TAXON 24(I): 53-66. FEBRUARY 1975 EVOLUTION OF GRASSESAND GRASSLANDS IN SOUTH AMERICA Arturo Burkart* Summary This is a discussion of the South American grasslands from the standpoint of their evolution and composition. The tribes are considered in relation to climate, and grasses are classified as mega-, meso-, or microthermic with respect to their temperature requirements. The principal grassland regions are three: (A) Tropical and Subtropical, which include the Llanos of the Orinoco River system and the Campos Cerrados of Central Brazil; (B) Temperate, including the Pampa of Argentina and the Campos of Uruguay; and (C) Cold Country Grasslands, which are the Steppes of the high Andes and Patagonia, and also the Pairamos of Colombia and Ecuador. Some attention is given to the floristic composition of each of these regions. The subject of endemism is dealt with, as well as the problem of disjunct distribution. Included is a discussion of changes brought about by agriculture and ranching in historic times, and what may be expected in the future. INTRODUCTION The Gramineae, with about 6oo genera and some 6ooo species, is one of the largest families of flowering plants. It is a truly cosmopolitan group, and remarkable because of the capacity of its members to form the domi- nant vegetation over large areas of the earth's surface. The terms steppes, savannas, prairies, pusztas, campos or pampas all refer to vegetation types in which grasses are dominant. To quote Ronald Good (1953; p. 53) "Pride of place must certainly go to the Gramineae . ., the great family ... Not only do the grasses reach to the furthest land in the north and to the borders of Antarctica in the south, but their degree of distribution is usually particularly complete and continuous. Almost alone among the families of flowering plants they form the dominant element in vegetation over great areas of the world, and nearly everywhere else too, except perhaps in some forest regions, the proportion of these plants is high." South America, as we shall see, is no exception, having rich and extensive grasslands. If we include both pure and mixed types, savanna or parkland, this subcontinent possesses nearly 3,500,000 square kilometers of grassland. Approximately a quarter of its entire area is, or was originally, natural grassland (with, of course, many herbaceous dicotyledons and members of other groups associated). In the other vegetation types: Tropical Rain- forests, Subtropical Forests or Woods, Pairamos, Puna, Catingas, Chaco, Monte, and Deserts or Semi-deserts, grasses are well represented, but not dominant. An exception is, however, the Bambuseae, which are very conspicuous in hot humid or mountainous forest regions, and even in the * Instituto de Botinica Darwinion, Lavarden 0oo, Correo Martinez, San Isidro, prov. Buenos Aires, Argentina. FEBRUARY1975 53 humid temperate forests of south-central Chile (Parodi, 1945). Along with the Leguminosae, Rosaceae, Vitaceae, Solanaceae, Cruciferae, and Cucurbitaceae, the Gramineae are closely connected with history, pre- history, and the development of human civilization. Long before man appeared on earth, grasslands must have played a decisive role in the evolution of such important and diversified animals as the ruminants, equidae, and other herbivores - mainly in the Old World. In South America, grasslands developed extensively without supporting large herds of vertebrate herbivores - except perhaps during the Tertiary, with great extinct mammals - up to the time of the Spanish and Portuguese conquests. The influence of prehistoric man also was very limited. The natural grass- lands of South America are, therefore, a nearly undisturbed result of the great biological processes: evolution, competition, and migration, since their origins in the second half of the Mesozoic. During the entire Tertiary period, i.e. some 6o,000,000 years or more, the development of the grass- lands was governed by the principal environmental factors: climate, geol- ogy, soils, topography, and diverse biological influences (parasites, polli- nators, symbionts, etc.). In spite of the great number of publications devoted to grasses, innu- merable questions posed by the morphology, taxonomy, genetics, cytology, anatomy, embryology, physiology, ecology, etc., of these plants are still unanswered. They inspired Agnes Arber's (1934) admirable book, "The Gramineae;" a work of timeless value. Modern grass systematics has changed substantially the schemes of Nees, J. C. Doell, Bentham & Hooker, E. Hackel, and A. S. Hitchcock, superseding the classifications based purely upon exomorphology and inflorescence characters. It exemplifies the high- est development of an integrated or synthetic taxonomy, being a model not yet attained in other families. The modern system for the Gramineae (cf. Potztal in Melchior, Engler's Syllabus ed. 12. 1964) is the result of a combined effort of workers in the special fields mentioned above, and makes use of basic contributions by authors such as: N. P. Avdulov, R. J. Roshevitz, H. Prat, J. R. Reeder, L. R. Parodi, F. A. McClure, G. L. Stebbins, T. Tateoka, and others. New lines of research are developing, as for example the recent discovery of a number of entomophilous olyroid grasses in the tropical rain forest of northern South America (Soderstrom & Calderon, 1971). This changes the traditional picture of grasses which were thought always to be wind pollinated, and, moreover, breaks down the no less traditional concept of a correlation between petaliferous, showy flowers, with zoophily against perianth reduced "amentiferous" flowers and anemophily. Electron micros- copy, applied to plastids, pollen, and other cell structures, is able also to contribute substantially to a deeper understanding of the natural affinities or diversities. PHYLOGENY AND TAXONOMY OF THE GRASS FAMILY With respect to origins, we may focus on the main processes of evolu- tion: mutation, hybridization, introgression, chromosome changes - the domain of genetics, the science of heredity and variation. Or we may seek to describe the actual historical developments and connections by which the different taxonomic groups of grasses (i.e. tribes, genera, species) origi- nated from extinct progenitors. This is properly phylogeny, but to arrive 54 TAXON VOLUME 24 at meaningful conclusions, we need a fairly complete fossil record. Unfortunately, this is not the case in the Gramineae. The oldest remains are dated, with doubt, as Jurassic (Bambusidium liasicum Heer; Poacites (?), according to Roshevitz, 1969). Several fossils are considered to be Cretaceous, and they include some genera recognized as present-day Arundo, Phragmites, Bambusa (Gothan & Weyland, 1954). Two conclu- sions seem warranted: (I) the grass family appeared simultaneously with the early Angiosperms during the Cretaceous evolutionary explosion, and (2) their development and migration dates back to early and middle Tertiary times (Oligocene, Miocene, etc.). It continued vigorously during the Quaternary. From this later period, Roshevitz (1969) mentions many well known genera: Arundinaria, Setaria, Oryzopsis, Phragmites, Bromus, Festuca, Agropyron, Hordeum, Alopecurus, and Beckmannia. At the present time, however, paleontology has not revealed the multiple steps in the phy- logeny of the different grass groups. Even so, it is perhaps significant that the few earliest grass fossils, which have been identified with certainty, be- long to tribes or subfamilies considered primitive by modern agrostologists: Phragmitoideae (Arundo, Phragmites) and Bambusoideae (Bambusa, Chus- quea). In South America, several grass fossils have been identified. Menendez in Fittkau et al. (1969) reported Poacites from the Cretaceous, and Parodi & Frenguelli (1941) described a fossil bamboo of the genus Chusquea from Chubut. Such remains demonstrate the very old and independent develop- ment of the grasses in our subcontinent. But again, the fossil record is so meagre that it is of slight help in reconstruction of the phylogenetic events. It remains for us, therefore, to utilize the indirect but classical method, for the construction of an acceptable phylogeny of the grass family. We must rely on comparative morphology in its broad sense, including anat- omy, dermatography, palynology, embryology, as well as cytogenetics, ecology, and phytochemistry, in our attempt to discover the natural affinities. To venture a "genealogical tree" is a risky enterprise, but I think the possibilities in this case are equal or better than in any other plant group. Roshevitz (I946, cf. translation, 1969) gives an interesting picture of grass phylogeny, beginning with the primitive Streptochaeteae and other tribes of the subfamily Bambusoideae. The ancestors of Gramineae can perhaps be defined as Proto-Commelinales (Farinosae, pro parte), especially from pre-Restionaceous and Flagellariaceous stock. Roshevitz' conceptions, revolutionary at the time, are still valid in the main lines, e.g., the prim- itiveness of "Phragmitiformes", (now Phragmitoideae) is generally con- sidered to be a connecting link between Bambuseae and the three (now four) great, more modern and herbaceous groups, at present called Pooideae (Festucoideae), Eragrostoideae ("Eragrostiformes") and Panicoideae ("Pan- iciformes"). From the latter I prefer to separate as a parallel, perfectly equivalent group, the Andropogonoideae (Tribes Andropogoneae, Maydeae, and probably Arthropogoneae), founded by R. Pilger and E. Potztal. The characters of leaf anatomy and