lyo STEBBINS 



of the grass family, which already existed in the Cretaceous period, speaks 

 against such an assumption. In other plant groups, such as the conifers, as 

 well as in numerous orders of animals well represented in the fossil record, 

 the fossil evidence shows clearly that the primitive members of the group 

 concerned died out relatively early in its evolutionary history, and the mod- 

 ern genera are all specialized in one way or another. 



In the grasses, both Prat and Bews (1929) have considered the bamboos 

 to be ancestral to the other tribes. This viewpoint is reached because bam- 

 boos undoubtedly have more primitive flowers; i.e., their floral structure 

 approaches the generalized monocotyledonous, or lily-like, type more than 

 does that of any other members of this highly specialized family. But al- 

 though the bamboos are undoubtedly primitive in floral characters, they are 

 nevertheless highly specialized in many other ways. They have a complex 

 pattern of branching, highly specialized leaves, often on distinct petioles 

 (which are, however, absent from the young shoots), and an elaborate system 

 of rhizomes. Their tree-like habit of growth is itself probably a secondary 

 specialization, as suggested by Agnes Arber (1934), since it is entirely dif- 

 ferent from the growth habit of any other trees, either mono- or dicotyle- 

 donous, and the groups of monocotyledons which on various grounds are be- 

 lieved to be most closely related to the Gramineae (Flagellariaceae, Restio- 

 niaceae, primitive Liliaceae) are all herbaceous. The earliest members of the 

 grass family, therefore, probably had flowers somewhat like those of some 

 modern bamboos, but their growth habit and general appearance were in all 

 likelihood more like that of the more familiar herbaceous grasses. 



Hence, the following picture emerges of the broad outlines of the course 

 of evolution in the grass family. The earliest grasses were herbs with stems 

 having few to many nodes, relatively simple racemose or paniculate in- 

 florescences, and spikelets with numerous florets, the bracts or glumes being 

 undifferentiated like those of primitive bamboos. The flowers themselves 

 were trimerous, with a perianth of three and perhaps six members, six 

 stamens in two series, and an ovary with a single ovule and three styles and 

 stigmas. They were probably already wind-pollinated, but descended from 

 ancestors with a well-developed perianth and insect pollination. They prob- 

 ably existed in the middle of the Cretaceous period, judging from the scanty 

 fossil evidence that is available, and inhabited regions with a tropical or 

 subtropical climate. From this primitive, now extinct group evolved a series 

 of lines adapted in different ways to various habitats, as typical of adaptive 

 radiation in all groups with a known fossil record. Three of these lines had 

 unusual success. The first two developed primarily in the tropics. These were 

 Panicoideae and the Chloridoid-Eragrostoid line. They probably came from 

 closely related ancestors and evolved in similar directions. Their evolution 

 was characterized by the retention of primitive features in the leaf epi- 

 dermis, caryopsis, embryo, and seedling, but by specializations in the leaf 



