1 74 STEBBINS 



Anomochloa, Streptochaeta, and Pariana in the American tropics, and sev- 

 eral strange endemics in Madagascar, described by A. Camus. These oddities 

 are a continual source of worry to taxonomists who wish to construct a neat, 

 well-ordered system. To the evolutionist familiar with several different 

 families, they are recognized as being a common phenomenon. They ap- 

 parently represent either the last vestiges of evolutionary lines which were 

 successful in the past and are now nearly extinct, or the culmination of 

 evolutionary trends which have been successful in only one specialized 

 ecological niche. 



SYNOPSIS OF THE PRINCIPAL CHARACTERISTICS ON WHICH 

 THE NEW SYSTEM IS BASED 



Figure 2 shows the evolutionary interrelationships of the principal tribes 

 and genera of grasses. This diagram has not been made in the form of the 

 customary evolutionary tree, with connections between the groups which 

 are supposed to represent extinct common ancestors. In a family like the 

 grasses, of which the fossil record is negligible, the actual connections be- 

 tween the modern groups can never be known and speculations about them 

 may be more misleading than enlightening. I have adopted, therefore, a 

 scheme which has been proposed by several authors and elaborated in par- 

 ticular by Sporne (1956). The modern groups are represented as cross sec- 

 tions of the branches of the evolutionary tree, with the distance between 

 branches representing as nearly as possible my conception of the degree of 

 resemblance or difference between the groups. Since this conception is based 

 on many characters, it should be in some sort of multidimensional form, and 

 any attempt to reduce it to the two-dimensional shape of a single surface 

 naturally produces distortion. Broken lines on the diagram indicate, there- 

 fore, that the groups on either side of them are farther apart from each 

 other than their position on the diagram would indicate. An attempt to 

 express different degrees of evolutionary specialization has been made by 

 placing a star in the position of the hypothetical primitive, generalized 

 extinct type which formed the beginning of the family and by arranging the 

 groups at distances from this star which correspond roughly to their degree 

 of evolutionary specialization. 



In fig. 3 and 4, the condition with respect to eight of the most significant 

 diagnostic characters is illustrated for the three major subdivisions of the 

 family, as well as for the relatively primitive group Bambuseae. Table 1 

 lists the condition with respect to these characters of four of the smaller 

 groups and of four well-known genera which do not fit well into any of the 

 eight groups listed in the diagram and table. The assignment of these genera, 

 as well as of about 20 others, to tribal or subfamily groupings must await 

 further study and may always be largely a matter of subjective judgment. 



