A revision of the genus Prionanthium ( Poaceae : Arundineae )

Priononanthium revised and three species are recognized. All three species have a chromosome number of n =7. The genus is unual. in having multicellular, secretory glands on the glumes and a wide range of inflorescence types. Priononanthium  is considered to be arundinoid on the basis of its leaf anatomy, ciliate ligule, spikelet morphologyand less strongly, chromosome number. Its chromosome number, multicellular glands, two florets per spikelet, type of palea and  rachilla extension relate it more closely to  Pentaschistis than to any other extant arundinoid genus. A fully illustrated taxonomic account is presented.


INTRODUCTION
IM onanthium , a genus of three annual species, found only in the Cape Province, is one of the rarest grass gene ra of southern Africa and one of the species has been listed as endangered (Hall & Veldhuis 1985).The total number of separate populations known is 14.While col lecting cytological specimens of grasses from the Cape, we were able to locate populations of all three species and this stimulated the following taxonomic revision.Ellis (in prep.) will report in detail on the leaf blade anatomy.

HISTORICAL RFVIFW
The genus l*rionanthium was first described by Desvaux (1831), who was unable to relate it to any other genus known to him.He named the only species P. rigidum and gave its distribution as 'India Oriental', (India an the East Indies).This was clearly an error, for the illustration that he provided clearly indicates that P. rigidum is the Cape species now known as P. dentatum.Prionanthium was undoubtedly based on Thunberg's collection from the Cape, the only collection known of t ie species until 1975, although this collection was not specifically cited by Desvaux.However, Thunberg's collection had already been described by Linnaeus the Younger (1781) as Phalaris dentata so that the correct r .T / , v !l1C tyP° sPccies-as Pointed out by Henrard (IV41), is Prionanthium dentatum (L.f.) Henrard.
For more than two centuries, the genus was asso ciated with a mixture of genera (often recognized as subtribes (cf.Pilger 1954;Zotov 1963)) that would be termed avenoid and arundinoid in modern terms (cf.Clayton & Renvoize 1986).Following Hubbard's (1948) formal recognition of the tribe Danthonieae, Chippindall (1955) was the first to explicitly and exclusively associate Prionanthium with arundinoid genera in the modern sense of Clayton & Renvoize (1986).This placement has been followed by De Wet (1956), Brown (1977), Ren voize (1981,1986), Clayton & Renvoize (1986) and Watson et al. (1986).MORPHOLOGY There are several morphological features which merit discussion beyond that provided in the formal descrip tions of the taxonomic section.
Multicellular, presumably secretory, glands occur on the glumes of all three species.They are of two kinds: stalked in P. dentatum (Figures 1A, B; 2C) and P. eck lonii (Figure 1C, D) and sessile in P. pholiuroides (Figures IE,F,H;2D).The stalked glands are approximately cylindrical in shape with sloping bases.They are com posed of 16 rows of cells in circumference.The apex is rounded with a shallow, basin-like depression in the centre surrounded by a low, heavily cutinized lip.The apices of the glands illustrated for P. ecklonii (Figure 1D) have apparently been slightly distorted through shrinkage.At the bottom of the basin-like depression is a rectangular grid formed by the raised, rigid, anti clinal adjacent cell walls of small, square cells (Figure 1H).The stalks of the glands tend to be longer toward the apex of the glume in both P. dentatum (Figure 1A) and P. ecklonii (Figure 1C).The function of the glands in Prionanthium is not known.In some freshly collected specimens of P. pholi uroides the glands were glistening with a thin layer of moisture which I assume to indicate active secretion.In very glandular species of Pentaschistis, such as P. aspera (Thunb.)Stapf, P. angulata (Nees) Adamson, P. angusti folia (Nees) Stapf and others, the plants may be sticky and/or produce an unpleasant smell.The volatile sub stance that is produced is not known.I did not detect any kind of odour in any species of Prionanthium.In Pentaschistis, at least in those species that produce it copiously, the volatile substance may be an anti-herbivore mechanism.It is possible that in Prionanthium the glands produce a sticky substance that would allow mature spikelets with caryopses to be dispersed through ad hesion to animals.I could not investigate this matter because all populations that I observed in the field were too young.
Inflorescences in Prionanthium display a distinct re duction series.The inflorescence of P. dentatum is the most unspecialized.It is a condensed panicle that is superficially spike-like.The short panicle branches are tightly appressed to the axis of the panicle.In P. ecklonii most o f the inflorescence branches are reduced to pedicels only and they occur in pairs, one short-pedicelled (near ly sessile) the other longer-pedicelled.Only at the base o f vigorous inflorescences is one order o f elongated in florescence branches developed, and the spikelets are paired on these branches, except for the apex, as they are on the main rachis o f the inflorescence.Towards the apex o f the inflorescence the paired arrangement is lost and the spikelets are solitary.In P. pholiuroides the solitary arrangement o f the spikelets is maintained throughout the inflorescence.In very depauperate speci mens o f this species the inflorescence may be reduced to a few, or exceptionally, to a single spikelet.The fruit of Prionanthium has never been described.That of P. dentatum still remains unknown but in both P. pholiuroides (Figure 5A, B) and P. ecklonii (Figure 5C, D) it is a caryopsis with the pericarp fused to the seed coat.It is narrowly lanceolate in outline, slightly convex on the embryo side and grooved on the hilum side.The hilum is linear.In P. pholiuroides the hilum groove is nearly as long as the caryopsis and is uniformly deep (Figure 5B).In P. ecklonii the hilum groove is */8 as long as the caryopsis and is deep in the lower half of the caryopsis but rather shallow in the upper half (Figure 5D).The embryo is V 4-3/10 as long as the caryopsis.The caryopsis is held loosely between the lemma and palea.In all respects the caryopsis morphology is typical of arundinoids.
The manner is which the spikelet disarticulates at caryopsis maturity is, however, not entirely typical of arundinoids.As is the norm in the subfamily, disarticu lation takes place between the florets, and the glumes are long-persistent.However, the florets containing the caryopsis are not shed immediately but each of the two florets of a spikelet is held between the incurved margins of the subtending glume.Although the process of dis articulation has not been observed beyond this point in the field, inferences from mature plants of P. ecklonii and P. pholiuroides which contained a high percentage of mature caryopses in their spikelets, suggest that the following events are probable: the florets containing the caryopses remain enclosed within the glumes for a long time after the caryopses have matured and after the plants have died.Presumably the glumes break off simply due to mechanical wear as the plants disintegrate during the dry season.Mature inflorescences, with most spikelets bearing caryopses, which were soaked in water did not show any noticeable hygroscopic activity of the glumes or florets under laboratory conditions.Further more, it is likely that the upper floret breaks off first since the upper glume is very prominently hinged and breaks off more easily than the lower one.This process seems consistent with the tougher, chartaceous nature of the glumes compared with the delicate, hyaline florets, and this suggests that the function of protecting the caryopsis has been taken over from the lemmas and paleas by the glumes.

CYTOLOGY
No chromosome counts have ever been reported for any species of the genus.Young inflorescences were col lected in Camoy's fixative in the field.Anthers were squashed in aceto-carmine to which a small amount of iron-acetate was added.Photomicrographs of chromo some complements were made with a Reicherdt Univar microscope.
All species had n = 7 with regular bivalent pairing and no meiotic irregularities (Table 1; Figure 6A -D

GENERIC RELATIONSHIPS
Although Chippindall (1955) expressed reservations about the classification of Prionanthium in the Danthonieae (= Arundineae, cf.Clayton & Renvoize 1986), it has not been challenged since that time (De Wet 1956;Ren voize 1981;Watson et al. 1986;Clayton & Renvoize 1986).I agree with this assessment based on data from leaf anatomy, gross morphology and chromosome number.
Leaf anatomy has previously been studied by De Wet (1956) and Renvoize (1981Renvoize ( , 1986) )  From the published results it is clear that Prionanthium is typically arundinoid in its anatomy.It has a non-Kranz leaf anatomy that typifies most arundinoid genera, in cluding double bundle sheaths, non-radiate mesophyll with a maximum lateral cell count greater than four, adaxial ribs, bulliform cell groups not associated with colourless cells, finger-like microhairs with tapering distal cells, domed subsidiary cells and sinuous long cells in the abaxial epidermis, and dumbbell-shaped to nodular silica bodies.Two unusual features for the subfamily were noted by Renvoize (1986): 1, the chlorenchyma extends between the outer bundle sheath and upper sclerenchyma girders or between upper and lower sclerenchyma girders, a feature also noted by Watson et al. (1986); 2, the upper and/or lower epidermal cells are large and thin-walled.Although unequivocally arun dinoid, the anatomical evidence available to date is not sufficient for a detailed comparison with other possibly related genera.
All Prionanthium species have a ciliate ligule, a characteristic that is shared with all other arundinoid genera recognized by Clayton & Renvoize (1986).
The relationship of Prionanthium to other arundinoid genera has only been explicitly discussed by Clayton & Renvoize (1986: 165), who consider it to be one of the primitive arundinoid genera, along with Tribolium, Urochlaena, Elytrophorus, Spartochloa, Notochloe, Zenkeria, Piptophyllum , and Styppeiochloa.They based this assessment on the short glumes and multinerved lemmas with entire tips of these genera.Furthermore, they (Clayton & Renvoize 1986: 171) noted that 'P.dentatum hints at a distant relationship with Tribolium', w ithout indicating on which character(s) they made this judgement.Presumably they suggested this relationship because P. dentatum has the most fully developed panicle of the three species and its proportionally short and broad spikelets superficially resemble those of the Lasiochloa alliance of Tribolium.
Four spikelet characters point to a relationship be tween Prionanthium and Pentaschistis: the occurrence of well differentiated multicellular glands, two florets per spikelet, a small rachilla extension above the upper floret, and similar paleas.Since multicellular glands are rare in the family, I consider this character especially important in relating Prionanthium to Pentaschistis.
Although Prionanthium and Pentaschistis share a number of features, their relationship is not a close one, as many differences in spikelet characteristics demon strate.Of the three characters explicitly mentioned by Clayton & Renvoize (1986) as typical of the primitive Arundineae, Prionanthium does not differ strongly from Pentaschistis in two: 1, short glumes (in both genera the glumes are as long as the spikelet and are as long as or longer than the uppermost floret); 2, awnless lemmas (all species of Prionanthium, 10 out of ±65 species of Pentaschistis)', 3, multinerved lemmas (3-5-nerved in Prionanthium and 5 -1 1-nerved in Pentaschistis).
The primary basic chromosome number of the Arundinoideae has been considered to be x = 12 (Clayton & Renvoize 1986).However, it is more likely that this is a secondary base number derived by polyploidy, since a number of arundinoid genera are now known with n = 6:  Stebbins (1956) and Hunziker & Stebbins (1987) also consider x = 6 to be the basic chromosome number of the Arundinoideae.Less common base numbers in the subfamily are x = 7 and 13 in Pentaschistis (Tateoka 1965;Hedberg & Hedberg 1977).Prionanthium with n = 7 has a number that is unusual for arundinoids but one that is matched at the diploid level, n = 7, in part of the genus Pentaschistis (De Wet 1954, 1960;Du Plessis & Spies 1988;Spies & Du Plessis 1988).There is also a resemblance between the two genera in the size of the chromosomes, among the largest chromosomes in the subfamily.

CONSERVATION STATUS
All three species of Prionanthium have been included in the South African red data book (Hall & Veldhuis 1985).The status of both P. ecklonii and P. dentatum (as P. rigidum) is listed as uncertain, whereas P. pholiu roides is listed as endangered.From the fieldwork and herbarium research that I conducted, it is apparent that P. pholiuroides is the most widely distributed of the three species (Figure 7).The population of P. pholiu roides reported on by Hall & Veldhuis (1985) was re visited on 21 September, 12 October and 16 November 1987 and still existed as a population of thousands of individuals.However, the local distribution was very spotty and clumped.Similar sites only a few kilometers away lacked any sign of P. pholiuroides.It had also managed to establish itself in a small area of disturbance along the roadside.The main population grows in a heavily over-grazed depression invaded by introduced species of Acacia.We discovered a modest range extension for this species {Davidse 34048, 34053).Its total distributional range, although very disjunct, is approximately 250 km, and not 60 km as noted by Hall & Veldhuis (1985).At this new site, the species grew on a limestone outcrop, especially in small sandy soil pockets on the eroded limestone.This indicates that its habitat requirements may not be quite as specialized as believed by Hall & Veldhuis (1985).It seems quite probable that the spe cies has been largely eliminated from localities such as the Stellenbosch Flats, which have been extensively urbanized.It was common there from the 1920s to at least the 1940s, judging by the Duthie and Rehm col lections, and although I believe it likely that some plants may still exist in the flats between Gordon's Bay, Cape Town and Stellenbosch, these are likely to be so few and isolated that the species must for practical purposes be considered extinct in this area.Thus, although the species is on the whole rare, it may not be in immediate danger o f extinction.By contrast, P. dentatum and P. ecklonii appear to have a much more limited distribution and on this basis must be considered to be endangered.

Not surprisingly for annual species, samples o f P. p h o liuroides (Ellis 5482A ) and P. ecklonii (Ellis 5503A )
collected in November 1987 showed good caryopsis production.In both species, at random, five spikelets from five different plants were examined for caryopsis production in both the upper and lower florets.In P. pholiuroides the lower florets had 92% caryopsis pro duction and the upper 88% .In P. ecklonii 96% of the lower florets and 92% of the upper florets produced caryopses.
All three species appear to be able to exist in areas of moderate disturbance such as that encountered in the typical grazed veld o f the Cape.All three species also grew on roadsides which may actually undergo less dis turbance than adjacent grazed veld since animals general ly do not have access to roadsides.However, road con struction projects could have drastic effects on these small roadside populations since the vegetation may be completely scraped off by heavy machinery at any time.Culms 30-430 mm tall, unbranched or branched only at the base, glabrous.Sheaths short-pubescent: ligule 0 ,5 -1 ,4 mm long, the membrane 0 ,0 5 -0 ,2 mm long, the cilia 0 ,4 -1 ,3 mm long; blades 15-105 x 0 ,5 -3,0 mm, short-pubescent.Inflorescence a condensed, cylindrical spike-like panicle, 5 -7 5 mm long, reduced to a single spikelet in depauperate individuals; rachis cylin drical, pubescent; branches tightly appressed to the rachis, shortened upward; pedicels short stumps or to 1 mm long.Spikelets 3 ,2 -5 ,2 mm long, not obviously borne in pairs, laterally arranged; glumes 5-7-nerved, the side facing the racliis pubescent (Figures 1 A, 2C), the hyaline margin broader than the chartaceous centre, the keel with prominent stalked glands, the tip shortly awnpointed; lemmas evenly pubescent, 3-nerved, shortly awn-pointed; paleas glabrous on the margins, pubescent on the back; anthers 1 ,8-2,3 mm long.Chromosome number: 2n = 14 (Figure 6A, B).
Only known from the Nieuwoudtville area of the western Cape at an elevation of ± 700 m in Western Mountain Karoo (Figure 7).
Vouchers: Davidse 33394, 33396;Ellis 2452, 5416, 5417.Nees (1841) and Chase & Niles (1962), among others, considered Thunberg's (1794) Phalaris dentata to be new and different from the Phalaris dentata of Linnaeus (1781).This is incorrect because both names are based on Thunberg's collection from 'Bockland' (= Bokkeveld).This is clear from Juel's (1918) account explaining how Linnaeus published hundreds of new species based on Thunberg's Cape collections.I interpret Thunberg's usage to be based on Linnaeus's (1781) name without proper attribution and using a different and original descrip tion.For this reason, I consider all other combinations made with the epithet dentatum to be based on Lin naeus's (1 781) initial usage of this name.
Until Ellis re-collected it in 1975, this species was only known from the type collection made by Thunberg in 1 773.Thunberg's collection, however, was a large one and a number of duplicate specimens were distributed to other herbaria so that very good type material exists.
Only known from the south-western Cape Province at low elevations, rare, in Coastal Renosterveld (Figure 7).This species was until recently only known from three collections, two made by Ecklon and one by Drege.When Nees (in Lindley 1836) described Prionachne eck lonii, he did not indicate a type and only a general geo graphical distribution was provided.However, from his use of Ecklon's name as the epithet and the fact that all three of the collections had been available to him by 1836 (Gunn & Codd 1981), it is clear that he was basing his species on these collections.This is confirmed from a later publication (Nees 1841), dedicated to Drege, Ecklon and Zeyher, where he redescribed and renamed this species as Chondrolaena phalaroides with two varieties, var.dentata and var.edentula, and explicitly cited the Ecklon and Drege collections.He also referred to his 1836 treatment of l*rionaehne in the synonymy.Unfor tunately he did not directly refer to P. ecklonii but made the new, but illegitimate, combination Prionachne den tata.This new synonym was based on Linnaeus's (1781) Phalaris dentata, which was also cited in synonymy, but was in fact misapplied.The fact that Nees explicitly states that he was publishing a substitute name for his Prionachne indicates that he considered Chondrolaena phalaroides and frionachne ecklonii to be the same species.Accordingly 1 have chosen to lectotypify Prionanthium ecklonii with the Ecklon collection from the Olifantsrivier.This preserves the usage that Stapf (1899) initiated in his excellent treatment of the genus in Flora capensis.
Nees (1841) did not truly understand the differences between the two species that had been collected by this time.This is apparent from the fact that he synonymized Phalaris dentata (= Prionanthium dentatum ) with his new species Chondrolaena phalaroides (= Prionanthium ecklonii).When he renamed Prionachne ecklonii as Chondrolaena phalaroides, he recognized two varieties, var.dentata and var.edentula.Under var.dentata he listed all but one of the combinations and usages of this name, including his own illegitimate new combination, Prionachne dentata.He excluded Thunberg's (1794Thunberg's ( , 1813) ) interpretation oi Phalaris dentata because he felt that Thunberg's description, which was substantially different from that of Linnaeus (1781), could not apply to the same species.In fact, Thunberg's descriptive state ments queried by Nees, spike 'subpaniculatum and 'glumas ovatas', do apply to Prionanthium dentatum, especially in contrast to the inflorescence and glume shapes o f P. ecklonii.Actually, Thunberg's entire glume description (glumae ovatae, concavae, glabrae, bimarginatae vel cinctae, intra marginem linea dupHci aut triplici elevata, viridi, sinuato-serratae) applies very well to the two glumes considered as a single unit.Further more, Thunberg (1794) explicitly cites the locality of the collection (crescit in Bockland) which leaves no doubt that it was based on his own collection from the Cape, which was, as noted before, also studied by Lin naeus (1781).
Nees's var.edentula should be referred to Prionan thium pholiuroides (see the discussion following that species).
Only known from the Cape Province, at low eleva tions, rare, primarily in shallow depressions that tem po rarily collect water (Figure 7).When Nees (1841) described Chondrolaena phalaroi des var.edentula, he described the glumes as non-dentate (i.e., lacking stalked, multicellular glands) and the spikelets as entirely alternate (i.e., spikelets borne singly on the rachis).This description applies well to Prionanthium pholiuroides, especially when contrasted to P. ecklonii.Nees also noted that he only saw a fragment of this taxon, intermixed with other specimens.Presumably this fragment was included with one of the three cited specimens, although he does not specify which one.All three were presumably destroyed since all his Poaceae specimens were acquired by B in 1855 (Stafleu & Cowan 1981).I have not seen similar fragments or specimens among duplicates of the Ecklon and Drege collections that I have examined.Nees suspected it to represent a new species, but, presumably because of the fragmentary nature of the specimen, decided to recognize it at a lower rank.
A recently discovered population near Struisbaai {Davidse 34048, 34053) is quite variable in spikelet morphology.It has the smallest spikelets of all known populations, 3 ,1 -5 ,2 mm versus 4 ,5 -7 ,0 mm for all others.However, more im portantly, individual plants within the population have spikelets with glumes that vary from completely eglandular and glabrous to those that are typically glandular and pubescent on the inner side, to glandular-glabrous, or eglandular-pubescent. Since these plants are seemingly segregating for these characters, these variations are not taxonomically re cognized.The Malmesbury population represented by Davidse 33459 is variable with regard to glabrous and pubescent spikelets.The only population, apparently now extinct, with glands on nerves other than the mid nerve of the glumes, is the Stellenbosch Flats population sampled by Duthie from 1925 to 1928.

Note added in proof
Stalked, multicellular glands also occur on the back of the lower lemma in many species of Panicum sect.
FIGURE 4. -Lodicules of Pricy nanthium species.A, P. den tatum, Davidse 33394, in ner and outer view of a pair of lodicules with cut off, hairy lemma at the base, X 150; B, P. pholiu roides, Spies 3678, inner view of old, slightly disr torted lodicule, X 150.C-D, P. ecklonii, Davidse 34018: C, inner view of lodicule, X 175; D, pair of lodicules at the base of the palea with lemma and ovary cut off, X 85.
Stolonifera [Zuloaga.F.O. &*Sendulsky.T. 1988.A revi sion of Panicum subgenus Phanopyrum section Stoloni fera (Poaceae: Paniceae).Annals o f the Missouri Botani cal Garden 75: 4 2 0 -4 5 5 ] of tropical America.Although the walls of the stalk are similar to those of Prionanthium, the central depression is nearly smooth and lacks the rectangular grid.It is probable that these Panicum glands are secretory, but this has not yet been conclusively demonstrated.

chromosomes are generally considered to be genetically inert, they often affect fertility, especially when present in high numbers (Jones & Rees 1982). Nevertheless, their occurrence in the population with the most variable spikelet morphology is noteworthy. Although this varia tion is probably due to simple segregation and recombi nation, the possibility o f a B-chromosome effect cannot be entirely ruled out without further research.
* * ' % m D -F, P. pholiuroides.D, Davidse 33983, anaphase I, n = 7; E-F, Davidse34048; E, diakinesis, n = 7 + IB; F, metaphase I, B-chromosome excluded from the metaphase plate.Arrows point to B-chromosome.
Watson et al. (1986)ndWatson et al. (1986)in an unspecified species.(Al though De Wet (1956) noted that he had studied P. eck lonii anatomically, the specimen that he cites as a voucher apparently is one of the Duthie collections made in the Stellenbosch Flats and actually represents P. pholiuroides.His reference to 'type' is inexplicable if interpreted as 'type collection' since none of the Prionanthium type collections originated in the Stellenbosch area.Perhaps he meant to indicate the kind or 'type of P. ecklonii originating in Stellenbosch', which, as I have just indi cated, represents P. pholiuroides.)A detailed study of leaf blade anatomy will be presented by Ellis (in prep.).