Evidence for anthropic selection of the Sheanut tree(

Vitellaria paradoxa)

P. N. LOVETT1 and N. HAQ2, *1

The Shea Project, COVOL Uganda, P.O. Box 833, Lira, Uganda (E-mail: covol@bushnet.net);2Environmental Research Group, International Centre for Underutilised Crops, The Universityof Southampton, Highfield, Southampton, SO17 1BJ, UK (*Author for correspondence: E-mail:haq@soton.ac.uk)

Key words: tree domestication, unconscious selection, Vitellaria paradoxa Gaertn.

Abstract. The Sheanut tree (Vitellaria paradoxa Gaertn.), a multi-purpose species highly valuedfor the oil obtained from its seeds, is commonly maintained in the semi-arid parklands of sub-Saharan West Africa. An inventory in the West Gonja District, Northern Region, Ghana, revealedthat on intensively farmed land this species constituted 79.7 ± 7.2% (Basal area = 2.19 ± 0.64m2 ha–1) of the woody biomass, on low intensity farmland 84.2 ± 10.0% (2.16 ± 0.57 m2 ha–1)and only 10.2 ± 3.3% (0.92 ± 0.23 m2 ha–1) in unmanaged woodland, with similar environ-mental characteristics. No significant differences were found between total Sheanut treedensities on different land use intensities, although as a proportion of all trees surveyed, largetrees were more common on farmed land. Participatory surveys revealed that these populationsare a direct result of anthropic selection as local farmers eliminate unwanted woody species onfarmland, leaving only those Sheanut trees that meet criteria based on spacing, size, growth,health, age and yield. Characteristics that could affect population dynamics during traditionalmanagement and harvesting including short viability seeds and cryptogeal germination are alsodiscussed with reference to unconscious selection. Tree improvement is currently constrained,as true to type varieties are difficult to propagate. It is proposed that Sheanut trees on farmlandare semi-domesticated having been subject to long-term anthropic selection during cycles oftraditional fallow and crop cultivation.

Introduction

The Sheanut tree, Vitellaria paradoxa Gaertn. (Family – Sapotaceae, syn.Butryospermum paradoxum), is a major component of the woody flora of theSudan and Guinea savannah vegetation zones of sub-Saharan Africa. Thespecies range forms an almost unbroken belt approximately 5,000 km longby 500 km wide from Senegal to Uganda (Bonkoungou, 1987). Vitellariaparadoxa is an important source of many non-timber products – the nutritiousfruits are eaten, the bark is used for traditional medicines and the latex is usedfor making glue. The wood is used for charcoal, construction, for furnitureand as pounding mortars (Dalziel, 1937; Abbiw, 1990). It is also of note thatthe pestles for use with these mortars, are poles from the so-called ‘male’Sheanut tree (Lophira lanceolata Van Tiegh. ex Keay), with which Vitellariais sometimes confused. The most valued product (usually harvested, processedand sold by women) is the Shea butter (francophone: le beurre de Karité)

Agroforestry Systems 48: 273–288, 2000. 2000 Kluwer Academic Publishers. Printed in the Netherlands.

extracted from the dried kernels, which is the main oleaginous product inmany areas of this species semi-arid range (Lamien et al., 1996). The oil iswidely utilised for local domestic purposes such as cooking, as a skin mois-turiser or as an illuminant. Shea butter is also utilised commercially as aningredient in cosmetic, pharmaceutical and edible products (Abbiw, 1990).Due to the importance of this species, Vitellaria has been included on thepriority list of African forest genetic resources by FAO (FAO, 1977, 1984,1988).

As a commodity of trade, Shea butter has had a very long history in WestAfrica, described by Ibn-Batutta, an Arab traveller in the 14th Century (Mainy,1953) and by Labat in 1728 (Dalziel, 1937). In 1797, the Scottish explorerMungo Park noted that traditional management was to selectively maintainSheanut trees on land cleared for cultivation (Park, 1799). This form of man-agement has probably been practised for many centuries prior to these obser-vations, as recent evidence suggests that at a site in Northern Burkina Faso,ca. 1000 AD, Sheanut trees were being maintained on cultivated land(Neumann et al., 1998). The Sheanut tree has a sacred status where utilisedand has often been described as a major component of the indigenous yam,sorghum or millet based farming systems practised in West Africa (Clark,1980; Boffa, 1995; Harlan, 1992; Pullan, 1974; Kessler, 1992). Although thebest specimens are commonly found on farmed land, available evidencesuggests that Sheanut trees have been managed in traditional farming systemsfor centuries. This species has usually been described as ‘wild’ as opposedto being a domesticated species (Abbiw, 1990; Baker, 1962; Hall et al., 1996).

Archaeological evidence for the domestication of plants in West Africa hasbeen scarce and dates of ca. 5,000 yrs bp are usually offered (Blumler, 1996),however, linguistic evidence has suggested dates at least 8,000 yrs bp (Ehret,1984). What is certain, is that many annual crops, including African Rice,Yam, Sorghum, Millet and Tef, originated in sub-Saharan Africa, a locationthat Harlan (1971) refers to as a ‘non-center’ of agricultural origins due tothe large expanse of this area. It has also been suggested, following long termassociations with man, that a number of African tree species are also in anintermediary state of becoming domesticated, but since this is an ongoingand iterative process there is no clear point at which a species can be said tobe no longer wild. These include the Boabab (Adansonia digitata L.), theWhite-barked Acacia (Faidherbia albida (Del.) A.Chev.), the West Africanoil palm (Elaeis guineensis Jacq.), the Cabbage tree (Moringa stenopotalaBak.F. Cuf.) and the Locust bean tree (Parkia biglobosa (Jacq.) Don.) (Harlan,1992; Jahn, 1991; Z. Teklehaimanot, pers. com., 1999).

In Northern Ghana, tradition dictates that the Sheanut tree is rarely plantedand most individuals result only from natural regeneration. Vitellaria alsohas recalcitrant seeds, a widely out-crossing breeding system and a lengthymaturation period (up to 15–20 years). Therefore, tree improvement through‘methodical’ selection is a laborious option and has probably not beenattempted by local farmers who also describe this species as wild. This does

274

not preclude other types of ‘human-induced change’ as reports describeVitellaria populations where larger ‘superior’ individuals are common onfarmland but regeneration abundant only on fallow land (Abbiw, 1990; Hallet al., 1996). Assuming that the maintenance of healthy trees in a man-madeenvironment, following the removal of unwanted individuals during cyclicalfallow clearance, offers an analogous domestication mechanism to that whenseeds of annual crops are reserved for planting the following season. Thesestudies offer evidence for the domestication of the Sheanut tree following‘unconscious’ or ‘natural’ selection in environments controlled by anthropicactivities.

The majority reports compare Vitellaria populations on cultivated landwith those on fallow land previously disturbed by anthropic activities, asopposed to ‘wild’ populations in completely unmanaged areas (Boffa, 1995;Schreckenberg, 1996; Osei-Amaning, 1996). The lack of small diameter treeshas also been given as evidence for the degradation of West African parklands(Gijsbers et al., 1993). Therefore, this study aimed to use a quantitative inven-tory to compare Vitellaria populations, in an environmentally homogenousarea, on sites with different intensities of landuse (high, medium and low).Using participatory interview methods with local farmers, this study furtheraimed to investigate the role that anthropic activities have had on the man-agement, conservation and possible domestication of Vitellaria paradoxa inthe farmed parklands of Northern Ghana.

Materials and methods

Tree inventory

Using Landsat thematic images (January 1986), an area encompassing dif-ferent landuse intensities was selected in the West Gonja District, NorthernRegion, Ghana (Figure 1). Following discussions with long-term residents,it was also established that few changes had occurred in the previous tenyears (apart from the introduction of mechanised ploughing and fertiliser usenear Bole). The intensities of landuse were defined as follows: low – unman-aged woodland south of the hamlet of Jintigi, medium – scattered farmlandaround the small village of Mandari and high – intensively farmed land aroundthe more densely populated district capital of Bole. Using soil maps (FAO-UNESCO, 1977; FAO, 1968), 1:50,000 maps (Survey of Ghana, 1966) andinformation from a local geological survey (S. Meadows-Smith, pers. comm.,1997), it was determined that all locations had similar soil types (FAO soilunit Lf31-a, petric phase), altitude (ca. 300 m above sea level) and geology(pre-Cambrian granites). Using data from Meteorological Services Depart-ment, Legon (C. Adams, pers comm., 1995), climatic similarity was assumeddue to the close proximity (within 25 km) of all locations to Bole (1092 ±72.9 mm yr–1, 33 to 21

°C).

275

1 square hectare plots were surveyed on randomly selected farms, betweenMay and September 1997, after obtaining permission from local landownerson medium intensity farmland (around Mandari village) and on high inten-sity farmland (near Bole town). To prevent potential confusion with farmedland that had been left fallow and areas of unknown management history, plotswere located only on land currently being cultivated. Due to high tree densityand transport limitations, smaller 0.25 ha plots were located in the low inten-sity area (south of Jintigi hamlet, in woodland about twenty miles west ofBole) using random positions on a geological survey grid. All woody stemsinside these plots (≥ 7 cm diameter at breast height, dbh) were measured andidentified by species.

Participatory questionnaires

With the aim of assessing traditional agroforestry practices in Northern Ghana,and to determine whether the West Gonja study sites were representative ofGhanaian parklands, participatory interviews were held with 130 people during39 interview sessions in three regions of Ghana. Since gender was an impor-tant factor with regard to Sheanut related activities, women were also asked

276

Lighter areas indicate vegetation clearance (cloud cover minimal)

Figure 1. Landsat thematic image of study area (18th January 1986) in the Northern Region,Ghana.

to express their views. Women made comments at 59% of interviews althoughthey were the sole or main respondents at only 15%. Information was obtainedon the farm and tree management systems employed in the area (e.g. farmsize, crops grown, fallow duration, utilisation of tree species, favouredVitellaria traits and selection at land clearance). Care was taken to ask ques-tions without suggestion and to crosscheck answers. Answers were also con-firmed by local knowledge gained during two years of field research in thearea.

Tree selection on farmed parklands was similar throughout Northern Ghanaand almost pure stands of Sheanut trees were frequently seen in fields plantedwith yams, millet, sorghum, maize, cassava or legumes. Most farmers stillpractised the traditional method of hand ploughing, but deep mechanisedploughing was more frequent near larger habitation centres in intensivelyfarmed areas. Animal traction was, however, the preferred power source inthe Upper-West region of Ghana. Farms were generally similar, although thedistances from the home to farm in the study area were significantly greaterthan at other locations (P < 0.01). The other notable difference was that farmsin the Upper West region were generally larger and fallow periods were shorter(P = 0.07–0.10), indications of a higher intensity of landuse than that occur-ring in the study area (Table 1).

Data analysis

To compare species diversity, Simpson’s diversity index was calculated. Thisindex provides an integrated value for the two components of α-diversity,variety and evenness (Hamilton and Perrott, 1981). The results are presentedin the form 1-D per plot, where values approaching 1 indicate areas withmany species, each at low density, and a value of 0 indicates an area with onlyone species. For two areas with an equal number of species, higher valuesfor 1-D indicate evenness. The percentage contribution to woody biomass wascalculated as the total basal area (BA) per hectare of each species divided bythe total BA per hectare of all species.

BA [m2 ha–1] = α(dbh in cm/200)2 × 10,000/plot size in m2

The Sheanut tree population structure was examined by plotting the meandensity of stems occurring at each location in each 7 cm dbh size class. UsingSPSS for Windows ver. 8.0, ANOVA was used to determine significant dif-ferences between quantitative site variables and post-hoc pairwise multiplecomparisons were performed using Tukey’s honestly significant differencetest.

277

278

Table 1. Participatory interviews and farm descriptors in the Northern Region, Ghana.

Region Location Villages Interviews Men Women Total Farm size House to farm Time farmed Time fallow/Acres miles yrs yrs

Brong Ahafo Wenchi 3 10 18 17 35 5.1 ± 2.1 2.0 ± 0.5 2.9 ± 0.7 8.1 ± 2.7Northern Bole 5 21 41 31 72 5.3 ± 1.4 3.6 ± 0.7 3.0 ± 0.3 7.2 ± 2.0Upper West Nandom/Wa 8 08 14 09 23 6.4 ± 1.9 1.5 ± 0.6 3.9 ± 0.7 4.3 ± 1.7

Results

Woody species composition of sites

A significant variation in the total number, and abundance of each tree speciesoccurred between sites. In particular, species commonly utilised for theiredible or other non-timber products, were found in greater numbers onfarmland. Parkia biglobosa (Don) was not found in any woodland plots andlarge individuals of Diospyros mespiliformis (Hochsetter ex A. de Condelle)– a species locally taboo to cut – were more common near Bole (Table 2).The number of individual trees, total woody biomass, species number anddiversity was also significantly lower in farmed areas (Table 3).

Results for Vitellaria only, show no significant variation in numbers of indi-viduals (dbh ≥ 7 cm) between sites. The basal area and total number of largeindividuals (dbh > 20 cm), however, were significantly greater on high inten-sity farmland near Bole than in natural woodland (Table 4). It was evidentthat few small (dbh < 13 cm) individuals of Vitellaria were maintained onland cleared for cultivation, particularly in farms mechanically ploughed, andring barking was observed as a method used to kill occasional mature trees.In woodland, the population structure showed many similarities to other‘natural’ systems, with abundant small, probably immature individuals andfew large trees (Figure 2). At this site, 82.6% of all Sheanut stems sampledwere small trees (dbh < 20 cm) although some plots contained only treesgreater than this size class. All mature trees in farmland, greater than 20 cmdbh, were seen to bear at least some fruit during the harvest season. Inwoodland, even after extensive searching of the whole area during the sameperiod, only one individual with fruit was found (dbh = 18 cm). Mature fruitcollection from this individual required vigorous shaking and viability washigh even four weeks after collection. On farmed land, fruit fell to the groundwhen mature (April–June) but all collections lost viability after three weeks.

Indigenous agroforestry management

When farmers were questioned whether they perform selective managementon Sheanut trees, the following results were obtained (as a mean of all threeregions):

• 72 ± 37% practised some form of selection during land clearance forcultivation.

• 37 ± 37% admitted to having ever cut any productive mature trees.• 58 ± 24% protected immature trees on cultivated land.• The majority of farmers said they would leave dense stands and cultivate

elsewhere.

This confirms field observations described above since male farmers saidthat although cutting of Sheanut trees was locally taboo, they still removed

279

280

Table 2. Top ten woody species at sites in North Ghana.

Bole (high intensity farmland) Mandari (med. intensity farmland) Unmanaged Woodland

Species Use1 %2 Species Use % Species Use %

Vitellaria paradoxa EBFM 79.7 Vitellaria paradoxa EBFM 84.2 Berlinia grandiflora BF 32.2Parkia biglobosa EF 07.1 Hannoa undulata FMC 03.6 Unknown species #15 – 12.2Diospyros mespiliformis EFM 04.4 Sterculia setigera M 02.4 Vitellaria paradoxa EBFM 10.2Azadirachta indica M 03.3 Parkia biglobosa EF 02.0 Pericopsis laxiflora F 10.1Daniellia oliveri SM 01.6 Daniellia oliveri SM 01.8 Isoberlinia doka BF 08.0Hannoa undulata FMC 00.7 Diospyros mespiliformis EFM 01.6 Lannea acida EF 06.3Pericopsis laxiflora F 00.6 Vitex doniana E 00.9 Afzelia africana BM 04.4Sterculia setigera M 00.5 Lannea acida EF 00.7 Combretum molle F 02.8Lannea acida EF 00.4 Bombax costatum C 00.7 Terminalia avicennioides F 02.5

1 Local utilisation of products (E – Edible, B – Building, F – Fuel, S – Soil improver, M – Medicine, C – carving).2 Average percentage contribution to average basal area at each site.

both unwanted immature and mature Sheanut trees from the land they wishedto cultivate. The high confidence limits are probably indicative of farmersbeing afraid of admitting to a stranger that they cut this valuable economicspecies since traditional agroforestry practices often conflict with that of thestate forestry laws.

Of those farmers who practised any form of selection, 77% described thecriteria used to keep an individual tree were usually based on age and fruitproductivity, stating that mature trees were kept unless unproductive due toold age or poor health. 19% of ‘selective’ farmers revealed that they wouldkill trees that significantly affected annual crops, by reducing light availability,and if at high density, the healthiest, most productive trees were preferentiallymaintained. 36% of farmers protecting regenerating saplings stated that theywould maintain those that were taller than 1 m at land clearance, providingthey were healthy, fast growing and well spaced with relation to othertrees.

When describing the ‘best’ trees, farmers had a wide range of preferencesand often gave more than one character. Good health, low competitive effectson crop yield, large sweet fruit, fast growth and resistance to mistletoe weredescribed as preferred traits. Trees with sweet fruit and high yield were equallythe most popular traits and both were mentioned by 45% of respondents. Someconfusion existed as to the preferred seed (whole nut) size, 38% of malefarmers said that small seeds yielded more oil whereas 28% said large seedswere better. Farmers in the Upper West Region also described a ‘Black’ varietyof Sheanut tree, which supposedly had high yields and dark buds.

All women, apart from one interviewee, said that small seeds would yieldmore oil than the same volume of large seeds. Fat analysis of seeds collectedfrom different mother trees throughout Northern Ghana (n = 140 trees, fourseeds per tree) revealed that no correlation (r2 < 0.01, P > 0.33) existedbetween mean fat content per tree (range = 29 to 62% by dry kernel weight)and mean seed size per tree (from 1.74 to 3.74 cm long) (Lovett and Haq, inprep.). The only woman who did not prefer small seeds said that she was alazy market trader, and as Sheanuts were sold by volume, it was easier to filla basin with large seeds.

281

Table 3. Comparison of biomass at sites in North Ghana – all woody species ≤ 7 cm dbh.

Site (intensity) No. of Trees Total BA Species1 1-Dplots ha–1 m2 ha–1 plot–1

Bole (high) 15 029 ± 13a 2.7 ± 0.7a 04 ± 2a 0.30 ± 0.13a

Mandari (med.) 06 061 ± 24a 2.6 ± 0.6a 08 ± 3ab 0.37 ± 0.19a

Woodland (low) 06 353 ± 61b 9.6 ± 2.0b 13 ± 2b 0.81 ± 0.03b

Means by column with no similar superscript letters indicate Tukey’s honestly significantdifference (P < 0.01).1 Not possible to give species per hectare due to different plot sizes.

282

Table 4. Comparison of biomass in sites in North Ghana – Vitellaria paradoxa only.

Site (intensity) No. of Trees Stems % Stems BA % contribution BAplots ha–1 ha–1 > 20 cm dbh m2 ha–1 to total BA Tree–1

Bole (high) 15 21 ± 6a 24 ± 7a 72.0 ± 9.4a 2.19 ± 0.64a 79.7 ± 7.2a 0.11 ± 0.02a

Mandari (med.) 06 44 ± 11a 48 ± 12a 53.1 ± 15.1ab 2.16 ± 0.57ab 84.2 ± 10.0a 0.05 ± 0.01b

Woodland (low) 06 41 ± 28a 46 ± 30a 40.2 ± 32.9b 0.92 ± 0.23b 10.2 ± 3.3b 0.05 ± 0.04b

Means by column with no similar superscript letters indicate Tukey’s honestly significant difference (P < 0.05).

Local knowledge on the Sheanut tree was extensive and 88% of respon-dents said they knew where the best (the healthiest and most productive) treesgrew. 40% of these farmers described topographic situation as the most impor-tant influence and the best trees were always said to be on higher land. 28%said that surrounding vegetation and soil type were equally important andthe best individuals were said to be found on cultivated land with well-drainedstoney or loamy soils. Only 4% mentioned that fire had any detrimental effecton Sheanut trees, stating that if vegetation was burnt during the floweringseason (November–January) the fruit yield would be reduced.

Women and children are the main collectors of mature Sheanut fruits and

283

Points without similar lower case letters, by size class, indicate Tukey’s honestly significantdifference (P < 0.05)

Figure 2. Size Class distribution of Vitellaria paradoxa, Northern Region, Ghana.

are given priority rights to harvest from land currently being farmed by familymembers. Open access rights are granted in fallow or unmanaged areas,although women often described that it was easier and less hazardous to collectfallen fruit from cleared land. During germplasm collections, it was notedthat few viable Sheanuts remained on cleared land as women gather every-thing they find in the early morning. Harvested seeds are depulped and imme-diately boiled before sun drying. This processing stage is said to improveyields and oil quality by preventing germination and fungal infection of seeds.Despite this extensive knowledge and intense local competition for Sheanuts,a traditional two-week ban on all harvesting was strictly enforced once fruitabscission had commenced. The reasons behind this were explained by localresidents and a number of alternatives offered. One viewpoint suggested thatit freed up human resources during a busy cultivation period. Another expla-nation was that the ban allowed land pacification ceremonies to take placeand ensured good rainfall. The last reason, described by an influential memberof the community, was that the strictly enforced fines allowed all women tohave an equal share of the harvest when the ban was lifted.

Discussion

Traditional anthropic agroforestry practices have clearly had a significantimpact on both the woody species composition and, on the population struc-ture of trees on farmland in the study area. It is also clear that the majorityof trees recorded on cultivated land are maintained because their non-timberproducts are locally valued. The most frequent species on farmed land wasthe Sheanut tree (Vitellaria paradoxa), which often formed over 80% of thewoody biomass. Field observations revealed that the main selection stageoccurs when fallow or woodland is cleared, as most immature individualsare removed and only certain large trees are maintained on cultivated land.Over time, this has led to significant changes in the Vitellaria population struc-ture as compared to that expected in the ‘natural’ environment, and a muchlarger proportion of mature productive trees (dbh > 20 cm) are present onfarmed land than in unmanaged woodland. Similarities of the medium inten-sity landuse site (around Mandari) with unmanaged areas, in terms of totalspecies numbers, Vitellaria biomass and population structure may be anartefact of small sample sizes or different plot sizes. Alternatively, it mayindicate that landuse is either better managed, at an intermediary stage ofintensity with longer fallow periods or that farmland has been expanding intorecently unmanaged woodland.

Important considerations for Vitellaria conservation and management areimplied by the differences between the high (Bole) and medium (Mandari)intensity farmland because deep ploughing reduces regeneration and damagesmature tree root systems. The local elders near Mandari strictly enforcetraditional methods and it is apparent that this site has a healthy population

284

structure. Farmland around Bole, however, has a much higher incidence oftractor ploughing and also lacks regeneration. Fallow length also appears tobe decreasing at this site and many fields are now permanently farmed withthe use of fertilisers. Field observations and interviews in other regions alsosupport this view since a similar Vitellaria population structure was observedon farmland near Nandom in the Upper West region of Ghana, an area wherecattle ploughing is widespread and fallows are shorter.

Interviews with local farmers, revealed that male farmers are activelyselecting specific Sheanut trees when land is cleared for cultivation fromfallow or woodland. The selection criteria used for mature trees are basedprimarily on fruit productivity (as a function of age, health and size) and com-petitive effects with annual crops (determined by tree size, leaf density andspacing). These views are further supported by similarities in the mean totalwoody biomass and Vitellaria biomass between farmland sites, which suggestthat farmers are actively maintaining a careful balance between crop and treeproductivity. Unwanted mature trees are usually killed by ring barking andallowed to dry before being cut for household uses (building poles, mortarsand charcoal). New recruits for farmland Sheanut populations are selectedfrom regeneration by not cutting and then protecting from fire, during thecyclical clearing of fallow land (uncultivated for 5–10 years in the study area).The criteria used to select immature Sheanut trees were based on spacing,health and speed of growth (only saplings > 1 m kept).

Apart from characters given above, others preferred traits included fruitpalatability, seed size and resistance to mistletoe, but since this species is notplanted, these were not obviously part of the selection process. Men statedthat they like large sweet fruit but were not convinced as to which seed sizewas better for oil production. Women, who are the sole processors of Sheabutter in the area, clearly stated that small seeds give a higher yield of oil byvolume. Laboratory analysis (Lovett and Haq, in preparation) showed no cor-relation of seed size with oil content (percentage by dry weight), supportinga preference for seeds with a high mass to volume ratio. It is also conceiv-able that oil is more easily extracted using traditional methods from certainseed types and local views deserve further research.

Other ‘unconscious’ anthropic selection pressures on Vitellaria are alsopossible following intensive harvesting and land cultivation during the ger-mination period. This is particularly relevant in the study area since Osei-Amaning (1996) suggests that most immature trees in fallow around Boleresult from coppice. Although further research is required to determine theexact period when seedlings actually become established during the farmingcycle, it is suggested that the traditional two week ban, coupled with thoroughcollection of viable seeds, would encourage survival of seeds which absciseon maturity and germinate immediately. It is also of note that traditional banson early harvesting are a common occurrence in many areas of West Africa(J-M. Boffa, pers. com., 1999). Mature seeds that remain on the tree or hadlong viability would be more likely to be collected for oil production since

285

germinated seeds are usually discarded. Cryptogeal germination of Vitellariaalso occurs during the cultivation period and new shoots do not appear aboveground until planting and weeding has been completed four to eight weekslater. Since the land is usually burnt four to six months later, this unusual char-acteristic, which probably originally evolved in response to fire (Jackson,1974), may now be protecting seedlings from the intensive but shallow handcultivation commonly practised throughout the region. As with other culti-vated crops, it is therefore possible that natural dispersal mechanisms ofVitellaria have adapted to survive in environments which alternate betweenintensive cultivation and fallow land.

The length of time for ‘domesticated’ adaptations to evolve in a popula-tion varies between species, though it has been postulated that populations ofcultivated wild lentils could develop 100% indehiscence in as little as twentygenerations (Blumler and Bryne, 1991). The Gonja tribe conquered the indige-nous Vagla and Safalba tribes near Bole, after migrating from Mali in thesixteenth century. Agriculture was probably already well established at thispoint as the latter tribes are still given a degree of control over land and treemanagement. It is also likely that population densities, though not as high aspresent, were quite large up until recently as there has been considerablemilitary activity and tribal migrations in the area (Goody, 1969). Assumingan average farm cycle of 15 years and similar agricultural practices, this periodwould have allowed at least thirty Vitellaria generations to have establishedsince the arrival of the Gonja.

Evidence presented confirms that Vitellaria populations, in the West Africanfarmed parklands, are a direct result of anthropic selection following manycenturies of traditional cultivation and fallow. Although ‘methodical’ selec-tion has probably been limited as preferred varieties are not planted or spe-cially protected. It is proposed that the maintenance of healthy, productivetrees well suited to these agroforestry systems, coupled with intensive har-vesting and cultivation across the species range, has led to Vitellaria paradoxabecoming semi-domesticated following ‘unconscious’ and ‘natural’ selec-tion mechanisms.

In recent years, much attention has been paid to those woody plants termedas ‘Cinderella’ species which provide economic or environmental benefitsthrough traditional utilisation but have been often been neglected by main-stream domestication (Leakey and Newton, 1994). The focus of this study isjust such a species, with regard to current agroforestry requirements, Vitellariaparadoxa still has many undesirable characteristics. Trees take up to 20 yearsto mature, the seeds are recalcitrant and the heterozygous populations producea highly variable crop in terms of quantity and quality. Unfortunately, superior‘true-to-type’ individuals are difficult to maintain as the species has an out-crossing breeding system and vegetative propagation methods have onlyrecently been developed (Bonkoungou, 1987; Grolleau, 1989; Lovett et al.,1996). Analysis of diversity followed by identification and multiplication ofindividuals with superior characteristics (from both domesticated and wild

286

populations) will allow ‘methodical’ selection and continuation of a treeimprovement process first started by African farmers thousands of years ago.

Acknowledgements

We express our gratitude to: staff at the Cocoa Research Institute of Ghana,SheaGold Ltd, the Botany Department, University of Ghana, the many farmersand other Ghanaians, for their kind assistance during fieldwork and to Dr Jean-Marc Boffa for comments on earlier drafts of this paper.

References

Abbiw DK (1990) Useful Plants of Ghana, West African Uses of Wild and Cultivated Plants,pp 66–67. Intermediate Technology Publications and The Royal Botanic Gardens, Kew,London

Baker HG (1962) Comments on the thesis that there was a major centre of plant domesticationnear the headwaters of the River Niger. Journal of African History 3(2): 229–233

Blumler MA and Byrne R (1991) The ecological genetics of domestication and the origins ofagriculture. Current Anthropology 12(1): 23–54

Blumler MA (1996) Ecology, evolutionary theory and agricultural origins. In: Harris DR (ed)The Origins and Spread of Agriculture and Pastoralism in Eurasia, pp 25–50. UCL Press Ltd,London

Boffa J-M (1995) Productivity and Management of Agroforestry Parklands in the Sudan Zoneof Burkina Faso, West Africa. Ph.D. thesis, Purdue University, West Lafayette, Indiana.101 pp

Bonkoungou EG (1987) Monographie du Karité, Butryospermum paradoxum (Gaertn. F.)Hepper, espèce agroforestière à usages multiples. Ougadougou, Burkina Faso, Institute deRecherche en Biologie et Ecologie Tropicale. 69 pp

Bonkoungou EG (1997) Report on seed collection of Vitellaria paradoxa in Burkina Faso, Mali,Senegal and Uganda. ICRAF and IPALAC. 19 pp

Clark JD (1980) Early human occupation of African savanna environments. In: Harris DR (ed)Human Ecology in Savanna Environments, pp 41–71. Academic Press, New York

Dalziel JM (1937) The Useful Plants of West Tropical Africa, pp 350–354. Crown Agents,London

Ehret C (1984) Historical/linguistic evidence for early African food production. In: Clark JDand Brandt SA (eds) From Hunters to Farmers: The Causes and Consequences of FoodProduction in Africa, pp 26–35. University of California Press Berkeley

FAO (1968) Ghana, Land and water survey in the Upper and Northern Regions (Volume 1).FAO, Rome

FAO-UNESCO (1977) Soil Map of the World, Volume VI. Africa, UNESCO, ParisFAO (1977) Appendix 8, Forest Genetic resource priorities. 8. Africa. Report of the Fourth

session of the FAO panel of Experts on Forest Gene Resources, held in Canberra, Australia,9–11 March 1977, pp 62–64. FAO, Rome. 75 pp

FAO (1984) Appendix 8, Forest Genetic resource priorities. 12. Africa. Report of the Fifthsession of the FAO panel of Experts on Forest Gene Resources, held in Rome, Italy, 8–11December 1981, pp 86–89. FAO, Rome. 79 pp

FAO (1988) Appendix 5, Forest Genetic resource priorities. 10. Africa. Report of the Sixthsession of the FAO panel of Experts on Forest Gene Resources, held in Canberra, Australia,held in Rome, Italy, 8–11 December 1985, pp 65–69. FAO, Rome. 79 pp

287

Gijsbers HJM, Kessler JJ and Knevel MK (1994) Dynamics and natural regeneration of woodyspecies in farmed parklands in the Sahel region (Province of Passore, Burkina Faso). ForestEcology and Management 64: 1–12

Goody J (1969) Comparative Studies in Kinship: Marriage Policy and Incorporation in NorthernGhana, pp 152–153. Routledge and Kegan Paul Ltd, London

Grolleau A (1989) Contribution a l’etude de la multiplication vegetative par greffage du Karité(Vitellaria paradoxa Gaertn. f. = Butryospermum paradoxum Hepper). Bois et Forets desTropiques 222: 38–40

Hall JB, Aebischer DP, Tomlinson HF, Osei-Amaning E and Hindle JR (1996) Vitellariaparadoxa: a monograph. School of Agricultural Sciences Publication Number: 8. Universityof Wales, Bangor. 105 pp

Harlan JR (1971) Agricultural origins: centers and noncenters. Science 174: 468–474Harlan JR (1992) Crops and Man. 2nd ed., pp 65–99, 177–191. Madison, Wisconsin: American

Society of Agronomy/Crop Science Society of AmericaHamilton AC and Perrott RA (1981) A study of altitudinal zonation in the montane forest belt

of Mt. Elgon, Kenya/Uganda. Vegetatio 45: 107–125Jackson G (1974) Cryptogeal germination and other seedling adaptations to the burning of veg-

etation in savanna regions: The origin of the Pyrophytic habit. New Phytologist 73: 771–780Kessler JJ (1992) The influence of Karité (Vitellaria paradoxa) and néré (Parkia biglobosa)

trees on sorghum production in Burkina Faso. Agroforestry Systems. 17: 2, 97–118Lamien N, Sidibé A and Bayala J (1996) Use and commercialization of non-timber forest

products in western Burkina Faso. In: Leakey RRB, Temu AB and Melnyk M (eds)Domestication and Commercialization of Non-Timber Forest Products in AgroforestrySystems, Non-wood Forest Products No. 9, pp 51–64. FAO, Rome

Leakey RRB and Newton AC (eds) (1994) Domestication of tropical trees for timber and non-timber products. MAB Digest 17, UNESCO, Paris. 97 pp

Lovett P, Azad AK, Paudyal K and Haq N (1996) Genetic diversity and development of prop-agation techniques for tropical fruit trees. In: Smartt J and Haq N (eds) Domestication,Production and Utilisation of New Crops pp 286–287. ICUC, UK

Mainy R (1953) Notes historiques autour des principales plantes cultivées d’Afrique Occidentale.Bulletin de l’Institut Français d’ Afrique Noire. 15: 684–730

Neumann K, Kahlheber S and Uebel D (1998) Remains of woody plants from Saouga, a medievalwest African village. Vegetation History and Archaeobotany. 7: 57–77

Osei-Amaning E (1996) Management of Vitellaria paradoxa in Guinea Savanna Rangelands inGhana. Ph.D. thesis. University of Wales. 199 pp

Park M (1799) Travels into the interior districts of Africa performed under the direction andpatronage of the African Association in the years 1795, 1796 and 1797. In: Swift J (1983)Travels into the Interior of Africa, pp 1–264. Eland, London

Pullan RA (1974) Farmed parkland in West Africa. Savanna 3(2): 119–151Schreckenberg K (1996) Forests, Fields and Markets: A Study of Indigenous Tree Products

in the Woody Savannas of the Bassila Region, Benin. Ph.D. thesis, University of London.326 pp

Survey of Ghana (1966) 1:50,000 topographic maps of Ghana. Accra

288