Annals of Botany 89: 67-76, 2002
© 2002 Annals of Botany Company
Functional Heterostyly in Tylosema esculentum (Caesalpinioideae)
1Department of Biological Sciences, Queen Mary, University of London, 2Thusano Lefatsheng Gaborone, Botswana and 3School of Biological Sciences, University of Sussex, Brighton
* For correspondence. Fax +44 (0)20 8983 0973, e-mail m.l.hartley{at}qmul.ac.uk
Received: 11 April 2001; Returned for revision: 16 May 2001; Accepted: 14 September 2001.
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
Tylosema esculentum is a long-lived perennial species endemic to arid areas of southern Africa. Its potential as a crop species has long been recognized as a result of the high oil and protein content of its seeds. The reproductive biology and breeding systems of the species were investigated in wild and experimental populations growing in Botswana. Field observations confirmed that the species is heterostylous with the pistil and anthers exhibiting reciprocal heights in the two morphs, although pollen size and sculpturing do not vary. The wet, non-papillate stigma characteristic of the species is the first to be reported in the Caesalpinioideae. In vivo and in vitro diallel crossing experiments demonstrated that a diallelic self-incompatability system exists in T. esculentum. The major site of pollen tube inhibition in the intramorph crosses was found to be in the style. This is the first report of functional heterostyly in the Fabaceae and of a confirmed self-incompatibility system in the Caesalpinioideae. Three separate lines of evidence, the monitoring of fruit development in open-pollinated plants, fruit set in diallel crossing experiments, and observations made in wild populations, demonstrated that fruit set and, by implication, seed set, are very low in this species. Floral abscission was a major limitation to the production of mature pods but there were also significant losses at other developmental stages of fruit production. The results suggest that low seed set may be an adaptation of the species to an environment in which rainfall is scarce.
Key words: Tylosema esculentum, marama bean, morama bean, Caesalpinieae, heterostyly, seed set, self-incompatability, ovule, fruit production, pollination.
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
As a first step toward the domestication and selection of high yielding genotypes in a possible crop species, we have investigated the breeding system and reproductive biology in Tylosema esculentum, a long-lived perennial legume species adapted to the arid zones of southern Africa (Fig. 1). Without an understanding of the breeding system to facilitate crossing, genetically sustainable improvement would be difficult to achieve. Where the potential product is seed-based, the need to determine the factors affecting seed quality and production is particularly important. T. esculentum is endemic to Botswana, Namibia and South Africa. The species, a member of the Caesalpinioideae subfamily of the Fabaceae, produces a nutritious and palatable food, the marama bean. The bean which compares well in protein and oil content with both soya and peanut (Ketshajwang et al., 1998) is collected and eaten by local people.
|
Little work has been done on the breeding systems and reproductive biology of the sub-family Caesalpinioideae. Only 47 of the 152 genera have been investigated (Owens and Stirton, 1989). Only in the tribes Caesalpinieae and Cassieae have sufficient species been examined to provide general conclusions (Owens, 1989). It has been suggested that six of the eight genera investigated—including Bauhinia, closely related to Tylosema—have gametophytic self-incompatability (SI) (Arroyo, 1981). Tylosema, in the tribe Cercideae, has been little investigated but heterostyly, unknown elsewhere in the Fabaceae, has been reported in all four species of the genus (Coetzer and Ross, 1976). The primary objective of this research was to determine the type of breeding system operating in this species through in vivo and in vitro crossing experiments and to examine the factors influencing seed set.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
Source of materials
All breeding experiments and the majority of observations were carried out at the Thusano Lefatsheng research station, 35 km west of Gaborone, Botswana (25°37'E, 24°41'S). Approximately 60 6-year-old plants raised from seed collected in Kweneng District and grown in an experimental plot at the station were used in the study. Collection of seed and leaf samples from natural populations was carried out at 34 widely distributed sites in Botswana. Twenty-two of the sites were located in the districts of Kweneng and Ghanzi, the areas of greatest density (a list of sites with GPS readings is available from the authors). In Botswana, T. esculentum is generally confined to unstructured sandy soils lacking a humus layer, which are never covered by standing water (Cole and Brown, 1976).
Determination of style and anther heights
Measurements of the stamens, style and ovary were taken to ascertain whether there were consistent differences between plants. Three flowers from each of four long style (LS) and four short style (SS) plants grown at the research station were sampled. The frequency of the two morphs in three natural populations (sites 15–17 in the Kweneng District) was also estimated by examining the styles of a total of 373 plants. Individual plants from each site were sampled at 5 m intervals.
Pollen and stigma/style morphology
Samples from four flowers of each of the two morphs were prepared for examination under the scanning electron microscope (SEM). SEM facilities were not available in Botswana so samples were fixed in the field in 3 : 1 ethanol/acetic acid and stored in 70 % ethanol. At Sussex, UK, fixed samples of pollen and gynoecia were placed in 1 % osmium tetroxide for 2 h at room temperature, then rinsed in distilled water and dehydrated in an ethanol series. After critical point drying and sputter-coating with gold, they were examined in a Leo S420 SEM at 20 kV. Digital images were printed on a Codonics NP 1600 network dye-sublimation printer.
Floral development and fruit production in open-pollinated plants
A total of 360 flowers from 45 immature inflorescences from each of nine plants (five SS and four LS) grown at the research station were tagged and monitored throughout the flowering and seed production stages during a 3 month period from December to February. The production of flowers and the duration of flowering were observed. Development of the tagged inflorescences was examined daily until the flowers had withered. After pod formation, observations were made once a week. The levels of floral abscission, fruit set and fruit abortion were recorded 10 and 40 d after pollination. Fruit set was defined as the swelling of the ovary to form a small pod. Fruit abortion was recorded when an immature pod abscised.
In vivo self pollinations
During the previous year, 153 immature flowers from five LS and five SS plants were isolated in cellophane bags and monitored for seed set. None of the flowers set any seed. To confirm the apparent lack of self pollination, a further selfing experiment was set up the next year to run concurrently with the intermorph cross-pollination experiments (see below). At least ten flower buds from each of another ten plants (five LS and five SS) were isolated in bags and tapped each day to encourage pollen transfer. At 10 and 40 d after pollination the flower heads were examined for seed set.
In vivo cross pollinations
Individual flowers were isolated in bags to protect them from open pollination. The complete lack of seed set in selfing experiments suggested that it was unnecessary to emasculate flowers for cross pollination. When a bagged flower was fully open and the anthers had dehisced, anther mucilage covered the pollen. Manual pollinations were then carried out. Freshly liberated pollen and mucilage were transferred from the designated male parent to the stigma of the female parent using fine forceps. Pollinations were carried out in the morning or early afternoon because the mucilage was often dry after this time. Four x four diallel crosses were carried out: intermorph crosses (long x short and the reciprocal) and intramorph crosses (long x long, short x short and their reciprocals). For each intermorph cross, a minimum of 12 flower buds per plant was bagged and manually pollinated the following day using pollen donated from each of four genotypes. For intramorph crosses, six to eight flowers from each of four genotypes were pollinated. Fruit set was assessed 10 and 40 d after pollination.
In vitro self and cross pollinations
Unpollinated gynoecia were removed from fully open young bagged flowers and inserted into agar in a Petri dish. Stigmas were pollinated manually with fresh pollen and the plates were covered and left on the bench at 31–37 °C during the day but were maintained at about 12 °C overnight as the ambient temperature tended to melt the agar. They were moved back onto the bench in the morning. Twenty-four to thirty hours after pollination, the gynoecia were fixed in 3 : 1 ethanol/acetic acid for 24 h before being stored at 4 °C. They were stained with aniline blue and were viewed under a Zeiss Axiophot microscope and photographed using EPL400X film under a 363 UV filter.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
Style and stamen heights
Examination of a total of 373 plants from three populations in the Kweneng District was carried out to determine the ratio of SS to LS plants. In all plants examined, the style and anther were at different levels. The three field populations of T. esculentum studied, site 15 (29 LS and 39 SS plants), site 16 (96 LS and 110 SS) and site17 (51 LS and 48 SS plants) appeared to contain approx. equal proportions of LS and SS morphs. Chi-squared tests on the populations confirmed the hypothesis that there was a 50 : 50 ratio of LS and SS flowers (P = 0·22).
The flowers contained a single style, two stamens and eight staminodes surrounding the style (Fig. 2). There were two distinct morphs: pin morphs with a LS and short stamens (Fig. 3) and thrum morphs with an SS and long stamens (Fig. 4). Measurements revealed that LS plants had pistils that were approx. 40 % longer than pistils in SS plants (Table 1). Both the style and the region adjoining the base of the ovary and the pedicel contributed to the difference in pistil length. Stamens in the LS plants were less than half the length of those found in the SS plants. There was no overlap in either the pistil or stamen length between the two morphs. However, the length of anthers and ovaries was very similar in both morphs. The consistency of these results was borne out by examination of many other plants at the research station and in the field.
|
|
|
|
Cell size and structure in stylar epidermis and pollen
Stigmatic papillae were not found to be present in either morph (Fig. 5A and B). Cell size, however, was found to vary significantly between the two morphs (Fig. 5C and D). LS stylar epidermal cells were almost twice the length of SS stylar cells (Table 1) while cell widths were similar. These data suggest that the differences in style length resulted from cell elongation rather than an increase in cell number.
|
Pollen size and sculpturing
Variation in pollen size is frequently found in distylous species where long styles are associated with large pollen grains and short styles with small pollen grains. As fresh or fixed pollen was not available at the appropriate stage of the project, measurement of pollen grains was made using the SEM micrographs. These did not, however, reveal differences in pollen size (Table 1) or morphology between morphs (Fig. 5E and F). The pollen was released as single grains.
Floral development
At the research station, the period of flowering of T. esculentum generally coincided with the rainy season which occurs between October and February (Mbewe, 1992). The peak of the flowering period varies considerably from year to year occurring, for example, in mid-February (1989), November (1990) and December (1994). Monitoring of floral development revealed that inflorescences comprised a mean of eight flowers (range 2–12) arranged in a cluster. An average of two to three flowers opened per day over a period of up to a week, by which time most of the flowers in an inflorescence had opened. Between 0 and 6·6 % buds failed to open (Table 2). As flowers withered within 3–4 d of opening, not all the flowers of an inflorescence were open at any time. The anthers dehisced longitudinally releasing pollen (Fig. 5G and H). On warm, sunny days at the research station, anthesis had occurred by the time the first buds opened between 0700 and 0800 h. On cool cloudy days, anthesis did not occur until midday or did not occur at all. Within 1 h of anthesis, pollen was covered by a thick exudate as reported by De Frey et al. (1992).
|
Breeding behaviour
Fruit set was used as an indicator of compatibility in all crosses. In T. esculentum, the ovary normally contains two ovules (Fig. 6), although four and even six ovules were found occasionally. Mature pods usually contain two seeds but occasionally only one was present. In the latter instance, there was often an indication that a second seed had aborted. The lack of selfing ability in T. esculentum was confirmed. No seed was set in any of the 100 self pollinations.
|
In the intramorph 4 x 4 diallel crosses (LS x LS and SS x SS) a total of 163 flowers from four genotypes were pollinated manually. Of these, 17–20 % abscised within 2 d and the development of the remainder was monitored for a further 30 d (Table 3). Fruit set was extremely low. In the LS x LS crosses, 98·7 % of the pollinated flowers failed to set any pods. Only in one of these pollinations did a single pod develop. In the SS x SS crosses, 92 % of pollinations failed to set any pods. Of the nine successful crosses in the latter, seven involved one individual plant (420), which set fruit in three of four SS x SS crosses.
|
In the intermorph 4 x 4 diallel crosses (LS x SS and SS x LS) a total of 281 flowers from four genotypes were pollinated manually. Between 23·5 and 30 % of the pollinated flowers abscised within 2 d. However, 10 d after pollination, 27·2 % of LS x SS crosses had set fruit in 14 of the 16 cross combinations, 2·5 times the amount of fruit set in the SS x SS crosses and 22 times that of the LS x LS crosses. In the reciprocal cross SS x LS, fruit set only occurred in 7·8 % of the pollinations and only in six of the 16 cross combinations. The number of pods surviving in compatible crosses diminished significantly between 10 and 40 d after pollination, particularly in the LS x SS crosses. A
2 test between within morph and between morph crosses showed a difference in fruit set at 10 d (2 = 8·9, P = 0·03 and) but not at 40 d (2 = 0·2, P = 0·645). It was noted that seasonal rainfall had stopped within 2 weeks of completion of the pollination experiments and this may have contributed to fruit abortion. The fact that fruit set was some 2·5 to 22·6 times higher in intermorph pollinations compared with intramorph pollinations provides significant evidence of self incompatibility. In vitro experiments were therefore carried out to investigate the site of inhibition in these pollinations.
In vitro pollinations
These pollinations were based on four LS and four SS genotypes in 4 x 4 diallel combinations. In these in vitro experiments the pollen germination rate was high in both the cross and self pollinations. The intramorph SS x SS data with and without plant 420 are presented for comparison. Of a total of 79 pollinations examined from the four crosses, only six showed an absence of pollen tube growth (Table 4). In the intramorph crosses, pollen tube penetration of the stigma occurred in all of the remaining pollinations but was strongly inhibited in the style and ovary. In LS x LS crosses, 83 % of pollinations were inhibited in the style (Table 4). In only one pollination did a pollen tube appear to enter the ovary, but not the ovule. In the other intramorph cross (SS x SS), approx. 45 % of pollinations showed pollen tube inhibition in the style and about one-third in the ovary. Penetration of the ovule also appeared to occur in one SS x SS pollination.
|
In both of the intermorph crosses, more than half of the pollinations resulted in pollen tubes growing successfully through the style into the ovary. However, only 8–18 % of these penetrated an ovule (Table 4). Although the number of pollinations with pollen tubes penetrating the ovary was considerably higher relative to the intramorph pollinations, it was nevertheless low overall and may go some way to explaining the low seed set recorded in field experiments. However, there are a number of potential biases in the data above. It is often very difficult to trace pollen tubes as far as the ovule. Nor is it certain that the gynoecia remained in the growth medium long enough for pollen tube growth to be completed. In addition, it is possible that the excision of the gynoecia might have had an inhibitory effect on tube growth. These biases may have led to an overall underestimation of pollen tube growth.
When data for pollen tube growth in each of the two crosses in the intramorph and intermorph pollinations are aggregated two trends are clear (Table 4). First, pollen tube inhibition in the intramorph crosses occurred mainly in the style with approx. two-thirds of pollinations being halted here compared with only one-third of intermorph crosses. Secondly, the major site of pollen tube growth inhibition for intermorph crosses was the ovary where tube growth was arrested in nearly half the pollinations. In the intermorph crosses, almost five times more pollen tubes reached the ovule than in the intramorph crosses. It is worth noting that plant 420 accounted for four out of six SS x SS pollinations in which pollen tubes entered the ovary (see Table 2). Crucially, with or without the results from this exceptional plant, ovule penetration appeared to be a very rare event in intramorph pollinations, occurring only once in an SS x SS cross. This genotype, which could be a self-compatible recombinant, also produced the exceptional results in the in vivo crosses. If this genotype is excluded from the aggregated results, the difference between inter- and intramorph pollinations is even more apparent (Table 5).
|
The data support the hypothesis that pollen tube growth ceases earlier in intramorph crosses than in intermorph crosses (P < 0·002; non-parametric two-tailed Mann–Whitney U test). When plant 420 is excluded from the intramorph crosses and the Mann–Whitney U test is applied again the result is even more significant (P < 0·0001). Comparison of the results of the intramorph crosses with and without plant 420 did not show any significant differences (P > 0·42). A
2 test supported this result: 2 = 14·1 (3 d.f.) (P = 0·003).
Fruit production
In general, fruit production was found to be extremely low although individual plants showed considerable variation in their fecundity. It was not possible to measure seed production within the time frame of the field experiments. In the nine plants observed, between 64·2 and 100 % of the flowers monitored abscised within 7 d of opening (Table 2). In two LS and two SS plants (numbers 335, 413, 414 and 301), all 125 of the flowers monitored failed to set fruit. In three other plants (numbers 403, 353 and 200), abscission of fruits occurred predominantly before the pods had reached 3 cm. In one SS plant (417), the very high initial fruit set (68 %) fell to only 2·6 % by the time pods had were full size (4 cm). Early fruit set exceeded 20 % in only two of the nine plants studied, an SS plant (417) and an LS plant (403). Final fruit production exceeded 10 % in only one plant (SS, 415). In this plant, three of the five inflorescences monitored produced mature pods, 17·6 % of the flowers set seed and 14·7 % produced mature pods. This was the only individual in which fruit abortion was not significant. In general, floral abscission was the major limitation to the production of mature pods, but there were significant losses on at least one plant at each stage.
The in vivo diallel crosses also featured high levels of fruit abortion. In the intermorph crosses between LS x SS plants, 67 % of young fruits that had formed by 10 d after pollination had abscised when they were examined 30 d later. Final fruit production was therefore only 8·8 % of total pollinations (Table 3). In the reciprocal cross, 55 % of young fruits had aborted 30 d later; thus, final fruit production was only 3 %. The possibility that selection occurred for high quality embryos or to eliminate recessive lethals (Wiens et al., 1987; Bawa and Buckley, 1989) could not be assessed within the time scale of the experiment. Nor was it possible to determine whether the high level of abscission was a result of resource limitation (there had been little rain over the previous month). However, in experiments in Texas, Powell (1987) did not find late fruit abortion to be a factor in the levels of fruit set in watered plants. This would suggest that lack of water rather than genetic factors was involved in the late abortion found in these experiments.
The level of fruit production was also recorded for plants sampled from 17 wild populations drawn from sites widely distributed across the species’ range in Botswana. Up to 30 plants were sampled in each population. The proportion of plants with one or more fruits ranged from 3 to 64 %, with an overall population mean of 19 %. The figure for fruit set may have been somewhat biased relative to the whole population as sampling did tend to concentrate on healthy, mature plants so plants with pods were probably over-represented in the samples. As most of the pods were less than 3 cm long, i.e. not yet fully mature, the possibility that local people had already harvested a significant number of pods did not arise.
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
T. esculentum has the morphology, life history and breeding behaviour characteristic of an outcrossing species exhibiting heterostyly allied to diallelic incompatibility. It is an insect-pollinated perennial species. There are two floral morphs with stamens and style at reciprocal heights. In diallelic species, the evolutionarily stable strategy (ESS) is based on an equal number of long style and short style (thrum) plants when the morphology is related to the breeding system (Maynard-Smith, 1978). The equal ratio of pin and thrum plants found in T. esculentum supports the hypothesis that heterostyly is associated with the breeding system in this species.
Although the flowers are not trumpet-, bell- or tube-shaped as commonly found in heterostylous species, the arrangement of the eight staminodes and two anthers surrounding the style helps direct the potential pollinator to the correct position for legitimate pollen transfer. The staminodes may also limit the deposition of illegitimate pollen onto thrum styles. It would be expected that both the pin and thrum flowers would invest the same amount of resources in reproduction. The fact that the number of pollen grains per anther does not vary significantly between pin and thrum flowers (De Frey et al., 1992) is consistent with the finding reported here that—unusually for a heterostylous species—pollen grains were the same size in each floral morph.
As yet there have been no fully confirmed reports of self incompatibility in the Caesalpiniodeae (Owens, 1989). However, while breeding experiments suggest that most species in the Caesalpiniodeae are self-compatible, it has been suggested that gametophytic self incompatibility is established in six of eight genera distributed over four of the recognized tribes in this family (Arroyo, 1981). This breeding system is correlated with wet stigmas and binucleate pollen (Heslop-Harrison, 1981) which are both present in the Caesalpinioideae. However, gametophytic self incompatibility cannot be diallelic and, if the heterostyly of T. esculentum is functional, a sporophytic self-incompatibility system would be predicted. No suggestion of sporophytic incompatibility in Caesalpiniodeae has previously been made although heterostyly, unknown elsewhere in the Fabaceae, is now confirmed in all four species of the Tylosema genus. The virtual absence of seed set following self pollination, together with pollen tube inhibition in intramorph crosses indicate that T. esculentum is self incompatible, although it is possible that strong dissortative mating occurs.
SEM analysis showed that T. esculentum does not have the characters expected of a member of the tribe Cercideae, sub-family Caesalpinioideae. Owens and Lewis (1996) found that all members of this tribe they investigated had wet papillate (WP) stigmas. However, there was no evidence for a papillate cell layer in T. esculentum. As would be predicted for wet non-papillate (WN) stigmas, the Tylosema stigmas seem unable to discriminate compatible from incompatible pollen.
Seed production is an important component of ‘fitness’, the ability of an individual to produce viable offspring. In T. esculentum, which may have potential as a commercial seed crop, this is of prime interest. Three separate lines of evidence: the monitoring of fruit development; fruit set in diallel crossing experiments; and data from wild populations all demonstrate that fruit set and, by implication, seed set, are very low in T. esculentum. For a flower to be successfully cross-pollinated, it must provide enough reward to attract a pollinator but not so much that the insect does not then move on to the flowers of another plant (Heinrich and Raven, 1972). In T. esculentum, an individual plant can spread over an area as large as 16 m2 and therefore pollination is often likely to be from the nearest plant (Levin, 1978). It has been shown that pollen of T. esculentum remains viable under laboratory conditions for up to 7 d (Monoghan, 1993). However, in the field experiments carried out in Botswana we found that the anther mucilage dried up by the end of the first day. This will probably reduce the likelihood of pollen adhering to potential pollinators and suggests that the time available for pollen transfer is likely to be short.
A pollinator survey revealed that insect activity was very low despite observations being carried out at the height of the flowering season. Previous studies have suggested that pollinator specialization is most developed in arid regions (Heinrich and Raven, 1972), but no single species predominated in this study. Although Mbewe (1992) suggested that ants may be the main pollinators, it is actually more likely that they are nectar robbers of T. esculentum rather than pollinators (Beattie et al., 1984). The lack of pollinators may explain the abundance of flowers in this species as it has been shown that where pollinator activity is uncertain, species often have an abundant display of flowers. The surplus flowers may act as a plentiful reward to attract the scarce pollinators (Stephenson, 1981). However, the large display of flowers observed in T. esculentum is unusual in heteromorphic species as this character is not usually associated with abundant displays (Darwin, 1877).
The prevalence of low fruit set is particularly important with regard to the possible cultivation of T. esculentum as the potential crop would be seed-based. The evolutionary pressures on a long-lived perennial in an arid environment would not consistently encourage high seed set. Low seed set and seed abortion in poor years may allow the plant to invest its resources in survival rather than in reproduction. In a year with ample, well-spaced rainfall, the plant may invest more of its resources in reproduction. In an environment where the water supply is very erratic, the ability to initiate seed production but abort at any stage in the process if water becomes limiting allows the plant to alter its reproductive strategy at a late stage. In 1993, pod maturation in the wild was very delayed in some areas. In the Ghanzi area in north-west Botswana where some late rainfall occurred, seed set was high but in the Kweneng area in south-central Botswana where there was no rain after flowering, very few pods reached maturity (M. Mbewe, pers. comm.). In the inflorescences monitored in the experiments reported here, fruit abortion occurred at all stages of development. As the pollinations were carried out over a period of 2·5 weeks as flowers opened, plants may have been affected at different stages by the lack of rain. Experiments in Texas have shown that when the water supply is adequate, seed set in this species may increase substantially as a result of a reduction in fruit abortion (Powell, 1987). The effect of lack of water may have masked the working of genuine diallelic incompatibility. In a country such as Botswana which has little rainfall and poor access to water for irrigation, selection for genotypes that produce the maximum number of seeds for the minimum input of water is essential if T. esculentum is to become a successful domesticated crop. Clearly, floral production is not the factor limiting seed set.
The cause of the low fruit set in thrum flowers when cross pollinated with pin pollen, compared with the reciprocal cross, is difficult to determine. The thrum stigmas are obviously less accessible for pollination. In T. esculentum, the style is surrounded by eight staminodes so it is difficult to place pollen on the thrum stigma without disturbing and possibly damaging the flower. In the field, the inaccessibility of the thrum stigma may be a factor as the number of pollinators is low.
The incompatibility reaction in T. esculentum does not occur at the stigmatic surface. The exact location of inhibition is not clear. It may be that the in vitro technique affected the later stages of pollen tube growth and so biased the results. However, the results of the in vitro experiments were not incompatible with the other experiments. The other possible source of bias may be the length of time the tubes were allowed to grow. Minimum growth rates have been found to be 1·75 mm h–1 (Richards, 1997), which would mean that even 24 h (the shortest amount of time allowed for growth) would have been ample for growth of even the LSs which were 14 mm at most. Owens (1990) suggests that the WN stigma surface lacks an epidermis, possibly lost from a once-WP type of stigma during evolution. With the loss of the epidermis, it appears that the stigmatic surface has lost its ability to discriminate between legitimate and illegitimate pollen. Rather the incompatibility reaction must occur in the lower style or in the ovary. This mechanism is usually found not in sporophytic systems but in the one-locus gametophytic systems (Richards and Mitchell, 1990). However, this result does correlate with the findings of Lewis and Jones (1992) who concluded that two different recognition systems exist in heterostylous species, one analogous to gametophytic multi-allelic SI and the other analogous to multi-allelic sporophytic SI. The lack of penetration of the ovule by the pollen tubes suggests that the low level of seed set is not an effect of an early acting, post-zygotic mechanism.
The unexpected high level of successful intramorph pollinations in SS plant 420 requires some explanation. If, as in most heterostylous species, the pin flowers have the genotype SS and the thrum flowers Ss, only in thrum flowers will recombination in the SI locus have an effect. Recombination may be the explanation for the high number of successful SS x SS fertilizations of flowers in plant 420. This would explain the number of recombinants found in this heterostylous species, as it is expected that diallelic incompatibility allied to heterostyly will have closely linked loci determining the various aspects.
Although some characters within T. esculentum do not conform to classic heterostyly, the results described here suggest that there may be a sporophytic incompatibility system associated with distyly ensuring out-crossing within T. esculentum. Although late-acting abortion must be considered as a possible contributory factor for the results presented here, the experiments in Texas (Powell, 1987) suggest that, in well-watered plants, late-acting abortion is not a significant factor. Dissortative mating must also be considered. However, the shape of the flowers is not as restrictive as seen in classic heterostyly and is unlikely to be completely effective in restricting illegitimate pollinations. This is the first report of functional heterostyly in the Fabaceae. It is also the first report of a self-incompatibility system in the Caesalpinioideae.
|
ACKNOWLEDGEMENTS |
|---|
We thank the Overseas Development Administration for funding this research. We thank also the Mercers’ Livery Fund, London, for support. We would particularly like to thank all those at Thusano Lefatsheng who contributed to this project.
|
LITERATURE CITED |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
-
Affre L, Thompson JD, Debussche M. 1995. The reproductive biology of the Mediterranean endemic Cyclamen balearicum Wilk. (Primulaceae), Botanical Journal of the Linnean Society 118: 309–330.[CrossRef]
Arroyo MTK. 1981. Breeding systems and pollination biology in the Leguminosae. In: Polhill RM, Raven PH, eds. Advances in legume systematics, part 2. Kew: Royal Botanic Garden.
Bawa KS, Buckley DP. 1989. Seed–ovule ratios, selective seed abortion and mating system in leguminosae. In: Stirton CH, Zarucchi JL, eds. Advances in legume biology. Monographs in Systematic Botany from the Missouri Botanical Garden, St Louis.
Bawa KS, Webb CJ. 1984. Flower, fruit and seed abortion in tropical fruit trees: implications for the evolution of paternal and maternal reproductive patterns. American Journal of Botany 71: 736–751.
Beattie AJ, Turnbull C, Knox RB, Williams EG. 1984. Ant inhibition of pollen function: a possible reason why ant pollination is rare. American Journal of Botany 71: 421–426.[CrossRef]
5 Botswana Technology Centre. 1991. Final report on morama beans processing. Commissioned by Thusano Lefatsheng, Gabarone.
Boyes DC, Nasrallah J. 1993. Physical linkage of the SLG and SRK genes at the self-incompatibility locus of Brassica oleracea. Molecular and General Genetics 236: 369–373.
Casselman AL, Vrebalov J, Conner JA, Singhal A, Giovannoni J, Nasrallah ME, Nasrallah JB. 2000. Determining the physical limits of the Brassica S locus by recombinational analysis. Plant Cell 12: 23–33.
Coetzer LA, Ross JH. 1976. Tylosema. Trees in South Africa 28: 77–80.
Cole MM, Brown RC. 1976. The vegetation of the Ghanzi area of western Botswana. Journal of Biogeography 3: 169–196.[CrossRef]
Darwin C. 1877. The different forms of flowers on plants of the same species. London: John Murray.
De Frey HM, Coetzer LA, Robberse PJ. 1992. A unique anther-mucilage in the pollination biology of Tylosema esculentum. Sexual Plant Reproduction 5: 298–303.
Engels S. 1984. Morama. The Bulletin of Agricultural Research in Botswana 2: 93–97.
Heinrich B, Raven PH. 1972. Energetics and pollination ecology. Science 176: 597–602.
Heslop-Harrison Y. 1981. Stigma characteristics and angiosperm taxonomy. Nordic Journal of Botany 1: 401–420.
Keegan AB, van Staden J. 1981. Marama bean, Tylosema esculentum, a plant worthy of cultivation. South African Journal of Science 77: 387.
Ketshajwang KK, Holmback J, Yeboah SO. 1998. Quality and compositional studies of some edible Leguminosae seed oils in Botswana. Journal of the American Chemists’ Society 75: 741–743.
Levin DA. 1978. Pollinator behaviour and the breeding structure of plant populations. In: Richards AJ, ed. The pollination of flowers by insects. London: Academic Press, 133–150.
Lewis D, Jones DA. 1992. The genetics of heterostyly. In: Barrett SCH, ed. Monographs on theoretical and applied genetics 15. Evolution and function of heterostyly. Berlin: Springer-Verlag, 129–151.
Maynard-Smith J. 1978. The evolution of sex. Cambridge: Cambridge University Press.
Mbewe MN. 1992. Domestication of morama beans (Tylosema esculentum, Fabaceae). International Conference on Development of New Crops, 8–12 March, 1992, Jerusalem, Israel.
Monaghan B. 1993. Assessment of genetic variation in morama populations as a basis for morama improvement in Botswana. Project report to Faculty of Agriculture and Forestry, University of Melbourne.
Owens SJ. 1990. The morphology of the wet, non-papillate (WN) stigma form in the tribe Caesalpinieae (Caesalpinioideae: Leguminosae). Botanical Journal of the Linnean Society 104: 293–302.
Owens SJ, Lewis GP. 1996. Stigma morphology in the Leguminosae: The wet papillate (WP) stigma in Caealpinioideae. Kew Bulletin 51: 119–131.
Owens SJ, Stirton CH. 1989. Pollen, stigma and style interactions in the leguminosae. In: Stirton CH, Zarucchi JL, eds. Advances in legume biology. Monographs in Systematic Botany from the Missouri Botanical Garden, St Louis. 105–112.
Powell AM 1987. Marama bean (Tylosema esculentum, Fabaceae) seed crop in Texas. Economic Botany 41: 216–220.
Richards AJ. 1997. Plant breeding systems, 2nd edn. London: Chapman & Hall, 58–59.
Richards AJ, Mitchell J. 1990. The control of incompatibility in distylous Pulmonaria affinis Jordan (Boraginaceae). Botanical Journal of the Linnean Society 104: 369–380.
Stephenson AG. 1981. Flower and fruit abortion: proximate causes and ultimate functions. Annual Review of Ecology and Systematics 12: 253–279.[CrossRef][Web of Science]
Wiens D, Calvin CD, Wilson CA, Davern CI, Frank D, Seavey SR. 1987. Reproductive success, spontaneous embryo abortion and genetic ‘load’ in flowering plants. Oecologia 71: 501–509.[CrossRef]
Wunderlin R, Larsen K, Larsen SS. 1987. Reorganization of the Cercideae (Fabaceae: Caesalpinioideae). Copenhagen: The Royal Danish Academy of Sciences and Letters.
CiteULike 
Connotea 
Del.icio.us What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||