CHAPTER II
PARKLAND DYNAMICS

Parklands are extremely dynamic systems which may develop over many generations, reflecting changes in the physical and socio-economic environment. In spite of the importance of these systems, however, there is an almost complete lack of quantitative data in the literature concerning actual trends in parkland extent, density and age distribution over past years or decades. This chapter reports on various biological, cultural, socio-economic and policy conditions which have been conducive or detrimental to the maintenance and expansion of agroforestry parklands. Specific details on the biophysical factors affecting tree-crop interactions are provided in Chapter 3, while the impact of Sahelian forestry policies on parkland management is covered in Chapter 5.

Changes in tree density over time

In West Africa, the cultivated area has expanded as population has increased at an annual rate of 2.7 percent over the past decades. Frontier areas and river basins freed from onchocerciasis have been opened up for cultivation. In the Sahelian countries, agriculture has also expanded northwards, by about 150 km in Niger (Breman and Traoré, 1986, cited in Breman and Kessler, 1995). One might well assume that where the traditional practice of tree conservation in fields has been retained, the area of parklands should have increased.

While little is known about changes in parkland extent, it appears that tree density and regeneration in parklands have declined. For example, the annual rate of decrease in the density of large trees throughout the village of Petit Samba, Burkina Faso, was 0.15 trees/ha from 1957 to 1984 and 0.57 trees/ha from 1984 to 1988 (Gijsbers et al., 1994). Whether density declined in its parklands was not specifically investigated; there was, however, a clear lack of small diameter classes. In Sob, Senegal, total parkland tree density decreased from 10.7 in 1965 to 8.3 trees/ha in 1985 (Lericollais, 1989). Decreasing tree cover in fields was also reported in Yatenga, Burkina Faso (Marchal, 1980). In contrast, an increase in density from 12.9 to 15.2 trees/ha and 6.1 to 6.7 trees/ha respectively, was documented in two large experimental sites of the Kano Close-Settled Zone of Nigeria between 1972 and 1981 (Cline-Cole et al., 1990). Another study outside the Sahel and Sudan zones reveals that the volume of on-farm woody biomass in western Kenya has increased between 1986 and 1992 at an annual rate of 4.7 percent (Holmgren et al., 1994). It did not specify, however, what percentage of the total woody volume was represented by scattered trees in fields.

Fig. 2.1. ‘Construction’ of a Borassus flabellifer parkland, Banfora, Burkina Faso
S.J. Ouédraogo

A study by Fairhead and Leach (1996) in Kissidougou, a prefecture in the forest-savanna mosaic zone of Guinea, illustrates the importance of not jumping to conclusions about human impact on the landscape. In an area of ‘derived savanna’ in which forest islands were assumed to be relics of a more extensive forest long since degraded by local people, the comparison of aerial and satellite photographs taken in 1952 and 1992 showed that the area of forest and secondary forest thicket vegetation has remained remarkably stable and, in places, has expanded in surface area, sometimes considerably (50–500 percent). The study shows that many of the forest islands around villages were in fact established by local populations, and that the open savanna has also been enriched with more woody species. Even though it does not specifically deal with parklands, the study is useful in suggesting that scientists and policy-makers can misread decades of forest history and that methods used by farmers to enrich their landscapes can be obscured and marginalized. It also underlines the importance of historical research, using oral and documentary evidence, to understand vegetation changes in agroforestry parklands.

Data suggest that parklands have been degraded in term of tree density and that lack of regeneration threatens their sustainability.

It is difficult to draw clear and definite conclusions about general parkland trends, given the very small number of quantitative reports available. Caution is also needed in interpreting results as plot boundaries are not always described and overall numbers can conceal significant variations at the site, field or soil unit scale. Nevertheless, the literature is dominated by a general, qualitative agreement among researchers and practitioners that tree densities have declined significantly in Sahelian parklands and woodlands since the droughts of the 1970s. This assessment is very often consistent with villager perceptions. Given the long generation time for trees, the lack of young age classes in parklands should be a warning of a serious risk of degradation of these systems. However, as this report demonstrates, parklands are not uniformly in decline and show a great deal of resilience and potential for recovery and sustainability.

Even given favourable conditions, tree densities are likely to vary over time. As trees only have significant productive or environmental value after they have attained a certain size, parkland management takes place over several decades. The management period may include the practice of fallow for Vitellaria and Parkia parklands or protection of tree regeneration in the more permanently cultivated F. albida parklands. Often, tree density is related to the length of time an area has been farmed (Pullan, 1974) with density in newly cleared areas being higher than in old parklands (Otegbeye and Olukosi, 1993).

There are several reasons for this:

• Density management at the time of field establishment takes place during clearing and for several years afterwards. On a sample of oldest sections of fields cultivated for the first time in Thiougou, Burkina Faso, Boffa (1995) showed that field age as reported by farmers was the factor which explained the most variability in tree density. The annual rate of density decline was 1.1 trees/ha. The practice of continuing tree selection was not always reported, however. Because forestry agents have long imposed severe sanctions against tree cutting activities in Sahelian countries, farmer participation in parkland management may be underreported or maintained at below optimal levels.

• Annual labour availability for tree and bush cutting may be limited.

• Non-productive individuals are eliminated over the years. In Thiougou, farmers evaluated the potential productivity of V. paradoxa trees over a period of two to six years or based on immediate observation of tree condition, nut, leaf and bark characteristics (Boffa, 1995).

• Trees are also cleared to ease the passage of draught animals (Bagnoud et al., 1995a; Boffa, 1995). In addition, tree sites are more fertile than surrounding areas for several years after tree death, as evidenced by better crop development. Farmers probably manage these sites actively in the context of micro-variability of soil fertility over the cultivation cycle.

• Finally, stands are thinned as trees grow larger and cover too much of the understorey cropping area. The difference between areas characterized by a large number of small trees and areas with a few trees with large crowns may be explained by differences in duration of occupancy (Pullan, 1974).

A decline in tree density over time may therefore be an inevitable process in the early part of the maturation of some parklands, such as Vitellaria parklands. The variety of objectives that farmers may be pursuing in establishing ‘appropriate’ parkland densities is also clear. Interwoven with the time dimension are environmental parameters which significantly affect tree growth and regeneration as well as parkland density, and which farmers integrate in their management decisions. Appropriate densities will be maintained in the long term only if farmers deliberately compensate for tree senescence and natural mortality with protected natural (or planted) tree regeneration. But farmers also manage the tree component of their farms in relation to other productive resources and in the context of various external influences, which make parkland management more complex. Additional factors responsible for changes in surface area, tree density and species composition in West African parklands are discussed below.

Natural factors

Drought

The effect of droughts in the 1970s and 1980s on tree survival and growth has been severe not only in natural woodlands but also in agroforestry parklands. Average annual rainfall during this period was at least 150 mm lower than during the pre-1970 period in seven village research sites in the Central Plateau and southwestern part of Burkina Faso (Lowenberg-Deboer et al., 1994). This general pattern is found throughout the Sahel and Sudan zones of Africa and has caused isohyets to shift south. The effect of drought on woody species has been more intense in the Sahel than in the Sudan, and on the upper parts of the landscape than in valleys (Breman and Kessler, 1995). Since the 1970s the occurrence of parkland tree species has declined in the northern part of their range. This is the case of V. paradoxa (Ohler, 1982, cited in Breman and Kessler, 1995). Ouédraogo and Alexandre (1996) also report that this species can now only be found in bottomlands in the Bam Province (600 mm rainfall) of Burkina Faso. However, drought does not always affect trees negatively, as illustrated in Box 2.1.

Livestock

Livestock is often assumed to play a determining role in breaking seed dormancy and increasing germination for parkland species such as F. albida. Researchers have therefore monitored the itinerary of seeds from fruit fall, which takes place during the dry season, from January to April in semi-arid West Africa, to seedling establishment. Contrary to popular belief, results suggest that livestock considerably reduces the number of viable seeds. In replicated stable experiments, only 2, 10 and 11 percent of healthy F. albida seeds ingested by sheep, goats and livestock respectively and found in faeces had retained their viability (Depommier, 1996b). Similarly, 84 percent of seeds consumed by enclosed sheep were digested in Ethiopia (Tanner et al., 1990). A much higher percentage (65 percent) were recovered in ox faeces in Sudan (Radwanski and Wickens, 1967) but the experiment was not replicated.

Fig. 2.2. Peulh herders let cattle browse pruned Faidherbia albida branches before piling and collecting the wood
R. Peltier

After ejection in livestock faeces, some seeds will then germinate in the dry season and die shortly afterwards due to lack of rainfall or termite attack (Depommier, 1996b). The number of F. albida seeds available in faeces on farms in Watinoma, Burkina Faso, was only 500–1125 seeds/ha, and was higher under tree crowns where livestock seeks shade in the dry season than in open areas. The germination rate of seeds having undergone intestinal transit was only slightly lower (close to 80 percent) than either seeds treated with sulphuric acid and soaked in water for 24 hours, as normally recommended, or control seeds (both averaging over 90 percent). However, transit through animals resulted in significantly delayed and more gradual germination, which reached a maximum only after several weeks. The difference may be due to the fact that only seeds with the hardest tegument and thus the hardest-to-break dormancy resist chewing and digestion. Because the first rains are erratic and often followed by drought periods, this delay is a substantial advantage for survival.

Therefore, rather than increasing or accelerating germination of F. albida seeds, livestock actually reduces potential germination. It does, however, contribute to wide dissemination throughout village fields (through manuring contracts with herders or spreading of manure collected in enclosures or compost pits), as well as enhancing seed conservation and survival by delaying and extending germination. Livestock has a further negative impact on regeneration by trampling seedlings, and causing the partial or complete elimination of tree shoots. Its effects, along with drought and farmers deliberately cutting or destumping shoots during field operations in the rainy season, can result in very limited regeneration rates (Depommier, 1996a).

Pests

Drought, pests and exotic tree species have exerted pressure on parkland species.

Pests have also weakened some parkland tree populations. Vitellaria paradoxa trees are infected by parasites of the Loranthaceae family. Four species of Tapinanthus (dodoneifolius, globiferus, ophiodes and pentagonia) are widespread in Mali and Burkina Faso. Surveys indicate that up to 95 percent of Vitellaria trees may be infested in Burkina Faso (Boussim et al., 1993a) and in Mali (Maiga, 1989). The effects of the parasites can include decreased wood quality, lowered resistance to pathogen attacks, decreased fruit production, and death of the tree. These parasites also affect other parkland species but to a lesser degree. In Watinoma and Dossi, Burkina Faso, 5–10 percent of F. albida trees were affected (Depommier, 1996a).

Potential sexual regeneration of parkland species through seeds can also be drastically altered through insect and bird attacks. In Burkina Faso, F. albida seeds are perforated by insects of the Bruchideae family, such as Caryedon sp. and Bruchidius auratopubens which feed on the seed embryo and cotyledons and destroy the seed's viability. Seed damage reached 75 percent on some F. albida trees in Watinoma (Depommier, 1996a), or comparable numbers in other studies cited by this author. A series of insects reviewed in Sallé et al. (1991) affect buds, leaves and reproductive organs, as well as branches and wood of V. paradoxa. The same situation probably applies to other parkland species. Likewise, birds such as Lamprocolius spp. removed seeds in 85–95 percent of F. albida pods from given trees (Depommier, 1996a).

Fig.2.3. Azadirachta indica invasion around Faidherbia albida trees in Dossi, Burkina Faso
S.J. Ouédraogo

Species composition and age structure in parklands are also evolving under the competitive influence of other tree species (Ganaba, 1996). Neem (Azadirachta indica) was introduced to West African Sahelian countries in the late 1910s through the English-speaking coastal countries. Since then, it has considerably expanded through planting for a variety of uses (primarily for shade, construction, medicinal and veterinary uses), and shows profuse and invasive levels of natural regeneration. This species grows better than others on rocky, lateritic and shallow soils and can withstand rainfall as low as 150 mm (von Maydell, 1983). It is disseminated by a variety of birds which induce faster germination by removing the fruit pulp.

Neem thrives in some parklands such as those of the Bulkiemdé Province, Burkina Faso. Farmers first planted it there on land unsuitable for crop production, then on areas close to compounds where densities of up to 4.4 trees/ha were recorded (Yélémou et al., 1993). The progression of neem is also marked in village and bush fields in Watinoma, Burkina Faso (Ouédraogo, 1994), and in Sob (550 ha), Senegal, where its numbers have increased from 3 in 1965 to 247 in 1985 (Lericollais, 1989). However, it competes aggressively with local parkland species for space and nutrients and dominates parkland regeneration. Dense concentrations of neem seedlings or suckers are commonly found around species such as F. albida, V. paradoxa, P. biglobosa, Adansonia digitata and Ceiba pentandra as a result of bird dissemination. The phenomenon is generalized throughout the Sahel and Sudan zones and appears to be detrimental to parkland species like F. albida (Depommier, 1996a). At the same time, in several places in West Africa, neem is reported to be declining, with signs of withering such as dead limbs, yellowing leaves, and a ‘giraffe neck’ appearance with only terminal leaves left on the tree. This condition is thought to be caused by a number of insects and fungal attacks. Neem decline is particularly severe in Niger where 30–40 percent of sampled trees are affected, and may be related to the narrow genetic base of the first introductions there (Direction Nationale de l'Environnement, 1998).

Economic and socio-cultural driving forces

Products from parkland trees are a significant economic resource. Farmer interest in maintaining and regenerating parklands therefore depends on the value of their products relative to other products of the land, or alternative income-generating activities.

In northern Côte d'lvoire, management of Vitellaria tree densities responds rapidly to changes in the relative prices of its products (Louppe and Ouattara, 1996). When Vitellaria nuts or butter sell for high prices, regeneration is promoted. Conversely, if fuelwood prices outstrip those of the tree's other products, trees tend to be felled and sold on the fuelwood market. The increased commercial exploitation of fruits from parkland species can result in local-level regulatory action to maintain the production capacity of tree stands, as Wiersum and Slingerland (1997) observed for Detarium microcarpum forests in southern Burkina Faso. Similarly, Bergeret and Ribot (1990) noted that the proportion of useful and multipurpose tree species protected in fields increased between 1968 and 1987 in Kumbija, Senegal (though density data were not reported). Farmers appeared to respond to the decreased availability of desired species in village forests, due to droughts and the intrusion of tree cutters for charcoal making, by regenerating trees on lands they could still control.

Fig. 2.4. Fruit of Vitellaria paradoxa ssp. nilotica held by a farmer participant of the Shea (Yao) project in Lira, northern Uganda
E. Masters

In southern Niger, Mayahi arrondissement (Maradi department), where rainfall ranges from 200 mm in the north to 450 mm in the south, has experienced spectacular changes in the management of woody vegetation in fields over the past decades (Joet et al., 1998). The low population density and relative abundance of woody vegetation in range and fallow lands near villages led farmers to clear woody vegetation for cultivation and remove (construction and fuel) wood and fodder without planning for the renewal of resources which appeared unlimited. In the 1980s, as a result of drought and demographic increase, the disappearance of fallows and woodlands and the unavailability of wood on these collective lands prompted farmers to adopt regeneration practices in their own fields. Improved clearing techniques, in which the sexual or vegetative regrowth of woody plants in fields is protected, were introduced and extended in the region by several projects in the 1980s and have spread spontaneously to several dozen villages in Mayahi.

Farmers in the Ahmar mountains around Alemaya in eastern Ethiopia increase the area interplanted with F. albida because they need to obtain cash earnings and several products from the same land unit (Poschen, 1986). However, one of the major cash crops, ch'at (Catha edulis), is susceptible to leaf diseases under shade and its growth phases overlap too much with those of the trees. As landholding size tends to decline with increasing population density, Poschen (1986) fears that tree density in fields will decline as the need to intensify ch'at cultivation is felt.

In Wolokonto, Burkina Faso, Borassus aethiopum parklands, characterized by 90 percent of trees being in the early development stage, are expanding both spatially and in density due to the high income generated from palm wine (Cassou et al., 1997). Significant tree regeneration and existing extraction skills currently ensure the sustainability of these parklands. However, Cassou and colleagues (1997) raise concerns about its future maintenance by younger generations attracted to higher urban salaries, plus the decline in wine demand resulting from the rising number of Muslim farmers. Likewise, Bernard et al. (1996) hypothesize that the local trade in Vitellaria and Parkia tree products partly accounts for the development of these parklands in northern Côte d'lvoire, where densities have increased over the past three decades.

Noticeable increases in parkland density were also observed between 1972 and 1981 in the Kano Close-Settled Zone of Nigeria which has a close connection with the urban economy and population densities in excess of 200 people/km² (Cline-Cole et al., 1990). The woody vegetation might have been expected to be severely stressed by an accentuated reliance on trees for fodder, food, and wood after the severe droughts in 1969–73. Instead, the large number of small diameter trees was evidence that farmers had been making a spontaneous effort to conserve, regenerate and plant trees in response to commercial opportunities for fuelwood and other products.

In some areas of Uganda, the Vitellaria tree is cut for charcoal-making in spite of its economic importance as a source of cooking oil (Masters and Puga, 1994). Where a strong cultural tradition of Vitellaria conservation exists, trees are maintained and used for butter production. The introduction of effective labour-saving technologies and the expansion or creation of markets to improve profitability have therefore been a priority for development projects focusing on the conservation and improved utilization of Vitellaria and other parkland species such as Balanites aegyptiaca, Parkia biglobosa, Tamarindus indica, Azadirachta indica and Jatropha curcas.

According to Hervouët, (1980, cited in Dallière, 1995) F. albida parklands were abandoned starting in the early 1900s among the Bissa, around 1940 among the Senoufo, around 1960 among the Bwaba and around 1970 among the Bissa and Samo in Burkina Faso in response to a conjunction of factors including the monetarization of the economy. Vimbamba (1995) argues that the cutting prohibition associated with species such as Khaya senegalensis led farmers in some areas to cut down and sell F. albida trees in order to pay taxes and make wooden utensils. This explanation may be debatable in view of the relatively poor quality of F. albida wood (Depommier, 1996a), but it may be linked to the local unavailability of species providing higher wood quality.

Where traditional products from parkland trees can be substituted by cultivated crops or items purchased at the market, farmers may be less motivated to regenerate parklands. In Yatenga, Burkina Faso, Marchal (1983) noted that milk was becoming increasingly dominant as a fat source and substituting for oil from Vitellaria trees. Schreckenberg (1996) observed that the Anii and Logba groups in Benin prefer the taste of palm oil to that of Vitellaria when they can afford it. Across the border in Togo, both these oils are being replaced by groundnut oil (Sauvaget, 1981), which is less variable in inter-annual yield and less arduous to process. Likewise, Maggi cubes or fermented soybeans represent potential commercial substitutes for fermented Parkia seeds. Larger market organization may also sustain the development of parkland trees. The small size of local markets appears to constrain the rising mango production in Rakaye on the Central Plateau of Burkina Faso (Vimbamba, 1995).

The value of parkland tree products can also influence farmers' objectives in parkland management. Before the drought in Burkina Faso, exploitation and reduction of the environmental effect of trees were the primary motives for pruning Parkia trees. Nowadays, Parkia fruits have become more valuable due to the relative decrease in density of Parkia individuals over the period and their more extensive commercialization. Consequently, farmers manage Parkia trees as a significant resource in their own right and mostly carry out pruning to promote fruit production rather than to reduce the potentially negative impact of the canopy on crop yields (Timmer et al., 1996).

Fig. 2.5 The COVOL (Cooperative Office for Voluntary Organizations of Uganda) hand press for extracting shea oil, Lira, Uganda
E. Masters

There have also been changes in the economic condition and socio-cultural traditions which sustained parklands. These are having an impact on management practices. Unlike the self-sufficiency prevailing several decades ago when parklands were developed or maintained, villages are no longer economically isolated. Important population movements have taken place during the 1970s and 1980s, including emigration from rural areas toward cities, coastal states and the southern parts of Sahelian countries. Villagers often rely on remittances from relatives living in these settlement zones. Younger generations are faced with the existence of opportunities in urban centres. In contrast, management of village lands in the past was self-contained and governed by strict traditional rules.

Some authors argue that the lessened villager interest in products and services offered by trees may be reflected in parkland degradation patterns observed in some places (Lericollais, 1989; Seignobos, 1996). However, one should also be concerned with how rural emigration affects demands for agroforestry products. Food habits regarding such products as P. biglobosa soumbala, V. paradoxa butter, Bombax costatum calyces, etc., are not necessarily altered as migrants come to live in urban areas. Links between village residents and urban migrants may then serve as a means of relating supply and demand and provide a channel for informal or market exchange for these products. Although the widening of village economies and urban migration may appear to have a detrimental effect on resource management, they also open up new opportunities for the processing and marketing of parkland products due to increased market size and demand for more elaborately processed or packaged products, etc. (Ndoye et al., 1997; FAO, 1995). These factors represent driving forces for the domestication of parkland species, as discussed in Chapter 4.

Prosopis africana parklands, as developed by the Musey of northern Cameroon, represent a somewhat unique system that was motivated primarily by socio-cultural factors rather than by agronomic and/or forest production benefits (Bernard, 1996). Social recognition among the Musey revolved around accomplishments in warfare and hunting. The horses they raised enhanced a warrior's social image and were prepared and venerated during rituals before combats. Prosopis africana was another indicator of social distinction. It was systematically protected by the group and was exclusively cut for funerals to decorate the tombs of respected clan members who were sometimes buried with their horses. Traditional taboos and ceremonies related to the way of cutting Prosopis trees upheld the tree conservation practice. However, the collective importance of these traditions has waned with the disappearance of tribal warfare, the prohibition of hunting fires, and the transfer of farmer interest from ponies to oxen for animal traction. Younger generations tend to discount animist beliefs in favour of introduced religions. Farmers also report that schooling has diverted children from chasing crop-damaging birds away from crops. Control of bird damage is now achieved through tree removal. The ongoing degradation of P. africana parklands reflects the erosion of socio-cultural values which sustained this agricultural system until a few decades ago. The future of the species in Musey agroforestry systems will depend, among other factors, on the potential economic benefits (associated with its fodder properties and high timber quality) that can be derived from its integration with crops.

Another case of socio-cultural reproduction of parkland agroforestry practices is the development of F. albida parklands in Watinoma, Burkina Faso (Depommier, 1996a; Vimbamba, 1995). When he took up his position about 30 years ago, the current village headman, who originated from a nearby area with a strong F. albida tradition, started regenerating the species in his own fields. In the general context of land degradation, and exposure to R&D activities, he was imitated by villagers, who became convinced of the species' role in restoring the fertility of their soils, and revived the F. albida tradition.

Agricultural development policies

In past decades, a number of developments aimed at increasing farm productivity have had important indirect effects on parklands. This section illustrates how the conception of the general role of trees in agricultural development models, technological developments such as animal traction, the intensified production of cash crops and use of inputs, cultural innovations and fertilizer subsidies have influenced the condition of parklands.

Role of trees in agricultural development models

Extension services have only recently advised farmers to integrate trees into agricultural systems. In the 1950s and 1960s, agricultural development models in West Africa promoted geometrically shaped fields devoid of woody cover where single cropping was prescribed. Trees, especially of local origin, were ignored or even considered as obstacles to the development of intensive and productive systems. In places, financial credit for agriculture was given only if all trees were cleared from fields. Trees had little role to play in land improvement schemes and were thought to be incompatible with animal traction or more intensive mechanization. Such models pervaded strategies of research, extension and development as well as commodity parastatal agencies during this period. References to these early land management projects by agricultural agencies are given for Senegal (Lericollais, 1989) and Cameroon (Seignobos, 1996). This trend was also evident in Zambia and Zimbabwe until 1980 or so (Campbell et al., 1991). Insufficient consultation among natural resource disciplines including agronomy, soil science and forestry probably hampered the integration of trees in the agricultural agenda. Awareness of the insufficiencies of these early development projects gradually led to more consideration of the role of trees in sustainable agricultural systems, and natural tree regeneration was included in some projects (Montagne, 1984; Taylor and Rands, 1991). Only more recently, however, have agricultural extension and development agencies begun to support the maintenance and regeneration of trees in fields (Joet et al., 1998).

Fig. 2.6 Oxen traction for sowing cotton in Parkia biglobosa parklands, Dolekaha, Côte d'lvoire
C. Bernard

Mechanization

Farmers using animal traction generally maintain a lower tree density in their fields, but this varies according to location and production objectives. Trees are eliminated to give draft animals space to manoeuvre and to reduce ploughing or ridging problems where tree roots are superficial. Traditional cultural practices were substantially altered under direct outside influence in Gallais' example in Mali (1967). Bergeret and Ribot (1990) also report from Senegal that government subsidies encouraged farmers to clear their fields of all trees and stumps when animal traction was introduced. A similar process of tree removal was reported in central Gambia (Weil, 1970, in Freudenberger et al., 1997). In three cotton-producing villages of southern Mali, densities of trees in fields were almost twice as high (about 16 to 23 trees/ha) on farms where mechanization was absent or incomplete as in those well equipped with animal traction (about 8 to 13 trees/ha) (Bagnoud et al., 1995a).

Fig. 2.7. Mechanized cultivation in Vitellaria paradoxa parklands, Bole, Ghana. Note lack of woody regeneration.
P. Lovett

In northern Côte d'lvoire, soil cultivation by tractors led to the drastic elimination of trees in the early years after the introduction of cotton (Bernard et al., 1996). As farmers later abandoned the use of tractors and adopted less expensive animal traction, they encouraged tree regeneration. Currently, tree density is slightly higher in farms with animal traction (15 trees/ha) than among hand-cultivating farmers (12 trees/ha) but not significantly so. In the Vitellaria parklands of Thiougou, a frontier village in southern Burkina Faso, farmers using hand cultivation, donkey and ox traction had tree densities of 31, 25 and 21 trees/ha respectively (Boffa, 1995). The difference was statistically significant between farmers using manual labour and those using oxen. The relatively high density of trees in Thiougou, even in fields cultivated with oxen, may be related to the recent conversion of lands to agriculture and the limited size of parkland trees, as well as the fact that production is almost exclusively oriented towards staple rather than cash crop production. The lighter plough and greater manoeuvrability of single animals in donkey traction had less of a depressive effect on parkland tree density. These examples show that, depending on the degree of agricultural intensification, trees need not be incompatible with animal traction.

The extent to which animal traction is hampered by trees depends on the rooting patterns of parkland species. Faidherbia albida is generally not considered a significant obstacle in the northern part of the Sudan and Sahel zones, because of its deep tap rooting habit, yet under higher rainfall conditions superficial rooting can be a constraint to mechanization (Bernard et al., 1996). Animal traction can also promote tree regeneration. In the shallow or hydromorphous soils of Watinoma and Dossi, Burkina Faso, regeneration of F. albida suckers was common on superficial roots where they had been partly or completely cut by a plough (or hoe) (Depommier, 1996a). This type of vegetative reproduction was believed to form a significant part of local Faidherbia albida parkland regeneration.

Crop introduction and development

Intensive cash crop production supported by the use of chemical fertilizers has led to parkland degradation in several places. In Mali, between the Niger and Bani rivers, the intensification of cotton production had a detrimental impact on parklands (Gallais, 1967). Following its introduction during World War II, cotton was associated with staple crops in bush fields. In the early 1950s, the French parastatal textile company, Compagnie française de développement des textiles (CFDT) successfully promoted mechanized cotton production in compound fields. As a result, farmers reduced the F. albida cover in these fields and cereal crops were transferred to bush fields. These conditions probably made production more vulnerable to drought and soil productivity less sustainable (Gallais, 1967). Farmers had also cleared all tree cover including F. albida in dry-season onion plots in bottomlands.

The decline of field tree density to the current level of 5 trees/ha in the M'borine area of Senegal is partly attributed to the introduction of groundnuts as a short fallow cash crop, as well as the decline of indigenous fallowing methods (Postma, 1990, in Shepherd, 1992). Tree density was more intensely reduced in cotton fields located in valleys than in other valley fields in Petit Samba, Burkina Faso (Gijsbers et al., 1994). In western Burkina Faso, intensive cropping systems under F. albida cover, which allowed high population densities on limited cultivated zones, have been neglected with the development of extensive cotton production in bush fields. Such a process has been felt in the last 20 to 25 years in the village of Dossi (Depommier, 1996a). In the Central African Republic, clearance of V. paradoxa and P. biglobosa trees in cotton fields was supported by SOCADA, the national agricultural development agency which focused on improvement of cotton production (Depommier and Fernandes, 1985).

In the village of Holom in the Bec de Canard area of northern Cameroon, Prosopis africana parklands, a rare form of parklands in Africa, have deteriorated, as indicated by a 50 percent overall tree mortality rate and a virtual absence of Prosopis regeneration in fields. Cotton introduction is held responsible for this decline (Bernard, 1996). The displacement of traditional staple crops resulted in a reduced practice of fallowing (in time and land area), which previously allowed for tree regeneration. In the soil management schemes promoted by SODECOTON, the company that superseded CFDT, the fertility enhancement function of fallows was replaced by the use of chemical fertilizers. In addition, trees are generally systematically excluded from plots in SODECOTON's technical package, whether directly or through the impact of animal traction.

Furthermore, plot layout and management patterns applied by the cotton agency do not conform to traditional land division and tenure arrangements. Thus the drastic expansion of cotton production (58 percent increase in area in two years) has been accompanied by an increase in land loans. Cotton is grown in blocks of land which may not correspond to former field boundaries and which are relocated on an annual basis, resulting in complex tenure situations. The alteration of land tenure rights also causes a modification of tree management strategies. For instance, the highest density of dead or dying P. africana trees in Holom-Nguéring is found on plots loaned for cotton production to farmers of neighbouring quartiers (quarter) and villages. Some of these borrowers argue that they are descendants of the founder of the first settlement of Holom and claim primary landholding rights over cotton plots. Bernard (1996) hypothesizes that the gradual elimination of trees may be a way of removing existing signs of traditional tenure and gaining exclusive land rights for the production of this cash crop.

Finally, in order to compensate for the reduction of the area previously cultivated with staple crops, farmers adopted muskuwaari, a dry-season sorghum variety. This is planted in periodically flooded depressions with clay soils, and on the whole its labour input does not overlap with that of cotton production. But farmers systematically clear all trees on and around bottomlands where muskuwaari is grown, to control bird depredation. The substantial displacement of rainfed crops by cotton and the transfer of cereals into bottomlands have thus been detrimental to the maintenance and expansion of P. africana parklands in this area.

However, the negative impact of modern technologies on parklands may relate to the way they have been applied rather than their incompatibility. Other evidence exists to show that technological change can have positive effects on parklands. In northern Côte d'lvoire, Parkia and Vitellaria parklands around the village of Dolekaha showed a significant increase in tree density while cultivation area remained stationary between 1962 and 1993 (Bernard et al., 1996). And this parkland development took place during the booming growth of cotton and maize production stimulated in part by input availability. Bernard and colleagues postulate that cash crop production and the associated use of chemical fertilizers have extended the cultivation period, thus contributing to the growth of these parkland trees, which was previously constrained by the short (two or three year) cropping cycles. The growing trade in Parkia and Vitellaria products also favoured their protection.

The area of F. albida parklands is expanding in northern Cameroon, where tree density is increasing in existing parklands, trees are regenerating in relic parkland areas, and new parklands are appearing at the southern edge of the Mandara mountains (Seignobos, 1982). Faidherbia albida progresses in a northwest direction from the Tupuri area toward the Diamaré region. This is partly due to the development of a muskuwaari sorghum variety which can be transplanted from inundated bottomlands to uplands where F. albida parklands are developing. It is also thriving north of Massa country among Bege populations.

In Musey society in southwestern Chad, there is currently a shift from horse breeding to a system of livestock raising and cropping. Villages are becoming more sedentary and these changes are reflected in an increase of F. albida parklands (Seignobos, 1982). Significant F. albida regeneration is also reported south of Zinder, Niger, in areas which were extensively cleared for groundnut production during colonial and early postcolonial times (Taylor and Rands, 1991).

Fertilizer policies

Agricultural policies in the late 1960s and 1970s which emphasized the use of green revolution technologies, including improved varieties and mineral fertilizers, also had a significant impact on tree management practices in some areas. The economy of Senegal's North Central Peanut Basin has relied on peanut production since its introduction in colonial times. Until the early 1980s, emphasis was placed on state intervention through input and credit provision, setting of commodity prices and control of industry and agricultural development agencies. Seyler (1993) found that the impact of fertilizers tended to substitute for the fertility effect of F. albida where it occurs. Low fertilizer prices and a state-protected agricultural sector inadvertently discouraged farmers' continued investment in F. albida regeneration in western Senegal (Seyler, 1993).

The analysis of United States Agency for International Development (USAID) data from 1965 to 1989 revealed a significant relationship between fertilizer subsidy and area cultivated in peanuts in the Départements of Thiès and Diourbel. Both increased between the mid-1960s and the late 1970s and then stabilized in the early 1980s in the transitional period when the first structural adjustment loan was approved but reforms were not fully implemented. Finally both variables decreased significantly with the implementation of the structural adjustment programme in the second half of the 1980s which focused on state disengagement, commodity diversification, and privatization. Questionnaire data collected in the same region suggested that the area cultivated in peanuts declined between 1985 and 1989 while the area under cereals and cowpeas increased, and the use of manure and F. albida regeneration in fields increased significantly (Seyler, 1993).

Agricultural development policies have also had a drastic impact on parkland systems…

Based on retrospective estimates of tree age from diameter measurements, F. albida densities were shown to be inversely related to the amount of fertilizer subsidy. In a first stage, farmers favoured the use of cheap fertilizer for maintaining soil fertility rather than emphasizing the conservation of F. albida cover in their fields. As fertilizer subsidies were reduced or eliminated with the new agricultural policy in the mid-1980s, they resorted to traditional methods of fertility conservation such as manuring and protection of F. albida trees. Furthermore, farmers reported that the close row spacing in peanut cropping makes intercropping with F. albida impractical and that, unlike the hand harvesting of millet and sorghum, the mechanical harvesting of peanuts also results in the loss of F. albida seedlings. Therefore the decrease of peanut cultivation area associated with structural adjustment has had a positive impact on F. albida parkland densities (Seyler, 1993). Farmers also seemed to respond to the high input cost and resulting decrease in peanut income by protecting and increasing the fruit production of other on-farm tree resources. For instance, Ziziphus mauritiana fruits are sold in local markets as well as in Thiès and Dakar, and fruits of Balanites aegyptiaca are sold to oil pressers in the area (Seyler, 1993).

… for example by encouraging the clearance of trees to facilitate animal traction.

The above results were confirmed by a financial analysis of technological interventions for the region, including various scenarios of intensified (high input) agriculture or F. albida parkland technology. Analyses revealed that the net present value of the introduction/maintenance of F. albida was 40 percent higher than the fertilizer/high input model. Therefore, when fertilizer is inexpensive or highly subsidized, it makes more financial sense for farmers to spread low-cost fertilizer in their fields than to maintain F. albida populations. In contrast, farmers rely on F. albida parkland or other traditional soil conservation technologies when fertilizer is accessible only at market prices. General agricultural and economic policy had more influence on the conservation and improvement of the agroforestry parkland system in Senegal's North Central Peanut Basin than policies and projects focusing on the system itself (Seyler, 1993). It is also recognized that in Zambia and other southern African countries the removal of government subsidies on inorganic fertilizers led to the need for Sesbania sesban-based improved fallows (ICRAF, 1995a). Consequently, macroeconomic policies, including those concerned with agricultural input pricing, need to be carefully designed in order to ensure sustainable agriculture production and the maintenance and improvement of the natural resource base.

The analysis of F. albida parkland dynamics in this and the previous section does not, however, imply that mineral fertilizers and F. albida have similar functions and are mutually exclusive, and that the development of the one will lead to the decline of the other. The organic carbon and nitrogen inputs of F. albida can be considered complementary to the mineral inputs of NPK fertilizers. Furthermore, the species is found not only in nutrient-deficient soils, but is also maintained in fertile sites throughout the Sahel (Depommier, 1996a).

Can the results of Senegalese F. albida parklands be extrapolated to other areas and other parkland systems? That such policy changes could have a positive effect in other systems such as cotton cultivation in Vitellaria and Parkia parklands, is less obvious. The F. albida parkland system in western Senegal has several peculiar characteristics. The tradition for maintaining these parklands is deeply rooted. (However, the departure from subsidized fertilizers also had a positive effect on Wolof farms whose agroforestry tradition may be less deeply rooted.) Secondly, the enriched soil fertility and improved crop production which could be expected from parkland species other than F. albida may not compare with those observed in the almost monospecific stands of F. albida in Senegal. Nevertheless, policies may similarly promote soil conservation or fertilization practices as a complement to agroforestry technologies. To the extent that appropriate agricultural policies may activate parkland agroforestry practices the question remains of interest, and the particular conditions warranting such changes are worth studying in detail throughout the zone of parkland occurrence.

Demographic relations

By shortening or eliminating fallows, increasing population pressure threatens the sustainability of Vitellaria paradoxa and Parkia biglobosa parklands.

There appears to be a clear association between population density and particular agroforestry parkland types. This section analyses this for the two most studied types: F. albida parklands, on the one hand, and V. paradoxa and P. biglobosa parklands, on the other. While this may clarify the peculiarities of each system, one should nevertheless bear in mind that given villages or regions often display a combination of both types.

There are many indications that F. albida parklands are able to sustain high population densities. Hallaire (1976) referred to systems including permanently cultivated fields under F. albida cover on terraces of the Mandara mountains in northeastern Nigeria and northwestern Cameroon and stall-fed cattle, with population densities as high as 300 inhabitants per square km. Recent population increase and development of rainfed agriculture around the Mandara mountains resulted in greater conservation and improvement of parklands which were previously abandoned during conquest by Fulbe herders (Seignobos, 1982). In the Hararghe Highlands of Ethiopia, F. albida parklands are common, due to a combination of high human population density, scant availability of uncultivated land and restriction of free grazing. However, these parklands are almost non-existent in the neighbouring Chercher highlands where different conditions occur (Poschen, 1986).

Likewise, in the Jebel Marra massif in Sudan, the degradation of the intensive F. albida system practised by the Fur was primarily attributed to a decline of population density over past decades (Miehe, 1986). Older villagers in the Koronga area reported that, 50 years ago, the population density was higher and the area cultivated permanently for more than 20 years was larger, while tree density in fields remained high and species composition constant. Currently, one may observe a relative scarcity of old trees on farmland and a lack of tree regeneration. This process is aggravated by a combination of socio-economic factors and by droughts. Traditional values, customs, and diets as well as the authority of local chiefs have eroded with monetization of the economy, commercialization and wage labour. With political security, part of the population leave the montane areas where the intensive F. albida parkland system is practised and move towards the plains to seek wage labour (Miehe, 1986). As pressure on land decreases, the remaining farmers adopt more extensive bush fallow or shifting cultivation.

Pélissier (1980a) also highlighted the fact that the Sérer in Senegal (700 mm rainfall) and the Brame and Mandjak in Guinea Bissau (1 500 mm rainfall) established a F. albida parkland system in order to support their high population densities of 50–100 and 100–150 inhabitants/km² respectively. In contrast, the Balant group, a similar sedentary agropastoral society located between the above-mentioned regions in Casamance, Senegal, did not develop F. albida parklands. And the Massa Goumay group, while having created F. albida parklands around Koumi on the Logone river, did not reproduce them after migrating to the Chari banks where population density was three times lower (Seignobos, 1982). In 36 villages in the administrative districts of Thiès, Diourbel, Fatick and Kaolack in Senegal's Peanut Basin, Seyler (1993) found a positive relationship between population pressure ranging between 50 and 100 inhabitants per square kilometre and F. albida crown cover, regeneration rates, and seedling survival in agricultural fields. In Niger, Lemaître (1954, cited in Felker, 1978) observed several cases where well-stocked F. albida parklands were associated with higher population densities (25–40 people/km²) than areas with no or little F. albida farm component (10–20 people/km²). In these cases, farmers appear to increase their investment in labour and soil fertility enhancement activities with rising population pressure.

In specific conditions, and increase in population density appears to have opposite effects in Faidherbia albida and Vitellaria-Parkia parklands.

Faidherbia albida parklands are suited to permanent cultivation. The alternation of short cultivation and fallow intervals is detrimental to them. In fallows, field borders or grass strips, F. albida's development suffers from competition from perennial grass species during the late part of the growing season, and fire. It is often absent in uncultivated zones of savanna, even those bordering fields (Peltier, 1988, cited in Bernard et al., 1996; Harmand et al., 1996). According to farmers, promoting the development of large F. albida trees in parklands is also a way of controlling its invasive development. Livestock grazing, the cutting back of F. albida shoots during field preparation, as well as ploughing operations cause persistent heavy resprouting and suckering (Bernard et al., 1996; Depommier, 1996a; Seignobos, 1996). Large trees appear to limit the number of seedlings and suckers and are preferred to control this undesired regeneration. And being adapted to permanent cultivation, F. albida parklands can support relatively high human population levels.

In contrast, Vitellaria and associated Parkia parklands rely on alternating fallow and cultivation periods. Naturally, this implies that the number of people they can sustain per unit area is relatively low. Fallow phases allow for seedling establishment and early growth as well as soil fertility restoration, while subsequent selection and land cultivation result in relatively good growth due to protection from fire and perennial grass competition. The traditional sequence of cultivation and fallow periods as it took place before colonization was reconstituted in southern Chad (Gilg, 1970). Villages would adopt a 4-year cropping cycle followed by a 12–15 year fallow, which was repeated two or three times before the population would move to nearby sites to allow for restoration of soil fertility. Gilg (1970) also indicated that villages would return to sites cultivated in the past. The persistence of V. paradoxa in burned fallows is also partly due to its resilience to fire (Hopkins, 1963) and its fire-adapted germination patterns (Jackson, 1974).

In turn, increasing population, concurrent decrease in available arable land, and higher cropping intensity shorten or eliminate fallow periods, thus threatening the sustainability of V. paradoxa and P. biglobosa parklands. For instance, bush fields in Guetela and N'Tossoni in southern Mali have been cultivated permanently for the last 10 to 30 years and 10 to 15 years, respectively. This is in contrast to the traditional rotation of five years of cropping and seven to ten years of fallow which is still observed in Pourou, a village with less intensive agriculture (Bagnoud et al., 1995a). While permanent (intensive) cultivation contributes to the maintenance of existing trees, it impedes tree regeneration. Thus small diameter (10 and 20 cm) size classes represented 32 and 30 percent, 8 and 7 percent, and 13 and 0 percent of all trees in Pourou, N'Tossoni and Guetela, for Vitellaria and Parkia species, respectively. For the same reason, overall Vitellaria and Parkia tree densities are usually lower in compound than in bush fields. Small diameter classes, however, may be more common in bush fields, which are usually fallowed, than in permanently cropped compound or village fields (Dallière, 1995; Mahamane, 1996; Ouédraogo and Devineau, 1997). Vitellaria acquires a suckering ability after about four years of sufficient root and stem growth (Ouédraogo, 1994), and cultivation before this stage destroys it. While traditionally there is a contrast in tree density between compound and bush fields, the acquisition of carts tends to extend agricultural intensification to bush fields as well (Bagnoud et al., 1995a). The presence of large trees is related to permanent site occupancy and long cultivation and fallow periods. However, old Vitellaria individuals seem not to be present in very old fallows. Nandnaba (1986) ascribed this phenomenon, observed in Nazinga, Burkina Faso, to their poor resistance to competition. In the light of Nandnaba's results, the traditional practice of long-term (up to 150 to 200 years) fallow resulting from long-term shifts in village location suggested by Gilg (1970) would imply that Vitellaria trees would no longer be dominant at this stage but would be part of a mixed woodland.

While the above paragraphs have contrasted the capacity of two parkland types to sustain population density, it should not be concluded that high population pressure necessarily leads to the degradation of parklands relying on farm-fallow cycles. The established conviction in many government forest policy bodies that population settlement and land use result in forest degradation is, increasingly, being disproved. In Kissidougou, Guinea, forest island density in 45 districts was positively correlated with population density (Fairhead and Leach, 1996). Similarly, there was a positive correlation between population density and the volume of planted woody biomass on farm at the district level in Kenya (Holmgren et al., 1994). Within reasonable limits, farmers' response to rapid population growth can therefore lead to more intensive sustainable land management practices.

Summary

The spread of parklands has probably increased in line with the expansion of the cultivated area throughout the Sahel and Sudan zones of West Africa in the last decades. At the same time, a combination of qualitative and quantitative data suggests that parklands have been degraded in terms of tree density and that a lack of regeneration threatens their sustainability. However, decline of parkland cover is not uniform and these systems display a high degree of resilience. There is a deplorable lack of the comparative density and age composition data at distinct time intervals necessary to assess their dynamic condition. A decline in tree density is sometimes inherent in farmer strategies for long-term parkland development to reach appropriate tree-crop combinations and high fruit production, and does not necessarily imply degradation.

Parkland systems respond dynamically to a variety of natural, economic, socio-cultural, technical, demographic and political parameters. Factors including drought, pests and exotic tree species have exerted pressure on parkland species, whereas livestock contributes to seed dissemination and survival. When farmers perceive that trees and their products become valuable due to increased economic worth, greater demand or declining availability, they actively invest in the protection and reproduction of parklands. They also support stronger institutional arrangements and maintain the necessary knowledge base for the management of these systems.

In contrast, farmers tend to neglect their forest resources and favour alternative agricultural practices, items of consumption and income-earning activities when these yield higher benefits in terms of lower cost, higher revenue, lower labour expenditure, higher subsistence priority, better taste, etc., than those from surrounding parkland trees. External parameters such as markets, external pressure on village resources, migration and relations with urban centres strongly influence the relative value of parkland trees. In recent decades, socio-cultural motives, which have been a strong factor for the perpetuation of parkland agroforestry practices, have tended to lose ground against the monetization of the economy and associated technological change, leading to parkland recession.

Agricultural development policies have also had a drastic impact on parkland systems. Until recently, research, development, extension and commodity parastatal agencies encouraged the clearance of trees to facilitate animal traction. Superficial tree roots can impede the progress of ploughs, and most of the data available point to lower tree densities on farms equipped with animal traction than among manual cultivators. However, animal traction is far from incompatible with parkland agroforestry, as demonstrated by some fairly high parkland densities reported in Burkina Faso and Côte d'lvoire. Cash crops, such as cotton, groundnuts and maize, were also introduced with the recommendation to clear parkland trees. They tend to displace traditional staple crops and replace fallow practices, which are responsible for tree regeneration, with the use of fertilizers. Alternatively, the use of fertilizers sometimes allows for longer cultivation periods between fallow cycles during which parkland trees can thrive. Cash cropping may encourage more exclusive control of fields and exacerbate tenure conflicts where land is used or claimed by multiple actors. When the historical basis of land access is in question, the conservation of trees, interpreted as a sign of past occupancy, may be jeopardized. Fertilizer policies may also have a significant impact on traditional fertility maintenance and tree management practices. There are indications that removal of agricultural input subsidies in the late 1980s fostered F. albida regeneration in Western Senegal.

An increase in population density appears to have opposite effects in Faidherbia albida, and Vitellaria-Parkia parklands. The Faidherbia system thrives with permanent cultivation, a virtual absence of fallows, and the intensive cropping practices resulting from high population numbers on a limited land area. As the need for intensive land use disappears with released pressure on land, the Faidherbia albida parkland system tends to dwindle. In contrast, the Vitellaria-Parkia parkland type depends on the alternation of fallow periods, which permit soil fertility restoration and tree regeneration, with cultivation intervals favouring the growth of selected trees. By shortening or eliminating fallows, increasing population pressure threatens the sustainability of Vitellaria paradoxa and Parkia biglobosa parklands.