G. UNDERWOOD 



mid-body scale-row counts for Bothrops caribbaeus from St 

 Lucia as 25-29 (mode 27) and for Bothrops lanceolatus from 

 Martinique counts from 29-33 (mode 31). The Philadelphia 

 Academy specimen, with 25 rows, is at the lower end of the 

 range for St Lucia specimens. Beyond reasonable doubt 

 therefore Cope had a specimen of C. errabunda which had 

 half swallowed a St Lucia Bothrops caribbaeus. 



Relationships of Clelia errabunda 



The species currently assigned to Clelia can be arranged at 

 several levels on the basis of the derived states of the 

 respiratory system, the vertebral scale row and the ventral 

 scale counts. This is set out in Table 3. 



C. bicolor. The three anterior supralabials, the low ven- 

 tral scale counts, the undifferentiated vertebral scales, the 

 presence of a left lung and a short trachea are primitive 

 features. The high maxillary tooth count appears to be a 

 derived feature. 



C. rustica. This species also has undifferentiated verte- 

 bral scales, a left lung (a mere vestige in BMNH 81.7.2.9) and 

 a short trachea. C. rustica and the following species are 

 derived in relation to bicolor in respect of two anterior 

 supralabials and higher ventral counts. 



C. errabunda, clelia, equatoriana and scytalina share the 

 derived feature of absent left lung. The latter three species 

 are further derived than errabunda in respect of the enlarged 

 vertebral scales and extended trachea. 



The type specimen of Clelia clelia Daudin (1803) is not 

 known to survive; however, the type locality is given as 

 'Suriname'. It is therefore assumed that specimens from the 

 northern coast of South America are typical clelia. Specimens 

 in The Natural History Museum from this area, from Central 

 America, from Rio Condoto on the Pacific slope of Colombia 



(1), from La Paloma nr Santiago R., Ecuador (1) and from 

 most of the rest of South America show the 19:19:17 scale 

 row pattern. However, a second specimen from Rio Condoto 

 and specimens from Ecuador (Guayaquil and east of Loja, 

 2), Peru (3) and Manacapuro on the Amazon (1) show the 

 17:19:17 scale-row pattern. These are indistinguishable from 

 typical clelia in respect of the other characters considered 

 here. Their occurrence in a north-western area of South 

 America with near overlap with the 19:19:17 (Rio Condoto) 

 form on the Pacific slope of Colombia does not look like an 

 accident of sampling (Fig. 5). The form in eastern Brazil 

 (plumbed) lacks spines on the hemipenis, some specimens 

 from the southern part of the range (occipitolutea) are pale in 

 colour, so clelia is evidently a variable species. Roze (1959) 

 reports a specimen from Venezuela with counts of: 

 21:22:19:17 and Chippaux (1986) reports a specimen with 

 21:19:17 rows from French Guyana. 



After the above account was prepared I received from 

 Zaher (personal communication) a photocopy of a portion of 

 Bailey's unpublished PhD thesis. It is evident that at that time 

 he regarded the island Clelia as sufficiently distinct to merit 

 subspecific status. He too did not believe the Guyana locali- 

 ties of the Sabine specimens. 



DISCUSSION 



The species clelia, equatoriana, scytalina and errabunda share 

 uniform dark adult colouration and absence of a small left 

 lung. Most other pseudoboine snakes have a small left lung 

 and a more varied colour pattern. Within this group of four 

 species errabunda is primitive to the others in that the 

 vertebral scale row is not enlarged and the trachea is short. 



Fig. 5. Localities of specimens from northern South America and the Lesser Antilles. Solid symbols = precise localities; hollow 

 symbols = approximate localities; circles = Clelia clelia; triangle = Clelia errabunda; 17, 19 = no of scale rows on neck of C. clelia. 

 B = Barbados, D = Dominica, Gr = Grenada, Gu = Guadeloupe, L = St Lucia, M = Martinique, V = St Vincent. 



