CRITERIA. 243 



precede both exceptionally hardened and unusually fleshy ones. The definitely winged 

 bracts in A. canescens are the result of a development quite different from that repre- 

 sented by irregular appendages and indicate a divergent phylogenetic branch, as will be 

 further noted when this species is reached in the taxonomic sequence. 



Inflorescence.— All stages in the development of complete dioecism are present in 

 Atriplex, beginning with the condition in which the sexes are mixed in small axillary clus- 

 ters. As the two kinds of flowers began to separate, the staminate came to occupy the 

 upper axils, the pistillate the lower ones. The staminate inflorescence is sometimes 

 very long and branched (bracteosa, wrighti, etc.), sometimes much reduced and yet 

 nearly or quite pure (fruticulosa, leucophylla, and some forms of peniandra). Partial 

 dioecism has been attained at several places in the evolution of the species (phyllostegia, 

 powelli), but it becomes complete only in two herbaceous species (decumbens, matamor- 

 ensis) and in the shrubs. In the latter group, ificomplete dioecism is not rare in several 

 of the less highly evolved forms (barclayana, nultalli, lentiformis) . 



The size of the staminate inflorescence and its degree of branching are useful as specific 

 and subspecific criteria in only a few cases. At other times there is considerable varia- 

 tion even within a single subspecies (A. peniandra typica and A. p. muricata) and hybrids 

 are suspected between species with long inflorescences and others with short ones {A. 

 davidsoni, p. 306). 



Leaves. — As in most large genera, the leaves are of much value in that they furnish 

 corroborative evidence useful in the identification of species, but not to be relied upon in 

 making the major groupings. They are usually alternate, but in a few species {decum- 

 bens, matamorensis) they are mostly opposite. In A. parishi the position varies independ- 

 ently of other characters. It has been determined by field observation and verified by 

 garden experiment that the lower leaves in most and probably all of the species of the 

 argentea and pusilla group are always opposite. Therefore the separation of species in 

 these groups on the basis of their "alternate" lower leaves, as has been sometimes done, 

 is not warranted. Such errors doubtless are the result of the incomplete nature of many 

 herbarium specimens, but it is not safe to assume that all of the leaves on a plant were 

 alternate merely because the remaining ones happen to be so. In some species the leaves 

 are closely sessile, in others decidedly petioled. The chief difficulty in using this feature 

 lies in the frequent interposition of intermediate species or minor forms in which the 

 leaves are narrowed below to a petiole-like base. A strong tendency is noted in some of 

 the more primitive herbaceous species to the formation of hastate and cordate bases to 

 the leaves, and this recurs in a few of the shrubby species. All gradations from a strongly 

 narrowed to a broadly hastate base are found in A. patula and' nearly all stages are repre- 

 sented also in A. phyllostegia and A. truncata. The amount of dentation or lobing, the 

 size, the shape, and the texture are useful in some cases for purposes of classification. 



Pubescence. — All of the North American Atriplexes are glabrous, except for a more or 

 less dense scurf consisting of modified trichomes. When the scurf is sparse or early decid- 

 uous, the herbage is greenish in color. This condition is pronounced only in the subgenus 

 Euatriplex (especially A. hortensis and A. patula), in the more primitive species of sub- 

 genus Obione (notably A. dioeca and A. phyllostegia) and in A. bracteosa. The remaining 

 species are densely scurfy, at least on the lower surface of the leaves, and the herbage is 

 therefore gray or whitish in appearance. This difference in color is an aid in distinguish- 

 ing on sight between some species otherwise much alike in all but minute technical char- 

 acters. However, it must be used with due regard to the presence of at least a sparse 

 scurfiness on the juvenile herbage of all forms. 



Habit. — With the exception of the anomalous A. calif arnica and the two introduced 

 Australian species, all of the subgenus Euatriplex and all of Obione up to the pentandra 

 group, as diagrammed in the chart of relationships (p. 238), consists of herbaceous annuals. 



