Studies in the Sphaerocarpales (Hepaticae) from southern Africa. 3. The genus Riella and its local species

Hepaticae, Riella Mont., R. affinis M.Howe & Underw.. R. alatospora Wigglesworth. R. capensis Cavers. Riellaceae Engl., R. echinospora Wigglesworth,  R. purpureospora Wigglesworth, Riellineae R.M.Schust., southern Africa. Sphaerocarpales Cavers, subgenus  Riella . subgenus  Trabutiella Porsild


INTRODUCTION
The genus Riella Mont. was first recorded from southern Africa by Cavers (1903). Dried mud. containing crustacea, the raison d'etre for its collection, was taken from a shallow pond near Port Elizabeth. Eastern Cape, in 1897 by a Mr Hodgson and sent to Owens College. Manchester (via Plymouth), where it was placed in a small aquarium. In a few weeks a number of green shoots had grown out of the mud. Upon fruiting, they were recognized as hepatics and sent to Cavers at Yorkshire College. Leeds. He identified and described them as a new species. R. capensis.
In 1926 and 1932 Miss E.L. Stephens sent a number of South African Riella plants together w ith algae and samples of mud to Manchester. Three new species of Riella from the Cape were isolated, one of w hich was from Valkenberg Vlei. unfortunately w ithout ripe spores. A fourth species was from Schonken's Salt Pan near Brandfort in the Free State. These species were studied and described by Wigglesworth (1937). Her descriptions were subsequently supplemented by Proskauer (1954), w ho also identified a Pocock specimen from a farm dam. 4 miles (6.4 km) from Grahamstown. as the widely dis tributed R. affinis. thus adding a fifth species to the southern African records of Riella. Amell (1957Amell ( , 1963 did not record any new species, but listed two new col lections from Namibia, the whereabouts of w hich have not been traced. Unfortunately, very few collections have been made in the last 30 40 years. The occurrence of these plants is stated to be rare and sporadic, which can perhaps 'be attributed to the fact that minor changes in the environment can result in their dis appearance' (Schuster 1992). Rapid urbanization in southern Africa in recent years, has also led to the destruction of natural habitats, particularly on the Cape Flats.
The gametophvtes of Riella are delicate, short-lived and highly susceptible to differing environmental condi tions. which, if not leading to their disappearance, may cause major changes in their size and form. The spores, however, are exceedingly resistant, surv iving in the dried state for years. With ornamentation that is regarded as species-specific, spores are essential for identification, as few Riella species show w ell-marked. distinguishing veg etative characters. Sometimes only the mode of branching may be of some significance. This study w as undertaken, even in the absence of recent collections, because infor mation gained by means of newer techniques such as SEM micrographs of the spores (and thalli) of southern African species has not been published before. It also completes this series of studies in the local Sphaero carpales, w hich were excluded in my treatment of the Marchantiidae for the Flora o f southern A frica (Perold 1999a).

MATERIAL AND METHODS
The same procedures as outlined in Perold (1999b) were employed in the preparation ot the material tor examination and photography by compound light micro scope and scanning electron microscope.
Throughout this treatment of the Sphaerocarpaceae and Riellaceae 1 have used the term 'stem', although it is usually referred to as 'axis' or 'rib'. Some of the species descriptions and illustrations of Riella provide rather less detail than others, because of a lack of suit able material.
Differences and similarities between the five southern African Riella species are presented in tabular form (Table 1).
Plants aquatic, usually entirely submerged, thallus developing in vertical plane, bilaterally symmetrical in plane of wing, secondarily asymmetrical. Stem (or axis) slender, erect, elongate, simple or furcate, in section ellipsoid or subround, invested with unistratose wing along its dorsal side. Wing undulate or ruffled, thin, over arching stem at coiled apex. Scales unistratose, leaflike, mostly dimorphic, the two forms not always easy to dis tinguish: lateral leaf scales basally attached to stem and formed at juncture of wing and stem, at maturity on both sides of wing and often associated with young involu cres; ventral leaf scales produced along ventral surface of stem, smaller and often constricted in middle, with lam inar attachment to stem by single cell or row of cells. Cells thin-walled, here and there with a single oil body. Rhizoids generally only borne basally on stem, all smooth-walled.
Asexual reproduction by gemmae from ventral side of stem and similar to ventral leaf scales.
Dioicous or rarely monoicous. Antheridia individual ly developed along thickened, free margin of wing, sunken in pockets, ovoid, on very short pedicel. Archegonia single, when fertilized enclosed in flask shaped involucres, smooth or rarely fluted with parallel, longitudinal ribs or lamellae, in acropetal sequence, usu ally to right and left of wing. Sporophyte with globose capsule, short seta and spherical or ± uniformly wide foot, cleistocarpous; spores released on decay of capsule wall and involucre. Spores large, single, not permanent ly united in tetrads, brown or purple to red; distal face ornamented with fine or coarse spines, tips truncate, sometimes wider below and basally connected by mem branes. rarely with prominent wing; proximal face with few to many, finer spines. Nutritive cells present, 4nucleate. Elaters absent.

DISCUSSION
Riella affinis is a widely dispersed species, originally known from Grand Canary; then from Stanford Uni versity campus, California; Uttar Pradesh. India (as R. vishwanatai); near Grahamstown. South Africa ( Figure  4), and finally, from Israel (unpublished, according to Lipkin & Proctor 1975). Schuster (1992) states that it was also reported from Argentina by Hassel de Menendez (1959), but this is incorrect, as R. affinis is not mentioned in this paper, which deals with R. americana M.Howe & Underw.
On account of its ribbed involucre. R. affinis is easily recognised and has been placed in subgenus Trabutiella. together with two other species from elsewhere, but it is the only monoicous one. The two other species are R. cossoniana Trabut (= R. paulsenii Porsild. placed in syn onymy by Lipkin & Proctor 1975) and R. garnundiae Hassel de Menendez. Schofield (1985: fig. 15-2A, B) illustrates what he calls Riella affinis. 'A' is of an 'antheridium-producing gametophore" and 'B \ 'detail of marginal chambers with antheridia'. He cites Wigglesworth (1937) Proskauer (1954). Thompson (1942) refers to 'explosive discharge of the antherozoids' in both culture and in temporary mounts, but Proskauer (1955) comments that he had never observed explosive discharge of antheridia in Riella. Proskauer also found wider variation in spore size, 70-130 fim diam., than I did. He remarks that, 'spore size is of limited diagnostic value in the order,' an admonition it would be well to keep in mind. Spore orna mentation. on the other hand, particularly as illustrated on SEM micrographs, is crucial to correct identification. Proskauer (1955) reports that the specimen collected by Dr Pocock on 31 March 1953, was from 'Farm dam, 4 miles from Grahamstown on Cradock Road ... about '/4 mile down from road'. On the label of another Pocock specimen collected on 10 May 1953 (BOL20503), this locality ( Figure 4) is given as 'Dam on Table Hill Farm. Cradock Road. 6 miles'. I visited the area in October 1999 and found the name of this farm now to be "Table Farm, Hilton' owned by the White family. I collected some mud from the bed of the Palmiet River, where the mostly dried up river runs under a bridge on the road (R350) from Grahamstown. Subsequently the techniques for cultivation, as described by Proctor (1972) and Hassel de Menendez (1987). were followed, but without success.  Figure 5A), occasionally with clusters of branches, possibly developed from numerous adventi tious shoots. Stem ( Figure 5D) in cross section ± ovoid or slightly flattened dorsally and rounded ventrally, 350 400 |im or 9 cell rows thick, 280-300 j.tm wide, outer cells rectangular or isodiametric, rather smaller, 25 -50 x 22.5-37.5 jim, inner cells angular, 50 60 x 42.5 50.0 Urn. Wing unistratose, sometimes bistratose at join with stem, margin slightly eroded, overarching stem apex, 1.8-3.4 mm wide, undulate, narrowing below and soon disappearing; cells near stem 5-or 6-sided. 62.5 95.0 x 37.5 42.5 f.im, near margin smaller and mostly 4-sided.  Involucres (Figures 5B, N; 6A, B) up to 5 produced in acropetal sequence along stem, obovoid, ± smooth, up to 2125 |im long, 1450 jim wide across widest part, nar rowing upwards to beak, ± 175 jim wide and surrounded by as many as 13 crowded, slightly projecting cells (Figure 50), below also contracted toward stalk, cells in unistratose involucral wall ( Figure 5R (im wide and 4-seriate, below only 50-55 pm wide and uniseriate. Foot not bulbous, width nearly uniform along its length, 420-430 x 220-240 pm, in cross section sur rounded by calyptra ( Figure 5P). Spores 105-125 (im diam., including prominent discoid w ing, ± 20 pm w ide. increasing to ± 27.5 pm w ide at angles, golden brow n. ± triangular; distal face ( Figure 7A-C) with ± 11 or 12 irregular rows of spines across diam.. 5.0-7.5 pm long, frequently dilated above, truncate, extending onto wing and ± 32 projecting beyond margin. 5.0-7.5 pm between spines, central ones linked by basal connecting mem branes forming indistinct reticulations; proximal face ( Figure 7D-F) somewhat raised as centrally flattened dome, w ithout triradiate mark, medianly w ith irregularly spaced papillae and marginally sprinkled with granules which extend onto wing, laterally surrounded by ± con cave wing, with radiating striations and marginally pro jecting spines. Wigglesworth (1937) reports that there is a striking difference from the other Riella species in the appear ance of the majority of cultured young plants of R. alatospora because they become heart-shaped at the top. instead of protruding at only one side as w as usual in R purpureospora Wigglesworth. I cannot comment on this, not having observed plants in culture.

DISCUSSION
In some of the specimens I examined, the marginal 5 or 6 row s of cells along the w ing w ere w ithout chloroplasts. which may perhaps be ascribed to the effects of partial drying.
In Wigglesworth (1937) there is a typographical error in the length of'the plant as it is given in mm (3.5) instead of cm. She gives the size of the spores as ± 120 pm. whereas Proskauer's (1955) measurements varied from 80 to 140 pm and my own from 105 to 125 pm.
Riella alatospora is easily distinguished by spores with a prominent discoid wing. Whether it has survived in ponds on the Cape Flats ( Figure 4) is a matter of con jecture: at least it will not ha\e been another victim of 'anonymous extinction' (Campbell 1989). thanks to the laudable efforts of the ladies Stephenson and W iggles worth.

DISCUSSION
Cavers did not give a Latin description of his new species, as it only became compulsory with the 1935 ICBN code (Briquet 1935); subsequently, Wigglesworth supplied a Latin description in 1937. Cavers (1903) described the stem as circular in cross section, which probably would be more representative of the species than my section of it ( Figure 8C). He also stated that he would describe the developmental stages of R. capensis, but I have not found a reference to such an article. He expressed the opinion that his new species came nearest to R. helicophylla Mont. from Spain, Algiers and Tunis.
Hassel de Menendez, in her 1959 paper, points out the differences between R. americana and R. capensis, the latter much branched, the male plants with up to 100 (or more) antheridia, the female plants with up to 50 sporan gia on a single plant, spore diam. 80 ^m and the spines 8 |im long. Sim's (1926) record of it from Cape Town is evidently incorrect (Wigglesworth 1937). Proskauer (1955) assigns a Pocock specimen collect ed on 12 December 1952 at the Palmiet River, Table  Rock Farm, seven miles from Grahamstown on the Cradock Road, to R. echinospora, but then refers to the marginal spines of the spores as showing light webbing, which, in my opinion, would place it nearer to R. capen sis. 1 have seen no such webbing in R. echinospora spores, which are subround. His determination is there fore suspect. Wigglesworth's (1937: fig. 7) illustration of the involucre of R. capensis, is rather less attenuate toward the beak, than those that I examined; her figure 9 of the proximal spore face, suggests some webbing at the base of the spines, whereas her figures 11 and 12 of R. echi nospora spores clearly rule out the possibility of any webbing. Furthermore, she supposes that the course of growth in R. capensis plants followed the same lines as that of R. alatospora.
My visit to Port Elizabeth in October 1999 in an effort to find more material of R. capensis proved unsuccessful.
It was recently brought to my attention by Dr W.R. Harding, that Coetzer (1987) had reported R. capensis from Rocher Pan on the west coast. This collection has not been traced and the determination could not be veri fied.

DISCUSSION
It is possible that Riella echinospora is more wide spread than just the Brandfort area, as Amell (1957Amell ( , 1963 also recorded it from a Volk collection in Namibia at Haribes, Marienthal, 4In seichtem Wasser auf feinem Sand, haufig\ Proskauer (1955) also reported that 'a sporeling with attached spore probably belonging to this species was isolated during class work at Berkeley from a culture prepared from soil gathered by Dr Pocock on the Cape Flats'. Regarding his reference to the specimen from 'the Palmiet River, Plants erect, 20-60 mm tall, stems sparsely to fre quently furcate, some branches again furcate, occasion ally with several pseudodichotomies close together, the daughter stems growing new wings; stalked adventi tious shoots formed anywhere along parent stem ( Figures 13A, B: 14C). Stem in cross section ( Figure  13C) subround, 330-350 fim or 10 cell rows thick, 350-380 nm wide, outer cells isodiametric, 25.0-37.5 x 20.0-32.5 nm, inner cells angular, mostly larger, 30-60 x 30-40 urn. Wing unistratose, but bistratose at join with stem, overarching stem apex, 1.3-2.75 urn wide, undulate, narrowing below and then disappearing alto gether, leaving basal part of stem wingless; cells near stem 5-or 6-sided, 50.0-87.5 x 30-50 nm, near margin smaller, 4-or 5-sided, 20.0-32.5 x 15-20 |im; scattered throughout wing, numerous small cells filled with an oil body ( Figure 13D). Scales dimorphic ( Figure 14F Dioicous. Male plants somewhat smaller than female plants. Antheridia numerous, in a single, linear series ( Figure 14A, B), in pockets of up to 23, in acroscopic sequence along wing margin, but sometimes interrupted (Figure 130), ovoid, 350 x 250-270 ^m, discharging through individual ducts, mostly sloping toward and opening by pores at edge of wing; cells in wing covering antheridia rather larger than those at periphery of antheridia. Archegonia ( Figure 13P) at maturity with 4 cover cells at apex of neck, these swollen and separating from each other, leaving neck open for entrance of antherozoids. Involucres (Figures 13Q, R; 14D, E) up to 7 produced in acropetal sequence along stem, pyriform, ± smooth, 2375-2575 |im long, ± 1600 ^m wide across widest part, abruptly narrowing to beak, ± 200 nm wide and surrounded by ± 12 cells in an irregularly protruding row, below also contracted toward stalk, cells in involu cral wall 5-or 6-sided, 40-60 x 32.5-45.0 jim. Stalk ± 375 x 250 nm, obliquely attached to stem, upper ± 125 jim occupied by basal part of foot. Calyptra multistratose, in cross section up to 4 layers of cells surrounding inner haustorial cells of foot ( Figure 13U). Capsule globose, up to 1200 urn diam. at maturity, wall red or mauve, unis tratose, cells ( Figure 13S) 4-or 5-sided, 42.5-70.0 x 30.0-47.5 ^im. Seta ( Figure 13T) ± 210 jim long, dark red, upper ± 100 ^m expanded, funnel-shaped, narrow below, only ± 35 lm wide, uniseriate. Foot ( Figure 13T) ± 260 ^m long, gradually expanding from narrow upper part to ± 200 ^m wide below. Spores 82.5-117.5 jim diam., including spines, without wing, purple or red, ± tri angular; distal face ( Figure 15A-C), with ± 12 rows of spines across diam. and 25-30 projecting around margin, mostly stout and truncate, rarely acute, 7.5-10.0 urn long, 5-10 ^m between spines, basally connected by mem branes forming irregular reticulations; proximal face ( Figure 15D-F) raised, not flat, marginally with basally webbed spines, sometimes webbing very prominent, appearing almost wing-like, 7.5-12.5 jim wide, with spines projecting outwards from it, triradiate mark occa sionally nearly complete, but mostly only present toward angles, rest of face irregularly dotted with up to 15 low spines per facet.

DISCUSSION
So far, R. purpureospora is the only species to have been collected in recent times, viz. by Dr W.R. Harding. It is quite a robust plant and is probably the easiest species to identify because of its purple or red spores and capsule wall, as well as the pronounced webbing between the bases of the marginal spines on the proximal spore face.
The reference to the separation of the four cover cells of the archegonial neck in my description is from Thompson (1942) who referred to R. qffinis, but it is equally applicable to other species, i.e. R. purpureospora.
The Harding specimen was collected at Blouvlei. Cape Town vicinity (Figure 4), in an ephemeral pan which con tains water between April and September, together with Pseudalthenia aschersoniana and Bolhoschoenus maritimus. The pH of the water was 9.6. the alkalinity (as CaCOi) 246 mg per litre and the salinity (as Na) 3813 mg per litre (W.R. Harding pers. comm.). This species has not been treated in this study, as there are no ripe spores in the original collection and it has not been collected again. As mentioned in the introduction, the spore ornamentation is essential for correct identifi cation o Riella species.

ECOLOGY
The five Riella species known from southern Africa, are from widely scattered localities, ranging from the summer rainfall area of central Free State (R. echinospo ra), to the winter rainfall areas of the Cape Flats (R. alatospora and R. purpureospora), in Western Cape, and extending to parts of Eastern Cape (R. affinis and R. capensis), which receive sparse summer and winter rains. The vegetation types in these localities, according to Low & Rebelo (1996). are the following: central Free State: Dry Sandy Highveld Grassland; Western Cape. Cape Flats: Sandplain Fynbos; Eastern Cape. Port Elizabeth area: Mesic Succulent Thicket; north of Grahamstown: Eastern Mixed Nama Karoo.
Riella species grow in temporary or permanent pools, vleis or intermittent streams, containing fresh or brackish water. Hassel de Menendez (1987) on the other hand, found that Argentinian Riella species did not grow in temporary dry ponds, but rather in lakes, some of which are artificial Riella thalli cannot, however, withstand desiccation, even for a short while.
It is thought that Riella spores may be transported by wind or by birds (Hassel de Menendez 1987). Schuster (1992) is of the opinion that it is unlikely, although theo retically possible, that thalli (and spores) may be dissem inated on the feet of w ading birds from one site to the next. Apparently spores can pass through their gut unharmed and may be transported in this wav over dis tances limited to under 80 100 km.