On the origin and evolution of feathergrasses (Stipeae, Poaceae)

N.N. Tsvelev

[This is an English translation of an article that Tsvelev published in 1977. I found it thought-provoking and insightful. This translation is being posted so that English speakers interested in the tribe may read his ideas].

A more detailed study of the speciation and classification of the "actual victors in the fight for existence among plants" (Lavrenko, 1947, p. 6) has great scientific and practical interest. In particular, much of the speciation of the herbaceous phytocoenoses of the steppes and meadows of different types, includes many genera of grasses (Poaceae) of which the genus Stipa, comprising more than 300 species, is widely known. E. M. Lavrenko (1947), in a list cited by him of species in the plant associations of the Eurasian steppe regions lists 8 species of Stipa. However this number should be considerably larger, particularly if one includes not only the species of the lowland steppes, but also those of the varied upland coenoses in which feather grasses are encountered almost to the upper limit of plant life.

There are very few objective data concerning the historical development of grasses. In the fossil record, Stipa is known from the Oligocene (MacGinitie 1953). In the Oligocene deposits of Colorado in the U.S.A., imprints of the panicles of the feathergrass Stipa florissanti (Knowlt.) MacGinitie (= Muhlenbergia florissanti Knowlton) were examined by the American agrostologist Agnes Chase who considered them identical with the extant species, Stipa mucronata Kunth. The latter grows from Mexico to Ecuador on dry mountain slopes of the central zone of mountains and belongs to one of the primarily South American section Stipa sect. Stephanostipa [now included in Nassella] although based, on the structure of its awns, it is quite close to the Asian species S. richteriana Kar. et Kir.

 The most recent discoveries of fossil feathergrasses are known from lower Miocene deposits in the U.S.A. (El ias 1942, 1946). At this place, fossils of the related South American genera Nassella Desv. and Piptochaetium C. Presl were discovered, as well as the remains of other American prairie grasses. This indicates that the American prairie had already formed by this time. We note that the evolution of the Eurasian steppes is generally associated with the Miocene. There is no doubt that evolution of the genus Stipa, like other genera of the Stipeae Dum., occurred considerably before evolution of the lowland steppes, but this conclusion is based on indirect evidence: anatomical indications and the distribution of contemporary species of the genus.

The primary need is to become familiar with the limits of the tribe Stipeae and its interrelation with the other tribes of grasses. At the present time, the following 16 genera are included in the tribe: Achnatherum Beauv., Psammochloa Hitchc., Ptilagrostis Griseb., Stipa L., Macrochloa Kunth, Orthoraphium Nees, Eriocoma, Piptatherum, Piptochaetium Presl, Nassella, Oryzopsis Michx., Parodiella J. & C. Reeder [= Lorenzochloa J. & C. Reeder] , Aciachne Benth., Streptachne R. Br., Stephanachne Keng, and Pappagrostis Roshev., although not all of these are generally recognized . For instance, the genus Achnatherum comprising several sections (Tsvelev 1972, 1974a) is usually included in Stipa.

The boundary between these two genera is indeed very unclear, but the species of Achnatherum are obviously more primitive that those of Stipa s.s. They have a short, acute callus that is  less highly specialized than the long acuminate callus of Stipa. Other differences between Stipa and Achnatherum are not evident in all species but are always in the direction of greater specialization. Consequently, Achnatherum can be considered more primitive than typical species of Stipa. The margins of the lemmas of Achnatherum usually do not overlap each other, leaving the palea exposed; at the apex of the lemma there are often two sharp teeth; the awns are relatively short, often not clearly geniculate and not deciduous, and the glumes that are usually not as long acuminate as they are in the majority of species of Stipa. In addition, the species of Achnatherum are more primitive than those of Stipa in ecological terms. Most species of Achnatherum are plants of savanna-like associations which were present before the steppes evolved. Many of the species of Achnatherum for which the chromosome number is known have 2n=24, but for A. bromoides (L.) Beauv. the number is 2n=28, the basic number of chromosomes probably equaling 7. There is reason to consider 7 as the basic chromosome number for all grasses (Tsvelev 1975a). For this reason, the characteristic number for Achnatherum, 2n=24, can be considered the result of aneuploid loss of one chromosome in each of the four genomes that compose it.

A substantial majority of the species of Stipa (including all species of sects. Stipa and Smirnovia have 2n=44 for a chromosome number, clearly a secondary increase in comparison with the 24 chromosomes and undoubtedly the result of repeated hybridization among some, possibly even distantly related, ancestors. Stabilization of the hybrids in his case could occur by means of introgression or amphiploidy. One of the presumed ancestors could have had 2n=20 and the other 2n=24, as in species of Achnatherum. In Stipa s.s. (Tsvelev 1974a), 2n=24 is found in the most primitive section, Achnatheropsis Tsvelev, of which the type species is S. sibirica (L.) Lam., which is intermediate between Achnatherum and Stipa. This section is about equally similar to both these genera but, based on its chromosome number, it is closer to Achnatherum than to Stipa.

We have already noted earlier (Tsvelev 1974a. p. 13) that the most mesophyllic species of this section is S. extremiorientalis Hara and, based on the structure of its callus, it is clearly related to Achnatherum. Therefore. following Keng (Keng 1959), it is more appropriate to include this section in Achnatherum, changing its name to Achnatherum sect. Protostipa Tsvelev, sect. nov. (=Stipa sect. Achnatheropsis Tsvelev 1972. Novosti Syst. Vysh. Rast. 9:56). Species of the flora of the U.S.S.R. that should be referred to Achnatherum are A. sibiricum (L.) Keng (=Avena sibirica L. 1753. Sp. pl. :79). A. extremiorientale (Hara) (= Stipa extremiorientalis Hara 1939, Journ. Jap. Bot. 15: 459) and A. confusum Litv.) Tsvelev (=Stipa confusa Litv. 1929. Izv. AN U.S.S.R., ser. 7. 8: 53). It should be noted that most Achnatherum species, including sect. Protostipa, have extravaginal shoots whereas all typical Stipa have very thick clumps with only intravaginal shoots. The formation of these clumps should be considered an indication of higher specialization for xerophytic climates. Moreover, the presence of membranous cataphylls, the modified leaves associated with the formation of the extravaginal shoots of Achnatherum, is also the result of xerophyllization (Tsvelev 1974b). One can assume that the xerophyllization of the less well adapted ancestors of feathergrasses was accomplished in two ways: by the formation of the membranous cataphylls at the base of the shoots for protecting the fragile meristem and by formation of very thick clumps with only intravaginal shoots.

The small genus Ptilagrostis, in every respect very similar to Achnatherum, is distinguished from the latter only by its hairy, rather than scabrous, awns. Taking into account the alpine distribution of the species [with the exception of P. pelliottii (Danguy) Grub. which has secondarily descended to the plains of central Asia] and their relatively small size, one can assume that this genus i s the result of cryophyllizati on of some of its ancestors that were similar to Achnatherum but already xerophyllic to a considerable extent. The transition to scabrous trichomes on the awns to short pilose hairs that increase the evaporative surface could be accomplished only in relatively humid conditions such as the alpine regions. Nine closely related species, widely distributed in the mountain regions of Asia and North America, belong to the genus Ptilagrostis.

The monotypic central Asian genus Psammochloa is also very closely related to Achnatherum. Its great specialization is related to its adaptation to its habitat in moving sand. Two other oligotypic genera (containing but 2 species) which are closely related to Stipa: the western Mediterranean Macrochloa and the Khami Himalayan high mountain species Orthoraphium, are closer to Stipa in the structure of their callus but they have several features in common with Achnatherum that are indications of primitiveness. These genera are oftenincluded in Stipa, but in our opinion, they deserve recognition as separate genera.

Several groups of genera, more distantly related to Stipa sensu lato, are included by several authors under the name Oryzopsis. Tsvelev (1972, p. 57) and Freitag (Freitag 1975) showed that the oligotypic North American genus Oryzopsis with the type species 0. asperifolia Michx. not only differs in morphology from the remaining genera that many contemporary authors combine with them, but it also occupies quite an isolated position in the Stipeae. All indications of the reproductive organs of this mesophyllic forest genus are indicative of great specialization. From this it can be concluded that this line evolved a long time ago and is not ancestral to any of the more xerophyllic genera of the tribe. The old Mediterranean genus, Piptatherum, represented in the USSR by numerous species, is apparently absent from America (with the exception of the introduced species P. miliaceum (L.) Coss.), according to the monograph by R. Y. Roshevits (1951). The genus includes mesophyllic forest species ( for example, P. virescens (Trin.) Boiss. but a significant majority of the species are adapted to savanna-like vegetation types and to open rocky habitats. The basic difference between Piptatherum and Achnatherum is the very blunt rounded callus of the smooth, shiny lemma which should be considered more specialized than the sharp conical callus of Achnatherum. One can assume that the callus in the Stipeae has evolved in two different directions from the primitive form found in Achnatherum: to the long pointed acuminate callus of Stipa and to the very short rounded callus of Piptatherum. It should be noted that the morphological boundary between Piptatherum and Achnatherum is not completely clear. A few species of Achnatherum, for example, the Chinese A. chinense (Hitchc.) Tsvelev) with sort, easily deciduous awns, are very close to Piptatherum.

The North American genus Eriocoma (with type species E. hymenoides (Roem. & Schultes) Rydb.), judging by the structure of the callus, occupies an intermediate position between Achnatherum and Piptatherum. American authors usually include it in Oryzopsis, together with Piptatherum, but it could equally well be placed in Achnatherum in a section of its own. We prefer to consider Eriocoma as an independent genus, substituting in North America for the somewhat more highly specialized Eurasian genus Piptatherum.

Piptatherum also has an analog in South America, Piptochaetium, which is very rich in species and only just enters North America. The callus in Piptochaetium is more developed than in Piptatherum and the awn does not fall off as it does in many species of Achnatherum. However, in relation to the texture and shape of the lemmas, we believe it to be more highly specialized than Piptatherum. Another South American genus, Nassella, is close to Piptochaetium, but it is sti11 more highly specialized with respect to the structure of its lemma.

Other, mainly monotypic genera of Stipeae (the South American Parodiella [=Lorenzochloa] and Aciachne, the central Asian Stephanachne and Pappagrostis and the Australian Streptachne) are significantly more isolated, very highly specialized, and are in need of additional research for reliable clarification of their systematic position which is not at all clear at present.

 Accordingly, Achnatherum can perhaps be recognized as the most primitive genus i n the Stipeae, closely related to the genus Stipa but displaying completely distinct relationships with the genera that are close to Piptatherum. Let us now consider, very briefly, the relationship of the tribe Stipeae to the other tribes of grasses for this might help clarify the origins of the tribe Stipeae and the genus Stipa itself. First of all, one must note that the Stipeae were, for a long time, included with those genera of the Aveneae which have single flowered spikelets (the tribe Agrostideae of many authors) and with a few closely related genera which, at the present time, make up the tribe Aristideae. There was some rationale for this treatment for it is quite common to have single flowered spikelets as a result of reduction from ancestral species with many flowers per spikelet. However, in other respects, the evolution of these three groups has been quite distinct. The evolution of the Agrostideae (genera Agrostis L., Calamagrostis Adans. etc.) has been for a long time, probably since the PalEocene, into relatively high mountains from mesic (damp) environments. Apparently this evolution was quite slow and uniform since the extant genera of this tribe still exhibit the primitive festucoid features of its ancestors and in no way, other than the presence of the single floret, do its members differ from the tribes Aveneae and Poeae with their many-flowered spikelets. The taxa which gave rise to the contemporary genera of the tribe Aristideae, on the other hand, evolved very early, possibly at the end of the Cretaceous or the beginning of the PalEocene, by way of xerophyllization and, specifically, the so-called eragrostoid and panicoid grasses acquired cooperative photosynthesis which is very efficient in arid climates (Karpilov 1970) and the associated aristoid and eragrostoid leaf anatomy with it. The Aristideae is mainly a Gondwanaland tribe and its possible ancestors may have been extinct representatives of the tribe Danthonieae which is widespread primarily in the southern Hemisphere. Basically, the xerophyllic tribe Stipeae , in comparison with the Aristideae, completely retained the typical festucoid anatomy (Carodil et al. 1973) and the normal type of photosynthesis. S. A. Nevsky (1937, p. 223) considered it possible to derive the Stipeae from the Danthonieae, forgetting however, the fact that the Stipeae are, in some respects, more primitive than the Danthonieae (a manifestation of heterobathmy) and for that reason, could not be derived from them. For instance, the majority of the Stipeae  have 3 well-developed lodicules whereas the Danthonieae have 2 rather poorly developed lodicules. In addition, many Stipeae have a rather long epiblast on the embryo (which is sometimes even divided at the apex (Worsdell 1916) that in our opinion (Tsvelev 1975b) also is a very primitive feature. These primitive features distinguish Stipeae from their other candidate for ancestors, the Aveneae. It is true that recently 3 lodicules have been detected in the genus of the Aveneae that was recently distinguished from Helictotrichon Bess., Danthoniastrum (Holub) Holub, the only species of which,  D. compactum (Boiss. et Heldr.) Holub, occurs on limestone rocks in the Balkan mountains and the western Caucasus.

 In comparison with the majority of other genera of Aveneae, this genus has deeply two lobed lemmas with a well-developed awn coming from the gap between the lobes. This last feature is characteristic of Danthonieae and, according to Nevsky (1937), was also peculiar to the ancestors of the Stipeae . Apparently, the primitive Aveneae, which would have been similar to Danthoniastrum, were closer to the ancestors of the Stipeae . However, such ancestors could only be the most primitive forms which have 3 well-developed lodicules, a long epiblast, and probably a number of other primitive features peculiar to the earliest grasses. Other characteristics of Stipeae which distinguish them from Aveneae (including Agrostideae), are examples of specialization connected with xeromorphogenesis. These consist of generally smaller chromosomes, smaller and almost spherical pollen grains (Solntseva 1967), frequent transitions to facultative or even obligate cleistogamy, the structure of the lemmas and paleas which ensure protection of the flowers from unfavorable conditions (Yakovlev, Solntseva 1965). The even smaller chromosomes, chiefly peculiar to the tropical tribes of grasses, are often considered more primitive than the larger chromosomes that are characteristic of the festucoid grasses. However, we consider the reduction in size of the chromosomes, like the reduction in size of the pollen grains, an indication of specialization, the result of xeromorphogenesis, which furthers the reduction in size of the florets and the structures in them (Tsvelev 1975a).

In this way, the anatomical-morphological peculiarities of Stipeae undoubtedly indicate its significantly greater proximity to the so-called festucoid group of grasses, most richly produced in the extra-tropical regions of both hemispheres. Starting from here, it is possible to assume that the ancestors of the Stipeae, like the ancestors of other festucoid grasses, evolved for a long time in relatively high mountains and were mesophyti c plants (Tsvelev 1975a). In the lower mountain areas and on the plains they apparently evolved later than the bambusoid, oryzoid, arundinoid, eragrostoid, and panicoid grasses. This probably explains the absence of cooperative photosynthesis (so advantageous in arid regions); although to assume the paraphyletic origin of grasses does not account for the fact that not all early grasses could have evolved cooperative photosynthesis.

At the same time, the presence in Stipeae of the aforementioned primitive characteristics speak to the  fact that their ancestors began to adapt to the low mountains and plains earlier than the ancestors of the other festucoid tribes of grasses, possibly even as early as mid-palEocene. The initial evolution of the Stipeae , apparently went very fast and unevenly, resulting in the early and very high specialization of distinct taxa together with very pronounced heterobathmy. Since the majority of the characteristics of the Stipeae are probably the result of xerophyllization it is possible to think that the formation of the early Stipeae , similar in appearance to most of the primitive species of Achnatherum, occurred in conditions of a drying climate that possessed some seasonality earlier than the savanna or savanna- like associations.

Both during the descent into the plains and during the later migrations of the ancestors of the Stipeae and the earliest Stipeae, the species hybridized with each other, forming a very large collection of new species and genera that were less specialized than their progenitors because of this hybridization (Tsvelev 1975b). Diploid ancestors of the feathergrasses with 2n=10, 12, and 14 are now almost non-existent although in another, more youthful tribe of festucoid grasses, the Triticeae, the ancestors of several large allopolyploid genera are also extant (Tsvelev 1975b). We earlier concluded (Chromosome numbers of flowering plants, 1969) that all contemporary Stipeae are of hybrid origin. Diploid species of the tribe could have become extinct, leaving a niche for less specialized (because of hybridization) descendants. But it is also possible that they did not exist, that the whole tribe had a hybrid origin.

Moving on to the question of evolution Stipa and the comparatively more primitive Achnatherum and Ptilagrostis which are closely allied to it, one must consider, in greater detail, the geographic distribution and ecological characteristics of these genera which, for convenience, we will call feathergrasses regardless of their generic affiliation. It is convenient to do this by continental blocks, beginning with the richest in feathergrass, Eurasia.

The richest center is near, middle, and part of central Asia [1] , including in the latter other mountain ranges of inner China, where there is the greatest variety of feathergrasses both in number of species and number of section (8 sections of Stipa, 4 of Achnatherum, and Ptilagrostis). Somewhat poorer in feathergrasses are the Mediterranean countries were only 4 sections of Stipa and 2 species of Achnatherum, belonging to 2 different sections, grow.  To the north of the ancient Mediterranean at the limits of the forest zone, feathergrasses are preserved only in a few locations as relicts of postglacial xerothermic periods. It is interesting to note that in the western portion of Eurasia the very northern most locations of feathergrasses, on the island of Gotland and along the river Och, belong the plumose feathergrasses of sect. Stipa, S. pennata L. Both the plumose feathergrasses of sect. Stipa and the glabrous awned species of sect. Leiostipa extend as far as the Northern Urals, but in Siberia the plumose feathergrasses fade out; only the glabrous-awned species are found at the northern limits of the genus in the central Yakut. There is also in Yakut, extending almost to the tundra, one of the species of the cryophyl lous genus Ptilagrostis, P. monghol ica (Turcx. ex Trin.) Griseb., but in central Yakut and Kamchatka Achnatherum confusum is found.

The plumose feathergrasses of sect. Stipa, which represent the most evolved species in the genus, in general are more common in the more western, less continental region of Eurasia. This is explained in part by the significantly larger evaporative surface of their awns. Probably for the same reason, they set seed significantly earlier than the glabrous-awned species, doing so while there is still considerable soil moisture. In the east, they extend as far as Pribaikal and Hangai, being found there, however, only in some isolated locations. The feathergrasses of sect. Leiostipa, on the other hand, are less frequent in western Europe, including the Medi terranean, than the plumose feathergrasses.

South of the mountain regions of central Asia, in tropical Industan, Indonesia, and Indochina, feather grasses, like other festucoid graases, are non-existent except in maritime regions of eastern Asia where there are a few species of Achnatherum which is more primitive than Stipa.

It is hardly possible to take the region of greatest species diversity of feathergrasses as being the area of their center of origin. Indeed, they have had a very long period of evolution and may have originated in regions where there are now very few representatives, e.g., the territory of Angarhad, or even where they are now completely absent. It is only possible to assume that in the mountainous regions surrounding the Tethys, especially those on the east, evolution of the feathergrasses proceeded particularly rapidly. It is possible to infer the features of the earliest feathergrasses from those of the extant, relatively mesophyllous, species of Achnatherum. They probably had flat, but sufficiently stiff, leaf blades, small but numerous spikelets, hairy lemmas with weakly bent, but not deciduous, scabrous awns, and short, pointed, glumes. The transition from Achnatherum to Stipa was accomplished through specialization of the callus and adaptation of the awns to zoochory. Evolution of the feathergrasses proceeded simultaneously towards the elongation of the spikelet and all its parts, this compensating usually for the reduction in number of spikelets in the panicle. The lengthening of the awns was usually correlated with the elongation of the glume apices which, in species with long-awned lemmas, was more effective in preventing premature dropping of the floret (?).

The alpine orogenies and the drying up of the eastern part of the Tethys at the palEocene-Eocene boundary probably played a major role in the subsequent evolution of the Eurasian feathergrasses. The early species of feathergrasses were raised by mountain formation to great elevations; they then adapted to the niches formed, including rocks and talus. Apparently on rocky and stony mountain slopes the transition from zoochory to anemochory began in some species by the transformation of the scabrous trichomes of the awn to short hairs. The longest-awned feathergrasses of sect. Leiostipa also became partly transformed to anemochory which was especially advantageous on rocky mountain slopes because they are avoided by large animals. Eventually the diaspores of many smooth-awned feathergrasses became adapted to dispersal by both large animals and the wind. Probably during the orogeny of the high mountains the cryophyllic genus Ptilagrostis evolved from some relatively weakly specialized species of Achnatherum via the transformation of the scabrous trichomes to short hairs. The substitution of zoochory by anemochory in the treeless regions of high mountains also proved to be a valuable evolutionary achievement.

Even in humid climates the evolution of petrophyl1ous plants usually leads to the evolution of new xeromorphic structures. Thus it is understandable that the accumulated xerophyllic adaptations of the feathergrasses pre-adapted them to rocky places and, in the miocene and Pliocene, enabled them to become established in arid and semiarid regions of the plains and foothills. The smooth-awned feathergrasses of sect. Leiostipa (S. capillata L., S. sareptana A. Beck, S. holosericea Trin. etc.) retained the zoochoric mode of dispersal, stopped earlier in their development than the feather grasses of sect. Stipa and Smirnovia Tsvelev which include many obligately anemochoric species with long-hairy awns.

We think that at the very least the majority of feathergrasses of the Eurasian steppes were formed during the alpine orogeny although it is impossible to exclude completely the possibility of the origin of several species by way of hybridization when the earliest mountain and steppe species were brought together. Thus the sandy race of the typical feathergrass, S. pennata subsp. sabulosa (Pacz.) Tsvelev, probably evolved after the descent of the common ancestor, S. pennata s.l. to the plains, although it is found on granite outcrops in the foothills.

Species of the two principal Eurasian sections of feathergrass - Stipa and Leiostipa - developed basically independently and in parallel. However, the second of these is more primitive and was formed significantly earlier than the alpine orogeny. Its first species probably had short spikelets and relatively short awns similar to the extant species S. bungeana Trin. and S. richteriana Kar. & Kir. It is entirely possible that long-awned species of - Leiostipa (S. capillata etc.) evolved only at the time of the alpine orogeny, together with the plumose feathergrasses of section Stipa, although such long- awned species as S. pellita (Trin. & Rupr.) Tsvelev may have evolved on the sandy shores of theTethys Sea during the Oligocene. Already among the earlier small-spikeleted and short-awned species of sect. Leiostipa were some that made the transition to feathergrasses of sect. Barbatae Junge, which has short-haired awns on moderately small lemmas which were often hairy over the entire surface. The section Barbatae is also certainly more ancient than sect. Stipa, and many of its species evolved in the highland areas of the ancient Mediterranean, again up to the beginning of the alpine orogeny.

It is interesting that the transition from scabridities on the awn to hairs, as in sect. Leiostipaa, occurred in two ways: 1) starting from the upper parts of the awn (this lead to the formation of typical Barbatae feathergrasses, and also to more advanced feathergrasses of sect. Stipa and 2) starting from the base of the awn (in this way went possibly the Eurasian feathergrasses, S. holosericea, S. richteriana, and others (which merit segregation in a separate section and also, the majority of South American and Australian feathergrasses). The small, ancient Mediterranean sect. Stipella Tsvelev, type S. capensis Thunb. (= S. tortilis Desf.) probably arose as far back as the Oligocene as a result of ephemerization of feathergrasses with awns that have hairs on the lower portion. Although S. capensis was described from South Africa, where it is very common, it can hardly be doubted that it was carried there by the first colonists. Also, one annual species, S. annua Mez, was described from South America, but it is doubtful that it belongs to sect. Stipella unless it is alien to South America.

The majority of species of the ancient-Mediterranean sect. Barbatae extend into near and Middle Asia. But in Central Asia there are three very distinct species in this section: S. breviflora Griseb., S. orientalis Trin., and S. purpurea Griseb. The last of these species, which extends into the highest mountains of Tibet and regions around this mountain system, probably merits assignment to a unique, monotypic, section.

Section Smirnovia Tsvelev (Tsvelev 1974a, c.20), with S. caucasica Schmalh., S. lingua Junge and others, occupies a central position among feathergrasses. It is characterized by a once-geniculate awn and leaf with a short ligule which changes from the base into a thick row of hairs. This last character is not characteristic of other genera of Stipeae and generally is rare among festucoid grasses but it is very common among arundinoid and ergrostoid grasses of the tribes Danthonieae, Aristideae, and Cynodonteae. It is undoubtedly a sign of great specialization (the result of xeromorphogenesis), such that sect. Smirnovia must be acknowledged to be still more specialized than sect. Stipa. Its species extend primarily into middle and Central Asia but they occur also in southern Siberia, in near Asia and in the Caucasus. They all have obligate anemochory and the feathery awns of someof them (for example, S. longiplumosa Roshev.) are no shorter than those of species from sect. Stipa. One can suppose that the first species of this section evolved on islands in the eastern part of the Tethys up until the alpine orogeny.

 For the territory of the Soviet Union, I (Tsvelev 1974a) recognized 71 species of Stipa, Achnatherum, and Ptilagrostis, and for Central Asia - 46 species (Tsvelev 1968), and for the USA, as is well known, 36 (Hitchcock 1951). This last number does not include the species of Mexico, an area that is very rich in endemics, nor South American species, so that it is possible to deduce that North America is not poorer in feathergrasses, relatively, than Eurasia. However, the evolution of feathergrasses in North America has not advanced as far as in Eurasia, (obviously, a consequence of the lesser aridity of North America during the Eocene) and the most specialized sections, sects. Stipa and Smirnovia are not present in North America. Among the extant feathergrasses the most prevalent are species of section Leiostipa, whose short awns do not reach such great lengths as in Eurasia. There is one typical representative of the ancient Mediterranean sect. Barbatae present in North America , S. neomexicana (Thunb.) Scribner which is relatively close to the Eurasian members of the section such as S. barbata Desf. and S. armeniaca P. Smirn. Stipa neomexicana grows on stony and rocky slopes in the southern part of the northern Cordillera of the USA and Mexico. It is unlikely that it was carried here from Europe and later transformed in place. However, it is possible that its development in North America paralleled the development of the European species from distant ancestors of sect. Barbatae, which inhabited the coasts of the Tethys sea up until the formation of the Atlantic Ocean.

Several of the feathergrasses of the USA and Mexico, e.g., S. speciosa Trin. & Rupr., are closer to the South American than the Eurasian feathergrasses, and belong to sections not represented in Eurasia. The genus Achnatherum is represented in North America by several species, as it is in eastern Asia, which are not always clearly distinct from Stipa. Thus, the distinctive S. tenuissima Trin. (occurring in the southern USA, Mexico, and Argentina) has lemmas 2-3 mm long with greatly overlapping edges (as in NasselIa and Stipa) but with a callus as in Achnatherum. The apices of its glumesare strongly elongated as in species of Stipa. Probably this species forms an independent monotypic section which may belong to either Stipa or Achnatherum. The cryophyllic genus Ptilagrostis is represented i n North America by one species P. porteri  (Rydb.) Tsvelev, which inhabits high mountains sites in Colorado oand is very close to the eastern Siberian P. alpina (Fr. Schmidt). Sipl.

We believe that the evolution of the feathergrasses in North America proceeded simultaneously and in parallel to their evolution in Eurasia, although they did not attain such a high degree of specialization in North America as in Eurasia. This assumption is confirmed by the almost simultaneous formation in the Miocene of both the steppes of Eurasia and the prairies of North America, in both of which phytocoenoses several feathergrasses played a prominent role. We do not see sufficient reason for assuming that the feathergrasses migrated from Asia to North America across the Bering Sea. The flora of northeastern Eurasia and western North America was already very similar at the beginning of the Paleozoic and apparently developed in parallel, with a very low possibility that species exchange occurred.

South America is very rich in feathergrasses. Even in 1925, 89 species were known (Hitchcock 1925). Recently 43 species were listed for the province of Mendoza in Argentina (Roiz 1964) and 29 for the province of Buenos Aires (Cabrera, Torres, 1968). The diversity is quite high, but is quite different from what is found in Eurasia. For instance, in South America, the sects. Stipa, Barbatae, and Smirnovia are completely absent as is the long-awned section Leiostipa, but there is a group of species Ptilostipa consisting of the special section in which, as in Smirnovia, once geniculate awns up to 15 cm long and ligules of short hairs are found. Another group of species, Pappostip a, possesses relatively short, once- geniculate, awns which are long-haired in the lower twisted part but scabrous in the upper portion. feathergrasses of this group, represented in North America by S. speciosa Trin. & Rupr., resemble species of the Eurasian section Pseudoptilagrostis Tsvelev but undoubtedly merit their own section. Several other groups of species also merit sectional rank, having at the base of the awn a 'crown', a concave platform with the joint inside. In the original species ('Jarava') isolated in its own genus Jarava Ruiz & Pavon, the lemma has a tuft of long hairs at the apex and there is a quite short, scabrid awn. The genus Achnatherum is here likewise represented by several sections, a few of which possess a crown at the base of the awn. Taking into account the great variety of South American feathergrasses which have become adapted to the basically extra-tropical regions of this continent, we do not consider it possible to accept the hypothesis of their penetration into South America from the north as with other genera widespread in the extra- tropical regions of the northern hemisphere  e.g., Festuca. As in other analogous situations (Tsvelev 1971) we would rather think rather that the evolution of South American feathergrasses proceeded in parallel with and simultaneously with the evolution of feathergrasses in the northern hemisphere. The abundance of feathergrasses in the southern part of South America, like that of other festucoid grasses, is easily explained by the presence of high mountains from which they can migrate to the plains as the climate deteriorates and the plants with higher temperature requirements become extinct. We should still note that the final connection of South America with North America by the Panama isthmus occurred, according to recent information (Clayton 1975) in the Pliocene (until recently it was assumed there was a transitory connection between the Americas by the Paleocene), and to think that then the feathergrasses not only moved through the closed tropical forests of the Panama Isthmus, and that this would give such a diversity of species as is found in extra tropical South America is not reasonable. There is far greater reason to suppose that, after the union of North and South America in the Pliocene, several species from the South American sections of grasses penetrated into Mexico and southern U.S.A.

It is still more difficult to accept the possibility of a migration of feathergrasses from north to south from Eurasia to Australia; the latter has around 10 endemic sections of Stipa which are distributed primarily in the south and on Tasmania (Townrow 1970). It would be easier to accept the migration of feathergrasses into Australia from the south of South America through Antarctica but the Australian feathergrasses are clearly closer to the Eurasian species of Leiostipa (S. holosericea, and S. richteriana), Stipel la (S. capensis) an Regelia (S. regeliana Hack.) than the sections of South American feathergrasses.

Concerning Africa (excluding the part of it that lies along the Mediterranean Sea where many Mediterranean species of feathergrass (S. barbata, S. pellita etc. are encountered) is poorer in feathergrasses than Australia, let alone South America. In South Africa, excluding the introduced S. capensis, the feather grasses are represented by only one species Achnatherum capense (Nees) Nevski (= S. dregeana Steud.) This species belongs to sect. Aristel la (Trin.) Tsvelev and is very close to the near Asian species Achnatherum longearistatum (Boiss. et Hausskn.) Nevski and A. turcomanicum (Roshev.) Tsvelev and is represented here by 2 subspecies of which the type subspecies is widespread in the low mountains but the comparatively primitive subspecies A. capense subsp. elongata (Nees) Tsvelev comb. nova (= Lasiagrositis elongata Nees 1941, F1. Afr. Austr. :168) is in the higher mountain regions of South Africa (in part on the Drakon mtns). In addition, disjunct populations of the latter subspecies are found in the mountains of Kenya and Tanganyika where they grow in evergreen xerophyllic forests and savannas at elevations of 2000-2500 m (Clayton 1970, p.115).

Achnatherum capense is one of the least specialized extant species of Stipeae . It has extravaginal shoots, rather wide, flat leaf blades without projecting ribs and large panicles with numerous spikelets. In this case it is reasonable to consider that it had its origin in some common ancestor of the sect. Aristella which migrated to South Africa from the near Asia along the mountain systems, although this seems to me unlikely. In addition to A. capense s.l., there is an endemic feathergrass in the mountains of tropical-Africa, in Uganda, S. tigrensis Chiov. which belongs to sect. Leiostipa. It is very close to S. nitens Ball, a widespread species of the Atlas mountains, and also relatively close to the old Mediterranean species S. holosericea s.l. Based on the structure of the awns (which are shorter than those of s. holosericea), these species are also close to S. capensis and several Australian species of Stipa.

We see the explanation for the relative paucity of feathergrasses in South Africa not in the absence of the possibility of their migration from the north (they were greater than in the case of Australia and South America), but in the absence of high mountains which would provide high mountain festucoid grasses with a path to the neighboring plains during deteriorating and increasingly arid conditions. Moreover, South Africa occupied the central position in Gondwanaland until its division into separate continental blocks and apparently arid conditions were already predominant on it.

If we accept the possibility of the prolonged development, in parallel, of the ancestors of the contemporary taxa in various territories, then these circumstances may explain the paucity of South African festucoid grasses and also of many other relatively moisture loving families of angiosperms. It is entirely possible that South Africa underwent repeated drying periods in much later times also, even during the Pleistocene. The majority of festucoid grasses living here could not endure these xerothermic conditions because of the absence of adequate refugia in high mountains.

 Thus, proceeding from contemporary geographical distribution of feathergrasses (Stipa) and the genera of the Stipeae closest to it, we can deduce their most probable origin and subsequent, more or less parallel development. From these data, it is probable that all extant feathergrasses are hybrids which have stabilised, having arisen not only as a result of closely related but also distantly related hybridization of species with various number of chromosomes. The first feather grasses apparently lived in savanna like habitats or were dead ends, already possessing xeromorphic characters i.e., were as if preadapted to subsequent separation at first onto the upland steppes, lowland steppes, prairies and pampas which were formed in the Miocene. The evolution of the feathergrasses from the very beginning proceeded very rapidly so they attained a high degree of specialization and an almost complete cessation of further evolution, except for evolution by hybridization. The occurrence of heterobathmy in feathergrasses is evidence for their rapid evolution.

Although the evolution of the feathergrasses proceeded more or less in parallel, it came to an end on various continents at very different levels. The highest evolutionary level was attained by the feathergrasses of Eurasia. Relatively close to this level are the feathergrasses of South America where there was a secondary center of speciation. The evolution of the feather grasses of South Africa came to an end at the lowest, most primitive level, the level of the genus Achnatherum.

 

Translation by K. Gonzalez and M.E. Barkworth, Utah State University, Logan UT 84322-5210. Translation made available by M. Barkworth, Dept. of Biology, Utah State University, Logan UT 84322-5305. We do not guarantee the accuracy of the translation although we are sure that, in general, it represents what Tsvelev wrote. Editing this translation in 2007 makes me even more appreciative Tsvelev's insight and comments than I was before and I was very impressed before.