Leaf anatomy of the South African Danthonieae (Poaceae). XV. The genus Elytrophorus

The leaf anatomy of Elytrophorus globularis Hack, and E. spicatus (Willd.) A. Camus is described and illus­ trated from freshly fixed material from SWA/Namibia and Botswana. It is shown that these two species are ana­ tomically indistinguishable. It is suggested that they are conspecific. and that E. spicatus possibly represents juve­ nile plants with immature inflorescences. The anatomical evidence strongly refutes a chloridoid relationship for E lytrophorus but appears to support arundinoid affinities for the genus. Striking anatomical and ecological simila­ rities exist between Elytrophorus and Sacciolepis huillensis (Rendle) Stapf. N o significant leaf anatomical differ­ ences separate Elytrophorus from 5. huillensis and some of the other C3 panicoid taxa and. consequently, Elytro­ phorus may represent a link between the Arundinoideae and the Panicoideae.


INTRODUCTION
Elytrophorus Beauv. is a genus of unusual little grasses found in tropical Africa, India to South China and Australia, with the centre of distribution apparently in tropical Africa.The genus is therefore *Botanical Research Institute.Department of Agriculture and Water Supply.Private Bag X101, Pretoria 0001.
restricted to warm tropical areas of the Old World surrounding the Indian Ocean.Some authors have distinguished four species in the genus (Loxton 1976;Schweickerdt 1942).Other workers uphold only two species: E. globularis Hack, and E. spicatus (Willd.) A. Camus (Chippindall 1955;Clayton 1970;Smook & Gibbs Russell 1985).Both these species occur in southern Africa where they are restricted to the tropical northern most parts of the region.In SWA/Namibia, Elytro phorus is found in Ovamboland and the Grootfontein.Okahandja and Caprivi Districts, and in Bo tswana it occurs in the Mababe Depression and the Okavango Delta of Ngamiland.E. globularis has also been collected at Mosdene along the Nyl River in the Naboomspruit District of the Transvaal.
Both species are water-loving and are found ex clusively on the edges of rainwater pans, ponds, de pressions and in rice fields, particularly on the pe riphery of these shallow water bodies when moist mud is exposed as the water evaporates and recedes.Damp hydromorphic clay soil is preferred and the plants even thrive in the cracking clay.Elytrophorus can withstand a certain degree of inundation and can survive in standing water up to 0,2 m deep and is considered to be a true hydrophyte (Schweickerdt 1942).In ideal situations Elytrophorus can form dense communities, the individual plants varying in height from 10 mm to 0,5 m, depending on the pre vailing moisture conditions.
Elytrophorus exhibits an unusual combination of anatomical features which have been described by Schweickerdt (1942) , Jacques Felix (1962), Clifford & Watson (1977) and Palmer & Tucker (1981).The objective of this paper is to describe and illustrate the leaf blade anatomy of both species and to relate this to the anatomical diagnoses of the subfamilies of the Poaceae as defined by Clifford & Watson (1977) and Renvoize (1981).The natural relationships of Elytrophorus are not readily apparent and agrostologists differ as to which subfamily and tribe this genus should be assigned to.The anatomical evi dence will be fully discussed in an attempt to resolve this question.

MATERIALS AND METHODS
Plants of Elytrophorus were collected in SWA/Na mibia and Botswana during the late summers (April or May) of 1977, 1981 and 1983.Herbarium voucher specimens were prepared for verification by the Na tional Herbarium (PRE).Segments of leaf blade material were removed in the field and immediately fixed in FAA (Johansen 1940).
Transverse sections, 10^ thick, were prepared after desilicification in 30% hydrofluoric acid (Breakwell 1914), dehydration following the method of Feder & O'Brien (1968) and infiltration and em bedding in Tissue Prep (Fisher Scientific).These sections were stained in safranin and fast green (Jo hansen 1940).The manual scraping method of Met calfe (1960) was used to prepare scrapes of the abax ial leaf epidermis.These were either stained in safra nin or double-stained in methylene blue and ruthe nium red.The anatomical structure was recorded photographically using a Reicherdt Univar micro scope and Ilford Pan F film (50 ASA).
In the anatomical descriptions which follow, the standardized terminology of Ellis (1976Ellis ( , 1979) ) will be used, together with the following abbreviations:

Differences between the species of Elytrophorus
In his treatment.Schweickerdt (1942) considered the genus Elytrophorus to comprise four species: E. globularis and E. spicatus as well as E. africanus Schweick.and E. interruptus Pilg.The latter two species are now considered to be synonyms of E. globularis (Clayton 1970) and consequently, the present study includes the two species which repre sent all currently recognized members of the genus.
The anatomical details described by Schweickerdt (1942) agree closely with the observations of this study.The only significant departure concerns the mention of aerenchymatous cells traversing the lacu nae.Schweickerdt considers this tissue to be a strik ing characteristic of the genus.In the present study, however, no aerenchyma or stellate cells were ob served in any of the ten specimens examined.Most specimens had lacunae located between the vascular bundles (Figures 3-6) but these cavities were never seen to be traversed by colourless aerenchyma cells.In addition, in some specimens of both species, the lacunae were not even fully developed (Figures 1 &  2, 13 & 14) although the central mesophyll between the vascular bundles was more diffuse, with larger intercellular air spaces appearing to represent the in itial stages of cellular breakdown prior to the forma tion of the typical lacunae.If this is so.then the lacu nae of Elytrophorus are lysigenous cavities arising by the dissolution of entire cells during the later ontoge netic stages of the leaf.The replacement of these broken down mesophyll cells by aerenchyma cells at this late stage of leaf differentiation appears un likely.Schweickerdt (1942) mentions aerenchyma tissue in all four species he studied although the cells are only illustrated in E. spicatus.The specimens exam ined by him were prepared from dried herbarium material and he remarks that tissue recovery was not satisfactory and this is reflected in his camera lucida drawings.His details and measurements of the softer tissues, in particular, may not necessarily be reliable and accurate.
This may also explain another difference between the findings of this study and those of Schweickerdt (1942).In the latter study diagnostic anatomical dif ferences were detected between E. globularis and E. spicatus whereas in the present study no differences were observed.According to Schweickerdt (1942), E. spicatus is characterized by having a leaf blade of 0,3-0,36 mm thick, with both adaxial and abaxial ribs and furrows and with bulliform cells between all bundles.E. globularis is said to differ in having a thinner blade (0.15-0.28 mm), neither surface being ridged, and well developed bulliform cells only oc curring in the region of the midrib.This interspecific variation appears to be contradicted by the illustra tion of E. globularis (Jacques Felix 1962), which shows large triangular adaxial ribs as w'ell as lacunae, whereas an illustration of E. spicatus (Clifford & Watson 1977) shows neither ribs nor lacunae.In the above studies only a single specimen from each species was examined and intraspecific variation could not be ascertained with much confidence.In the present study, however, variation has been shown to occur within each of the species and the sample of each species studied exhibited as much va riation as that considered by Schweickerdt (1942) to justify separation of the two species.Examples are.as mentioned, the differences in adaxial rib and la cuna development in different specimens of E. glob ularis (Figures 3 & 5).
The present study shows that E. globularis and E. spicatus are indistinguishable on anatomical grounds.These results, and the observation that both species may occur at the same locality at the same time, throw doubt on the validity of upholding two separate species.E. spicatus may merely repre sent juvenile plants with younger or immature inflo rescences.This hypothesis requires testing.

General
The classification of Elytrophorus has been the subject of much debate in the literature.Some authors consider it to belong to the Chloridoideae.and Chippindall (1955) and Bor (1960)  playing a curious mixture of festucoid and panicoid features as well.Prat (1960) and Decker (1964) placed Elytrophorus in their unplaced genera al though Decker noted similarities with the Danthonieae.Jacques Felix (1962) isolated the genus in a separate tribe, the Elytrophoreae, belonging to his series the Arundinoidae.This classification has been upheld by most modern authors and Elytrophorus is usually assigned to the Arundinoideae in the tribe Danthonieae (Clayton 1970;Loxton 1976) or the tribe Arundineae (Renvoize 1981).Renvoize (1981) considers Elytrophorus to conform closely to the coherent arundinoid core group which is virtually synonymous with the Danthonieae.

Affinities with the Chloridoideae
Evidence from leaf in transverse section Elytrophorus has a double bundle sheath, as do the chloridoid grasses, but the outer sheath is thinwalled and non-Kranz, lacking specialized chloroplasts.It is therefore a C3 genus.as is confirmed by 13C/12C ratios for E. globularis of -26,23% (Dinter 7390) and E. spicatus of -25,70% 2089).As far as is known, all chloridoid grasses are C4w'ith only one possible exception (Ellis 1984) and therefore, have strongly radiate mesophyll and a maximum lateral cell count of less than four.In Ely trophorus this count is greater than 10 and, although the mesophyll displays a tendency to be radiate, it is of the Isachne type (Metcalfe 1960) with several lay ers of elongated, diffuse cells with many air spaces all arranged in a somewhat radiate manner (Figures 2 & 14).This type of mesophyll is unknown in the Chloridoideae where a single layer of compact, tabu lar cells surrounds each bundle.In the chloridoid type of anatomy the bulliform cells are usually asso ciated with deeply penetrating fans of colourless cells, whereas in Elytrophorus none of these colour less cells occur.The evidence from leaf transections does not indicate a chloridoid connection for Elytro phorus.

Evidence from abaxial epidermis
Elongated microhairs with short basal cells, and much longer, tapering distal cells are common in Elytrophorus and were observed on all specimens examined in this study.The structure of these micro hairs is illustrated in the accompanying photomicro graphs (Figures 8,10,11,12,15 & 16) and even more clearly in the scanning electron micrographs of Palmer & Tucker (1981).This structure differs sig nificantly from the chloridoid type which is always egg-shaped with shorter, inflated distal cells (Clif ford & Watson 1977;Renvoize 1981).Elytrophorus also lacks long cells with sinuous walls which are typ ical of chloridoid grasses.The subsidiary cells of Ely trophorus are dome-shaped (Figure 10) or low dome-shaped (Figures 8,12 & 15), whereas in the Chloridoideae they are predominantly triangular.Chloridoid grasses often have papillate epidermides whereas Elytrophorus does not, at least at the level of resolution of light microscopy.Palmer & Tucker (1981) illustrate tiny, warty papillae visible only at higher magnifications with the scanning electron microscope but these are, nevertheless, not of the chloridoid type.The horizontally elongated nodular to sinuous type of silica body found in Elytrophorus is unknown in the Chloridoideae where silica bodies are not elongated and are usually saddle-shaped, but may be cross-shaped, square or shortly dumbbell shaped.In epidermal structure, therefore, Elytro-phorus bears no resemblance whatsoever to the chloridoid condition.
Leaf anatomical evidence, therefore, does not support chloridoid phylogenetic affinities for Elytro phorus and the classification of this genus in the Chloridoideae cannot be supported.

A ffinities with the Arundinoideae
Other workers place Elytrophorus in the rather illdefined subfamily Arundinoideae (Jacques Felix 1962;Clayton 1970;Renvoize 1981).This subfamily cannot be defined as precisely as the other four sub families and lacks reliable diagnostic features.Nevertheless, a diagnosis of the Arundinoideae is possible (Clifford & Watson 1977;Renvoize 1981) and. in most respects Elytrophorus conforms very well to this definition.
Arundinoid microhairs are finger-like with taper ing distal cells, the subsidiary cells are domed and the epidermis is not papillate -all characteristics of the epidermis of Elytrophorus.There are points of difference, however, where Elytrophorus does not conform to the arundinoid definition.The straightwalled long cells of Elytrophorus are an example, as are the nodular to sinuous or crenate silica bodies.These character states are more typical of the festucoid subfamily but they are not unknown in the Arundinoideae.The silica bodies of the arundinoid grasses are horizontally elongated and may be nodu lar.cross-or dumbbell-shaped and.consequently do not differ greatly from the Elytrophorus condition.In epidermal structure, therefore.Elytrophorus gen erally resembles the arundinoid type closely.
In leaf transverse sections the same is true and Elytrophorus diverges little from the arundinoid type.Both are non-Kranz (with a few arundinoid ex ceptions).have double bundle sheaths and non-radiate or slightly radiate mesophyll and have a maxi mum lateral cell count greater than four.In addition the arundinoid grasses are also characterized by hav ing adaxial ribs, as does Elytrophorus.Bulliform cells not associated with colourless cells is another characteristic common to both these groups.The leaf anatomy of Elytrophorus, as seen in transverse section, therefore, conforms very closely to the arundinoid type and there is no anatomical evidence for excluding Elytrophorus from this subfamily.The leaf anatomical data of this study support the classifi cation of Elytrophorus in the Arundinoideae.as re commended by most modern authors.

A ffinities with Sacciolepis and other C? Panicoideae
Although the above evidence may be convincing, other factors suggest caution in postulating arundi noid affinities for Elytrophorus.A similar distribu tion in hot.tropical areas is unknown in the other C3 South African Danthonieae as discussed by Ellis et al. (1980).Apart from the ubiquitous Phragmites, Elytrophorus is the only C3 arundinoid grass found in the northern tropical regions of southern Africa.In the hydrophytic habitats favoured by Elytrophorus the only other C^ grasses belong either to the Oryzeae or the Paniceae.Little anatomical resemblance exists between the oryzoid grasses and Elytrophorus but the C3 type of panicoid anatomy of genera such as Acroceras and Sacciolepis and the leaf anatomy of Elytrophorus show striking similarities.Sacciolepis huillensis (Rendle) Stapf. in particular, is indisting uishable from Elytrophorus in leaf anatomy.These grasses share an identical habitat and physiognomy and the S. huillensis specimens examined in this study (Ellis 3716 & 3717) were collected together with E. spicatus (Ellis 3718) at the same locality at the same time.This observation may.or may not.be significant and deserves further discussion.

5.
huillensis has nodular silica bodies, no intercos tal short cells, long cells with straight or only slightly sinuous anticlinal walls and domed stomata.The microhairs are also elongated with a tapering distal cell, although the basal cell is slightly larger than that of Elytrophorus.In transection the anatomy of both species is virtually identical except, perhaps, that 5. huillensis displays a tendency for the leaf to be some what thicker in the midrib.The work of Nixon (1953) confirms this anatomical structure for S. huil lensis.
The anatomical resemblance between these two taxa.presently classified in two different subfami lies.may reflect convergent evolution in response to identical habitats, but the resemblance may also be phylogenetically significant.The leaf anatomy of Elytrophorus is strikingly similar to that of many of the C, panicoid taxa.This type of anatomy is fully described in Ellis (1986) and evaluated in relation to the panicoid grasses.The only anatomical differ ences between Elytrophorus and many of these pan icoid species are the very sinuous long cells of the C3 forest species in particular and the elongated but dumbbell-shaped silica bodies, although the nodular type may occur in some of these species.

Conclusions
The anatomical indications of this study are that Elytrophorus should possibly be assigned to the group of C3 panicoid taxa rather than to the Dantho nieae.The advisability of Elytrophorus being trans ferred to the Paniceae on morphological grounds needs to be carefully examined.It is of interest to note that the embryo of Elytrophorus is panicoid (Jacques Felix 1962) but the chromosome number of x=13 is most unusual for the Poaceae.In addition, the large group of C3 panicoid grasses, including many species of Panicum as well as other genera, may warrant recognition at suprageneric level as they all share a similar basic leaf anatomy not found elsewhere in the Panicoideae.This group, together with Elytrophorus, may represent a primitive pani coid group forming a link between the Arundinoi deae and the Panicoideae.
Elytrophorus is.therefore, a most interesting genus from a phylogenetic viewpoint and further stu dies on this genus may even help elucidate some as pects of evolution in the Poaceae as a whole.A bet ter understanding of the systematics and taxonomy of Elytrophorus should help clarify our concepts of the grass subfamilies and their interrelationships.