Cladistical analysis was performed only on the data of our “postlarval” semaphoront of the Paguroidea. These data were analyzed using PAUP version 4.0 beta 11 (Swofford, 2002), utilizing data matrices (Table 2) created in MacClade version 3 (Maddison & Maddison, 1992). Cladograms (Figs. 6, 7) were generated using the Branch-and-bound search option with the following parameters: addition sequence, furthest; branches collapsed if maximum branch length is zero; initial upper bound computed via stepwise; “Multrees” in effect; topological constraints not enforced; midpoint rooting in effect; and all characters unordered and with equal weight (1). In interpreting these cladograms, however, it must be emphasized that for reasons previously explained, the analyses was done without the use of an out-group (unrooted), so no ancestry can be identified. Unrooted trees such as these can only indicate relationships and not order of descent (Swofford et al., 1996; Hall, 2001). And while we refer to the positional alignment of taxa as sister groups, we do not intend to suggest phylogenetic relationships. As may be seen from the discussion below, we have, for the most part, relied upon character adjacencies and have followed character transformations in our determinations of directional change.
Despite our unconventional use of cladistic analyses and the limited postlarval data base, the trees generated (Figs. 6, 7) have shown certain points quite clearly. The first, and unquestionably the most significant is the distinct separation of lithodids from species of Pagurus. This separation is in marked contrast to the results presented by Cunningham et al. (1992: 539, fig. 1) who interpreted their DNA and other data to suggest that Lithodes aequispinus (as L. aequispina) and Paralithodes camtschaticus were nested within the genus Pagurus. As may be seen in Fig. 6, the less calcified Hapalogaster is the sister group to the more heavily calcified genera Lopholithodes, Paralomis and Lithodes, while apparent links between the lithodids and Paguristes and Discorsopagurus are unresolved. The former is a shell-inhabiting genus, the latter a polychaete worm tube dweller. However, these results cannot be interpreted as indicative of true, albeit distant, genetic relationships, but simply as the closest taxa that could be determined by a very small database. Among species of Pagurus, the sister taxa relationship between Pagurus bernhardus and P. ochotensis is to be expected; as adults they are both included in the bernhardus group of McLaughlin (1974). The less resolved relationships of the remaining Pagurus species reflects the polyphyly recognized within this genus. Although the cladogram does give some suggestion of relationships, it is not possible from it alone to determine whether evolution has proceeded toward or away from Pagurus.
The tree generated for only the megalopal and first crab stages (Fig. 7) is even less informative, but the distinct separation of the lithodids from the other paguroids is clearly maintained. Intrageneric similarities in developmental patterns are supported by the nesting of the species of Paralithodes and Cryptolithodes, whereas the nesting of Acantholithodes hispidus and Oedignathus inermis is most probably attributable to the onset of decalcification or dechitinization at crab stage 1 in both taxa. Data from additional early juvenile stages would undoubtedly result in distinct separation, as is already apparent in Phyllolithodes papillosus. The unresolved branches for Placetron wosnessenskii and the two species of Hapalogaster can most probably be attributed to the slower rates of pleonal transformation in the early postlarval stages of these taxa. Though unresolved, a suggested link between the Lithodidae and the remaining paguroids through Porcellanopagurus filholi Filhol, 1885b, may reflect a real pathway, or simply an independent “carcinization” event. Species of Porcellanopagurus share with all lithodids the complete absence of male pleopods, and with the Hapalogastrinae a reduced pleon.
A comparison of data from the megalopal and first crab stage as opposed to that from the megalopa and two early crab stages indicates that the reduction in taxa had little appreciable effect on the general pattern of relationship between lithodids and other paguroids. However, among the lithodids, the more general relationships suggested from information provided by just the first crab stage changed appreciably when two stages were considered. When only megalopa and first crab stage data were available, intergeneric relationships were rather poorly resolved. This is not surprising, as the effect of inherent heterochrony in pleonal plate development in particular, is reflected in the larger sample size. There was only a single deletion from the other paguroid data base, yet relationships, at least within Pagurus, with more than one crab stage provided definite indications, as noted above, of the polyphyly of this genus.