no. 3645 PORTUNIDAE — STEPHENSON, WILLIAMS, LANCE 17 



An additional feature is the presence of large red spots on the 

 posterior portion of the carapace. 



Many of these are without great evolutionary significance and do 

 not appear early in keys. Others do (e.g., 14, 29, and S3) and even have 

 been used for subgeneric distinctions (e.g., 1 {). Knowledge of the charac- 

 ters of a wide range of Indo-West Pacific species (Stephenson and Camp- 

 bell, 1959; Stephenson and Rees, 1967) suggests that many additional 

 features result from parallel evolution (e.g., expansion of the antero- 

 external angle of the merus of the third maxilliped). 



The problems of recognising subgenera of Portunus by classical 

 criteria have been detailed (Stephenson and Campbell, 1959). The 

 present study, by showing gradations in affinities among several groups 

 or complexes and by adding to the list of features showing parallel 

 evolution, does not make this recognition any easier. 



General Conclusions 



With the work carried out against a background of classical knowl- 

 edge of the group, we found it not surprising that most of the detailed 

 conclusions do not greatly distort the accepted patterns. The most 

 significant of these conclusions appear to be as follows: 



(1) Analogous pairs of species of portunids with Pacific and 

 Atlantic forms are far from recognisable in most cases. 



(2) There are about 15 closely related American species, herein 

 called the P. xantusii group, which comprise nine Pacific and six 

 Atlantic species. This group appears to have originated from Pacific 

 ancestors. 



(3) The genus Callinectes appears to have had an Atlantic 

 origin. 



(4) One primarily Atlantic species, P. sayi (£§), belongs to the 

 predominantly Indo-West Pacific P. pelagicus group. P. sayi is 

 possibly the species in the group most closely related to the genera 

 Arenaeus and Callinectes. 



(5) There are grave doubts over the validity of separating 

 Arenaeus from Callinectes; these genera merit monographic treatment. 



Possibly the most important conclusions concern methodology. In 

 working on the American forms of Portunus, Callinectes, and Arenaeus 

 we hoped to develop techniques that could be applied to the very 

 numerous Indo-West Pacific species of Portunus, whose complex 

 interrelationships are difficult to clarify by the traditional approach. 



The results have shown that (1) for the recognition of groups, the 

 method embodied in the 3-dimensional models is entirely acceptable; 

 of these models, the second (based on three axes only) appears slightly 

 preferable; (2) for the detailed consideration of affinities within the 

 groups, the tabular method is adequate and gives less distortions; 



