Upper Eocene (Karasawa 2000: table 1), and, although they clearly belong to the Dorippidae , their taxonomic placement may need further study. Dorippe rissoana Desmarest, 1817 ( Desmarest 1817: 509; 1822: 119, pl. 10, figs. 1–3), from an unknown origin, is a doubtful fossil according to Desmarest himself, and it is very close to, if not synonym with Dorippe quadridens . † Medorippe ampla Garassino, De Angeli, Gallo & Pasini, 2004 ( Garassino et al. 2004: 260, figs. 5–7; De Angeli & Garassino 2006b: 40; De Angeli et al. 2009: 174, fig. 6), from the Upper Miocene of Italy, and † M. cf. ampla from the early Pliocene of Italy ( Garassino et al. 2012: 27) show the enlarged carapace and the anterolateral spine characteristic of this extant genus; the generic status of † M. tanabei Karasawa, 2000 , from the Miocene of Japan, is less certain (Karasawa 2000: fig. 2). The carapace of † D. margaretha Lôrenthey in Lôrenthey & Beurlen 1929, from the Miocene of Hungary and Portugal, resembles that of extant genera as Medorippe ( Müller 1984a: 66, pl. 34; Holthuis & Manning 1990: 7; Karasawa 2000: 811), but may also belong to another dorippid genus. “? Dorippe aff. lanata ” Vega Ferreira, 1965, from the Helvetian of Portugal ( Förster 1979a: 93), could be a synonym of † D. margaretha (see Holthuis & Manning 1990: 7). † Dorippe carpathica Förster, 1979 ( Förster 1979a: 91, fig. 3, pl. 2, fig. 3), known from the Miocene, suggested to be a homoloid ( Müller 1984a: 66, 100), left in Dorippe by Holthuis & Manning (1990: 7), and tentatively assigned to Neodorippe by Müller (1996: 9, as † N.? carpathica ; see Schweitzer et al. 2010, as † N.? carpathica ), has the narrow and simple carapace of Neodorippe , but this assignment remains questionable, the eventuality of an ethusid should be also taken into account ( Müller 2006). † Heikeopsis tuberculata ( Morris & Collins, 1991) ( Morris & Collins 1991: 5, fig. 1a, 1b, as † D. (D.) frascone tuberculata ), from the Miocene of Sarawak, has a well-preserved ventral surface; its thoracic sternum was first figured upside down, thus misinterpreted in the description, but, when correctly oriented (Collins in Collins et al. 2003: pl. 1, fig. 5b), it clearly shows the strongly concave suture 4/5, one of the typical features of the Dorippidae . † Paradorippe sp. , from the Pliocene, shows the same characteristic suture 4/5 ( Karasawa & Nobuhara 2008: fig. 2.12). The carapace of † Nobilum wenchii Hu & Tao, 1996 ( Hu & Tao 1996: pl. 11, fig. 9), from the Miocene, is longer and narrower than in extant N. histrio . † Dorippe ornatissima Müller, 2006 , from the Miocene of Hungary, with a strongly ornamented carapace, is a dorippid but it cannot be assigned to a precise genus ( Müller 2006: 41, pl. 1, fig. 1). † Dorippe judicis Gripp, 1964 , from the Lower Miocene, not listed by Schweitzer et al. (2010), could be an ethusid according to Müller (2006: 42). The identity of the remains of † D. astuta Fabricius, 1798 , recorded by Van Straelen (1938: 91) from the Pliocene of Indonesia is unknown. † Titanodorippe Blow & Manning, 1996 , from the Middle Eocene, is only known by a cheliped propodus. The Eocene † Bartethusa Quayle & Collins, 1981 (type species: † B. hepatica Quayle & Collins, 1981 ( Quayle & Collins 1981: 738, pl. 104, fig. 4), from the Upper Eocene of the Isle of Wight, has a short and wide carapace that explains the assignment of this genus to Dorippidae instead of Ethusidae by De Grave et al. (2009: 31) and Schweitzer et al. (2010: 79). † Eodorippe Glaessner, 1980 , may also be included in Dorippidae (see below). A fragment of carapace from the Miocene is clearly a dorippoid ( Karasawa, Nakagawa & Kaede 2011: 32, fig. 2.7).

† Ethusa berica De Angeli & Beschin, 2008 ( De Angeli & Beschin 2008: 22, fig. 5; De Angeli et al. 2010: 154, fig. 7) (†“ Ethusa n.sp. ” Marangon & De Angeli, 2007: 73, fig. 1B), from the Lower Oligocene of Italy, is a typical ethusid and could be the oldest species known to date for the genus. Müller (1984b: 26) and Vía Boada (1988) reported the probable presence of a fossil subspecies of E. mascarone from the Miocene of Spain. The Japanese † E. chibai Karasawa, 1993 ( Karasawa 1993: 44, pl. 8, fig. 12; 1997: 41, pl. 8, fig. 5), from the early Pliocene, was described as close to E. indica Alcock, 1894 . † Ethusa octospinosa Müller, 2006 , from the Middle Miocene of Hungary, with a small, narrow carapace and characteristic frontal teeth, is clearly an ethusid ( Müller 2006: 42, pl. 1, figs. 1, 2). † Ethusa evae Müller & Collins, 1991 (p. 66, fig. 3h, pl. 4, figs. 1, 2), from the Late Eocene of Hungary, has a typical dorippoid carapace. † Ethusa popognensis De Angeli, Garassino & Pasini, 2009 (De Angeli et al. 2009: 175, fig. 7), of the late Miocene of Italy, is an ethusid with a sculpted carapace. An Ethusa sp. and another dorippoid recorded by Artal & Gilles (2007: 8, figs. 2E, 3C) from the Miocene of France are under study by these authors.

An unnamed mangrove crab from the Cenomanian of Egypt referred to the † Necrocarcinidae ( Schweitzer, Lacovara, Smith, Lamanna, Lyon & Attia 2003: 890, fig. 1) has two very elongated pereopods (probably P2 and P3), a much shorter P1 ending in a stout chela, the P4 and P5 seemingly absent (only a reduced one may be present), but the ventral part is not preserved. This fossil could belong to Dorippoidea , but in any case it does not belong in † Necrocarcinidae , which is a podotreme family ( Guinot, Vega & Van Bakel 2008; Van Bakel, Guinot, Artal, Fraaije & Jagt 2012).

† Sodakus Bishop, 1978 , was included in the Dorippidae by Bishop (1978: 608), De Grave et al. (2009: 31), and Schweitzer et al. (2010: 80). In the type species † S. tatankayotankaensis Bishop, 1978 ( Bishop 1978: 608, figs. 3, 4, pl. 1, figs. 1–6), from the Maastrichtian of South Dakota, the first two abdominal somites are narrow and dorsal, the P4 and P5 coxae are large and dorsally located, the well-preserved thoracic sternum is relatively narrow, with a triangular anterior shield (sternites 1–3) inserted between the mxp3, a long sternite 4, characters that may in part belong to a dorippoid but also to a dakoticancroid. The facies of the carapace and the thoracic sternum of † S. mexicanus Vega, Feldmann & Villalobos-Hiriart, 1995 ( Vega et al. 1995: 245, fig. 4), from the Lower Maastrichtian of Mexico, are clearly not those of a dorippoid. For now † Sodakus is excluded from the Dorippoidea .

A dorippoid status is here recognised for the † Telamonocarcininae Larghi, 2004 , that we elevate to the familial rank, † Telamonocarcinidae new status (see Appendix II), to tentatively receive three genera: (1) † Telamonocarcinus Larghi, 2004 , with two Cenomanian species, † T. gambalatus Larghi, 2004 ( Larghi 2004: figs. 5, 6, 7.2–7.8; see also Garassino et al. 2008: 61) as type species, and † T. binodosus ( Collins, Kanie & Karasawa, 1993) (previously placed in † Eodorippe ); (2) † Eodorippe Glaessner, 1980 , with † E. spedeni Glaessner, 1980 , from the Campanian-Maastrichtian as type species; (3) † Tepexicarcinus Feldmann, Vega, Applegate & Bishop, 1998 , with † T. tlayuaensis Feldmann, Vega, Applegate & Bishop, 1998 , from the Albian of Mexico, as type species. These extinct genera were included in Dorippidae by De Grave et al. (2009: 31) and Schweitzer et al. (2010: 79, 80) without a separate rank.

The developed P2 and P3 of Telamonocarcinidae Larghi, 2004 new status, with long, wide meri and carpi and with long, curved dactyli, are typically those of the living Dorippidae (see figures in Holthuis & Manning 1990). The P4 and P5, when known, are indicated as not subchelate, being similar to those of the extant Ethusidae (see Castro 2005). The male abdomen is composed of six free somites plus telson in † Telamonocarcinus (as in extant Dorippidae ), of five somites plus telson in † Tepexicarcinus (as in extant Ethusidae , with fused somites 3–5); the abdomen is unknown in † Eodorippe . With its narrow and weakly sculpted carapace (see Collins et al. 1993: fig. 2.6), the reconstructed † Telamonocarcinus gambalatus (see Larghi 2004: fig. 5) resembles a dorippid, whereas † T. binodosus resembles an ethusid. † Eodorippe spedeni Glaessner, 1980 ( Glaessner 1980: 183, fig. 13, 13a), with a well preserved, typically dorippid carapace, has a “long and narrow, spatulate, pointing forward rostrum” ( Glaessner 1980: 185) that is not found in extant Dorippoidea . But this so-called rostrum could simply correspond to the endostomial gutter that projects anteriorly in the median frontal incision in extant dorippids; however, this elongated gutter, which is visible dorsally, is never as developed as figured in † E. spedeni ( Glaessner 1980: 183, fig. 13a). The shape of the rostrum was one of the significant characters that allowed Glaessner (1980) to tentatively include † Eodorippe in † Torynommatidae (see also Feldmann & Keyes 1992: 10, 68). Glaessner (1980: 186, 187) referred to the “resemblances between Eodorippe and Palicus Philippi ” and suggested “a derivation of Dorippoidea from Tymoloidea”. This explains why Glaessner (1980: 190) hypothesised “a Late Cretaceous origination of the heterotrematous Dorippidae from Tymoloidea”, i.e., from Cyclodorippiformia, namely from the Podotremata. † Eodorippe is clearly a dorippoid.

The absence of gonopores on the well-preserved and complete P3 coxae of female † Telamonocarcinus indicates a eubrachyuran condition, as confirmed by the features of the thoracic sternum, which are similar to those of extant dorippoids and clearly different from Cyclodorippoidea ( Larghi 2004: 535) . It is thus likely that † Telamonocarcinidae , which possesses a “mixture” of both Dorippidae and Ethusidae and is probably not monophyletic, contains the most ancient known dorippoids, with characters considered to be plesiomorphic such as the prominent rostrum. The hypothesis that † Telamonocarcininae “were perhaps the ancestors of modern dorippids” ( Larghi 2004: 535) finds support in the reduced and probably dorsal P4 and P5 ( Larghi 2004: 536). A P5 as the only reduced pereopod could allow the interpretation of a retroplumoid.

† Tepexicarcinus tlayuaensis Feldmann, Vega, Applegate & Bishop, 1998 , from the Albian of Mexico, first suspected to be a homolid (Feldmann et al. 1998: 86, fig. 7), has reduced and subdorsal P4 and P5, hence its assignment to † Telamonocarcininae ( Larghi 2004: 534, 536; Vega et al. 2006: 28, fig. 4, pl. 2, figs. 2–12) and to Dorippidae ( De Grave et al. 2009: 31; Schweitzer et al. 2010: 80). The faint figures of Vega et al. (2006) do not suggest a dorippid but remind of an ethusid. The status of these genera and species will remain doubtful until the nature of their thoracic sterna and sutures is known (see Appendix II).

There is still much to be done to understand the true organisation of early Cretaceous crabs, which may be podotremes (e.g., such as some torynommatids) or to either eubrachyurans belonging to poorly known extinct families or basal heterotremes with living representatives such as dorippoids, retroplumoids, or palicoids.

The podotreme status of † Orithopsidae ( Gymnopleura , †Palaeocorystoidea), a family often considered close to Dorippoidea , was recognised by Guinot, Vega & Van Bakel (2008: 711) and Van Bakel, Guinot, Artal, Fraaije & Jagt (2012: 63). † Orithopsis Carter, 1872 , the onomatophore (type genus) of † Cherpiocarcinus Marangon & De Angeli, 1997 , and † Silvacarcinus Collins & Smith, 1993 ), which were once referred to either † Necrocarcinidae or to their own family († Orithopsidae ) within Dorippoidea ( Förster 1968: 177; Fraaije 2002: 913; Schweitzer, Feldmann, Fam, Hessin, Hetrick, Nyborg & Ross 2003: 32; De Grave et al. 2009: 31; Schweitzer et al. 2010: 81), are podotremes. The association of † Necrocarcinidae (podotreme) with the Calappoidea De Haan, 1833 (see Schweitzer & Feldmann 2000a: 241; Collins 2003: 85) or with the Dorippoidea (both heterotreme eubrachyurans) is not supported by evidence.

Superfamily Hymenosomatoidea . A separate superfamily Hymenosomatoidea is supported by the combination of morphological, reproductive, spermatozoal, larval, ecological as well as molecular features that all are unique within Brachyura . The relationships with Majoidea , at least with some families, are however undeniable (see below; Position of the Hymenosomatoidea within the Brachyura ; Affinities between Hymenosomatoidea and Inachoididae ).

The poorly calcified, thin carapace forms a sort of lid and generally bears a lateral furrow (“hymenosomian groove” or “hymenosomian rim”; Figs. 43A , 47D, E ) that encircles the dorsal surface, often divided into plates by deep grooves, internally corresponding to thickenings that reinforce the surface, but sometimes hardly grooved ( Yang & Sun 1998). Such plates are unique to Brachyura , and are even present in the zoeae of Elamena cimex Kemp, 1915 , the carapace of which is formed of three triangular plates ( Krishnan & Kannupandi 1988, as Trigonoplax cimex Kemp, 1915 ; see Kemp 1915). The nature of the carapace and of the lateroventral part of the body of Hymenosomatidae remains a mystery that could only be resolved by the location of the ecdysis line and by an ontogenetic study.

Several characters are unusual: (1) vulvae variously displaced anteriorly ( Fig. 48C ); (2) thoracic sternum with sternites 1–3 individualised and forming a narrow shield prolonging between the mxp3; sternites 4–8 considerably enlarged; (3) sutures 4/5–7/8 equidistant, short, and laterally restricted; (4) sterno-abdominal cavity of males generally reduced in length; (5) male and female abdomens never having more than five somites, always with the formation of a pleotelson (somite 6 fused to telson) and, often, fusion of additional somites to the pleotelson [e.g., in male and female Crustaenia palawanensis ( Serène, 1971) , with somites 3–5 fused to pleotelson, thus abdomen consisting of only three elements]; (6) press-buttons for abdominal locking situated on the undivided part of thoracic sternum; (7) intercalary platelets, either completely defined, articulated, and moveable, thus similar to dromiid uropods in male Amarinus and Odiomaris ( Fig. 29D, E ), or more often fused, with variably discernible delimiting sutures, which may be not visible at all in derived taxa ( Lucas 1980: fig. 7; Guinot & Richer de Forges 1997: figs. 4A–E, 6B–E); (8) abdominal sockets located at the base of the pleotelson (somite 6 fused telson) thus belonging to the last abdominal somite (somite 6), as usual; (9) reduction in the number of females pleopods, and even complete loss ( Lucas 1980); (10) abdominosternal configuration in mature females with brood cavity and pseudovulvae ( Ng & Chuang 1996: 60, fig. 26D); (11) thoracic sternum/pterygostome junction (variously complete, rarely absent), as a result of the long extension of sternite 4 ( Figs. 42B , 43C ); (12) distinctive Milne Edwards openings, tightly sealed by the flabelliform coxa ( Guinot 1979a: fig. 30A; Guinot & Richer de Forges 1997: figs. 1D, 3); (13) axial skeletal system with dorsoventral bipartition into two equal parts above and below junction plate, and parallel arrangement of phragmae in anteroposterior plane, ending in regular, symmetrical partition, the sella turcica being the only transversal binding ( Fig. 47D–F ; Secretan 1998: 1763, figs. 19, 20); (14) G2 with wide proximal part ( Melrose 1975: fig. 2E).

The clearly demarcated and variously mobile platelets, which bring to mind dromiid uropods, are assumed to be “vestigial” dorsal uropods, each bearing ventrally a socket for abdominal locking. The persistence of uropods in male Amarinus ( Fig. 29D, E ; Holthuis 1968: 115; Lucas 1980: fig. 7A–D; Ng & Chuang 1996: figs. 2D, 3E; Guinot & Richer de Forges 1997: figs. 4B–D, 6B–E; Guinot & Bouchard 1998: fig. 27; Rahayu & Ng 2004: fig. 2E; Naruse, Mendoza & Ng 2008: fig. 1e; Guinot 2011a: fig. 3A–C), Odiomaris ( Ng & Richer de Forges 1996: fig. 7B; Guinot 2011a: figs. 1A, 2B, D), as well as in Elamenopsis guinotae Poore, 2010 ( Poore 2010: fig. 2, table 1) and Hymenosoma hodgkini Lucas, 1980 ( Lucas 1980: 170, fig. 7I), is clearly plesiomorphic, concurrently with a number of other plesiomorphic characters, such as the disposition of the eyes (absence of orbits), of the cephalic appendages and of the proepistome (either absent or varying from reduced to large), the disposition of the five free abdominal somites (plus pleotelson) in both sexes, and the regularly compartmented axial skeleton ( Fig. 47D–F ).

In establishing the subfamily Odiomarinae for Odiomaris and Amarinus, Guinot (2011a) clearly stated that the appendages of the sixth abdominal somite are practically present in all Decapoda , but with different patterns: as a biramous appendage, foliaceous ramus, rasp, ventral lobe, dorsal plate, intercalated platelet, or socket. The recognition of Brachyura by the absence of uropods is actually an oversimplification. The eubrachyuran socket corresponds to a “vestigial uropod”, as well as the dromiid uropod ( Guinot & Bouchard 1998). The brachyuran uropod actually presents a wealth of character states: the dromioid dorsal uropod (without a socket) and the odiomarine uropod with a functional socket, as well as the eubrachyuran socket that is the complementary part of the typical press-button system. The transformation of the uropod-socket, which seemingly has required a substantial modification, is exemplified by the configuration of Odiomarinae , where the uropod varies from completely articulated to variously incorporated to the sixth somite. The transformation series of the moveable intercalated platelets into fused plates, followed by their complete integration to the abdomen in a pleotelson, illustrates the stages towards the apomorphic condition found in male Hymenosomatinae McLeay, 1838 (as defined by Guinot 2011). When the delimited platelets have disappeared, the sockets remain excavated on the ventral surface of the sixth somite, as a variously deep depression, delimited by a calcified margin. This conforms to the condition found in all Eubrachyura. In the odiomarine post-pubertal females, as in most eubrachyuran mature females, the widened abdomen does not bear sockets any longer and it is not held against the thorax. The diversity of Hymenosomatoidea is translated by several characteristic patterns: moveable platelets, which is the plesiomorphic condition ( Odiomarinae pro parte); simple and no longer articulated platelets, with variously distinct sutures ( Odiomarinae pro parte); structures dorsally recognisable at the pleotelson’s base as in Hymenicoides (see Kemp 1917: fig. 21; Guinot & Richer de Forges 1997: fig. 4E; Naruse & Ng 2007a: 18) and H. robertsi (see Naruse & Ng 2007a: figs. 1b, 5a) and Limnopilos Chuang & Ng, 1991 ( Naruse & Ng 2007a: figs. 1d, f, 8a); and variously marked prominences, the external indications being lost (other hymenosomatoids). All of these structures play the same role in covering the acute buttons of the press-button system and firmly locking the abdomen. The uropod, however, can be no longer articulated, with variously marked remnants of the sutures, and finally becomes entirely incorporated in the sixth somite at the pleotelson’s base, remaining still recognisable by an inflated area or being without a visible external indication. The odiomarine uropod represents the only case of a dorsal plate among the Eubrachyura (for the case of Capartiella , see Affinities between Inachoididae and Inachidae ).

The diversity of Hymenosomatoidea is seen in the extreme variability of the G1, from simple, thin, curved, with a narrow apex to thick, twisted, and ending in a complex, inflated, lobed apex ( Melrose 1975; Lucas 1980; Ng & Chuang 1996). The G2, although not often described or figured, seems more constant, always short, with a wide base and a tube-like distal portion ( Melrose 1975: 16, fig. 2E; Naruse, Ng & Guinot 2008: fig. 8g; Naruse & Komai 2009: 178, fig. 3g).

The thoracic sternum/pterygostome junction varies in Hymenosomatoidea , and the Milne Edwards openings are large ( Fig. 42B ; Guinot & Richer de Forges 1997: 462, figs. 1D, 2A; Ng et al. 1999: figs. 1C, 2E; Guinot 2011a: fig. 1B, C; 2011b: fig. 1B; Ng et al. 2011: fig. 2B). A junction may be absent, e.g., in Crustaenia palawanensis and Neorhynchoplax mangalis (Ng, 1988) , incomplete with only a weak latero-anterior extension of sternite 4, as in N. dentata Ng, 1995 , N. prima , and Elamena truncata (with a slight but acute latero-anterior extension of sternite 4), or complete as in Trigonoplax unguiformis ( Lucas 1980: 154, 186, fig. 5F; Ng et al. 1999: 90; Guinot 2011b: fig. 1B). The Milne Edwards openings, entirely filled by the flabelliform coxa, are thus close to the cheliped coxae in the case of a narrow junction or distinctly separated as in Odiomaris pilosus and Amarinus lacustris ( Figs. 42B , 43C ). The hymenosomatoid disposition may be considered an intermediate stage towards the dorippid condition, with pterygostomial slits ( Fig. 42C ).

The spermatozoal ultrastructure of Hymenosomatoidea is informative in many respects. The spermatozoa of two species of Odiomaris Ng & Richer de Forges, 1996 , O. pilosus and O. estuarius , and of Elamena vesca Ng & Richer de Forges, 1996 , have the components typical of eubrachyuran (heterotreme and thoracotreme) spermatozoa but significantly differ in at least nine major characteristics from those of all other brachyuran taxa that were investigated ( Richer de Forges et al. 1997; Jamieson & Tudge 2000). The more noteworthy are: the presence of an epiopercular dome; an acrosome smaller in volume than the nucleus, longer than wide, and strongly emergent, being surrounded only basally by nuclear material; the thin, putative inner acrosome zone that is anteriorly almost septate owing to several longitudinal corrugations; and the unique helical and posterolateral disposition of the nuclear arms. The combination of spermatozoal characters, collectively and often individually, is so markedly distinctive from that of the families with which Hymenosomatidae has traditionally been associated, the heterotreme Majoidea and thoracotremes such as Varunidae , Ocypodidae , and Gecarcinidae ( Jamieson et al. 1995) , that Richer de Forges et al. (1997: 238, 239) recognised a “hymenosomatid-type of spermatozoon”. A “majid-hymenosomatid” relationship was not supported by spermatozoal ultrastructure, because the two families differed in the nine distinctive characters of Hymenosomatidae ( Jamieson & Tudge 2000; Tudge et al. 2013). One of these hymenosomatid characters, the almost septate condition of the inner acrosome zone is, exceptionally, approached in the majoid Cyrtomaia furici Guinot & Richer de Forges, 1988 (Inachidae) but could not be considered a convincing synapomorphy between the two families ( Jamieson et al. 1998: 199, 204, 205; see also Hinsch 1973). Majoids formed the sister-group to all other Eubrachyura in an analysis of only spermatozoal characters ( Jamieson et al. 1995: fig. 1A), whereas the dorippid Neodorippe was the plesiomorphic sister-group to all other investigated eubrachyurans in a cladistic analyses of combined morphological and spermatozoal characters ( Jamieson et al. 1995: fig. 1B; see also Jamieson 1994: 390, 391; Jamieson & Tudge 1990, 2000). The highly developed projection of the acrosome from the nucleus in the hymenosomatid spermatozoon recalls the totally emergent acrosome of podotremes. The projection (emergence) of the acrosome from the nucleus is complete in podotremes, strong in hymenosomatid, and partial in dorippids and majoids. A less pronounced emergence is seen in Anomura ( Tudge 1995; Jamieson & Tudge 2000), dorippids ( Jamieson & Tudge 1990, 2000) and majoids ( Jamieson et al. 1998; Jamieson & Tudge 2000; Tudge et al. 2013). Emergence of the acrosome may represent the plesiomorphic condition in Brachyura whereas the complete (podotremes) or strong (hymenosomatoids) emergence may represent an independently apomorphic development from this state, whereas most heterotremes and thoracotremes have completely withdrawn the acrosome. From a purely spermatological perspective, Odiomaris , Elamena , and, provisionally, all hymenosomatoids must be excluded from Thoracotremata (see Position of the Hymenosomatoidea within the Brachyura ). Both Hymenosomatoidea and Cryptochiroidea , which are similar in having male sternal gonopores (like in thoracotremes), have sperm features that are equivocal with regard to their placement within Brachyura ( Tudge et al. 2013) .

Larval data, in particular the absence of a true megalopal stage, a feature unique among the Brachyura , also indicate the exceptional nature of Hymenosomatoidea ( Williamson 1982; Williamson & Rice 1996). This exception to the general developmental pattern of Brachyura is verified in all hymenosomatids, e.g., Neorhynchoplax kempi (Chopra & Das 1930) ( Salman & Ali 1996, as Elamenopsis kempi ), Halicarcinus varius (Dana, 1851) ( Horn & Harms 1988) , Elamena sindensis Alcock, 1900 ( Tirmizi & Kazmi 1987), E. matheoi ( Desmarest, 1823) ( Al-Kholy 1959) , Hymenosoma orbiculare (see Dornelas et al. 2003). The number of freeliving zoeal stages is reduced to no more than three stages in estuarine and marine hymenosomatoids. The freshwater Amarinus lacustris has a direct development, hatching occurring before the last crab instar is reached, and the larvae remaining among the maternal pleopods until they moult to juvenile crabs ( Wear & Fielder 1985: 38), whereas the estuarine A. paralacustris has three zoeal stages and no megalopa ( Lucas 1971). A reduced number of large eggs may be enclosed in a large brood cavity and even within the cephalothorax in some marine hymenosomatoids. This phenomenon is made possible by the empty internal space devoid of skeletal phragmae, their absence in the median part of the thoracic sternum allowing the sternal surface to be strongly excavated in the female and even to reach the level of the carapace wall ( Melrose 1975; Lucas 1980; Ng & Chuang 1996; Guinot & Richer de Forges 1997; Naruse et al. 2005). The larval development may be abbreviated, and even direct as in Neorhynchoplax mangalis (see Lucas 1980; Ng 1988a; Ng & Chuang 1996). The cave-dwelling Sulaplax ensifer possesses the largest eggs (mean 1.19 mm, n = 10) and the smallest clutch size (17 eggs) known for any hymenosomatoid; a female Neorhynchoplax bovis (Barnard, 1946) contained 13 juveniles under the abdomen ( Barnard 1950: 72; Naruse, Ng & Guinot 2008: 31). The zoeae present particular, even unique characters, such as reduced antennules and antennae, antenna with a single lateral seta assumed to be a vestigial exopodite, only a vestigial coxal endite on the second maxilla armed with a single seta, loss of the outer three processes on the telson, and loss of the pleopods ( Al-Kholy 1959; Rice 1980; Muraoka 1977; Terada 1977; Fukuda 1981; Tirmizi & Kazmi 1987; Dornelas et al. 2003: 2593, table 1). The carapaces of the zoeae of Elamena cimex Kemp, 1915 , are formed of three separate plates ( Krishnan & Kannupandi 1988).

The abdomen of the hymenosomatoid zoeae, which consists of five somites (see Williamson 1982: 64) and is devoid of pleopods in all stages, belongs to two basic types ( Wear & Fielder 1985: 38). Type 1, a small, slender abdomen, with a telson much longer than broad, ending in a short, narrow fork, characterises Amarinus paralacustris (see Lucas 1971: fig. 5E, F, as Halicarcinus paralacustris ), Halicarcinus australis (see Lucas 1971: fig. 1; Rice 1980: fig. 19a, b), H. cookii (see Wear & Fielder 1985: figs. 97, 98), H. coralicola (Rathbun, 1909) (Terada 1977: fig. 2H, as Rhyncoplax coralicola ), H. messor ( Stimpson, 1858) (Fukuda 1977: fig. 2A1, A2, I1 I2, 3A, I, as Rhynchoplax messor ), H. orientalis Sakai, 1932 (Fukuda 1977: fig. 1A1, I1; Terada 1977: fig. 3H), H. ovatus Stimpson, 1858 ( Lucas 1971: fig. 4A, E, I), H. planatus (see Richer de Forges 1977: figs. 18A, B, H, 19A, B, 20E, F; Wear & Fielder 1985: figs. 103, 104), H. rostratus (Haswell, 1881; see Low 2012a: table 1 for the date) ( Lucas 1971: fig. 5A, E), H. varius (Dana, 1851) ( Wear & Fielder 1985: figs. 99, 100; Horns & Harms 1988: figs. 1A–C, 3G–I), H. whitei (Miers, 1876) ( Wear & Fielder 1985: figs. 101, 102), Hymenosoma depressum Hombron & Jacquinot, 1846 ( Wear & Fielder 1985: figs. 110, 111), and H. orbiculare (see Dornelas et al. 2003: figs. 8, 9). Type 2, with abdominal somites 4 and 5 laterally extended, somite 5 being expanded into a broad plate overlapping the telson, characterises Elamena cimex (see Krishnan & Kannupandi 1988: figs. 1A, B, 2 l, m, 3 l, m, 4n, o), E. mathaei ( Desmarest, 1823) ( Gurney 1942: fig. 119A, D; Al-Kohli 1959: figs. 29, 37), E. momona Melrose, 1975 ( Wear & Fielder 1985: figs. 118, 119), E. producta Kirk, 1879 ( Wear & Fielder 1985: figs. 120, 121), E. truncata ( Stimpson, 1858) (Terada 1977: fig. 4H), Neohymenicus pubescens (Dana, 1851) ( Wear & Fielder 1985: figs. 106, 107), Trigonoplax longirostris McCulloch, 1908 ( Wear & Fielder 1985: figs. 114, 115). These abdominal patterns in the zoeae probably correspond to the distinct higher-ranked lineages that may be recognised in the family (Naruse, Guinot & Ng, manuscript).

The Hymenosomatoidea is unique in Brachyura by the absence of a true megalopa in marine and brackish species as well as in the freshwater representatives ( Broekhuysen 1955; Richer de Forges 1977; Melrose 1975; Muraoka 1977; Lucas 1971, 1972, 1975, 1980; Rice 1980, 1981a; Wear & Fielder 1985; Gore 1985; Rabalais & Gore 1985; Rabalais 1991; Horn & Harms 1988; Kai & Henmi 2008). The stage described by Boschi et al. (1969) as a megalopa of Halicarcinus planatus and the atypical “megalopa” of Trigonoplax unguiformis described by Fukuda (1981) are actually the first juvenile stage. The omission of a megalopa (similarly reported in some Pinnotheridae ) seems not to be linked to the suppression of the free larval stages that occurs in the freshwater species since it has been reported in all the studied hymenosomatoids, including those with free zoeae. The failure to develop pleopods during the complete zoeal phase is probably associated to this absence of a megalopal stage ( Rice 1980: 314). The juveniles, which lack natatory pleopods and uropods, are, however, able to swim using their walking legs ( Wear & Fielder 1985: 37), the arrangement of their antennules and antennae and their bifurcated (zoea-like) telson suggest, nevertheless, a megalopa ( Lucas 1971: 485). According to Lucas (1971: 484), the larvae of Halicarcinus ovatus Stimpson, 1858 , and H. rostratus (Haswell, 1881) tend to remain in the vicinity of the habitat of the adults, thus “there is little need for a transitional megalopa stage to locate the adult habitat”, giving the advantage of a reduced pelagic life. Lucas (1975: 99) hypothesised that “the first juvenile instar may be the result of considerable modification of a megalopa to a crab-like condition, rather than omission of a developmental stage”. Rabalais & Gore (1985: 76, 77, 86; see also Rabalais 1991: 218) regarded the first juvenile hymenosomatoid crab as an “advanced zoea” despite the presence of rudimentary setae without a natatorial function and considered a “direct” developmental pattern. According to Felder et al. (1985: 183), the first juvenile crab is equivalent to a “postlarval” stage (or decapodid; see Anger 2001) in the form of a benthic megalopa. The suppression of the megalopa could be not an adaptation to the environment but considered instead an ontogenetic character inherited from a common ancestor, as in the case of the Jamaican sesarmid Metopaulias depressus (see González-Gordillo et al. 2010) (See Carcinisation and its outcomes). Thus, in the case of Hymenosomatoidea this characteristic permits assuming that the lineage has evolved from a single, deeply rooted marine ancestor, probably with an abbreviated development, and now inhabiting the sea as well as numerous types of non-marine habitats.

According to Rice (1980: 314, 315), the combination of larval characters clearly distinguishes Hymenosomatoidea not only from Majoidea , but also from all other brachyuran zoeae (with the exception of dorippoids and pinnotheriods), the hymenosomatoid developmental sequence seemingly resulting from the retention of the ancestral early larval stages and the loss of the later ones. Rice’s first conclusion that their advanced zoeal features (tendency to reduce carapace and abdominal armature, simplification in the cephalothoracic appendages, reduction in the length of the telson fork) ally them with the most derived thoracotreme families rather than with the more derived majoid crabs is an indication of the perplexing traits exhibited by the group (for accuracy of brachyuran larval description see Clark et al. 1998). The interpretation of specialisations interpreted as “advanced features” in larvae ( Rice 1980: 315; Wear & Fielder 1985: 38) must be linked to the extreme carcinisation of an old group, having undergone a long evolutionary process. The adult characteristics, which were erroneously considered “recently acquired”, hence the assignation to the thoracotremes by Guinot (1978a; 1979a), but subsequently rectified by Guinot & Richer de Forges (1997), could be interpreted in the same manner. Contradictory traits (early and “advanced” features) likewise occur in the case of the larvae of Dorippidae . Thus, Rice (1980: 314, 319) briefly suggested a similarity between dorippids and hymenosomatids. The conclusions of Rice (1983: 326) that the larval morphology of hymenosomatids shows that they “could not have evolved from any of the extant thoracotrematous groups” are consistent with the placement of Hymenosomatoidea outside Thoracotremata and basal in Eubrachyura.

The biology of reproduction of these small-sized crabs, with several peculiar reproductive strategies, is noteworthy. All hymenosomatids have an extended reproductive season with multiple spawning, breeding occurring throughout the year ( Hill & Forbes 1979; Johnston & Robson 2005; Van den Brink & McLay 2009, 2010; McLay & Van den Brink 2009; Van den Brink et al. 2012; see also Hosie 2004). Copulation may occur before the puberty moult of the females and also include pre-pubertal males, exceptional traits in Brachyura . Observation of mating between a male and an ovigerous female of the tropical Limnopilos naiyanetri in an aquarium ( Fig. 43B ) illustrates this unique strategy. The synchronicity of ovarian maturity and embryonic development in Halicarcinus planatus shows that females are able to re-mature their ovaries while they are ovigerous, a rather unique feature, whereas a high number of successive spawns is an adaptive advantage to the extreme sub-Antarctic environment ( Diez & Lovrich 2010; see also Gorny 1999; Vinuesa et al. 2005; Boschi & Gavio 2005; Vinuesa & Ferrari 2010); Ferrari et al. 2011. Females of Halicarcinus cookii show a precocious mating in the penultimate instar and are able to lay fertilised eggs after their pubertal moult ( Van den Brink & McLay 2010). The duration of the reproductive period of H. planatus decreases towards the southern range limits, the populations of the Beagle Channel (Tierra del Fuego) only having two main periods of reproduction, which correspond with two larval cohorts in the zooplankton ( Diez et al. 2012). Moreover, the postlarval development of H. planatus females suggests an anomalous growth process ( Vinuesa & Ferrari 2008).

The occasional carrying behaviour observed in Hymenosomatoidea can be regarded as a “primitive” behaviour. The arthrodial cavities of the P4 and P5 of Hymenicoides are subdorsal and dorsal respectively, and the P5, which is shorter, may be placed above the carapace. Hymenicoides carteri lives among sponges and nibbles insect larvae and polychaete worms in the Ganges ( Ramakrishna 1977). Hymenosomatoids living among algae may be found carrying tufts of the red alga Corallina held by the P5 “which bend upwards and over just like those of the sponge crabs” ( Melrose 1975: 7). Halicarcinus innominatus Richardson, 1949 , was observed in aquaria hiding under a shell with its P5 gripping the valve ( Melrose 1975: 33, 34). A burying behaviour in sand or mud by means of efficient fringed pereopods is also known ( Barnard 1950: 71; Melrose 1975: 7; Lucas 1980: 222) (see Concealment behaviour: Carrying behaviour; Burying and burrowing in the Eubrachyura).

The Hymenosomatoidea , which probably contains the smallest known brachyurans, less than 2 mm of carapace width, consists of only one family, including 20 genera and 120 species, but species diversity is higher since, for instance, eight new species have been described in 2008 and 2009. The genetic and morphological analyses by Edkins et al. (2007) and Teske et al. (2009) furthermore suggested that the southern African “crown crab”, Hymenosoma orbiculare , represented five distinct species. According to G. Paulay (in Castro 2011: 111) Hapalocarcinus marsupialis actually consists of a complex of several species.

The wide morphological variations of the hymenosomatid vulvae and gonopods as well as the heterogeneous organisation of the carapace, rostrum, proepistome, epistome, mouthparts, and the male and female abdomens ( Melrose 1975; Lucas 1980; Ng 1991; Ng & Chuang 1996; Naruse et al. 2005; Naruse & Ng 2007a; Naruse, Ng & Guinot 2008; Naruse, Mendoza & Ng 2008) provide evidence for the presence of several distinct lineages in Hymenosomatidae (T. Naruse, D. Guinot & P.K.L. Ng, manuscript).

The group is distributed worldwide but it is particularly widespread throughout the southern hemisphere, also circumpolar in the subantarctic region ( Richer de Forges 1976, 1977; Gorny 1999; Yeo et al. 2008; Boschi & Gavio 2005; Diez et al. 2012). Hymenosomatoids are particularly abundant in Australia ( Walker 1969; Chapman & Lewis 1976; Lucas 1980; Davie 2002; Poore 2004) and New Zealand where they may be confined to lakes ( Chilton 1882; Melrose 1975; McLay 1988; Towers & McLay 1995), and are common in fresh waters of India, from rivers close to sea and under tidal influence or far inland ( Kemp 1917; Ramakrishna 1977; Naruse & Ng 2007a; Ng et al. 2011) and southern and eastern Africa ( Teske et al. 2009). Hymenosomatoids, generally with cryptic habits, are found in a wide variety of environments: the open ocean to a depth of 500 m ( Melrose 1975; Richer de Forges 1993), in muddy sediments, estuarine and brackish waters, permanent inland fresh waters ( Kemp 1917; Takeda & Miyake 1971; Hill & Forbes 1979; McLay 1988; Towers & McLay 1995), from frehwater caves in Indonesia (Ng 1991; Naruse, Ng & Guinot 2008), from an anchialine cave in Philippines ( Husana et al. 2011), at an altitude of 1600 m in New Guinea ( Holthuis 1968: 112), and even in arid zone pools ( Kemp 1917; Ali et al. 1995, 2000), from mud-covered littoral rocks, coral reefs ( Melrose 1975; Ng & Jeng 1999a); may be also symbiotic with echinoderms, including sea cucumbers and brittle stars ( Takeda et al. 1976: 33; Poore 2004: 390; Kai & Henmi 2008: 342). Some species are the most abundant soft-bottom brachyurans where they occur, as in the case of Neorhynchoplax kempi (Chopra & Das, 1930) , which is one of the most important components of the invertebrate fauna in southern Iraq ( Ali et al. 1995, 2000, as Elamenopsis kempi ). Odiomarinae are restricted to fresh or low-salinity waters ( Marquet et al. 2003; Juncker & Poupin 2009). Some species such as Cancrocaeca xenomorpha and Samarplax principe Husana, Tan & Kase, 2011 , exhibit true troglomorphic adaptations (Ng 1991; Deharveng et al. 2002; Husana et a l. 2011; see also Naruse, Ng & Guinot 2008). Based on their wide geographical and vertical distributions (from a 500 m depth to as high as 2000 m altitude) and wide variety of habitats, hymenosomatoids are probably the most ecologically diverse group of crabs. According to Teske et al. 2009, the evolutionary history of the southern African Hymenosoma could be explained by dispersal from tropical towards temperate regions, with range expansions into formerly inhospitable habitats during warm climatic phases, followed by adaptations and speciation during subsequent cooler phases (see also Forbes & Hill 1969; Hill & Forbes 1979). Another explanation is a Gondwanan origin of Hymenosomatoidea , as assumed here ( Guinot 2011b; see Position of the Hymenosomatoidea within the Brachyura ).

Hymenosomatoids lack a fossil record (Lucas 1970: 275; 1980: 225; Feldmann & McLay 1993: 447, table 1; Teske et al. 2009: 24; Schweitzer et al. 2010: 93), but they are all so small and soft-shelled that the absence of fossils cannot be evidence of a relatively recent origin. Contrary to Walker (1969) and Lucas (1970, 1980), who suggested a recent origin, Chilton (1915) hypothesised that Amarinus lacustris was an ancient species that arose prior to the break-up of Gondwana, thus possibly Cretaceous. This hypothesis conforms to all data listed above. Despite the recent genetic data by Teske et al. (2009) supporting a post-Gondwanan origin, an early origin of the family, in combination with high specialisations, is assumed here. If this hypothesis is confirmed, the marine heterotreme brachyurans predate the fragmentation of Gondwana.

Molecular and spermatozoal data are consistent with the interpretation of Hymenosomatoidea as the more basal heterotreme clade that is proposed below (see Position of the Hymenosomatoidea within the Brachyura ; Affinities between Dorippoidea and Hymenosomatoidea ; Affinities between Hymenosomatoidea and Inachoididae ; Affinities between Inachoididae and Inachidae ).

Superfamily Leucosioidea . The transformation series of the penis from coxal to coxo-sternal ( Fig. 30 ; see Modalities of penis protection: Coxo-sternal penial tube) may be compared to the ability of the thoracic sternum to expand laterally. In Leucosiidae the anterior region of the sternite 4 shows a transformation series of the thoracic sternum/pterygostome junction ( Fig. 42D ), clearly linked to the respiratory system.

One of the strongest synapomorphies of the Leucosioidea is the morphology of the respiratory system, the inhalant and efferent openings being rather close to one another below the fronto-orbital region. The afferent branchial channel is deeply excavated along the sides of the endostome and covered by the mxp3 exopodite. The external border of this afferent channel (endostegal channel) is represented by the pterygostomial crest, which is not the true pterygostomial margin but the external border of the oral-pterygostomial margin ( Garstang 1897b: 215; Ihle 1918: 186–197; Monod 1956: 83; Guinot 1978a: 282; 1978b: 12). The efferent channel occupies the median endostomial groove. The plesiomorphic condition is found in the Iphiculidae , which is typically leucosioid by having the afferent channel alongside the endostome and lacking the Milne Edwards openings, without, however, a connection between the sternum and pterygostome, the immobile mxp3 coxa closing any possible aperture. In other leucosioids there is a junction of the thoracic sternum (sternite 4) with the pterygostome, which separates the bases of the chelipeds from the mxp3, and the inhalant water opening is situated in front of the buccal cavern ( Fig. 42D; H . Milne Edwards 1836–1844, Atlas , pl. 24, figs. 1a, 2a, 3a, 3c, 4a, pl. 25, figs. 1a, 1e, 2f, 3a; Ihle 1918: 192, fig. 107; Bouvier 1940: 143). Different character states of the thoracic sternum/pterygostome junction are found in Leucosiidae , the outgrowth of sternite 4 in front of the chelipeds varying from narrow (plesiomorphic condition: Myra , Pseudomyra , Ilia ) to wide ( Philyra ; see Ihle 1918: fig. 115) to much extended (derived condition: Cryptocnemus , Euclosia , Leucosia ) ( Fig. 42D ; Ihle 1918; fig. 107; Guinot 1966: 760, 761, figs. 23, 24; 1967: 836; 1978b: 9, 11). The leucosioid respiratory system is clearly linked to a burying habit. Absent in the Iphiculidae is the large egg brood chamber present in many Leucosiidae . The leucosiid brood chamber is like a box with the margins tightly joined to the edges of the sterno-abdominal cavity, so firmly joined that the abdomen is broken when lifted; a peculiar canal, the branchio-sternal canal, allows the oxygenation of the eggs ( Drach 1955). This reflects an adaptation to protect the eggs during burying. Another family of burying crabs, the Corystidae , which lacks a marked abdominal sexual dimorphism with an unspecialised female abdomen ( Fig. 3B ), has not evolved such a protection. The respiratory currents are also fundamentally dissimilar in the Corystidae and the Leucosioidea . The Matutidae and Calappidae , which also bury, lack such a brood cavity.

The large sternum/pterygostome junction, which prevents any afferent opening in front of the chelipeds, was, among others, a diagnostic feature used by H. Milne Edwards (1837a: 99, 118) to establish its “ Leucosiens ”. A similar junction, responsible for the absence of pre-chelipedal inhalant openings, was considered characteristic of the “ Raniniens ” ( Fig. 42E; H . Milne Edwards 1837a: 191; Atlas , pl. 21, fig. 2). Despite a similar junction, the leucosiid respiratory system (anterior inhalant opening) was nevertheless clearly distinguished by H. Milne Edwards (1839: 134–136, figs. 2–4) from the raninoid ones, in which posterior inhalant openings located between the tergite of first abdominal somite and the P5 coxa are present in some families. According to Ortmann (1892: 556, 557, pl. 26, fig. 11b), the thoracic sternum/pterygostome junction is a character shared by Raninidae and Leucosiidae (grouped in the same division, Leucosiinea). The junction line between the thoracic sternum and pterygostome, which is present in all the Raninoidea ( Fig. 42E ) and most Leucosioidea ( Fig. 42D ), shows a variable extent in genera of both groups. This line was called sutura carapaco-sternalis by Ihle (1915: 66; 1918: 192, fig. 107). A sternum/pterygostome junction is the rule in the Raninoidea (excepted for † Marylyreidinae , basal) but it varies from weak to large; in the Leucosioidea such a junction is not found in Iphiculidae , which shows the plesiomorphic condition, and its development also varies. The respiratory system of the Raninoidea (Podotremata) , with the absence of Milne Edwards openings, is linked to their burying behaviour. The Leucosioidea (Eubrachyura) are also burying crabs (see Concealment behaviour: Burying in the Raninoidea ; Burying and burrowing in the Eubrachyura), but the respiratory systems of the Raninoidea and Leucosioidea are clearly different in terms of functional morphology. The Dorippidae , which also bury, shows a sternum/pterygostome junction with, however, a different pattern and it is obvious that the highly specialised dorippid inhalant opening ( Fig. 42C ; Ihle 1916: 103, fig. 45; Bouvier 1940: 199, fig. 140) evolved from typical brachyuran Milne Edwards openings, as those of Ethusidae ( Fig. 42A ).

A second synapomorphy of Leucosioidea is the strict condylar location of the penis, which emerges from the extremity of the P5 coxo-sternal condyle ( Fig. 30 ; see Modalities of penis protection; Condylar protection; Table 4). The Leucosioidea shows a particular transformation series from coxal to coxo-sternal penial condition ( Fig. 30 ). In Leucosia , for example, sternites 7 and 8 are substantially expanded (paralleling the sternal expansion occurring at the sternite 4 level in most Leucosioidea , see above) and join over the P5 coxa, and thus provide an additional protection to the penis, which is enclosed within the coxo-sternal condyle ( Fig. 30C, F ).

Superfamily Majoidea . A male gonopore with a posteriormost location in relation to sternite 8 characterises all the Majoidea , e.g., Oregoniidae ( Fig. 1A ), Inachidae ( Figs. 49B, C , 50A ), Inachoididae ( Fig. 50C, E ), Epialtidae , Pisidae and Majidae ( Fig. 50G ), as well as Hymenosomatidae ( Figs. 29A, B , 58A, B, E ) when included in Majoidea . A trend towards a coxo-sternal condition occurs in particular in Inachidae ( Fig. 49B ) and, at a lesser extent, in some Inachoididae (See Modalities of penis protection: Coxo-sternal protection; Affinities between Inachoididae and Inachidae ).

There is considerable instability in the taxonomy of Majoidea , which consists of about 950 species distributed in 200 genera grouped in six families by Ng, Guinot & Davie (2008: 98) and Ahyong et al. (2011: 187) and, in addition, Pisidae . Substantial changes can be expected. Despite several additions in recent years (e.g., Richer de Forges & Ng 2009 a, b, 2012; Windsor & Felder 2009, 2011; Guinot 2012; Richer de Forges & Corbari 2012) the taxonomic relationships between the families remain unresolved. Many subfamilies were recognised in the Majidae ( Dana 1851a; Miers 1879; Young 1900). The large family Inachidae as listed by Ng, Guinot & Davie (2008: 110) without subdivisions is polyphyletic, whereas Števčić (2005), who only recognised a subfamily rank Inachinae (included in Majidae ), recognised 17 inachine tribes. Guinot (2012) recently proposed the resurrection of several subfamilies besides the Inachinae emend, i.e., Podochelinae Neumann, 1878 , Anomalopodinae Stimpson, 1871 , and, possibly, Eucinetopinae Števčić, 2005. The Inachidae , and Inachoididae are treated here separately and in more detail (see Male gonopores among selected taxa of Eubrachyura: Family Inachidae ; Family Inachoididae ; Monophyletic Heterotremata: Superfamily Majoidea ; Affinities between Inachoididae and Inachidae ).

Marques & Pohle (2003), in re-evaluating larval support for the monophyly of majoid families, found that most majoid families were paraphyletic (except for Oregoniidae ), with Oregoniidae and the Inachidae + Inachoididae groups (save for Macrocheira De Haan, 1839 ) forming a clade in unconstrained analyses. Based on CO1 and 18S sequences, Sotelo et al. (2009) found Schizophrys White, 1848 , to be basal to Maja (see also Simeó et al. 2010). A molecular phylogeny using three loci (16S, COI, and 28S) from 37 majoid species by Hultgren & Stachowicz (2008b) supported several relationships predicted from larval morphology, including a monophyletic Oregoniidae branching close to the base of the tree, a close phylogenetic association among Epialtidae , Pisidae Dana, 1851 , Tychidae Dana, 1851 , and Mithracidae MacLeay, 1838 , some support for the monophyly of Inachidae and Majidae , and no support for the other families. Another analysis by Hultgren et al. (2009), using sequence data with three loci and 53 characters of larval morphology from 14 genera representing 7 majoid families, obtained similar results: the branching of Oregoniidae at the base of the tree, confirming that this family is one of the oldest majoid lineages (see also Rice 1980, 1988; Clark & Webber 1991; Pohle 1991; Marques & Pohle 1998) but without resolving its position relative to the remaining majoids, and different trees for the resolution of other relationships, namely the Inachidae and Tychidae relative to the remaining majoids. The adult morphological characters traditionally used to classify majoids into different families were thus found to be subject to convergence in diverse cases. Hultgren et al. (2009: 448) pointed out the importance of a rigorous testing in order to evaluate the monophyly of Majoidea and to properly recognise the higher-ranked systematics of the group, in particular whether it includes the Hymenosomatidae . Analyses based on arginine kinase sequences combined with those based on the mitochondrial cytochrome oxydase I gene provided strong support for the superfamily Majoidea but data were in conflict with the current assignment of some genera to the Mithracidae and Pisidae ( Mahon & Neigel 2008) . Recent molecular analyses of Mithracidae based on three mitochondrial genes in addition to morphological characters ( Windsor & Felder 2009, 2011) highlight important results that necessitate a reappraisal of the current classification. The species of Epialtidae and Pisidae of the Californian-Oregonian region were considered separate families by Wicksten (2012) because they are distinct in morphology and habitat.

Porter et al. (2005: table 1, fig. 2; see also Crandall et al. 2009: 296), based on a molecular estimation of decapod phylogeny, pointed out a divergence of Majoidea from the rest of Brachyura approximately 254 million years (i.e., Permian) and recovered Majidae as an older lineage than Cancridae . The record of an unequivocal majoid, from the Cenomanian of France, the † Priscinachidae Breton, 2009 , a family close to Inachidae and with punctae evoking traces of lost setae (see Breton 2009), demonstrated the first occurrence of a majoid in the Upper Cretaceous that contrasts to the traditional Eocene origin previously acknowledged ( Glaessner 1969: R440, R502). Majoidea is thus recognised here as a deeply rooted lineage, with the Inachidae + Inachoididae and the Oregoniidae as the more ancient groups. See Affinities between Hymenosomatoidea and Inachoididae ; Affinities between Inachoididae and Inachidae .

Many majoids protect themselves from predation by camouflage strategies such as decoration behaviour and colour change (see Concealment behaviour; Decoration behaviour). The energetically costly production of the hooks used as holding material and the time spent searching for and maintaining decoration material, constrain allocation trade-offs with functions such as growth and reproduction, hence an important role in majoid evolution ( Wicksten 1983, 1993, 2012; Stachowicz & Hay 1999, 2000; Thanh et al. 2003; Berke & Woodin 2005, 2008; Berke et al. 2006; Hultgren & Stachowicz 2008a, 2009; Sallam et al. 2011).

Superfamily Orithyioidea . The complicated taxonomic history of all the families that were traditionally included in Oxystomata De Haan, 1835 (Oxystomes H. Milne Edwards, 1834) reflects the difficulty in recognising monophyletic high-ranked taxa and clear relationships (see Bellwood 2002a, b, for detailed historical account).

The placement of the monogeneric Orithyiidae , known only by Orithyia sinica , has been a source of debate. The family has been placed either within Calappoidea ( Ortmann 1892; Balss 1957; Alcock 1896; Kim 1973; Sakai 1976; Dai & Yang 1991; Chen & Sun 2002) or Dorippoidea ( Ihle 1918; Guinot 1978a, 1979a; Bellwood 1996, 2002b; Guinot & Bouchard 1998; Schweitzer & Feldmann 2000a; Martin & Davis 2001; Schweitzer et al. 2010). The placement of the Orithyiidae in its own superfamily, Orithyioidea , proposed by Števčić (2005: 102), was accepted by Ng, Guinot & Davie (2008: 125, figs. 93–95) and De Grave et al. (2009: 37). Studies on the larval characters of Orithyia have not been conclusive ( Hong 1976; Rice 1980: 316, 356; 1981a: 293). According to Rice (1980: 356, 357), larval stages of Orithyia are so distinct from those of all other crabs, including Calappidae , that they “suggest an isolated position that should be recognised by separate family status, at least”, and that “extant dorippid zoeae have several advanced features that exclude them from the possible ancestry of Orithyia ”.

Bellwood (1996) diagnosed Orithyiidae as a monophyletic group supported by nine synapomorphies. Additional characteristics of the orithyiids are: a full condylar protection of the penis, which emerges directly from the extremity of the large coxo-sternal condyle of the P5 coxa as a thick, short, and cylindrical sclerotised tube enclosing the ejaculatory duct, the articulation of the condyle with the sternite being concealed by the penis ( Fig. 31A ; see Modalities of penis protection: Condylar protection); a penial papilla folded inside the tube, supposedly devaginating for mating; the thoracic sternum circular, with a narrow somite 8 ( Ng, Guinot & Davie 2008: fig. 95); the vulvae externolateral, not hidden under the short abdomen; and the male abdomen short and lacking structures for a locking system. A vestigial biramous uropod on abdominal somite 5, consisting of a long ramus on one side and only a bud on other side ( Fig. 31F ), has been found in a male O. sinica , 80.0 × 74.0 mm (MNHN-B11612). Orithyia sinica , the “tiger crab”, which has a high potential for aquaculture, particularly in Korea ( Kim & Chung 1990; Koo et al. 2005; Jee et al. 2007), has been widely studied but its reproductive biology remains largely unknown.

Bellwood (1996) included the Orithyiidae in Dorippoidea , but the absence of Ethusidae in the cladistic analysis unfortunately limited her conception of the Dorippoidea . Although a relationship between orithyiids and dorippoids cannot be completely discarded, Orithyiidae lacks the synapomorphies that are shared by dorippids and ethusids (see Superfamily Dorippoidea above). According to Bellwood (1996: 183) the “smooth sclerotized sheath around the penis” was the most distinctive synapomorphy recognising the clade Dorippe + Orithyiinae. The penis actually shows as a short papilla in several families of Eubrachyura including Atelecyclidae , Bythograeidae , Calappidae , Carpiliidae , Cancridae , Cheiragonidae , Corystidae , Matutidae , Menippidae , Parthenopidae , Pilumnidae , Platyxanthidae , Thiidae , and Xanthidae . In other heterotreme families the penis is longer and variously enclosed in a sclerotised sheath. In Orithyia the penis is but a short tube, dorsally covered by a sclerotised sheath and ending in a soft papilla ( Fig. 31A ).

Orithyia could be related to the Dorippidae , with which it shares a short and incompletely folded abdomen, although proportionally more reduced in Dorippidae , but, however, not covering the vulvae and devoid of locking structures in Orithyia . Larval characters could also support orithyiid-dorippid relationships, the zoeae of Orithyia having long cephalothoracic spines as well as telson forks (Hong 1979; Rice 1980). Both dorippid and orithyiid zoeae have a combination of characters unknown in any other groups and showing a “degree of advancement” typical of the thoracotreme families but found also in the Inachinae ( Rice 1980: 317, 319). These larval traits are regarded here as those of ancestral heterotremes (see Position of the Dorippoidea within the Brachyura ). The zoeae of Matuta (e.g., Rajabai 1959; Hashmi 1969; Terada 1987) differ from the calappid zoeae and similarly resemble those in some of the thoracotreme families ( Rice 1980: 316).

The Milne Edwards openings of Orithyiidae , which show as narrow, deep slits concealed behind the cheliped bases (sternum/pterygostome junction absent), are distinct from the separate apertures in Dorippidae , which are characterised by a complete sternum/pterygostome junction ( Fig. 42C ), in contrast to Ethusidae , which do not show marked modifications and have normal afferent openings ( Fig. 42A ). The disposition of the penis of Orithyiidae , with a full condylar protection ( Fig. 31A ), is distinct from that of the Dorippoidea , in which the penis does not emerge from the extremity of the coxo-sternal condyle and has its proximal portion protected within a bulb. Another important difference between Dorippoidea and Orithyiidae is the penial arrangement of both Dorippidae and Ethusidae , with multistate characters of the coxo-sternal condition that correspond to a posterior widening of the thoracic sternum ( Figs. 15 – 22 ), whereas there is no well-developed posterior sternal widening in Orithyiidae .

The Orithyiidae is distinctive among Heterotremata by a unique combination of characters: thoracic sternum circular, wide but remaining narrow posteriorly, flat; sternites 1, 2 forming a triangular shield, separated from sternite 3 by a suture; sternites 4–7 rather developed; sternite 8 much less developed; sutures 4/5–7/8 long, separated only medially by narrow space ( Guinot 1979a: pl. 14, fig. 7; Bellwood 1996: fig. 2D); episternites 4–7 narrow, elongated, delimited ( Ihle 1918: fig. 91); median line on sternites 7 and 8; sterno-abdominal cavity long, smooth; male and female abdomens very short; vulvae lateral, exposed ( Guinot 1979a: pl. 14, fig. 9; Guinot & Bouchard 1998: 648, fig. 15B); absence of locking-abdominal structures ( Guinot & Bouchard 1998: 648, fig. 15A); G1 finger-like ( Dai & Yang 1991: fig. 57); male gonopore very far from suture 7/8, in a posteriormost location in relation to sternite 8; peculiar axial skeleton ( Guinot 1979a: 103, 260, pl. 14, figs. 7, 8). Additional features are the buccal cavern, with extremely specialised mouthparts (see below) and the modifications of the pereopods, P2–P4 with styliform, carinate dactyli, P5 paddle-like.

The Matutidae (see revision by Galil & Clark 1994) shares with Orithyiidae several features, including developed anterior thoracic sternites, delimited episternites 4–7 ( Ihle 1918: fig. 90), male gonopore in a posteriormost location in relation to sternite 8, and a finger-like G1 ( Dai & Yang 1991: figs. 55, 56). There are also some similarities in the respiratory system of the two families, with the presence of Milne Edwards openings as long and narrow slits in front of the cheliped bases, and mxp3 exopodite lacking flagellum ( Bellwood 1996: fig. 3C, D). In the Matutidae the exhalant channels are located on each part of a median septum (H. Milne Edwards, Atlas, 1834 , pl. 20, fig. 4; 1836–1844, Atlas , pl. 7, fig. 1a, b; Garstang 1897b; Ihle 1918: 169, fig. 87; Števčić 1983: 167). Orithyia is more specialised, with an unpaired septum separating two distinct exhalant channels transformed into two tubes, looking like two short tunnels ending in two rounded orifices (H. Milne Edwards 1837a: 111; 1836– 1844, Atlas , pl. 8, fig. 1a; Ihle 1918: 169, fig. 88; Ng, Guinot & Davie 2008: fig. 94). The endostomial septum of Matutidae could have the same origin as that of the more complex arrangement of Orithyiidae . The penis emerges from the extremity of the coxo-sternal condyle in Orithyiidae ( Fig. 31A ; see Modalities of penis protection: Condylar protection), whereas the perforation is located in a deep notch on the anterior margin of the P5 coxa in Matutidae .

The distal articles of the P2–P5 are modified in Orithyiidae , notably the P2–P4, with thick, sharp, carinate dactyli (for digging), P4 and P5 with short, flattened carpus, and P5 with paddle-like dactylus (for swimming). The P2–P5 are also strongly modified in Matutidae , the P2 and P5 being paddle-like. These features are clear adaptations for burying and swimming in both Orithyiidae and Matutidae ( Seidel 1976; Basson et al. 1977; Števčić 1983; Perez & Bellwood 1989). Such a modification of the P2 and P3 is rare in Eubrachyura, the Matutidae being characterised by a peculiar pre-copulatory grasping position (the male grasps the chelae of the female) allowing the pair to continue to swim and bury freely. The first two abdominal somites that are similarly rigid and carinate in both families are probably used to anchor the body to the substratum.

The Orithyiidae (sensu Guinot 1979a: pl. 14, figs. 7–9; Bellwood 1996: fig. 2D) displays a thoracic sternal pattern 5, subpattern e ( Fig. 56K ): sutures 4/5–6/7 interrupted although with relatively close interruption points; suture 7/8 more widely interrupted than preceding sutures; median line extending along sternites 7, 8; high median plate at sternite 7 level, partially on 8, before connecting to a rather wide sella turcica; endosternites 5/6, 6/7 with invaginated extremities as in glove fingers. The Matutidae (see Guinot 1979a: pl. 14, figs. 4–6; figure 2C of Bellwood 1996 is incorrect) has a peculiar thoracic sternal pattern 5 (not represented in Fig. 56 ): suture 4/5 interrupted, although with rather close interruption points; suture 5/6 seemingly complete but actually with a narrow gap between the interruption points; suture 6/7 with distal extremities anteriorly directed and interrupted; suture 7/8 seemingly complete in males but interrupted in females; median line discontinuous, being long on sternite 5, short on sternite 6, then interrupted, becoming continuous along sternites 7, 8; median plate low on sternite 5, high and vertically directed along posterior sternites before connecting with narrow sella turcica; and endosternites connecting medially. The sternal suture 7/ 8 in its horizontal position has an anteriormost location relative to the P5 coxal perforation in both Orithyiidae and Matutidae , a significant similarity.

It is hypothesised that the characters shared by Orithyiidae and Matutidae are the result of common ancestry, and that both families belong to the same lineage, with a more highly specialised elaborate respiratory system in Orithyiidae . Carcinisation is moderate in Orithyiidae , which is evidenced by sutures 4/5–7/8 being weakly interrupted and the wide sternal plastron remaining posteriorly narrow. There is only a moderately wide sternite 8 in Orithyiidae ( Bellwood 1996: fig. 2D), smaller in Matutidae ( Ihle 1918: fig. 90). A shortening of the abdomen has taken place in Orithyiidae , hence the empty portion of the sterno-abdominal cavity, the exposed vulvae, and the loss of any locking structures. The exposure of vulvae in Orithyiidae , Corystidae , Cheiragonidae , and Bellia (Belliidae) probably evolved in parallel in these families.

The Orithyiidae , puzzingly absent from fossil records (see Schweitzer & Feldmann 2000a: 243; Schweitzer, Feldmann, Fam, Hessin, Hetricks, Nyborg & Ross 2003: 31, 32; Schweitzer et al. 2010: 80), seems to be the only living representative of a subclade having a long evolutionary history. The Matutidae , with a “normal” abdomen shteltered in a deep and narrow sterno-abdominal cavity, vulvae and abdominal-locking structures, is supposedly more recent. Fossil matutids are known from the Oligocene, Miocene and Eocene ( Müller & Galil 1998; Karasawa 2002; De Angeli & Marchiori 2009; Fraaije et al. 2012). †Eomatutinae De Angeli & Marchiori, 2009, was established for the oldest fossil, † Eomatuta De Angeli & Marchiori, 2009 , from the Middle Eocene ( De Angeli & Marchiori 2009: 106, figs. 2, 3), with a densely tuberculated carapace and the absence of an epibranchial spine.

Superfamily Palicoidea . The Palicidae was considered assignable or close to Dorippidae ( Faxon 1895: 38, as Cymopoliidae ; Bouvier 1897a: 785, 787 footnote, as Palicés; 1897b: 3, 11, as Palicae or Palicinés; 1898: 105 footnote, as Palici; A. Milne-Edwards & Bouvier 1900: 21; 1902: 40, as Palicae; Guinot 1978a: 249; Moosa & Serène 1981: 22, as Palicinae and Crossotonotinae). Larval characters of Palicidae have shown a relationship with Dorippoidea ( Cano 1891b, 1893a; Gurney 1942; Bourdillon-Casanova 1960). Rice (1980: 319; see also Rice 1981a), however, recognised a “degree of advancement” in the larval characters close to that of the catometope families but also pointed out unique features suggesting a distinct evolutionary history. Larval morphology supports the recognition of the two families, Palicidae and Crossotonotidae , within Palicoidea , the zoeae hatching in an advanced stage of development for two characters (biramous mxp3 and pereopods with a bilobed chela) and distinguishable by three characters ( Clark et al. 2012). The Palicidae was sometimes placed close to Retroplumidae (e.g., Balss 1957: 1633, 1661), a view tentatively followed by Guinot & Quenette (2005) and Guinot & Breton (2006: 619, 620).

In a study of the molecular phylogeny of Grapsoidea sensu lato, Schubart, Cuesta, Diesel & Felder (2000: 181, fig. 1, as Palicidae consisting of Palicinae Bouvier, 1898, and Crossotonotinae Moosa & Serène, 1981), using Palicus obesus (A. Milne-Edwards, 1880) as outgroup, did not result in the inclusion of Palicoidea in Thoracotremata. Molecular sequences corroborated the monophyly of Palicoidea , with its taxonomic position nevertheless still remaining unresolved ( Schubart, Neigel & Felder 2000a: 826, 827, fig. 1). A suprafamilial status for Palicidae and Crossotonotidae was given by Števčić (2005: 104), Ng, Guinot & Davie (2008: 127), De Grave et al. (2009: 37), and Castro (2010). From molecular analyses using 16S rDNA sequence data (the same gene used by Schubart, Neigel & Felder 2000a) of Palicus caronii and Crossotonotus spinipes, Wetzer et al. (2009: 487 , fig. 1, table 2) excluded Palicoidea from the Thoracotremata in their phylogenetic tree, but placed the group as basal to thoracotremes, “basal to the other (non out-group) crabs”.

The Palicoidea clearly belongs to the heterotreme assemblage. The modification of sternite 8 for penis protection arose homoplasically in Palicoidea ( Fig. 32 ) and Chasmocarcinidae ( Fig. 24 ) (see Modalities of penis protection). The affinities between Palicoidea and related taxa have presented a dilemma, tentatively resolved to a certain extent here (see Affinities between Palicoidea , Retroplumoidea and Hexapodoidea ). There are complex relationships between the carapace, thoracic pleurite 8, thoracic sternite 8, and the first abdominal somite in Palicoidea . The wide variety of the G1 among Palicidae (see Castro 2000) warrants a re-examination of the taxonomic relationships within the family. The worldwide distribution of Palicidae (eight genera with nearly 60 known extant species, see Ng, Guinot & Davie 2008: 127), in addition to several plesiomorphic characters (see Axial skeleton; Thoracic sternum), suggests a deeply rooted lineage. The socket, which is either posterior on abdominal somite 6 as in Palicus caronii ( Fig. 50H ) and Pseudopalicus bidens Promdam & Nabhitabhata, 2012 ( Promdam & Nabhitabhata 2012: fig. 3A, C) or in the median portion in other species, as well as the lateral displacement of the skeletal phragmae ( Fig. 45A ), and the sternal pattern 5, subpattern 5d ( Fig. 56J ), are characteristic of Palicidae .

The Crossotonotidae , which is a smaller family comprising two genera and six species restricted to the Indo-West Pacific region ( Castro 2000), is distinguished from Palicidae by the condition of somite 8 and, to a lesser extent, the arrangement of somite 7. The crossotonotid modifications concern basically sternite 8 (secondarily sternite 7), the nature of P5 (position, orientation, relative size, no carrying function), and the shape of the penis. In addition, the modification involves the first two abdominal somites, which are wide in both male and female crossotonotids ( Castro 2000: 445, fig. 45d), in contrast to the very narrow and dorsoventrally compressed somites in male and female palicids ( Castro 2000: 445, fig. 2C, D).

Additional fossil Heterotremata. † Goniochele Bell, 1858 (type species by monotypy: G. angulata Bell, 1858 , from the Eocene) is particularly well documented thanks to the presence of well preserved male and female abdomens of † G. angulata (see Bell 1858: 25, 26, pl. 4, figs. 8, 9; Carter 1898: 23, pl. 1, fig. 6), the dorsal P4 and P5 of † G. angulata evident from the disposition of the coxae ( Bell 1858: pl. 4, fig. 5), and the thoracic sternum of † G. angulata and † G. madseni Collins & Jakobsen, 2003 ( Carter 1898: 23; Collins & Jakobsen 2004: pl. 3, figs. 2a, 4a). † Goniochele was referred to † Necrocarcinidae (included in Calappoidea) by Schweitzer & Feldmann (2000a: 241) and to † Orithopsidae (included in Dorippoidea ) by De Grave et al. (2009: 31, fig. 2G) and Schweitzer et al. (2010: 82), both families actually podotreme ( Guinot, Vega & Van Bakel 2008: 711; Van Bakel, Guinot, Artal, Fraaije & Jagt 2012: table 1). The female sternum of † G. angulata and † G. madseni , with the sternites 1–3 forming a triangle between mxp3 and suture 5/6 lacking a strong curve, definitively excludes being a dorippid. The absence of any references to the presence of vulvae on sternite 6 in both figured sterna of † G. angulata and † G. madseni is puzzling, leaving open the question that the genus could be a podotreme; the “ventral” characters, however, in particular the rather wide thoracic sternum, seem diagnostic of Eubrachyura. The dorippoid affinities of † Goniochele suggested by Guinot, Vega & Van Bakel (2008: 709) have been recognised by the establishment of a separate family within Dorippoidea , † Goniochelidae Schweitzer & Feldmann, 2011 ( Schweitzer & Feldmann 2011a).

The difficulty of the taxonomic position of early crabs is demonstrated by the case of the Maastrichtian † Binkhorstia Noetling, 1881 , which consists of † B. ubaghsii (Van Binkhorst, 1857) , the type species, and † B. euglypha Collins, Fraaye & Jagt, 1995 . The genus was assigned to Dorippidae (see Glaessner 1969: R492; Quayle & Collins 1981: 738) but also referred to various families or subfamilies, either podotreme or supposedly eubrachyuran: † Carcineretidae (see Wright & Collins 1972: 91; Collins et al. 1995: 199; Fraaye 1996a; 1996b: 271; Jagt et al. 2000: 40; see also Wright 1997: 138), Cyclodorippidae (see Feldmann & Villamil 2002: 721), † Necrocarcinidae (see Collins 2003: 85; Collins & Jakobsen 2004: 71; Larghi 2004: 530), and more recently in † Torynommatidae (see Glaessner 1980: 181; De Grave et al. 2009: 29; Schweitzer et al. 2010: 78). The ventral surface of † B. ubaghsii (see Van Bakel, Jagt, Fraaije & Coole 2003: 87, fig. 1), showing in particular a wide male thoracic sternum and a rather long male abdomen inserted in a true sterno-abdominal cavity, have provided evidence for the non-podotreme nature of † Binkhorstia . Moreover, typical eubrachyuran vulvae are certainly present in the female (B. van Bakel, pers. comm. 2008). The heterotreme † Binkhorstia could have some affinities with Retroplumoidea , as shown by the wide thoracic sternum, spatulate rostrum, flattened P2–P4, and P5 that are reduced and placed laterally to the first abdominal somite. Several characters, however, do not support an assignment to either Retroplumoidea or Dorippoidea : the strongly heterochelous and homodontous chelipeds, absence of cristiform chelae, presence of a molariform tooth on the larger chela being indicative of a shell-breaking behaviour ( Van Bakel, Jagt, Fraaije & Coole 2003: fig. 1.1, 1.2; Dietl & Vega 2008: 292), and a narrow, vertically oriented thoracic sternite 7 in the male. † Binkhorstia could belong to a not yet described extinct eubrachyuran family (Artal et al., unpublished).

The discovery of a complete thoracic sternum in † Componocancroidea Feldmann, Schweitzer & Green, 2008, from the Middle Cretaceous (Albian) is remarkable. The presence of vulvae in † Componocancer Feldmann, Schweitzer & Green, 2008 , as assumed here, makes impossible an assignment to the Podotremata. That † Componocancer may represent “an early intermediary between the dromiaceans and the eubrachyurans” ( Feldmann, Schweitzer & Green 2008: 508) is unsustainable in several respects. Firstly, the “ Dromiacea ”, the “ Brachyura Primigenia ” of Alcock (1900a), cannot have evolved into Eubrachyura. Moreover, the “triangular sternum in which the sternites are unfused laterally […] strongly reminiscent of some palinurid and scyllarid lobsters” and described as a “unique sternal architecture” ( Feldmann, Schweitzer & Green 2008: 505, 506; see also Feldmann et al. 2007: 28), bears only a slight resemblance to that of a podotreme. It could simply represent a plesiomorphic heterotreme state. A narrow, segmented thoracic sternum with elongated, narrow, triangular sternites 1–3 is found in some extant heterotremes, e.g., Cancridae (Rathbun 1930: pl. 83; Davie 1991: figs. 2, 5, 7; Schweitzer & Feldmann 2000c: 224), Atelecyclidae ( Guinot 1979a: pl. 9, figs. 4–6; Tavares & Cleva 2010: figs. 5E, 7D), Cheiragonidae ( Salva & Feldmann 2001: figs. 32B, 33B), Trichopeltariidae ( Guinot & Bouchard 1998: fig. 14A; Salva & Feldmann 2001: figs. 18C, 19A, 20C, 21B; Tavares & Cleva 2010: figs. 5F, 6B, 7B, 22D, 27C, 31D, 34D, 35C), extant and fossil Corystidae ( Guinot 1979a: pl. 9, fig. 1; Guinot & Bouchard 1998: fig. 12A; Van Bakel, Jagt, Fraaije & Wille 2004: 101, fig. 3, pl. 4, figs. 1, 2, 6–8; Van Bakel, Jagt, Artal & Fraaije 2009: fig. 2E). The combination of anterior sternites forming a broadly triangular, compartmented shield, with enlarged, unfused posterior sternites (sutures 4/5–7/8 all interrupted), characterises for instance the Dorippidae and Ethusidae ( Fig. 42A, C ). The condition in † Componocancer with reduced and dorsally directed sternites 7, 8, supposedly indicative of reduced, dorsally carried P4 and P5 (P5 being smaller), could suggest dorippoid affinities. A comparison of † Componocancer with two other extinct taxa included in Dorippidae by De Grave et al. (2009) and Schweitzer et al. (2010) could be valuable: (1) † Tepexicarcinus Feldmann, Vega, Applegate & Bishop, 1998 , from the Albian, with an incomplete thoracic sternum, and reduced, subdorsal P4 and P5 (Feldmann et al. 1998: fig. 7; Vega et al. 2006: fig. 4, pl. 2, figs. 2–12); and (2) † Sodakus Bishop, 1978 , from the Maastrichtian, with a comparable thoracic sternum ( Bishop 1978: fig. 3C; 1986b: 135; Vega et al. 1995: fig. 4.2). † Componocancer was diagnosed by interrupted sutures 3/4, 4/5, 5/6, 7/8 (“sternites 7 and 8 fused axially”) and complete suture 6/7 ( Feldmann, Schweitzer & Green 2008: 506, fig. 2D). Such a complete suture 6/7 does not satisfactorily support a dorippoid affiliation (see Monophyletic Heterotremata: Fossil Dorippoidea ). Morphological comparisons with other brachyurans lack in the description of † Componocancroidea , and the related putative suprafamilial or familial brachyuran taxa have not been outlined. The reduced and dorsal last pereopods limit the choice of the eubrachyuran groups to which the family may be assigned. We agree with De Grave et al. (2009: 31), who listed † Componocancridae Feldmann, Schweitzer & Green, 2008 , as basal in Heterotremata. The † Componocancridae provides new evidence that the Eubrachyura appeared at the very least in the Albian, much earlier than previously thought. The Upper Cretaceous † Marocarcinidae Guinot, De Angeli & Garassino, 2008 , shows apparent similarities with † Componocancridae , e.g., developed sternite 3, interrupted sternal sutures 4/5, 5/6, 7/8 and complete suture 6/7 ( Guinot, De Angeli & Garassino 2008: figs. 1B, 2; Martill et al. 2011). The two families, however, are distinctive: (1) the reduced and dorsally positioned sternites 7 and 8 of † Componocancridae , supposedly resulting in reduced, dorsal P4 and P5, contrast with the ovoid sternum and the well-developed, “normal” last pereopods of † Marocarcinidae ; (2) the suture 7/8 is incomplete in † Marocarcinidae at least in females, complete in † Componocancridae ; (3) the † Componocancridae lacks the median line (see Thoracic sternum) that is present along the posterior sternites in † Marocarcinidae , at least in females. The † Marocarcinidae is incertae sedis in De Grave et al. (2009: 46) and simply included in Eubrachyura without suprafamilial attribution in Schweitzer et al. (2010: 70).

The corystid † Hebertides jurassica ( Guinot et al. 2007b) is a heterotreme from the Miocene (instead of from the Middle Jurassic) ( Taylor et al. 2012).

Well-preserved ventral structures in fossils now provide sufficient data for an assignment to Podotremata or Eubrachyura, but this was neglected for a long time. The case of † Dromilites americana Rathbun, 1935 , is representative: the ventral surface figured by Rathbun (1935: pl. 17, fig. 2) was clearly that of a eubrachyuran, but the species was kept in Dromiidae by Schweitzer et al. (2010: 64) and Franṭescu et al. (2010: 264), who considered “the placement of Rathbun (1935) as being correct” instead of following the heterotreme placement (in Goneplacidae ) proposed by Armstrong et al. (2009). Franṭescu et al. (2010: 264) remarked that the specimens illustrated by Armstrong et al. (2009) as † Tehuacana americana ( Rathbun, 1935) represented more than one species. The nomen † Marydromilites , established without a description or definition by Števčić (2005: 134), is not valid ( Ng, Guinot & Davie 2008). †Tehuacanini Števčić, 2011, is a tribe of Geryoninae Colosi, 1923, in Geryonidae , created by Števčić (2011: 132). See Other extinct putative Podotremata.

Monophyletic Brachyura

The Brachyura is a monophyletic group encompassing Podotremata and Eubrachyura, this monophyletic condition implying an exclusive common ancestor for all its components, and thus without multiple origins.

The Brachyura is in fact the only group among the traditional suborders of Decapoda to be generally recognised as monophyletic ( Scholtz & Richter 1995; Crandall et al. 2000; Schram 2001; Dixon et al. 2003; Ahyong & O’Meally 2004; Schram & Dixon 2004; Porter et al. 2005; Ahyong et al. 2007; Tsang et al. 2008; Chu et al. 2009b; Toon et al. 2009). As in earlier classification schemes, and on the basis of anomuran larval features and molecular data, the Dynomeniformia (under the nomen Dromiacea ) was still considered to belong in Anomura until relatively recently ( Gurney 1942; Kircher 1970; D.I. Williamson 1976; Fincham 1980; Lang & Young 1980; Rice 1980; 1981a; Laughlin et al. 1982; Cronin 1986; Spears & Abele 1988; Spears et al. 1993; Spears & Abele 1996; Camp et al. 1998). This anomuran dromiid hypothesis is now refuted (e.g., Guinot & Tavares 2001; Lemaitre & Tudge 2003; Ahyong & O’Meally 2004; Ahyong et al. 2007; Tudge et al. 2012), and the inclusion of Dynomeniformia ( Dromiacea ) in Brachyura is currently accepted, the major problem being the status of podotreme crabs (see Monopphyletic Podotremata). Lemaitre & McLaughlin (2009: 128), quoting Tavares (2003) and Ahyong et al. (2007), concluded to a “still conflicting evidence on the relationships between the Anomura and the ‘primitive crabs’ or Podotremata”. The phylogeny of the Anomura is a highly variable infraorder that, as the infraorder Brachyura , remains a source of conflicting interpretations ( Lemaitre & McLaughlin 2009; Ahyong, Schnabel & Maas 2009; Tsang et al. 2011).

Calibrations based on marine brachyuran fossils date their last common ancestor before the fragmentation of Gondwana ( Klaus et al. 2011; see Earliest brachyurans below). Major events have occurred in the evolution of Brachyura , and the sequence to explain the brachyuran organisation is only a matter of postulation. Most of these events are (not in sequence): reduction and folding of the abdomen, with changes in the intersomital articulations; loss of the tail fan (i.e., of the rami of the uropods on either side of the telson), causing the abandonment of the escape reaction; loss of locomotory pleopods; internal organs mainly excluded from the abdomen; reorganisation of sexual organs, which led to a suite of secondary morphological modifications; arching of the posterior thoracic somites; and development of a holding- or locking-abdominal mechanism. These events are interrelated, and it is not easy to understand one without the others. The evolution towards a crab-like body implies the broadening of the cephalothorax, in particular that of the thoracic sternum (see Thoracic sternum; Carcinisation and its outcomes).

The Podotremata synapomorphically shares a paired spermatheca, and Eubrachyura is supported synapomorphically by a paired vulva. The common ancestry of these two major clades is supported by several synapomorphies, which include: (1) reduction of the abdomen, with the loss of the internal organs other than the intestine and, in males, of the posterior abdominal appendages; (2) folding of the abdomen under the cephalothorax against the thoracic sternum, with only a flexible articulation between the somites ( Fig. 51 ); (3) holding of the abdomen due to various devices, at least involving abdominal somite 6; (4) modification of the uropod (no longer biramous and forming a tail fan with the telson as in other Decapoda ; see Wilson 1996) to form a dorsal plate, or ventral lobe (Dynomeniformia; see Guinot & Tavares 2003: 110, table 1), or socket on abdominal somite 6 (Homoliformia, Lyreididae , and Eubrachyura); (5) pairing of the seminal receptacles (podotreme spermatheca and eubrachyuran seminal receptacle), instead of medially located as in other Reptantia ( Ahyong & O’Meally 2004: 691; Guinot & Quenette 2005: 173); (6) protrusion of a penis from the male gonopore; (7) sperm transfer via a pair of gonopods (G1 and G2) acting together with the penis ( Dixon et al. 2003: 952); (8) development of sternite 4, always larger than the preceding sternites; (9) presence of an endosternal intertagmal phragma that connects the tagma/thorax and the tagma/abdomen to thoracic interosternite 7/8 (“brachyuran sella turcica”, as redefined here); (10) sperm ultrastructure, with acrosome shortened to nearly spheroid shape and generally with loss of corrugations of perforatorial chamber wall ( Jamieson et al. 1995; Jamieson & Tudge 2000; see also Tudge 2009).

Other features that give additional evidence to the monophyly of Brachyura are: (1) reduction of the rostrum and broadening of front; (2) progressive formation of orbital cavities in which the eyestalks can be protected (see Cephalic condensation); (3) reduction in size and folding of antennules and antennae in individual fossae; (4) junction of the pterygostomial region of carapace with epistome; (5) contact of the branchiostegite with the pereopod coxae (peculiar configuration, however, in Raninoidea , which are gymnopleure); (6) transformation of P1 into strong chelipeds; (7) expansion of the mxp3 ischium and merus to form a plate-like covering of the other mouthparts; (8) loss of the pair of pleopods on the first female abdominal somite ( Gymnopleura ), although a vestigial, uniramous pleopod 1 remains in some podotremes, e.g., in Dynomeniformia, Homoliformia, and only a few Cyclodorippiformia (e.g., Xeinostomatinae Tavares, 1992 ; see Tavares 1992b: 514, as Xeinostominae ; 1994a: 120); and (9) development of apodemes forming walls of muscle-cavities in the thorax linked to the locomotory thoracic limbs ( Bourne 1922b: 27, 72).

Other brachyuran synapomorphies have been listed by Dixon et al. (2003: 966). The posterior border of the carapace that was indicated as straight (see Dixon et al. 2003: 948, 966, table 1) is actually concave in most podotremes ( Figs. 39A , 51C, E, F ), generally straight in other brachyurans.

Another characteristic of the Brachyura is sideways and diagonal locomotion, a highly specialised adaptation that involves a number of unique morphological and physiological features ( Schäfer, 1954; Lochhead 1961; Clarac & Coulmance 1971; Martinez et al. 1998; Martinez 2001; Schreiner 2004; Vidal-Gadea et al. 2008; Vidal-Gadea & Belanger 2009). Terrestrial walking behaviour benefits from the walking legs being free from abdominal holding in heterotremes and, in thoracotremes, from reproductive functions. Forward locomotion is, however, the rule, or at least frequently used, in podotremes; some rare heterotremes (e.g. dorippids) also move forwards, whereas mictyrids are a notable exception in thoracotremes (see Locomotion; Carcinisation and its outcomes).

Reduction, folding and holding of the abdomen. The reduction of the abdomen in all Brachyura is achieved with the subsequent elimination of its locomotory function. The brachyuran abdomen clearly evolved from a well developed, long, and wide abdomen similar to that in the long-bodied Decapoda . Carcinisation that leads to crab-like forms in Anomura is similarly achieved via a progressive broadening of the cephalothorax and shortening of the abdomen. The anomuran male abdomen is nevertheless held partially “tucked under” the cephalothorax and fully folded consecutively to a further significant reduction ( McLaughlin & Lemaitre 1997; Tsang et al. 2011; Reiman et al. 2011) but it is never locked as in Brachyura .

The male brachyuran abdomen remains proportionally long and wide, filling the whole space between the pereopods, with an elongated telson reaching the mxp3, in the podotremes retaining many plesiomorphies, e.g., Dromiidae pro parte; Homolodromiidae (in which the telson can be only slightly shorter than the rest of the abdomen); Homolidae and Poupiniidae ; Dynomenidae pro parte ( Acanthodromiinae , Paradynomeninae ). It is shorter in the remaining Dynomenidae (Dynomeninae, Metadynomeninae), in Latreilliidae , in †Etyoidea, and in †Palaeocorystoidea. In other podotremes, i.e., Cyclodorippoidea and Raninoidea ( Fig. 51D ), the male abdomen is still even shorter, interpreted here as a secondary adaptation to burying behaviour.

The brachyuran abdomen was indicated as “flexed in first two pleonic somites” by Dixon et al. (2003: 951, 961). The exact location of abdomen flexion varies among different groups ( Guinot 1979a: fig. 1). It has often been mentioned that the abdomen of primitive crabs was not folded, an assertion that needs a clarification. True, the first somites can be dorsal to various extents, a plesiomorphic condition as in Homolodromiidae and Homolidae , but the remaining somites are flexed and the abdomen is held, as a rule, thanks to a holding system in podotremes as well as in Eubrachyura.

Folding of the abdomen may be incomplete in some Podotremata. The male abdomen retains a strong general curvature in Homolodromiidae ( Fig. 51C ), where it is not firmly held in more basal genera. In Homolidae ( Fig. 51E ) the abdomen maintains a proximal curvature, is long, filling entirely in length the sterno-abdominal depression, and is held by the mxp3 and the pereopods. Abdominal flexion occurs more anteriorly, at the somite 5 level, in Hypoconcha arcuata Stimpson, 1858 , and H. panamensis Verrill, 1869 ( Dromiidae Hypoconchinae). The abdomen is short and not held in Raninoidea ( Lyreididae excepted). The short abdomen of raninoids as in Ranina ranina ( Figs. 38C , 51D ) or Notopoides latus , instead of being folded, is kept in an extended position thanks to a median row of opposed protuberances between somites 2 and 3, 3 and 4, and 4 and 5, each pair acting as an abutment and the complete structure forming an “obstruction system” against the folding of the abdomen. It is assumed that it is a secondary acquisition as part of the major structural modifications linked to a highly specialised burying behaviour ( Van Bakel, Guinot, Artal, Fraaije & Jagt 2012: fig. 49A–C). The more original condition in the Recent fauna is only encountered in the basal family Lyreididae , where the moderately short male abdomen is inserted into the sterno-abdominal depression and fixed by hook-like projections. The †Palaeocorystoidea displays a rather long and fixed abdomen. Elongation of the carapace and the narrowing of the thoracic sternum have had repercussions on the abdomen: shortening, loss of the holding structures, variable rigidity, and even a new function as in Ranina ranina ( Fig. 51D ), where the small, stiff abdomen is rhythmically active during digging (see Burying in the Raninoidea ). In Cyclodorippoidea the abdomen is short, only covering the posteriormost sternites, and held by mechanisms that vary among the families and that are different from those of all other Brachyura ( Tavares 1994b; Guinot & Bouchard 1998), see below.

The first abdominal somites may also remain as a prolongation of the cephalothorax in some Eubrachyura. A striking case is Corystes cassivelaunus , where the first two abdominal somites form a rigid dorsal plate, which is continuous with the carapace, whereas the rest of the short abdomen is flexed and not fixed ( Fig. 3A ). The abdomen of Corystes probably assists the legs in digging into sand ( Garstang 1896; Hartnoll 1968b, 1972a). The eubrachyuran abdomen is flexed at different locations: between the carapace and the first abdominal somite (which corresponds to a complete folding), between the first and second somites (the first remaining dorsal, e.g., in Ocypode ), between the second and third somites, or in the middle of the third somite. † Ocypode italica Garassino, De Angeli, Pasini & Tangocci, 2010 , from the Middle Pliocene of Tuscany, shows a similar abdominal folding ( Garassino et al. 2010: fig. 4C).

Whereas an abdomen occupying the space between the pereopods is a plesiomorphic character, the formation of a special excavation for its insertion is considered apomorphic. Thus, the brachyuran abdomen is either folded back lying between the pereopods in a space referred to as the sterno-abdominal depression (Dynomeniformia, Homoliformia, Lyreididae , †Etyoidea, †Palaecorystoidea), or it is lodged in a variously excavated and well-defined special cavity, the sterno-abdominal cavity (Cyclodorippiformia, †Dakoticancroidea, Eubrachyura) (see Sterno-abdominal depression and sterno-abdominal cavity). The eubrachyuran male abdomen (and that of immature females) is secured in a flexed position and generally tightly held in place against the ventral surface of the thorax by various devices, the most common being the press-button locking mechanism. One of the functions of the abdomen is to protect vulvae, gonopods, penes, pleopods, and eggs (see Elements of Morphology).

Contrary to other decapods, which have similar male and female abdomens, the brachyuran abdomen is usually sexually dimorphic. The male abdomen is narrower than that of the female, which is wider, bears several pairs of pleopods, and is used for the incubation of the eggs. Sexual dimorphism in external morphology is recognisable early in development, for instance at the second stage juvenile crab stage in Erimacrus isenbeckii (see Sasaki & Kuwahara 1999: 32, tables 2, 3). Sexual dimorphism is particularly marked in some Brachyura , as when the abdomen of the ovigerous female is modified into a large cover to form a brood chamber, sometimes closed like a box (e.g., Leucosiidae, Palicoidea ). Sexual abdominal dimorphism is weak in primitive brachyurans, in particular in basal podotremes such as Dynomenidae (see Guinot 2008). Female and male abdomens are not much different from each other in Mictyris .

The raninoid and eubrachyuran female abdomen lacks a pair of pleopods on the first somite, in contrast to the female abdomen of Homolodromioidea , Dromiioidea, Homoloidea , and a few Cyclodorippoidea (e.g., Xeinostomatinae Tavares, 1992 ; see Tavares 1992b: 514, as Xeinostominae ), which bears a vestigial, uniramous, not egg-carrying pleopod 1. The pleopod of the abdominal somite 5 may be vestigial and not oviferous in Phyllotymolinidae , such as in Phyllotymolinum crosnieri and Lonchodactylus messingi Tavares & Lemaitre, 1996 ( Tavares 1993a: 287, fig. 12; Tavares & Lemaitre 1996: 465).

The unfolded abdomen of Decapoda requires the simultaneous extension and flexion of each somite, thanks to particular tegumental joining structures and hinges by which they are articulated with one another. These movements necessitate an accentuated flexure of the anterior portion of the tergites and the differentiation of a special structure, the procurrent lamina, in continuity of the arthrodial membrane. The complete structure makes available a space for the infolding of the articular membrane during extension, so the soft anterior portion of each tergite can slide under the calcified posterior portion of the preceding tergite ( Drach & Jacques 1982). In longbodied forms such as Nephrops ( Fig. 51A ) and Astacus each abdominal somite, as well as the complete abdomen, remains curved ( Huxley 1879; Glaessner 1969). The reduced brachyuran abdomen, which retains the crustacean ring-like segmentation, does not require, however, these movements, and is essentially different by the arrangement of the articulation hinge between the somites. It can be entirely extended and straightened (sometimes except for the first somites) thanks to the membranous, flexible intersomital articulations, most of which are not externally visible. This fundamental difference allows the brachyuran abdomen to remain permanently flexed beneath the cephalothorax and to be closed as a single unit. The ability to flex the abdomen is present in all brachyuran taxa. The primitive Brachyura offer different character states of the abdominal articulation. The Homolodromiidae plesiomorphically retains in its two genera narrow external membranes between the first five somites, their size decreasing posteriorly ( Fig. 51C ). Although the disjointed tergites of the somites in question do not overlap, the somites are convex and exceptionally moveable so the whole homolodromiid abdomen is strongly curved. A similar condition characterises Homolidae , where a rather well developed membrane separates abdominal somites 1–4, in particular the first two ( Fig. 51E ). Moreover, Homolodromiidae shows extended, variously disjointed abdominal pleura (Guinot 1995: 175, figs. 5 et seq.), a characteristic that is also present in some Cyclodorippoidea (see Tavares 1994) but lost in Eubrachyura. In contrast, only a very small membrane may be visible in the dromiid abdomen ( Fig. 51F ), in particular between somites 2 and 3, a character that needs to be verified for the whole family. Ranina ranina ( Figs. 38C , 51D ), which has abdominal pleura laterally extended on each side of the condyle-glene, keeps a particular mobility of each somite thanks to narrow, articular membranes, the entire structure allowing the articulation of two rigid adjacent tergites. The raninid abdomen may, however, remain rigid thanks to the “obstruction system” described above. Presence of abdominal pleura in Raninoidea ( Fig. 51D ) attests the plesiomorphic, podotreme nature of the group. The enigmatic † Eocarcinus ( Fig. 51B ), from the Lias, possesses the articular abdominal membranes and pleura found in Recent Homolodromiidae ( Fig. 51C ), which is, nonetheless, not sufficient to establish with any certainty its assignment to Brachyura (see Monophyletic Brachyura: Earliest Brachyura ).

The articular membranes, in particular externally visible between somites 2 and 3 on a curved, arched abdomen at the location of the flexure of the tergites, tend to be lost in Eubrachyura, where the abdomen becomes flattened and straight. The articular membranes that are situated between all abdominal somites and on the complete breadth of each somite in Dorippidae (e.g., Medorippe lanata , Fig. 51G ) do not exclusively correspond to a particular curvature of the tergites. A review of the abdominal articulation deserves a study in all brachyuran families. In any case, the simple articulation of the abdominal somites in Brachyura , never with an extension and flexion of each tergite on each other and without complex joining system between the somites, is a synapormorphy of Brachyura . This organisation enables the fusion at various degrees of some abdominal somites.

The resting position, the natural posture, of a typical brachyuran male is having its abdomen permanently flexed and kept closely applied to the sternal plastron but unfastened only during mating. When observing living crabs, freshly collected individuals or specimens preserved in alcohol, it is rare, if not exceptional, to find individuals with their abdomen not fixed to the ventral surface. A maintained abdomen is the rule for males and also for pre-pubertal females. Fossil crabs, including those representing moults, may be found in situ with the abdomen locked against the body. The ventral inflexion of the abdomen occurs in other Decapoda such as anomurans, but an abdomen held by specific structures is exclusively a brachyuran adaptation. The holding of the abdomen may be considered a synapomorphy of Brachyura ( Guinot & Bouchard 1998) . The folded abdomen of eubrachyuran males (except for mating) and of pre-pubertal females is fixed by the press-button in such a manner that its maximal length is maintained, the corresponding socket being located on the last abdominal somite (somite 6); only the telson, at the extremity of which the intestine ends by the anal opening, is able to move, allowing defecation. In podotremes where the abdomen is held down at different levels, the maximum extension of the abdomen is similarly secured. Abdomen flapping, at least of the telson, is performed during defecation in some crabs, but it is not necessary because the mobile telson (eventually pleotelson) may be exclusively involved. A rapid abdominal flapping may be observed as an escape reaction, for example in Jonas species , which swim backwards when disturbed (P.K.L.Ng, pers. comm. 2012). Opening of the abdomen in males is essential for copulation ( Yaldwyn 1966), sometimes also after copulation ( Lindberg 1980: 267, 268, 277). It is not well known to what extent the extension of the male abdomen is required during recognition and courtship. Pairs of crabs flap their abdomens at various times before and after copulating, such as in Chionoecetes bairdi (see Donaldson & Adams 1989; see also Paul & Adams 1984), but, in contrast, there is no flapping of the abdomen prior to mating in either the male or female, such as in Hemigrapsus sanguineus (see Anderson & Epifanio 2010).

Abdomen flaps are frequent in females during egg bearing, the abdominal flapping increasing with embryonic development for oxygenation ( Fernández et al. 2002; Baeza & Fernández 2002; Brante et al. 2003; Fernández & Brante 2003; see also Ruiz-Tagle et al. 2002). A female Chionoecetes opilio was observed carried by a male flapping her abdomen to release the eggs ( Hooper 1986). Gecarcinid females release the zoeae by rapid abdominal fanning (Hartnoll 2010). The abdomen remains permanently closed after the puberty moult in mature females, except in ovigerous ones, in some families (e.g., Dorippoidea , Palicoidea , Retroplumoidea ).

Loss of the biramous uropod and its transformation. According to Hessler (1983) the evolution of the eumalacostracan (caridoid) abdomen from that of a phyllocarid through fusion of the seventh original abdominal somite to the sixth occurred in concert with the evolution of the tail fan, the sixth pleopods having evolved to uropods because they were the most posterior appendages (see also Schram 1974; Wilson 1996; discussion in Schram 2009). Uropods represent a pair of larval appendages on abdominal somite 6, and the sixth pair of pleopods combined with the telson to form the tail fan is considered an ancestral decapod structure ( McLaughlin et al. 2004). Uropods combined with a telson are present throughout Decapoda , except for Lithodidae Samouelle, 1819 (uropods lost in males and females), Lomisoidea Bouvier, 1894 (uropods vestigial) ( Balss 1940; McLaughlin 1983a, b; Gruner 1993; McLaughlin et al. 2004; Tudge et al. 2012), and for Brachyura , these groups having essentially a flexed abdomen.

The interpretation of the biramous structure of the appendages, particularly of the biramous uropods, is problematic (see Williams 2004), and the supposed absence of uropods in Brachyura merits discussion. According to Dixon et al. (2003: 954), the Brachyura have “telsons and uropods that are ‘de-specialised’, meaning that they are reduced (or absent) and apparently serve no major function”. It is not exactly correct to say that the telson is no longer functional in crabs. It anteriorly closes the sterno-abdominal depression or cavity, being sometimes involved in abdominal holding, and remains an articulated element when some abdominal somites (very often 3–5) are fused, thus representing the only mobile component of the male abdomen. The telson ventrally bears the anus near its extremity, employing its flexibility to eject the feces, since the abdomen is invariably immobilised (at least) at the level of somite 6 in nearly all Brachyura . The brachyuran telson has generally lost its specialisation for escape and digging in contrast to that of other decapods. In extant raninoids, most of which having an unlocked abdomen, however, a calcification of the distal intestinal region forms a plate, the “telson protection valve” used as a protection when the crab enters the substrate by back-burrowing ( Van Bakel, Guinot, Artal, Fraaije & Jagt 2012).

Compared to all other Decapoda , the Brachyura lacks a tail fan and biramous uropods, but retains uropods as dorsal plates ( Dromiinae , Sphaerodromiinae , †Basinotopinae, Dynomenidae ), ventral lobes (some Dromiinae , Hypoconchinae ), or are modified as sockets (all other Brachyura , with rare exceptions). It is probably inaccurate to diagnose the Brachyura by the absence of uropods, and, furthermore, the reference to “vestigial uropods” is an oversimplification. The structure of the uropods actually represents a wealth of character states, and brachyuran uropods may have a true function. The brachyuran abdominal somite 6 is variously modified in relation to the size and shape of the uropods (e.g., in Dynomenidae ; Guinot 2008: fig. 5). The brachyuran abdominal-holding system involves uropods in the form of dorsal plates, when present (most Dromiidae ), or sockets (Homoliformia, Lyreididae, Eubrachyura ). The modified uropods thus serve as a major function, the holding of the abdomen against the thoracic sternum, which is a strong support for a monophyletic Brachyura . The appendages of the sixth abdominal somite are therefore practically present in all Decapoda , although having different forms and functions: biramous appendages, foliaceous rami, rasps, lobes, dorsal plates, and sockets.

The presence of the uropods as dorsal plates in Odiomarinae , the more basal Hymenosomatoidea ( Fig. 29C– E ), is not fortuitous, but evidence of the persistance of an ancestral structure in an ancient lineage ( Guinot 2011a). This is the only known case in Heterotremata. For the case of the majoid Capartiella , see Fig. 49C, D ; Affinities between Hymenosomatoidea and Inachoididae ; Affinities between Inachoididae and Inachidae ). The only Brachyura without uropods are Cyclodorippidae , Phyllotymolinidae , and Cymonomidae . A socket is absent, but there is an abdominal modification at the somite 6 level (fused with the telson to form a pleotelson in Cyclodorippidae ). In the “sliding system” of Cyclodorippidae (e.g., Clythrocerus nitidus ) the margins of the deep abdominal cavity are anteriorly excavated: when inserted beneath the salient margin of the cavity, the pleotelson (abdominal somite 6 fused with telson) cannot be lifted ( Tavares 1994b: 210, fig. 35A–C; 1998: 116, figs. 7A–C, 8; Guinot & Bouchard 1998: 641; Bouchard 2000: 130, fig. 31A, B, table 10). In the “block system” of Phyllotymolinidae the narrow abdominal somite 6 is positioned between a pair of bulges or crests in sternite 6 so that the enlarged base of the telson is maintained, rending impossible its backward movement; moreover, the abdominal edges fit the sides of the cavity, resulting in a combination of coadaptation by juxtaposition and engagement ( Tavares 1994b: 208; 1998: 116, fig. 1B; Guinot & Bouchard 1998: 641; Bouchard 2000: 132, fig. 31C, table 10). In Cyclodorippiformia, where the male abdomen is placed posteriorly, the involved sternal part is therefore the thoracic sternite 6 at the P3 level, instead of the usual sternite 5. Specialised holding structures are unknown in Cymonomidae , the abdomen of which is secured by a right-angled fold, and in which uropods are absent in the megalopa ( Wear & Batham 1975). The condition of the megalopa has not been documented in other Cyclodorippiformia. The slight abdominal modifications of Cyclodorippiformia cannot be considered homologous to the sockets. Such coadaptation by juxtaposition and engagement is known in Eubrachyura, as in some Leucosiidae Ebaliinae Stimpson, 1871 (e.g., Ilia and Randallia Stimpson, 1857 ; see Guinot 1979a: 145, pl. 15, figs. 5, 6; Guinot & Bouchard 1998: 653, figs. 18, 19A), and there is an absence of any locking structure in Iphiculidae , which displays the plesiomorphic condition (e.g., Iphiculus and Pariphiculus , see Guinot 1979a: 146, 147; Guinot & Bouchard 1998: 653).

The absence of morphological adaptations to hold or lock the abdomen is very rare among the Brachyura . The monophyly of Brachyura is supported by the similarity of position (topographic and position in relation to other parts) and homology of the two different regions that are involved in abdominal holding. The main holding or locking structure always depends on the same thoracic somite, either appendage P2 or sternite 5 (with the exception of a press-button on sternite 4 of Homoliformia). The abdomen is mainly held at the somite 6 level, which bears the uropods or sockets, or the abdomen is modified laterally on its margins. The socket is already present in podotremes ( Lyreididae, Homoliformia , and †Dakoticancroidea) and is practically constant in Eubrachyura, with the consistent nature of the press-button system ( Guinot & Bouchard 1998). We believe that an abdomen that can be fixed by means of a simple, efficient mechanism coupled with a wide thorax and reproductive structures acting interdependently as a single complex is a key to the evolutionary success of brachyuran crabs.

The diagnosis of the Brachyura by A. Milne-Edwards (1861: 49), inspired from H. Milne Edwards (1832, 1834, 1837a, 1852), and summarised by Brocchi (1875: 7, 50) as “ tous les Décapodes dont le pénultième anneau de l’abdomen est dépourvu d’appendices mobiles chez l’animal arrivé à son développement complet ” (all the decapods with the penultimate abdominal somite devoid of any mobile appendage in the completely developed animal), is perfectly valid nowadays. Only the term “mobile” must be deleted in reference to Dynomeniformia (previously Dromiacea ), in some of which the uropod, preserved as a ventral lobe or modified into a dorsal plate, may be mobile.

In Erimacrus isenbeckii (Cheiragonidae) uniramous uropods with natatory setae are still present in the megalopa, shown as small buds in the first crab stage, and disappear in the second stage in both sexes. At this stage, a pair of locking prominences appears on the thoracic sternum in males and females ( Sasaki & Kuwahara 1999: 31, 32, fig. 2H, tables 1, 2: “rocking mechanisms” [sic] mentioned as located and often figured on sternite 6, but located on sternite 5 close to suture 5/ 6 in adults as in their fig. 10C). It is not known if a socket internally differentiates on abdominal somite 6, but the coincidence between the loss of the uropod bud and the appearance of a locking button is clearly linked to the formation of a corresponding socket. The detailed study of the morphological changes of larval stages and juveniles by Sasaki & Kuwahara (1999) thus provides the ontogenetic criterion for the transformation of the uropod into a socket and the homology between these two structures. In Pyromaia tuberculata (Inachoididae) , where the megalopa has a uropod that consists of an expodite distally bearing two natatory setae ( Luppi & Spivak 2003: 208, fig. 4N), the abdominal socket, not yet visisble in the first crab stage, is present in the second, third and fourth crab stages and absent from fifth to ninth crab stages, concurrently with the presence then absence of the buttons of the locking mechanism at the same stages ( Flores et al. 2002: 316, fig. 3, table 1). A morphological and histological ontogenetic study of abdominal somite 6 should provide useful information on the socket.

Male reproductive system. The external reproductive structures of brachyuran crabs exhibit a higher degree of complexity than those of the other Decapoda . The brachyuran male reproductive system consists of three organs that belong to two different, originally independent, and well differentiated parts of the body: a penis (distal portion of the ejaculatory duct) belonging to the cephalothorax, emerging from the appendage of thoracic somite 8 (P5 coxa) or from sternite 8, and a G1 and G2 belonging to the abdomen (appendages of abdominal somites 1 and 2). These three organs have co-evolved to morphologically and functionally work together as a copulatory complex ( Fig. 9 ). Notwithstanding variations in the width of the thorax and the length of the penis and G2, the interaction is successful. We agree with Dixon et al. (2003) that the arrangement of two gonopods, a G2 acting together with a G1, is a synapomorphy of Brachyura since the two gonopods do not interact in this way in other decapods.

Other sperm-transfer structures have involved in the remaining Decapoda , but not exactly the same as in Brachyura , where the penis functions with the pairs of gonopods, and sperm is deposited near or introduced in a pair of internalised female structures (spermathecae or vulvae). In Astacidea Latreille, 1802, Nephrops norvegicus (Nephropidae) and Astacus astacus ( Linnaeus, 1758) ( Astacidae Latreille, 1802 ), the G1 and G2 are combined together but, unlike Brachyura , they function as a single structure ( Guinot 1979a: 224, figs. 57–59A, B; Hobbs et al. 2007). In A. astacus ( Linnaeus, 1758) and A. leptodactylus Eschscholtz, 1823 , there is only a short, soft papilla (Erkan et. al. 2009a); in other species of Astacus , the penis may be protected by a hard, spooned-shaped cup; in Astacoides Guérin, 1839 , short “phallic papillae” are associated to the male gonopores ( Hobbs 1987: 9, figs. 6–8, 10). In these astacids the sperm is transferred to the sternal area or inside an unpaired spermatheca, instead of the internalised paired spermatheca of the Podotremata or the paired vulva of Eubrachyura. In Panulirus japonicus von Siebold, 1824 , the gonopore opens at the bottom of a deep fold between a genital accessory apparatus and the P5 coxo-sternal condyle, thus copulation seems to occur without the insertion of the male apparatus into the female gonopore, and fertilisation is external ( Nakamura 1993: 1489, figs. 1–12). The spermatophore is similarly deposited externally in many other palinurids ( Phillips et al. 1980). An internal insemination was hypothesised in the lobster Jasus lalandii (H. Milne Edwards, 1837) ( Palinuridae Latreille, 1802 ) thanks to tongue-like flaps normally closing the male gonopores but capable of erection for the introduction of internal spermatophore ( Fielder 1964: 163, fig. 2; Aiken & Waddy 1980). The absence of sperm in the oviduct and of a spermathecal enlargement observed in the same species by Paterson (1968) contradicts the previous postulations. An external fertilisation has been recently confirmed in Homarus americanus (H. Milne Edwards, 1837) (Nephropidae) by Aiken et al. (2004). The sexually modified first two pairs of “swimmerets” in male galatheids such as the squat lobster Galathea strigosa (Linnaeus, 1767) , which act together for sperm transfer, are differently shaped ( Heitier 1983).

The monophyly of Brachyura is thus further supported by the synapomorphy: combined action of G1, G2, and penis to inseminate a paired female receptacle in both Podotremata and Eubrachyura. Despite the speculative external fertilisation of podotreme crabs, the internal fertilisation of Eubrachyura corresponds to a highly modified system ( Bauer 1986).

The two gonopods and the penis are conservative traits and thus of significant importance in classification. They are less affected by environmental changes than by other external morphological characters because they function only at the moment of sperm transfer ( Türkay 1975a). Changes in the morphology of this complex supposedly co-evolved with the female reproductive structures. As the gonopods must reach and inseminate the vulvae, a coadaptation is expected at least between the male and female structures. Nevertheless, a complex lockand-key arrangement has yet to be demonstrated (J. Crane 1975: 465; Guinot 1979a: 249), there is at most a simple adjustment between the gonopod and the vulva. This adjustment, including increased sclerotisation and thickness of the G1, its orientation and torsion as well as the modification of its tip to facilitate insertion into the vulvar slit and to maintain the gonopod and the vulva in place, has been shown in Corystes (Corystidae) ( Hartnoll 1968b), Sesarmidae and Plagusiidae ( Hartnoll 1964, 1965), Ucinae ( Crane 1975), and in the varunid Helice / Chasmagnathus complex ( Sakai et al. 2006). In crabs with a vulvar sternal cover, or an operculum that needs to be temporarily pushed inwards as in Corystes and many thoracotremes, the G1 apex invariabily (save for possible exceptions) bears a horny, acute endpiece (e.g. Davie & Pabriks 2010; see Female sternal gonopores, or vulvae; Gonopods).

Earliest brachyurans. There is no clear understanding of the origin of the Decapoda , the stem forms being absent in the fossil record. Jurassic (Bathonian) formations in northwestern Scotland, which show burrows ending in funnel-like apertures, U- and L-shaped forms and possible basal dwelling chambers, are ichnofossils that potentially provide the earliest known record of activity by crabs (or shrimps) (P. Marshall 2003; Bracken et al. 2009).

The identities of the earliest brachyurans remain doubtful, leading to a number of hypotheses (e.g., Bouvier 1896; Van Straelen 1928a, b; Feldmann & Schweitzer 2010; Schweitzer & Feldmann 2010a). The Carboniferous † Imocaris tuberculata Schram & Mapes, 1984 , first suspected to be a podotreme brachyuran, a dromiid ( Schram & Mapes 1984; Schram 2009: 5, fig. 3), and † I. colombiensis Racheboeuf & Villaroel, 2003 , have been conversely considered putative peracarid pygocephalomorphs ( Rinehart et al. 2003; Racheboeuf & Villaroel 2003; Guinot et al. 2007b; discussion in Schram 2009; see also Hotton et al. 2002). A “dromiacean” assignment for †Imocaridae Števčić, 2005 (as †Imocarididae [sic]), is doubtful, the carapace lacking podotreme facies. Consequently, the placement by Porter et al. (2005: 366, fig. 2) of † I. tuberculata in Brachyura as the second oldest fossil reptant (after † Palaeopalaemon newberryi Whitfield, 1880 , from the late Devonian) has resulted into a basal position of the brachyuran lineage in their decapod phylogeny, contradictory to accepted hypotheses. A Triassic species from New Mexico, † Rioarribia schrami Rinehart, Lucas & Heckert, 2003 ( Rinehart et al. 2003), first supposed to be a eubrachyuran, was presumed not to be a decapod by Schweitzer & Feldmann (2005). These fossils, in any case, tell us little about the origin of Brachyura .

We agree with the conclusion of Porter et al. (2005) that many of the modern brachyuran families have had an origin earlier than the Eocene, many brachyuran radiation events occurring in the Cretaceous. Recent evidence from the fossil record indicates an explosive evolution in the Jurassic, in which more than ten podotreme families were already represented. Preliminary results of a multilocus molecular phylogeny of the Brachyura by Ahyong et al. (2010) suggested a Carboniferous origin, with a Permian origin for the heterotremes, including the monophyletic Old World freshwater crabs, an assertion that may rise reservations and is subject to confirmation.

The earliest known member of Brachyura is confined to the Early Jurassic and is a podotreme, as previously hypothesised ( Guinot & Tavares 2001). † Eoprosopon Förster, 1986 , with † E. klugi Förster, 1986 , referred to † Prosopidae ( Förster 1986; Wehner 1988; Müller et al. 2000), then to Homolodromiidae (see Guinot 1995), and considered Homolodromioidea incertae sedis by Schweitzer & Feldmann (2010a), goes back to the Pliensbachian and therefore is the earliest known brachyuran. “Prosopids” and allied crabs, which are usually not found within concretions and have been preserved as detached carapaces or limbs, often as casts only, thus represent the most primitive brachyurans, the numerous cryptic habitats within sponge-microbial mounds and coral reefs favouring their high diversification in the Late Jurassic of the Tethys ( Collins & Wierzbowski 1985; Wehner 1988; Müller et al. 2000; Franṭescu 2011; Schweigert & Koppka 2011).

Another one early brachyuran appears to be † Homolus auduini [sic] ( Eudes-Deslongchamps, 1835) (Eudes- Deslongchamps 1835: 39, pl. 1, figs. 4–6), from the Bathonian. It belongs in the living family Homodromiidae , and, as such, † Homolus Eudes-Deslongchamps, 1835 , would be one of the oldest known homolodromiid genera ( Hee 1924: 148, pl. 5, fig. 4, as † Pithonoton auduini ; Beurlen & Glaessner 1931: fig. 15, as † Protocarcinus auduini ; Förster 1986: 26, as † Prosopon auduini ). † Palaeinachus longipes ( Woodward & Salter, 1865) ( Woodward & Salter 1865: pl. 2, fig. 15; Woodward 1866: 493, pl. 24, fig. 1) is a synonym (A. Milne-Edwards & Bouvier 1902: 10, fig. 4; Van Straelen 1923b: 553, as † Homolus auduini ; 1925: 339, fig. 154, pl. 10, fig. 9, as † Avihomola auduini ; Glaessner 1929: 348, as † Protocarcinus auduini ; Wehner 1988: 30, as † Foersteria auduini ; Guinot 1995: 265, as † Foersteria auduini ; Guinot et al. 2007b: 243, as † Homolus audini [sic]). The taxon † Homolus auduini has priority over † Gabriella audini [sic] recently used by Collins et al. (2006: 125), † Gabriella Collins, Ross, Genzano & Mianzan, 2006 , being a replacement name for the preoccupied nomen Foersteria Wehner, 1988 . The original spelling auduini would be an incorrect original spelling to be corrected to audouini as the species was explicitly dedicated to Audouin, as “ Homole d’Audouin ” (see Eudes-Deslongchamps 1835: 39) (see Anonymous 1999, Code, Art. 32.5.1; Dubois et al. 2011: 60).

Close attention should be paid to the early Jurassic (Lower Pliensbachian) † Eocarcinus , onomatophore (type genus) of † Eocarcinidae Withers, 1932 , and †Eocarcinoidea Withers, 1932. It was often considered a basal podotreme ( Withers 1932, 1951; Glaessner 1933; Beurlen 1933; Števčić 1977; Förster 1979b, 1985a, b, 1986; Förster, Gaździcki & Wrona 1985; Wehner 1988; Guinot 1991, 1995; Vía & Sequeiros 1993; Vermeij & Lindberg 2000). On the other hand, † Eocarcinus has been regarded as a model for a link between the macruran and the brachyuran Decapoda ( Förster 1979b; Wägele 1989), or transitional between Glypheoidea Zittel, 1885, and the early brachyuran prosopids ( Müller et al. 2000; Krobicki & Zaton 2008), also proposed as predecessor of † Gastrodoridae or †Gastrodoroidea within the Anomura ( Van Bakel et al. 2008; Klompmaker et al. 2011a). The † Eocarcinidae was assigned to section “ Dromiacea ” by De Grave et al. (2009: 27) and Schweitzer et al. (2010: 57), but finally excluded from Brachyura by Feldmann & Schweitzer (2010). The “crab form” recognised in the Sinemurian (early Lias) by Vermeij (2008: table 1) may probably be referred to this family. In known specimens of † E. praecursor Withers, 1932 , the long, developed abdomen is always located on the prolongation of the cephalothorax and shows a marked curve of only the last somites. † Eocarcinus exhibits articulating membranes in the proximal portion ( Fig. 51B ), more similar to the configuration of the Homolodromiidae ( Fig. 51C ) than to that of a decapod such as Nephrops ( Fig. 51A ). The narrowness of the first abdominal somite of † Eocarcinus indicates a weak muscular mass, a necessary condition for a possible flexure. The reconstructions of † E. praecursor by Withers (1932: fig. 1) and by Förster (1979a: fig. 4; 1985a: fig. 5a) shows a too wide proximal abdomen compared to the published photographs ( Withers 1932: pl. 9, pl. 10, fig. 1; Förster 1979a: figs. 1, 3; Guinot & Tavares 2001: fig. 21). The fossilisation of † Eocarcinus with an unfolded abdomen could be due to the absence of holding structures ( Bouchard 2000: 64). The most important character to eliminate † Eocarcinus from Brachyura , the condition of uropods and tail fan, remains still questionable, the poor preservation of this region providing conflicting interpretations by Withers (1932), Förster (1979a) or uncertainty ( Guinot & Tavares 2001; Feldmann & Schweitzer 2010; Schweitzer & Feldmann 2010a). The absence of visible dorsal uropods must be explained by their shape, putatively as ventral lobes as in Homolodromiidae , as sockets as in Homoliformia, or complete loss as in Cyclodorippiformia and most Raninoidea . It is not known if the first Homolodromiidae had developed uropods as dorsal plates or under another form, the condition in their putative ascendants remaining undetermined so far by lack of preservation of the ventral structures. The wide abdominal somites of † Eocarcinus , with conspicuous pleura ( Fig. 51B ), reminds the homolodromiid abdomen, also with pronounced pleura ( Fig. 51C ). The term “epimere” was erroneously used by Feldmann & Schweitzer (2010) and Karasawa, Schweitzer & Feldmann (2011) to designate these abdominal pleura. The epimere is actually a membranous, unsegmented tegument that lines the carapace and is replaced, along the lateral wall of the body, by the semi-calcified pleurites (see Axial skeleton; Secretan 1977; Casanova et al. 1988). The homolodromiid male abdomen is entirely flexed and held, and ends in very long telson (Guinot 1995), contrasting with the extended abdomen ending in a short telson in † Eocarcinus . Withers (1951: 184) assumed that the P4 and P5 were reduced and carried on the back, whereas a camouflage behaviour could not be deduced from their position according to Förster (1979a: 21). The cheliped (P1) of † E. praecursor has long and wide fingers, with numerous teeth on the prehensile border of both fingers, stronger on the fixed finger. The merus/carpus articulation of the chelipeds of † Eocarcinus , with a nearly horizontal axis of rotation allowing their complete extension and forward movements (Feldmann & Schweitzer 2010: 248), resembles the condition of Homolodromiidae , in contrast with the pereopods of most Brachyura which show a horizontally inclination, performing a lateral locomotion. The orientation of the chelipeds of † Eocarcinus is indeed not dissimilar to that of the Homolodromiidae , the generalised decapod disposition. This configuration applies not only to the chelipeds but also to the P2 and P3, and is present in a small number of diverse basal Brachyura that walk forwards (see Locomotion). The absence of true orbits also characterises the Homolodromiidae . A number of characters makes obvious that the closest relative of † Eocarcinidae is the Homolodromiidae . The † Eocarcinidae , Homolodromiidae , and † Prosopidae were included in a “new” superfamily “ Prosopoidea superfam. nov. ” by Števčić (1977: 67) without taking into account that, according to the Code, a superfamily must keep the same authorship as the family, i.e., Prosopoidea von Meyer, 1860. The re-examination of the type material of † E. praecursor by Feldmann & Schweitzer (2010: 247, figs. 1A, 2B) provides, however, a character, a chelate P2, which would be inconsistent with a placement within Brachyura . If this is the case, † Eocarcinus cannot be referred to Brachyura . Feldmann & Schweitzer (2010) concluded to the inclusion of the †Eocarcinoidea in the Anomura (see also Van Bakel et al. 2008; Klompmaker et al. 2011a).

If † Eocarcinus proves to be a brachyuran (loss of the tail fan, not chelate P2), the most plausible relationship could be with Homolodromioidea . An eventual, although improbable, proximity with Homoloidea , in particular with Poupiniidae , has been discussed by Guinot & Tavares (2001). This hypothesis is rejected, although Homoliformia is a very ancient lineage whose extant descendants retain an ancestral axial skeleton with connections by interdigitations (like Homolodromioidea ), a condition more plesiomorphic than that of Dromioidea ( Secretan 1983: fig. 1; 1998: fig. 16; Guinot & Quenette 2005: figs. 18, 22).

It is crucial to solve the question of † Eocarcinus , because this fossil is used as a key calibration point in brachyuran phylogeny. Predicting the timing of evolutionary events from fossil, morphological, and molecular data is a challenging estimation problem ( Cranston & Rannala 2005) and needs to be based on accurate data. Teske et al. (2009: 25, fig. 4) considered † Eocarcinus as the oldest known true crab, about 190 million years old, whereas † Imocaris tuberculata was the oldest brachyuran according to the brachyuran tree of Porter et al. (2005). The Middle Eocene † Notocarcinus sulcatus Schweitzer & Feldmann, 2000 , considered the oldest fossil representative of Cancridae (Feldmann & Schweitzer 2006) and thus selected as calibration point in a molecular analysis of Hymenosomatidae by Teske et al. (2009), is a genus and species known only by a dorsal carapace.

The molecular analysis combining nuclear and mitochondrial sequences given by Bracken et al. (2009: 112, figs. 2, 3, table 1) supported the Brachyura as a monophyletic clade with marginal support; however, values significantly increased upon the removal of the homolid Paromola japonica (see Monophyletic Podotremata). The strong support for a clade uniting the Potamidae with Grapsidae and Ocypodidae in this analysis is surprising and currently diffcult to explain.

The molecular decapod phylogeny of Porter et al. (2005: 366, fig. 2), in which Anomura and Brachyura are basal reptant lineages, is a controversial hypothesis in contradiction to all previous analyses, and recognised as such by these authors themselves. We agree that such a hypothesis would force “a re-interpretation of many of the morphological characters currently used to define reptant lineages” ( Porter et al. 2005: 366). The recent additions to the already rich fossil brachyuran record should allow re-evaluation of the current polarisation of the chosen characters to evaluate decapod phylogeny. The difference in the polarisation of the characters could simply result from different methodologies, from the poor phylogenetic value of the morphological traits used in the analyses, and from inaccurate interpretation of the earliest fossils (see above). More studies are needed to explore the more recent proposals, in particular those of Tsang et al. (2008: fig. 2) and Chu et al. (2009b: fig. 1), diametrically opposed to Porter et al. (2005) ( Lemaitre & McLaughlin 2009: 127, see figs. 1–13).

Figures

FIGURE 1. A, Anatomical relationships of penis and gonopods in Brachyura exemplified by Chionoecetes opilio (Fabricius, 1788) (Oregoniidae) (modified from Beninger et al. 1991: fig. 1C; G1 basal setae not figured); B–D, schematic representation of the dispositions of male gonopore, penis, and ejaculatory duct (indicated by dotted line) in Brachyura: B, C, perforation of P5 coxa by the ejaculatory duct: B, coxal gonopore, with penis emerging from P5 coxa (Podotremata and Heterotremata pro parte); C, coxo-sternal condition, with coxal gonopore and penis emerging at level of thoracic sternal suture 7/8 (Heterotremata pro parte); D, E, perforation of thoracic sternum by ejaculatory duct: D, sternal gonopore at the suture 7/8 level; E, sternal gonopore far from suture 7/8, in posteriormost location in relation to sternite 8. a1, abdominal somite 1; cx5, P5 coxa; g, gonopore; G1, first gonopod; G2, second gonopod; p, penis; p5, fifth pereopod; 7, 8, thoracic sternites 7, 8; 7/8, thoracic sternal suture 7/8. Scale bar: 1 mm (A).

FIGURE 2. Spermathecae in Dromiinae. A, Dromia bollorei Forest, 1974, ovig. female 43.1 × 47.5 mm, holotype, Ivory Coast (MNHN-B21993); B, Dromia personata (Linnaeus, 1758), female 31.0 × 36.0 mm, France, St. Vaast (MNHN-B21964); C, Lauridromia dehaani (Rathbun, 1923), ovig. female 59.0 × 70.0 mm (MNHN-B6871); D, Stimdromia lateralis (Gray, 1831), female 14.3 × 15.7 mm, New South Wales (MNHN-B11206). g, female gonopore; s, aperture of spermatheca; s.d., sterno-coxal depression; 4–8, thoracic sternites 4–8; 7/8, thoracic sternal suture 7/8. Scale bars: 5 mm (C); 2.5 mm (D).

FIGURE 3. Exposed vulvae. A, B, Corystes cassivelaunus (Pennant, 1777) (Corystidae). A, male 35.2 × 24.6 mm, France, Bay of Biscay (MNHN-B5381); B, female 31.0 × 24.0 mm, France (MNHN-B5385); C, Telmessus cheiragonus (Tilesius, 1812) (Cheiragonidae), female 35.0 × 46.0 mm, Japan (MNHN-B5445): notice notch on abdominal somite 6; D, Pseudocorystes sicarius (Poeppig, 1836), female 50.0 × 44.0 mm, Chile (MNHN-B5407); E, F, Bellia picta H. Milne Edwards, 1848 (Belliidae), pre-adult female 29.0 × 25.5 mm [Tahiti, erroneous origin], dry (MNHN-B2). a6, abdominal somite 6; b.p., pressbutton; v, vulva; 4–6, thoracic sternites 4–6.

FIGURE 4. Vulvae in Goneplacidae (notice the complete thoracic sternal suture 6/7, incomplete suture 7/8, and median line line along sternites 7, 8). A, Carcinoplax longimanus (De Haan, 1833), female 29.1 × 37.8 mm, Taiwan (MNHN-B29648); B, Pycnoplax victoriensis (Rathbun, 1923), female 12.1 × 14.6 mm, Australia, Victoria (NMV J52213), remains of sperm plug; C, Thyraplax crosnieri (Guinot & Richer de Forges, 1981), female 16.7 × 21.9 mm, Philippines, Panglao I. (ZRC 2004.0720) (after Castro 2007: figs. 1, 18, 19 respectively). s.p., sperm plug; v, vulva; v.c., sternal vulvar cover; 4–8, thoracic sternites 4–8.

FIGURE 5. Vulvae in Hexapodidae (A–C) and Retroplumidae (D). A,?Hexapus sexpes (Fabricius, 1798), 10.6 × 18.0 mm, Persian Gulf (MNHN-B10209). B, Hexapodidae sp., female 3.0 × 4.7 mm, Gabon (MNHN-B27907). C, Thaumastoplax anomalipes Miers, 1881, female 4.0 × 6.0 mm, Congo (R.O.C.) (MNHN-B27913). D, Bathypluma forficula Saint Laurent, 1989, female 12.0 × 14.0 mm, Indonesia, Makassar Strait (MNHN-B11229). p.b., press-button; v.c., sternal vulvar cover; 4–7, thoracic sternites 4–7; 5/6, thoracic sternal sutur 5/6. Scales bars: 5 mm (A, D), 2 mm (B, C).

FIGURE 6. Complex vulvae. A, Cavoportunus dubius (Laurie, 1906) (Portunidae Portuninae), female 18.5 × 14.3 mm, Philippines, Bohol, Panglao I. (ZRC 2001.0627) (Photographs by S.T. Nguyen & P.K.L. Ng); B–D, Episesarma lafondii (H. Jacquinot in Hombron & H. Jacquinot 1846) (Sesarmidae), female 26.9 × 30.7 mm, Western New Guinea, Kamora River (ZRC2000.2066). B, vulvae; C, D, detailed views showing sternal vulvar cover with raised posterior wall, shorter inner wall, and folded central operculum. Photographs by P. Davie, Queensland Museum. o, operculum; r, rim; v.c., sternal vulvar cover.

FIGURE 7. Sperm plugs in Dromiidae. A, Austrodromidia octodentata (Haswell, 1882), female 44.0 × 47.0 mm, South Australia (AM P2311); B, Pseudodromia latens Stimpson, 1858, female 29.0 × 21.0 mm, South Africa (SAM A1725). g, gonopore; s, aperture of spermatheca; s.p., sperm plug; 6–8, thoracic sternites 6–8; 7/8, thoracic sternal suture 7/8. Scale bars: 5 mm (A); 2.5 mm (B).

FIGURE 9. Penes and two pairs of gonopods acting together as a copulatory complex in Brachyura. Callinectes sapidus Rathbun, 1896 (Portunidae, Portuninae). Copyright ©2003 Steven C. Zinski.

FIGURE 12. Penis in Homoliformia, located in demarcated excavation, covered by row of setae. Homola ranunculus Guinot & Richer de Forges, 1995 (Homolidae), male 26.4 × 21.6 mm, New Caledonia (MNHN-B19871). cx4, cx5, P4, P5 coxae; g, gonopore; m.l., median line; p, penis; 7, 8, thoracic sternites 7, 8; 7/8, thoracic sternal suture 7/8. Scale bar: 5 mm.

FIGURE 14. Derived coxo-sternal condition showing penis sheltered in gutter and location of “supplementary” suture 7/8 at complete dorsal junction of sternites 7 and 8 (both separated here to expose penis), “true” suture 7/8 situated below. Prionoplax ciliata Smith, 1870 (Panopeidae Eucratopsinae), male 13.4 × 20.3 mm, Panamá (MNHN-B24490). cx4, cx5, P4, P5 coxae; g, gutter; p, penis; pa, papilla; 7, 8, thoracic sternites 7, 8; 6/7, 7/8, thoracic sternal sutures 6/7, 7/8. Scale bar: 2 mm.

FIGURE 15. Coxal condition in Dorippidae. A–D, Medorippe lanata (Linnaeus, 1767), male 26.0 × 32.0 mm, Senegal (MNHN-B21493): A, genital region; B, penis, G1, G2 in situ; C, G2; D, G1 apex, setae not shown. E, Phyllodorippe armata (Miers, 1881), male 20.0 × 25.0 mm, Dahomey (=Benin) (MNHN-B16380): genital region. a1, abdominal somite 1; b, sclerotised bulb; ba, basipodite; c, coxo-sternal condyle; ca, carapace; co, coxopodite; cx4, cx5, P4, P5 coxae; en, endopodite; e7, episternite 7; G1, G2, gonopods 1, 2; pa, papilla; t, sclerotised tube; 7, 8, thoracic sternites 7, 8; 7/8, thoracic sternal suture 7/8. Outline of the abdomen indicated by dotted line. Scale bars: 5 mm (A, E); 2 mm (B, C); 1 mm (D).

FIGURE 16. Coxo-sternal condition in Dorippidae. Dorippe quadridens (Fabricius, 1793), male 33.0 × 33.5 mm, Vietnam (MNHN-B11177): A, genital region; B, genital region with penis exposed (distal half of penis not shown in its normal vertical position); C, G1, penis, G2 in situ; D, G1 apex; E, G2. a1, abdominal somite 1; b, sclerotised bulb; c, coxo-sternal condyle; ca, carapace; cx4, cx5, P4, P5 coxae; G1, G2, gonopods 1, 2; en, endopodite; pa, papilla; t, sclerotised tube; td, distal portion of penial tube; tp, inclined proximal portion of penial tube; 7, 8, thoracic sternites 7, 8; 6/7, 7/8, sternal sutures 6/7, 7/8. Scale bars: 5 mm (A–C); 1 mm (D); 2 mm (E).

FIGURE 17. Coxo-sternal condition in Dorippidae. A, B, Dorippe tenuipes Chen, 1980. A, male 21.0 × 25.0 mm, South China Sea (MNHN-B8937). B, male 19.0 × 20.0 mm, holotype of Dorippe miersi Serène, 1982, Vietnam (MNHN-B7279): A, genital region covered by abdomen; B, genital region uncovered by abdomen, penis covered by sternite 8. C, Paradorippe granulata (De Haan, 1841), male 11.0 × 11.0 mm, Japan (MNHN-B11181): genital region. D, Philippidorippe philippinensis Chen, 1985, male 25.0 × 30.0 mm, Philippines (MNHN-B18913): genital region covered by abdomen. a1–a3, abdominal somites 1–3; b, sclerotised bulb; c, coxo-sternal condyle; ca, carapace; cx4, cx5, P4, P5 coxae; e7, episternite 7; pa, papilla; td, distal portion of penial tube; tp, inclined proximal portion of penial tube; 6–8, thoracic sternites 6–8; 6/7, 7/8, sternal sutures 6/7, 7/8. Scale bars: 5 mm (A, D); 3 mm (B, C).

FIGURE 18. Coxo-sternal condition in Dorippidae. Dorippoides facchino (Herbst, 1785), male 25.0 × 31.0 mm, India, Pondichery (MNHN-B11178): A, genital region; B, genital region with penis exposed; C, penis, G1, G2 in situ; D, G1 apex. a1, a2, abdominal somites 1, 2; b, sclerotised bulb; ba, basipodite; c, coxo-sternal condyle; ca, carapace; co, coxopodite; cx4, cx5, P4, P5 coxae; en, endopodite; e7, episternite 7; G1, G2, gonopods 1, 2; pa, papilla; td, distal portion of penial tube; tp, inclined proximal portion of penial tube; 7, 8, thoracic sternites 7, 8; 6/7, 7/8, sternal sutures 6/7, 7/8. Outline of the abdomen indicated by dotted line. Scale bars: 5 mm (A); 2 mm (B); 3 mm (C); 1 mm (D).

FIGURE 19. Coxo-sternal condition in Dorippidae. A, B. Dorippoides nudipes Manning & Holthuis, 1986, male 18.0 × 20.0 mm, Madagascar (MNHN-B18276): A, genital region covered by abdomen; B, genital region with penis exposed. C, Neodorippe callida (Fabricius, 1798), male 12.0 × 12.0 mm, China, Amoy (MNHN-11170): genital region. a1, a2, abdominal somites 1, 2; b, sclerotised bulb; c, coxo-sternal condyle; ca, carapace; cx4, cx5, P4, P5 coxae; e7, episternite 7; pa, papilla; t, sclerotised tube; td, distal portion of penial tube; tp, inclined proximal portion of penial tube; 7, 8, thoracic sternites 7, 8; 6/7, 7/ 8, sternal sutures 6/7, 7/8. Scale bars: 5 mm (A, B); 2 mm (C).

FIGURE 20. Coxo-sternal condition in Ethusidae. Ethusa mascarone (Herbst, 1785), male 11.0 × 9.8 mm, Senegal (MNHN- B20933). A, genital region; B, P5 coxa, penis, G1, G2 in situ; C, G1 apex; D, G2. a1, abdominal somite 1; ba, basipodite; c, coxo-sternal condyle; ca, carapace; co, coxopodite; cx4, cx5, P4, P5 coxae; en, endopodite; G1, G2, gonopods 1, 2; p, penis; pa, papilla; 6–8, thoracic sternites 6–8; 6/7, 7/8, thoracic sternal sutures 6/7, 7/8. Outline of the abdomen indicated by dotted line. Scale bars: 2 mm (A, B, D); 1 mm (C).

FIGURE 21. Coxo-sternal condition in Ethusidae. A, Ethusina talismani A. Milne-Edwards & Bouvier, 1897, male 7.0 × 6.0 mm, lectotype, Azores (MNHN-B24049): genital region (proximal portion of penis not shown). B, Ethusa rosacea A. Milne- Edwards & Bouvier, 1897, male 12.0 × 11.0 mm, syntype, off Canary Is. (MNHN-B13545): genital region. C, Ethusa dilatidens Chen, 1997, male 18.0 × 18.0 mm, Philippines (MNHN-B18930): penis with papilla protruding from funnel-like structure. a1, abdominal somite 1; c, coxo-sternal condyle; cx4, cx5, P4, P5 coxae; f, funnel; pa, papilla; t, sclerotised tube; 6– 8, thoracic sternites 6–8; 6/7, 7/8, thoracic sternal sutures 6/7, 7/8. Scale bars: 1 mm (A); 3 mm (B); 1 mm (C). 6–8, thoracic sternites 6–8

FIGURE 22. Coxo-sternal condition in Ethusidae. Ethusina abyssicola Smith, 1884, male 12.6 × 11.0 mm, Massachusetts (MNHN-B24048): A, genital region; B, genital region with penis exposed by dissection; C, P5 coxa, penis, G1, and G2 in situ; D, G1 apex. a1, a2, abdominal somites 1, 2; c, coxo-sternal condyle; ca, carapace; cx4, cx5, P4, P5 coxae; en, endopodite; G1, G2, gonopods 1, 2; p, penis; pa, papilla; 7, 8, thoracic sternites 7, 8; 6/7, 7/8, thoracic sternal sutures 6/7, 7/8. Outline of

FIGURE 23. Sternal male gonopores in Thoracotremata. A, Cardisoma carnifex (Herbst, 1796) (Gecarcinidae), male, Madagascar (MNHN-B29589); B, Plagusia depressa (Fabricius, 1775) (Plagusiidae), male 40.0 × 44.0 mm, Senegal (MNHN- B11693): schematic view (see Fig. 33A) (abdomen indicated by dotted line); C, Ucides cordatus (Linnaeus, 1763) (Ucididae), male 70.0 × 93.0 mm, Brazil (MNHN-B13237) (abdomen indicated by dotted line); D, Grapsus grapsus (Linnaeus, 1758) (Grapsidae), male 57.0 × 60.0 mm, Clipperton I. (MNHN-B11538): schematic representation after dissection; E, F, Ocypode cursor (Linnaeus, 1758) (Ocypodidae), male 38.0 × 49.0 m, Senegal (MNHN): E, before dissection along dotted line; F, after dissection showing ejaculatory duct perforating somite 8, instead of P5 coxa. c, P5 coxo-sternal condyle; cx5, P5 coxa; e.d., ejaculatory duct; e7, episternite 7; g, gonopore; p, penis; p5, pereopod 5; 7, 8, sternites 7, 8; 7/8, thoracic sternal suture 7/8.

FIGURE 24. Coxo-sternal condition in Chasmocarcinidae. A, B, Trogloplax joliveti Guinot, 1986 (Trogloplacinae), male 20.0 × 27.0 mm, paratype, Papua New Guinea, New Britain (MNHN-B12794). A, thoracic sternum (abdomen removed); B, details of sternites 7 and 8, portion of penis (dotted line) enclosed in the “double plate”. C, Chasmocarcinus typicus Rathbun, 1898 (Chasmocarcininae), male 9.0 × 12.0 mm, Brazil (MNHN-B29900): penis (proximal portion dorsally visible) enclosed in invagination of sternite 8 after perforating P5 coxa, and emerging as papilla (left); P5 coxa removed to expose ejaculatory duct (right). c, coxo-sternal condyle; cx4, cx5, P4, P5 coxae; e.d., ejaculatory duct; e.p., exposed part of penis; g.i., gutter for carapace-thoracic sternum interlocking mechanism; l, invagination line of sternite 8; m.l., median line; p, penis; pa, papilla; p.p., proximal portion of penis; s.c., sterno-abdominal cavity; 4–8, sternites 4–8; 4/5–7/8, thoracic sternal sutures 4/5–7/8. Scale bar: 5 mm (C).

FIGURE 26. Thoracotreme condition in Heloecius cordiformis (H. Milne Edwards, 1837) (Heloeciidae). A, male 15.0 × 23.0 mm, Australia, New South Wales (MNHN-B21489): male gonopore sternal, with penis emerging close to P5 coxa; B, dissected male (same data, approximately same size): P5 (left) with penis protruding from sternite 8. c, P5 coxo-sternal condyle; cx5, P5 coxa; c.p., slightly sclerotised proximal portion of penis; pa, papilla; 8, sternite 8.

FIGURE 29. Male gonopore (A, B) and dorsal platelets of the abdomen (C–E) in Hymenosomatidae. A, Cancrocaeca xenomorpha Ng, 1991, male 3.9 × 4.6 mm, paratype, Indonesia, Sulawesi (MNHN-B24450); B, Odiomaris pilosus (A. Milne- Edwards, 1873) (Odiomarinae), male 20.0 × 22.6 mm, New Caledonia (MNHN-B26146); C–E, same data as B: abdomen in dorsal and ventral views; notice on pleotelson (somite 6 fused to telson) the vestigial uropod as intercalated, dorsal platelet, with socket situated beneath, and uropod (D) with electron microscope (all after Guinot & Richer de Forges 1997: figs. 5B–D, 9A, B). a.c., sterno-abdominal cavity; cx5, P5 coxa; g, gonopore; G1, first gonopod; p, uropod as dorsal platelet; pt, pleotelson; s, socket; 6–8, thoracic sternites 6–8; 6/7, 7/8, thoracic sternal sutures 6/7, 7/8. Scale bar: 100µm (E).

FIGURE 30. Coxal to coxo-sternal condition in Leucosiidae, penis enclosed inside P5 coxo-sternal condyle (abdomen removed). A, Ebalia nux A. Milne-Edwards, 1883, male, Bay of Biscay (MNHN-B20067); B, Pseudomyra mbizi Capart, 1951, male 25.0 × 20.0 mm, Congo (R.O.C.) (MNHN-B17570); C, Philyra pisum De Haan, 1841, male 21.0 × 20.5 mm, Japan (MNHN-B16982). D–F, schematic representation of A–C, portion of P5 coxa covered by episternites 7 and 8 (dotted line in C and F). a8, p8, anterior and posterior expansions of sternite 8; c, P5 coxo-sternal condyle; cx4, cx5, P4, P5 coxae; e7, e8, episternites 7, 8; g, gonopore; 7, 8, thoracic sternites 7, 8; 6/7, 7/8, thoracic sternal sutures 6/7, 7/8.

FIGURE 31. A–D. Condylar protection: penis within P5 coxo-sternal condyle. A, Orithyia sinica (Linnaeus, 1771) (Orithyiidae), male 80.0 × 74.0 mm, North China Sea (MNHN-B11612); B, Sayamia germaini (Rathbun, 1902) (Parathelphusidae), male 35.9 × 44.0 mm, holotype of Potamon (Parathelphusa) germaini, Vietnam (MNHN-B5162): penis emerging from extremity of long condyle; C, Leurocyclus tuberculosus (H. Milne Edwards & Lucas, 1842) (Inachoididae), male 33.6 × 36.8 mm, Urugay (MNHN-B9366); D, Stenorhynchus lanceolatus (Brullé, 1837) (Inachoididae), male 34.0 × 16.0 mm, West Africa (MNHN-B21437). E. Non-condylar protection in Dilocarcinus pagei pagei Stimpson, 1861 (Trichodactylidae), male 40.0 × 50.0 mm, Bolivia (MNHN-B12813): penis, with thick sheath, emerging from oblique superior border of condyle. F. Vestigial pleopods on male abdominal somite 5 of Orithyia sinica, same data as in A. a5, a6, abdominal somites 5, 6; c, coxo-sternal condyle; cx5, coxa of P5; e7, episternite 7; G1, first gonopod; I, prominence of sternite 8; m.l., median line; p, penis; pa, papilla; t, sclerotised sheath; te, telson, v, vestigial pleopod; 6–8, thoracic sternites 6–8; 7/8, thoracic sternal suture 7/8. Scale bars: 5 mm (A, F); 1 mm (partial enlargement of A).

FIGURE 32. Coxo-sternal condition in Palicidae (A, B, E), Crossotonotidae (C, D), schematic representation. A, B, Palicus caronii (Roux, 1830), male 9.0 × 11.0 mm, Canary Is. (MNHN-B16335). A, thoracic sternum and abdomen in situ; B, detail of A. C, D, Crossotonotus spinipes (de Man, 1888), New Caledonia (MNHN-B26811): detail of sternites 7 and 8. C, male 36.0 × 44.0 mm; D, female 29.0 × 32.0 mm. E, Palicus zonatus (Rathbun, 1893), male 10.0 × 13.0 mm, Gulf of California (MNHN- B26745): ventral view, abdomen removed; invaginated sternite 8 has been partially cut to expose underlying penis (left). a1, abdominal somite 1; b, press-button; cx4, cx5, P4, P5 coxae; e3–e7, episternites 3–7; l, invagination line of sternite 8; m.l., median line; p, penis (schematically represented too thin, straight in A and B, correctly represented thick and at an angle in Fig. E); p5, pereopod 5; 6, 7, thoracic sternites 6, 7; 8, thoracic sternite 8 in the form of double plate. Scale bar: 3 mm (E).

FIGURE 33. Thoracotreme disposition. A, Plagusia squamosa (Herbst, 1790) (Plagusiidae), male cw 55.0 mm (dissected), French Polynesia (MNHN-B8889): trajectory of vas deferens directly connecting gonad to sternal gonopore. B, Ostracotheres tridacnae (Rüppell, 1830) (Pinnotheroidea), male, 9.0 × 8.6 mm, Egypt, Suez (MNHN-B10578): thoracic sternum; C, Pinnotheres pisum (Linnaeus, 1767) (Pinnotheroidea), male 4.7 × 4.6 mm, Bay of Biscay, La Rochelle (MNHN-B10604): male gonopore (B, C, after Guinot 1979a: figs. 24D, 53G respectively). cx5, P5 coxa; g, sternal gonopore; j, lateral extension of sternite 4 resulting in incomplete thoracic sternum/pterygostome junction; m.l., median line; mxp3, external maxillipeds; p, penis; t, telson; v.d., vas deferens; 3–8, thoracic sternites 3–8; 4/5–7/8, thoracic sternal sutures 4/5–7/8. Scale bar: 1 mm (A).

FIGURE 35. Metasesarma aubryi (A. Milne-Edwards, 1869) (Sesarmidae), male 18.1 × 21.5 mm, Red Sea (MNHN-B13230): penis in rest position (abdomen lifted), exposed above the G1. cx4, P4 coxa; e6, episternite 6; g, male gonopore; G1, first gonopod; G2, second gonopod; i.a.2, internal folding of abdominal somite 2; i.s., incomplete sulcus of thoracic sternite 7; p, penis; 7, 8, thoracic sternites 7, 8; 6/7, 7/8, thoracic sternal sutures 6/7, 7/8. Scale bar: 3 mm.

FIGURE 36. Sternal penis protection in Varunidae. Varuna litterata (Fabricius, 1798), male 41.2 × 46.0 mm, New Caledonia (MNHN-B-22720): penis emerging from large orifice on thoracic sternite 8. A, invaginated papilla; B, extended papilla. o, sclerotised sheath acting as an operculum; pa, papilla. Scale bar: 1 mm.

FIGURE 37. Podotreme condition and spermatheca at extremity of suture 7/8 in fossil podotreme †Basinotopus tricornis Collins & Jakobsen, 2004 (Dynomeniformia Dromiidae †Basinotopinae), Eocene of Denmark. A, thoracic sternum; B, detail. cx1–cx3, P1–P3 coxae; e4, e5, episternites 4, 5; g, gonopore; s, spermatheca; 4–8, thoracic sternites 4–8; S6/7, S7/8, thoracic sternal sutures 6/7, 7/8. Scale bars: 5 mm. Modified from Collins & Jakobsen 2004: fig. 1b.

FIGURE 38. Non-gymnopleurity in extinct †Palaeocorystoidea (A) in contrast to gymnopleurity in Raninoidea, in extant (B, C) as well as fossil (A) species. A, †Ferroranina dichrous (Stenzel, 1945) (†Palaeocorystoidea †Palaeocorystidae), height of image 48 mm, Middle Cretaceous (Cenomanian-Turonian), Eagle Ford group, Texas: branchiostegite in contact with the coxae; B, Lyreidus tridentatus De Haan, 1841 (Raninoidea, Lyreididae), male 37.1 × 20.7 mm, Philippines (MNHN-B13364): exposed pleurites 5–7 forming flat plate; C, Ranina ranina (Linnaeus, 1758) (Raninoidea Raninidae), female 85.8 × 73.1 mm, Madagascar, Tulear (MNHN-B31983): exposed pleurites 5–7 as excavated plate, dorsally surrounded by P5. a, abdomen; b, branchiostegite; c, carapace; cx, coxa of pereopod; P1–P5, pereopods 1–5; 5–7, exposed pleurites 5–7. Scale bars: 5 mm (B, C). (A, courtesy of B. van Bakel).

FIGURE 39. †Notopocorystes stokesii (Mantell, 1844) (†Palaeocorystoidea †Palaeocorystidae), Middle Albian, France. A, indeterminate sex, carapace, dorsal view (MAB k. 2872); B, D, female 29.0 × 20.0 mm (overall) (MAB k. 2873): B, thoracic sternum; D, posterior part of sternum with spermathecae and “double peg” for abdominal-locking mechanism; C, female, height of image 17 mm (MAB k. 2874), with gonopore on P3 coxa and characteristic crescent-shaped sternal suture 4/5. cx2– cx4, P2–P4 coxae; c.s., crescent-shaped suture; d.p., double peg; e4, e5, episternites 4, 5; g, gonopore; G1, first gonopod; g4, gynglyme for P1 on thoracic sternite 4; g5, gynglyme for P2 on thoracic sternite 5; g6, gynglyme for P3 on thoracic sternite 6; s, spermatheca; 4–6, thoracic sternites 4–6. (Courtesy of B. van Bakel).

FIGURE 40. Evolution of the abdominal-locking mechanism among the Gymnopleura: ancestral condition in extinct †Palaeocorystoidea (F–I) and derived condition in extant Raninoidea of the family Lyreididae (A–E). A, Lyreidus tridentatus De Haan, 1841, ovig. female 38.0 × 23.0 mm, Australia, New South Wales (MNHN-B13316); B, L. brevifrons Sakai, 1937, male 49.4 × 23.4 mm, New Caledonia (MNHN-B24319): ventral surface, abdomen in locking position (A), abdomen removed (B); C, D, Lysirude channeri (Wood-Mason, 1885), Philippines (MNHN-B11562): sexual dimorphism of episternite 5 hook: strong teeth in male (C, 25.5 × 17.4 mm), blunt, eventually lost, and hook setose in large female (D, 33.7 × 20.5 mm); E, Lyreidus stenops Wood- Mason, 1887, male 33 × 18 mm, Philippines (MNHN-B13367): hook with 2 strong teeth; F, G, †Notopocorystes stokesii (Mantell, 1844), Middle Albian, indeterminate sex, 35 × 22 mm (overall), France (MAB k. 2870): F, ventral surface without abdomen; G, detail of double peg on episternite 5; H, I, †Eucorystes iserbyti Van Bakel, Guinot, Artal, Fraaije & Jagt, 2012, Middle Albian, indeterminate sex, height of image 6.0 mm, France, Marne (MAB k. 2871): H, thoracic sternum without abdomen; I, detail of double peg on salient episternite 5, notice the two recurved teeth. c.s., crescent-shaped suture; d.p., double peg; cx2, cx3, P2, P3 coxae; e4, e5, episternites 4, 5; g4, gynglyme for P1 on thoracic sternite 4; g5, gynglyme for P2 on thoracic sternite 5; G1, first gonopod; h, episternite 5 hook; to, tooth; 4, 5, thoracic sternites 4, 5. C–E, scanning electron micrographs. Scale bars: 100 µm (C, E), 1 mm (D). (C–E, courtesy of J.-M. Bouchard; F–I, courtesy of B. van Bakel).

FIGURE 41. Cladogram of crown group brachyuran relationships based on morphological data. Other possible trees are discussed in the text. Subsections Dynomeniformia: Homolodromioidea (Homolodromiidae), Dromioidea (Dromiidae, Dynomenidae); Homoliformia: Homoloidea (Homolidae, Latreilliidae, Poupiniidae); Cyclodorippiformia: Cyclodorippoidea (Cyclodorippidae, Cymonomidae, Phyllotymolinidae); Gymnopleura: †Palaeocorystoidea (†Palaeocorystidae, †Camarocarcinidae, †Cenomanocarcinidae, †Necrocarcinidae, †Orithopsidae), Raninoidea (Raninidae, Lyreididae). Synapomorphies: (1) uniramous uropods; (2) G1 and G2 acting together and with penis; (3) paired spermatheca; (4) uropod as a ventral lobe or a dorsal plate; (5) structure of abdominal locking-system on thoracic sternite 4 (“homoliform press-button”); (6) uropods completely lost; (7) female gonopore on thoracic sternite 6 (vulva); (8) eubrachyuran sella turcica; (9) uropod modified into a socket acting with typical press-button system on sternite 5; (10) oxystome condition; (11) both P3 and P4 with modified distal articles; (12) structure of abdominal locking-system on thoracic sternite 5; (13) double peg on thoracic sternite 5; (14) gymnopleurity; (15) male gonopore on thoracic sternite 8. “?” means "no synapomorphy". A third brachyuran synapomorphy “brachyuran sella turcica” could be added to (1) and (2). Hymenosomatoidea is not taken into account.

FIGURE 42. Milne Edwards openings and cases of thoracic sternum/pterygostome junction. A, Ethusa magnipalmata Chen, 1993 (Dorippoidea Ethusidae), male 12.4 × 11.6 mm, holotype, Loyalty Is. (MNHN-B21524): no junction, “normal” Milne Edwards openings; B, Amarinus lacustris (Chilton, 1882) (Hymenosomatidae Odiomarinae), male 4.7 × 4.7 mm, New Zealand, Oratia stream (ZRC2010.0238): sternum/pterygostome junction, Milne Edwards opening sealed by largely exposed, falbelliform, whitish mxp3 coxa, and separated from the cheliped; C, Medorippe lanata (Linnaeus, 1767) (Dorippoidea Dorippidae), female 17.8 × 23 mm, Mediterranean Sea (MNHN, skeleton prepared by S. Secretan): Milne Edwards openings as pterygostomial slits; D, Coleusia signata (Paul'son, 1875) (Leucosioidea Leucosiidae), male 24 × 21 mm, Suez Canal (MNHN- B31894): large junction, absence of Milne Edwards openings; E, Ranina ranina (Linnaeus, 1758) (Raninoidea Raninidae), female 85.8 × 73.1 mm, Madagascar, Tulear (MNHN-B31983): large junction, absence of Milne Edwards openings. cm, mxp3 coxa; e, sternal extension; j, sternum/pterygostome junction; l, leucosiid abdominal-locking structure; m, Milne Edwards openings; p, sternal perforation; pt, pterygostome; p.b., press-button; s, dorippid Milne Edwards opening as pterygostomial slit; v, vulva with operculum; 3–6, thoracic sternites 3–6; 3/4, 4/5, thoracic sternal sutures 3/4, 4/5. Scale bars: 3 mm (C), 5 mm (D). (B, photograph by J.C. Mendoza).

FIGURE 43. Hymenosomatidae. A, B, Limnopilos naiyanetri Chuang & Ng, 1991, Thailand, observed in aquarium in Singapore: (A) crawling; B, male mating with an ovigerous female (photographs by R. Yeo, courtesy of P.K.L. Ng); C, Amarinus lacustris (Chilton, 1882), male 4.7 × 4.7 mm, New Zealand, Oratia stream: largely exposed, white, flabelliform coxae of mxp3 (ZRC2010.0238). Photographs by J.C. Mendoza.

FIGURE 45. Skeletons with parallel endosternites and narrow sella turcica. A, Pseudopalicus declivis Castro, 2000 (Palicidae), male, New Caledonia (MNHN-B30492): medially attenuated endosternites, remaining lateral portions covered by pleurites; marked median plate; B, Bathypluma spinifer Saint Laurent, 1980 (Retroplumidae), male, Philippines (MNHN-B37017): cx5, P5 coxa; m, median plate; p, pleurite; P5, pereopod 5; s, sella turcica; s.c., convexity of sterno-abdominal cavity; t, endosternite.

FIGURE 46. Axial skeleton of Medorippe lanata (Linnaeus, 1767) (Dorippidae), male, Mediterranean Sea (MNHN). A, carapace (partially removed) covering all pleurites except exposed lateral portions of pleurites 5–7; B, detail of lateroposterior region of carapace to show exposed, calcified external portion of pleurites 5–7, with setting gutter for carapace; thick line indicates carapace border; C, skeleton with endosternites, median plate, and sella turcica. a1, first abdominal somite; b, sclerotised penial bulb; c, carapace edge; ce, ventral extension of carapace posterior edge; cp5, cp6, calcified portion of pleurites 5, 6 covered by carapace; cx2–cx5, coxae of P2–P5; c6, c7, coxo-pleural condyles of P3, P4; e, epimere (membrane); e6, extension of pleurite 6 covered by carapace; ep5–ep7, exposed pleurites 5–7; e4/5–e6/7, endopleurites 4/5–6/7; g, setting gutter of carapace; ga, gap between laminae of median plate; m, articulating membrane; m.p., median plate; P4–P8, pleurites 4–8; 8, exposed portion of sternite 8; s, sella turcica. Skeletons prepared by S. Secretan.

FIGURE 47. Skeletons of Dorippidae (A, B), Inachidae (C), Hymenosomatidae (D–F), and Inachoididae (G–I). A, B. Medorippe lanata (Linnaeus, 1767), Mediterranean Sea, dorsal view. A, male, after removal of carapace; B, female, after removal of pleurites; C, Inachus dorsettensis (Pennant, 1777), female, Mediterranean Sea (MNHN): sagittal section. D–F, Odiomaris pilosus (A. Milne-Edwards, 1873), New Caledonia (MNHN): D, dorsal view after partial removal of carapace; E, axial skeleton, dorsoventral view after removal of pleurites; F, axial skeleton, sagittal section. G, H, Paulita tuberculata (Lemos de Castro, 1949); G, male 23.0 × 23.0 mm, French Guiana (MNHN-B19511): exposed pleurites 5–8 visible after removal of carapace; H, female: axial skeleton, sagittal section. I, Leurocyclus tuberculosus (H. Milne Edwards & Lucas, 1842), Brazil (MNHN): sagittal section. a, abdomen with pleopods; a1, first abdominal somite; c, carapace; cx5, P5 coxa; e, endopleurite; e.p., exposed pleurite; e5–e8, exposed pleurites 5–8; g, setting gutter of carapace; h, hymenosomian rim; j, junction plate; m, median plate; p, pleurite; pi, pillar; pl, pleopod; s, sella turcica; st, thoracic sternum; t, endosternite; v, location of vulva; y, eye; 3/4–6/7, endosternites 3/4–6/7; 4–8, pleurites 4–8. C, F, H, I: skeletons prepared by S. Secretan.

FIGURE 48. Location of vulvae. A–C, anterior displacement: A, Anasimus latus Rathbun, 1894 (Inachoididae), female 18.4 × 14.0 mm, French Guiana (MNHN-B17807); B, Capartiella longipes (Capart, 1951) (Inachidae), ovig. female 9.4 × 8.1 mm, Nigeria (MNHN-B19602). C, Halicarcinus planatus (Fabricius, 1775) (Hymenosomatidae), female 14.0 × 19 mm, South Pacific, Campbell I. (MNHN-B25960) (schematic); D, Hiroia krempfi Fize & Serène, 1956 (Cryptochiridae), female, Palau: vulvae not widely separated and sternal suture 7/8 complete (schematic, modified from Kropp 1990: fig. 6c); E, Pseudohapalocarcinus ransoni (Fize & Serène, 1956) (Cryptochiridae), female, Guam: vulvae located apart on broad thoracic sternite 6 (schematic, modified from Kropp 1990: fig. 12c). j, sternum/pterygostome junction; m.l., median line; v, vulva; 3–8, thoracic sternites 3–8; 4/5–7/8, interrupted thoracic sternal sutures 4/5–7/8; w, sternal wall.

FIGURE 49. Capartiella longipes (Capart, 1951) (Inachidae), male 8.2 × 6.5 mm, Dahomey (= Benin) (MNHN-B19590). A, carapace, dorsal view: notice exposed pleurites 5–8; B, ventral view; C, thoracic sternum, abdomen lifted, G1 lowered; D, detail of pleotelson with moveable flap supposedly homologous to vestigial uropod. a1, a5, abdominal somites 1, 5; b, pressbutton; c, carapace edge; cx1, P1 coxa; e5–e8, exposed pleurites 5–8; f, flap; G1, first gonopod; pt, pleotelson (somite 6 fused to telson); r, roof formed by thoracic sternite 8; 3–8, thoracic sternites 3–8.

FIGURE 50. A–G. Thoracic sternum (A, C, E, G) and ventral surface of abdomen (B, D, F) of Inachidae (A, B), Inachoididae (C–F), and Majidae (G). A, B, Inachus dorsettensis (Pennant, 1777), male 20.0 × 18.0 mm, France, Saint Martin de Ré (MNHN-B17337); C, D, Collodes leptocheles Rathbun, 1894, male 17.5 × 13.2 mm, Gulf of Mexico (MNHN-B14055); E, F, Pyromaia tuberculata (Lockington, 1877), male 17.0 × 12.2 mm, Baja California (MNHN-B9290); G, Prismatopus harmandi (Bouvier, 1906), male 46.0 × 24.5 mm, holotype, vicinity of Tokyo (MNHN-B22330). H. Palicus caronii (Roux, 1828), male 6.0 × 7.0 mm, eastern Atlantic (MNHN-B16335): ventral surface of abdomen with long socket on somite 6. b, press-button.

FIGURE 51. Articulation of abdominal somites, dorsal view. A, Nephrops norvegicus (Linnaeus, 1758) (Nephropidae), eastern Atlantic (MNHN); B, †Eocarcinus praecursor Withers, 1932 (†Eocarcinidae), Lower Jurassic, England, paratype (BM 23091); C, Dicranodromia sp. (Homolodromiidae), Philippines (MNHN-B28939); D, Ranina ranina (Linnaeus, 1758) (Raninidae), Madagascar (MNHN-B7662); E, Homola ranunculus Guinot & Richer de Forges, 1995 (Homolidae), New Caledonia (MNHN- B19871); F, Sternodromia spinirostris Miers, 1881 (Dromiidae), male, Congo (R.O.C.) (MNHN-B7863); G, Medorippe lanata (Linnaeus, 1767) (Dorippidae), male, West Africa (MNHN). a1–a3, abdominal somites 1–3; cx5, P5 coxa; m, articular membrane; p, abdominal pleura; t1–t3, abdominal tergites 1–3 with calcified anterior (a) and membranous posterior (P) parts; t, telson.

FIGURE 53. Carrying behaviour in Dromiidae (A, B), Homolidae (C, D), and Cyclodorippidae (E). A, B, Tumidodromia dormia (Linnaeus, 1763), Mayotte, carrying sponges; C, Homola orientalis Henderson 1888, French Polynesia, with the carried sponge (detached); D, Lamoha inflata (Guinot & Richer de Forges 1981), French Polynesia, carrying a sea anemone; E, Deilocerus planus (Rathbun, 1900) carrying piece of shell (see Wicksten 1982). A, B, photographs by J. Dumas; C, D, by J. Poupin; E, sketch by M. Wicksten.

FIGURE 56. Diagrammatic representation of patterns of thoracic sternal sutures 4/5–7/8 and modalities of median line (median plate not taken into account) in Eubrachyura. A, pattern 1: sutures 4/5–7/8 complete (uninterrupted), median line along sternites 5–8, may extend on sternite 4; B, pattern 2: only suture 4/5 interrupted, median line along sternites 6–8; C, pattern 3: sutures 4/5, 5/6 interrupted, median line along sternites 7, 8; D, pattern 4: sutures 4/5–6/7 interrupted, median line on sternite 8; E, generalised pattern 5: sutures 4/5–7/8 incomplete, median line variable among subpatterns a-f; F, pattern 6: sutures 4/5–7/8 interrupted, suture 6/7 complete, median line along sternites 7, 8; G, subpattern 5a: sutures 4/5–7/8 interrupted, median line absent; H, subpattern 5b: sutures 4/5–7/8 interrupted, median line along sternites 3–8; I, subpattern 5c: sutures 4/5–7/8 interrupted, median line along sternites 5–8, sometimes partially on sternite 4; J, subpattern 5d: sutures 4/5–7/8 interrupted, median line along sternites 5–7; K, subpattern 5e: sutures 4/5–7/8 interrupted, median line along sternites 7, 8; L, subpattern 5f: sutures 4/5–7/8 interrupted, median line on sternite 4. Each thoracic sternum with same width; all sutures represented parallel, equidistant, and similar. Some additional patterns are not figured.

FIGURE 58. Male reproductive system in Hymenosomatoidea (A, B, E), Pinnotheroidea (C, D), and male gonopores in Cryptochiroidea (F). A, B, dissection of Odiomaris pilosus (A. Milne-Edwards, 1873), male 11.0 × 14.0 mm, New Caledonia, Wé Waalu River (MNHN-B32604): ejaculatory duct and penis (highlighted in B); C, D, dissection of Dissodactylus crinitichelis Moreira, 1901, male 4.6 × 7.0 mm, Brazil (MZUSP 24231): ejaculatory duct (highlighted in D). E, testis and vas deferens as seen through the transparent carapace of Trigonoplax unguiformis (De Haan, 1839), male 6.1 × 7.3 mm, Western Australia (USNM 184973); F, Utinomiella dimorpha (Henderson, 1906): male gonopores perforating sternite 8 far from suture 7/8 (modified from Henderson 1906: pl. 8, fig. 8, as Cryptochirus dimorphus). cx1, cx5, P1, P5 coxae; e.d., ejaculatory duct; g, gonopore; p, penis; t, testis; 4–8, thoracic sternites 4–8; 4/5–7/8, interrupted thoracic sternal sutures 4/5–7/8. Scale bars: 1 mm (A, B); 4 mm (C, D).