Natural hybridization in the genus Eriosema ( Leguminosae ) in South Africa

Although hybridization is today widely recognized as having played a major role in the evolution of the plant kingdom, there is still disagreement (Heiser, 1973) as to how it should be defined. For example, Sibley (1957) defined it as interbreeding between populations in secondary contact, regardless of tax­ onomic rank, whereas Solbrig (1970) differentiated clearly between the crossing of different taxa, calling that between species, or taxa of higher rank, ‘hybridization’, and that within species ‘recombina­ tion’. In this paper the term hybridization is used in Mayr’s (1942) sense of ‘the crossing of individuals belonging to two unlike natural populations that have secondarily come into contact’. Most authors who have written on this subject recognize two main types of hybridization: spon­ taneous hybridization and introgressive hybridiza­ tion. Naturally occurring interspecific hybrids, presumably individuals of the first filial generation (henceforth referred to as the F,), are known in all major groups of plants and in all well studied floras (Grant, 1971). According to this author the F, generation may be fertile, semi-sterile, highly sterile, or completely sterile, and in all but the last case it can produce some later-generation progeny. The partially fertile F, hybrid may reproduce sexually, be selfing, or there may be crossing with sister hybrid plants, or backcrossing to one or both parental species. The resulting second-generation progeny can then go on to cross with one another and with the original plants. This results in a hybrid swarm, an extremely variable mixture of species, hybrids, backcrosses, and later-generation recombination types.


INTRODUCTION
Although hybridization is today widely recognized as having played a major role in the evolution of the plant kingdom, there is still disagreement (Heiser, 1973) as to how it should be defined.For example, Sibley (1957) defined it as interbreeding between populations in secondary contact, regardless of tax onomic rank, whereas Solbrig (1970) differentiated clearly between the crossing of different taxa, calling that between species, or taxa of higher rank, 'hybridization', and that within species 'recombina tion'.In this paper the term hybridization is used in Mayr's (1942) sense of 'the crossing of individuals belonging to two unlike natural populations that have secondarily come into contact'.
Most authors who have written on this subject recognize two main types of hybridization: spon taneous hybridization and introgressive hybridiza tion.Naturally occurring interspecific hybrids, presumably individuals of the first filial generation (henceforth referred to as the F,), are known in all major groups of plants and in all well studied floras (Grant, 1971).According to this author the F, generation may be fertile, semi-sterile, highly sterile, or completely sterile, and in all but the last case it can produce some later-generation progeny.The partially fertile F, hybrid may reproduce sexually, be selfing, or there may be crossing with sister hybrid plants, or backcrossing to one or both parental species.The resulting second-generation progeny can then go on to cross with one another and with the original plants.This results in a hybrid swarm, an extremely variable mixture of species, hybrids, backcrosses, and later-generation recombination types.
In this study the term 'spontaneous hybridization' is used in reference to the production of occasional or sporadic natural hybrids, whereas the term 'intro gressive hybridization' refers to the repeated backcrossing of a natural hybrid to plants of one, or both, parental populations.
Three hybrid situations have been encountered in the present study.These are: (a) spontaneous hybridization between two species, with the hybrid progeny sterile; (b) introgressive hybridization between two spe cies, with the hybrid progeny semi-sterile or fertile; (c) hybrid swarming between four species with the hybrid progeny of variable fertility.Representative examples of spontaneous and intro gressive hybridization have been selected from the hybrid catalogue and are presented in detail.The only hybrid swarm encountered is briefly referred to, as its detailed analysis was considered beyond the scone of this work.The three hybrid situations listed above form the framework on which the hybrid cata logue has been compiled.It is presented in the last section.
The structure of populations of plants derived through hybridization have been evaluated almost exclusively on the basis of morphological criteria (Levin, 1967).Their analysis has been possible by the development of a number of techniques such as Anderson's (1949Anderson's ( , 1953Anderson's ( , 1956Anderson's ( , 1957) ) pictorialized scatter diagrams and hybrid indices and to a lesser ex tent Hatheway's (1962) weighted hybrid index.How ever, as Levin (1967) clearly pointed out, these tech niques, although they provide considerable informa tion about the gross population structure, the direc tion and extent of gene flow, and the correlation and segregation of species characters, they nevertheless may fail to provide a basis of ascertaining the specific nature of each hybrid.As a consequence of domi nance, character cohesion, epistasis, complex mode of inheritance, varying degrees of expressivity and phenotypic plasticity Levin (1967) considers any judgements concerning parentage of individual organisms to be unreliable.Clearly comparative mor phology alone is inadequate in portraying the dimen sions and significance of hybridization: it can however, indicate its physical presence and, from a practical taxonomic viewpoint, can provide an ade quate preliminary tool of analysis.
For over 30 years Anderson's paradigm of charac ter coherence, the basis of his concept of introgres sive hybridization, has been very useful to taxono mists dealing with hybrid populations.There is now, however, sufficient data to suggest that this para digm should be considerably modified (Flake, Rudloff & Turner, 1969;Adams & Turner, 1970;Heiser, 1973;Anderson & Harrison, 1979;Grant, 1979;Wells, 1980).The Andersonian theory has been suc cinctly summarized by Grant (1979) as follows: 'Character coherence is a diagnostic feature of natural hybrid populations.The character coherence is due primarily to multifactorial linkage and secon darily to pleitropy and environmental selection.Selection in nature normally favours parental types and thus works in the same direction as linkage'.What has troubled Grant (1979) and other authors is the limiting scope of Anderson's theory as well as the misleading interpretations that can arise from its faulty techniques.The latter has been clearly shown by Wells (1980).The biggest stumbling block in Andersonian theory is the finding (Grant 1979) that different hybrid populations may have marked dif ferences in character coherence.Grant has suggested, therefore, that character coherence is not a constant, but rather a variable feature of natural hybrid popu lations.Two new techniques are now available to detect this (Grant 1979;Wells, 1980).The nub of Grant's conclusion is that multifactorial linkage and recombination are opposing tendencies and that in hybrid populations they work in balance such that their respective expressions of character coherence and character recombination will reflect a balance or compromise between opposite extremes.He suggests that 'the balance is affected by various internal and external factors.The components of the recombina tion system of the plant group comprise the internal factor (see Grant 1975, Chap.3); the chief external factor is natural selection.The interplay between these factors determines the point of equilibrium be tween character coherence and character recombina tion and so will vary from hybrid population to hybrid population depending on the controlling fac tors involved.
This study has utilized Andersonian techniques as these were all that were available at the time.Unfor tunately the present author has no further opportuni ties to continue this study and presents the results in the hope that someone will reasses the problem in greater depth using chemical and cytological data and a revised Andersonian paradigm.
Eriosema is one of the few papilionoid genera in southern Africa which is known to have undergone hybridization.The reason for this propensity is un known.It is interesting that Rhynchosia, a genus close to and often confused with Eriosema, has not a single recorded case of hybridization.Judging from reported cases hybridization among papilionoid legumes is an apparently rare phenomenon.This may, however, be due to a lack of appreciation by taxonomists of its role in plant variation.The chemotaxonomic studies of Baptisia by Alston and Turner and co-workers remain the classic case of hybridization analysis in the Papilionoideae (Alston, 1959;Alston, Turner, Lesters & Horne, 1962;Alston & Turner, 1963;Alston & Hempel, 1964;Alston, 1965) and a useful model to follow if suitable facilities are available.

SPONTANEOUS HYBRIDIZATION
While collecting specimens on vacant commonage at Hayfields, Pietermaritzburg (Fig. 1), I found a large population of yellow flowered E. salignum E. Mey., growing below a ridge on a gentle slope.South of this was an extensive population of prostrate, bright red, orange and yellow flowered E. cordatum E. Mey.In the adjacent area between the two species and also within the E. cordatum population were about twenty semi-erect, pale pink and yellowflowered plants (Fig. 2), most of which bore withered flowers and fruits.Some of the fruits were fully formed but the enclosed seeds, had shrivelled.The vesture of hairs of these twenty proved to be inter mediate between those borne by plants of E. salig num and E. cordatum that grew in the same general area.I decided to analyse the entire population.
After considering the possible influence of slope, shade and soil type, I laid out four transects that pro vided an adequate sample of the area.I collected all plants of Eriosema that grew within 1 m of these lines.At 50 m intervals, I then made right-angled transects along which I also collected all plants within 1 m distance from the transect.
As fruiting and flowering were almost completed at time of first locating the population, emphasis had to be placed upon the morphological form of vegeta tive characters.A record was made of each plant's habit of growth, the degree of senescence of the stipules and the type and density of stem and leaf ves ture.The length and width of the terminal (upper) leaflet of the mature trifoliolate leaf subtending the first produced inflorescence, and the length of the petiole of the same leaf were also noted.The qualita tive morphological characters are listed in the left hand column of Table 1.Each character was divided into three states.Those states representative of plants of E. salignum were given the value 0; those represen tative of plants of E. cordatum carried the value 2, while intermediates were represented by value 1.One hundred and fifty two plants were scored for these character states.The sum of the index values for all character states of an individual comprises its total index value.Theoretically the total index value for plants of E. salignum was thus 0, that for plants of E. cordatum was 10, while for hybrid plants scores from 1-9 were possible.Plants with similar total index  values were grouped into a frequency distribution (Fig. 3).The 21 intermediate plants all fell into class 5 of the frequency distribution.The 61 plants which had individual total index values of 0 were considered to belong to E. salignum, whereas those plants which had a total index value of 10 were considered to belong to E. cordatum.Only one group of plants was found to be intermediate, namely those falling within class 5. Quantitative parameters were combined with the qualitative character states by constructing a pictorialized scatter diagram (Fig. 4).In this diagram the value of the ratio of leaflet length/leaflet width was plotted against the length of the petiole.

HYBRIDS SALIGNUM A A A A A A A A A A A ▲ A ▲ ▲ ▲ A A A A A A A A A
In Fig. 4 the hybrids (open circles) are seen to occupy an area of the graph intermediate between the main area of variation occupied by each putative par ent.The hybrid is not entirely discrete spatially as a number of specimens of both putative parents fall within its range.
Fig. 4 and Table 1 both indicate the presence of a group of plants with characters that are not those of E. salignum nor those of E. cordatum but more or less intermediate between these.
A resampling of this population on the 22-10-1975 produced a plant (Fig. 5) which, although clearly within the range of the intermediate plants referred to above, showed some features that are worth special mention.The stipules, instead of senescing, persisted and remained green.They also varied from free and overlapping at the base (Fig. 5.5) to fused for two-thirds their length (Fig. 5.6).It would be ad vantageous to monitor an area such as the one des cribed, for hybrids and hybrid variation throughout several years to attempt to obtain information relat ing to maternal and pollen parents of the hybrids that grew and flowered at different times and especially to ascertain whether the nature of the spontaneous hybrids varied over time.
In the Hayfields area the location of the hybrid plants in relation to the populations of both putative parental species suggested that the hybrids arose from seed shed by E. cordatum.Plants of E. cor datum always set abundant seed, whereas those of E. salignum produced a poor, erratic seed set.This level of seed set has been found to be consistent for these taxa throughout their range.Bruchid damage was also heavier in seeds of E. salignum.

INTROGRESSIVE HYBRIDIZATION
Of common occurrence in Eriosema in South Africa are populations in which two species are read ily recognized but in which a range of 'intermediates' are also present.
The example discussed in this section was located near Sobantu village, Pietermaritzburg (Fig. 6).Two species occurred: E. cordatum and E. preptum C. H. Stirton.
The habitat consisted of heterogeneous soil condi tions that ranged from clay in burnt grassveld to deep sand on the lip of a bare roadbank.The latter site formed the limit of distribution.The roadbank was fully exposed to the sun, whereas the grassland had a fair number of plants of Acacia that provided shade.
Plants were analysed as in the example decribed previously.A field assessment of their range of varia tion revealed two characters with marked quantita tive variation, namely the length of the inflorescence and the length of the terminal leaflet of a mature trifoliolate leaf.Measurements were recorded for 136 plants.
The qualitative characters are listed in the left hand column of Table 2.Each character was divided into 3 or 4 states.The numerical value of the character states provides an index value.Total index values were compiled for each plant and then used in the compilation of a frequency distribution (Fig. 7) by the method used previously.The scoring of individ ual plants was such that a typical plant of E. preptum would score 0, one of E. cordatum would score 9, whereas plants with characters of either putative parent would score from 1 to 8. Fig. 7 shows that nearly 50 % of the plants scored fell between the two putative parents.Fig. 8 is typical of the plants falling in group 4 in Fig. 7. From Fig. 7 it appears that there was backcrossing of the interspe cific hybrids with plants of E. preptum.
The quantitative characters were combined with the various qualitative character states and are pre sented in a pictorialized scatter diagram (Fig. 9).It The hybrid plant (half the specimen was photo graphed) shown in Figs 8 & 10 is a robust intermedi ate, and of all the different 'introgressants' is, with others like it, the least likely to be found in the prox imity of either parent.
The hybrids were found scattered throughout the total population but tended to be aggregated in micro-localities not particularly favoured by either parent.
Although the overall population at Sobantu Vil lage has been interpreted as exhibiting signs of introgressive hybridization, it is quite likely that this may be an incorrect explanation as a number of alterna tives are theoretically plausible.Firstly, individuals having characters of two species may represent the remnants of the ancestral population out of which the two species differentiated (Dobzhansky, 1941).This does not appear to be the case here as the intro gressants were mostly sterile and were found in the more recently disturbed areas of the total population.
A second alternative might be that the variation arose through mutation.Heiser (1973) pointed out that this is probably not uncommon and quoted Mayr (1942) as saying that it is known that the basic potentialities of related species tend to be similar and so the mutational channels are therefore more or less prescribed.This, however, also seems unlikely in view of the numbers of 'intermediates' observed.
The third possible explanation is more complex.It involves population intergradation.There are two types of intergradation.Primary intergradation (Mayr, 1942) occurs in populations that are in con tinuous contact, whereas secondary intergradation occurs between populations that have come together after a previous separation.Heiser (1973) pointed out that the former involved recombination and the later hybridization.Secondary intergradation could result in introgression.But as Mayr (1963) also pointed out, these two types of intergradation are not easy to dis tinguish, and because of this, the effects of primary intergradation would be difficult to distinguish from those of introgressive hybridization.Heiser (1973) quoted Barber & Jackson (1957) as stating that 'in a region of great ecological change one can expect simultaneous clinal variation in the frequencies of genes at a number of loci.Variability under the con trol of selection may reach a peak and decay on both sides of this peak'.This would lead to the loose association of variables that Anderson (1953) con sidered diagnostic of introgression.This would then mean that highly variable populations resulting from primary intergradation would be quite similar to those resulting from introgression.Both putative parents are widely distributed in the drier areas of Pietermaritzburg.For the most part they are allopatric but at least three localities are known where they overlap.In all the overlapping localities there is hybridization of the type reported in this section.It seems that these examples conform to Mayr's (1963) statement, that 'belts of highly variable populations in a meeting zone between rather uniform popula tions are almost invariably zones of secondary in tergradation'.Heiser (1973) mentioned two further possible inter pretations of what is known as introgression.The first involved segmental allopolyploidy where segre gates approaching one or both parents may occur.From the available cytological evidence (Baudet, 1977;Turner & Fearing, 1959), it seems that poly ploidy is absent in the subtribe Cajaneae to which Eriosema belongs.It seems unlikely, therefore, that the intermediates of the Sobantu population were segregates produced by a polyploid species.
Heiser's second interpretation concerned the pro duction of introgression-like effects which might result from inbreeding and selection in autogamous plants following an initial hybridization.Heiser (I.e.) stressed that although occasional hybridization bet ween autogamous species did allow for the possibility of introgression, the lack of backcrossing after the initial hybridization would imply that the variation observed, while like introgression, was not the out come of this phenomenon.
Field and laboratory experiments which involved bagging and emasculation of flowers revealed that both putative parents of the Sobantu Village popula tion failed to set seed if the inflorescences were isolated from pollinators.
One real difficulty in interpreting the Sobantu pop ulation is that there is, as Heiser (1973) pointed out, some confusion between the definition of introgres sion and of hybrid swarming.Whether the Sobantu population consisted of hybrid populations out of genetic contact with the putative parents, or whether it consisted of hybrid plants repeatedly backcrossing to the parental forms clearly cannot be solved without detailed genetic experiments.The unlikely possibility also exsists that the introgressants are nothing more than F, hybrids showing a marked degree of environmental plasticity.

HYBRID CATALOGUE
In the hybrid catalogue I have endeavoured to in clude all the putative hybrids that have been found in Eriosema in southern Africa (Fig. 11).This hybrid list will, I hope, encourage further studies within Eriosema.I hope it will also indicate to future workers that hybridization probably contributes materially to the extensive variation represented within Eriosema in southern Africa.
Each pair of putative parents has been named and numbered as representing a case of interspecific hybridization.
Under the numbered putative parents (for example H,), each individual hybrid specimen known has been cited, together with its putative parents where these have been suspected.Each set of citations if followed by a general discussion.Dieterlen 866, a hybrid, was the basis for the incor rect inclusion of Eriosema parviflorum E. Mey. in the flora of Lesotho.Dieterlen 104 (PRE) has five specimens on the sheet but only the robust specimen is a hybrid.Rogers 12199e is included with uncer tainty.
Putative hybrids between E. cordatum and E. salignum (yellow-flowered form) have been found over a wide range yet the plants that are of suspected hybrid progeny are everywhere remarkably uniform.No specimens have been found which might be con sidered backcrosses between the intermediate hybrid and either of the two putative parents.
Hybrids, in general, are not common and thus are likely to be rare in any locality.The Stirton 736 Hayfields population, however, numbered over 20 plants spread over four hectares.Some of these plants had very large rootstocks and were considered to be at least five years old.Field studies in the Hayfields population showed that E. salignum and E. cordatum usually had very little overlap in the time and duration of anthesis but on cloudy days both species showed fully reflexed flowers (i.e.flowers ready for pollination with the standard folded back) throughout the day.Sporadic hybrids were found to occur in the locality of this population and may well have been the result of such an environmental influence.It is possible that isolating mechanisms other than time of flower reflexion may be in general operation to maintain the integrity of the two parental species.This needs fur ther study.
Field studies during the present work have sug gested that plants of E. cordatum and E. salignum hybridize more frequently than any other known cases of hybridizing species in the genus in southern Africa.
E. cordatum E. Mey.x E. salignum E. Mey.(yellow-flowered form) can be separated from its putative parents by the characters listed below.The hybrid situation on Mr I. G arland's farm at M tunzini was particularly interesting.The hybrids were numerous and occurred wherever the putative parents were sympatric.There was a very large hybrid population at the bottom o f the front lawn of his house.The hybrids were intermediate in nearly all characters, the most striking being the intermediate root system (Fig. 12).In this figure, the hybrid plant lies between the two putative parents: it can be seen to have features both of the straight daucate The Stirton 1409-1411 population was analysed under Introgressive Hybridization.In the Sobantu Village and Scottsville populations there was marked hybrid vigour and an intermediate type o f rooting system.The main differences between the hybrids and the putative parents are summarized below'.It must be stressed, however, that in dealing with an introgressed population the choice o f characters are to some extent arbitrary.Overlap must therefore be ex pected.The facies o f the Stirton 1442-1445 population is similar in many ways to that described for the Stirton 1409-1411 population o f E. cordatum E. Mey.x E. preptum C. H. Stirton.In both instances the habitat had been subjected in the past to burning.The soil cover and type was also rather variable.Stirton 1442a (sheet 1) showed the greatest hybrid vigour.This population may be backcrossing to either parent as Stirton 1442a (sheet 2) and Stirton 1442b were in termediates between Stirton 1442a (sheet 1) and E. transvaalense (Stirton 1445) and E. angustifolium (Stirton 1444) respectively.The hybrids were spread randomly throughout the population and were readi ly distinguishable on habit and flower colour.E. transvaalense has pink and yellow flowers and is a prostrate, softly pubescent densely matted plant whereas E. angustifolium is an erect, stiffly haired yellow-flowered species.The hybrids tended to have pale pink flowers and to vary from prostrate, ascen ding to erect in habit.

E.
transvaalense and E. angustifolium are sympatric species but this small population is the only one in which I have found hybridization between the two species.This population was spread over a disturbed, re cently-felled Eucalyptus plantation.Stirton 1259 had yellow flowers and was semi-erect, whereas Stirton 1260 was pale pink-flowered, numerous stemmed and prostrate.These two collections formed the limits of the range of variation.There was a wide range of 'intermediates' between the four 'putative' parents.I attempted to analyse this population using the techniques described earlier but could not make any sense of the results.As there was no time avail able to pursue the problem it is reported here in case difficulties may be encountered in identifying plants collected in the Gingindlovu area.

Fig. 11 .
Fig. 11.-Known distribution of hybrid populations of Eriosema in southern Africa.
rootstock o f E. cordatum and of the constricted rootstock of E. salignum.(Intermediate rooting systems are very characteristic o f hybrids in the genus as a whole.)There was a marked hybrid vigour in this Mtunzini population.The hybrids were tall, suberect or ascending and stood out against other plants of the population.Mey.x E. preptum H3 Eriosema cordatum E. C. H. Stirton So u t h A f r i c a .-2930(Pietermaritzburg): Hilton College farm lands (-CB), Khan s.n.(NU 45857); behind S.P .C .A .kennels, Scottsville, Pietermaritzburg (-CB), Stirton 713 (PRE , this is a mixed collection containing both putative parents) and Stirton 1243 (PRE), putative parents Stirton 1241 1242 (PRE); near Sobantu Village (-CB), Stirton 1411 (PRE), putative parents Stir ton 1409 & 1410 (PRE).

A
CK N O W LED G EM EN T S The bulk of this study was undertaken in 1974 as part of a Masters thesis at the University of Natal, Pietermaritzburg. 1 am grateful to the former Head of Department of Botany, Professor C. H. Bornman, and his staff for facilities and assistance; to the Secretary of the Department of Agricultural Techni cal Services and the Director, Botanical Research In stitute for one year's study leave; to L. Cowan for Figs 5 & 8 and for assistance in the field, and finally to Professor K. D. Gordon-Gray for her enthusiastic supervision and inspiring introduction to biosystematic botany.UITTREKSEL Beide spontane en introgressiewe verbastering kom in die natuur in Eriosema in Suid-Afrika voor.Daar word verslag gegee van een geval van hibriede-geswerm en 'n lys van ses hibriede word aangebied en tot in besonderhede bespreek.

cordatum, near Sobantu Village, Pietermaritzburg (sampled 1974-11-20); the terminal leaflet of the leaf subtending the second produced in florescence was measured. 10, representative specimens of the Eriosema population near Sobantu Village, Pietermaritz burg
: 1, E. preptum; 2, part of hybrid intermediate showing hybrid vigour; 3, E. cordatum.312 STUDIES IN TH E LEG U M IN OSA E -PA PILIO N O ID EA E O F

SOU TH ERN A FRICA can
be seen that plants of E. cordatum mostly had longer terminal leaflets than did plants of E. preptum.The tendency of the hybrids to produce longer inflorescences than those of either putative parent is also apparent.What is not clearly shown is the hybrid vigour of the putative hybrids, but this may be seen in Fig10.

at connate at base, to base to 3 /» total -Vatotal length length Under surface of leaf-Under surface of Under surface of lets woolly with longer leaflets dull with leaflets shiny with appressed hairs on serni-erect hairs on sparsely scattered veins veins and in inter-erect hairs costal areas Stem pubescence dense, Stem pubescence Stem pubescence appressed, finely dense, shaggy and shaggy ferruginous hairy appressed. or white, patent. Flowers yellow Flowers orange (or Flowers red and pale pink) and yellow yellow
This evidence suggests that there may be two distinct taxa presently included in E. salignum and this pro bability is still being investigated.Case H2 hybrids are more erect with very dark green shiny upper leaf surfaces and have ovate leaflets with the bases cor date.The leaflets o f Case H2 hybrids are more com monly unifoliolate than in the Case H I hybrids.