Cytogenetic studies in the genus Pentaschistis ( Poaceae : Arundinoideae )

Cytogenetic studies of 45 specimens, representing 16 taxa of the genus Pentaschistis (Nees) Spach confirmed two basic chromosome numbers (x = 7, 13) for the genus. Chromosome numbers for five species and one subspecies are described for the first time, i.e. P. aimides subsp.  jugorum (n = 2x = 14). P. colorata (n = 2x = 14). P densifolia (n = x = 7), P. linui (n = 6x = ± 42), P. rigidissima (n = x = 7, n = 3x = 21) and P viscidula (n = 3x = 21). Polyploidy occurs frequently and new ploidy levels are described m four of the species, namely P airoides (Nees) Stapf subsp. airoides (n = 3x = 21), P. cir- rhulosa (Thunb.) McClean (n = x = 7), P. eriostoma (Nees) Stapf (n = 3x = 39+0-4B) and P rupestris (n = 4x = 28) The majority of species form young polyploid complexes. There seems to be no correlation between cytogenetic data and mor­phological groupings within Pentaschistis.

The aim of this study was to determine whether the cytogenetic studies support the morphological groupings of Linder & Ellis (1990) and Ellis & Linder (1990); to determine the degree of polyploidy in the genus and to use ploidy levels to determine the age of the polyploid complexes in the genus.

MATERIALS AND METHODS
The material used during this study was collected in the field.Voucher herbarium specimens are housed in the Geo Potts Herbarium.Bloemfontein (BLFU) and/or in the National Herbarium.Pretoria (PRE).
Young inflorescences were collected and fixed in Carnoy's fixative (Carnoy 1886).The fixative was replaced by 70% (v/v) ethanol 24-48 hours after fixa tion.Anthers were squashed in 2% (m/v) aceto-carmine (Darlington & La Cour 1976).Slides were permanently mounted by freezing them with liquid CO2 (Bowen 1956), followed by dehydration in ethanol and mounting in Euparal.At least twenty cells per specimen were stud ied for each meiotic stage, except where otherwise indi cated.
With the annotation of chromosome numbers we fol lowed the example set by the series 'Index to plant chro mosome numbers' (Goldblatt 1981), where chromosome * Department of Botany and Genetics.University of the Orange Free State.P.O Box 339, 9300 Bloemfontein t To whom correspondence should be addressed MS received 1997-09-15 numbers derived from meiotic studies are presented as the gametic (n) number and chromosomes from mitotic studies as the somatic (2n) chromosome number.The average number of chiasmata per bivalent was consid ered to be the number of chiasmata per chromosome and were calculated by dividing the mathematical average of the chiasmata per cell by the haploid chromosome num ber of the plant.Chromosome configurations were cal culated as the average number of each configuration per cell.The number of B-chromosomes is presented as the minimum and maximum number of B-chromosomes observed per cell.The numerical values of the chromo some abnormalities were obtained by calculating the average number of a certain abnormality per cell.Genomic relationships were calculated according to the models of Kimber & Alonso (1981).

RESULTS AND DISCUSSION
Forty five specimens, representing the different groups in Pentaschistis, except Group 5. have been stud ied.The results are presented alphabetically in table form for each group (Table 1).

Number o f specimens per ploidy level
All published ploidy levels (Hedberg 1957;De Wet 1960;Tateoka 1965a & b;Davidse et al. 1986;Du Plessis & Spies 1988;Spies & Du Plessis 1988;Du Plessis & Spies 1992;Spies et al. 1994) were used to determine the number of specimens per ploidy level among the different species of the genus Pentaschistis.The frequency of specimens per ploidy level was also plotted for the two different basic chromosome numbers in Pentaschistis (Figure 4).Meiotic analyses of the different Pentaschistis species revealed haploid chromosome numbers of seven, multiples of seven (Figures 1, 2), thirteen and multiples thereof (Figure 3).The low number of species with a basic chromo some number of thirteen can be attributed to the fact that only specimens from South Afnca have been studied.All species with a basic chromosome number of thirteen, except P. eriostoma.grow in central Afnca (Du Plessis & Spies    Polyploidy is common in the genus Pentaschistis.Nineteen of the 45 specimens studied were polyploid.Higher ploidy levels were detected in some of the cells of these specimens.This could be caused by cell fusion (Spies & Van Wyk 1995) which is not uncommon in the genus Pentaschistis (Figure 5).Pentaschistis airoides subsp.jugorum and P. colorata were tetraploid.The genome homology of these two species was determined and the observed chromosome configurations corresponded best with the expected values for the 2:2 model (Table 2) of Kimber & Alonso (1981).It indicates that two sets of genomes are present and each set consists of two genomes.The relative similarity of the genomes within a set is 0.5 and the relative affinity between the sets is expressed by an x-value, which may vary between 0.5 (differences between sets are similar to differences within a set) and one (sets are totally different) (Kimber & Alonso 1981).The x-values for the specimens studied were one or tending towards one (Table 2), thus indicating little to no homology between the two sets of genomes (for example AABB).Based on the specimens used during this study, these species are alloploid.However, because of the occurrence of an occasional quadrivalent, P. colorata may be classified as a segmental allotetraploid.The occurrence of quadrivalents indicates that the genomes of this species correspond to some extent.; The genome homology of the tetraploid specimens of P. pallida was determined.The observed and expected values corresponded best to the 2:2 model for two of the specimens, namely Spies 3828 and 5292, with x-values of one or tending towards one (Table 2).This indicates that little or no homology exists between the two sets of genomes, thus classifying these specimens as allotetraploids (for example AABB).However, the occurrence of quadrivalents indicates that the genomes of this species do correspond to some extent.Therefore, Spies 3828 is classified as a segmental allotetraploid (AAA'A') with a x-value of 0.7 (Table 2).The observed and expected values for the other two P. pallida specimens (Spies 3859 and 5381) correspond ed best to the 2:1:1 model, with x-values of 0.8 and 1 respectively (Table 2).The 2:1:1 model indicates that three sets of genomes are present, one set consists of two genomes and the other two of one genome each.The x-values for the specimens studied were one or tending towards one (Table 2), indicating little or no homology among the three sets of genomes (for exam ple AABC or AABB').However, because of the occur-.rence of quadrivalents in these two specimens (Spies 3859 and 5381) they are classified as segmental allotetraploids (AAA'A").Therefore, the specimens of P. p a l lida are either alloploids or segmental alloploids tending towards alloploidy (for example AABB.AABC, AABB' or AAA'A").The two different genomic constitutions for this species may indicate that there is a high degree of variation within the species.
In P. eriostoma (Figure 3A-E) three different ploidy levels were detected, namely diploid, tetraploid and hexaploid, that correspond with the published literature (Spies & Du Plessis 1988;Du Plessis & Spies 1992).Heptaploidy was also observed by Du Plessis & Spies (1992).In the diploid (n = x = 13) specimens of P. erios toma 13 bivalents were observed.The genome homolo gy of the tetraploid specimens of P. eriostoma was deter mined.The observed and expected values corresponded best to the 2:1:1 model with an x-value of 0.9 (Table 2), indicating that little homology exists among the three sets of genomes.This species is thus an allotetraploid species (for example AABC).However, a low frequency of quadrivalents occurred in the tetraploid.indicating that the genomes of this species do correspond to some extent.Pentaschistis eriostoma is, therefore, classified as a segmental allotetraploid (for example AAA'A").
All the existing chromosome data of Pentaschistis was used to determine the number of specimens per species per ploidy level.The number of specimens per ploidy level of the two different basic chromosome num bers was also determined (Figure 4).This was done to determine the degree of maturity (it is the ratio between diploid and polyploid specimens, as well as the level of polyploidy) in the polyploid complexes in the different species and in representatives of the two basic chromo some numbers.
In old polyploid complexes only polyploid levels are observed (Grant 1981).A few species (P.airoides subsp.jugorum ; P argentea\ P. aristifolia; P barbata subsp.barbata\ P. colorata\ P. lima; P malouinensis; P mannii\ P. minor; P. aff.patula\ P rupestris & P. viscidula), were classified as old polyploid complexes.It is important to note that in almost all the old polyploid species, only one or a few specimens were examined.More specimens should be studied to verify these hypotheses.The classi fication of young polyploid complexes is better support ed, since more specimens were examined.Representa tives of both the basic chromosome numbers (x = 7; x = 13) were classified as young polyploid complexes, with sim ilar frequency ratios (Figure 4).The ploidy levels obtained in this study, combined with existing data, sug gest that the genus Pentaschistis is a young polyploid hybrid complex (Figure 4).Hybridisation could have resulted in the formation of a basic chromosome number of x = 13 for P eriostoma and P borussica.However, morphological and anatomical data (Linder & Ellis 1990;Linder et al. 1990;Ellis & Linder 1990) supports the inclusion of P. borussica in the genus Pentaschistis.It is thus more probable that polyploidisation and subse quent aneuploidy of a Pentaschistis species, or hybridis ation between two different Pentaschistis species and subsequent aneuploidy, resulted in P. borussica.This is not true for P. eriostoma.Morphological and anatomical data placed P. eriostoma in group 6.According to Linder & Ellis (1990) and Ellis & Linder (1990) this is the group of species with uncertain affinity.It is more probable that P. eriostoma was formed by hybridisation of two unre lated species.Genomic in situ hybridisation (GISH) analysis, using various putative parental genomes as probes, should be used to determine the origin of the x = 13 basic chromosome number.
Univalents were observed in nearly all of the studied specimens (Figure 6A-F).Lagging chromosomes were observed during anaphase I in P. aristidoides, P. curvifolia, P. eriostoma, P lima, P. pallida, P. rigidissima, P. rupestris, P. tomentella, P viscidula and one of the two unnamed Pentaschistis species (Figure 7A).A small number of micronuclei (Figure 7E-H) was observed throughout the whole genus Pentaschistis.Anaphase I or II bridges (Figure 7B-D) were observed in five of the specimens studied.This is most probably the result of paracentric inversion.The frequency of these abnormal ities was so low that it should not affect the fertility of the specimens.
In P. pallida (Spies 4406), a diploid (n = 7+0-1B), five bivalents and one quadrivalent were observed.In this diploid, seven bivalents are expected.A quadrivalent was also observed in a diploid specimen of P. eriostoma (Spies 5370).This phenomenon can be the result of a balanced translocation.If complete pairing between the two nonhomologous chromosomes occurred, a quadrivalent would result.Three types of division are possible during anaphase I.If adjacent type I or II division occurred, there would be no fertile pollen.During adjacent type I division, the two centromeres belonging to non-homologous chro mosome pairs move to the one pole and the other two to the other pole.During adjacent type II division, the two centromeres from a homologous chromosome pair move to each pole.If alternate type division occurred, all the pollen would be fertile.During alternate type division, the chiasmata of all the chromosome pairs lie on the metaphase plate and the centromeres of the non-homolo gous pairs move to the same pole (Schulz-ShaefTer 1980).
The fertility of P. pallida (Spies 4406) and P. eriostoina (Spies 5370) pollen is thus dependent on the type of division that occurs.All the material in our laboratory is fixed after collection, so no pollen germination fertility test could be done.More specimens of these two species must be collected to study whether this phenomenon occurs frequently or if it was a unique event.
In conclusion, although cytogenetic differences were observed between the different species, the cytogenetic data did not support or reject the groupings suggested by Linder & Ellis (1990) and Ellis & Linder (1990).The basic chromosome numbers of x = 7 and x = 13 suggest that P. eriostoma and P. borussica must be further removed from the other Pentaschistis species.Since P. eriostoma shows little cytogenetic, morphologic or anatomic similarity to the other species, we do suggest that P. eriostoma is not closely related to the other species of the genus Pentaschistis.The data also indicate that most Pentaschistis species form young polyploid complexes.Species classified as old polyploid complex es were usually inadequately studied.

TABLE 1 .
-List of Pentaschistis specimens studied with their respective gametic chromosome numbers, ploidy levels, figures where they are shown, localities and voucher specimens (cont.) 1992).The basic chromosome number of seven predomi nates in the southern African species, whereas x = 13 pre vails in the remainder of Africa (Du Plessis & Spies 1992).

TABLE 2 -
Genomic relationships in tetraploid Pentaschistis specimens according to the models of Kimber & Alonso (1981) Values indi cate sum of squares between observed and expected values for each model; whereas values in parentheses indicate x-values for each model a.