Quantitative descriptors of variation in the fruits and seedsof

Irvingia gabonensis

R. R. B. LEAKEY1, J.-M. FONDOUN2, A. ATANGANA3 andZ. TCHOUNDJEU4

1

Centre for Ecology and Hydrology (CEH), Bush Estate, Penicuik, Midlothian, Scotland, EH260QB, UK (E-mail: RRBL@ceh.ac.uk); 2 Institut de la Recherche Agricole pour le Développement(IRAD), BP 2123, Yaoundé, Cameroon; 3 Department of Plant Biology and Physiology,University of Yaoundé I, BP 812, Yaoundé, Cameroon; 4 International Centre for Research inAgroforestry (ICRAF), BP 2123, Yaoundé, Cameroon (E-mail: Z.Tchoundjeu@camnet.cm)

Key words: Bush mango, Dika nut, domestication, ideotype

Abstract. Methods were developed to quantify variation in the fruit, nut and kernel traits usingthe fruits from four trees of Irvingia gabonensis, an indigenous fruit tree of west and centralAfrica. The measurement of 18 characteristics of 16–32 fruits per tree identified significantvariation in fruit, nut and kernel size and weight, and flesh depth. Differences were also iden-tified in shell weight and brittleness, fruit taste, fibrosity and flesh colour. Relationships betweenfruit size and weight with nut and kernel size and weight were found to be very weak,indicating that it is not possible to accurately predict the traits of the commercially-importantkernel from fruit traits. Seven key qualitative traits are recommended for future assessments ofthe levels of genetic variation in fruits and kernels. These traits describe ideotypes for fresh fruitand kernel production.

Introduction

Recent work in the humid lowlands of West Africa has established that thereare substantial local and regional markets for the non-timber forest products(NTFP) of certain indigenous trees, such as Irvingia gabonensis (Aubry-Lecomte ex O’Rorke) (Ndoye et al., 1997). Some NTFPs from Cameroon (e.g.Gnetum leaves, Irvingia kernels) have even entered the international marketand a recent market survey in Europe has emphasised their high value.

Studies on the biological variability of indigenous fruit tree species, theirpropagation using cheap and simple methods appropriate for rural develop-ment projects, and their suitability for domestication have been progressivelyincreasing in West Africa over the last 20 years (reviewed by Leakey et al.,1990; Okafor and Lamb, 1994). In the last five years, the International Centrefor Research in Agroforestry (ICRAF) initiated a coordinated initiative acrossthe Region, based on the priorities of subsistence farmers and national researchscientists (Franzel et al., 1996). This species prioritisation activity involvedfield activities such as farmer preference, product ranking, and the valuationand ranking of priority species (Franzel et al., 1996). From this process, thetop priority species identified for humid West Africa was Irvingia gabonensis

Agroforestry Systems 50: 47–58, 2000. 2000 Kluwer Academic Publishers. Printed in the Netherlands.

(Bush mango/Dika nut). Domestication of indigenous fruits through agro-forestry is seen as one of three important issues in the transformation of landuse in Africa (Sanchez and Leakey, 1997), through the establishment of abetter balance between food security and natural resource utilisation. It shouldalso help to alleviate the poverty that drives deforestation (Leakey and Simons,1998) and is an important element of the ‘Woody Plant Revolution’ (Leakey,and Newton, 1994). To be successful in these terms, domestication has to belinked to commercialisation and market expansion. This has to be done inways that will provide smallholder farmers with socio-economic (Leakey andIzac, 1996), policy (Leakey and Tomich, 1999), and marketing benefits(Leakey, 1999).

In Nigeria, Okafor (1974) identified two varieties of Irvingia gabonensis,one with sweet, edible fruits and the other with bitter fruits. The seed cotyle-dons (kernels) of both varieties are, however, used as a food thickening agent,although those of the latter are more important commercially. Harris (1996)subsequently revised the taxonomy calling the bitter variety Irvingia womboluVermoesen. In Cameroon, the trade in these kernels to Gabon, Nigeria,Equatorial Guinea and the Central African Republic has been valued atUS$260,000 per annum (Ndoye et al., 1997). The kernels are an importantsource of both a polysaccharide, which forms the glutinaceous thickeningagent, and an oil. The oil content of these kernels, however, varies from 51%to 72%, depending on the geographic origin of the fruits (reviewed by Leakey,1999). More knowledge of this sort of variability is needed if marketopportunities are to be expanded in support of domestication.

The geographic range of I. gabonensis is from Nigeria to Congo, while thatof I. wombolu is much greater: Senegal to Uganda (Harris, 1996). In order tocapture much of the genetic variation, germplasm collections were imple-mented during the fruiting seasons of 1994/95 (Ladipo et al., 1996). Thesecollections which were made in collaboration with the authorities and farmersin the host countries, were targeted at visible variation and farmers’ percep-tions of superiority. They were made at 10–30 locations per country in thehumid forest belt of Ghana, Nigeria, Cameroon and Gabon, exchangedbetween countries and established as living genebanks at three sites (Ibadanand Onne in Nigeria; Mbalmayo in Cameroon) in 1995. Each genebankcontains approximately 60 accessions (single progenies).

The germplasm collections are being evaluated by ICRAF and partnersfor growth and phenological traits and will become part of a programme ofgenetic testing, with selection of superior individuals. Molecular studies ofthe extent of genetic diversity in these populations are underway to providea baseline for the domestication programme. These studies by Institute ofTerrestrial Ecology have determined that the centre of diversity for I. gabo-nensis is in southern Cameroon, while that of I. wombolu is in south eastCameroon and western Nigeria (Lowe et al., in press), areas postulated asforest refugia during the last Ice Age. For the latter species, the centre ofdiversity could extend either further east or west into areas not covered

48

by this study. As would be expected for outbreeding species, the geneticdiversity within progenies was high, while that between populations was low.This information provides a fundamentally important basis on which to builda farmer-oriented cultivar selection programme.

The second phase of the domestication programme is being initiated inNigeria and Cameroon. In these two areas, field teams are working withfarmers to identify, select and multiply superior trees. To date over 2000mature plus-trees have been identified and these are being propagated vege-tatively by air-layering (Tchoundjeu et al., in press). The first rooted marcottsfrom these selected trees are currently in nurseries and will become stock-plants for further multiplication and, subsequently, these cultivars will beestablished in on-farm trials. In addition, village nurseries are being estab-lished so that the participating farmers are the beneficiaries of the programmein accordance with the Convention of Biological Diversity.

Vegetative propagation techniques are being used for the capture andmultiplication of superior phenotypes as putative cultivars. Currently, studiesare in progress in Nigeria and Cameroon to improve existing methods of airlayering, grafting and budding, as these are all means of propagating fromthe mature crowns of selected trees. However, an attractive alternative wouldbe to propagate by cuttings as this avoids grafting incompatibilities andpotentially gives high multiplication rates. Simple, low technology methodshave been developed in Cameroon for juvenile shoots of tropical trees (Leakeyet al., 1990), including Irvingia gabonensis (Shiembo et al., 1996), but it isknown that mature shoots are more difficult as they become physiologicallyand ontogenetically old. Research is needed to try to rejuvenate mature crownshoots physiologically, while retaining the benefits of reproductive capacitydue to ontogenetic age.

ICRAF’s tree domestication programme for I. gabonensis is currentlyfocussing much of its resources on the vegetative propagation of trees thatfarmers are identifying as superior. There is at present, however, little knowl-edge about farmers perception of genetic variation or their capacity to identifyand capture the potential for improvement through the formation of cultivarsbased on the propagation of elite trees in a wild population. Evidence fromfarmer interviews suggests that there is substantial variation in certain fruittraits, such as fruit size and the taste of fresh fruit pulp, as well as somevariation in the size of the commercially-important kernels. Attempts toidentify elite trees are, however, subject to both the cooperation of farmersand their willingness to disclose knowledge of local trees. The most commonlyavailable information is about variation in fruit sweetness, but this usuallylimited to the differentiation into two classes: sweet and bitter. Farmersindicate that kernel size is important on the market, although there is as yetno indication if this is recognized by price discrimination. It seems to begenerally considered that large kernels will be found in large fruits.

The purpose of the present study was to develop a methodology for thelater quantitative characterization of tree-to-tree variation in I. gabonensis

49

fruits in Cameroon and Nigeria. This assessment of genetic variation is aimedat the determination of:

– the level of diversity available to farmers within the area of their communalownership;

– how farmers’ evaluate superiority in their trees; – farmers’ concepts of genetic variation in fruit traits and their ‘improve-

ment’;– how farmers criteria for selection matches with recognized horticultural

traits of fruit quality, yield, and chemical composition;– the levels of selection intensity being applied by farmers; and– the relationships between the variability of fruit/kernel traits and market

prices.

Methods and materials

Fruits were collected from four I. gabonensis trees in August 1998 fromdifferent locations in Cameroon (Table 1). Fruits were scarce this year as manytrees were not fruiting or fruiting poorly. Consequently, a sampling proce-dure was only tested in the case of Tree 2 Ngalan, where the area below thetree crown was divided into four quadrants. Eight ripe, but undamaged, fruitswere collected from each quadrant at a point 2/3 of the distance from the trunkto the edge of the crown along transects at 90° from each other. The meanweight of 5, 10, 15, 20, 25, 30 fruits from this sample were compared withthat of a larger sample of 90 fruits in order to determine optimal sample size.

On the day of collection, the fruits were weighed using a small portablekitchen scales graduated to 2 g, and measured using calipers graduated to0.1 mm. Fruit flesh depth was measured by a spike attached to the samecalipers. The spike was inserted until it hit the hard nut at the centre of thefruit. The external measurements of the fruits were made in three dimensions(length, width and thickness) and the flesh depth measurements were madeon both sides of the fruit in these three dimensions (Figure 1).

The next day, the fruits were depulped to expose the endocarp (nut) and arecord made of fruit taste (scored 1[bitter]-5 [sweet]), colour (yellow ororange) and fibrosity (scored 1 [nonfibrous]-5 [fibrous]). The nuts were thenleft to dry for seven to 10 days until the remnants of flesh had dried. They

50

Table 1. Locations of I. gabonensis trees that provided fruits for this study.

Tree no. Village Farmer No. of fruits Latitude Longitude

1 Ngomedzap Market 16 3°16

′ 11°11′1 Nkoevos Deux Moise Nkomo 22 2°56′ 11°26′1 Ngalan Odile Essong 26 2°54′ 11°12′2 Ngalan Odile Essong 32 2°54′ 11°12′

were then weighed and measured using the calipers in the same three dimen-sions as the fruits. The nuts were then carefully broken open so that the seeds(kernels) could be extracted. The fresh kernels were then weighed using alaboratory balance (Mettler Toledo PB 3002) and measured in the same threedimensions using the calipers. The weight of the nutshell was determined bythe difference between the nut and kernel weights, and a ‘shell brittleness’factor derived as 50/shell weight. Statistical analyses of the weight and lengthdata were done using Microsoft Excel 97, and skewness was calculated usingGENSTAT 5 Version 4.1.

Results

Sample size

A comparison of the standard errors of the mean values of the first 5, 10, 15,20, 25, and 30 fruits from the 32 fruit sample from Tree 2 at Ngalan, foundthat the mean weight of a sample of 20 or more fruits differing by less thanthe standard error (Table 2). The weights of the 30 individual fruits forminga sample were normally distributed (skewness = –0.02).

51

Figure 1. Diagram of the positions of fruit, endocarp (nut) and seed (kernel) measurement(Lg = length [T= top, B= bottom]; W = width [L = left, R = right]; Th = Thickness [F = front,B = back]).

Fruit weight and size

Mean fruit weight, fruit length, fruit width and fruit thickness differed sig-nificantly between trees (Table 3). Fruits from Tree 1 at Ngomedzap weregreater in length than they were in width or thickness, while those of Tree 1at Nkoevos Deux were greater in width than they were in length or thickness(Table 3). The overall relationship between fruit weight and fruit lengthwas linear (r2 = 0.923), as it was in trees from Ngalan and Nkoevos Deux(Table 4). Similar relationships were found between fruit weight and fruitwidth or thickness.

52

Table 2. Effect of sample size on the mean weight and standard error of I. gabonensis fruitsfrom Tree 2 at Ngalan.

Sample size Mean fruit weight (g) Standard error Skewness

5 fruits 44.0 ± 4.7 –0.7610 fruits 44.7 ± 2.5 –0.8115 fruits 43.7 ± 2.0 –0.5020 fruits 42.7 ± 1.8 –0.2925 fruits 42.7 ± 1.6 –0.4430 fruits 43.9 ± 1.6 –0.02

Table 3. Tree-to-tree variation in fruit, nut and kernel weight (g) and size (mm) in Irvingiagabonensis.

Village Ngomedzap Nkoevos Deux Ngalan: Tree 1 Ngalan: Tree 2

Mean ± SE Mean ± SE Mean ± SE Mean ± SE

Fruit weight 103.8 1.40 58.2 2.88 36.3 2.29 43.3 1.62Fruit length 058.4 0.51 47.0 0.81 42.4 0.90 43.1 0.53Fruit width 056.6 0.30 55.6 0.99 44.2 1.09 45.4 0.53Fruit thickness 049.4 0.42 46.8 0.93 36.2 0.79 42.3 0.63Nut weight 011.9 0.55 07.7 0.42 06.0 0.36 06.0 0.20Nut length 042.0 0.61 40.2 0.86 31.7 0.71 31.7 0.39Nut width 033.3 0.74 36.9 0.74 30.0 0.94 27.9 0.38Nut thickness 022.2 0.35 19.3 0.46 19.0 0.63 19.0 0.28Kernel weight 003.5 0.14 03.1 0.19 03.3 0.30 01.8 0.12Kernel length 032.0 0.55 26.5 0.73 25.5 0.59 23.1 0.41Kernel width 021.4 0.72 22.0 0.68 21.8 1.06 18.7 0.45Kernel thickness 011.6 0.35 11.3 0.26 12.5 0.43 10.1 0.22Nut shell weight 008.5 0.19 04.6 0.24 02.6 0.53 04.2 0.14

Nut weight and size

Nut weight was significantly different between trees from the different villages(Table 3), with those from Ngomedzap being on average twice as heavy asthose from Ngalan. In general, nuts were significantly longer than they werewide, and wider than they were thick (Table 3).

Kernel weight and size

The patterns of variation in the weight of kernels were different from thoseof fruits and nuts, as the differences between kernels from Tree 1 Ngalan,Ngomedzap, and Nkoevos Deux were not statistically significant (Table 3).The greater weight of kernels from Tree 1 at Ngalan was matched by theirgreater thickness. Kernels varied most in length and least in thickness(Table 3).

Shell weight

Shell weight varied greatly between trees, with those of Ngomedzap beingthree times heavier than those from Tree 1 at Ngalan (Table 3). The shells ofthe latter had also been found to be the most brittle during depulping, with

53

Table 4.

Traits Tree Equation r2

Fruit weight v Fruit length Ngomedzap y = 0.13x + 44.80 0.135 Nkoevos Deux y = 0.26x + 31.89 0.843Ngalan – tree 1 y = 0.36x + 29.25 0.849Ngalan – tree 2 y = 0.30x + 30.22 0.807

Fruit thickness v Flesh depth Ngomedzap y = 0.27x + 1.84 0.278Nkoevos Deux y = 0.31x + 0.20 0.655Ngalan – tree 1 y = 0.22x + 2.24 0.219Ngalan – tree 2 y = 0.32x + 0.70 0.612

Fruit length v Nut length Ngomedzap y = –0.48x + 69.72 0.150Nkoevos Deux y = –0.36x + 57.05 0.116Ngalan – tree 1 y = 0.21x + 23.02 0.086Ngalan – tree 2 y = 0.01x + 31.11 0.0003

Fruit length v Kernel length Ngomedzap y = –0.40x + 55.07 0.126Nkoevos Deux y = –0.27x + 39.09 0.090Ngalan – tree 1 y = 0.11x + 21.07 0.036Ngalan – tree 2 y = 0.14x + 17.25 0.032

Fruit weight v Kernel weight Ngomedzap y = 0.04x – 0.127 0.128Nkoevos Deux y = –0.03x + 4.569 0.162Ngalan – tree 1 y = 0.01x + 2.782 0.014Ngalan – tree 2 y = 0.02x + 1.010 0.067

the shells of five of the 30 fruits of Tree 2 Ngalan breaking during thedepulping process.

Fruit flesh depth

Flesh depth was not uniform over the whole fruit. It was least in the lengthdimension of the fruits (i.e. top and bottom) and greatest in the thicknessdimension (see Figure 1 for explanation). In all dimensions, except top andbottom, flesh depth was the same on opposite sides of the fruit. Flesh depthat the top of the fruit was less that at the bottom because of the attachmentof the peduncle. Fruits from Tree 1 at Ngalan had the least flesh depth.

Relationships between flesh depth and fruit size (length/width/thickness)were found to be weak (Table 4), but strongest in the fruit thicknessdimension.

Relationships between fruit, nut and kernel traits

Nut and kernel length was shown to be virtually independent of fruit length,as was kernel weight of fruit weight (Table 4). The relationship between nutlength and kernel length was however, stronger, but still relatively weak(Table 4).

Fruit taste, colour and fibrosity

There were marked differences in fruit colour (yellow for Ngomedzap andNkoevos Deux to orange for Ngalan trees 1 and 2), fibrosity (score 1 forNgomedzap and Nkoevos Deux to 5 for Ngalan trees 1 and 2) and taste (score1 for Ngalan tree 1 to 4 for Ngomedzap and Nkoevos Deux) between trees.

Multi-trait assessment

To visualize the combination of the above traits, as a single ‘variety’ the keytraits can be expressed as a web diagram (Figure 2a), showing that in thissample of only four trees, there were differences in the relative superiority/inferiority between traits and combinations of traits.

Discussion

Prior to this study, ideas about the traits of interest for the domestication ofIrvingia gabonensis were based on descriptive accounts of the variation infruit characteristics between trees (e.g. Ladipo et al., 1996). There is still nogood assessment of the range of genetic variation in traits of likely impor-tance, although a study is in progress (Shiembo et al., 1998).

The present study establishes which of a number of possible fruit, nut and

54

55

Figure 2a. Tree-to-tree variation webs for four trees of Irvingia gabonensis in Cameroon.

Figure 2b. Fruit and nut ideotypes for Irvingia gabonensis.

kernel traits of I. gabonensis will be the most useful for future studies aimedat quantifying the extent of tree-to-tree variation in a larger and more exten-sive study of the variation in I. gabonensis populations.

It is clear from the present study, that there is considerable phenotypicvariation in almost every parameter measured and that a few traits are closelyrelated. The lack of a relationship between fruit and kernel weight and sizeis clearly very important, as it indicates that, contrary to farmers’ presump-tions, it is not possible to accurately predict the kernel traits from anobservation of the size and weight of the fruit.

With respect to the identification of desirable traits for further characteri-zation of I. gabonensis fruits, it is particularly interesting to observe thestronger relationship between flesh depth and fruit thickness, compared withfruit length or width. Future fruit measurements within on-going studies ofgenetic variation in I. gabonensis will be restricted to measurements of fruitlength and width, flesh depth in the thickness dimension, fruit, nut and kernelweights. The variation in nutshell weight found in the present study may beof special significance as kernel extraction is a labour intensive activity (about1 kg/person/hour: Acworth, unpublished), done by women. Shell brittlenesswould therefore be a socially important trait. Easily cracked nuts have beenreported twice before, once from Gabon and once from Cameroon. Howeverthis trait may be relatively uncommon, as it has not been previously observedby the ICRAF tree domestication team.

The lack of a relationship between fruit and kernel traits also suggeststhat, as in many other tree crops (Dickmann, 1985), there is the opportunityto identify a small number of key traits that together would form an ‘ideotype’that combines a number of highly desirable characteristics of potentialcommercial value. Currently, fruits of I. gabonensis, are most important as asource of kernels for use as a food thickening agent, and of lesser impor-tance as a fresh fruit. Consequently, it would be useful to identify traits thatwould contribute to the creation of both ‘fruit’ and ‘kernel’ ideotypes. Useof the above-identified traits in different combinations could achieve this(Figure 2b). In this case, the ‘fruit ideotype’ would be described by largevalues for flesh depth, fruit weight; good taste and low fibrosity, while thekernel ideotype would be described by high values for kernel weight, shellbrittleness (low shell weight) and drawability (the thickness of the glutina-ceous polysaccharide exudate formed by boiling the kernels). Obviously, asinformation becomes available these ideotypes can be refined. For example,flesh colour may become important if large sweet fruits are marketed, whileother traits associated with suitability for wine making might also becomeimportant (see Leakey, 1999). Similarly, the kernels are also a good sourceof vegetable oil and it may be that the kernel ideotype could be subdividedinto two; one rich in high quality oil and the other with desirable levels/qualityof the polysaccharide.

When the data from the present study are displayed to illustrate theircloseness towards these two ideotypes, it is interesting to observe that one

56

(Tree 1 at Ngomedzap) has some of the characteristics of the fruit ideotype,and that another (Tree 1 at Ngalan) has two of the charactistics of the kernelideotype (Figure 2b). From such a small sample of trees, this result gives muchencouragement that ideotype selection is a worthwhile approach to the domes-tication of this species. Future studies will apply the ideotype concept to muchlarger tree populations of I. gabonensis.

Acknowledgements

This publication is an output from a research project (R7190 Forestry ResearchProgramme) funded by the Department for International Development of theUnited Kingdom (DFID). The authors are indebted to DFID for funding theproject, understanding that DFID can accept no responsibility for anyinformation provided or views expressed. They also wish to thank theircollaborators in UK Overseas Development Institute (Dr Kate Schreckenbergand Miss Charlotte Boyd) and ICRAF/IRAD Agroforestry Research Networkin Cameroon and Nigeria (Dr Tony Simons, Dr Bahiru Duguma and Mr JosephKengue) for their help and encouragement.

References

Dickmann DI (1985) The ideotype concept applied to forest trees. In: Cannell MGR and JacksonJE (eds), Attributes of Trees as Crop Plants, pp 89–101. Institute of Terrestrial Ecology,Monks Wood, Huntington, PE17 2LS, UK

Franzel S, Jaenicke H and Janssen W (1996) Choosing the right trees: setting priorities formultipurpose tree improvement. Research Report 8, ISNAR, The Hague, The Netherlands,87 pp

Harris DJ (1996) A revision of the Irvingiaceae in Africa. Bulletin du Jardin Botanique Nationalde Belgique 65, 143–196

Ladipo DO, Fondoun, J-M and Ganga N (1996) Domestication of the bush mango (Irvingiaspp.) some exploitable intraspecific variations in west and central Africa. In: Domesticationand Commercialization of Non-Timber Forest Products in Agroforestry Systems, pp 193–205,Leakey RRB, Temu AB, Melnyk M and Vantomme P (eds) Non-Wood Forest Products,No. 9. FAO, Rome, Italy

Leakey RRB (1999) Potential for novel food products from agroforestry trees. Food Chemistry66, 1–14

Leakey RRB and Simons AJ (1998) The domestication and commercialization of indigenoustrees in agroforestry for the alleviation of poverty. Agroforestry Systems 38, 165–176

Leakey RRB and Izac A-MN (1996) Linkages between domestication and commercializationof non-timber forest products: implications for agroforestry, In: Domestication andCommercialization of Non-Timber Forest Products in Agroforestry Systems, pp 1–7, LeakeyRRB, Temu AB, Melnyk M and Vantomme P (eds) Non-Wood Forest Products, No. 9. FAO,Rome, Italy

Leakey RRB and Newton AC (1994) Domestication of ‘Cinderella’ species as the start of awoody-plant revolution. In: Leakey RRB and Newton AC, Tropical Trees: The Potential forDomestication and the Rebuilding of Forest Resources, pp 3–4. HMSO, London, UK

Leakey RRB and Tomich TP (1999) Domestication of tropical trees: from biology to economics

57

and policy. In: Buck LE, Lassoie JP and Fernandes ECM (eds) Agroforestry in SustainableAgricultural Systems, pp 319–338. CRC/Lewis Publishers, New York, USA

Leakey RRB, Mesén JF, Tchoundjeu Z, Longman KA, Dick JMcP, Newton AC, Matin A, GraceJ, Munro RC and Muthoka PN (1990) Low-technology techniques for the vegetativepropagation of tropical trees. Commonwealth Forestry Review 69, 247–257

Lowe AJ, Gillies ACM, Wilson J and Dawson IK (In press) Conservation genetics of bush mangofrom central/west Africa, implications from RAPD analysis. Molecular Ecology

Ndoye O, Pérez MR and Eyebe A (1997) The markets of non-timber forest products in the humidforest zone of Cameroon. Rural Development Forestry Network Paper 22c, ODI, London

Okafor JC (1974) Varietal delimitation in Irvingia gabonensis (Irvingiaceae). Nigerian Journalof Forestry 4, 80–87

Okafor JC and Lamb A (1994) Fruit trees: diversity and conservation strategies. In: Leakey RRBand Newton AC (eds) Tropical Trees: the Potential for Domestication and the Rebuilding ofForest Resources, pp 34–41. HMSO, London, UK

Sanchez PA and Leakey RRB (1997) Landuse transformation in Africa: three determinants forbalancing food security with natural resource utilization. European Journal of Agronomy 7,15–23

Shiembo PN, Newton AC and Leakey RRB (1996) Vegetative propagation of Irvingia gabo-nensis, a West African fruit tree. Forest Ecology and Management 87, 185–192

Tchoundjeu Z, Duguma B, Foudoun J-M and Kengue J (In press) Strategy for the domestica-tion of indigenous fruit trees of West Africa: case of Irvingia gabonensis in southernCameroon, In: Proceedings of 5th Annual Conference of the Cameroon Committee ofBiosciences, 17–19 December 1997, Yaoundé, Cameroon. Cameroon Journal of Biology andBiochemical Sciences

58