APOMIXIS AND SEXUALITY IN FORSSK. AT DIFFERENT PLOIDY LEVELS (GRAMINEAE)

SP. BIRARI*

Cytogenetics Laboratory, College of Agriculture, Pune, Maharashtra State,

Received January 1, 1979 /Accepted November 15, 1979

The twenty-chromosome, diploid species Themeda Emery (1957) in a 2n = 40 chromosomal race of anathera (Nees) Hack., T. tremula Nees ex Steud., T. from India. According to them, the strigosa (Ham) A. Camus, T. hookeri (Griseb.) A. tetraploid cytotype of T. triandra with moderate Camus, T. mooneyi Bor and two forty-chromosome, meiotic irregularities is a facultative apomict. Merewe tetraploid species, T. villosa (Nees) A. Camus and T. (1957) published on tile embryology of T. triandra caudata (Nees) A. Camus with normal meiosis, are from South . Without any mention of chromo- sexual. Two more tetraploids, T. quadrivalvis (L.) O. some number and type of meiosis, he recorded T. Ktze and T. triandra Forssk., one fifty-chromosome triandra to be sexual and apomictic. Woodland (1964) pentaploid, T. longispatha Raiz et Jain, two sixty- undertook studies on the mode of reproduction in chromosome hexaploids, T. dacruzii Birari and T. diploid and tetraploid T. australis from . arundinacea (Roxb.) with moderate degrees of meiot- According to him, diploid T. australis is sexual and ic irregularities, are facultative apomicts. Eighty-, the tetraploid a facultative apomict. Tlfis paper re- ninety- and one-hundred-and-ten-chromosome, octo-, ports on file mode of reproduction in the diploids nona- and 11 -ploids of T. triandra with severe meiotic T. anathera, T. tremula, T. strigosa, T. hookeri, T. irregularities are obligate apomicts. Sexual species mooneyi, the tetraploids T. villosa, T. caudata, T. produce reduced 8-nucleate bipolar Polygonum-type quadrivalvis, T. triandra, the pentaploid T. longispatha, of embryosacs in the ovules. In the facultative apo- the hexaploids T. dacruzii, T. arundinacea and in micts, most of the ovules exhibit the simultaneous octo-, 9- and 11-ploid cytotypes of T. triandra, in development of unreduced, 4-nucleate monopolar relation to chromosome number (ploidy) and chro- Panicum type and reduced, 8-nucleate Polygonum mosomal association in meiosis. type embryosacs. In a few cases the sexual embryo- sacs degenerated and only Panicum type matured. In obligate apomicts there was early degeneration of Material and methods MMC, dyads and, in few instances, mature sacs of the Polygonum type and only four-nucleate Panicum The material included 37 accessions (Tab. 1) collected type embryosacs developed. The ovules in this group from 7 states of India, and belonging to 12 species of of cytotypes produced 1 to 5 embryosacs of the Themeda as identified by the author and subsequent- Panicum type. ly confirmed by the Director, Botanical Survey of India Calcutta, and the Director, Royal Botanic Introduction Gardens, Kew, England. The specimens of all acces- sions were deposited in the herbarium of tile Regional Studies on the mode of reproduction in relation to Botanist, Botanical Survey of India, Western Circle, chromosome number were reported by Brown & Pune. Embryological preparations were made by the * Present address: Head, Dept. of Agricultural Botany, traditional microtome method and by Bradley's mod- Konkan Krishi Vidyapeeth, Dapoli 415712, Maharashtra, ified ovule squash technique (D'Cruz & Reddy, India 1967). Flemnfing's triple stain technique was used for

Genetica 54, 133-139 (1980). 0016-6707/80/0542-0133 $1.40. 133 © Dr. W. Junk B.V. Publishers, The Hague. Printed in The Netherlands. Table 1 Species studied and collecting data

Species Accession Location Date of Number collection

Themeda anathera 1-773 Shivalik, Uttar Pradesh Dec 1969 Themeda tremula 1-731 Foot of Pratapgar, Maharashtra Nov 1968 Themeda strigosa 1-777 Gorakhpur, Uttar pradesh Jan 1970 Themeda hookeri I-775 Raisily Koraput, Orissa Jan 1970 Themeda mooneyi 1-776 Nelpada-Koraput, Orissa Jan 1970 Themeda villosa 1-686 Ranchi, Bihar Jan 1969 1-687 Shillong Assam Jan 1969 1-689 Gauhatti-Assam Jan 1969 1-772 Shivalik, Uttar Pradesh Dec. 1969 Themeda caudata 1-778 Mussory, Uttar Pradesh Dec. 1969 1-641 Pune, Maharashtra Sep. 1968 1-647 Carla Cave, Maharashtra Sep. 1968 1-648 Khopoli, Maharashtra Sep. 1968 1-651 Kamshet, Maharashtra Sep. 1968 1-657 Palghar, Maharashtra Oct. 1968 1-664 MRA Panchgani, Maharashtra Nov. 1968 1-668 Mahabaleswar, Maharashtra Nov. 1968 I-674 Satpura, Maharashtra Nov. 1968 1-690 Gauhatti, Assam Jan. 1969 1-766 Palamau, Bihar Nov. 1969 1-767 Laxmipur, Orissa Jan. 1970 1-768 Rajendranagar,, Hyderabad Jan. 1970 Therneda longispatha 1-779 Dehra Dun, Uttar Pradesh Dec. 1969 Themeda dacruzii 1-774 Mussory hills, Uttar Pradesh Dec. 1969 Themeda arundinacea 1-780 Sastradhara, Uttar Pradesh Dec. 1969 Themeda triandra 1-897 Tamil Nadu, Madras Jan. 1970 1-638 Panhala Fort, Maharashtra Jan. 1970 1-683 Ooty, Tamil Nadu Nov. 1968 I-691 Hazaribag, Bihar Jan. 1969 1-770 Laxmipura, Orissa Jan. 1970 I-771 Raisily, Orissa Jan. 1970 1-637 Purandar Fort, Maharashtra Nov. 1968 1-688 Netherhat, Bihar Jan. 1969 1-640 Panchagani, Maharashtra Apr. 1968 1-665 Mahabaleswar, Maharashtra Nov. 1968 1-673 Pratapgar, Maharashtra Nov. 1968

staining the sectioned material (Johansen, 1940). detailed report on meiotic behaviour in diploid and Ovule squashes were stained in 1% acetocarmine. polyploid species of Themeda will be given else- where. However, to elaborate the observed rela- tions between the mode of reproduction and type Observations of meiosis encountered, the data on chromosome association at metaphase I and the type of re- Meiosis production have been summarized in Table 2. These data reveal that all diploids (2n = 20) and two tetra- Meiosis in the species and cytotypes of Themeda ploids, T. villosa and T. caudata (2n = 40)form bi- varied from essentially normal to highly irregular. A valents only, while the other two tetraploids T. qua-

134 Table 2 Chromosome association at meiotic metaphase I, tetrad formation and type of reproduction in species of Themeda

Species Chromosome + Average chromosomal configu- Reproduction number rations per cell at metaphase I 2n and types of tetrad

Themeda anathera 20 ! 0 II, normal tetrads sexual Themeda tremula 20 10 II, normal tetrads sexual Themeda strigosa 20 10 II, 0-3 B, abnormal tetrads and sexual triads Themeda hookeri 20 10 11, 0-3 B, normal tetrads sexual Thimeda rnooneyi 20 10 II, 0-3 B, normal tetrads sexual Themeda villosa 40 20 11, normal tetrads sexual Themeda caudata 40 20 II, normal tetrads sexual Themeda quadrivalvis 40 3 I + 13.5 II + 0.69 III+ 2.00 IV, facultative apomictic tetrads with or without micronuclei Themeda triandra 40 2.40 1 + 15.20 II + 1.80 IV + 2 B, facultative apomictic tetrads always with micronuclei 80 0.40 I + 30.40 I1 + 2 IV + 1.80 VI, obligate apomictic tetrads always with micronuclei 90 5.0 I + 26.84 II + 1.50 III + 1.83 IV obligate apomictic + 0.33 V + 1.67 VI + 0.50 VII + 0.17 VIII + 0.33 IX; tetrads always with micronuclei 110 3.13 I+31.8011+ 1.75 111+4.90 obligate apomictic IV + 2.50 VI + 0.38 1X tetrads with micronuclei Themeda longispatha 50 7.60 1 + 12.40 II + 2.04 III + 1.72 facultative apomictic IV + 0.92 V, tetrads with or without micronuclei Themeda dacruzff 60 16.40 II + 2.0 III + 2.12 IV + 2.12 facultative apomictic VI, tetrads with or without micronuclei Themeda arundinacea 60 1.40 I + 25.00 I1 + 1.00 1II facultative apomictic + 1.40 IV, tetrads with or without micronuclei

+Averages based on more than 50 PMC examined.

drivalis and T. triandra, the pentaploid T. longi- Mode of reproduction spatha and the hexaploids T. dacruzii and T. arun- dinacea form multivalents and univalents. Forma- The ernbryosac formation was normal in T. anathera, tion of higher associations appears the characteristic T. tremula, T. strigosa, T. hookeri, T. mooneyi, T. vil- feature of the 2n = 80, 2n = 90 and 2n = 1 I0 cyto- losa and T. caudata. In each ovule of these species, a types of T. triandra. Thus in the species of Themeda primary subepidermal cell derived from an archespo- chromosome association ranged from normal to rium differentiated into a megaspore mother cell highly irregular, with intermediate behaviour in two (MMC). It underwent first meiotic division and form- out of four 2n = 40 species and in the 2n = 50 and ed a linear tetrad. Three megaspores of file tetrad at 2n = 60 cytotypes. the micropylar end degenerated wl~e the chalazal

135 one developed. An enlarged chalazal megaspore under- were examined. Of these, 22 (17%) had single, well- went three successive mitotic divisions and formed developed Polygonum type sacs with underdeveloped two, four and eight nuclei respectively. The unorien- uni.nucleate aposporous cells, while 53 (35.70%) ted 8-nucleate embryosac oriented its nuclei at the ovules had both Polygonum and Panicum type sacs time of anthesis. The mature 8-nucleate reduced em- developed simultaneously. In 38 of these ovules, Po- bryosac with oriented nuclei was a typical bipolar lygonum and Panicum type sacs occurred in a 1:1 Polygonum type. A day after anthesis, the antipodals proportion, in 15 in a 1:2 proportion. In the third were found to divide into six nuclei and these antipo- group (55 ovules), there were only one to three dais underwent further successive mitotic divisions Panicum type of sacs. In pentaploid T. longispatha, and formed an antipodal complex. of the 100 ovules examined, 21 had a single Polygo- The normal sequence of megasporogenesis and num type embryosac with one or two nucleate apo- gametogenesis as described above was disturbed in sporous cells, while 70 (70%) ovules possessed both most of the ovules in T. quadrivalvis, T. triandra, T. sacs, well developed. Of these, 48 ovules had equal longispatha, T. dacruzii and T. arundinacea, while it numbers of Polygonum and Panicum type sacs, while collapsed completely in all ovules of the 8-, 9- and 22 had the Polygonum and Panicum type in a 1:2 11-ploid cytotypes of T. triandra. The ovules of the proportion. The third group of 9 (9%) ovules had species and cytotypes in which the sexual process only one to two aposporous Panicum sacs. In the was disrupted were found to form an unreduced four- hexaploid T. dacruzii, of the 185 ovules examined, nucleate monopolar aposporous embryosac by mitotic 51 (27.27%) had well developed single Polygonum divisions. The ovules of some of the species were type sacs while the second group of 99 (5351%) found to develop both sacs simultaneously. In the ovules had both types in the proportion of 1 : 1 in 65 species in which the sexual process in the ovules was ovules and 1:2 in 34 ovules. The third group of 35 completely disrupted and where multiple aposporous (18.92%) ovules had only 1 to 3 aposporous Panicum embryosacs were formed, the disruption could not be type sacs. In the other hexaploid, T. arundinacea, of studied in detail due to overlapping of the multiple the 125 ovules, 32 (25.60%) had well developed Poly- embryosacs, and hence the development at MMC, gonum type sacs with one or two uninucleate apospo- dyad and tetrad stages could not be recorded in sec- rous cells. In the second group of 60 (48%) ovules, 28 tioned material. However, ovules of these species and had a 1:1 and 32 had a 1:2 proportion of Polygonum cytotypes were examined at the time of anthesis by and Panicum type of embryosacs. In the third group the ovule squash technique and the results obtained of 33 (26.40%), the ovules had only Panicum type are given below. More than 150 ovules were studied sacs (1 to 3). In the octoploid, nonaploid and 11- in each diploid T. anathera, T. tremula, T. strigosa, T. ploid cytotypes of T. triandra, all ovules had only hookeri, T. mooneyi and in the two tetraploids, T. Panicum type sacs (1 to 5). Of the 175 ovules exam- villosa and T. caudata. Ovules of all the accessions in ined in the octoploid cytotype all had Panicum type each of the above species had only single reduced Po- sacs (1 to 3), while in nonaploid and 11-ploid T. trian- lygonum type embryosacs, indicating a sexual mode dra. all 250 ovules examined had Panicum type sacs of reproduction. In T. quadrivalvis 240 ovules were (1 to 5). examined in 13 accessions and according to the num- Thus, the species and cytotypes of Themeda had ber and type of embryosac in the ovule, they were two extreme groups of ovules: those with sexual and classed in three groups. The first group (40 ovules = those with aposporous sacs. The species possessing 16.66%) had a well developed Polygonum type with only sexual or only aposporous sacs have been classi- underdeveloped uni- or binucleate aposporous cells. fied as sexual and obligate apomicts respectively. The In the secound group (102 ovules = 42.45%) there species having intermediate ovules, i.e. producing was simultaneous development of both types of sacs. both types in the same or in different ovules indepen- Of these, 90 and 12 ovules had the Polygonum and dently have been classified as facultative apomicts. Panicum type of sacs, in a 1:1 and 1:2 proportion According to these categories, the diploid T. anathera, respectively. In the third group (98 ovules = 40.84%) T. tremula, T. strigosa, T. hookeri, T. mooneyi and there were only one to four Panicum type of sacs. In the two tetraploids T. villosa and T. caudata are strict- the tetraploid cytotype of T. triandra, 130 ovules ly sexual types. On the other hand, T. quadrivalvis, T.

136 triandra 2n = 40, T. longispatha, T. dacruzii, T. arun- and obligate apomicts respectively. The present re- dinacea are facultative apomicts. The octo-, nona- and sults differ from those of Celarier & Harlan in that of 11-ploid cytotypes ofT. triandra are obligate apomicts. the four tetraploids reported, two, having normal meiosis, were sexual and the other two, having abnor- Discussion mal meiosis, were facultative apomictic. Babcock & Stebbins (1938), Stebbins (1941 and In the species and cytotypes of Themeda, the degene- 1950), Powers (1945), Gustafsson (1946), Manton ration of the sexual products MMC, dyad and occa- (1952) and Brown & Emery (1958) have emphasized sionally mature sacs indicated the disruption of the that there is a correlation between apomixis and po- sexual process. The ovules of some species in which lyploidy. Brown & Emery have restricted their studies degeneration occurred formed aposporous embryo- to the family of grasses. Akerberg (1950), Nygren sacs. The species in which the sexual process was not (1950), Powers (1945) and Miintzing (1940) sug- disturbed produced a normal reduced 8-nucleate, gested that apomixis was due to a delicate genetic Polygonum type of embryosac. The ovules in some balance. of the species formed both aposporous and sexual Powers, while studying apomixis in Parthenium has sacs simultaneously. suggested a genetic model which involves three pairs The species of Themeda had a very classical series of factors. Stebbins (1950) accepted the hypothesis of chromosome numbers (ploidy levels) ranging from of Powers but felt that apomixis in other was 2n = 20 to 2n = 110. Their meiotic and reproductive more complex. Gerstel et al. (1950 and 1953) stated behaviour is shown in Table 2. that two doses of genes for apomixis were dominant The reproduction in T. anathera, T. tremula, T. over one dose of genes for sexuality in Parthenium. strigosa, T. hookeri, T. mooneyi, T. villosa, T. caudata Burton & Forbes (1960) working on Paspalum was sexual. The development of the female gameto- came to the conclusion that apomixis was recessive phyte was similar to that described by Woodland to sexuality and the inheritance was tetrasomic. (1964) in T. australis (2n = 20). It was aposporous Harlan et al. (1964) working with tetraploid Dichan- apomictic in T. triandra, T. quadrivalvis, T. longi- thium and Bothriochloa concluded that apomixis spatha, T. dacruzii, T. arundinacea, as ovules of these was dominant to sexuality and was controlled by not species and cytotypes formed 4-nucleate unreduced more than a single gene per genome. Reddy (1967) aposporous Panicum type embryosacs and had an working with crosses in Dichanthh~m, suggested that essentially similar aposporous development of the sexuality and apomixis were different phenomena female gametophyte to that reported by Celarier & controlled by two groups or blocks of genes. He Harlan (1957) in Dichanthium, Bothriochloa, Capel- further suggested that the genes for apomixis were lipedium, by Fischer et al. (1954), Snyder et al. (1955) cumulative, while those for sexuality were non-cumu- inPennisetum, by Brown & Emery (1958) in 17 species lative. of the genera Bothriochloa, Capellipedium, Dichan- The workers listed above, wltile working on the thium, Heteropogon, Hyperrahania and Themeda, inheritance of sexuality and apomixis have only taken Woodland (1964) in T. australis 2n = 40 and by Brown into consideration the mode of reproduction (sexual &Emery (1957)in T. triandra 2n = 40. Embryological or apomictic) and the somatic chromosome comple- studies in apomictic types made so far in Gramineae, ment (ploidy). They did not consider the meiotic revealed the formation of two types of unreduced behaviour. There are indications that polyploids aposporous embryosacs; those having four nuclei as with normal meiosis are sexual like the diploids in Panicum-type sacs and the other with 8 nuclei as while polyploids with abnormal meiosis are apomictic. in Hieracium-type sacs. Eight-nucleate, unreduced In tile mode of inheritance of reproduction, there- Hieracium-type of embryosacs were not encountered fore, not only the ploidy of the individual but also in the aposporous apomictic species under study. its meiotic behaviour will have to be taken into con- Celarier & Harlan (1957) reported that diploids were sideration, as the genes responsible for the meiotic sexual and polyploids (including tetraploids)were behaviour appear to influence the formation of tile apomicts in Dichanthium, Bothriochloa and Capel- type of embryosac either wholly or partially. Further lipediurn. According to them, tetraploids and hexa- study in this direction would be necessary, which ploids were irregular in meiosis and were facultative 137 would involve a programme of interspecific hybridi- the Regional Botanist, Botanical Survey of India, zation under close cytogenetic observation. Western Circle, Pune; the Director of the Botanical Bews (1929) suggested that in Africa the advanced Survey of India, Calcutta and the Director, Royal genus Themeda covers immense areas of sub-tropical Botanic Gardens, Kew, England. Shri Prabhakar S. grassland. Gluckman (1951), Bayer (1954) and De Shimpi is acknowledged for typing the manuscript. Wet (1960) reported T. triandra to be the dominant This research has been financed in part by the grass of . Bor (1960) reported that grant made by The United States Department of Themeda spreads over the Indian territory under di- Agriculture, under PL 480 Project A-7 ~R-130. verse climatic conditions and habitats. Hayman (1960) stated that T. australis (R. Br.) Stapf, com- monly known as Kangaroo grass was the only which References had a distribution extending into the temperate re- gions. Its habitat was wide, extending from banks of Akerberg, E. (1950). The progeny of artificial hybrids be- tween Poa pratensis, L. and Poa alpina, L. Proc. 7th Int. seasonal water courses in central Australia to altitudes Congr. Bot. : 327-328 upto 5000 ft. No other species in the Australian flora Babcock, E.B. & G.L. Stebbins Jr. (1938). The American had a distribution more extensive than T. australis. species of Crepis: their relationships and distribution as Hooker (1896) cited that Anthistiria imberbis, i.e. affected by polyploidy and apomixis. Carnegie Inst. T. triandra var. Royelei Hook. of the North West Washington Publ. 504. 200 pp. Bayer, A.W. (1954). The ecology of grassland. In: D. Mere- Himalaya (Simla Hills) was distributed in the tempe- dith ed., Grasses and pasture of South Africa. Johannesburg rate zones and rose up to 7500 ft altitude at Kumaun. (1954). Stapf (1919) found T. triandra var.punctata on Soyra Bews, J.W. (1929). The world grasses. Their differentiation, mountain in the Nile region of Africa at the altitude distribution, economics and ecology. Longmans, Green of 9200-9850 ft. In the present studies the 37 acces- and Co. Ltd. London. Birari, S.P. (1973). New species of Themeda Forssk. from sions belonging to twelve species of Themeda were India J. Bombay nat. Hist. Soc. 70: 346-347. collected from all over India from regions with diverse Brown, W.V. & W.H.P. Emery, (1957). Apomixis in the Gra- climatic conditions. The wide adaptability of the mineae tribe Adropogoneae: Themeda triandra and higher chromosomal races and their dominating Bothriochloa ischaemum. Bot. Gaz. 118: 246-253. Brown, W.V. & W.H _P. Emery (1958). Apomixisin Gramineae, place in the different floras might have its basis in the .Am. J. Bot. 45: 253-263. polyploidy and the partially apomictic reproductive Burton, G.W. & I. Forbes (Jr.) (1960). The genetics and mani- system. The genus Themeda reproduces by means of pulation of obligate apomixis in common Bahia grass apomictic embryo development. Normal meiosis in (Paspalum notatum Flugg.) Proc. 8th Int. Grassld Congr.: some polyploid collections suggests that an occasio- 66-71. Bor, N.L. (1960). The grasses of Burma, Ceylon, India and nal normal embryo and pollen grain could be pro- Pakistan. Pergamon Press, Oxford. duced to form sexual seeds. Such a partially apomic- Celarier, R.P. & J.R. Harlan (1957). Apomixis in Bothrio- tic system could explain the large number of chromo- chloa, Capillipedium. Phytomorphology 7: 93-102. some races encountered in Themeda and as indicated clausen, N.J. (1954). Partial apomixis as an equilibrium sys- by Clausen (1954) may explain the variabilities of the tem in evolution. Caryologia (Suppl.) 6: 469-479. D'Cruz, R. & P.S. Reddy (1967). A modified Bradley's squash genus in any population where more than one species technique for studying embryosac in Gramineae. Stain occurs. Technol. 42: 237-240. De Wet, J.MJ. (1960). Cytogeographyof Themeda triandra The author wishes to express his deep indebtedness in South Africa. Phyton 15: 37-42. to Prof. Dr. Riu D'Cruz for his guidance and encour- Fischer, W.D., E.C. Bashaw & C.E. Holt (1954). Evidence for apomixis in Pennisetum ciliare and Cenchurus setigerus. agements; to Dr. A.B. Joshi, the Vice-Chancellor, Agron. J. 46: 401-404. Mahatma Phule Krishi Vidyapeeth, Rahuri, for valu- Gerstel, D.U. & W.M. Mishanoo (1950). On the inheritance able criticism during the preparation of the manus- of apomixis in Parthenium argentatum. Bot. Gaz. 112: cript and to Dr. M.V. Thombre for several helpful 96-109. suggestions. Gerstel, D.U. & W.IVl. Mishanoo (1953). An additional note on the inheritance of apomixis in Guayule. Bot. Gaz. Thanks are also due for the sincere help in the 115: 89-93. identifications of different species of Themeda by

138 Gluckman, E. (1951). Cytotaxonomic studies in Themeda triandra Forssk, Unpublished thesis Witwatersrand Univ. So. Aft. Cited in D. Meredith's Grasses and Pastures of South Africa, Johannesburg, (1954), by R. De V. Pienar under the title 'The chromosome numbers of some In- digenous South African and introduced Gramineac'. Gustafsson, A. (1946). Apomixis in the higher plants. I. The mechanism of apomixis, Lunds Univ. Arsskr. N.k] Avd. 2. 43: 71-178. Harlan, J.R., M.H. Brooks, D.S. Borgaonkar & J.M.J. De Wet (1964). Nature and inheritance of apomixis in Bothrio- chloa and Dichanthium. Bot. Gaz. 125: 41-46. Hayman, D.L. (1960). The distribution and cytology of the chromosome races of Themeda australis in Southern Australia.Aust. J. Bot. 8: 58-68. Hooker, J.D. (1896). Flora of British bldia (Gramineae) 7: 210-220. Reev & Co, London. Johansen, D.A. (1940). microtechnique. Mc-Graw Hill, New York. Manton, I. (1952). Problems of cytology and evolution in the Pteridophyta. Univ. Press, Cambridge, 316. Merewe, R.B. (1957). 'N embryologiese studie van Themeda triandra Forssk. Jl S. Air. Bot. 23:139-149. Muntzing, A. (1940). Further studies on apomixis and sexua- lity in Poa. Hereditas 26: 115-190. Nygren, A. (1946). A preliminary note on cytological and embryological studies in arctic Poae. Hereditas 36: 231-232. Powers, L. (1945). Fertilization without reduction in guayule (Parthenium argentatum Grey.) and hypothesis as to tile evolution of apomixis and polyembryony. Genetics 30: 323-346. Reddy, P.S. (1967). Study of mechanism and inheritance of apornixis in Dichanthium Willemet. Thesis Univ. Pune (1967). Stapf, O. (1919). Gramineae. In: Prain's Flora of Tropical Africa 9: 1-768. Stebbins, G.L. (1941). Apomixis in angiosperms. Bot. Rev. 7: 507-542. Stebbins, G.L. (1950). Variation and evolution ht plants. Columbia Univ. Press, New York. Snyder, L.A., A.R. Hernandez & H.E. Warmke (1955). The mechanism of apomixis in Pennisetum ciliate, Bot. Gaz. 116: 209-222. Woodland, P.S. (1964). The floral morpbology and embryo- logy of Themeda australis (A. Br.) Stapf. A ust. J. Bot. 12: 157-172.

139