Cytogenetic studies in the genus Tribolium (Poaceae: Danthonieae). IV. Section Uniolae

The section Uniolae of the genus Tribolium Desv. comprises four species, namely T. alternans (Nees) Renv., T. amplexum (Nees) Renv., T. brachystachyum (Nees) Renv. and T. uniolae (L. f.) Renv. Tribolium aiternans, T. amplexum and T. uniolae are morphologically similar and form a hybrid swarm. Thus, only two species should be recognized, namely T. brachystachyum and a T. uniolae hybrid swarm . Diploid, tetraploid and hexaploid specimens have been observed in the T. uniolae hybrid swarm, whereas all the T. brachystachyum specimens were tetraploid. Segmental alioploidy, tending towards alioploidy, is present in both species.


Introduction
The genus Tribolium Desv. comprises twelve species and is subdivided into three sections (Visser & Spies 1994a), of which only the section Uniolae will be discussed in this paper. Tribolium alternans (Nees) Renv. is distributed in sandy soil, in river flats in the south-western Cape Province, whereas T. amplexum (Nees) Renv. is distributed in disturbed sandy soil in the same area (Gibbs Russell et al. 1990). Tribolium brachystachyum (Nees) Renv. is restricted to the mountains between Paarl and Franschhoek. Tribolium uniolae (L. f.) Renv. is widely distributed and grows from the eastern extremity (Port Elizabeth) of the distribution area of the genus to the western extremity, the Atlantic Ocean (Gibbs Russell et al. 1990).
Cytogenetic studies on Tribolium suggest a basic chromosome number of six (Spies et al. 1992;Visser & Spies 1994 a-c) or a basic chromosome number of seven reported by De Wet (1960), based on a T. uniolae specimen with 2n = 28. We attribute the deviation to the possible presence of B-chromosomes in De Wet's (1960) specimen, since no similar deviation was observed in any of the more than 200 specimens studied in our laboratories.
The meiotic chromosome behaviour of the species in the section Uniolae was studied in order to determine the type of ploidy present among the different species, as the type of ploidy present indicates whether hybridization played any role during the development of the different species in this section. In addition to the cytogenetic data, morphological data was included in an attempt to determine the phylogenetic relationships among the various species in the section Uniolae.

Materials and Methods
The materials used were collected in the field. Voucher herbarium specimens are housed in the Geo Potts Herbarium (BLFU), Department of Botany and Genetics, University of the Orange Free State, Bloemfontein, and the National Herbarium (PRE), Pretoria. These specimens and their localities are listed in Table 1. The methods used during this study included meiotic analyses and factor analyses of morphological attributes (Visser & Spies 1994c). Chiasma frequencies are regarded as the average number of chiasmata per bivalent and were calculated as the total number of chiasmata per cell divided by the haploid chromosome number of the cell. This method was used because it is congruous to other calculations done, especially with determining genome homology within an individual (Kimber & Alonso 1981).
All the species in the section Uniolae are perennial, tufted grasses, growing to a maximum of 700 mm. The inflorescences form compact spikes. The spikelets are distichously arranged, 4 -7 mm long and are 4 -9-flowered.
The glumes of T. aiternans, T. amplexum and T. uniolae specimens are glabrous to sparsely pubescent, whereas the glumes of T. brachystachyum are densely pubescent. The glumes of herbarium specimens of T. brachystachyum and T. uniolae in the section Uniolae were studied under an electron microscope. All the species in this section possess two types of trichomes on the spikelet, namely large trichomes on the glumes and a smaller fringe of trichomes on the lemma. The large trichomes are long, glassy and taper off apically, whereas the shape of the smaller trichomes differ between the various species (Visser & Spies 1994a). Tribolium alternans, T. amplexum and T. brachystachyum possess club-shaped trichomes, whereas it varies from tapered to club-shaped in T. uniolae (Visser & Spies 1994a).
Factor analyses of the morphological data revealed that T. ' alternans, T. amplexum and T. brachystachyum differ morphologically ( Figure 1). However, T. uniolae overlaps to such an extent with both T. alternans and T. amplexum, that morphological separation of these three species is impossible. All the Table 1 List of species cytogenetically studied, voucher specimen numbers and localities according to the degree reference system ( , Presence of cell fusion, resulting in more than one haploid chromo-. some number per specimen. specimens, except those of T. brachystachyum, can be regarded as part of a T. uniolae hybrid swarm. We regard a hybrid 'swarm as a complex mixture of parental forms, hybrids, backcross types and segregation products (Grant 1981). Therefore, we recognize only two species in the section Uniolae for the purpose of this study, i.e. T. brachystachyum and a T. uniolae hybrid swarm. Four T. brachystachyum specimens were cytogenetically studied (Table 1). This study supports the findings of Spies et ai. (1992) that this species is tetraploid (n = 2x = 12) ( Figure   2A--C,K,L). The average chiasma frequency for this species is 1.5 (Table 2). Various meiotic chromosome abnormalities have been observed in r . brachystachyum (Table 2). These abnormalities include univalents during metaphase I ( Figure  2D), chromosome laggards during anaphase I ( Figure 2E), micronuclei during telophase I ( Figure 2F) and II, precocious chromosome segregation during metaphase I and II ( Figure  2G), various anaphase I and II bridges ( Figure 2H,I), unsynchronized second divisions ( Figure 2J) and uneven segregation of chromosomes during anaphase I ( Figure 2K). B-chromosomes were present in all of the specimens studied (Table 1).
S.-Afr. Tydskr.Plantk., 1994, 60(5)  Chromosomes were described as B-chromosomes when their number deviated from the expected chromosome number and their meiotic chromosome behaviour deviated from the expected behaviour of additional aneuploid chromosomes. The behaviour of B-chromosomes in rribolium was discussed by Spies et ai. (1992). It is impossible to recognize B-chromosomes in all cells of a specimen. However, when certain meiotic stages or certain cells of a particular meiotic stage indicate the precence of B-chromosomes without doubt, chromosomes deviating from the expected behaviour in other cells of the same specimen were regarded as being B-chromosomes. This was only done where the 'abnormal' chromosome behaviour did not occur in any specimen without any B-chromosomes. The genome homology of two T. brachystachyum specimens have been determined according to the models proposed by Kimber and Alonso (1981). The genome analyses indicated that the observed chromosome configurations corresponded best with the expected frequencies for the 2:2-model, with xvalues of 1 (Table 3). The 2:2-model indicates that two sets of genomes are present. Each set consists of two genomes and the relative similarity of the genomes in a set is considered to be 0.5. The relative similarity between the sets of genomes is expressed by an x-value that may vary between 0.5 (differences between sets are similar to differences within a set) and 1 (sets differ very much). The x-values for the T. brachystachyum specimens was 1 (Table 3), thus indicating no homology between the two sets of genomes. Based on the specimens used during this study, this species is an alloploid species.   The T. uniolae hybrid swann was cytogenetically represented by 36 specimens (Table 1). Polyploidy was frequently encountered (Figure 3 A-F). One specimen was diploid (n = x = 6) ( Figure 3A), 31 were tetraploid (n = 2x = 12) ( Figure 3B) and four were hexaploid (n = 3x = 18) ( Figure 3C).
Various meiotic chromosome abnonnalities were observed. These abnonnalities included the following: univalents (or Bchromosomes) during metaphase I ( Figure 3E), chromosome laggards during anaphase I (Table 2), micronuclei during telophase I and II (Table 2), precocious segregation of one bivalent on the metaphase plate, anaphase bridges (Table 2), uneven segregation of chromosomes during anaphase I ( Figure 3F) and Table 3 Genomic relationships in the tetraploid T. brachystachyum and T. uniolae specimens according to the models of Kimber and Alonso (1981) • The x-value is followed by the sum of squares of the deviation between the observed and the expected values for each model, in parentheses. Nought to four B-chromosomes are present in the specimens from the T. uniolae hybrid swann (Table 1). The number of Bchromosomes varies per specimen and per polyploid level. The B-chromosomes fonn univalents during metaphase ( Figure  3E), which are either on the metaphase plate or away from it. The B-chromosomes on the metaphase plate fonn laggards during anaphase I that result in micronuclei during telophase I.
Genome analyses of nine specimens revealed that the 2:2model of Kimber and Alonso (1981) fitted the specimens to the greatest degree, with x-values of 1 or approximately 1 in all specimens, except Spies 4589 (x-value = 0.65) and Spies 4612 (xvalue = 0.75) (fable 3). With the exception of ili:se two segmental alloploid specimens, all other specimens were alloploids.

Discussion
The species of the genus Tribolium exhibit great genetical and morphological variation (Spies et al. 1992). This variability within a species is particularly evident in T. uniolae in the section Uniolae. This perennial species is a tufted grass, varying in length from 100 to 600 mm. The lengths of the inflorescences vary from 8 to 70 mm and are often branched at the base. The degree of hairiness differs extensively. Specimens can either be glabrous or hairy on the stems and/or the leaves and/or the panicle. Specimens with all these combinations of characters have been observed.
Due to the wide range of morphological variation in T. uniolae, this species includes many characters associated with T .  phological differences have been observed between T. uniolae and T. brachystachyum. The number of trichomes on the glumes varies for T. uniolae, whereas for T. brachystachyum, they are always dense. The structure of the trichomes on the glumes corresponds for the two species (Visser & Spies 1994a). Both species have club-shaped trichomes at the base and on the edge of the lemma (Visser & Spies 1994a), but the degree of hairiness between the T. uniolae specimens differs. Tribolium brachystachyum is very setaceous and contains trichomes on the leaves, the blades and the inflorescences, whereas T. uniolae is glabrous or setaceous in any combination of these three parts. Morphologically, T. uniolae varied for every character examined, except that none of the specimens had trichomes on the stems of the plants.
The morphological overlap between the species in this section, the large morphological variation observed in T. uniolae, the presence of sexually reproducing specimens, facultative apomixis (Visser & Spies 1994b) and the degree of variation in polyploid levels, indicate that T. uniolae forms a hybrid swarm, including T. alternans and T. amplexum Therefore, no coru;istent morphological distinction exists between the species included in this hybrid swarm. The low frequency of diploid individuals in T. uniolae suggests that it is a mature hybrid swarm.
In an attempt to confirm the morphological suggestions for hybridization in the section Uniolae, meiotic chromosome behaviour and the type of polyploidy present in different specimens of the section were studied. A wide range of haploid chromosome numbers have been observed for the section Uniolae, namely n = x = 6 (T. uniolae) , n = 2x = 12 (T. brachystachyum and T. uniolae) and n = 3x = 18 (T. uniolae).
These somatic chromosome numbers confirm the basic chromosome number of six for the genus and the section Uniolae (Spies et al. 1992).
The genomic constitution of the various polyploid specimens were studied with the use of Kimber and Alonso's (1981) models. Tribolium brachystachyum has been classified as a segmental allotetraploid species , tending towards alloploidy (Table  3). However, the presence of quadrivalents indicates that the genomes of this species must, to some extent, be similar.
The genomic constitution of the various T. uniolae specimens differed. The 2:2-model fitted the specimens to the greatest degree (Table 3), with x-values of I or tending towards 1. The only exceptions were Spies 4589 (x-value = 0.65) and 4612 (x-value = 0.75). Both these specimens have a very high number of quadrivalents per cell in comparison to all the other specimens (Table 2), thus indicating a greater degree of genome homology in these specimens. Nevertheless, all specimens are either alloploids or segmental alloploids and hybridization in all of them is, therefore, implied. The alloploid and segmental alloploid origin indicates that hybridization played an important part in the evolution of the section Uniolae.
Tribolium uniolae is usually found in mountainous areas. Five specimens were collected on Piketberg (Spies 4585,5008,5009,5010 & 5011). Due to the absence of other mountain ranges that link this locality to the mountain ranges in the southern Cape, Piketberg is geographically isolated. Three different polyploid levels have been observed in these specimens, namely diploid (Spies 4585), tetraploid (Spies 5008,5009 & 5010) and hexaploid (Spies 5011). The whole known range of polyploid levels observed for T. uniolae is , therefore, found in this geographically isolated location, making this area an ideal location for studying evolution in the species.
Morphologically, the five specimens differed. The prostrate diploid specimen is morphologically similar to one of the tetraploid specimens (Spies 5010). Both specimens are relatively small and tender, lack trichomes on the inflorescences and have curly leaves. The other tetraploid specimens (Spies 5008 S.-Afr.Tydskr.Plantk., 1994, 60(5) & 5009) correspond with the hexaploid specimen (Spies 5011), except for the trichomes on the latter specimen's inflorescences. The increasing morphological variation at higher ploidy levels supports a hybrid origin for these specimens.
The apparent ability of T. alternans, T. amplexum, T. brachystachyum and T. uniolae to hybridize and to exploit the advantages of hybrid species complexes or hybrid swarms, with ranges of chromosome numbers and genomes, is ancient in the grasses (De Wet 1986). Natural hybridization is common in the grasses and the variability within a hybrid population increases (Ehrendorfer 1980). This level of genetic variability allows the grasses to take advantage of new habitats (Ehrendorfer 1980), and, in the case of Tribolium, disturbed habitats such as road sides and agricultural ground (Spies et al. 1992).

Conclusions
The basic chromosome number of the species in the section Uniolae is x = 6. A wide range of ploidy levels have been observed in this section, namely diploid (T. uniolae), tetraploid (T. brachystachyum and T. uniolae) and hexaploid (T. uniolae). Two types of polyploidy have been identified, namely alloploidy (T. brachystachyum and T. uniolae) and segmental alloploidy (T. uniolae). This fact, as well as the results of the principal-component analysis, indicate that hybridization played a major role during the evolution of these species. The cytogenetic and morphological evidence lead to the recognition of a T. uniolae hybrid swarm, including the species T. alternans, T. amplexum and T. uniolae. It is therefore suggested that the number of species in this section should be reduced to two, namely T. brachystachyum and a T. uniolae hybri.d swarm.