Comment on "The early evolution of the tetrapod humerus".

In their analysis of humeral evolution in early tetrapods (taken here as meaning vertebrates with limbs), Shubin et al. (1) described a suite of derived characters that represents a significant advance on previous analyses (2– 4). This makes their negative assessment of GSM 104536, the humerus provisionally attributed to Elginerpeton pancheni (2, 4, 6), all the more surprising. The specimen (see Fig. 1, A to F) derives from the late Frasnian (mid-Late Devonian) locality of Scat Craig in Scotland, and is thus about 10 million years older than the Catskill Formation humerus ANSP 21350 that forms the centerpiece of the Shubin et al. study (1, 7). It is associated with tetrapod lower jaws, snout fragments, shoulder girdle fragments, ilia, a femur, a tibia, and a probable tetrapod neural arch (2, 4, 6, 8). The Scat Craig specimens form the earliest known group of postcranial tetrapod elements, and GSM 104536—if correctly interpreted—is the earliest known tetrapod humerus. After presenting their list of derived humeral characters, Shubin et al. (1) state, in the caption of figure 2, that “[t]he putative humerus of Elginerpeton . . . becomes difficult to interpret in light of these observations, because it has none of the synapomorphies shown in Fig. 2. By lacking the shared characters present in the most basal node, the identification of Elginerpeton as a tetrapod humerus would require both the evolution of unique features and the loss of all the apomorphies shared by tetrapods and their successive two outgroups.” They reference the provisional description of GSM 104536 from 1991 (2), but the full description from 1998 (4) is not cited. The Shubin et al. statement that GSM 104536 has no tetrapod characteristics whatsoever begs the question of why I ever identified it as a tetrapod, and their claim is a contradiction of the published descriptions of the specimen (2, 4). The derived characters described by Shubin et al. for Panderichthys tetrapods include a dorsoventrally flattened shaft; an enlarged and posteriorly inclined ectepicondylar ridge; a broad, shallow proximal depression for scapulohumeral muscle insertion; and an enlarged area for muscle insertion above the radial condyle. The characters described for tetrapods alone include a distally expanded ectepicondylar process, a distinct recess and incipient crest at the proximal union of ectepicondylar and entepicondylar processes, an enlarged and proximodistally expanded entepicondylar process, and a ventral crest confluent with the posteromedial rim of the entepicondylar process. Shubin et al. claim that all of these features are absent in GSM 104536. My 1991 description (see Table 1) explicitly compared GSM 104536 with the humeri of Eusthenopteron and early tetrapods and identified equivalents of a number of these synapomorphies. The “thin, flat entepicondyle, continuous with flat body of humerus” and “sharp preaxial margin” in that description are collectively equivalent to the Shubin et al. “dorsoventrally flattened shaft” plus “enlarged and proximodistaly expanded entepicondylar process” [see figure 2 in (1)]. My “narrow, tall ectepicondyle” overlaps with their “enlarged and posteriorly inclined ectepicondylar ridge” and “distally expanded