CYTOGENETICS AND EVOLUTION OF THE GRASS FAMILY 1 69 



unidirectional; the polyploids are, as a rule, derived from relatives with 

 lower chromosome numbers. Furthermore, although hybrid polyploid or 

 amphiploid species are constant and may even be ancient in the geological 

 sense, the diploid ancestors of many such species still exist and can be recog- 

 nized by their external morphology. Once the suspected ancestral species 

 have been identified, the supposed evolutionary course of hybridization and 

 chromosome doubling can be repeated in the garden or greenhouse, and the 

 hypothesis about phylogeny can be subjected to the acid test of prediction 

 and verification by experiment. No other type of evolutionary phylogeny 

 on the level of species or genera can as yet be subjected to this rigorous test. 

 Although experiments of this type have been performed in many different 

 plant families, the grasses provide particularly favorable material for them, 

 not only because of their high percentage of polyploidy, but also because 

 their species are widespread and common, and most of them can be raised 

 and hybridized under controlled conditions with relative ease. 



The few phylogenetic patterns of polyploidy in grasses which have been 

 analyzed by this method have had an important bearing on problems of plant 

 geography, particularly the interpretation of past migration and dispersal 

 on the basis of the distribution of modern species. In some instances the 

 ancestry of amphiploid or hybrid polyploid species has been diagnosed, and 

 one or both of the immediate parents of the amphiploid have been found to 

 occur in areas widely separated from its own area of distribution. Such 

 discrepancies call for an explanation in terms of altered geographic distribu- 

 tions in past geological epochs, since we must assume that at the time of its 

 origin, the amphiploid occurred together with both of its parental species. 

 If, in attempting to explain these alterations in distribution, we become 

 thoroughly familiar with the fossil evidence and with geological evidence 

 about past climates and the distribution of land and water, we can often use 

 polyploids to provide valuable evidence about this phase of historical plant 

 geography. 



In the preceding section, the broad outlines of the new orientation in 

 grass taxonomy and evolution have been sketched. The subsequent sections 

 will be devoted to amplifying this sketch with specific examples and will pose 

 some problems for future investigators. 



GENERAL TRENDS OF EVOLUTION IN THE GRASS FAMILY 



The nearest approach to a complete system of classification based upon 

 these newer characters was made by Prat (1936), who illustrated most of the 

 characters mentioned above and constructed a phylogenetic tree on the basis 

 of them. The difficulty with this "tree," as with nearly all such efforts, is 

 the assumption that the ancestral type of a given group has been preserved 

 to the present day and can be identified among living forms. The very age 



