The Eriosema cordatum complex . II . The Eriosema cordatum and E . nutans groups

The Eriosema cordatum E. Mey. complex is segregated into a number o f species. E. cordatum E. Mey. is retained as a polymorphic species and two allied species, E. lucipetum C. H. Stirton and E. zuluense C. H. Stirton are described as new. Four additional species E. buchananii Bak. f., E. nutans Schinz, E. psoraleoides (Lam.) G. Don and E. parviflorum E. Mey. are revised and excluded from the E. cordatum group.

In 1975 Stirton concluded that E. cordatum could be segregated into five taxa which, pending further study, he named A to E. Taxon A is retained as a polymorphic species.Taxon B and taxon C are united into a new species E. lucipetum.Taxon D is described new as E. zuluense.Taxon E has already been described as E. gunniae C. H. Stirton (Stirton, 1981), so will not be dealt with further.The remain ing species are correctly identified as E. buchananii, E. nutans, E. parviflorum and E. psoraleoides.

Vegetative parts
Habit.This complex comprises perennial herbs, suffrutices and shrubs.E. psoraleoides is the only true shrub and often grows up to 1,5 m high.The suffrutices are low growing woody plants with annually regenerated growth points; E. nutans, E. buchananii, and E. parviflorum.The remaining taxa all die back completely during the winter months to regenerate aerial parts each spring.Growth form may be erect, ascending, decumbent or prostrate.Rootstocks.All species develop perennial root stocks which, being quite variable in form, provide excellent key characters for fieldwork, especially in the absence of flowers and fruits.E. cordatum is readily separated from all other species by its distinc tive daucate rootstock (Fig. 1).E. psoraleoides has the most robust and branched rootstock (Fig. 2).There is usually a short pseudo-stylopodium at the apex of the rootstock.Root nodulation occurs in all the species treated in this paper (Grobbelaar, Van Beyma & Todd, 1967;Grobbelaar & Clarke, 1972;Stirton, 1975).Freshly exposed rootstocks, with roots and nodules still intact, will, if left in water for 24 hours, produce a pungent mustard-like odour.The nodules swell considerably.Stems.The perennial herbs have strict stems that are either terete or angular.Up to fifteen stems may arise from the short pseudo-stylopodium at the apex of the perennial rootstock.Complex branching is common in the suffrutices and in the shrub E. psoraleoides.Two species are worth describing in detail.
The main stem in E. parviflorum is very short.This results from the suppression of the leader con comitantly with the production of lateral branches.The lateral branches grow out perpendicularly to the main stem and are produced close to the ground.Their growth begins once the leader has reached ap proximately 50 cm in height.The side branches grow very quickly.If the leader dies back the plant assumes the shape of a low bowl.The characteristic low spreading nature of this suffrutex is possibly at tributable to the sudden retardation of growth in the leader stem.
In contrast to that of E. parviflorum the stem of E. psoraleoides is erect and continues to lengthen once branching begins.However, if the leader is damaged, the lateral branches soon compete to lead.This results in the production of a short rounded shrub in contrast to the usual tall, erect, somewhat spreading shrub.Short rounded shrubs are commonly found in Kwazulu and on closer investigation the leader will be found to have been damaged by a borer insect.E. psoraleoides also produces tall, densely branched shrubs, with a compact appearance.This form results from the production of numerous watershoots that arise directly from the pseudo-stylopodium and this form can be expected to be found in areas subject to burning.Vesture.The nature of the hair covering is determin ed by the length, direction, form, and quantity of the hairs considered collectively.As St earn (1973) in dicated, these characters should be stated individual ly as there are more types of vesture than there are terms to designate them.It is most unfortunate that there is no standard hair terminology for, as Roe (1971) pointed out, this would enhance the tax onomic value of hairs.As vesture forms a useful character in the characterization of taxa in the pre sent study, I have decided to follow Grear's (1970) treatment of the American species of Eriosema by adopting Lawrence's (1960) definitions of hair types.In Eriosema all hairs are simple.Both glandular and non-glandular hairs are represented.There are two types of glandular hairs: firstly, very short stalked globose glands (white, yellow, orange or red) that vary considerably in size among the different species; secondly, uniseriate hairs with bulbous bases.The non-glandular hairs are uniseriate and vary in colour from hyaline to white, grey, tawny or ferruginous.
Leaves.Trifoliolate leaves are more common than unifoliolate leaves.Occasional tetrafoliolate leaves occur in E. psoraleoides.Cotyledons are always unifoliolate.In all taxa the first two leaves of each stem are nearly always unifoliolate.Subsequent leaves may be unifoliolate or trifoliolate.Grear (1970) reported a similar situation in America.He mentioned that the 'first-formed leaves in trifoliolate species, even in strong shoots of old plants, are always unifoliolate, even up to a fifth or higher node'.The South African species commonly referred to as unifoliolate are now known to include trifoliolate-leaved plants as well.Since many herbarium col lections are made before the fourth internode has elongated, it is often difficult to decide whether the first formed three unifoliolate leaves are the basal leaves of a unifoliolate or a trifoliolate species.Similarly, it commonly happens that the basal leaves have abscised by the time the fourth and higher inter nodes have elongated.Such plants are often referred to as trifoliolate-leaved plants.These so-called trifoliolate-leaved species may include unifoliolate leaved plants, but less rarely so than the so-called unifoliolate-leaved species including trifoliolateleaved plants.For these reasons the use of leaf struc ture as used by previous authors has been dropped as a key character for the delimitation of taxa in the £.cordatum complex.
The size, shape, vesture and texture of the leaves have been found to be useful definitive characters.However, as will be indicated later, both size, shape and vesture exhibit a remarkable variation under dif ferent environmental conditions.An appreciation of this variation alone will do much to help in the under standing of the anomalies encountered in the keys of Harvey (1862) and Burtt Davy (1932), both of whom relied heavily on the shape, foliolation and vesture of leaves as diagnostic criteria.
The shapes of leaves seen in this complex have been classified according to Hickey's (1973) 'Classification of dicotyledonous leaves' (see also Dilcher, 1974).As this system is not readily available, a length-breadth ratio is sometimes provided in brackets after the description of each shape.This allows for a rapid conversion of Hickey's terms to the equivalents of any other system.All the leaf measurements given are those of mature terminal leaflets, or if there are no trifoliolate leaves, then of the mature unifoliolate leaf.Petioles.Petioles are either subsessile or elongated.Most are channelled and ridged on the adaxial sur face.Length has been found to be a useful character in some instances, but in others this parameter shows marked plasticity.Stipules.Stipules are a key character in Eriosema.They may be free and laterally positioned in relation to the petiole, that is, if the petiole was held tightly against the stem the stipules would flank it (Fig. 3a); or they are connate and opposite the petiole, that is, would not flank the petiole if it was held tightly against the stem (Fig. 3b).
The use of the stipule as a key character requires careful observation as in some species the connate stipules are wont to split to the base, and may therefore be mistaken for a pair of free stipules.This disadvantage is readily overcome however by a careful inspection of the hair line along the margin of the stipule (Fig. 4).It can be seen that in free stipules the hair line extends uninterruptedly to the base (Fig. 4.1a) whereas in connate stipules that have split, the hair lines end abruptly (Fig. 4.2b).The split margin is also uneven (Fig. 4.2c).Connate stipules usually split as a result of the expansion of nodal tissue.
The stipules are either tightly clasped to the stem, patent, or recurved.In some taxa the stipules persist but senesce rapidly almost before the young leaflets expand, whereas in other taxa the stipules persist but remain green until almost the end of flowering.In E. psoraleoides the stipules are often caducous.These are valuable field characters.

Reproductive parts
Inflorescences.In all taxa the inflorescence is an ax illary raceme bearing (1 -)-50 flowers.Flower number is too variable to be of much diagnostic sig nificance.The length of the peduncle is useful in diagnosing E. psoraleoides as in this species it is always less than one third of the length of the raceme.Flower bracts are either caducous (£.psora leoides) or persistent (all other spp.);shorter (£.nutans), or equal to (£. cordatum) the calyx.Bract vesture may be sericeous, pilose, pubescent or stiff patently hairy.Flowers.The component parts of the flower vary considerably in shape, colour, pubescence, gland encrustment and dimensions.Flower colour has proved to be a useful distinguishing character.Flowers are consistently either pure yellow, or red and yellow, or pale pink with purple stripes, or red, yellow and orange, or orange and yellow.Herbarium collections of all but the yellow-flowered taxa turn purple on drying: yellow' flowers retain their colouration.

Flowers usually hang downwards before they open.
As they open they tend to rise into the horizontal plane but return to their former position when they close.(a) Calyx.All taxa have a five-lobed campanulate calyx with distinct triangular or lanceolate teeth (Fig. 5a).Three parts of the calyx are referred to in this study: calyx lobes, calyx teeth and calyx tube.The fused portion of the calyx is called the calyx tube (Fig. 5a) whereas the unfused portions are called the calyx teeth (Fig. 5b).The fused and unfused portion of each segment together is called a calyx lobe.The lobes are measured from the apex of each tooth to the base of the tube directly beneath each tooth.Other authors use the term lobe instead of tooth.Three types of calyx lobes have been distinguished.1. Keel lobe (Fig. 5a. 1) This lobe lies directly beneath the keel and is thus abaxial (Fig. 5.1). 2. Lateral lobe (Fig. 5a.2).There are two lateral lobes.These lobes are situated one on each side of the flower (Fig. 5b.2).They are always equal in length and upcurved towards the vexillar lobes.3. Vexillar lobe (Fig. 5a.3).There are two vexillar lobes.These lobes are situated on the upper (adaxial) part of the flower.The vexillar lobes may be markedly connate or free and this is a useful character in some taxa.
The length, pubescence and degree of fusion of the calyx lobes; the shape and length (in relation to calyx tube) of the calyx teeth; and finally the length of the calyx lobes (in relation to the length of the flower) are all useful diagnostic criteria.(b) Standard.The standard varies in colour, size, shape and structure.Hickey's (1973) classification has been used to determine shapes.Length-breadth (1/b) ratios are given as was done for leaves.Shape varies from narrow obovate (2:1) to wide obovate (1.2:1, £. psoraleoides), or oblong (2:1, £. buchananii).The apex may be rounded, emarginate or hooded.
Auricles vary in size, and range from prominent in flexed flaps (E.buchananii) to small slightly inflexed flaps (E.cordatum).The appendages of the standard are important diagnostically.These are situated on the inner surface just above the auricles (E.cor datum), well above auricles but less than half-way up the blade (E .zuluense), or high up and above the middle of the blade (E.buchananii).The appendages may be free (E.buchananii), partly connate at base (E.parviflorum) or connate (E .nutans).E. psoraleoides and E. lucipetum (anomalous popula tion Stirton 1340) are both characterized by the absence of any appendages.
The back of the standard is nearly always pubes cent and glandular.In E. psoraleoides it may also be glabrate.Two unusual specimens Strey 5048 and Vahrmeijer 1121, which have been included in E. cor datum also have glabrate or glabrous standards.
The colour of the standard varies but it is constant within the species under study.It is yellow in E. cor datum, E. lucipetum, E. zuluense, E. nutans, E. par viflorum and E. psoraleoides.In E. buchananii it is pale pink lined with purple.E. cordatum has a dull red outer surface with the inner surface orange or yellow and distinctly purple-veined.
A characteristic of all taxa is the presence of a semi-circular patch of intense yellow situated just above the appendages.This patch is noticeable even in the yellow flowered taxa where it tends to be a more intense yellow.It is referred to as a 'nectar guide' in this study.(c) Wings.At first glance the wings appear to be uniform.Their infrequent use as a diagnostic character has probably been due to the difficulty en countered in describing them.The wing consists of a lamina which may be cultrate (E.nutans), cultriform (E.cordatum), or narrow oblong (E.psoraleoides).The single auricle may be straight (E.psoraleoides), or forward sloping (E .nutans).The claw is either at tenuate (E.nutans) or strongly recurved (£.zuluense).
The wings may be shorter, equal to or longer than the keel blades.Colour varies from yellow, yellow with flushes of pink, to orange.(d) Keel blades.The keel blades show various degrees of fusion.The shape is difficult to describe but ranges from narrow oblong narrowing towards the claw (3: 1, E. psoraleoides) to variously curved to almost falcate abaxially.The apex may be obtuse, falcate or rostrate.The auricle is less defined than in the wings.It may be recurved, vertical or forward sloping.The claw is attenuate.The keel blades are usually wider than the wings.The keel blades may be equal to (E .psoraleoides), shorter than (most species) or longer than (E.parviflorum) the wings.The lamina may be sparsely or densely encrusted with glands.The base line of the keel blade may be glabrous or hairy.The keel blade is generally greenish white in colour.(e) Androecium.The stamens are diadelphous with the vexillar stamen free.The vexillar stamen has a knee-like thickening near the base.The thickness of this knee is usually proportional to the depth of the claw of the standard.The fused stamens are collec tively referred to as the staminal sheath.The anthers form two whorls when the sheath is closed as alter nate filaments are unequal in length.Anthers are uni form, dehiscing longitudinally.(0 Gynoecium.The length of the gynoecium and the variation in its components are important characters. The ovary may be stalked, subsessile or sessile.It is biovulate (rarely triovulate in E. psoraleoides) with two campylotropous ovules.The type and degree of vesture on the ovary wall is a useful character.The hairs may extend less than half the length of the valves.The style is glabrous, or minutely hairy and glandular, or eglandular.It is variously thickened at, or beyond, the point of curvature.The height of cur vature is a useful character and is calculated as the vertical distance from the base line of the style to the level of the stigma (Fig. 6).The erect portion of the style may be straight or incurved.The stigma is capi tate, inserted.(g) Discoid floral nectary.Whilst floral* nectaries have been frequently reported in papilionoid legumes, particularly in the tribe Phaseoleae (Waddle & Lersten, 1974), their presence in the genus Erio sema is poorly documented.This is probably owing to their small size and also to the difficulties involved in the interpretation of their structure.Phillips (1951) mentioned that the ovary in Eriosema is 'usually sur rounded by a small cupular disc'.The only other report found was Waddle et al's (1974) report of the presence of a discoid floral nectary in Eriosema rufum Don.There are no reports on the presence of floral nectaries in the American species (Grear, 1970).Discoid floral nectaries have been found in all the species studied in the present work.
The ontogeny of discoid floral nectaries remains obscure.Moore (1936) thought that they resembled a staminal sheath and were more closely associated with the staminal whorl than with the gynoecium.Waddle (1968) suggested alternatively that the disc nectary was not a reduced whorl but an outgrowth of the receptacle.The solution to this puzzle of mor phological interpretation will involve, as Waddle et al. (1974) pointed out, a careful ontogenetic study.The discoid floral nectaries in all the taxa studied here were securely adnate to the base of the staminal sheath.In view of the lack of ontogenic evidence the discoid flora nectary has been considered in the des cription of species as in independent structure.
The sizes and shapes of the floral nectaries are variable.The apical margin may be wavy, erose, smooth, finely dentate or revolute.The type of margin appears to be correlated with the quantity of nectar secreted.Nectaries with irregular margins were seen, in the field, to produce more nectar at one time than the smooth margin nectaries.Frey-Wyssling (1955) has reported that the relative propor tions of xylem and phloem in nectaries affects the type of secretion.Nectaries which are supplied most ly by phloem tissue tend to secrete nectar with a high sugar concentration.It is difficult to interpret the meaning of the lack of vascular tissue reported in the nectary of Eriosema rufum Don (Waddle et al., 1974).A study is needed to correlate types of vascu lar tissue with nectar secretion in Eriosema and then to correlate the nectaries with pollination mechan isms.A useful study could be made of the structure, morphology, stomatal distribution and ontogeny of nectaries in papilionoid legumes.I am not aware that such a survey has been made for the family as a whole.
Fruits.The fruits are compressed 2-valved, 2seeded dry dehiscent pods.The valves are obliquely beaked, oblong (2:1) to wide oblong (1, 5:1), or trapeziform.The valves are smooth and shiny inside, and usually sericeous, velutinous, pilose hairy or shortly pubescent outside.The valves dehisce along upper and lower sutures until they remain attached only at the base.Most of the South African species dehisce with an explosive action.This may be due to an environmental influence as pods of the same species have been seen to dehisce very slowly during cloudy overcast conditions.The separated valves coil very tightly once the seed has been shed.Seeds.Mature seeds are smooth, dull or shiny, com pressed, transverse or oblique.Shape may be oblong, reniform or somewhat mango-shaped.Seed colour varies from black or deep purple to chestnut brown to khaki, with or without speckles and mottles.Polymorphic seeds are commonly produced in a species but although both types are never found on the same plant they may occur in the same popula tion.In E. parviflorum the black seeds are thicker and narrower than the speckled khaki seeds (see under Insect infestation).
The hilum of the Eriosema seed is linear and ex tends almost the whole length of the seed.The sub horizontal funicle is attached at the inner extremity.The thick bilabiate flange flanking the hilum is refer red to as a strophiole (after Grear, 1970).The attach ment of the seed of the valve wall can be seen in Fig. 7.
Seeds swell to almost double their size when they imbibe water.This effect is also seen in the unripe pod.When the pod is green and fully developed the two seeds almost fill it, but as the pod blackens and dries out the seeds shrink to almost half their previous size.

ECOLOGY
Throughout this study field observations have been aimed primarily at assessing the range of varia tion within the taxa under study.If the field observer repeatedly meets a particular phenomenon, or range of variation, in the individuals or populations under study and this is frequently associated with a par ticular set of environmental or local habitat condi tions, there is some justification for associating these where either one, or the other, is encountered in new field situations.Therefore a field observer experienc ed in the study of a particular group of plants, should be able to draw upon his experience when meeting new plants of the same genus in new localities.Under these conditions his intuitive assessment of variation in relation to general habitat conditions should be reasonably accurate, despite his inability to relate a particular feature of variation to a particular en vironmental factor.
If, in the consideration of variation within the species in this study, an environmental factor, or plexus of factors, are suggested as being related to some particular morphological form, or feature of it, this is the outcome of field observation and is regard ed as a reference point to be kept under considera tion, rather than as a factual statement of cause and effect.
This section collates various field observations that have been useful in understanding the species under study.Selected examples of infraspecific variation are presented in order to provide as broad a picture of the species as possible.Fortunately most of the ex amples presented are indicative of the types of varia tion, and phenomena, to be found in the whole E. cordatum complex.Where relevant more detailed ac counts are given under each species.

Field observations
Plants growing in full sun tended to be shorter and more compact than plants of the same species grow ing under different intensities of shade.Similarly plants that grew' in burnt veld had a more stunted form than did plants of the same species which grew in adjacent unburnt veld.These phenomena, which apply to all species in the E. cordatum complex, were fully discussed in a previous paper (Stirton, 1977).
E. cordatum was observed to exhibit a markedly greater variability than any other species in the com plex.Plants that grew in the same population showed a variation in the size, shape and foliolation of the leaves, degree of compactness of flowers on the in florescence, length and thickness of the peduncle, shape and senescence of the flower bract and finally the size of the standard.Fig. 8 shows the variation which can occur in the length of the inflorescence within the same population.
It was found that plants of E. cordatum that grew in tall grass had bigger leaves, longer internodes, shorter inflorescences and smaller standards, than those plants that grew in short grass in more open areas.In other populations the peduncle was very long in tall grass and short in short grass.Allsopp (1947) wrote that plants of E. cordatum become suppressed if the veld remains unburned, or is heavily grazed and trampled.This information ap pears to have been based only on the inspection of field sites.
The combined influence of wind and sun appear to affect Eriosema plants.In prolonged hot windy weather these plants tended to die back from the growing points and to shed their basal leaves.This observation is important since the first dichotomy of the available keys differentiates between unifoliolate and trifoliolate leaves.It is now known that many species, E. cordatum in particular, produce their first trifoliate leaf at the fifth or higher node.Such a plant, if collected before the trifoliolate leaves had been produced, would be difficult to key out.Similarly for a trifoliolate-leaved plant that had already shed its basal leaves.The loss of unifoliolate leaves by a predominently trifoliolate-leaved plant may of course be due to natural ageing.
Hot windy days have also been observed to wither flowers.Two populations of E. cordatum were observed to have failed to set seed for an entire season because the flowers withered just as the cor ollas emerged from the calyx.

Insect infestation
Insect larvae, particularly Agromyzidae, cause ex tensive damage to various plant parts.Eleven dif ferent types of larvae have been collected and await identification and I am indebted to Dr B. R. Stuckenberg of the Natal Museum for his advice.He informed me that most species of Agromyzid flies lay their eggs in particular plants and are host specific to those plants.The observation that some larvae of these insects are found only in certain species has been a useful field guide for the field indentification of species.These studies when pursued promise to be most rewarding.The different Agromyzid larvae form distinctive patterns (mines) beneath the cuticle or epidermis.Such patterns are often most noticeable on herbarium specimens.
Further studies are also needed on the insects which cause extensive damage to seeds.In E. par viflorum both laboratory and field observations revealed that the production of polymorphic seeds was closely associated with the infestation of insects in fruit pods.Pods with black seeds were rarely af fected, whereas in pods with speckled, greenish yellow seeds, the seeds rarely reached maturity because of heavy insect damage.These seeds should be analysed comparatively for chemicals that inhibit predation.

GEOGRAPHICAL DISTRIBUTION
The geographical distribution maps presented are based mainly on the author's field collections and available herbarium material.This commonly used approach to the compilation of distribution maps has certain pitfalls that should be recognized.Firstly the distribution map is not necessarily a reflection of the taxon's extant distribution.It stands to reason that the compilation of a distribution map is an act based on an a priori hypothesis that the map will provide the basis for some deductive argument.If this is so, then a map compiled from herbarium collections spanning a century or more is indeed a shaky base for any type of interpretation.This awareness of a space time relationship is seldom considered in the com pilation and interpretation of distribution maps.The use of unchecked specimen citations gathered from the literature hardly needs any comment.The second pitfall is that a map compiled from all the available herbarium material is open to severe criticism if, as White (1971) so testily states, its interpretation con cerns unwarranted speculations as to its inclusive variation.The reason is that the known complexity of genetic systems, including hybridization, poly ploidy and apomixis, is such that any particular expression of a phenotypic character could be theor etically possible over any interval of time and at any particular point of a taxon's range.Such expression may have little or no significance in an understanding of the taxon's origin or distribution pattern.The chances of this situation occurring are greater the older are the specimens used.The third pitfall con cerns the frequent attempts by botanists to discover variables of ecological significance by a reference to distribution maps.Although not known as such, this phytocartogeographic approach has an equivalent in the zoocartogeographic approach of zoologists.An example of the problem and the confusion that may develop is shown in the critical assessment by Stuckenberg (1969) and Van Dijk (1971) of Poynton's (1964aPoynton's ( , 1964b) ) faunal study of amphibia in southern Africa.The final pitfall is obvious and common to most approaches.It concerns the correct delimitation of taxa.Even one incorrectly named specimen in a distribution map can lead to incorrect interpretations.
The known distributional ranges of all the species studied in this complex is given in Fig. 9.This com plex occurs mainly in the Transvaal, Swaziland, Natal and the eastern Cape Province.Very few col lections are known from Lesotho, Botswana and the Orange Free State.No representatives of the complex occur in South West Africa/Namibia or the western and northern Cape Province.
Eriosema cordatum and E. psoraleoides, have the most widespread distribution of the group in southern Africa, whereas E. nutans is less widely dispersed.The only really restricted distributions are those of E. buchananii, E. zuluense, and E. lucipetum.Four species of the complex extend beyond southern Africa.These are E. buchananii, E. nutans, E. parviflorum and E. psoraleoides.

TAXONOMY
The descriptions of Eriosema species that follow are based on the author's examination of herbarium material and of populations in the field in South Africa.
A selected citation of representative specimens of each species recognized in this study is given in the enumeration of species.This was prepared by the selection of a single voucher specimen from amongst all specimens studied to serve as a record of each quarter degree square through which the taxon was naturally distributed.1. Eriosema zuluense C. H. Stirton, sp. nov., E. cordato E. Mey.affinis, sed dentibus calycinis triangularibus, tubo brevioribus (raro eum aequantibus), floribus luteis, appendiculo interno vexillari manifeste supra auriculos sito distinctam.
Eriosema zuluense is endemic to Zululand and stretches from Josini Dam and Hluhluwe in the north to Mhlabatini, Hlabisa, Melmoth, Eshowe and Ngoye Forest in the south (Fig. 11).vesture, habit, branching and its triangular calyx teeth shorter than the calyx tube.Acocks 13067 is unusual in that the stipules are rapidly caducous and the leaflets narrower.The plants grow in dense grassland on rocky hillsides, particularly dolerite outcrops, between 300-400 m.Flowering occurs between October and December.
The specific epithet zuluense refers to the region throughout which they are distributed.
This difficult species has long been the dumping ground for many specimens of Eriosema in South African herbaria.The two factors which have pro bably contributed most to this confusion are the oc currence of marked phenotypic plasticity within the taxon and the occurrence of extensive hybridization (Stirton, 1981) of this species with other species.
E. cordatum remains the most heterogenous group in the E. cordatum complex and includes within its range of variation the previous varieties cordatum and guenzii.This taxon is the most widespread Eriosema in southern Africa (Fig. 12).Preliminary fieldwork has enabled me to outline some of its com plex variation.Fig. 12 is a pictorialized distribution map showing variation in leaf shape and size, relative lengths of leaf petiole and leaf rachis, stem vesture and shape of calyx lobes (Fig. 13).The areas 1-3 shown in Figs 12 & 13 are merely the known ranges of certain types of variation and do not necessarily in clude all the collections found within them.Plants tended to be decumbent and predominantly unifoliolate along the coastal areas (3).There was a general increase in size of the leaves from south to north.Coastal plants growing south of Port Edward were both unifoliolate and trifoliolate.In the Mpangazi area in northern Zululand there were a few collections which were notable for their larger flowered and sparsely haired, unifoliolate condition.This area requires further investigation.Inland from the Natal coast plants became smaller and predominantly trifoliolate.More stems were produc ed.Area 2 is an approximation of the area in which erect trifoliolate leaved plants can be expected to oc cur [for example, Stirton 1462 (Fig. 14)].Area 1 is the range of small semi-erect unifoliolate leaved plants which differed from the coastal unifoliolate leaved plants in their pubescence, leaf shape and habit.These plants, for example Stirton 1465 & 1468 were restricted to the high altitude areas of the eastern Transvaal.They appeared to have a close af finity to E. gunniaeC.H. Stirton.Although included here, their identity remains open.They warrant fur ther investigation.
Transvaal plants were predominantly decumbent, trifoliolate, and small-leaved.Leaflet apices were more acute than in the Natal populations.Lateral leaflets were narrower, more asymmetrical and more oblique than lateral leaflets of Natal plants.As ex ceptions occurred randomly throughout the area no safe conclusions could be drawn from this data.
Stem vesture was too variable to be reliable but, considered as a whole, it was noted that the majority of plants had either tawny or ferruginous, upward or downward pointing hairs.This was a feature which, considered with the presence of a daucate rootstock, connate stipules (invariably splitting), strongly reflexed red and yellow flowers, stiff haired calyx and flower bract, and with the wing greatly exceeding the rostrate keel, could be used to separate this species from all other Eriosemas in southern Africa.
The morphological diversity of this species has been discussed in some detail elsewhere (Stirton, 1981) and in the section on Ecology.In summary it may be said that no useful purpose will be served at present if it is divided into a number of sub-groups.If it is recognized that the species is polymorphic and if future field studies are aimed at understanding pat terns within the variation encountered, much will have been done towards understanding the complex as a whole.Vernacular names for this species include: Uqontsi (Bryant, 1909), Zulu; Leshetla -soft bone, Lesapo le letsehali -the female bone, and Setloli se sehola -the big jumper (Phillips, 1917;Jacot Guillarmod, 1971), Sesotho.Both Zulus and Sesothos are reported to use this plant for medicinal purposes.Bryant (1909) has an interesting chapter on impotency and barrenness.He wrote: 'With all primitive peoples, all that pertains to the sexual functions, in volving as it does the propagation of the species and the piCservation of the tribe, is a matter of para mount inportance.Impotency on either side is with them more than a disgrace, it is a calamity.Should the male organs fail altogether to produce the seminal fluid, the roots of the imPindisa (Rubia eordifolia) are boiled and drunk at bedtime, resulting in an early emission.A hot milk infusion of the roots of the Uqontsi (Eriosema cordatum and E. salignum) herb has a similar effect'.
Both Phillips (1917) and Jacot Guillarmod (1971) reported that the Sesotho mixed the Leshetla with other plants and used this as a medicine to stimulate bulls in spring.Phillips (1917) noted further that E. cordatum E. Mey. is less powerful a stimulant than E. salignum E. Mey.
Eriosema cordatum is a plant of diverse habitats and is quick to colonize disturbed habitats; areas where it readily hybridizes with other species (Stir ton, 1981).Flowering occurs throughout the summer months.
Perennial herb, 10-30 cm tall, flowering in early summer.Stems erect or decumbent, branching near the base, tawny or ferruginous.Leaves mostly 1-foliolate on erect plants and 3-foliolate on decum bent plants, basal leaves always 1-foliolate.Leaflets 8,5-11,0 cm long, 4,5-7,0 cm wide, laterals smaller, base cordate, apex acute, terminal and 1-foliolate leaves symmetrical, narrowly to broadly ovate, laterals smaller and oblique, strigose or sericeous, lower surface dull or shiny, covered with either light yellow or red glands.Stipules 10-15 mm long, 4-5 mm wide, ovate-lanceolate, oblique, free, clasping the stem, pubescent.Petiole 5-7 mm long.Racemes axillary, (10) 20-40-flowered, exceeding leaves, peduncle 5-12 cm long.Flowers yellow, rarely pink and yellow, bracts more than half the length of the flower, caducous.Calyx 6-7 mm long, lobes equal, teeth triangular or lanceolate, longer than the tube,   This species is easily separated from E. cordatum by its free stipules.In areas where the two species overlap there exist plants which may be intermed iates.Such plants, because they have free stipules, are included here in E. lucipetum.They tend to have pale pink and yellow flowers (for example Stirton 1416).Further studies may indicate that they are hybrids.
Eriosema lucipetum is rather variable.Plants from the westernmost part of the range are characterized by their decumbent habit, leaves with upper surface strigose and under surface shiny with a dense cover ing of small reddish glands, and calyx ferruginous hispid (Fig. 17).The northern populations usually comprise erect plants with sericeous leaves, the under surface dull and variously covered in white or pale yellow' glands; the calyx being white or tawny veluti nous.The western plants may bear yellow, pink and yellow and very occasionally red and yellow flowers whereas northern and eastern plants are always yellow-flowered.It is the western plants which overlap in distribution with E. cordatum.
Stirton 1344 is a problem specimen to place in E. lucipetum.Its general facies is that of the erect form of the northern and eastern parts.It differs from these however in its sharply acute ovate leaves and its differently shaped standard petal that lacks appen dages on the inner face.Among the erect plants of the eastern populations there is also an increasing tendency towards the decumbent habit the further east and south one goes.These areas are poorly col lected however and it may be that more than one species may be involved.Collections are still needed between the range of E. lucipetum and that of the group of yellow-flowered specimens included tem po: arily under E. zuluense.(See under that species.)The latter specimens are included under E. zuluense.owing to their fused stipules and shortly-triangular calyx teeth.
E. lucipetum is a plant of rocky grassveld, especial ly rocky banks and along roadsides, usually between 1 900 -2 100 m.Flowering begins in October with a peak in Novemer-December and then tails off during March.
The specific epithet lucipetum refers to the seem ingly light-seeking habit of the inflorescences as they emerge from their rocky habitat.
Eriosema psoraleoides extends from West Africa across Central Africa to Sudan, then southwards from east Africa across parts of Zambia to Angola, through Zimbabwe, Mozambique, Botswana, Swazi land into South Africa (Transvaal, Natal).It also occurs in Madagascar, the type country (Fig. 18).Eriosema psoraleoides is one of the most common ly collected Eriosema species in southern Africa.It is a very distinctive species and is easily recognized.It has been confused occasionally with E. salignum E. Mey., E. burkei Benth.and E. nutans Schinz, but is readily identified by its leaf shape and pubescence, minute stipules, absence of appendages on the stand ard, and its congested, short peduncled inflores cence.The species is restricted to the lower-lying areas of Swaziland, the Transvaal and Natal.It is usually found along roadsides, disturbed crop lands and forest margins, and along coastal sandflats in coarse grassland with scattered shrubs.It has also been recorded in swampy conditions.Field studies have shown that although variable in different environ ments, it was nevertheless consistent in its important diagnostic characters.Following the road from the Malkerns Agricultural Research Station in Swaziland to Mbabane, I noticed a decrease in size, branching, and number of flowers per inflorescence.A similar gradient occurs between coastal and inland plants.
Eriosema psoraleoides can produce flowers all year round but its flowering is heaviest between October to May with peaks between October to December and March to May.During the period of the second peak the Natal populations, as a whole, flower much later than the Transvaal populations, reaching a peak in April as compared with March.In the first peak the Natal populations flower earlier in September as compared with October-November for the Trans vaal.
On the Natal coast plants growing south of Mandini have been observed to flower at different times to those plants growing north of the Tugela River (C.J. Ward, 1975, pers. comm.).
The Venda common name for this species is mutangasiwa (Westphal 38, Codd 6881).In Afrikaans it is referred to as the geelkleurtjie, an allusion, according to Smith (1966), to their yellow flowers and general appearance to the European Laburnum.
Both Gerstner (5002) and Moll (5345) recorded that Tonga and Zulu tribesman cook and eat the ripe seeds.A medicinal use has been observed by Codd (6881) who noted that roots were used to cure inter nal disorders.These plants apparently are invoked by some Swazi tribesman during storms as a protector against lightning.A less obvious usefulness is the reputed efficacy of their leaves and roots when used as a remedy against venereal diseases; instances hav ing been noted in the Congo, Zambia and Nigeria (Watt & Breyer-Brandwijk, 1962).The same authors also report that in West Africa peasants rub leaves on their dogs in order to control lice.
In South Africa this species is confined to a few localities in the northern Transvaal (Fig. 21).It ap pears to be restricted to the North-eastern Mountain Sourveld Veldtype.It occurs between 900 and 1 490 m.The Transvaal is the southernmost limit of distri bution.It is also recorded from Kenya, Uganda, Tan zania, Zambia, Malawi and Zimbabwe.Verdcourt (1971) in his treatment of Eriosema for the Flora of Tropical East Africa reported that he had seen only one specimen of var.buchananii from South Africa.This specimen, Scheepers 814, has pro ved however to be the most atypical of the South African specimens.It has larger more distinctly ovate leaves, longer peduncles as well as a shorter lighter patent indumentum.This may be due to environmen tal influences as similar attributes have also been noted in field studies in other species in the E. cor datum complex.Scheepers 814 groups well with the  These comments hold for all the other specimens of E. buchananii in South Africa which are clearly distinct and easily separable from E. cordatum.I suspect that their relationship is probably not as close as previously envisaged.
In South Africa E. buchananii has been confused more often with two other species rather than with E. cordatum.Here it has been known under the name E. nutans Schinz [ = E. polystachyum (A.Rich.)Bak.] to which it has a very strong resemblance, and also to E. montanum Bak.f. which does not extend as far south as South Africa.E. buchananii may be readily separated from the rest of the complex by the presence of an appendage situated above or at the middle of an oblong standard, the large incurved flattened auricles and the falcate stipules with the base narrower than the middle.The appendage is characteristic as it is divided into two shallow, crescentiform ridges which lie closely one on either side of the central axis of the standard.E. nutans is separated by its smaller auricles and its undivided ap pendage which occurs just above the auricles merging into them.
Apart from the comment by Mogg (13992) that his plant was 'common in grassland ridges of gneiss', there is very little ecological information for the species in South Africa.Scheepers 814 has the following note 'north facing slope, intense sunlight to misty or windy, periodically moist well drained and aerated to dry, arid shallow soil tending toward lateritic type'.
E. buchananii flowers from November to January, with a peak in December.
There is another variety in east Africa, var sub prostratum Verde.This does not occur in South Africa.Gillett 17669 (K) from Tanzania, which has been included under var.subprostratum seems to me to be worthy of some rank.It has smaller flowers, lacks the typical buchananii stipules and has a dif ferent pubescence.
In South Africa E. nutans is confined to the Transvaal and Swaziland (Fig. 23).Elsewhere it oc curs in Zaire, Sudan, Ethiopia, Eritrea, Kenya, Uganda, Tanzania, Malawi, Zambia, Mozambique and Zimbabwe.Burtt Davy (1932) accepted both E. nutans Schinz and E. polystachyum (A.Rich.)Bak.He remarked that E. nutans Schinz 'is a close ally of E. polystachyum (A.Rich.), Bak.f. ms'.Under E. polystachyum (A.Rich.), Bak. he stressed its close affinity to E. richardii Benth.ex Bak.f.Later, however, he cited Rogers 14510 as E. polystachyum but labelled it E. richardii.His treatment of these various taxa can be understood and appreciated after reading a note on pp 15-16 of his flora.In this note on 'Synonyms and References' Burtt Davy com mented on some of the difficulties he encountered with synonomy.His example fortuitously concerned Eriosema polystachyum.He said: 'For the sake of space, also, only so much synonomy has been given as seemed requisite to correlate the species with the " Flora Capensis" and the " Flora of Tropical Africa" , or with more recent revisions and monographs.The principle has been adopted that the same name should not be used for more than one plant, even though the first name so applied has been reduced to the rank of a synonym.This rule has much to commend it in view of the fact that dif ferences of opinion will always exist as to the correct ness of reducing a name to the status of a synonym, and the consequent liability that the plant to which it was first applied will be restored to specific rank.An example in point is that of the Tropical African Eriosema polystachyum (A.Rich.)Bak. ( 1871 E. nutans has been interpreted differently in dif ferent parts of Africa, a not unusual result of the slow, spasmodic development of African botany.This name has been used rarely in South Africa.Here the most popular name has been E. polystachyum (A.Rich.)Bak.In referring to practice outside South Africa, Verdcourt (1971) made the following com ment: 'During the past thirty years this (£.nutans Schinz) has mostly been accepted as a variety of E. buchananii but I am certain that the two are best kept distinct, the latter differing in large leaves, much broader stipules, different position of the standard appendages and less constancy in flower colour.The flowers of E. nutans are yellow but only a few plants of E. buchananii have been seen with yellow flowers; they are mostly cream, pink, or purple lined with purple'.
As the presence of £. buchananii has now been firmly established in the Transvaal, I might add that Verdcourt's comments hold for South Africa, except that here plants of E. buchananii have not been recorded as yellow-flowered.
Over 50 specimens had been incorrectly named in the past.This material had been referred to E. psoraleoides (£.cajanoides), E. parviflorum, E. zeyheri and E. squarrosum.E. psoraleoides with its very small stipules, wide obovate standard without appendages, different keels, wings and pistil is easily separated from E. nutans which has an oblong stan dard with appendages and long lanceolate stipules.The two species differ also in that in E. nutans the peduncle is longer than half the length of the in florescence.In E. psoraleoides the peduncle is less than one third the length of the inflorescence.E. par viflorum is distinguishable from E. nutans by its wide obovate (1,2: 1) standard, and its allopatric distribu tion (almost restricted to Natal in S. Africa).
The most noticeable variation which occurs in £. nutans is the presence or absence of glands.Two types of glands have been found; large and small red glands and yellow glands.Gland colour may be the result of ageing or an artefact of the poisoning of herbarium specimens.Glands are mostly confined to specimens from the northern Transvaal.Glandular and non-glandular plants both extend from Tanzania to the Transvaal.Calyces are quite variable ranging from sparsely short and short toothed, to long toothed and densely short pilose.Leaf shape and size are most variable.Stipules range from short triangular and slightly falcate, to long linearlanceolate and straight.This variation may be worthy of infra-specific categorization.However, in view of the numerous difficulties of synonmy which have resulted from regional demarcations of infra-specific ranks, I have left this to later workers who may see the species in a broader African perspective.
£. nutans grows mainly in the bushveld and open grasslands and has been found on south-facing hillsides, in disturbed areas such as dipping tanks, in cultivated fields and near marshy ground.It flowers from November to May with a definite peak in February to March.
The synonomy of Eriosema parviflorum is puzz ling.There is little agreement among the three latest studies, and a' the chosen example below' indicates, it is a problem the solution of which rests in a full African study.The chosen example concerns the treatment of the name Eriosema parviflorum var.sarmentosa Staner & De Craene. Keay (1973) assign ed this to £. parviflorum subsp.parviflorum, whereas Verdcourt (1971a) placed it under £.par viflorum subsp.podostachyum (Hook, f.) J. K. Mor ton.Jacques-Felix (1971), however, excluded it from £. parviflorum completely and made it a synonym of £. spicatum Hook.f.Unlike both Keay and Verd court, he did not recognize infraspecific categories in £. parviflorum.This example forms part of a con fused pattern.Rather than add to this confusion I have decided not to recognise any infraspecific categories in the region under study until the species has been studied over its entire range.
Eriosema parviflorum extends eastwards from west Africa across the Camerouns and Congo to the Sudan and east Africa, then southwards through Zambia and Mozambique into South Africa (Trans vaal and Natal) (Fig. 25).It also occurs in Madagas car.Eriosema parviflorum E. Mey. is mostly restricted to the alluvial coastal plain of Natal.It grows in open moist grassland, coastal forest margins, damp marshland and bushy sand dunes.It seems to favour a south-west aspect and an altitude range of 5-650 m.
There is a variation in robustness and leaf size in this species.The large leaved forms are rare.Stem pubescence, gland colour, size of stipules and number of flowers also vary to some extent, but apart from this the species is fairly constant throughout its range.
In Natal E. parviflorum has a tendency to produce floral aberrations.I have recorded and preserved an unusually large number of cases in which up to three standards were present in the same flowers.The most common aberration found was a variation in the shape of the pistil.Two forms occurred, sometimes in the same plant.The area of thickening may be at the point of flexure or above it.Beyond the point of flexure the style may be erect or strongly recurved.This may constitute an example of cleistogamy.Another common aberration is the lack of, or partial fusion of the stamens into a staminal sheath.Keel blades are often absent.This information is recorded in case difficulties are encountered in keying out the species.The wing is an unreliable character.
An unusual feature is the occurrence of 'sleep movements'.I have been unable to determine whether these are photonastic or thermonastic.They are definitely not thigmonastic.Fig. 24.2 and 24.3 shows the same plant during the day and later during the night.The leaflets expanded during the day can be seen to have closed tightly to the petioles at nightfall.
This species was recorded by Phillips (1917) for Lesotho.He based this citation on Dieterlen 866.A study of this specimen has revealed that it has characters intermediate between those of E. cor datum var.cordatum and E. salignum E. Mey.The possibility of hybridization cannot be ruled out (Stir ton, 1981).Plants such as Dieterlen 866 were pro bably common since Phillips (I.e.) referred to it by its Lesotho name, 'Leshetla la loti', a name different from those he recorded for the putative parents.I am certain that his specimen is not E. parviflorum E. Mey. and that this species does not occur in Lesotho.
E. parviflorum has no close affinities in South Africa.It has often been confused with E. nutans Schinz, but differs from that species in the shape of the standard and the shape of the seeds.
Plants flower from early September to as late as April.

Figs 1
Figs 1 -9 .-1, Eriosema cordatum, daucate rootstock, side branches absent.2, E. psoraleoides, multibranched rootstock.3, basic types o f stipules found in the genus Eriosema: a, side view of free stipules with petiole in plane o f paper, and T /S of stem at point indicated by dotted line showing how stipules flank petiole; b, side view of fused stipules clasping stem with petiole lying in the plane o f the paper, and T /S o f stem showing how stipules are fused and lie opposite the petiole.4, marginal hair line in stipules of Eriosema: 1, free stipules; a, hair line extending to base of stipules, 2, fused stipules show ing a split down the middle; b, end of hair line; c, uneven edge showing direction o f rupture.5a, diagramatic representa tion of an Eriosema calyx showing component parts; b, side view of an Eriosema flower showing the relative position of the three types of lobes: 1, keel lobe; 2, lateral lobes; 3, vexillar lobes.6, diagramatic representation of an Eriosema gynoecium 7, E. cordatum, fruit cut away showing seed attachment to valves.8, E. cordatum, plants showing a range of variation in length of inflorescene (5 plants collected from one population at the Scottsville Race Course, Pietermaritz burg).9, known distribution of the E. cordatum complex in South Africa.

Fig
Fig. 12. -Eriosema cordatum.Pictorialized distribution map showing variation in leaflet shape and size, length of peti ole and length of rachis.a, Galpirt 1031; b, Stirton 1468; c, Stirton 1462; d, Strey 5048; e, Stirton 1620; f, Baijnath 419; g, W^e/rss.n.; h, Huntley 90; i, Schelpe 1755; j, Mogg 15756; k, Scheepers 487.W here two leaflets are shown, the first is the terminal leaflet of a trifoliolate leaf and the second leaflet is one of the laterals o f the same leaf.If single leaves are shown, only unifoliolate leaves occur.The left hand bar refers to length of petiole and the right hand bar refers to length of leaf rachis.Area 1, area in which unifoliolate leaved plants are likely to occur in the Trans vaal; 2, area where erect plants are likely to occur; 3, area where unifoliolate cor date leaved plants grow in Natal.

304THE
ERIOSEMA CORDATU M C O M P L E X .II.TH E ERIOSEMA CO RDATU M AND E. N U TAN S GROUPS F ig. 26.-Huntley 250, large-leaved form of E. parviflorum col lected in the Ngoye Forest Reserve (

-Known distribution of Eriosema zuluense in South Africa.
This species is somewhat variable and additional collecting is needed.It is best exemplified by Acocks 12980, Codd 1885, Codd 2004, Ward 13067 and Ward 1548.E. zuluense is characterized by its F ig .11.
Richards 2362/A (Tanzania); Semsei 1652 (Tanzania).These specimens might deserve some in fraspecific ranking.Verdcourts inclusion of var.buchananii in the E. cordatum E. Mey.complex can be understood from his comments on the Scheepers 814 specimen.He referred to Scheepers 814 as being a 'small rounded shrub of upright habit with numerous ascending to suberect branches from the base, corolla very pale pink, veined with dark red purple'.In differentiating between E. cordatum and E. buchananii in South Africa, he made some rele vant remarks, based as it turned out, on a specimen atypical of its range.He commented: 'This is most interesting because the same collector has collected true E. cordatum at Letaba [Scheepers 487, 3 Oct. 1958 (K; PRE)]; he described the flowers as yellow veined with dull red, the standard dull red outside.The two are strikingly similar in general appearance save that Scheepers 814 has much denser patent in dumentum on the stem.The two are, however, equal ly strikingly different in their standards; not only is there the difference in the appendages already pointed out, but the shape is also quite different in each.E. buchananii has a distinctly rectangular stan dard with the auricles pronounced, above which it is very slightly narrowed; E. cordatum has a distinctly obovate standard, gradually narrowed into the claw, the auricles being less obvious.Moreover the claws of the other petals are much longer in E. buchananii than they are in E. cordatum \ Scheepers (1974 pers.comm.)felt sure his Nos 487 and 814 were different -'certainly they are very distinct in habit'.He commented on E. buchananii's 'more stiffly upright habit' as opposed to E. cordatum's 'characteristically laxly sprawling habit'.
Rhynchosia cajanoides Guill.&Perr.(1833), it is quite possible that the Transkeian E. polystachyium is not conspecific with E. cajanoides, in which case Meyer's name would have to be restored and Richard's name would have to be changed, and with consequent confusion.Propably anticipating this, Bentham proposed in mss, the name Eriosema richardii, which is adopted here'.[N.B.E. polystachyium E. Mey. is a synonym of E. psora leoides (Lam.)G. Don].