Stomatal area as an anatomical criterion for the determination of chromosome number in the Eragrostis curvula complex

Twenty stomatal areas of each of 55 Eragrostis curvula (Schrad.) Nees plants were determined. An increase in polyploid level is shown to be moderately correlated with an increase in stomatal area. However, the extent of overlap in stomatal areas between different polyploid levels is too great to use this character for the determination of the polyploid level above the diploid level. All diploid E. curvula plants have an area of less than 280 /x.2, whereas the tetraploid plants have areas greater than 320 /x2. It is therefore possible to identify diploid E. curvula plants on the basis of their stomatal area.


INTRODUCTION
A correlation exists between cell size and the ploidy level o f the cell as polyploid cells are usually larger than the cells o f th eir d ip lo id counterparts (Stebbins, 1950;A lla rd , 1960).This fact has led cytogeneticists to search fo r suitable cells which could be used to determine chromosome num ber accurately w ith ou t a study o f the chromosomes.Blakeslee & A ve ry (1937) used pollen size to distinguish between different polyploid levels in Datura stramonium.De W et (1954) used stomatal length as a criterion fo r a sim ilar study in the genus Danthonia.These studies have not always given the desired results.The most successful results have always been obtained from induced po lyplo id y where the chromosome number o f a single plant has been altered by the use o f chemicals such as colchicine (Blakeslee & A ve ry, 1937;Biswas & Battacharyya, 1976).
The aim o f the current investigation was to determ ine to what extent this m ethod could be used fo r predicting chromosome numbers in the Eragros tis curvula (Schrad.)Nees complex.

MATERIAL AND METHODS
Plants o f the E. curvula complex were collected in the fie ld throughout South A fric a and were transplanted in the Pretoria N ational Botanical Garden (V o rste r, 1978).The chromosome numbers o f these cultivated plants were determ ined by V orster (1978).
Leaves o f 55 o f these plants (Table 1)

RESULTS AND DISCUSSION
The average stomatal area o f each plant is shown in Table 1.These stomatal areas were plotted graphically (Fig. 1).A correlation factor o f 0,6774 was found to exist between chromosome num ber and stomatal area fo r each plant.This correlation is represented graphically in Fig. 1.
From Table 2 it is clear that the only homogene ous group in this study was the d ip lo id (2n = 20) group.This group showed a standard deviation o f 21,79.From Fig. 1 it is evident that the d ip lo id group is the only group w ith almost no overlap w ith any other group.The only overlapping that occurred was between the diploid group and one pentaploid plant (Vorster 334).D u rin g a cy to g e n e tic a l re examination that follow ed this anatom ical study, it was found that this plant was indeed a d ip lo id and not a pentaploid.A s the V o rste r collection was made and transplanted before 1978, the possibility exists that this plant could have reverted to a diploid plant o r that the plant was w rongly numbered.
A m eiotic analysis o f the unpaired chromosomes o f Vorster 334, showed an average o f 5,75 unpaired chromosomes per cell (V o rs te r, 1978).U n fortu n ate ly no embryosac analysis was done on this plant during the present study.I t is clear, therefore, that this is a very unstable plant and, although the degree o f apom ictic reproduction is not known, the chromosome num ber o f this plant could have been altered in various ways.From these results it appears that it can be predicted w ith a high degree o f confidence that any E. curvula plant w ith a stomatal area o f less than 280 /j? w ill be diploid.
A n unexpected high standard deviation was encountered in the te tra p lo id (2n= 40) group.This results from the fact that the tetraploid sample can be divided into tw o groups and a big gap o f more than 50 (x2 exists in the area measurements between Vorster 422 and Vorster 989.The first group corresponds w ith the regression line drawn in Fig. 1.The second group (Vorster 989, 750, 303, 797 and 210) have very large stomatal areas.The average stomatal area (502,37 /a2) o f the second group is almost as large as the average stomatal area (510,41 fi2) o f the octoploid (2n =80) group.It is possible that this m ight be a case o f divergent evolution occurring, or that two d iffe re n t sources exist from which these tw o differen t groups have evolved.It  should prove interesting to have a closer look at m orphological differences between these tw o te traploid groups.It may also be significant to note that all plants in this 'abnorm al' group were originally collected in mountainous areas.
A large degree o f variation (standard deviation = 87,99) exists in the pentaploid (2 n= 5 0) group.Because the pentaploid group tends to be genetically unstable (V orster, 1978), it w ill be necessary to undertake a cytogenetical study on this group again, to determine whether chromosome instability is responsible fo r this excessive variation.
The abnorm ally large stomatal areas found in some o f the tetraploid and pentaploid plants are responsible fo r the fact that the hexaploid group has a relative lower average stomatal area than both above-mentioned groups.The hexaploid (2n=60) group is relatively homogeneous w ith regard to the stomatal area (standard deviation = 44,59) and this fact is also reflected in th e ir relative norm al m eiotic behaviour where an average o f 1,64 unpaired chromosomes per cell were found (V o rste r, 1978).
W ith the exception o f Vorster 854, the heptaploid (2n=70) group shows relatively little variation.This fact corresponds w ith the observation that hepta ploid plants in this study have an average o f 4,52 unpaired chromosomes fo r each m eiotic cell (V orster, 1978).Pentaploid plants showed an average o f 5,71 unpaired chromosomes per cell.The heptaploid plants are, therefore, genetically more stable than the tetra p loid plants.This fact is clearly reflected in this study by the more lim ite d variation o f stomatal areas o f the heptaploid plants in comparison w ith the pentaploid plants.
In the octoploid (2n = 80 ) group only one deviation from the norm al acceptable d istribution was found.This plant (Vorster 242) was studied cytogenetically and found to be tetra ploid (2n=40).This might be due to secondary haploidization or a wrong number m ight have been used somewhere during different collecting and transplanting opera tions.

CONCLUSIONS
From the results o f this study, it is clear that an increase in stomatal area is correlated w ith an increase in polyploid level (Fig. 2).H ow ever, the extent o f overlapping o f stomatal areas between d ifferent polyploid levels o f E. curvula is too great fo r the prediction o f the polyploid level by this means.The only exception in this regard is the diploid group.A ll d ip lo id plants had areas o f less than 280 n 2, whereas the other polyploid levels had values above 320 /a2.
The only published example where cell size has been used directly to determ ine polyploid level, is w ith induced polyploidy (Blakeslee & A ve ry, 1937;Biswas & Battacharyya, 1976).The present results indicate, therefore, that it m ight be possible to use cell size as a crite rion fo r determ ining chromosome numbers, but only when w orking w ith very closely related plants exhibiting little m orphological diver gence and originating from sim ilar environments.The De W in ter collection included in this study, was sampled from a small area and consists o f m orphologically almost identical plants but having different chromosome numbers.This group shows a correlation coefficient o f 0,953.It m ight therefore be possible to use this method successfully i f the follow ing selection crite ria are strictly applied:

INTRODUCTION
The construction o f Richards Bay H a rbo u r resulted in the Richards Bay Lagoon (28°49' S and 32°05' E ) being divided in to a harbour zone and a natural sanctuary by the construction o f a berm w all and the opening o f a new m outh fo r the nature reserve during 1975.Begg (1978) reports on the many abiotic and bio tic changes occurring after the h a b ilitio n o f the Mgobezeleni Lake (near Sodwana Bay) fo llow in g the construction o f a bridge that impeded tid a l flo w and caused the swamps to be flooded.D uring studies on N atal estuaries (W ard, M S) and on the conservation p rio ritie s between Richards Bay and M lalazi M outh (Weisser, M S) the vegetation o f the southern shores was rem apped, sampled w ith eight releves and compared w ith findings o f V en ter (1972),   M any Phoenix reclinata trees were cut by Z u lu people living nearby to tap the sugar-rich sap.The damaged plants coppice freely fro m the base to form dense thickets.
The Barringtonia racemosa C om m unity was quantitatively studied by V en te r (1972), who took samples at three places nearby: at M zingazi L a k e , west were sampled during M arch 1981.A fte r 24 hours fixatio n in F .A .A .fixative, epiderm al scrapes o f the abaxial epidermis were prepared fo llo w ing the method o f M etcalfe (1960) w ith slight m odifications.Safranin and M ethylene Blue were used fo r staining the epiderm al scrapes.T w enty stomata from each plant were measured.Measurements o f the stomatal area were made from photomicrographs.A K o ntron M 0 P -A M 0 2 image analyser was used fo r determ ining the stom atal area.* Botanical Research Institute, Department of Agriculture and Fisheries, Private Bag X101, Pretoria, 0001.
Fig. 1.-Correlation between chromosome numbers and stomatal areas in selected representatives of the Eragrostis curvula complex.
1) C ollect only m orphologically sim ilar plants 2) C ollect only in one geographical area

Fig
Fig. 1.-Remnants of Phoenix!Hibiscus and Barringtonia ra cemosa Communities on southern shore of Richards Bay destroyed by increased tidal range and salinity following opening of new mouth at Richards Bay.In background landward side of Bruguiera gymnorrhiza Zone.
This suggests a landward extension o f the M angrove C om m unity.Open parts o f the destroyed zone were colonized by epipelic ( = growing on m ud) algae.V en ter (1972) described the Phoenix!Hibiscus C om m unity p rio r to its destruction as occupying a small zone im m ediately behind the Bruguiera gymnorrhiza C om m unity at the south-eastern shore o f Richards Bay.Landw ard it bordered on the Barringtonia racemosa C om m unity (see Fig. 3 in V enter 1972) and on prim ary or secondary Dune Forests.It was a dense th icket form ed m ainly by the palm Phoenix reclinata, Hibiscus tiliaceus and the fern Acrostichum aureum.O ther species present were Rapanea melanophloeos, Ficus trichopoda, F. capensis, Syzygium cor datum, Bruguiera gymnorrhi za and Avicennia marina.The last tw o species were in poor condition and were probably remains o f the neighbouring Bruguiera gymnorrhiza C om m unity which, in the course o f tim e, was apparently displaced by the Phoenix!Hibiscus C om m unity.Clim bers were Mikania cordata, Ipomoea cairica, I. congesta, Dioscorea sylvatica and Rhus nebulosa, the firs t tw o species being pa rticu larly common.Field layer species recorded were Acrostichum aureum, Nidorella auriculata, Phragmites australis, Typha latifolia subsp.capensis, Cyperus alter nifoliusy Scadoxus magnificus and Blumea lacera.

F
i g .2.-View across part of de stroyed zone.In this area Phragmites australis and Bruguiera gym norrhiza adapted well to changed conditions and are now increasing.

TABLE 2 . -Average stomatal area (fi2) of each plant for the polyploid levels of the Eragrostis curcula complex
U IT T R E K S

TABLE 1 .-Eight relevds of the Bruguiera gymnorrhiza and Phoenix!Hibiscus Communities in the study area. The values correspond to Braun-Blanquet cover-abundance estimates (Braun-Blanquet 1964). The plots 25 m2, were approximately level, with muddy substrate and usually with stagnant water (dated 1980.11.20-24)
Phoenix reclinata and Acrostichum aureum were found to occur near the southern corner o f the bay, on slightly higher parts farthest fro m the sea (Table1, releve 445).A plant that has lately increased in the affected area is Phragmites australis w hich, in places, form s dense stands between the dead trunks and debris (Fig.2and Table1,Releves 441,442,443,446).