CHAPTER IV
IMPROVED PARKLAND MANAGEMENT

Approaches to halt and reverse land degradation and desertification in the Sahel in the 1970s and early 1980s focused primarily on externally driven forest protection and tree-planting. Indigenous forest management systems were generally assumed to be destructive. As governments' forest management approaches have become more integrated, however, and increasingly elicit local participation, the variety and value of indigenous forest management systems have gradually been recognized. Knowledge of the techniques used, and the scale, purpose, and impact of traditional management practices in the Sahel is still limited (Savenije, 1993, cited in Breman and Kessler, 1995), yet indigenous knowledge in this field is a necessary point of departure for the development of improved practices (A.S. Ouédraogo, 1995). This chapter reviews existing management practices in agroforestry parkland systems and the available research findings on promising techniques. These technologies aim at improving parkland tree density and the production of the tree component and/or crop production.

Parkland management practices

Assisted tree regeneration

Given the variable results and high cost of projects focusing on tree planting, there is increased interest in the protection and stimulation of natural tree regeneration where mother trees are available (Taylor and Rands, 1991; Montagne, 1996; Sumberg, 1990, cited in van den Beldt, 1996; Joet et al., 1998). The objective of assisted regeneration is to encourage farmers to identify, protect and stimulate the growth of naturally regenerating shrubs and trees in their fields. Young woody plants are staked (often with painted stakes so that they are easily visible) and protected from grazing, tillage and fire. If the objective is parkland formation or enrichment, woody plants will need to be protected for many years, while more limited protection is sufficient if they are being regenerated to satisfy fuel and construction wood needs.

In comparison to tree planting, for which returns may often require five to ten years, natural regeneration in dry zones of the Sahel has the advantage of providing relatively short-term (two to three years) benefits in the form of wood from pruning/thinning and an environment which is better able to sustain crop production. It is also technically easy to apply and reproduce, is relatively lowcost and requires little community organization (Taylor and Rands, 1991). Where these short-term benefits are reported, north of Maradi in central Niger, as well as further south around Mayahi, most species resprout vigorously, show rapid growth and can thus have productive and environmental functions within a few years (Joet et al., 1998). The type and quantity of wood harvested are also sufficient to meet household needs for fuel and construction. However, in neither of these two cases do regeneration techniques necessarily lead to the establishment of parklands. Project personnel observe that they are practised in drier and more marginal areas than those where parklands are generally found, or using species that are not traditionally considered as parkland species, or because the primary purpose in areas of high population density is wood and fodder production rather than intercropping.

Fig. 4.1 Vitellans paradoxa subsp. nilotica regeneration being actively protected in a house field, Adwari, Uganda
J.-M Boffa

The benefits of natural regeneration are clear. As a result of the 1981–1985 Gao Project in Dosso, Niger, Faidherbia albida densities in fields have dramatically increased from 4.9 trees/ha in 1981 to 29 trees/ha in 1992 (Montagne, 1996). As evidenced by the presence of small saplings in fields in 1992, farmers have now adopted the technique and extended the protection of regeneration to other tree species as well. The technique is also applied in other areas of Niger. As a result of improved clearing methods, densities of up to 100 shoots/ha of species including Guiera senegalensis, Combretum glutinosum and Piliostigma reticulata are maintained in the fields of several hundreds of villages around Mayahi, Niger (Joet et al., 1998). Exposure to extension work by technical services and development projects has been one of the factors determining village adoption of the technology. Protection of shrubs such as Guiera senegalensis has also been encouraged further south in Burkina Faso, on degraded land where tree cover is particularly low (Lowenberg-Deboer et al., 1994).

Farmers plant trees usually confined to the proximity of homes where tenure is more secure, soils are better fertilized and seedlings are protected from browsing, drought and fire.

In Senegal, farmers had several suggestions for promoting farmer motivation to increase F. albida regeneration. Farmers protecting regeneration could be exempted from all or part of the rural tax. Peanut seeds could be offered in exchange for tree regeneration and protection. Contests with food or cash prizes could be organized within or between villages to reward farmers with the highest rates of regeneration. Finally, one or two village farmers trained in silvicultural techniques could serve as local forestry resource people (Seyler, 1993).

Planting of parkland species

Traditionally, parkland species which easily regenerate naturally are not planted. For instance, in Dori, Burkina Faso, 90 percent of all dominant parkland trees are regenerated naturally rather than planted (ICRAF, 1996). However, the importance of planting as a technique for parkland establishment and regeneration depends on parkland type. Borassus aethiopum parklands such as those of Wolokonto, in southwestern Burkina Faso, are believed to have been planted originally, even though the species is locally present in natural stands. Planting is still actively practised every two or three years in order to replace trees which do not survive tapping (Niang, 1975), and even results in increased density and spatial extension of parklands. Fallen Borassus fruits are gathered and heaped in fields, and sprouted seeds are then planted, often in lines for easier mechanized cultivation and for plot delineation (Cassou et al., 1997). On the Seno plain of Dogon country in Mali, Adansonia digitata trees are planted in compounds and nurtured before they are transplanted along the edge of cultivated fields (Sidibé et al., 1996).

In addition, numerous references indicate that farmers actively regenerate tree species when the benefits of their investment are guaranteed, as a response to land degradation and the rarefaction of trees in their agricultural landscape, or because of the subsistence and economic value they provide. Deliberate tree planting takes place mostly in the vicinity of family compounds, where browsing of animals, fire and harvesting can be controlled and where the use of household refuse, animal manure, and plant residues contributes to higher soil fertility than in distant fields. In many cases farmers prefer to plant exotic tree species because of social customs and the long time indigenous species take to reach maturity and yield expected benefits. Generally they prefer to rely on wild trees to meet their local species needs, although there are local variations.

In Petit Samba, Burkina Faso, all male interviewees declared that they planted exotic tree species, while only a few transplanted indigenous Vitellaria paradoxa, A. digitata and Parkia biglobosa wildings from under mother trees to compound areas where they could be watered and protected (Gijsbers et al., 1994). Around 85 percent of compounds and 16 percent of (non-compound) fields surveyed in 21 villages of central Mali had planted trees, while the non-compound figure was 38 percent in six villages of the Bandiagara area. In compounds, the primary species was the exotic Azadirachta indica (neem) while farmers planted (or transplanted) B. aethiopum, A. digitata, F. albida and neem in fields (McLain, 1991a). In Houet, Burkina Faso, and Benin, P. biglobosa is sometimes sown at the same time as crops (A.S. Ouédraogo, 1995). In Senegal, farmers often plant neem, mango and B. aethiopum on field borders or in fields next to family compounds (Seyler, 1993). Kessler and Boni (1991) report that farmers plant several tree species in compound fields in Burkina Faso, for example P. biglobosa in Sissili, A. digitata in Kossi, B. aethiopum in Comoé and H. thebaica in Seno. Vitellaria paradoxa is virtually never planted except in rare cases such as at Kokologho, Burkina Faso. The practice of growing neem to produce roofing poles has become widely established in northern Ghana over the last 20 years on farmer initiative (Norton, 1987, in Shepherd, 1992).

Fig. 4.2 Expansion of Borassus aethiopum parkland through active regeneration (foreground)
R. Faidutt

In Nigeria, neem and Eucalyptus are planted in parklands of Katsina State (Otegbeye and Olukosi, 1993). In Sotoko, planting of P. biglobosa was practised by 30 percent of farmers interviewed (Popoola and Maisanu, 1995). In the densely populated Close-Settled Zone of Kano, 71 percent of respondents of all ages reported planting trees in farmed parklands. Species included mango, P.biglobosa, Ficus thonningii, neem, Ceiba pentandra and F. albida in descending order of importance (Cline-Cole et al., 1990). Montagne (1986) presents numerous local instances of nursery creation, multi-purpose tree plantations, fallow enrichment by direct seeding and marketing of seedlings in Niger as a result of individual initiatives. The establishment of fruit species in groves is also often reported. In Wolokonto, southwestern Burkina Faso, mango, citrus, guava and cashew trees have traditionally been planted in small clumps next to farmer compounds, and this is now also practised further out in parklands (Cassou et al., 1997).

In response to drought and the decline of aquifer levels, farmers in the Peanut Basin of Senegal have tested a method for successfully establishing mango trees on Dek Dior soils. They introduce a mango seedling in a planting hole where a young B. aethiopum individual is already growing. Thus, the water that the latter species is able to extract at certain depths will benefit the growing mango seedling. The size of the selected Borassus plant depends on production objectives. If grown for its palm fronds, a small plant is chosen which will soon be stifled by the developing mango. If fruit and wood production are intended, a larger plant is selected which will successfully co-develop with the mango tree (Freudenberger, 1993b).

Regeneration can also result from indirect human intervention. It appears that various stands of P. biglobosa result from seeds deposited by soldiers, porters and forced labourers after eating them. Such is the case of stands in Noumoudara and Koloko (Kenedougou) in Burkina Faso, which are attributed to the army of Samory Traoré, or those in villages such as Koutoura, Nianaba and Bonyolo, which are thought to have been created by the forced labour involved in the construction of the Ouagadougou-Abidjan railway (A.S. Ouédraogo, 1995; Ki, 1994).

Where appropriate, planting of parkland trees should be encouraged. Variable survival and growth performance of trees in plantations are sometimes primarily related to soil and less to genetic variability as found in an F. albida trial in Niger (Geiger et al., 1994). In order to reduce plantation costs, increase success rates, and ensure good productivity of crops under trees, planting of parkland trees should be done selectively on high fertility microsites. These favourable microsites can be identified with farmers through growth observations of a cereal crop in the previous season and seedlings planted predominantly on these sites (van den Beldt, 1996).

Improved fallows

Improved fallows, whereby economically useful and fertility-improving trees are planted before cropping is discontinued, appears to be a promising alternative to long natural fallows.

Mineral fertilization alone is not sufficient to sustain crop yields in savanna soils, because of the degradation of physical, chemical and biological properties associated with the decline in soil organic matter content (Piéri, 1989). Fallowing therefore appears to be a necessary practice and is one of the few means available to subsistence farmers to maintain satisfactory physical and chemical soil fertility. In areas of high population pressure, however, fallows are increasingly being abandoned or reduced in length. Improved fallows aimed at the biological recycling of mineral elements and the build-up of soil organic matter seem to provide a potential alternative to long natural fallows (Harmand, 1998).

In this technique, trees are planted a few years before fields are fallowed in order to give seedlings a competitive advantage over other regenerating plant species, while not competing severely with crops. Protection from grazing and fire (and from plant competition) is necessary in fallow enrichment areas. Depommier and Fernandes (1985), for example, mention that P.biglobosa is one of the few species which is sometimes sown directly at the end of a rotation in order to enhance the subsequent fallow. Seignobos (1982) also mentions that in Cameroon F.albida is cut back every two or three years and maintained as a shrub in order to shorten fallow duration. To ensure adoption by farmers, improved fallows need to be able to generate useful products (wood, gums, fruits, fodder, etc.) during the fallow time and restore soil fertility in a shorter time than natural fallows.

In northern Cameroon, Harmand and Njiti (1998) found that Eucalyptus camaldulensis was not appropriate for short enriched fallows, as soil characteristics were found to deteriorate with poor incorporation of organic matter into the soil and lower soil porosity. In contrast, herbaceous fallow and Senna siamea had a positive effect on soil carbon, but only Acacia polyacantha, a nitrogen fixing species, led to a marked improvement in carbon and nitrogen content six years after planting (four-year fallow). The capacity of A. polyacantha to store nitrogen which is mineralized more rapidly in soil organic matter and root biomass was reflected in higher yields of subsequent crops than in other fallow types. However, the absence of a visible effect on cation exchange capacity after four years' fallow shows that fertility improvement during fallow is slow. Harmand and Njiti suggest that Senna siamea fallows should be converted to cultivation without the use of fire in order to conserve the system's nitrogen and that the selection of a few coppiced stems could ensure shoots for a subsequent fallow cycle. Pruning of A. polyacantha maintained in fields would reduce shade and root competition with crops while providing farmers with all-purpose wood. Acacia senegal is another N-fixing candidate for improved fallow which is well adapted to a variety of Sudanian sites and has the advantage of generating the economic product, gum arabic, after four years.

Improved fallows apparently have good ecological and economic potential in the search for more productive and sustainable agroforestry parkland systems. Additional research is necessary to better define the processes and conditions (including species) under which improved fallows can accelerate soil fertility restoration. The integration of various forms of this technology in farming systems should also be experimented with, and management prescriptions are needed which focus on optimizing annual and tree crop production.

Fire protection

Fire in parklands can damage parkland trees directly, reduce wood and fruit production and affect the tree regeneration and fertility restoration potential of fallows. Fires in the uncultivated bush or fallows may occur at any time during the dry season and be very hot due to the presence of large amounts of fuel. In fields, however, controlled burning at the start of the dry season may be less damaging because fuel amounts are relatively low. In the Bassila region of Benin, significantly more trees were burned in the bush than in fields, although there was no significant difference in the timing of fires (Schreckenberg, 1996).

Fig. 4.3 Naturally regenerated stand of Vitellaria paradoxa following protection from fire and cultivation, Tolon, Ghana. P. Lovett

Fire tolerance in Vitellaria parklands is ensured by the thick bark as well as its cryptogeal germination system whereby the plumular region of the seedling is below the ground surface (Jackson, 1968). The removal of associated species by fire also influences the subsequent seed germination of Vitellaria individuals positively. Nevertheless, Hall et al. (1996) reviewed evidence indicating that, despite its fire tolerance, fire protection is beneficial to Vitellaria stands provided that plant competition is eliminated. When fire was excluded from areas climatically suitable for closed forest, Vitellaria regeneration was reduced. In contrast, in drier conditions V. paradoxa stands regenerated more completely in the 20 years following clearance when fires were excluded than under an annual burning regime. Early burning in the dry season was also less detrimental to regeneration than later burning. In the Anara Forest Reserve of Nigeria, growth was vigorous in pollarded Vitellaria trees subject to fire, while coppicing was poor in fire-protected plots (Onochie, 1961, cited in Osei-Amaning, 1996).

Fire also has a significant effect on the time to first fruiting as well as annual fruit production. In the absence of fire, Vitellaria trees start producing fruit after 15 to 20 years, but fructification will take twice as long for trees which are frequently burned (Delolme, 1947). In the high Niger Valley, early fires are used to stimulate fruit production in Vitellaria if it stops bearing fruit for two to four years (Maguiragua, 1993, cited in Cissé, 1995). Farmers sometimes say that fire can stimulate the upward flow of sap when it is well dosed. According to Ruyssen (1957) and Hopkins (1963, cited in Hall et al., 1996), flowering and leaf flush come earlier in burned than unburned trees, yet this does not necessarily imply higher fruit production. Similarly, fire had a positive effect on flowering in the Bassila region of Benin, although the average estimated fruit yields of burned and unburned trees were not significantly different (Schreckenberg, 1996). However, intense fires can also devastate fruit production. The timing of fires appears to be very important in determining their effect on fruit production. As trees are due to come into flower or have started flowering, late fires (after late January) can generally annihilate fruit yields, while there is no yield difference between sites burned early and those protected from fire, as observed in Bondoukui-Bavouhoun, western Burkina Faso (Serpantié, 1997a). People in northern Ghana also prefer early burning in order to eliminate the risk of fire during flowering (Hall et al., 1996). Deposition of soot and ash on stigmatic surfaces may also impair germination and functioning of pollen grains (Osei-Amaning, 1996). Fire exclusion associated with weeding resulted in higher fruit production than a fire treatment to eliminate undergrowth (Adomako, 1985).

Parkia biglobosa is also referred to as a fire-resistant species, yet no conclusive evidence exists showing that fire has a significant effect on survival and reproduction (Hall et al., 1997). In Bassila, Benin, heavy fire damage was associated with no flowering at all, while light burning after bud formation did not have a negative impact on fruit production (Schreckenberg, 1996). Local farmers believed that P. biglobosa as well as Tamarindus indica produce better in the total absence of fire. In Burkina Faso, Parkia trees are often protected from bush fires by removing fuel, flattening the grass and establishing belts of green vegetation around the trees (Ki, 1994). For instance, Gourmantché labourers may be paid to cultivate a piece of land around stands of trees which will serve as a fire-break or simply to clear the area around individual trees. The Gourmantché, Lobi and Bobo practise early burning and flatten the grass to keep flame height low. The use of green vegetation belts around trees is also practised among Gourounsi farmers. Although more intensively used for stands of P. biglobosa, some of them are also applied to individual trees.

The immediate effects of burning on top-soil nutrient availability are generally positive, but nutrient loss through volatilization, leaching and wind and water erosion can also prevail. Fire control leads to more nutrients being stored in the woody biomass than in the soil, and reduces mineral cycling. It contributes to an increase of woody plant diversity, canopy cover and productivity in the Sudan zone (Breman and Kessler, 1995). Bush plots protected from fire for three years produced over twice as much woody biomass as unprotected plots in the Sudan zone of Cameroon (Peltier and Eyog-Matig, 1989). Burning was recommended after an Eucalyptus camaldulensis fallow in Cameroon to compensate for poor organic matter incorporation into the soil (Harmand, 1998). However, fire did not significantly improve soil chemical properties after Senna siamea and Acacia polyacantha fallows, which are characterized by higher external recycling of nitrogen and mineral elements. Fire can be beneficial on nutrient-deficient ferralitic soils in humid zones where minerals are immobilized in above-ground plant compartments. Breman and Kessler (1995) recommend that burning should be avoided as much as possible in the Sahel region. Instead, field clearance residues should be spread over the fields as mulch after the largest woody parts have been removed.

Silvicultural techniques

To increase production of parkland trees, sahelian farmers commonly protect, fence and water seedlings, select vigorous shoots and prune trees.

After their selection and protection, the development of parkland trees is aided by traditional silvicultural techniques. Farmers usually weed the area around chosen individuals. Young P. biglobosa trees are generally protected from animals by a barrier of branches, straw or thorns in Burkina Faso (Ki, 1994). Planted or valuable naturally regenerated trees in the residential area can also be watered. Close to half of 230 farmers interviewed in central Mali reported that they practise one or more tree maintenance techniques including pruning, fencing, watering and manuring (McLain, 1991a). To ensure rapid F. albida seedling establishment and growth to a desired shape, farmers will clean small saplings of branches which could pull them down (Montagne, 1984). In Burkina Faso, Janodet (1990) reported that some farmers cut the seedlings to the ground in the first year and cut all but the most vigorous stem in the second year in order to develop a strong taproot. Faidherbia albida shoots can be very bushy in parklands because seedlings or suckers are repeatedly browsed by livestock in the dry season and even cut back annually by farmers during the rainy season. In the meantime, the root system develops vertically until it reaches the aquifer, and a selected shoot will grow rapidly above ground past browsing height if deliberately protected. In contrast, F. albida has an ascending tree architecture in plantations (Peltier's note, in Seignobos, 1996). Seignobos (1982) also noted that where high densities of bushy F. albida occur, farmers can easily establish parklands by selecting a single branch on these trees if land for cultivation is needed.

Pruning/debranching

The purpose of pruning (cutting back certain branches) may include wood, fodder and mulch production, improved fruit production, reduction of shade on understorey crops, longer tree lifespan, as well as control of parasitic plants such as Tapinanthus spp. in affected species. However, as discussed in Chapter 5, these benefits are often not recognized by outsiders, including foresters.

Pruning of P. biglobosa is common in Nigeria (Tomlinson et al., 1995) and was practised on 10 to 30 percent of all Parkia trees in 21 villages on a north-south transect in central Burkina Faso (Timmer et al., 1996). Its main purpose now in Burkina Faso is to improve tree survival and productivity especially in the northern sites, whereas its primary function used to be for wood exploitation and to reduce the tree's influence on the environment. This indicates that Parkia fruits themselves represent an increasingly important resource worth managing for. Most pruning activities take place in April and May rather than after the rainy season on trees older than 30 years. Light pruning (removing one or several main or secondary branches) was as frequent as intensive pruning (removing most or all main branches). With increasing tree age farmers tend to prune trees more intensively and more often, with the aim of rejuvenating them and reducing shade. Pruning is also more intense in village fields and valleys than in bush fields and slopes, probably because trees are better protected and manured, or produce more fruit there. The fear of forestry agent sanctions, the time lag needed (at least three years) for enhanced fructification, the lessened tree resilience in drought years, and potential tree tenure conflicts were the primary reasons for not pruning more extensively.

Lopping (pruning of smaller branches and twigs, often for fodder) is commonly practised in F. albida parklands which include a pastoral component. In Watinoma and Dossi, Burkina Faso, two villages with high and low livestock and herder population densities, respectively, lopping was practised on 50 percent and 25–30 percent of F. albida trees (Depommier and Guérin, 1996). Excluding small diameter trees, which are generally not pruned, pruning frequency on larger trees was 70–80 percent in both villages. Pruning intensity, or canopy reduction, increased with higher grazing pressure as well as with increasing proximity of residence. In Sob, Senegal, F. albida canopies had been heavily reduced between 1965 and 1985 by 38 percent on average (13 to 66 percent) (Louppe et al., 1996). They are also heavily lopped in Sudan (Miehe, 1986). In central Burkina Faso, pruning takes place mostly during the second half of the dry season, culminating in March-April when fodder is most needed. After an early dry season operation, farmers/herders may prune large trees again at the end of the dry season to harvest the very nutritious and palatable new shoots.

Repeated pruning of F. albida tends to stimulate leaf production. In Burkina Faso, leaf regrowth one year after total pruning, which was correlated with canopy size, was complete where trees were regularly pruned but reached only 60–80 percent of previous leaf biomass where pruning was not practised systematically (Depommier and Guérin, 1996). Pruning resulted in an increased ratio of leaf biomass to total biomass and a canopy area reduction of 10–40 percent. Regular pruning also appeared to increase canopy regrowth capacity as shoot diameters were up to twice as large where it was practised. In Burkina Faso, pruning resulted in a two to three month delay of F. albida phenology. When it was carried out in the second half of the dry season, leaves appeared at the start of the rainy season, fell in the second half and grew again in the following dry season. Pruning also induces tree rejuvenation and causes older spineless trees to grow spines. Overall, F. albida responds very well to pruning when this is not excessive. Cissé (1984) recommended a single pruning operation early in the season. Contrary to its effect on leaf biomass, however, pruning has a drastic negative effect on pod production. This was reduced by up to ten times between 1994 and following intense pruning in 1995 in Burkina Faso.

Fig. 4.4 Lopping of Faidherbia albida, Ngaparou, Senegal.
P. Danthu

Pruning of V. paradoxa is less common than that of F. albida and P. biglobosa. Vitellaria's slow growth rates are reviewed by Hall et al. (1996). Pruning is considered to be either not suitable (Kessler, 1992) or to have uncertain effects (Kater et al., 1992). Farmers claim that pruning will reduce fruit yields (Kessler, 1992). However, some large individuals in Nigeria respond very vigorously to pruning (Brun, 1996). The species also responds well to coppicing. Individuals of 15–30 cm diameter showed vigorous stump shoots (Chevalier, 1948). In southern Burkina Faso, lower branches up to 15 cm diameter had been cut off at their base in 56 percent of a large sample of V. paradoxa trees. This was done in order to allow manoeuvring of draft animals and/or improve tree form for the production of understorey crops rather than to stimulate tree growth or fruit production (Boffa, 1995). Bagnoud et al. (1995a) also report that the lower branches of Vitellaria and Parkia trees are pruned every two to four years to form ascending crowns in young trees and to afford more light to crops in older ones.

When growing on croplands, Detarium microcarpum is occasionally pruned to reduce shade over crops and to promote fruit production (Wiersum and Slingerland, 1997). Farmers in Ndam Mor Fademba, Senegal, also report that they prune Combretum glutinosum and Guiera senegalensis shrubs to hasten tree regeneration for firewood collection or to enhance agroforestry associations (Schoonmaker and Freudenberger, 1992). In northern Cameroon, Prosopis africana is systematically pruned in parklands before the rainy season in order to allow more light penetration for crops underneath. Excessive pruning intensity may be one of the causes of parkland degradation in this area (Bernard, 1996). Hyphaene thebaica and B. aethiopum are also lopped to supply palm fronds (Seignobos, 1982). Finally, Quercus rotundifolia, Quercus suber and Quercus faginea trees in the dehesa agroforestry systems of Spain are often pruned for improved acorn and wood production (Joffre et al., 1988).

Research on Tapinanthus species, epiphytic plant parasites of V. paradoxa as well as a variety of other parkland species, shows that pruning is currently the quickest and cheapest way to rid trees of this threatening pest. Tree limbs affected by the parasite should be cut above the infected site in order to eliminate the parasite's whole absorption system (Boussim et al., 1993b). Given the high proportion of trees affected and the wide zone of occurrence of Tapinanthus species, such measures are urgently needed and should be widely recommended by extension services.

Ringing

Ringing is another traditional practice used for stimulating fruit and seed production. It was observed in several instances on P. biglobosa in Burkina Faso (A.S. Ouédraogo, 1995; van der Vleuten, 1995, cited in Wiersum and Slingerland, 1997). When optimally done, a shallow 10 cm-wide ring of bark is cut from the trunk at breast height just before flowering. This technique is used on Parkia trees which farmers identify as producing no fruit or producing seedless fruit, and can also be combined with pruning. It is assumed that ringing favours flowering and fruit production to the detriment of vegetative production by blocking the transport of assimilates or changing the concentration of gibberelline growth hormones (Tromp et al., 1976; and Costes, 1983 cited in Wiersum and Slingerland, 1997). In the high Niger Valley farmers also practise 80 cm longitudinal cuts on A. digitata and introduce rock salt in order to stimulate growth (Maguiragua, 1993, cited in Cissé, 1995).

Coppicing and pollarding

Depommier and Fernandes (1985) report that parkland species (no particular species mentioned) in landscapes dominated by Vitellaria and Parkia species in the Central African Republic can be coppiced (cut at the base to encourage bushy shoot regrowth) and pollarded (cut above grazing height to encourage shoot regrowth) to limit crop yield depression or for gathering wood and other tree products, or even to provide support for yam vines. Trees in fields left to fallow are also managed for fuelwood production by coppicing. Grigsby and Force (1993) state that in the Upper Niger River Valley region of Mali, women coppice regenerating trees on three to five year rotations to obtain diameters equivalent to the size of a fist. Two or three rotations of fuelwood can be produced in the course of a fallow period allowing the restoration of soil fertility. Neem (Azadirachta indica) is commonly coppiced every year before the rainy season; it is used for fuel and construction wood and the leaves and smaller branches are sometimes used for mulching (Yelemou et al., 1993). In alley farming, sorghum yields were highest close to trees coppiced early and lowest close to trees coppiced late (Tilander et al., 1995). Detarium microcarpum is actively managed for production of wood which is particularly valued for construction purposes (van der Vleuten, 1995, cited in Wiersum and Slingerland, 1997). For this species, farmers assist in the growth of a single vertical shoot by cutting the others and regularly stripping the plant of lower leaves, side branches and other shoots. Once they reach usable size, Detarium plants can be coppiced repeatedly for several years. Pollarding of Ceiba pentandra was also reported in parklands outside the city of Zaria, Nigeria (Pullan, 1974). Faidherbia albida parklands of Masa country in northern Cameroon are associated with a layer of systematically pollarded Ziziphus mauritiana. Harvested wood and branches are used as construction poles and as material for fences erected to surround compound fields and channel livestock movement (Seignobos, 1996).

Fig. 4.5 Pollarded Azadirachta indica among Faidherbia albida trees in village fields of Dissin, Yoba Province, Burkina Faso.
S.J. Ouédraogo

Fig. 4.6 Faidherbia albida parkland with substratum of Ziziphus mauritiana trees, Yagoua, Cameroon.
C. Bernard

Tree fertilization

One could expect a similar positive effect of organic and/or chemical fertilizers on tree (wood and fruit) production to that observed on domesticated orchard species. Currently, priority use of available fertilizer resources is usually directed to crop rather than tree yield improvement. Little information exists therefore on this topic. Nevertheless, Simond (1930, cited in Hall et al., 1996) mentioned that the use of 5t/ha of cattle manure increased Vitellaria fruit yields by an average of 38.6 kg per tree. Hall et al. (1996) also cited a study conducted by Osei-Bonsu (1991) in which seedlings in tilled plots displayed better height (but not diameter) performance (mean height 6.28 cm) than those in zero-tilled plots (mean height 4.12 cm), but which was not visible before the third year of the experiment. The excavation of a space between roots of P. biglobosa to stock water during the rainy season was observed among the Gourounsi in Burkina Faso. This is done to remedy possible rainfall insufficiencies (Ki, 1994). The same author encountered indigenous pest-control operations in which a sharp object is introduced in sites of worm infection. A few farmers also used toxic products against insect attacks.

Manuring and application of rock phosphate in conjunction with mycorrhizal inoculation can benefit tree establishment and production.

Because of the common phosphorus deficiency of West African soils, research has also focused on the use of phosphorus fertilizers in conjunction with endomycorrhizae for better tree development. By expanding the volume of explored soil, infection with endomycorrhizal fungi stimulates F. albida growth (Ducousso and Colonna, 1992). Bâ et al. (1996) showed that non-inoculated F. albida seedlings can make direct use of Burkina Phosphate, a locally available natural tricalcic rock phosphate. However, inoculation with the endomycorrhizal fungus, Glomus aggregatum, significantly increases the efficiency of phosphate use. Inoculated seedlings displayed increased growth, phosphorus and nitrogen content, and a lower root/shoot ratio. Increased doses of rock phosphate increased phosphorus content in stems of inoculated seedlings but had no effect on their height.

Management techniques for improved crop production

This section focuses on tree-based techniques for improving crop production. Numerous references to soil and water conservation techniques (Reij et al., 1996), such as earth and rock bunds (Rochette, 1989), grass strips (Renard and van den Beldt, 1990) and zai (Lowenberg-Deboer et al., 1994) can be found elsewhere.

Pruning and coppicing

The possibility of increasing solar irradiation and taking advantage of nutrient enrichment around trees is a strong argument in favour of pruning parkland trees. Pruning P. biglobosa resulted in higher sorghum yields relative to unpruned trees (Kessler, 1992). In Senegal, pruning inverted groundnut productivity trends with increasing distance from Cordyla pinnata trees. Total biomass was 1 486, 2 084 and 2 110 g/m² around unpruned trees, while yields were 2 177, 1 829 and 1 787 g/m² for pruned trees (Samba, 1997). The same patterns were observed on Azadirachta indica in Saria, Burkina Faso. Sorghum grain and straw yields under the canopy of pruned trees were higher than at the distances of 1 and 3 radii, and yields under pruned canopies were higher than under unpruned trees (Zoungrana et al., 1993). Coppicing resulted in similar and significant grain yield increases around A. indica. Yields were 48 and 33 percent higher under and directly outside canopies (coppiced before planting) than in the open field (Tilander et al., 1995). Improved environmental conditions would appear to outweigh below-ground competition, which is possibly intensified with root morphology changes resulting from pruning (van Noordwijk et al., 1991, cited in Rhoades, 1995).

The effect of pruning F. albida on understory crop production has not been quantified so far. If nutrient inputs from leaf litter are primary factors for the ‘F. albida’ effect, one would expect repeated pruning to reduce or cancel this effect on crop productivity when compared with unpruned trees. According to Louppe (1990, cited in Depommier et al., 1992), pruning F. albida has a negative impact on crop production underneath. However, yield improvements have also been observed in the proximity of pruned F. albida trees (Depommier and Guérin, 1996). These authors suggest that the effect of pruning would need to be verified on parkland trees which have not been subject to pruning since establishment. This may be accomplished better on station than on farm.

The above data indicate that pruning offers good potential for improving crop yields under trees which would normally reduce them. At the same time, if carried out in moderation, pruning of F. albida need not necessarily reduce the tree's benefits too greatly. Data reviewed in earlier sections also suggest that the proportion of trees subject to this practice is moderate in the region, especially for species other than F. albida, and could therefore be extended. However, little is known of the influence of annual pruning on nutrient cycling and fertility of subcanopy soils, as well as crop performance, especially in the long term.

Organic inputs are necessary to maintain soil fertility in semi-arid West Africa. Positive results have been achieved with application of Azadirachta indica or Acacia lebbek leaf mulch.

Research addressing these issues will be instrumental in establishing precise recommendations for appropriate pruning intensity, frequency, and time according to species, tree age, latitude and rainfall conditions. Such studies should result from additional experiments as well as thorough investigation of indigenous knowledge in this area. As pruning may not only improve crop yields but also reduce potential fruit, leaf and wood production, studies need to investigate both these aspects in order to provide farmers with management prescriptions best suited to their production objectives.

Organic fertilization and mulching

Poor soil fertility is a primary constraint limiting crop production in semi-arid West Africa, even outweighing moisture deficiency in the long run (Bationo and Mokwunye, 1991a). In this area, the purchase of mineral fertilizers is generally limited in subsistence-oriented farming systems, leading to soil degradation through a decline in exchangeable bases, acidification and aluminium toxicity when used alone (Piéri, 1989). Research has therefore stressed the benefits of various organic additions such as crop residues and the applications of straw, compost and manure to increase the fertility of agricultural soils (Bationo and Mokwunye, 1991b). The use of manure and compost is, however, limited by livestock numbers and straw availability, and could be complemented with tree mulch.

Some years ago the use of organic material (household refuse, crop residues, animal manure) tended to be restricted to compound fields. More recently, farmers appear to be managing a larger area of more distant fields in a more intensive manner including the application of organic and chemical fertilizers (Vimbamba, 1995; Seyler, 1993). Some degree of manure/compost use is reported in most studies, e.g. in Vitellaria parklands closest to settlements in Mali (Baumer, 1994), in F. albida parklands in Burkina Faso (Depommier et al., 1992), in F. albida and Adansonia parklands north of Kano, Nigeria (Pullan, 1974), etc. It is transported to the field, dumped in piles and then spread evenly over an area with a radius of a few metres. In Senegal, manure is applied not at random but particularly in areas deemed infertile, and not under and around F. albida trees where production is favourable (Seyler, 1993). In areas where sources of organic matter have become scarce, farmers resort to collecting grass during the dry season which they spread in fields as mulch (Lowenberg-Deboer et al., 1994).

The application of tree mulch in crop fields is recognized as another way of maintaining or enhancing soil fertility in these systems. In semi-arid Burkina Faso, leaf mulch from Azadirachta indica (3.7 t dry matter/ha) or Albizia lebbeck (2.7 t dry matter/ha), corresponding to 75 kg nitrogen/ha enhanced sorghum yield up to fourfold (Tilander, 1993). The mulch effect on yields increased over the three years of application and was more pronounced with higher dosage. Smaller leaf quantities (25 to 50 kg nitrogen/ha), corresponding to the production of an alley-farmed field, also had a significant positive influence on yields in most cases. In addition, yields monitored on an annual basis were not sensitive to mulch composition (A. indica, A. indica + straw, or A. lebbeck), but over the three-year period A. indica leaves were superior to the two other mulch types. Timing of application also influenced yields in a way which seemed to vary according to rainfall distribution. In Malawi, highest quantities of inorganic nitrogen derived from annual F. albida litter and root inputs became available with the first rains. However, Rhoades (1995) noted that this period of nitrogen availability occurred before full crop root development and that the highly mobile nitrates may have leached below the crop root zone, thus bypassing crops completely. He therefore recommended the mixing of easily decomposed and nutrient-rich litter with resistant (high lignin content) plant material to favour a more gradual nutrient release.

Tilander also showed that A. indica mulch contained a higher amount of nutrients and showed release rates higher than (for nitrogen, calcium and magnesium) or similar to (for potassium) Acacia holocericea leaves (Tilander, 1996). Both nutrient-rich (A. indica leaves and A. indica leaves + compost) and nutrient-poorer mulches (wild grass and A. holocericea phyllodes) resulted in significant water conservation and temperature reduction. However, there were higher yields in nutrient-rich (A. indica leaves, A. indica leaves + compost, and compost alone) treatments than in other plots with nutrient-poor mulches. Highest yields were achieved with mulches combining high nutrient delivery as well as moisture conservation and temperature reduction. In the range of mulches tested, A. indica leaves performed best.

Fig. 4.7 Manure or compost is heaped and will be spread uniformly throughout the fields.
R. Faidutti

Samba (1997) reported a nursery experiment in which the application of 39, 78 and 156 kg of Cordyla pinnata litter/t of soil increased total millet biomass 17, 1.9 and 1.5 times over the control, but resulted in total peanut biomass decline of 11, 13 and 29 percent, respectively. According to Cissé (1995), leaves of V. paradoxa, F. albida, Khaya senegalensis, Daniella olivieri, Isoberlinia doka, Pterocarpus erinaceus and Afzelia africana have a fertilizing effect. They are harvested and composted with millet residues and manure. Foliar biomass production for V. paradoxa (mean dbh=42 cm) and Bombax costatum (mean dbh=39 cm) in parklands was estimated from litter sampling at 29.5 and 23 kg/tree, respectively, in southern Burkina Faso (Bambara, 1993). An estimate of Cordyla pinnata leaf production in parklands of the southern part of the Peanut Basin of Senegal was 337 kg/ha (Samba, 1997). Assessments of foliage production reviewed in Breman and Kessler (1995), and considered in Chapter 6, mostly focus on woody species of the Sahel zone of West Africa. In the same reference, nitrogen and phosphorus foliage concentrations are tabulated for a larger spectrum of woody species.

Preliminary data suggest that tree mulch can contribute to sustained crop production in agroforestry parklands, but that this requires more intensive tree management than commonly found on-farm. Additional research in this area should assess what amounts of nutrients can be expected from parkland leaf mulch according to species and density, as well as defining for an extended number of parkland species how (application timing, short-and long-term effects, etc.) application of leaf mulch contributes to soil protection and productivity goals. These findings could eventually lead to technical prescriptions of intensified parkland management including pruning regimes and optimal tree densities according to climatic zones and soil type. Of similar importance, the costs, benefits, and practical feasibility, in terms of labour and access to tree resources, of tree-harvesting and mulching practices also need to be considered.

Tree/crop associations

Higher soil fertility and improved microclimatic conditions (see previous chapter) under F. albida canopies present a more favourable agroecological environment for crop growth. Williams (1992) hypothesized that the conditions created by F. albida allow the planting of (cash) crops not otherwise common in the Sahel because of their lack of adaptation to high temperatures and low soil fertility. Gains from intensive cropping around trees would promote farmer interest in intensive tree management. Trials with maize and cotton under F. albida have been conducted in Sadoré by ICRISAT (van den Beldt, 1996). Dancette (1968) suggested that fields could be laid out in between F. albida rows, with cereals grown in tree rows and groundnuts in the alleys. Farmers do allocate specific crops to sub-canopy environments. Aware of the more pronounced effect of F. albida on millet than on groundnut production, farmers traditionally cultivated millet under F. albida even in groundnut fields in Sob, Senegal (Louppe et al., 1996). Where millet is a regular field crop, higher priced crops such as tobacco and sorghum are sown under F. albida canopies in the Madaroumfa area of Niger (Montagne, 1996).

Fig. 4.8 Association of a local tuber crop, ‘fabirama’, with a Parkia biglobosa canopy, Thiougou, Burkina Faso.
J.-M. Boffa

The sowing of crops or crop mixtures demanding higher soil fertility and moisture or tolerating shade in the proximity of Vitellaria and Parkia trees has also been recommended (Kessler, 1992; Kater et al., 1992). Farmers often choose to grow tobacco, cassava, yam, sweet potato and large-leafed vegetables under these trees (Teklehaimanot et al., 1997; Kessler and Boni, 1991; Janodet, 1990; Wiersum and Slingerland, 1997). Besides making more efficient use of microenvironmental conditions, farmers diversify resources and reduce the risks of crop failure through such tree-crop associations.

However, these crops are not cultivated on a large scale, because of low demand and time-consuming weeding requirements (Kessler, 1992). Although this technology may not be uniformly applicable on a large scale, research aimed at identifying optimal crop-tree combinations and the technical and socio-economic conditions required for their adoption is necessary.

Genetic improvement of parkland species

Planting trees is still a limited practice in the Sahel. In addition to a number of possible social and economic factors (discussed in later chapters) farmers have apparently been discouraged from doing so by several biophysical characteristics of indigenous tree species. Parkland species may have a slow growth rate, and a long juvenile phase before fructification. Thus, farmers feel that they will not benefit from planted trees during their lifetime and would rather rely on trees occurring naturally. In addition, because parkland trees are wild, average tree production is low and production of pods/fruits is highly variable from one individual to another and from year to year. For instance, pod/fruit production varies by a factor of 10 between consecutive years and between trees in F. albida (Depommier, 1996b), and by a factor of 5 between an average and best producing V. paradoxa (Boffa et al., 1996a) in Burkina Faso.

Fig. 4.9 Rooted hardwood cutting of Vitellaria paradoxa with new flower bud formation. P. Lovett	Fig. 4.10 ‘Timber’ variety of Vitellaria paradoxa growing in cleared woodland, Bomburi, Ghana. P. Lovett

Recently, there has been an increasing interest in domestication of agroforestry trees (Leakey and Newton, 1994). Domestication is defined as human-induced change in the genetics of a plant to bring it into wider cultivation through a farmer-driven or market-led process (Harlan, 1975; Simons, 1997b). Resulting in the delivery of improved trees, it has the potential to increase farmer interest in managing (cultivating) parkland trees more intensively as well as contributing to the productivity and sustainability of these systems. However, there is a limited understanding of the sort and quantities of tree germplasm needed as well as its impact on the adoption of parkland agroforestry practices. A major challenge in realizing the potential held in the domestication process depends on whether the resulting improved germplasm fulfils farmer needs. While the potential positive effects of domestication seem good because of the large variation existing within parkland species, farmers' decision-making processes concerning tree planting and management practices are still poorly understood (Simons, 1996).

Domestication strategies for individual species vary according to their uses, biology and target environments and vary in intensity on the continuum from the wild to the genetically transformed (Simons, 1997a). A central preoccupation of domestication work at ICRAF has been to assess user needs and preferences related to tree species. Priority-setting guidelines including the various stages leading to a gradual reduction of species considered have been made available (Franzel et al., 1996). In Senegal, Mali, Burkina Faso and Niger, farmers were asked to rank 15 preferred species. The 28 species identified out of the potential 60 show that there are common interests across the region. Top species for the whole region included Adansonia digitata, closely followed by V. paradoxa and Tamarindus indica (ICRAF, 1995a). Only two exotic species (Azadirachta indica and the now almost indigenous T. indica) were cited, suggesting heavy farmer reliance on and appreciation of local species.

The exercise showed marked differences among social groups, such as men and women, as well as younger and older men, and emphasizes the need to identify target groups clearly. Species selection is also dependent on purpose (conservation, subsistence, income generation, etc.); thus Bauhinia rufescens and Prosopis africana, which are among the most threatened and important fodder species, did not figure in the 15 top species. There are currently no comprehensive data comparing the economic, social and environmental costs and benefits of domesticating species occurring in forest systems with those of species found on farms. Until this knowledge gap is bridged, priority will tend to be given to farm species for which at least some biological and social information is more readily available (Simons, 1996).

The first steps toward the domestication of several West African parkland species have taken place in past years. Assessment of genetic diversity has been ongoing for F. albida in the last two decades (CTFT, 1988; Leakey and Newton, 1994; Bastide and Diallo, 1996). For P. biglobosa, germplasm conservation and evaluation activities, among others, have been carried out recently through an EU-funded project (Teklehaimanot et al., 1997). Germplasm collections of V. paradoxa were undertaken in 1985–1986 (Adu-Ampomah et al., 1995) and more recently in Ghana (Lovett, 1999; Lovett and Haq, 1999a), as well as in Burkina Faso, Senegal, Mali and Uganda through ICRAF. Prosopis africana seedlots have also been collected in SALWA countries, and B. aegyptiaca in Niger (ICRAF, 1994; Tchoundjeu et al., 1998). Evaluation and improvement activities on A. indica including collections in Ghana, Senegal and Tanzania have taken place within the International Neem Network (Thomsen and Sigaud, 1998). The understanding of vegetative propagation varies from species to species and some details are presented in Box 4.1.

Fig. 4.11 ‘Erect’ variety of Vitellaria paradoxa in farmed parklands, Sawla, Ghana. P. Lovett	Fig. 4.12 ‘Dwarf’ variety of Vitellaria paradoxa with fruit in farmed parklands, Tolon, Ghana. P. Lovett

Because the domestication process starts with farmers, the approach used to improve agroforestry trees presents not only biological but also socio-economic and political challenges. This is a departure from traditional industrial tree breeding and requires different methods. Focus is placed on the rapid availability of high quality germplasm for a diversified and stable production rather than the lengthy process of high selection intensity through generations of provenance and family trials. Generally speaking, depending on species and location, improvement should seek to enhance the quality of tree products, tree growth rate and pest resistance, and ensure species adaptability to potential planting sites (including minimal competition with crops) (Simons, 1997a). Rapid delivery of germplasm also requires that successful propagation techniques are developed in advance of the identification of elite material, so that germplasm displaying superior performance in identified priority traits is made available when demand is incipient. As long as characteristics and functions of trees desired by farmers are not comprehensively surveyed, germplasm with the largest possible genetic variation should be promoted.

For each species under consideration, the question is whether domestication is worthwhile and which strategy should be followed. Decision-making frameworks are needed which take into account the large and complex range of determinants involved (Simons, 1996). An important criterion justifying investment in a selection and seed production programme is the pre-existing demand for the species, and particularly whether the species is currently planted or not. Data presented in Chapter 7 show that markets for parkland products are substantial in terms of volume of products and income. Among other factors, market value depends on the relative lack of tree products due to the degradation of forest cover, and the relative economic profitability of tree regenerating practices.

Elements of the parkland classification presented in Chapter 1 can therefore be used to distinguish parkland and/or species types in order to establish general domestication strategies. Proto-arboricultural and planted parklands, where people already plant, transplant or deliberately disseminate species, may be the easiest target for domestication. Where they occur, a growing farmer demand will ‘only’ require improved germplasm with markedly superior performance. The case of Borassus aethiopum, which is actively harvested and (trans-) planted for food security and income generation, is a prominent and promising example. Farmers may also be keen to use improved germplasm of Azadirachta indica, and possibly Adansonia digitata, both of which they already actively regenerate. In each case, the improvement threshold will need to be defined with farmers in the context of other decision-making parameters.

Whether farmers perceive differences between individuals of the same species is central to the potential impact of tree improvement. Forms of domestication practised by local people and associated indigenous knowledge on genetic diversity should be a point of departure for genetic improvement research. Farmers vary in their appreciation of intraspecific variation, but much indigenous ethnobotanical knowledge still needs to be investigated. In V. paradoxa parklands, regenerated through cycles of cultivation and fallow, farmers carry out some selection, not only at clearing time but also in later years. One-quarter of a farmer sample in Thiougou, Burkina Faso, distinguished unproductive Vitellaria trees on the basis of tree condition (trunk with burn patches or base openings, diffuse or partially dead foliage), nut, leaf and bark characteristics, as well as the amount of nuts found under trees (Boffa et al., 1996b). Narrow leaves are considered a distinguishing sign of a ‘tam daaga’ variety, which produces small and nutless fruit, while the nuts of individual Vitellaria trees called ‘zoonpela’ fall before maturity. The remainder of the sample reported that they eliminate individuals which do not produce well, based on an evaluation period of two to six years. It is probable, therefore, that farmer selection over many generations has led to some degree of domestication in this and other parkland species (Lovett and Haq, 1999c).

Farmers are generally aware of the diversity which exists in highly valued parkland species. Throughout West Africa they recognize a variety of trunk, fruit, seed, pulp, flowering and fruit production characteristics in P. biglobosa. Two main types, ‘dark’ and ‘white’ are often distinguished (A.S. Ouédraogo, 1995). At least three varieties of A. digitata are recognized depending on bark colour. Leaves and fruit of the black or zirafin (Bambara) variety are valued more highly than the red or zirablé variety (Maguiragua, 1993, cited in Cissé, 1995). Another grey-bark or siradjé variety is mentioned in Sidibé et al. (1996). To what extent are these distinctions made in the use and marketing of products and germplasm of parkland species? Generally, the agroforestry germplasm market is imperfect in that no premium is paid for physical or genetic quality and intraspecific diversity tends to be under-appreciated (Simons, 1996).

Potential demand for improved germplasm is also linked to patterns of supply and prices of parkland products, and thus to the density of species available locally. It is not clear, for example, that genetic improvements are necessarily the most effective way of increasing the economic contribution of V. paradoxa (Box 4.2). Modelling the likely influence of an increase in supply on prices of products from V. paradoxa, A. digitata, P. biglobosa and B. aethiopum in western Burkina Faso, ICRAF (1997) found that prices would decline for all products as supplies increased, but in different modes according to areas as well as species considered. For instance, the highest and lowest price drops were for A. digitata and P. biglobosa respectively, and products retained higher prices in areas where the density of corresponding species was lower. This was also the case in a study by Lamien et al. (1996) in Burkina Faso. In response to seasonal variation in the price of V. paradoxa and P. biglobosa, women in Benin maintain their profits by adjusting the weight of sale units or, in the case of processed products, their ingredient concentration (Schreckenberg, 1996). Improvement should thus be spread over a wide range of species and could help farmers distribute their activities according to variations in demand for parkland species. Selection for precocious or double fruiting varieties may be beneficial. Improved storage and processing capacities may also help to counteract the decline of prices resulting from increased supply.

It is important to note that tree management is not uniform throughout given parkland types or species distribution ranges and may vary in intensity depending on a range of conditions. Parkland conservation and expansion efforts including domestication research will be relatively easier in areas where the intensity of indigenous tree management is high or rising. These niches, which are noted in various places in this report, need to be identified more precisely in semi-arid West Africa and targeted for tree improvement activities with farmer participation.

Lastly, little is known about germplasm supply. Domestication can only have an impact if pathways exist to provide farmers with improved germplasm (Simons, 1996). These include distribution (to the National Agricultural Research Systems-NARS, NGOs, communities, the private sector), dissemination (to farmers), and diffusion from farmer to farmer. The demand side including farmers, disseminators and distributors, as well as the supply side, needs to be assessed both quantitatively and qualitatively for parkland species in semi-arid West Africa.

In conclusion, domestication has an important role to play in the maintenance and expansion of agroforestry parklands, but there are wide gaps in the understanding of the needs and types of improved tree germplasm, and its potential impact on the reproduction of these systems. No single method exists to realize its potential immediately. Rather, gradual steps towards the establishment of domestication strategies which are specific to individual tree species, locations and markets are needed. Case studies can serve as models to be adapted to other species. A planting culture driven by demand for subsistence and/or commercial products is necessary for domestication to proceed actively. Research should therefore focus on, and establish, the corresponding links between farmer decision-making processes and conditions of tree resources, as well as characteristics of the demand and markets of parkland products. While increased supply may be central to domestication of scarce, overexploited species, efforts for the improved commercialization of other parkland products may require quality improvements at the production, storage, processing, and packaging stages. Articulate policies for increasing/enhancing the local use of parkland products and organizing and supporting the economic sector based on parkland products would be highly beneficial.

Management and conservation of parkland genetic resources

The concept of forest genetic resources refers to all (environmental, social, economic, cultural, and scientific) values of the heritable materials contained within and among forest species (Palmberg-Lerche, 1996). The conservation of forest genetic resources aims to guarantee their existence, evolution and availability for future generations. It not only implies preserving the present distribution of natural variation in forest species, but also requires the conservation of processes which promote and maintain their genetic diversity (Namkoong, 1991). Genetic variation in species is essential to their ability to evolve in a changing environment, and is central to tree breeding programmes which aim at further developing tree resources for human needs.

Fig. 4.13 Bombax costatum branches trimmed to harvest the flowers for use in a highly appreciated sauce, with negative consequences for regeneration.
J.-M.Boffa

As presented earlier in this report, there are a number of threats, such as drought, land clearing for agriculture, grazing, over-exploitation of wood, fruit extraction, etc., which endanger the sustainability of woody species in arid and semi-arid zones of West Africa. Given the large number of threatened species in these zones and the relatively limited labour and financial resources, national governments and collaborating international organizations see the need to define priority species and populations. For instance, based on the level of threats and socio-economic importance, 11 species were identified for priority conservation at the regional level at an FAO workshop on in situ conservation of forest genetic resources in Ouagadougou in 1994 (FAO, 1994). These species, most of which occur in agroforestry parklands, include in alphabetical order: Acacia nilotica, Acacia senegal, Adansonia digitata, Anogeissus leiocarpus, Borassus aethiopum, Dalbergia melanoxylon, Faidherbia albida, Khaya senegalensis, Parkia biglobosa, Pterocarpus erinaceus and Vitellaria paradoxa. A list of priority species to be conserved at the regional level based on socio-economic and ecological criteria is also regularly evaluated and updated by the FAO panel of experts on forest genetic resources (e.g. FAO, 1997).

The two main strategies for conservation of genetic resources are (i) conservation in situ, which involves maintaining the species in its native environment, and (ii) conservation ex situ, which implies conservation through seeds, live clones, pollen or tissue. These two approaches are complementary and are used together whenever possible. However, in situ conservation is preferred as it is technically and economically more feasible, as well as being based on farmer participation. In addition, it is potentially fully compatible with a continued economic use of forest resources and can contribute to the conservation of associated plant and animal species, and the ecosystems in which target species live.

A wide range of information is needed for successful in situ conservation programmes (Box 4.3). Of particular note is the pronounced lack of understanding of the factors which determine the genetic structure of natural populations. The genetic structure of forest populations is generally determined by biological and ecological parameters including distribution area, reproduction system (auto or allogamy, pollination type), seed dispersal, reproduction mode (sexual or asexual), and long-term evolutionary factors including drift, mutation, natural selection and migration (Zongo, 1998). Genetic variation in tree species can be assessed either through the study of phenotypic variation or through molecular and biochemical markers.

Conservation strategies should take into account factors which regulate genetic diversity. It follows that an understanding of these factors is necessary for the development of successful management guidelines and conservation efforts (Chevallier, 1998). In general, trees are known to display a higher genetic diversity than herbaceous plants both at the intraspecific and intrapopulation levels, but a lower diversity at the interpopulation level. Similarly, species with an extended distribution range tend to have a higher intraspecific and intrapopulation diversity but a lower interpopulation variability than endemic species. Wind dispersion of pollen favours similarity between populations, whereas seed dispersion by gravity has the opposite effect. Through the crossing of related or unrelated parents, and depending on the prevailing reproduction system, fruit also contribute to the genetic structure of populations. An excess of crossings between related parents in a random mating system creates more differentiated fruit, and thus populations which are genetically more distinct. This effect can be measured mostly during the establishment of populations, and declines afterwards. There are also annual variations in gene flow through pollen, as flowering density may vary from year to year. In allogamous, wind-dispersed species with a high proportion of genetic variability at the intrapopulation level, conserving a small number of populations will be enough. In contrast, a larger number will be required for autogamous species with high interpopulation diversity.

Fig. 4.14 Adansonia digitata flower visited by a bat. Bats play a significant role in the pollination of several parkland species.
R. Faidutti

Like most tropical and temperate species studied so far, parkland species including F. albida (Joly, 1991), P. biglobosa (A.S. Ouédraogo, 1995), Tamarindus indica (Thimmaraju et al., 1977), and probably V. paradoxa (Hall et al., 1996) appear to be predominantly outcrossing, with a lower percentage of genetic interpopulation than intrapopulation variation. While phenotypic variation in P. biglobosa seeds sampled in five Sahelian countries was higher within than between populations (A.S. Ouédraogo, 1995), interpopulation differentiation in 11 countries measured through electrophoresis was high (Teklehaimanot et al., 1997).

In northern Cameroon, the overlap of flowering periods among F. albida trees spread over a period of eight months is favourable to outcrossing (Zeh-Nlo and Joly, 1996). Because pruning extends the flowering period, it appears to favour outcrossing. Pollination could occur regardless of tree height or diameter, since no difference in flowering periods according to size groups was recorded. Therefore, gene flow is not limited among trees of various size classes. Furthermore, although cross-pollination between trees was higher during periods of intense flowering, the outcrossing rate remained high regardless of the density of flowering trees, unlike previous results on other insect-pollinated tree species.

The comparison of two distinct F. albida populations in Cameroon illustrates the potential influence of management intensity on genetic diversity (Zeh-Nlo and Joly, 1996). In Zamay, a village with low-intensity agricultural use and marginal pastoral activities, genotypes tend to be irregularly distributed spatially with groups of related trees. In contrast, in Kongola the spatial distribution of genotypes is regular and random with little similarity between neighbouring trees. The situation in the latter village, where cropping and herding activities are widespread, is assumed to result from the role of cattle in seed dispersal over the village lands, thus contributing to the spatial intermixing of genotypes.

These findings suggest that evaluation of genetic diversity should take into account local variations in both time and space. First, maximum genetic diversity will be obtained from seeds produced during the peak flowering season, while seeds produced at the beginning or end of the season may have a narrower genetic base. Secondly, collection schemes should take the spatial organization of genetic diversity into account to ensure better representativeness.

The structure of genetic diversity in F. albida at the village level is shaped by the intervention of livestock in tree regeneration as well as farmers' tree selection practices. At the continental scale, it is influenced by livestock transhumant pathways, human migrations, and resource use patterns by wild herbivores. Likewise, flows of P. biglobosa plant material across villages, regions and national borders through human migrations, commercial relations and transnational ethnic ties may have contributed to the relatively low phenotypic interpopulation variation observed (A.S. Ouédraogo, 1995).

The impact of practices used in parkland management on processes which influence genetic diversity remains widely unknown and provides numerous opportunities for research. For instance, studying the impact of parkland management practices, including clearing, fallowing, pruning, selective wood harvesting and fruit/seed extraction, would contribute to the identification of criteria and indicators for the conservation and sustainable use of agroforestry parkland systems.

Summary

Farmers have sustained agroforestry parklands through various management practices. Parkland species tend to be regenerated naturally rather than planted. Assisted regeneration can significantly increase tree/shrub densities, either with the objective of reconstituting a parkland cover or for the ongoing production of wood and fodder, especially in areas of high population density and pressure on wood stocks. Tree planting is usually confined to the proximity of homes where tenure is more secure, soils are better fertilized and seedlings are protected from browsing, drought and fire.

With the exception of Borassus aethiopum, which is actively planted to compensate for harvesting, planting primarily involves exotic species because of their fast growth rate and the availability of indigenous species in fields and fallows. Increasingly, however, farmers appear to participate in the planting/transplanting and dissemination of local tree species, which can be a useful way of taking advantage of the microsite variability in soil fertility in cultivated fields. Improved fallows, whereby economically useful and fertility-improving trees are planted before cropping is discontinued, appear to be a promising alternative to long natural fallows, but are still at the experimental stage.

Parkland species display some tolerance to fire. Fire may stimulate Vitellaria regeneration, but less so than an absence of burning with the removal of plant competition. The beneficial effect of fire protection is more pronounced in drier areas. Frequent burning delays first fruit production in young Vitellaria trees, but can stimulate flowering and fruiting in mature trees. However, no evidence points to fire having a role in increasing yields. The timing of fires is a key factor. Late fires are very destructive for growth, wood and fruit production, while burning early in the dry season has little or no effect and is therefore recommended. Several fire control techniques are applied for the protection of valuable trees such as Parkia biglobosa. The increase of soil nutrient availability after fire is immediate, superficial and highly subject to loss. While burning is a practical solution to reduce the abundant woody biomass at the end of a fallow cycle, fire control is generally advised in semi-arid West Africa as it contributes to an increased biomass and canopy cover, to the biological recycling of nutrients and to higher system productivity.

Sahelian farmers commonly apply silvicultural techniques to increase production of parkland trees. These include seedling protection and fencing, watering, and the selection of a vigorous shoot, particularly for Faidherbia albida. Parkia biglobosa is commonly pruned to improve survival and productivity, and for rejuvenation and shade reduction in old trees. Large Faidherbia trees are most often pruned for fodder production in silvipastoral zones of the Sahel. Pruning can be intensive (with an average canopy reduction of 40 percent) and stimulates leaf regrowth, causes an additional foliation peak during the rainy season and drastically depresses pod production. Other parkland species are pruned less intensively, often with the objective of improving tree form and enhancing understorey crop performance. Ringing is sometimes carried out to stimulate fruit and seed production. Coppicing and pollarding, practised both in fields and fallows, represent a way of limiting competition with intercrops and providing wood and other tree products in species with good vegetative growth such as Azadirachta indica and Detarium microcarpum. Parkland trees should be coppiced early in the dry season to minimize competition with crops. Fertilizers are rarely applied to trees, as crops have priority; consequently their effects have been little studied. Techniques including manuring and application of rock phosphate in conjunction with mycorrhizal inoculation can benefit tree establishment and production.

Available data for various parkland species show that crop productivity under pruned trees is higher than under unpruned trees and sometimes also higher than in open controls. The improved light and soil fertility conditions appear to outweigh below-ground competition which is possibly intensified with root morphology changes resulting from pruning. Pruning appears to hold good potential for improved parkland management, but better understanding of the comparative impact of repeated pruning on soil fertility and crop performance is necessary before recommending it widely. Its effect on the production of fruit, leaf, wood, and other useful products also needs to be quantified to provide a balanced picture.

Organic inputs are necessary to maintain soil fertility in semi-arid West Africa. Preliminary positive results have been achieved with 25–75 kg nitrogen/ha of Azadirachta indica or Acacia lebbek leaf mulch. Timing of application influences yields in a way related to rainfall distribution and stages of crop development. Highest yields are achieved with mulches combining high nutrient delivery with moisture conservation and temperature reduction. Despite the promising contribution of tree mulch technology, additional research is needed regarding its on-farm application and costs and benefits to farmers. Research is also needed into extending the existing farmer practice of selecting certain crops to take advantage of the specific fertility and microclimate conditions in the sub-canopy ecological niche.

Planting of indigenous trees might be more popular if they had faster growth rates, shorter juvenile phases and higher and more consistent yields. Improving these and other characteristics could take advantage of the great variation existing in parkland species. Guidelines for prioritizing species and characteristics for domestication research have been developed by ICRAF. These indicate that there are certain species of common interest across the Sahel, but also highlight the differences that may exist between social groups. Target groups for domestication must therefore be carefully identified. At the same time as dealing with biological problems, socio-economic and political challenges also need to be overcome. Thus a key to successful domestication of indigenous species and their adoption by farmers is ensuring that demand exists together with a potential supply system to produce and disseminate the improved germplasm. Parkland species which are intensively managed may be the easiest target for domestication. Research with farmers is necessary to identify the levels of improvement required to ensure uptake as well as to make use of their knowledge of intraspecific variation in certain species.

The continued existence of such intraspecific variation is important to allow both natural plant selection and evolution, and the use and breeding of woody plants for diverse, changing needs and environments. Conservation of forest genetic resources requires the preservation of the present distribution of natural variation as well as the processes that promote and maintain their genetic diversity. This can be achieved through ex situ conservation of seeds, live clones, pollen or tissue and/or in situ conservation of the species in its native environment, as it is practised in agroforestry parklands. The latter is technically and economically more feasible and also preferred because it involves farmers and leaves open the possibility of combining use and conservation of the resource. It does however require a better understanding of the factors that determine the genetic structure of natural populations, specifically their breeding mechanisms. Current research indicates that the majority of parkland species are outcrossing and show a high level of local variation (in both time and space) of genetic diversity.