Leaf anatomy of the South African Danthonieae (Poaceae): XIX. The genus Prionanthium

The leaf blade anatomy of Prionanthium dentatum (L. f.) Henr.,  P. ecldoni i (Nees) Stapf and  P. pholiuroides Stapf is described and illustrated. The transectional anatomy is non-Kranz with diffuse chlorenchyma. The adaxial epidermis has dome-shaped stomata, dumbbell-shaped to nodular silica bodies and elongated microhairs. The three species differ in the presence or absence of macrohairs, abaxial stomata and marginal linear, sessile glands. The latter character is shared with Pentaschistis triseta (Thunb.) Stapf and its allies, and relationships with  Pentaschistis are indicated rather than with any other extant arundinoid genus.


INTRODUCTION
The genus Prionanthium is a small genus of three species all endemic to the south-western Cape Province. It includes small, ephemeral or annual plants which are exceedingly difficult to locate and are considered to be rare and/or endangered (Davidse 1988). Taxonomically the genus does not pose any particular difficulties and it contains three morphologically distinct species: P. dentatum (L. f.) Henr. ( = P. rigidum Desv.), P. ecklonii (Nees) Stapf and P. pholiuroides Stapf. These three species are recognized by most previous workers (Chippindall 1955;Clayton & Renvoize 1986;Watson et al. 1986;Davidse 1988) although all these workers did not have access to material of all three taxa.
The recent revision by Davidse (1988) included all the anatomical voucher specimens used for the present study and the anatomical data reported on here is complementary to the morphology and cytology discussed in this revision. The present study is the first in which comparative leaf anatomy of all three species has been undertaken.
Although there is little difficulty with the delimitation of taxa within Prionanthium, the generic relationships of the genus are puzzling. Most authors now agree with the placement of the genus in the Arundinoideae, usually in the tribe Arundineae (= Danthonieae). Davidse (1988) convincingly argues the case for arundinoid relationships and, furthermore, demonstrates that Prionanthium and Pentaschistis are probably related. This conflicts with the suggestion of Clayton & Renvoize (1986)  belongs with the primitive arundinoid genera such as Tribolium, Urochlaena, Elytrophorus and others from Australasia. The phylogenetic implications o f the leaf anatomy of Prionanthium w ill be considered in this paper in order to shed more light on this problem. The purpose o f this paper is, therefore, to describe and illustrate the leaf blade anatomy of Prionanthium and to compare this structure with that of the other South African danthonioid species, as well as with earlier published results for Pri onanthium.
Little detailed attention has been given to the leaf anatomy o f Prionanthium in the published literature but the information available does indicate typical arundinoid anatomy. The anatomy is non-Kranz with a double bundle sheath, non-radiate mesophyll with a maximum lateral cell count greater than four, bulliform cell groups not associated with colourless cells, finger-like microhairs with tapering distal cells, domed subsidiary cells and dumbbell-shaped to nodular silica bodies (Hansen & Potztal 1954;Renvoize 1986;Watson et al. 1986). Renvoize (1986) notes two unusual features for the subfamily: the chlorenchyma is not interrupted by sclerenchyma girders above and below the vascular bundles, and the epidermal cells are large and thin-walled. De Wet (1956) notes that the epidermis of Prionanthium is panicoid (with linear microhairs) but that the anatomy is festucoid with the chlorophyll tissue being uniformly distributed between the bundles.

MATERIALS AND METHODS
Specimens of all three species of Prionanthium were collected in the south-western Cape. Herbarium voucher specimens were prepared for verification by the staff of the National Herbarium (PRE) where they are housed. These same specimens were also examined by Davidse for a revision o f the genus (Davidse 1988).
Central segments o f leaf blades were removed and immediately fixed in FA A. Transverse sections and epidermal scrapes o f these blade segments were prepared following the methods outlined in previous papers in this series (e.g. Ellis 1988).
The standardized terminology o f Ellis (1976Ellis ( , 1979 ; 2C); probably constitute part of marginal linear glands which appear to vary from 3 -7 cells wide and are located on a distinctly narrow and projecting margin. These structures are also visible in surface view ( Figure 1G, H) where the persistent nuclei are clearly visible. Figure  4A, C, E); end walls vertical; walls unthickened to slightly thickened. Cell shape and size consistent across intercostal zones; long cells adjoin one another or separated by short cells; no bulliform cells. Stomata: usually absent in P. pholiuroides ( Figure 3E, F Figure 4F); hairs longer than stomatal length; distal cell deciduous and thin-walled. Macrohairs: usually absent but always present in P. dentatum (Figure 1 E -H ); rarely present in P. pholiuroides-, unicellular, soft hairs with single inflated, hemispherical cell associated with the superficial base. Silica bodies: angular, irregular dumbbell-shaped to nodular; horizontally elongated; exclusively costal in very narrow (1 -3 files) costal zones; separated by costal short cells.

Differences between the species o f Prionanthium
The three species of Prionanthium can readily be distinguished from one another on anatomical criteria as summarized in Table 1. P. dentatum is unique in the genus in possessing macrohairs but it otherwise resembles P. ecklonii very closely in leaf anatomy: both species have similar outlines, ribs and furrows, abaxial stomata and microhairs. They also share the very unusual marginal, sessile glands which are currently only known in four or five species of Pentaschistis. In addition to the macrohairs, P dentatum and P. ecklonii also differ slightly in the arrangement of the vascular bundles and in the extent of the development of the sclerenchyma strands. P. dentatum and P. ecklonii, therefore, appear to be more closely related to each other than either is to P. pholiuroides. This agrees with the morphological evidence, particularly that of the glume glands where P. dentatum and P. ecklonii share similar stalked multicellular glands, whereas in P. pholiuroides the glands are sessile (Davidse 1988).

P. pholiuroides is distinct in the genus on the basis of a number o f features: the leaf outline is inrolled to
infolded; the adaxial furrows are narrow and cleft-like; there are fewer large vascular bundles in the leaf transection which is, consequently, narrower; the outer bundle sheath is often not entire but interrupted abaxially; the sclerenchyma strands may be conspicuous; the margin is not tapering and lacks the linear glands and abaxial stomata are absent or very rare. P pholiuroides can, therefore, be readily separated from both P. dentatum and P. ecklonii on leaf anatomy and this study fully supports the recognition of these three species in the genus.
The results presented here must be interpreted with caution as only single populations of both P. dentatum and P. ecklonii were examined in this study due to the rarity o f all the species. In the case o f P. dentatum the same population was sampled on three different occasions -in the spring of 1975, 1987 and 1988. All these specimens are virtually identical in leaf anatomy and very little variation was exhibited. However, two populations of P. pholiuroides were included in the study and they exhibit obvious ana tomical variation as illustrated in Figure 3. The Mamre population has more open leaves with smaller adaxial sclerenchyma strands than the Struisbaai population. Although these differences are visually rather striking, they do not represent significant differences, and characters such as vascular bundle number and arrangement, the cleft-like furrows and the absence of abaxial stomata re main consistent. It is of relevance to note that the Struis baai population also differed cytologically from the Mamre population in the occurrence o f a small B chromosome and this population was also the most variable in spikelet morphology (Davidse 1988). Nevertheless, it appears like ly that most of the anatomical features separating the spe cies will hold even when more material becomes available.   (Davidse 1988) and Prionanthium appears best placed in the Arundineae o f the Arundinoideae.
Within the Arundineae the generic relationships of Prionanthium are subject to differing interpretations. Clayton & Renvoize (1986) are of the opinion that Prionanthium, and in particular P. dentatum , suggests a relationship with Tribolium and the other primitive arundinoid genera. Leaf anatomical characters can be used to substantiate such a relationship, particularly as all the primitive South African arundinoid taxa display similar mesophyll characteristics-which will here be termed the orthophyllous or m esic anatomical type. This differs significantly from the sclerophyllous type found in genera such as Merxmuellera and Pentameris. These two basic arundinoid anatomical types are discussed in Ellis (1988a) and Prionanthium shares many anatomical similarities with primitive genera such as Tribolium and Urochlaena (Ellis 1988a). However, other arundinoid genera not included in the heterogeneous group of primitive genera (Clayton & Renvoize 1986) also have this orthophyllous type of anatomy. Chaetobrom us (E llis 1988b), Schism us, Karroochloa and some species of Pentaschistis are examples and, therefore, the possession of the m esic type of arundinoid anatomy does not necessarily appear to indicate affinity.

Davidse (1988) proposes a distant relationship between
Prionanthium and Pentaschistis based on several shared spikelet characters, the most important being the occur rence of well differentiated multicellular glands on the glumes o f all three species of Prionanthium and on the glumes, pedicels and leaves of several species of Pentaschistis (Linder et al. in prep.). Although stalked multicellular glands do not occur on the leaves o f Prionanthium, the presence of sessile, linear glands serves to strengthen the argument of affinities between these two genera.
The leaf glands do not appear to represent equivalent structures to the sessile glands o f the glumes of P. pholiuroides which appear to be similar to the multicellular glands but lack the stalks. The sessile glands on the leaves are linear structures with the glandular tissue not aggregated into raised, cylindrical cones as is the case with the glume glands. Prionanthium is. therefore, unique in the possession o f two gland types. In Pentaschistis both these gland types occur (Linder et al. in prep.) but not on the same plants or species and are characteristic o f two different species groups. On the basis o f this character alone Prionanthium, therefore, appears to be intermediate between the Pentaschistis species groups represented by P. triseta and P. thunbergii (Kunth) Stapf. Affinities with Pentaschistis are supported by the anatomical evidence, but a critical revision of Pentaschistis is needed before the phylogenetic links of Prionanthium becom e more lucid.