A historical review of hydrozoan classifications was compiled and discussed by Bouillon et al. (1992). From a methodological perspective, all proposed classifications for the Hydrozoa have been essentialistic, i.e., focused on grades of evolution. Bouillon et al. (1992) also observed that most previous classifications have not been adequate. Although these classifications based their arguments on the assembling of taxa within historical concepts of some proposed names; the idea that a classification should represent some actual phylogenetic knowledge has not been considered. As a substitution for those former inadequate classifications, Bouillon et al. (1992) resurrected the name Hydroidomedusae (originally proposed by Claus, 1877a, 1877b) stating, “Our proposal is to divide the class Hydrozoa into two subclasses, the Hydroidomedusae and the Siphonophorae” (Bouillon et al., 1992: 282). Thus, within the Class Hydrozoa, the siphonophores would be juxtaposed against all other hydrozoans. Although Bouillon et al. (1992) did not justify the existence of such a subclass Hydroidomedusae with any clearly defined features, several authors have followed them, e.g., Pagès et al. (1992), Pagès & Gili (1992), and Boero et al. (1996). Others prefer to use the term Leptolida for this same grouping, e.g., Cornelius (1995), Vervoort (1995), and Brinckmann-Voss & Arai (1998), a name whose use was already denied by Bouillon et al. (1992). Still others simply still consider Hydrozoa to simply include several unnested groups, e.g., Schuchert (1998) and Collins (2000).
Since Darwin hierarchical classifications indicate relationships of subordination within nested sets. Since Hennig, we focus on trying to identify monophyly, natural groups that reflect common ancestry. If we presume that the label “Hydrozoa” (here considered a monophyletic group) is hypothesized to be derived from a single ancestral lineage, then all descendants could be grouped in subordinate patterns. If any of these subgroups are joined together under a taxon name, e.g., the Hydroidomedusae of Bouillon et al. (1992) and posed against another, i.e., the Siphonophorae, this would then necessarily imply that members of these subgroups are related and share a common lineage. The uniqueness of such groups implies that they represent monophyletic lineages that are distinct delineated by a set of autapomorphies.
Are the Siphonophorae monophyletic vis-à-vis the other hydrozoans? This is not clear in Bouillon et al. (1992), and I can see no justification other than that of historical convenience. Students of the Hydrozoa do not deal generally with siphonophores, possibly because of their very derived morphology, ecology, and life history. Obviously, historical convenience is not enough of a reason to separate a group. Perhaps the most striking difference in the siphonophore lineage is its “high degree of determinancy of form” (Mackie et al., 1987: 110-113). The modular construction of siphonophores is based on definite growth zones along an anterior-posterior axis. Derivatives of both polypoid and medusoid stages are found attached to the axial free-floating animal (Kirkpatrick & Pugh, 1984). Observing the singular morphology of the Siphonophorae, one might conclude that this group is highly modified and so, phenetically distant from all other Hydrozoa. Indeed, such phenetic classification reflects similarity relationships, which are not necessarily congruent with, or the result of, rigorous phylogenetic relationships.
However, could siphonophores have a common putative ancestor with some other hydrozoan subgroup(s)? Under such a hypothesis, following an event of cladogenesis, the siphonophore lineage underwent an accelerated process of anagenesis and/or specialization of form, resulting in a long and highly differentiated lineage. In this regard, Petersen (1979: 109 more explicitly; 1990: 105) suggested a possible relationship between Siphonophorae and the Anthomedusae, Leptomedusae, and Limnomedusae. Schuchert (1996: 13) was even more explicit in his statements, considering that “affinities [of Anthomedusae] with the Siphonophora and Laingiomedusae are not established. All three taxa have gonads on the manubrium as a possible synapomorphy. Desmonemes occur only in the Siphonophora and Anthoathecata [= Anthomedusae] and may indicate a closer relationships. A taxon Hydroidomedusae (Bouillon et al., 1992) which includes all Hydrozoa except the Siphonophora is therefore most probably not a natural taxon” (my italics).
Indeed, Marques (1997), in a phylogenetic study of the Cnidaria (Fig. 1), concluded there is a relationship among Leptomedusae, Anthomedusae, and Siphonophorae. The Limnomedusae, with no autapomorphy, are a group of uncertain monophyly. They might possibly form a clade with the other hydrozoan subgroups (Trachymedusae, Laingiomedusae, Actinulidae and Narcomedusae). However, that issue is not relevant here. A single group composed of the Leptomedusae, Anthomedusae, and Siphonophorae is supported by the following synapomorphies: 1) polypoid polymorphism; 2) presence of basitrichous isorhizas (homoplastic with Anthozoa, considered by him equivalent to atrichous isorhiza), and 3) loss of the endodermal statocyts on medusae (a reversal, homoplastic with Laingiomedusae, a group that would be eventually included in the Anthomedusae see Schuchert, 1996). The sister-group relationship of the Siphonophorae and Anthomedusae would be supported by the presence of desmonemes, stenoteles (a doubtful character since the nematocyst is lacking in filiferan Anthomedusae; also homoplastic with Cubozoa, Actinulidae, and Trachymedusae), and gonads on the manubrium (a reversal from gonads on the radial canals in the more inclusive subgroups, and another homoplasy with Laingiomedusae). Fig.1 also shows the autapomorphies of Leptomedusae, Anthomedusae and Siphonophorae. It is interesting to note that the Anthomedusae are weakly supported as a subgroup, since one of the putative autapomorphies (perradial mesenteries in the medusa [Petersen, 1990]) is not accepted by some authors (e.g., it is not cited in Schuchert, 1996), and the other (arrangement of the polypoid tentacles in more than one whorl) is reversed in several groups of Anthomedusae. The uncertainty associated with these two characters leaves open the question whether or not the siphonophores are a group belonging within the Anthomedusae (raised by Schuchert, 1996 and also P. Schuchert, pers. comm.). This possibility could be verified in a more detailed analysis in which the subgroups of Anthomedusae would be included, thus probing for anthomedusan paraphyly. Nevertheless, all of the above hypotheses (Petersen, 1979, 1990; Schuchert, 1996; Marques, 1997) cast serious doubt on the validity of the taxon Hydroidomedusae as a phylogenetically-based clade. Furthermore, if we take into account a completely different database, such as molecular 18S rDNA sequences, phylogenetic analysis (using both parsimony and maximum likelihood) still corroborates the association of Leptomedusae, Anthomedusae and Siphonophorae (cf. Collins, 2000).
Fig. 1. Relationship among the Leptomedusae, Anthomedusae and Siphonophorae, after Marques (1997).