Cytotaxonomy of the Andropogoneae II. Subtribes Ischaeminae, Rottboelliinae, and the Maydeae

160 Cytologia 22

Robert P. Celarier

Department of Botany and Plant Pathology, Oklahoma A. & M. College, Stillwater, Oklahoma, U. S. A.

Received March 5, 1957

Introductory remarks, key to the subtribes, materials and methods, and the review of the subtribes Dinzeriinae and Saccharinae have been treated previously (Celarier, 1956b). Consequently, this report shall commence im mediately with the subtribes to be considered in this study. Subtribe III. Ischaenzinae Hack. Although the stout fused condition of the joints and pedicels marks this subtribe as rather more advanced than the Saccharinae, it almost always has two flowers, awned fertile spikelets and must be considered as rather primitive on morphological grounds. Pilger (1940) divides the subtribe into two groups, the Ischaenzininae and the Apludininae. In general this follows the plan of Stapf (1919), Bews (1929) and Keng (1939) and there is fair agreement among these workers concerning the materials considered in each division, although there are some differences regarding the status of various entries. The major exception is Keng's treatment in which Sehirna and Apluda are placed in the Euandro pogoneae and Arthraxon is included in the Ischaenzinae. Although Sehima and Apluda have been retained in the Ischaenzinae in this study, it has seemed advisable to follow Keng's example in regard to Arthraxon. It ap pears definitely to be related to Thelepogon in regard to the tuberculate or muricate lower glumes of the sessile spikelets and the cordate lanceolate leaves. Key to the genera of the Ischaeminae AA. Racemes several to many noded, not enclosed in a sheath (Ischaemi ninae). A. Racemes 3-many, digitate or nearly so. I. Rachis joints and pedicels more or less thickened, first glume of the sessile spikelet without tubercules or muricate, leaves not cordate...... Ischaemum subgen. Coelischaemum II. Rachis joints and pedicels slender, first glume of sessile spikelet tuberculata or muricate, leaves lanceolate cordate. 1. Sessile spikelets two flowered, lemma awned from between the sinus...... Thelepogon 1957 Cytotaxonomy of the Andropogoneae II 161

2. Sessile spikelets one flowered , lemma awned from the back or at base ...... Arthraxon B. Racemes binate, rachis joints and pedicels triangular ...... Ischaenzum Subgen. Euischaemum C. Racemes usually solitary. I. Pedicellate spikelets more or less reduced, glumes without wings. 1. Pedicillate spikelet neuter but almost as large as sessile. a. Joints and pedicels much thickened .. Ischaenzum Subgen. Digastriunz b. Joints and pedicels almost linear .....Sehima 2. Pedicellate spikelets reduced to the lower glume , much smaller than the sessile...... Kerriochloa II. Pedicellate spikelets male . 1. Upper glume of both spikelets with wing-like crest ...... Andropterum 2. Glumes not winged...... Pogonachne BB. Racemes with one node, three heteromorphic reduced spikelets contain ed in a sheath (Apludininae)...... Apluda Of the eight genera included in this subtribe only two, Ischaernunz and Arthraxon, have more than a few known species. With the exception of fschaenzum, which has 3-4 species in America, all are restricted to the Eastern Hemisphere with the greatest number of genera and species in South east Asia. Only four genera are known cytologically. Ischaemum L. This is an extremely large and complex genus with well over fifty species and distributed throughout the tropics of the world. Two subgenera, Euischaemum and Coelischaemum, are recognized by both Keng (1939) and Pilger (1940) and it appears entirely possible from Hubbard's report (1935) that a third group Digastriunz may be distinct. Several sections have been recognized and described in the subgenus Euischaemum (Pilger 1940). Subgenus Coelischaemum I. brachyatherum Fenzl. was available for this study from a single collection made in Southern Rhodesia. It was found to have twenty somatic chromosomes with ten bivalents at metaphase I (Fig. 1) and completely re gular meiotic behavior. I. glaucostachyum Stapf was studied by Gould (1956) from two South African collection. Both had 2n=20 but no report of the meiotic behavior was given. Subgenus Euischaemum I. rugosum Salisb. from a collection in Assam, India, was found to be 2n=18 with nine bivalents at diakinesis (Fig. 2) and metaphase I. It was completely regular throughout its meiotic divisions. A second accession of

Cytologia 22, 1957 11 162 R. P. Celarier Cytologia 22 this species came from a collection near Sao Paulo, Brazil. This was also found to have 18 somatic chromosomes with nine bivalents at metaphase I (Fig. 3) and regular meiotic behavior. I. tinzorense Kunth. was first studied by Bremer (1925) and was reported to have twenty somatic chromosomes. However, in this study, from materials collected at Taichung, Formosa, it was seen to be 2n=36. At diakinesis and metaphase I the normal condition was 18 bivalents (Fig. 4) but occasion ally two univalents were present. At anaphase and telophase I chromosome behavior was completely regular with 18: 18 distribution. I. ciliare Retz was studied from material collected at Turrialba, Costa Rica. This material yielded extremely poor preparations at diakinesis and metaphase I and it was only after considerable effort that fifteen cells were analyzed. In all of these the 2n number appeared to be 54, but only six cells had 27 bivalents. More frequently 2-4 univalents were seen (Fig. 5) and in one cell a quadrivalent was observed. I. diplopogon Hook. from the Aravalli Mts. of India was found to have a somatic number of forty. The meiotic divisions were completely regular throughout and there were consistently twenty bivalents at diakinesis (Fig. 6) and metaphase I (Fig. 7). I. arcuaturn Stapf from South Africa was reported by de Wet (1954) to have a somatic complement of twenty chromosomes. A more recent study (de Wet and Anderson 1956) has recorded 2n=50 indicating a polyploid series. No meiotic studies were made for either accession. I. guianense Kunth. was studied by Krishnaswamy (1941) and was found to be 2n=40. I. crassipes Thell. var. typicum was found by Moriya and Kondo (1950) to have 56 somatic chromosomes with 28 bivalents at metaphase I. However, an accession of this species, collected for this study by A. Moriya from Yakasidi, Japan, appears to be 2n=60. Although the chromosomes stained distinctly, their behavior was essentially regular, and their size was not ex ceedingly small, it was nevertheless difficult to determine the number with certainty. This seems to be principally due to the clumping of the chromo somes, especially at metaphase I, and, to a lesser extent, was the result of quadrivalent formation (Figs. 8, 9, 10). Because of this difficulty 75 cells were analyzed at diakinesis and meta phase I. Of these cells, four were found that had 28 bivalents and one quadrivalent, and 15 others were probably of this constitution (Fig. 8). Fifteen cells had 28 configurations (Fig. 9) and it appears likely that most of these were cases of two groups of two cells lying together. The remain ing 41 cells all seemed definitely to have 30 bivalents. Anaphase and telophase I were essentially regular, but occasionally two chromosome pairs appeared to be late in separating, and in one cell a bridge, without a fragment, was seen. It seems probable that these irregularities 1957 Cytotaxonomy of the Andropogoneae II 163

Figs. 1-12. Chromosome behavior in the Ischaeminae. 1350•~. 1, metaphase I in Ischaemum brachyatherum with ten bivalents. 2, diakinesis of I. rugosum from Assam with nine bivalents (chromosomes traced in India ink and the photograph bleached with K3 Fe (CN)6). 3, metaphase I of I. rugosum from Brazil showing nine bivalents. 4, diakinesis of I. timorense showing 18 bivalents. 5, metaphase I in I. ciliare with two univalents. 6, diaki nesis in I. diplopogon with 20 bivalents (two are very small). 7, metaphase I of I. diplo pogon with 20 bivalents (two lying together at arrow). 8, diakinesis of I. crassipes with 28 bivalents and one quadrivalent (arrow at quadrivalent). 9, prometaphase I of I. cras sipes with 28 configurations. 10, diakinesis in I. crassipes with thirty bivalents. 11, anaphase I in Sehima nervosum from India showing several lagging chromosomes. 12,

metaphases I of Arthraxon hispidus with 18 bivalents, 164 R. P. Celarier Cytologia 22

were due to the slower separation of the occasional quadrivalent. A detailed study of several accessions of this species may prove fruitful but with the present information it appears that this species is 2n=60 and has a basic number of ten (or five) rather than seven as suggested by Moriya and Kondo (1950). I. anthephoroides Miq. was studied by Kuwada (1915) and reported to have a somatic number of 68. A more recent study of this species (Tateoka, 1955) has recorded 72 somatic chromosomes and the author suggested aneu ploidy in the species. No meiotic studies have been reported and the species definitely needs a more detailed analysis. Although only ten species of the genus have been studied cytologically, it is already obvious that at least two basic numbers, nine and ten, are known and it is at least possible that another, seven, exists. It is very likely that this large genus may indicate more than one line of evolution and a detailed analysis of the morphological and cytological variation in the group should prove extremely fruitful. Sehima Forsk. This is a small genus with only eight or nine described species. It appears to be closely related to Ischaemum and was considered only a section of Ischaemum by Hackel (Hackel 1889). It is widespread in the Old World tropics occurring from West Africa to Australia, and one species, S. nervo sum Stapf, covers most of this area. S. nervosum was first reported by Sampath and Ramanathan (1949) as having 34 somatic chromosomes. However, in a later study Mehra (1955) demonstrated that the species has a polyploid series with a basic number of ten. He studied one accession from Australia with 2n=20 and one from India with 2n=40. In this report two accessions are considered, one from Hyderabad, India end the other from Kenya, East Africa. Both were found to have forty somatic chromosomes and thus supports Mehra's data that ten (or five) is the basic number. In both accessions one quadrivalent is occasionally encountered, but the predominant configuration is twenty bivalents. Both univalents at diakinesis and metaphase I and lagging chromosomes at anaphase and telophase were frequently seen in the Indian accession but were absent in the African material. Arthraxon Beauv. The genus Arthraxon is considered by most authors to consist of from 20 to 25 species but they are , for the most part, poorly understood and a thorough and complete taxonomic revision is needed. The genus is rather abundant throughout the tropics and subtropics of the Old World with the center of species diversity in Southeast Asia and Indonesia. A. ciliaris R. B. was studied by Avdulov (1928) and was found to have a somatic chromosome number of 36. 1957 Cytotaxonomy of the Andropogoneae II 165

A. hispidus Merr., which may be different from A. hispidus (Thunb.) Makino (Henrard 1941), was also studied by Avdulov (1928) and found to be 2n=36. Recently Tateoka (1954) has recorded the same number for A. hispidus Makino. No meiotic studies were made on either of these species. A. hispidus (Thunb.) Makino. var. cryptatherus (Hack.) Honda, was studied from one entry that had become naturalized in Washington, D.C. It was found to have 36 somatic chromosomes and nearly regular meiotic be havior. Although 18 bivalents was by far the most common configuration seen at metaphase I (Fig. 12), univalents, usually two (Fig. 13), were found in approximately 40% of the cells and one configuration was seen that sug gested a quadrivalent. Apluda L. Only one species, A. mutica L., is generally recognized. However, it is extremely polymorphic with several named varieties and is widespread throughout the tropics of southern Asia, Australia, and the Pacific Islands. Two cytological types have been reported, a diploid with 2n=20 recorded by Hunter (1934) and a tetraploid with 2n=40 by Avdulov (1928). In the present study one entry from Dehra Dun, India, was studied and it was found to be 2n=20. At diakinesis and early metaphase I, ten biva lents were observed in all cells studied (Fig. 14). However, there was con siderable difference in . the time of separation of the bivalents at late metaphase and early anaphase (Figs. 15, 16), and occasionally it appeared that there was a variable number of univalents. A large number of cells was observed at telophase I and they were always normal with 10: 10 distribution. The second division was also normal so that the most likely explanation of these univalents is early terminalization of some bivalents. Subtribe IV. Rottboelliinae Hubb. Although very closely related to the subtribe Ischaeminae it is rather more advanced in many morphological features. In most cases the two flowered fertile spikelets of Ischaeminae have been reduced to one and the awn of the fertile lemma has been lost. The subtribe is usually divided into two groups. One, the Vossiininae, is well agreed on by Stapf, Bews, Keng and Pilger, but there is somewhat less agreement on the other, the Rottboelliininae. Although the overall contents are the same with all of these workers (except that Bews places Eremochloa in the Ischaeminae), there is much disagreement at the generic level. Thaumastochloa is considered as part of Ophiuros by Keng, Stapf, and Bews; two genera, Lasiurus and Coelorhachis, are separated from Rottboellia by Stapf and Bews, Hackelochloa is considered Manisuris by Bews, and there are many other differences that could be cited. In the following key I have been inclined to accept the divisions as genera even though it is obvious that most of these problems are far from solved. 166 R. P. Celarier Cytologia 22

Key to the genera of the Rottboelliinae AA. Racemes in espatheate inflorescence and scattered along a common axis, rarely reduced to a single raceme (Urelytrum) and then with glume of pedicellate spikelet awned. Both spikelets similar (Vossi ininae). A. First glume (both spikelets) acute or obtuse, joints and pedicels 3 -angled and somewhat club-shaped. I. Racemes digitate to solitary, spikelets two flowered ...... Phacelurus II. Racemes panicled, spikelets one flowered...... Thyrsia B. First glume of at least one spikelet awned and pointed, joints and pedicels round or flat. I. Spikelets two flowered, sessile spikelet sunken in rachis. 1. Lower glume of all spikelets long caudate, rachis usually digitate, sessile spikelets disarticulating tardily and horizontally, callus glabrous...... Vossia 2. Lower glume of sessile spikelets eucaudate, producing long awn in pedicellate spikelet, rachis solitary, sessile spikelet disarticulating obliquely, callus bearded...... Urelytrum II. Spikelets one flowered, sessile spikelets not sunken in rachis. 1. First glume smooth, joints and pedicels elongated, pedicels 2-3 jointed...... Pseudovossia 2. First glume muricate, joints and pedicels continuous ...... Jardinea BB. Racemes solitary and terminal on simple and sparsely branched culm or inclosed in a false spatheate panicle, both spikelets mostly dissimilar (Rottboelliininae). A. Spikelets paired, one sessile the other pedicellate. I. Pedicels free from internodes 1. Racemes more or less villous, never cylindrical. a. Sessile spikelets two flowered, villous throughout...... Lasiurus b. Sessile spikelets one flowered, cilliate on keel ...... Elyonurus 2. Racemes glabrous and cylindrical or compressed a. Sessile spikelets sunken in a thickened rachis. + Spikelets of each pair similar, two flowered ...... Chasinopodium ++ Pedicellate spikelets much reduced, one flowered...... Coelorhachis b. Sessile spikelets not sunken in rachis, rachis slender, first glume rugose transversely, two flowered, pedicellate spikelet 1967 Cytotaxonomy of the Androp ogoneae II 167

Figs. 13-23. Chromosome behavior in the Ischaeminae and the Rottboelliinae. 1350•~. 13,

metaphase I in Arthraxon hispidus with 17 bivalents and two univalents. 14, metaphase

I of Apluda mutica with ten bivalents. 15, early anaphase I of A. mutica showing diff

erences in time of separation of bivalents. 16, later anaphase of A. mutica with all

bivalents separated except one. 17, metaphase I of Lasiurus hirsutus with 28 bivalents.

18, diakinesis in L. hirsutus showing 28 bivalents. 19, metaphase I of Elyonurus argen

teus with five large bivalents. 20, diakinesis of Rottboellia exaltata from South Africa with ten bivalents. 22, diakinesis of R. exaltata from Kenya with ten bivalents. 22-23, metaphase and anaphase I of Hacklochloa granularis showing seven bivalents and 7: X distri bution of chromosomes (chromosomes traced with India ink and bleached with K3Fe(CN)6). 1 AR IZ. P. Celarier Cytologia 22

reduced, glumes of both spikelets frequently delicately awned...... Rhytachne II. Pedicels fused to the internodes of the rachis. 1. Pedicellate spikelets well developed and similar to sessile, rachis tenacious...... Hemarthria 2. Pedicellate spikelet more or less reduced, rachis fragile. a. Joints and pedicels free. + Sessile spikelets one to the node...... Rottboellia + + Sessile spikelets two to the node....Ratzeburgia b. Joints and pedicels more or less fused. + Sessile spikelets two to the node...... Mnesithea + + Sessile spikelets one to the node. Sessile spikelet globose, lower glume pitted and tubercled, fi rst glume not winged...... Hackelochloa ** Sessile spikelet not globose , lower glume not pitted, first plume winged...... Manisuris B. Spikelets solitary, sessile (pedicellate either entirely suppressed or rudimentary). I. Pedicels free from the internode of the rachis...Eremochloa II. Pedicels fused to the internodes of rachis. 1. Spikes cylindrical, lower floret male, and with a palea ...... Ophiuros 2. Spikes dorsiventral, lower floret neuter, without palea ...... Thaumastochloa All of the genera in this subtribe are rather small, only Elyonurus has as many as twenty species and all others except Coelorhachis have fewer than ten. The distribution is general throughout the tropics with the greatest diver sity, both in genera and species, being found in tropical Africa and South east Asia. The group Vossiininae has no representatives in the Western Hemisphere. There is very little cytological information on any of these genera, only two of the Vossiininae are known, and seven genera in the Rottboelliininae have been studied.

Vossiininae Phacelurus Griseb. It is a very small genus with only four described species, all of which are apparently rare. Nevertheless it has a rather wide distribution from South ern Rhodesia, across Asia Minor, through the Himalayas and into China and Japan. P. latifolius Ohwi has been studied by Tateoka (1955) from a collection 1957 Cytotaxonnmy of the Andropogoneae II 169 at Ito, Japan. It was found to have forty somatic chromosomes but no meiotic studies were made. Urelytrum Hack. This is another small genus with some six or seven species all of which are restricted to Africa. U. squarrosum Hack. was studied from a South African collection by de Wet and Anderson (1956) and was recorded as 2n=20 .

Rotfboelliininae Lasiurus Boissier The taxonomic position of this material has long been a matter of dis pute and the species L. hirsutus has been placed by different taxonomist in Ischaemum, Rottboellia, Elyonurus and Coelorhachis. Although it is now rather generally regarded as a distinct genus it was considered to be only a section of Rottboellia by Pilger (1940) in the most recent monographic treatment. Although considered monotypic until recently there are now three des cribed species. They are well adapted to desert conditions and are distributed through the desert regions of Africa, Arabia, and India. L. hirsutus Boiss. was studied from one entry that was originally col lected in western Sahara. It was found to have a somatic number of 56 and regular meiotic behavior. Thirty cells were analyzed in detail at diaki nesis and metaphase I and all had 28 bivalents (Figs. 17, 18). At anaphase and telophase I only 28: 28 distribution of the chromosomes was observed and the second division was entirely regular. Elyonurus Humb. and Bonpl. This is the largest genus of the subtribe with more than twenty species. Although found in all the tropical regions of the world, it is apparently a much more important element of the tropical sevannas and plains of the Americas than of the Old World. E. argenteus Nees., and important weedy grass of the South African veld, has been studied and found to have ten somatic chromosomes(Celarier, 1957). Since a detailed description of the chromosomesof this species has already been presented, it suffices here to say that it represents a second genus in the Andropogoneae with a haploid number of five and represents the largest chromosomes so far encountered in the tribe (Fig. 19). The meiotic behavior is completely regular. Two other species, both American, are known cytologically. E. barbiculmis Hack. and E. tripsacoides Humb. and Bonpl. from south Texas were studied by Brown (1951) and both were found to be 2n=20. Coelorhachis Brogn. There are probably more than fifteen species in this genus and they are 170 R. P. Celarier Cytologia 22 found in the tropics of both hemispheres. Several species extend into the warm temperate regions of North America. C. cylindrica Nash. was first studied by Reeves and Mangelsdorf (1935) and they reported a 2n number of 18 with nine bivalents at metaphase I. This number was later confirmed by Brown (1951), and more recently by Gould (1956). C. glandulosa Stapf and Ridley was found by Avdulov (1931) to be 2n =54 , which he considered a hexaploid with a basic number of nine. Hemarthria R. Br. This genus contains some eight or nine species most of which are located in tropical Africa but one is found as far east as Australia. Two species have been studied cytologically. H. altissima Stapf et Hubb. from South Africa was reported by de Wet (1954) to have a 2n number of 20. No meiotic studies were made. H. sibirica Ohwi was studied by Moriya and Kondo (1950), as Rottboellia japonica, and was found to be 2n=18 with nine bivalents at metaphase I. According to the figure given by these authors the chromo somes appear to be rather large. This number has been confirmed by Ono and Tateoka (1953). Rottboellia L. Although a small genus with only 3-4 species, at least one, R. exaltata, is extremely polymorphic and widely distributed throughout the tropics. R. exaltata L. was studied from a South Indian collection by Krishnas wamy et al (1954) and recorded as having 36 somatic chromosome. No studies were made of the meiotic divisions. For the present study three ac cessions were available, two from South Africa and one from East Africa. Morphologically two distinct types were represented. One was a rather small erect plant represented by both South African accessions, and the other was a large, sprawling, decumbent plant. In all three accessions the somatic number was 20 and ten bivalents were consistently found at diakinesis and metaphase I (Figs. 20, 21). No irregularities were seen at anaphase or telophase. Since identification of this material is somewhat difficult, and since the author has not seen Krishnaswamy's material, it is not possible to state with certainty that the species reported on here is similar to the one studied by Krishnaswamy et. al. (1954). However, it is entirely possible that we have here a species with two basic numbers, five or ten and nine. A similar con dition has been encountered previously (Celarier, 1956c) but here also there is some question in regards to the identification of the species. Hackelochloa O. Ktze. A monotypic annual genus that is a widespread weed throughout the tropics. H. granularis O. Ktze. from a collection at Nairobi, Kenya, was studied. 1957 Cytotaxonomy of the Anlropogoneae II 171

Although the plants flowered profusely very little but material was taken at the proper stage. Only two slides were available for study and neither of these were good. It was obvious from these studies that the meiotic behavior was regular and that the chromosome number was small . However, it was only with difficulty that the number was ascertained and there still remains some doubt. Most cells appear to have seven bivalents at diakinesis and metaphase I (Fig. 14) and in those cases where more are indicated it seems probable that it is due to a difference in time of terminalization of the bivalents. The few cells seen at anaphase I appear regular with 7:7 distribution of the chromosomes (Fig. 15). Eremochloa Buse A small genus with 8-10 species and distributed in Southeast Asia, Indonesia, and Australia. One species, E. ophuiroides, which has been introduced into south Texas from the Orient, was studied by Brown (1950) and found to have 18 somatic chromosomes.

Maydeae Matthieu Although a considerable amount of work has been done on some members of the Maydeae, there are others that are very poorly understood, and the precise phylogenetic position of the group is in no way solved. That the Maydeae is very closely related to the Andropogoneae is generally recognized (Randolph 1955), and it seems quite likely, from the point of view of ex perimental taxonomy, that they should be considered together. The fact that hybrids have been produced between Saccharum and both Zea and Coix (Celarier 1956b) is suggestive of this relationship, in spite of the promiscuous breeding behavior of Saccharum. Although very little work has been done by the author on the Maydeae, it is included in this report because of this obvious relationship to the Andropogoneae. (See Table 2 for review of the cytology of the group). Key to the Genera of the Maydeae A. Female spikelets without hardened fruit cases or with cases formed principally from the thickened rachis segments of the spike. I. Male spikelets in terminal panicles, female in a sheath inclosing spike ...... Zea II. Spikes containing female spikelets in lower part, male spikelets in upper part...... Tripsacum B. Female spikelets totally enclosed in ovoid to spherical fruit cases, formed principally from hardened leaf sheath...... Coix C. Female spikelets enclosed in hardened fruit cases formed chiefly from the first glume. 172 R. P. Celarier Cytologia 22

I. First glume of the female spikelet trilobate, hilum linear ...... Trilobachne II. First glume of the female spikelet not cleft, or only shortly two dentate, hilum not linear . 1. Lower margins of seed enclosing a cavity in which the hilum fits, hilum only evident below...... Polytoca 2. No cavity at base of seed, hilum not hidden by margins of caryopsis. a. Margins of lower glume overlapping, enclosing the rhachis...... Sclerachne b. Margins of lower glume not overlapping, rhachis visible over the whole length...... Chionachne Chionachne R. Br. This is a small genus with only 4-5 known species and distributed in tropical areas from India to Australia. One species, Ch. koenigii Thwaites is found throughout this area and it is the only species that has been thoroughly studied cytologically. It was first reported by Reeves and Mangelsdorf (1935) from an unspecified location and under the name Poly toca barbata Stapf. They found it to have twenty somatic chromosomes. This number was confirmed by Nirodi (1955) from materials collected at Coimbatore, India and the meiotic behavior was shown to be regular. Ch. semiteres was also found to be 2n=20 by Janaki-Ammal (1945). Sclerachne R. Br. Only one species, S. punctata R. Br., is known and it appears to be restricted in distribution to Indonesia. Cytologically it was found to be 2n =20 by Reeves and Mangelsdorf (1935) but no meiotic studies were made. Polytoca R. Br. Six species are generally recognized and like Chionachne the genus is distributed from India to Australia. Only one species, P. macrophylla Benth., has been studied cytologically. It was first reported by Avdulov (1931) as having forty somatic chromosomes and this number was confirmed by Simmonds (1954) who reported the meiotic behavior as regular. These results were confirmed by the recent work of Nirodi (1955). Coix L. This genus seems to consist of complexes of types in which only a few species are distinct. One species complex, the cultivated C. lachryma-jobi L., is extremely variable with many recognizable varieties, but with intro gression between most of them in certain locations. Five to six species are generally recognized and they are more or less restricted to southeast Asia, although C. lachryma-jobi has been introduced throughout the tropics of the world. Four species have been studied cytologically and diploids, tetraploids and octaploids have been found. Coix aquatica Roxb. with n=5 was the first 1957 Cytotaxonomy of the Andropogoneae II 173 Table 1. Cytology of Ischaeminae and Rottboelliinae 174 R. P. Cealrier Cytologia 22 Table 1. Cytology of Ischaeminae and Rottboelliinae (Continued) 1957 Cytotaxonomy of the Andropogoneae II 175 diploid studied (Reeves and Mangelsdorf 1935) and recently Nirodi (1955) has reported the same number for C. poilanei Mimeur (?). The meiotic behavior of the latter was analyzed in some detail and was shown to be re gular throughout. Many varieties of C. lachrynza-jobi, collected from much of its range of distribution, have been studied (Reeves and Mangelsdorf 1935; Nirodi 1955). All of these were tetraploids with n=10 and meiosis was regular in those entries studied. Tetraploids as well as octaploids have been found in C. gigantea Koen, ex Roxb. (janaki-Ammal 1945). Meiosis of the octaploid

Table 2. Cytology of the Maydeae 176 R. P. Celarier Cytologia 22 was reported by Nirodi (1955) who observed one quadrivalent rather frequently and two occasionally. Otherwise the behavior was regular. With the excellent background work of Watt (1904), Valleys (1948) and especially the recent report of Nirodi (1955) the genus is now sufficiently well known that a detailed study of species relationships should be undertaken. Tripsacum L. Nine species, eight of which have been studied cytologically, are generally recognized for the genus. Although distributed throughout the Americas, the center of greatest diversity seems to be in Mexico and Central America (Randolph and Hernandez - Xolocotzi 1950). All of the species studied cytologically have been found to have a basic number of nine but no diploids have yet been found. Four species, T. fl oridanunz Porter ex Vasey, T. austale Cutler and Ander., T. maizar Hern. and Rand., and T. zopilatense Hern. and Rand. have been found only as tetraploids with 2n=36; three species, T. laxum Nash, T. latifolium Nash, and T. pilosunz Scribn. and Mer., are known only as octaploids with 2n=72; and one species, T. dactyloides L., has both 36 and 72 chromosome races. Meiosis is normal or nearly so in the tetraploid species but is extremely irregular in those octaploid species in which it has been studied (Longley 1924; Mangelsdorf and Reeves 1939; Randolph 1955; Simmonds 1954). Randolph (1955) has suggested that 72 chromosome species are amphiploids of tetraploids such as T. nzaizar and T. zopilotense. This cross has been successfully made (Randolph and Hernandez-Xolocotzi 1950) but unfortu nately none of the hybrids have flowered and no meiotic studies have yet been made. There is some evidence of a polyploid complex at the octaploid level (see Randolph 1955 for discussion) and this is further complicated by apomixis, at least in T. dactyloides (Farquharson 1955). Although the cytological and morphological evidence strongly suggests that Tripsacum is very closely related to some members of the Rottboelliinae, the fact that it can successfully hybridize with Zea mays (Mangelsdorf and Reeves 1939; Randolph 1955) makes its relationship to that genus obvious. Zea L. It has been almost fifteen years since the proposal of Reeves and Mangelsdorf (1942) that the genus Euchlaena Schrad. should be transferred to Zea L. It appears to the author that this transfer was based on the very best experimental evidence and although recognized by some taxonomists (Rollins 1953) there has been a reluctance by many to accept this change. Recently Darlington (1956) has suggested that they should be treated not only as the same genus but as one species. Although the explanations given by these authors for this transfer are quite adequate, it has recently been pointed out very forcefully by Randolph (1955) that the linkage studies of many workers as well as detailed cytogentic studies demonstrated that the 1957 Cytotaxonomy of the Andropogoneae II 177 chromosomes of teosinte and maize are essentially similar . Also the fertility studies of Rogers (1950) on hybrids involving numerous types of both Z. mays and Z. mexicana Reeves and Mang. have further demonstrated their relationship. The contention of some workers that they are different genera because they are completely distinct in nature does not seem to be entirely valid, since F1 hybrids are commonly found in certain areas (Collins and Kempton 1920; Collins 1921; Reeves and Mangelsdorf 1942) and since there is a good possibility that there has been much introgression of teosinte genes in races of Z. nays. Wellhausen et al (1952), from a detailed study of some 2000 varieties collected throughout Mexico, were able to distinguish 25 distinct races. Of these 25 races, thirteen were believed to contain some teosinte germ plasm. If this thesis is correct and if the origin of the Corn Belt Dents is similar to that proposed by Anderson and Brown (1952) it is probable that some teosinte genes are present here also. Three species are usually recognized and although their natural habitats are somewhat obscured by their dependence on cultivation, it seems more than likely that the center of origin of the genus is tropical America and its distribution throughout the Old World has been post-Columbian. Two of the species, Z. mays and Z. mexicana, have been shown by numerous workers to have 20 somatic chromosomes and regular meiotic be havior. The third species, Z. perennis Reeves and Mang., has 40 somatic chromosomes and was first recorded as having 20 bivalents at diakinesis and metaphase I. (Longley 1924). However, in a later paper Longley (1934) states that he finds a prevalence of tetravalent chromosomes at diakinesis. This species has been crossed several times with both diploid and tetra ploid Z. mays and the triploid and tetraploid hybrids have been analyzed cytologically and to some extent genetically. In the triploid hybrids a rather low frequency of trivalents are found and in most of these it appears that one chromosome is loosely attached to a more or less normal bivalent. Most frequently there are ten bivalents, with or without the loosely associated chromosome (Longley 1924). The work of Emerson (1929) using marker genetic stocks suggest "that the two toesinte chromosomes more commonly disjoin (from the bivalents) and pass one to each pole, while the maize chromosome is distributed at random." This type of pairing is confirmed by the tetraploid hybrids in which meiosis is regular and pollen fertility high. Also, genetic studies in this cross indicate that the bivalent associations are mostly by autosyndesis (Collins and Longley, 1935). It would be of considerable interest to known exactly what the pairing relationships are between the chromosome of Z. perennis and Z. 7nexicana. There is probably no other plant species that has received such extensive Cytolozia22, 1957 12 178 R. P. Celarier Cytologia 22

cytogenetical study as has Z. mays, but it is beyond the scope of this report to do more than just mention that a very excellent and concise review of the subject has recently been published (Rhoades 1955).

Discussion and conclusions

The most striking feature revealed in this review is the very limited amount of cytological information that is available. Of the 28 genera con sidered in the Ischaeminae and Rottbolliinae only thirteen have been studied cytologically, and even in the very well known Maydeae one of the seven genera has not been studied. Also, it has been mentioned that several of the genera are in bad need of a detailed morphological study, and one of these Ischaenzum, appears to be very important phylogenetically. With these limitations it is most likely that many conclusions regarding relationships may require readjustment as more information become available. There appears to be little doubt but that the Ischaeminae with its two fl owered spikelets and awned fertile lemma is more primitive than the Rot tboelliinae and probably ancestral to it. It seems equally clear that the Ischaeminae is closely related to the Saccharinae at some points and was probably derived from it. The Maydeae appears to be the most advanced of the group considered and although its relationship to the Ischaeminae and Rottboelliinae is not entirely clear it seems probable that the Maydeae was derived from one or both of these subtribes. Certain species of the genus Ischaenzum are obviously related to Microste giunz of the Saccharinae and are probably the most primitive members of the subtribe, however, other of the species are rather advanced morphologically. The subgenus Coelischaemunz with numerous racemes is in several respects closely related to Microstegium and is undoubtably one of the pri mitive forms in the Ischaeminae. It seems quite likely that this section has differentiated in various directions and it is possibly ancestoral to several genera in both the Ischaeminae and the Rottboelliinae. It is possibly the progenitor of Thelepogon, which differs from it by a reduction in size of joints and pedicels, extensive leaf modification, suppression of pedicellate spikelet and modifications of the sessile spikelet. Arthraxon may well have been derived from Thelepogon principally by reduction from two to one fl owered spikelets. Also, it seems probable that the subgenus Euischaemum was derived from Coelischaemum, possibly through the section Fasciculata, with a reduction in the number of racemes to two, a trend which is terminated with solitary racemes in the subgenus Digastrium. It is also entirely possible that Phacelurus and Thyrsia of the Rottboelliinae are derived from Coelischaemunz. 1957 Cytotaxonomy of the Andropogoneae II 179

The following diagram summarizes this hypothesis .

Since so very little is known concerning the Rottboelliinae it seems somewhat futile to speculate further until more information is available. The Maydeae, on the other hand, are fairly well known and much speculation regarding phylogeny of the group and the relationship of the entries to one another has already been made (Mangelsdorf and Reeves, 1939; Weatherwax and Randolph 1955). It appears profitable at this time to separate the Maydeae into three groups. a) The genus Coix. b) Other Maydeae from Southeast Asia. c) The American Maydeae These divisions are based on both morphological (Weatherwax 1926; 1950) and cytological (Nirodi 1955) information but it should be mentioned that much further work is needed on the southeast Asian genera before their status will be secure. A beginning of a study of the relationship of these groups to one another 12* 180 R. P. Celarier Cytologia 22 has recently been made. Harada et al (1954) has successfully crossed Coix lachryma-jobi L. var. susutama Honda with Zea mays L. but only when the former is used as the female. This appears to be due to failure of Coix pollen to germinate. Cytological studies have not yet been made of this hybrid and the results of such studies are anxiously awaited and should prove of considerable importance in this problem. Also, it is quite possible (Weatherwax 1950) that these groups are more closely related to different members of the Andropogoneae than they are to one another and at least one author (Iltis 1911) has suggested a polyphyletic origin for the Maydeae. Three basic numbers, five (or ten), seven, and nine, are found in the Ischaeminae and the Rottboelliinae and two of these, five (or ten) and nine are also found in the Maydeae. The occurrence of species with n=5 in both the Rottboelliinae and the Maydeae is somewhat suggestive of an old chain of relationship and it is not at all unlikely that other species with this number will also be encountered as more studies are made. Also several secondary lines of evidence have been shown to support the hypothesis that five is an important basic number in the Andropogoneae (Celarier 1956 a). Of considerable interest is the fact that the two basic numbers common to Rottboelliinae and the Maydeae are found in the genus Ischaemum; and it is also possible that a third number, seven, occurs here. These factors plus the morphological complexities of the genus makes it quite likely that Ischaemum is of considerable phylogenetic importance and a really through study of this genus may illuminate many questions. Polyploidy is obviously of much importance in the evolution of these subtribes since it is found to occur in each of the three basic numbers. Although few species in these subtribes have been sampled throughout their geographical range of distribution, several species (Apluda mutica, Ischaemum arcuatum, Sehima nervosum, Coix gigantea and Tripsacum dactyloides) have been found with two chromosome races. The meiotic behavior of most of the polyploids is regular, suggesting an amphidiploid type of origin. How ever, in the two genera (Coix and Tripsacum) in which a somewhat detailed cytological and morphological study has been made there is some indication of a polyploid complex; and, at least in Tripsacum, apomixis is known to occur. The significance of these factors can only be determined after a much more through study has been made.

Summary

1. An artifical key to the genera of the subtribes Ischaeminae, Rot tboelliinae, and the Maydeae is given. 2. A review of the cytology of these subtribes, including chromosome number and meiotic behavior whenever available, is recorded, 1957 Cytotaxonomy of the Andropogoneae II 181

3. Of the 35 genera in these subtribes only 19 have been studied and two of these, Lasuirus and Hackelochloa, are reported here for the first time. 4. Three basic numbers, five (ten), seven and nine, have been found in each of the subtribes with polyploidy occurring in each of the basic number groups. 5. A brief discussion of the phylogenetic patterns of the subtribes it presented. Both cytological and morpholgical factors are considered but it is decided that it is still premature to make definite conclusions since the samplings of the materials in these subtribes are so inadequate. 6. It does, however, appear quite likely that the genus Ischaenzunz is of fundamental importance in this problem and that a thorough understanding of this genus may furnish a key to the overall problem.

Literature Cited

Anderson, E. and Brown, W. L. 1952. Origin of Corn Belt maize and its genetic significance. in Heterosis. Ames, Iowa. Avdulov, N. P. 1928. Systematic karyology of the family Gramineae. Proc. All-Russ Botanists Congress. pp. 65-67. - 1931. Karyo-systematische Untersuchung der Familie Gramineen. Bull. Appl. Bot., Genet., and Pl. Breeding. Suppl. 43. Bews, J. W. 1929. The World's Grasses. London. Bremer, G. 1925. The Cytology of the Sugarcane 111. Genetica 7: 293-322. Brown, W. V. 1930. A cytological study of some Texas Gramineae. Bull. Torrey Bot. Club. 77: 63-76. - 1951. Chromosome numbers of some Texas grasses. Bull. Torrey Bot. Club. 78: 292 - 299. Celarier, R. P. 1956a. Additional evidence for five as the basic chromosome number of the Andropogoneae. Rhodora 58: 135-143. - 1956b. Cytotaxonomy of the Andropogoneae I. Subtribes Dimerinae and Saccharinae. Cytologia 21: 272-291. - 1956c. Cytology of Andropogon distachyus L. Bull. Torrey Bot. Club 83: 183-191. - 1957. Elyonurus argenteus, a South African grass with five chromosome pairs. Bull. Torrey Bot. Club. 84: 157-162. Collins, G. N. 1921. Teosinte in Mexico. Jour. Hered. 12: 338-350. - and Kempton, J. H. 1920. A teosinte-maize hybrid. Jour. Agr. Res. 19: 1-37. - and Longley, A. E. 1935. A tetraploid hybrid of maize and perennial teosinte. Jour. Agr. Res. 50:123-133. Darlington, C. D. 1956. Chromosome Botany. London. de Wet, J. M. J. 1954. Chromosome numbers of a few South African grasses. Cytologia 19: 97-103. - and Anderson, L. J. 1956. Chromosome numbers in Transvaal grasses. Cytologia 21:1-10. Emerson, R. A. 1929. Genetic notes on hybrids of perennial teosinte and maize. Amer. Nat. 63: 289-300. Farquharson, L, 1. 1955. Apomixis and polyembryony in Tripsacum dactyloides. Amer. Jour. Bot. 42: 737-743. Gould, F. W. 1956. Chromosome counts and cytotaxonomic notes on grasses of the tribe Andropogoneae. Amer. Jour. Bot, 43: 395-404. Gramer, E. A. and Addison, G. 1944. Meiose in Tripsacum australe Culter e Anderson. 182 R. P. Celarier Cytologia 22

Anais escole super. agr. "Luiz du Queiroz" Univ. Sao Paulo, Sep. 9: 213-224. Hackel, E. 1889. Andropogoneae in De Candolle, A. Monographiae Phanerogamarum VI. Paris. Harada, K., Murakami, M. Fukushima, A. and Nakazina, M. 1954. Breeding study on the forage crops; studies on the intergeneric hybridization between the genus Zea and Coix (Maydeae) I. Saikyo U. Fac. Agr. Sci. Rpt. 6: 139-145. Henrard, J. T. 1941. Notes on the nomenclature of some grasses. Blumea 4: 496-538. Hunter, A. W. S. 1934. A karyosystematic investigation in the Gramineae. Canad. Jour. Res. (C) 11: 213-241. Hubbard, C. E. 1935. Ischaemum, fragile R. Br. in Hook. Ic. Tabula 3263. Iltis, H. 1911. Uber einige bei Zea mays L. beobachtete Atavismen, ihre Verursachung durch den Maisbrand, Ustilago Maydis D. C. (Corda) and uber die Stellung der Gattung Zea im System. Ztschr. Induktive Absam. U. Vererbungslehre 5: 38-57. Janaki-Amoral, E. K. 1945. in Darlington and Janaki Ammal, Chromosome Atlas of cultivated plants. London. Keng, Y. L. 1939. The gross morphology of Andropogoneae. Sinensis 10: 274-343. Krishnaswamy, N. 1941. Untersuchungen zur Cytologie uud Systematik der Graminee. Botanischen Centralblatt 60: 1-56. Krishnaswamy, N., Raman, V. S. Shetty, B. V. and Chandrasekharan, P. 1954. Chromo some numbers of some Indian economic plants. Curr. Sci. 23: 64-65. Kuwada, Y. 1915. Uber die Chromosomenzahl von Zea mays L. Bot. Mag. Tokyo 29: 83 -89. Longley, A. E. 1924. Chromosomes in maize and maize relatives. Jour. Agr. Res. 28: 673 -682. - 1934. Chromosomes in hybrids between Euchlaena perennis and Zea mays. Jour. Agr. Res. 48: 789-806. - 1937. Morphological characters of teosinte chromosomes. Jour. Agr. Res. 54: 836-862. Mangelsdorf, P. C. and Reeves, R. G, 1939. The origin of Indian corn and its relatives. Texas Agr. Expt. Sta. Bull. 574: 1-315. Mehra, K. L. 1955. Chromosome number in the tribe Andropgoneae, Gramineae. Ind. Jour. Genet. and Pl. Breed. 15: 144. Moriya, A. and Kondo, A. 1950. Cytological studies of forage plants I. Grasses. Jap. Jour. Genet. 25: 126-131. Nirodi, Nalini. 1955. Studies on Asiatic relatives of maize. Annals of Mo. Bot. Gard. 42: 103-130. Ono, H. and Tateoka, T. 1953. Karyotaxonomy in Poaceae I. Chromosomes and taxonomic relations in some Japanese grasses. Sot. Mag. Tokyo 66: 18-27. Pilger, R. 1940. Gramineae Ill. Unterfam. Panicoideae in Engler U. Pradtl, Die Naturlichen Pflanzenfamilien. Vol. 2. Leipzig. Randolph. L. F. 1955. History and origin of corn II. Cytogenetic aspects of the origin and evolutionary history of corn. in Corn and Corn improvement. New York. - and Hernandez-Xolocotzi, E. 1950. Cytotaxonomic diversity of Tripsacum in Mexico. Genetics 35: 686. Reeves, R. G. and Mangelsdorf, P. C. 1935. Chromosome numbers in relatives of Zea mays L. Amer Nat. LXIX: 633-635. - and - 1942. A proposed taxonomic change in the tribe Maydeae (family Gramineae). Amer. Jour. Bot. 29: 815-817. Rhodes, M. M. 1955. The cytogenetics of maize. in Corn and Corn improvement pp. 123 - 219. Rogers, J. S. 1950. Fertility relationships in maize-teosinte hybrids. Texas Agr. Expt. Sta. Bull. 730: 1-18. Rollins, R. C. 1953. Cytogenetical approaches to the study of genera. Chronica Botanica 14: 133-139. 1957 Cytotaxonomy of the Andropogoneae 11 183

Sampath, S. and Ramanathan, K. 1949. Chromosome numbers in Indian economic plants III. Curr. Sci. 18: 408-409. Simmonds, N. W. 1954. Chromosome behavior in some tropical plants. Heredity 8: 139-146. Stapf, O. 1919. Gramineae in Prain, D., Flora of Tropical Africa. Vol. 9. London. Tateoka, T. 1954. Karyotaxonomy in Poaceae 11. Somatic chromosomes of some grasses. Cytologia 19: 317-328. - 1955. Karyotaxonomy in Poaceae III. Further studies of somatic chromosomes. Cytologia 20: 296-306. Vallaeys, G. 1948. Le "Coix Lacryma-Jobi". Bull. Agric. Congo Belge 39: 247-304. Watt. F. 1904. Coix spp. or Job's Tears: A review of all available information. Agr. Ledger 11: 189-229. Weatherwax, P. 1926. Comparative Morphology of the Oriental Maydeae. Indiana Univ. Studies No. 73. - 1950. The history of corn. Scientific Monthly LXXI: 50-60. - and Randolph, L. F. 1955. History and origin of corn. I. and II. in Corn and Corn Improvement. New York. Wellhausen, E. J., Roberts, L. M. and Hernandez X. E. in collaboration with Paul C. Mangelsdorf. 1952. Races of maize in Mexico. Their origin, characteristics, and distribution. Bussey Inst. of Harvard Univ. 223 pp.