Addendum 4. Salter’s revision of South African Oxalis (Oxalidaceae) and some new combinations from Herbertia 48 (1&2) – 1992

Department of Agricultural Development
Winter Rainfall Region, Elsenburg (RSA).

The genus Oxalis, as it occurs in South Africa, is discussed generally in relation to the revision by T.M. Salter (1944).  A brief account is given of Salter’s collecting activities, the distribution and variability of the species, and the problems of classification.  Some new combinations are made.  Particular attention is paid to the section Pardales Salter, and all the 11 species recognised by Salter (many described by him) are combined in one species, namely Oxalis pardalis Jacq.  Two species, Oxalis dentata Jacq. and Oxalis lateriflora Jacq., are reduced under O. livida Jacq.  Similarly Oxalis urbaniana Schltr. and Oxalis callimarginata Weintr. are reduced under Oxalis goniorrhiza Eckl. & Zey.

Keywords: Oxalis, taxonomy, speciation.

Oxalis is particularly well represented in South America, and the species are also a very prominent component of the flora of the Mediterranean region of South Africa.  This region extends from Luderitz in Namibia, southeastwards to near East London on the southeast coast (Bayer, 1974).  Annual precipitation over this area varies enormously from as little as 100 to over 1400mm/annum.  The area is divided by Rutherford and Westfall (1986) into four distinct biomes on the basis of summer aridity, seasonal distribution of rainfall and on plant growth forms.  Both geology and topography vary enormously so that landscape heterogeneity and soil forms provide great habitat contrasts over short distances.  These factors, together with historical events, probably account for the richness of the Cape flora.  Despite the potential and realised weediness (for example: O. pes-caprae L. and O. caprina L.), many species are spectacularly colourful in autumn and winter and are useful and interesting as horticultural subjects.  T.M. Salter (1944) revised the genus on the basis of his own intensive collecting.  His interest in fact started in 1931, and, although he collected until 1957, serious attention to Oxalis continued only until 1942.

Salter’s collecting activity is summarised in his four accession books where 1,740 specimens are listed.  A weakness in this collecting is in the geographical coverage.  He collected principally on the Cape Peninsula: the route to Caledon via Sir Lowry’s Pass and back via Franschoek, the route to Springbok in the north, and the areas from Mamre to Piketberg and Porterville.  One trip was made to the area between Caledon and George, another to west of Garies and two trips eastwards between Garies and Khamieskroon via Leliefontein.  Virtually no collecting was done in southern Bushmanland, in the Tanqua/Doornriver Karoo, the Little Karoo, the southern Great Karoo or even in the Worcester/Robertson Karoo.  The areas north and west of Steinkopf and east and west of Springbok were not sites of collection.  He finally recognised 208 species, stated that he was under no illusions about the completeness of his work, and admitted to the difficulties he experienced in deciding on what taxa to recognise.  The revision is unquestionably the product of the most painstaking and meticulous labour.  In using his revision to identify more than 1000 collections of my own, it became evident that Salter was remarkably successful in sorting out at least the material at his disposal.  Nevertheless, the revision is a curious paradox of grossly variable ‘group-species’ with many apparently rare and localised elements.  This may, in fact, be the position in the field, but the inordinate difficulty I experienced in trying to identify the 13 species that occur in the small area of the Karoo Botanic Garden at Worcester led me to attempt a wider evaluation of Salter’s revision.

Herbarium representation of the species may in some measure reflects the knowledge of the genus.  Sixteen (16) of the described species are known only from the types.  Salter, himself, described 13 species which remained known to him only from the type collections, and another 13 species are represented by only two collections, 20 species are represented by three cited herbarium specimens, while several species are represented by more specimens but all from the same locality.  Altogether 29 species were known from only one gathering, while 70 species were represented by three specimens or less.  By contrast 45 species were known from 20 specimens or more, and these include very variable ‘group-species’ such as Oxalis flava L., 0. purpurea L. and 0. obtusa Jacq.  The majority of the species, including six Namibian elements, occur in the winter rainfall region mentioned earlier.  12 species occur in the Eastern/Western Cape, while 16 species have a wider South African distribution.

Salter provides three identification keys in his revision.  The keys are based on the number of leaves per petiole, the articulation of the peduncle, leaf-shape, corolla shape, and a variety of similar morphological characters.  Salter stressed the value of bulb characters, but made very little use of this structure in circumscribing his species.  The annual replacement of the bulb and the characteristic sequence of outer protective and inner nutritive scales was wholly ignored.  Another character possibly useful as the basis for a sub-generic break, is that of the carpels being either fused or free.

Ornduff (1973) has made the only recent contribution to the taxonomy of the genus.  He made use of a wide range of morphological characters, as well as of fertility and chromosome number in describing one species (Oxalis dines) which Salter had dismissed as a variety of 0. disticha Jacq.  Ornduff suggested (private communication, 1974) that Salter may have recognised too few species.  However, my own experience with Oxalis suggests the converse, and also that the status of 0. dines is not in the least convincing.  Ornduff did not examine overall variability of 0. disticha or pollen fertility within his concepts of either this species or his new 0. dines.  He suggested that sexual reproduction in Oxalis is a rare event.  This may be true for some species, but my experience is that many species produce abundant seed, and seedlings are common.  Furthermore, there are many species which patently do not have the capacity to spread or even proliferate vegetatively.  Vegetative reproduction is undoubtedly responsible for weediness of those Oxalis species such as 0. pes-caprae, but this is not even necessarily a species character.  In 0. pardalis Sond. vegetative propagation results in the formation of a compound cluster of bulbs, but not in the physical relocation of plants.  In 0. zeyheri Sond. there are populations in which the individuals barely reproduce vegetatively while in others the plants are highly stoloniferous.  Salter (1944) reports this in both 0. flava L. and 0. glabra Thunb., and this is also the case in 0. livida Jacq.

The object of this paper is to present some conclusions drawn from a study of about 1000 specimens additional to Salter’s work.  In most cases living plants were brought into cultivation from where detailed notes and herbarium specimens were prepared.  The proportion of confident identifications based on Salter’s revision was a little above 60%.  My experience with the identification keys was most frustrating, and positive identifications were frequently made only after painful comparison with the complete description and reference to the locality.  Early in the work several new elements were identified outside of Salter’s collecting areas, and descriptions of these were even prepared for publication.  However, as I became more familiar with Salter’s species, so did it also become increasingly evident that such an approach would produce an irrational proliferation of names and total confusion in the genus.  The problem at this stage is not one of there being too few or too many species, but the wholly inadequate knowledge we have of their distribution and variability.

The following re-description and synonymy of three species is presented as an indication of the real state of affairs in the genus and as an indication of further problems which will be encountered in Oxalis.

1. Oxalis pardalis Sond., in Flora Capensis 1:345 (1860); Salter, in J.S.Afr.Bot. Supp. 1:234 (1944).  Type: Sandy places at Hex River, Burke et Zey. 253 (holotype, S,; isotypes, BOL, SAM).  0. capillacea E.Mey. ex Sond.  In Fl.Cap. 1:435 (1860); Salter, l.c.:222 (1944).  Type: Hex River, Burke et Zey. 251 (holotype, S).  0. confertifolia (0. Kunze) R. Knuth, in Eng.Bot. Jahrb. 41:27 (1937); Salter, l.c.:231 (1944).  Type: sine loc. Drege 7426 (holotype, S).  0. robinsonii Salter & Exell in J.Bot. 48:146 (1930); Salter, l.c.:236 (1944).  Type: Reutersbosch, Salter 376/57 (holotype, BM).  0. lineolata Salter, in J.S.Afr.Bot. 2:1 (1936), l.c.:226 (1944). Type: near Doornbosch in dry shale flats, Salter 5382 (holotype, BOL).  0. massoniana Salter, l.c. 2:3 (1936), et :227 (1944).  Type: in damp, muddy, shallow soil overlying flat rock at summit of Vanrhyns Pass, Salter 705 (holotype, BOL).  0. melanograpta Salter, l.c. 2:5 (1936), et :229 (1944).  Type: in shale on low koppie 18 miles north-east of Vanrhynsdorp, Salter 714 (holotype, BOL).  0. heildelbergensis Salter, l.c. 2:5 (1936), et :232 (1944). Type: 3 miles west of Heidelberg, Salter 2385 (holotype, BOL).  0. grammophylla Salter,.l.c. 2:8 (1936), et :234 (1944).  Type: Karamoe, Salter 4469 (holotype, BOL).  0. leptogramma Salter, l.c. 2:147 (1936), et :224 (1944).  Type: Worcester Division, near Orchard Siding, Salter 2178 (holotype, BOL).  0. camelopardalis Salter, l.c. 3:102 (1937), et :230 (1944). Type: Worcester, Rooihoogte Pass, Salter 2308 (holotype, BOL).

Plants: rarely acaulescent, 50-300mm high.  Bulbs: to 200mm deep in soil, compound, old tunics loosely adhering, new bulb comprising 5-20 linear acuminate, scaly protective scales up to 60mm long and 5mm wide, with 2-4 harder, acute, undulate, coriaceous tunics enclosing 4-6 elongate, acute, fleshy, nutritive scales.  Protective scales are clothed with retrorse brown hairs.  Rhizome and stem: 50-150mm long, glabrous, glandular or pubescent.  Leaves: epetiolate and cauline, to petiolate and terminally congested; leaflets glabrous, glandular, pubescent or ciliate, linear, linear-acute, obtriangular or elliptic, always with short, translucent, longitudinal idioblasts which turn black on drying.  Peduncles: 1-flowered, with alternate or subopposite, linear, idioblastic-striate bracts above middle or near calyx. Sepals: 36mm long, ovateobtuse, ovate-acuminate or lanceolate, idioblastic-striate.  Corolla: white, pink, apricot, red-purple, lilac or yellow; petals 13-22mm long, 643mm wide, lamina ovate or elliptic.  Filaments: the longer unequal, 5-5.7mm long, toothed or gibbous, glandular, anthers horizontal.  Ovary: carpels fused, 2-7 loculed, uniseriate, lateral shiny idioblastic pustules; style pubescent below, with glandular or simple hairs above.  Seeds: arillate, without endosperm.

Salter recognised 11 species which he grouped together in his section Pardales.  Five of these species were based effectively on a single collection or population.  Where specimen numbers increased, Salter recognised varieties and forms based primarily on flower colour and character of the petioles.  The principal characters used by Salter in his identification key for the section were leaflet number, leaves sessile or petiolate, petal width, sepal length, plication of bulb tunics, proportion of lamina to tube length, indumentum, size, exsertion of filaments, number of ovules and flower colour.

My first experience with Oxalis was the collection of material from new localities, and even localities visited and collected from by Salter, which invalidated his keys.  Initially it was felt that some of these new finds were so strikingly different that they could be described as new.  However, as my collection grew, so did the realisation that a more liberal attitude to the species was necessary.

The individual plant in this species comprises a deep-seated, compound bulb, each bulblet producing a single stem.  Individual stems may reach as long as 3OOmm, although in rare cases the plants may even be stemless.  The nature of the indumentum is surprisingly variable.  The rhizome may be either glabrous, incipiently glandular, glandular or glandular/pubescent.  The stem is similarly variable and may, in addition, be ciliate.  The leaves are also glabrous, pubescent or ciliate.  Occasionally there is a degree of viscidity, particularly of the stems.  This indumentum of rhizome, stem and leaf is not correlated and different combinations are observed in different populations.

Bulb structure was observed to be the most consistent element in that the new bulb is subtended by many sharply pointed and loosely adhering outer scales, hardening progressively to the last few scales which enclose the softer, white, fleshy, nutritive scales.  The outer tunics are always clothed with retrorse, brown hairs, although in one collection (Bayer 2301) these hairs appeared to be absent.  Oxalis massoniana was characterised by transversely plicate protective scales, but this element is simply an unusual ecotype growing on the Nieuwoudtville escarpment in shallow sand overlying deeper stone.  Thus, in this case the bulbs are affected by soil depth.  There was never any indication of vegetative propagation by stolons as occurs in other species.  Secondary bulbs arise in the axils of the scales on the rhizome above the primary, basal bulb-producing axil.  Contractile roots appeared in many collections, apparently as a function of planting depth, and may be a factor assisting re-establishment after predation in the wild.  Propagation and re-location is by sexual propagation from an exendospermous seed.

Salter described the stem in 0. confertifolia as woody.  There is no anatomical explanation for this and it may simply be the result of increased lignification to support a longer stem with more leaf mass.  Stem length is related to leaf distribution and to available light, and I could find no evidence to suggest that there were any notable discontinuities between populations with respect to stem character.

Salter distinguished 0. capillacea on the basis of the plants having 3-9 leaflets per petiole, and with leaves terminally congested and petiolate.  Several populations were recorded with 5, 6 or 7 leaflets per petiole and also in which the leaves were either sessile or petiolate.  Sessile leaves are, however, always cauline (along the length of the stem) as opposed to terminal.  All gradations were found between populations with wholly sessile and those with wholly terminal, petiolate leaves, and at one locality west of Swellendam (Bayer 1300) both these types of plants were observed at the one site.  Three new collections from the Ceres Karoo included one in which the plants had terminal leaves with broad elliptic or obovate leaflets, and in one of these the plants were acaulescent.  In Bayer 1300, the leaves were partly cauline and had seven leaflets each.  In Bayer 1285, such leaves were all cauline.  The linear to linear-cuneate leaf shape was fairly consistent for all populations, except those of the western Ceres Karoo.

Floral characters were all extremely variable, although consistent within populations.  Many different colours were recorded also with great consistency within populations despite all style lengths (suggesting sexual propagation?).  Flower size was variable between populations: from 22mm in the smallest to lOOmm long in the longest.  Petal width also varied from 4-14mm wide, without any apparent correlation between the two variables.  The long filaments were invariably toothed, occasionally only gibbose, unequal and glandular, while the ant hers on both long and short filaments were horizontally orientated.  The carpels are fused and the ovule numbers varied from 2-7.  Density of the idioblastic vesicles on the carpels was variable and occasionally inconspicuous.

The overall distribution of the species in question extends from at least Oudtshoorn and Klein Brak near Mossel Bay, throughout the south-western Cape, as far south and west as Bredasdorp and Infanta.  Elements are found in the western Cape from Hercules Pillar (a favourite stopping point of Salter’s who, however, did not find this species there) to Gouda, Piketberg and Porterville.  Only three collections have been made in the Little Karoo: one from the Burger Pass west of Montagu, another from southeast of Touws River, and a third from Dysselsdorp near Oudtshoorn.  A single bulbless specimen collected by R.D.A. Bayliss from near Joubertina is possibly this species.  The species is common in the southern Tanqua/Doornriver (Ceres) Karoo and also in the southern Koue Bokkeveld.  There appears to be a gap in the distribution of the species from here to the Botterkloof and Nieuwoudtville areas where it is again common and variable.  Salter described four forms of his Oxalis grammophylla from this area, as well as 0. melanograpta from the eastern Knersvlakte and 0. massoniana from the edge of the Nieuwoudtville escarpment.

Salter’s species, as well as many of the variants mentioned here, indicate the number of elements which could perhaps satisfy some requirement of discontinuous variability from the main body of the species.  0. confertifolia, a large, red-purple flowered element from around Swellendam and Heidelberg, could particularly be regarded as discrete.  At Mardouw, west of Swellendam, this sessile-leaved, trifoliate element grows immediately adjacent to a yellow, terminally-leaved, seven-foliate element.  This yellow form is found from Mardouw southwards and westwards to McGregor, and it breeds true from seed after hand-pollination.  However, around Mardouw there are populations with lilac, white, or salmon-pink coloured flowers, and in which the leaves are partly sessile and partly terminal.  These could be interpreted to be of hybrid origin and suggest that two distinct species are involved.  A similar situation arises at De Doorns where 0. pardalis as a yellow, sessilifoliate form grows a short way away from a coppery-pink terminal-leaved form.  Again at Rooihoogte Pass and Moddergat (near Villiersdorp), there is Oxalis camelopardalis as a glabrous, red-purple, sessilifoliate form, within kilometres of a series of populations of a ciliate, lilac-coloured, partly sessile and partly terminal-leaved form.  At Karooport (Ceres) there is a trifoliate, terminal-leaved form near to a five-foliate form.

It should be clear from the above that there is great difficulty in arriving at a taxonomic rationale which adequately expresses the situation.  A derived philosophy of a species comprising a series of interbreeding or potentially interbreeding individuals which are continuously variable in space and time can only be fitted if all the necessary information is available.  Obviously, a broad field study cannot accurately analyze or assess the many technical and genetic aspects of breeding or the detailed statistics of morphological variability.  Even if it could, there would still be imponderables such as the Ceres Karoo forms, which, in addition to some morphological criteria, are also uniquely night-flowering.  In fact a manuscript name was coined on the basis of their remarkable carnation-like scent.  But fragrance has since been observed in many other species as well.  The night-flowering character implies that this element has its own pollination syndrome and the capacity to eventually pursue a discrete species path.  Present recognition would simply beg the question of what the situation is regarding its morphological and geographic complements.  It certainly indicates the scale at which collecting needs to be done to sample adequately.

The conclusion reached here is that the most practical solution is to regard Oxalis pardalis also as a ‘group-species’, recognisable immediately by a characteristic and unique bulb.  The blackening idioblasts on the leaflets, sepals and on the ovaries are confirmatory.  0. purpurea, in which the leaves and flowers are in fact very different but which also has similar idioblasts, the hairs on the outer bulb-tunics are vertically orientated.  A further close relation between these species is most strongly nullified by the fact that in 0. purpurea the carpels are free and the ovules many and biseriate.

The extrapolation of the above treatment of Oxalis pardalis on the basis of Vavilov’s principle of homologous variation (Davis and Heywood, 1967) may have severe implications for other taxa in the genus.  However, Salter has already prepared the way for such a liberal species concept by his acceptance of ‘group-species’, and the above comments indicate that many more of his species would fall into the same category were they better known.

Additional material examined (deposited at NBG):

3119 (Loeriesfontein): 32km north of Nieuwoudtville (-AB), Bayer 2756; north of Nieuwoudtville (-AC), Bayer 2752; west of Grasberg (-AC), Bayer 1880; 10km north-east of Grasberg (-AC), Bayer 1875; east of Nieuwoudtville (-AD), Bayer 2551; Oorlogskloof (-AD), Bayer 2747; Doornbosch (-CC), Bayer 1902; 6km north of Doornbosch (-CC), Bayer 1899; north of Doornbosch (-CC), Perry 707.

3219 (Wuppertal): Tertia (-BC), Bayer 3536; north of Gydo (-CC), Bayer 2199, 2200; Skitterykloof (-CC), Bayer 2522; north-east of Karoopoort (-DD), Bayer 644.

3318 (Cape Town): Hercules Pillar (-DD), Bayer 2855.

3319 (Worcester): east of Saron (-AA), Bayer 2853; Gydo (-AC), Bayer 1909; Theronsberg Pass (-AD), Bayer 1732; southeast of De Doorns (-BC), Bayer 880; Smalblaarrivier (-BD), Bayer 626; Worcester (-CIB), Bayer 729; Fonteintjiesberg  (-BC), Bayer 1277; Chavonnes (-CB), Perry 207; Penderris (-CB), Bayer 1795, 1798; west of Meulplaas (-CD), Bayer 1789; Hammanshof (CD), Bayer 1790; Moordkuil (-CD), Forrester 50; Buitenstekloof (-DB), Bayer 1321; Roedekleigat (-DC), Bayer 1778; Elandsberg (-DD), Bayer 1291; Vrolikheid (-DD), Bayer 1285; Klaasvoogds (-DD), Bayer 1259.

3320 (Montagu): Harry Hill Road (-CC), Bayer 1300; 27km west of Swellendam (-CC), Bayer 1819; 2km north of Drew (CC), Bayer 1821; 1km north of Drew (-CC), Bayer 1820; 3km north of Drew (-CC), Bayer 1715; east of Ashton (-CC), Bayer 1719; west of Mardouw (-CC), Bayer 1716; 4km east of Ashton (-CC), Bayer 1725; 8km east of Ashton (-CC), Bayer 1714; Mardouw (-CC), Bayer 1727; 13km north of Mardouw (-CC), Bayer 1717; 15km west  of Mardouw (-CC), Bayer 1718.

3322 (Oudtshoorn): Dysselsdorp (-CIB), Bayer 5772.

3419 (Caledon): 25km east of Caledon (-BA), Bayer 1810; 15km east of Caledon (-BA), Bayer 1808; 36km east of Caledon (-BA), Bayer 1809; north of Krige (-BA), Bayer 1805; Olifantsdoornpoort (-BB), Bayer 1292; Skietpad (-DB), Bayer 2247; Mierkraal (-DB), Bayer 2270.

3420 (Bredasdorp): Napky (-AB), Bayer 1814; Rooivlei (-AC), Bayer 2254; Kathoek (-AD), Bayer 2553; 10km west of Heidelberg (-BB), Bayer 1738; 14km west of Heidelberg (-BIB), Bayer 1734.

3421 (Riversdale): 10km east of Heidelberg (-AA), Bayer 1741; Draaihoek (-BA), Bayer 2320; Cooper Siding (-BB), Bayer 1745.

3422 (Mossel Bay): 20km north-east of Mossel Bay (-AA), Bayer 1748; Klein Brak (-AA), Bayer 2301.


2. Oxalis livida Jacq. Oxal.Monogr. :39 t.8 (1794); Salter, in J.S.Afr.Bot. 6:11 (1940), et :84 (1944).  Type: Iconotype ut nominatus, Oxal.Monogr.:38 t.8 (1794).  Oxalis dentata Jacq. Oxal.Monogr. :39 t.7 (1794); Salter, l.c. :83 (1944).  Type: Iconotype ut nominatus, Oxal.Monogr.:39 t.8 (1794).  Oxalis lateriflora Jacq. Hort.Schoenbr. 2:41 t.204 (1797); Salter, l.c.:84 (1944).  Type: Iconotype ut nominatus Hort.Schoenbr. 2:41 t.204 (1797).  Oxalis phellandroides E. Mey. ex Meisn., in Hook. London Journ.Bot. 2:54 (1843).  Type: Postelynberg, Drege (?).

Plants: are shortly caulescent, often branched, usually sparsely glandular, 150-300mm high.  Bulb: ovoid, shortly pointed, up to 30mm diam., new bulb comprising 8-11 outer, dark-brown, protective scales; scales to l0mm wide with 5-8 raised ribs, expanded above into membraneous wings with fringed margins, 7-8 pale creamy-white nutritive scales with some hairs at tips, arranged in 2-3 parastichies.  Rhizome: 20-5Omm diam., glabrous, with or without lateral stolons.  Stem: up to 7mm diam., 10-15Omm long, occasionally branched, smooth or sparsely glandular. Leaves: few to 20, usually glabrous, terminal or occasionally cauline, petioles up to 6Omm long, sparsely glandular; leaflets three, shortly petiolulate, occasionally with hairs on petiolules and on midrib, obtriangular, usually deeply bibbed and livid below, often with glandular and simple hairs admixed below, noncallussed.  Peduncles: 2-3, usually cauline and preceding leaves, 50-15Omm long, glandular; lanceolate, callussed bracts at upper articulation; pedicels 20-40mm long, glandular.  Sepals: ovate-acuminate, with apical orange callusses, glandular and with adpressed short white hairs, often purple margined.  Corolla: pale lilac to purple, petals 12-2Omm long, glandular, funnel-shaped greenish tube; petals cuneate, obliquely subtruncate, often with purple margin below.  Filaments: the longer is subequal, toothed, sparsely glandular; anthers abaxially sub-vertical.  Ovary: carpels adnate and with suture to base, 1-2 loculed, sparsely glandular or with setose hairs above, style setose below, glandular above or smooth.  Capsule: not exserted.  Seeds: endospermous.

The species in the section Cernuae Knuth are characterised by an umbellate, many-flowered inflorescence.  Salter recognised 23 species in 5 sub-sections.  Of these, five are Namibian, six are broadly South African, and the rest occur in the winter rainfall region.  I have experienced difficulty in identifying the rather limited number of collections in this section, and much more collecting, particularly in northern Namaqualand, will be needed to resolve problems here.  The section Lividae Salter, clearly comprises only one species.  Salter distinguishes his three component species only on the behavioral characteristics of the bulbs.  In Oxalis livida the contractile root was said to operate vertically, in 0. dentata horizontally, whereas in 0. lateriflora there was said to be no contractile root and only lateral runners.  In my series of collections, there appeared to be only one species in which any such distinctions were obscure.  In some collections the bulbs in one clone were oriented either vertically or horizontally under the influence of contractile roots, while some bulbs were without such roots.  As mentioned before, the formation of a contractile root is often the response of the bulb to shallow placement in the soil.  Any other differences which could be inferred from the respective descriptions, such as the depth of the division of the bibbed leaflets, or on colour, are supercilious when viewed against the variation within Salter’s own ‘group-species’.  Salter could have merged these three species merely for consistency.  My experience with other species, as noted in the introduction, is that vegetative behaviour is variable within, and barely a basis for recognition of, species.

Thus, Oxalis livida is a species characterised by its bulb with up to 11 outer, protective scales which are ribbed below and which broaden above into expanded, membraneous wings with fringed margins; and 8-9 nutritive scales arranged spirally in the parastichies 2 and 3.  The bulb is not ‘beaked’.  The plants are usually clothed with glandular, or both glandular and simple hairs, and flowering usually occurs before vegetative growth.  The fleshiness of the leaves and purplish coloration of the short thick stems may also be distinctive.  There is a close resemblance vegetatively to 0. bifida Thunb. but this latter species has an usually branching stem and an one-flowered peduncle.

Additional material examined (deposited at NBG):

3318 (Cape Town): Tygerberg (-DC), Bayer 2731.

3319 (Worcester): west of Tulbagh (-AC), Bayer 4277; Montpelier (-AC), Bayer 3461; Boontjieskraal (-AC), Bayer 2728; Sandberg Hills (-BC), Bayer 836; Fonteintjiesberg (-CB), Bayer 1285b; Moddergat (-CD), Bayer 1252.

3420 (Bredasdorp): Stormsvlei (-AA), Fascio 15.


3. Oxalis goniorrhiza Eckl.& Zey. Enum. 1:89 n.699 (1836); Salter, in J.S.Afr.Bot. Supp. 1:276 (1944).  Type: Stellenbosch in Zwartland, EckI. and Zey. 699 (holotype?). 0. semiglauca Eckl.& Zey. Enum. 1:? (1836).  0. semiglauca (Eckl.& Zey.) Sond., in Fl.Cap. 1:319 (1860) (holotype?).  0. urbaniana Schltr. Eng.Bot.Jahrb. 27:157 (1900): Salter, l.c. :255 (1944).  Type: Saron, Schlechter 7859 (holotype, B).  0. callimarginata Weintr. Journ.Bot. 44:209 (1931); Salter, l.c. :256 (1944).  Type: Saron, Leipoldt in BOL 18619 (holotype, BOL).

Plants: caulescent, up to 200mm high, occasionally branched, glabrous to sparsely pubescent, or with glandular and simple hairs admixed.  Bulb: clustering, ovoid, up to 15mm long and 8mm diam., new bulb comprising sub-entire, membraneous, outer tunic; cochlear, 3-5 ribbed, hard, protective scale enclosing a second scale also with a raised mid-rib, occasionally a third more membraneous unribbed protective scale, two subequal, white, nutritive scales, all with sharp apical points.  Rhizome: 20-30mm long, occasionally with stolons.  Stem: up to 150mm long, glabrous to sparsely pubescent, occasionally also glandular, sometimes branching, with several epetiolate or sub-petiolate cauline leaves. Leaves: lower leaves cauline, terminal leaves shortly petiolate, petioles 5-15mm long; leaflets three, linear-cuneate, conduplicate, incised at apex, 5-8mm long, 2-3mm wide, glabrous to sparsely pubescent below with simple or admixed hairs, usually callussed with apical and sub-marginal callusses.  Peduncle: one-flowered, axillary, glabrous or sparsely pubescent with two alternate, lanceolate, callussed or inconspicuously callussed bracts near calyx.  Sepals: lanceolate, acute, purple margined, usually callussed, rarely glabrous.  Corolla: 15-2Omm long, glabrous, funnel-shaped, yellow tube; petals obovate, obliquely sub-truncate, white, pale pink or lilac.  Filaments: the longer unequal, toothed, sparsely glandular, anthers horizontal; the shorter with anthers sub-horizontal.  Ovary: carpels fused, 3-5 locules, uniseriate, glandular and ciliate above, style, ciliate below, glandular above.  Seed: arillate, without endosperm.

The three elements as recognised by Salter were accommodated in two different sub-sections of the section Angustatae.  Oxalis urbaniana and 0. callimarginata were placed in the sub-section Sessilifoliatae despite having relatively petiolate leaves, and 0. goniorrhiza in the sub-section Lineares despite a comparatively leafy stem.  Comparison of 0. goniorrhiza with 0. recticaulis Sond., which was also placed by Salter in the sub-section Sessilifoliatae, highlights this anomaly because 0. recticaulis has no more a leafy stem than does 0. goniorrhiza.  In fact in one specimen of the latter cited by Salter, the leaves are entirely terminal.  This kind of anomaly indicates just how difficult it is to use the revision for identifications.  Variation in 0. pardalis already indicates that petiolate versus non-petiolate, or sessile-cauline versus petiolate-terminal leaves, may be encountered within the same species, and this also occurs in 0. goniorrhiza.

Oxalis urbaniana and 0. callimarginata were both collected at Saron.  The latter was regarded as a stouter species with more distinct bulb ridging, distinctly callussed leaflets, shorter petal claws and shorter teeth on the filaments.  My own collections from there do not support two such closely related species and differ in detail from both of Salter’s circumscriptions.

Oxalis goniorrhiza is, therefore, regarded as a species characterised by a bulb with 4-6 acutely angled, vertical ridges, contained in a sub-entire membraneous outer tunic.  It occurs in damp, marshy situations and is usually white-flowered.  It is not yet certain if characters other than the bulb can be used to easily separate this species from 0. multicaulis Eckl.& Zey. and 0. recticaulis, which are very similar, if also in fact distinct from another.  The synonymy of the elements merged here and in the preceding two species, emphasises a weakness trenchant in at least the history of South African systematic botany.  This is the general failure to first determine what a species is, and then the develop a species concept which can properly accommodate the variability in and of the local flora.


Additional material examined (deposited at NBG):

3318 (Cape Town): Hercules Pillar (-DD), Bayer 2887.

3319 (Worcester): north of Saron (-AA), Bayer 2848; Kluitjieskraal (-AC), Bayer 1349; Wollesley (-AC). Bayer 2902; Romans River (-AC), Bayer 2903; Darling Bridge (CA), Bayer 2904; Worcester (-CB), Bayer 2907; Brandwacht (-CB), Bayer 689; east of DuToits Kloof (-CB), Bayer 871; north-west of Worcester (-CB), Bayer 843; Rooihoogte (CD), Bayer 678; Brandvlei (-CD), Bayer 860; Kwaggaskloof (-CD), Bayer 2887.



Oxalis pardalis Sond.
(=0. camelopardalis Salter).
(=0. capillacea E.Mey. ex Sond.).
(=0. confertifolia (0. Kunze) R. Knuth).
(=0. grammophylla Salter).
(=0. heildelbergensis Salter).
(=0. leptogramma Salter).
(=0. lineolata Salter).
(=0. massoniana Salter).
(=0. melanograpta Salter).
(=0. robinsonii Salter & Exell).

Oxalis livida Jacq.
(=0. dentata Jacq.).
(=0. lateriflora Jacq.).
(=0. phellandroides E. Mey. ex Meisn.).

Oxalis goniorrhiza Eckl.& Zey.
(=0. semiglauca Eckl.& Zey.).
(=0. urbaniana Schltr.).
(=0. callimarginata Weintr.).


I would like to thank P.V. Bruyns and E.G.H. Oliver for commenting on the manuscript.



Bayer, M.B. 1984.  The Cape Flora and the Karoo.  Veld & Flora 70:17.

Davis, P.H. and Heywood, V.H. 1967.  Principles of Angiosperm Taxonomy.  Oliver and Boyd. Edinburgh and London.

Ornduff, R. 1973. Oxalis dines, a new species from the western Cape.  J.S.Afr.Bot. 39:20.

Rutherford, M. and Westfall, R.H. 1986.  The biomes of South Africa.  Memoirs of the Botanical Survey of South Africa no.54.

Salter, T.M. 1944.  The genus Oxalis in South Africa.  J.S.Afr.Bot. Suppl. Vol.1.  National Botanic Gardens of South Africa, Kirstenbosch.