A NEW FAMILY OF TRIASSIC LOBSTERS (DECAPODA: ASTACIDEA) FROM BRITISH COLUMBIA AND ITS PHYLOGENETIC CONTEXT

Journal of Paleontology, Jan 2004 by Amati, Lisa, Feldmann, Rodney M, Zonneveld, John-Paul

3. Branchiocardiac Groove Origin. Methods for describing and interpreting carapace grooves vary. It is more important that homologous features are coded consistently than what name is given to each groove. For each taxon considered, a line drawing was made and the grooves were coded uniformly (Fig. 6). This process was also used for coding the postcervical groove.

The ventral extension of the branchiocardiac groove (a) has not been identified consistently among taxa. In Eryma (Forster, 1966, fig. 12) for example, this groove appears to extend ventrally to connect with the hepatic groove (b^sub 1^). The postcervical groove (c) arises from the branchiocardiac groove (a) above this point. In Nephrops, (Holthuis, 1974, fig. 24, B) the branchiocardiac groove (a) is absent and the postcervical groove (c) extends ventrally to connect with the hepatic groove (b^sub 1^). In other taxa, such as Hoploparia (Feldmann et al., 1977, pl. 3, figs. 1-6 and text fig. 7), some authors imply that the hepatic groove (b,) extends dorsally to connect with the cervical groove (e).

4. Intercervical Groove (c'). This term is not always used in the literature, especially for extinct taxa. Holthuis (1974) illustrated this groove as an anterior extension of the postcervical groove (c) on the nephropid carapace. In this analysis, the intercervical groove (c') was coded as present in erymids such as Eryma (Forster, 1966, fig. 12) and Enoploclytia (Forster, 1966, figs. 33-34).

5. Attachment Site of the Adductor Testis Muscle (x). One of the most distinctive features of Chimaerastacus is the definition of the attachment site of the adductor testis muscle by the ventral bifurcation of the branchiocardiac groove (a). The attachment site is also defined in Pseudopemphix, but by the ventral extensions of both the branchiocardiac (a) and postcervical (c) grooves.

Taxa used in cladistic analysis.-The purpose of this analysis is to determine which families belong to the infraorder Astacidea and to elucidate the relationships of these groups. Other taxa used in the analysis were included as "place holders" that allow us to test the monophyletic status of the Astacidea. Information regarding these taxa provides an interesting framework for future studies, but will not be commented on in great detail in the present work. Fossil genera were selected based on the availability of well-preserved and complete specimens. Taxa such as Protoclytiopsis antiqua (Birstein, 1958), Palaeopemphix (Gemmellaro, 1890), Litogaster turnbullensis (Schram, 1971), and Lissocardia (Forster, 1967) would have been interesting additions to the analysis, but the material is too fragmentary for comprehensive coding.

Results.-A parsimony analysis was conducted using PAUP* version 4.0b10 (Swofford, 2000). All characters were unordered and Penaeus and Stenopus were designated as out-groups. A heuristic search of 100 random replicates using TBR branch-swapping and ACCTRAN optimization was used. The 10,573 trees obtained from the heuristic search have a tree length of 70 steps, a consistency index of 0.571, a rescaled consistency index of 0.446 and a retention index of 0.781. Following the initial heuristic search, characters were reweighted using the rescaled consistency index and another heuristic search of 100 random replicates was run using TBR branch-swapping and ACCTRAN optimization. The result is five equally parsimonious trees with tree lengths of 40.495 steps, a consistency index of 0.642, a rescaled consistency index of 0.524 and a retention index of 0.815. A strict consensus of the five trees is shown in Figure 7. Figure 8 shows an optimized character distribution of one of the five trees using ACCTRAN optimization. This tree was chosen because it is the most resolved of the live and places the Nephropoidea lower in the tree than the Astacoidea, which is in closer agreement with the analyses of Scholtz and Richter (1995) and Schram (2001). An analysis with the same parameters was run using DELTRAN optimization and resulted in a strict consensus tree that is identical to the tree obtained using ACCTRAN optimization. The only significant difference in character distribution using DELTRAN optimization involves the fractosternum. Instead of arising at node 55 and then being lost at node 50 and node 47, the fractosternum is independently derived at node 44 and at node 54 and lost at node 50.