318 



GENUS ATRIPLEX. 



tissue which goes to make up thickness in the one case is compensated by the greater 

 lateral extension in the other. An earnest endeavor has been made to utilize these bract 

 features as criteria for the retention of each group in specific rank, but the too frequent 

 meeting of the characters, and the apparently total absence of correlated features seem 

 to make this impossible. Furthermore, there are some collections, such as Johnston's 

 3228 from Isla Partida, in which both compressed and thickened bracts occur on the 



Fig. 44. — Phylogenetic chart of the subspecies of Atriplex barclayana. 



same plant. This is possibly a case of hybridization, but, if so, the wide distribution of 

 both types over the same area could be expected to result in such frequent crossing as to 

 swamp the specific characters. As a final reason for merging the two groups may be 

 cited a parallel pair of forms under A. polycarpa, one with compressed, the other with 

 thickened bracts; yet here these characters are not concomitant with any others, nor can 

 the forms be assigned to different geographic areas. 



In considering these two groups of subspecies from an evolutionary viewpoint it seems 

 that the palmeri group is the more primitive. This is because the bracts are more like 



