I 68 STEBBINS 



their chromosomes, one forms a system strikingly similar to that based 

 on anatomy and histology, and equally different from that founded on the 

 traditional characteristics of the inflorescence. He therefore studied addi- 

 tional characters: the shape of the first seedling leaf, the organization of the 

 starch grains in the leaf and caryopsis, the organization of the resting nu- 

 cleus, and the geographic distribution of the genera. All these bore out the 

 system based upon anatomy and chromosomes. More recently, confirming 

 evidence has been obtained from studies of lodicules, caryopses, embryos, 

 root-hair development, and the reaction of germinating seeds to a weed- 

 killing organic chemical compound, IPC. We thus find that the realignment 

 of genera and tribes proposed by Avdulov is supported by nearly all the 

 characteristics which we can study, and so appears to reflect genetic and 

 evolutionary relationships better than the traditional system. This situation 

 has been recognized by an increasing number of taxonomists, particularly 

 Dr. C. E. Hubbard of Kew, England. He has pointed out the unnaturalness 

 of some traditional genera, such as Lepturus, which he has severed into four 

 genera belonging in different tribes. A radical rearrangement of genera and 

 tribes of grasses is in the making, some features of which will be presented 

 below. His system has been adopted by the authors of a recent Flora of the 

 British Isles (Clapham, Tutin, and Warburg, 1952). Pilger (1954) proposed 

 a system of classification which incorporated some of the newer characteris- 

 tics, but it remains unnatural in a great many respects. The changes men- 

 tioned above have affected chiefly the broader classification of higher cate- 

 gories of grasses. But the problems of grass taxonomy do not end with the 

 tribes and genera. Specialists who can recognize most genera of grasses at 

 first sight still have great difficulty in delimiting their species. When one 

 compares the diverse treatments which various monographs and floras pre- 

 sent for such genera as Festnca, Poa, Agropyron, Andropogon, and Panicum, 

 one realizes that even the experts have difficulty recognizing the species of 

 grasses. 



Cytogenetics has clarified the most important reason for this difficulty, 

 although this new approach has not simplified the problem of delimiting 

 species. Species delimitation of grasses is intrinsically difficult because inter- 

 specific boundaries have been blurred by hybridization and chromosome 

 doubling, or polyploidy. Because of the widespread action of these two 

 processes, the evolutionary "tree" of most grass genera is not a simple 

 branching affair, but a highly complex network. Most of the common species 

 of grasses are not descended from a single ancestral type, but contain in 

 varying proportions gene combinations derived from two, three, four, or more 

 separate and sometimes widely divergent ancestors. 



This situation, however embarrassing it may be to the taxonomist, pro- 

 vides a gold mine for the evolutionist who has enough patience, persistence, 

 and insight to analyze its problems. Evolution by polyploidy is essentially 



