Present distribution of forest and savanna can be interpreted by looking for parallels with environmental conditions at various spatial scales. Among them, climate and edaphic conditions determine water availability and are of paramount importance in the competition between the two vegetation types. The outcome of competition is modified by disturbances such as fire and shifting cultivation. In the following paragraphs we discuss these main environmental factors, and how they operate at different spatial and temporal scales. The ecology of savanna per se is far beyond the scope of this chapter and can be found elsewhere in the literature (e.g. Huntley & Walker 1982, Bourlière 1983, Walker 1987).
There is a general agreement that climate is the primary factor explaining the distribution of forest and savanna, at least at a continental scale (Adejuwon 1971, Swaine 1992, see also Furley 1992 for the Neotropics). The main limiting factor for the presence of forest is water availability, often expressed in terms of total annual precipitation and the length of the dry season. For instance, Aubréville (1962) states that the lowland tropical forest is mainly found in areas receiving more than 1400 -1500 mm of annual rainfall, but that this limit can be lowered to 1250 mm if the precipitation is evenly distributed throughout the year. A place with a dry season of 1 - 3 months (with monthly precipitation below 30 mm) can support forest depending on annual rainfall and soil type. According to Adejuwon (1971), the forest-savanna ecotone in western Nigeria is comprised within the 80 -100 rainfall days isopleth, and the 1250 mm mean annual rainfall isohyet. The 25 mm mean annual water deficit and the 1300 mm mean annual evapotranspiration isopleths all have the same shape and pass through this zone.
Edaphic conditions are of paramount importance in determining the forest-savanna equilibrium at a regional scale especially in areas where climate conditions are intermediate. In west-central Ghana, the limit between continuous forest and forest savanna mosaic is congruent with substrate: savanna dominates on Voltaian rocks and the forest dominates on the Birrimian basement complex, on which soils are richer in clay and nutrients (Swaine et al. 1976, Swaine 1992). A similar situation is found in Côte d'Ivoire, at the eastern side of the V-Baoulé where the boundary between continuous rainforest and forest-savanna mosaic corresponds to the boundary between schistous and granitic substrates, the former giving rise to soils being also richer in clay and having thus a better water-holding capacity (Spichiger 1975).
At the local scale, most authors agree that soil types (which are generally distributed repeatedly along successive toposequences) are of primary importance to explain the patchy distribution of forest and savanna (Morgan & Moss 1965, Aubréville 1966, Latham & Dugerdil 1970, Adejuwon 1971, Furley 1992). Savanna inclusions are always found on sandy, phosphorus-deficient soils, or very shallow, waterlogged soils (Adjanohoun 1964). In west-central Ghana, vegetation types were found to occupy distinct catenary positions and to be associated with characteristic soil types. Grassy, treeless savannas are often found on alluvial soils that are waterlogged in the wet season and lack water in the dry season. Levels of total phosphorus, exchangeable calcium and potassium, water content, and pH in the topsoil were all found to be higher under forest than under savanna. These differences in soil fertility, which correlate with tree cover and stature, were however suspected to be of secondary importance. The primary factors separating vegetation types are probably effective soil depth, soil texture and water supply (Markham & Babbedge 1979). Soil properties can also be transformed in feedback with the vegetation above it, by changes in soil fauna (e.g. termite species) and by cultivation and settlements (Fairhead & Leach 1996, 1998).
Fire is a key element in understanding the distribution and dynamics of savannas and the transition zone with forests. Although natural fires occur through ignition of the herbaceous layer of savannas via lightning, sparks from falling rocks and fermentation (Phillips 1974), almost all savanna fires are currently anthropogenic of origin. Indeed, man's use of fire has led to an increase of fire frequency in these systems which can hardly be considered natural and each year savanna fires run over almost the whole surface of the transition zone. However, the existence of fire-adapted flora confirms the importance of fire as a system determinant, even before man mastered fire (see also Bond & Van Wilgen 1996). Reviews on the effect of fire on the savanna ecosystem can be found in Phillips (1974), Gillon (1983) and Trollope (1984).
Forest pioneer species whose propagules originate from the forest elements scattered in the landscape are very aggressive in invading savannas and are frequently found as seedlings or saplings. These are generally found within a few metres from the forest edge but also sometimes several hundred metres apart depending on their dispersal mode (anemochory or zoochory) and site characteristics (longdistance establishment of forest species occurs in savannas with a high tree cover) (Spichiger 1975).
Several long-term experiments on the effect of fire have been initiated in the transition zone of West Africa. In Kokondekro, near Bouaké in Côte d'Ivoire, three 1-ha plots have been subjected to different fire regimes since 1937 (Figure 3.4). Annual early burning (beginning of the dry season) has promoted a shrub savanna (Figure 3.4A). Annual retarded burning (mid to late dry season) has promoted a savanna with a very loose shrub cover because of a greater fire intensity (drier fuel) and the fact that it destroys tree leaves that establish long before the rainy season (Figure 3.4B). Integral protection from fire has led to a genuine forest (Figure 3.4C) composed mainly of dense semi-evergreen (Guineo-Congolian) forest species (Monnier 1981). A similar experiment is being carried out in Lamto (south of the V-Baoulé) where a c. 60 ha savanna has been protected from fire since 1962. The results clearly show the same trend, although vegetation succession is in a less advanced stage (Vuattoux 1970, 1976, Devineau et al. 1984). In northeast Ghana, another fire protection experiment has led to a clear decrease of grass cover and a clear increase in tree number and stature, but on contrary to the Kokondekro experiment, there has been no establishment of Guineo-Congolian forest species. This is explained by the much more northern position of the plot, north of the transition zone, in a region where dense forest elements are very rare and thus forest seeds unavailable (Brookman-Amissah et al. 1980).
A reverse experiment has also been conducted. There has been an attempt to create a savanna inside the forest zone at Adiopodoumé, south Côte d'Ivoire, by felling and burning the original vegetation, sowing savanna species and burning the vegetation regularly. The result was a relative failure: there was a massive colonisation by ruderal forb species, which made it difficult to ignite fire and burn the vegetation (Adjanohoun 1964).
There is a general agreement that the savannas of the transition zone are not in equilibrium with climate and are a direct consequence of annual fires. The climatic climax would be semi-deciduous rainforest and further north (from c. 9°N) dry forests and woodlands. Guinean savannas are considered to be of secondary origin. This is also reflected in names such as: "savane de substitution" or "derived savanna" (Keay 1959). They are to be considered as a "fire-climax". It has been suggested by Aubréville (1966), Monnier (1981), and Swaine (1992) that before the massive use of fire by humans, the transi-tion was much more gradual; "in a primitive state there were quite extensive woodlands [...] intermediate between moist forest and myombo woodlands" (Keay 1959).
From the evidence accumulated above it is clear that fire plays a very important role in the resilience of savannas to invasion by forest species. It is also clear that fire reduces the density of savanna trees. The fuel biomass is lower under the shade of the tree crowns, and forest pioneer species have therefore better chances to become established. Fire also favours the herbaceous layer of savanna by promoting the resprouting of perennial grasses and the germination of annual savanna species (Hopkins 1963). In turn, a dense herbaceous layer increases fuel availability and thus dry season fire intensity.
Fires can also occasionally sweep under the canopy of the drier types of rainforest which are to be found in direct contact with burning savannas (for Ghana, see Swaine 1992). These fires are generally low in intensity
and are essentially fuelled by litter (ground fires, Phillips 1974). They extend to the canopy only during exceptionally dry conditions (Swaine 1992). Like logged forests, burnt forests are more prone to renewed burning (Hawthorne 1991). In Ghana, fires have recurred in several years since the main outbreak in 1983. Consequently, regrowth becomes predominantly herbaceous, but it is uncertain if such continued pressure will convert forest to true savanna (Swaine 1992).
Animals can have an impact on the forest-savanna transition through pollination and the dispersal of plant species, but also through seed predation, browsing, and the regulatory role of top predators on herbivores (Medellin & Redford 1992). Other vertebrates and especially birds play also key roles in these processes. In Africa, large populations of herbivores can open up the landscape (Kortland 1984).
Soil fauna is also of great importance. Termites have a large influence on nutrient cycling and soil displacement. Their mounds are often preferred sites for establishment of woody species (Abbadie et al. 1992, Howse 1992). Predation on colonies by specialised mammals or ant species ultimately result in mound decay and release of nutrients to the soil.
Apart from the intensification of the fire regime, there has been considerable debate on the impact of human activities and especially traditional shifting cultivation on the distribution of forest and savanna. It was generally assumed that shifting cultivation would lead to an extension of savannas by destroying woody vegetation (Aubréville 1966). In South America slash and burn agriculture, fire and deforestation have led to savannisation (see for example Kauffmann & Uhl 1990, Uhl & Kauffmann 1990, Miller & Kauffmann 1998). However, in the Rio Branco-Rupununi region of the northern Amazon, forest regrowth generally occurs along the forest-savanna boundary, in spite of persistent shifting cultivation (Eden & McGregor 1992). Field studies in Côte d'Ivoire, have even revealed that shifting cultivation at the forest edge actually promotes afforestation (Spichiger & Blanc Pamard 1973); field preparation reduces herbaceous domination in the fallow, and thus limits fuel availability. As the impact of fire is reduced, forest pioneer species can establish. Similarly, in southwest Togo, forests regenerate as well in savanna fallows (Guelly et al.1993).
Studies in heavily populated countries like Nigeria have widely questioned the origin of the derived savannas (Keay 1959). It has been suggested that at Ibadan, some 50 km inside the rainforest zone, savanna is derived from secondary forest. "In the absence of continual bush clearing for cultivation it is doubtful if fire alone could maintain the savanna indefinitely" (Clayton 1958). Aubréville (1966) was also convinced that cultivation (and fire) could explain the present distribution of savannas, like for instance for the southern extension of savannas in central Côte d'Ivoire (V-Baoulé). Monnier (1981) argued that this was not likely, given the low population density and the kind of tools that were available. Nevertheless, in some places like the western extension of the V-Baoulé, there is general agreement that the Pennisetum purpureum savannas are derived from cultivation of forests (Adjanohoun 1964). Fairhead and Leach (1995, 1996) have widely questioned the interpretation of the environmental history of the forest-savanna mosaic. According to the colonial and post-colonial governments in Guinea, the once continuous forest over the region of Kissidougou was cleared by local population and Malinke migrants. Based on oral and historical sources they demonstrate that this interpretation is not true. Instead, most of the human activities are promoting afforestation (farming system, deliberate planting of forest species in fallows, pasture, fire-breaks). Based on aerial photographs, they show that this has led to a 40% increase in forest cover during the last 50 years. In the northern part of the mosaic in Ghana, studies have demonstrated that even under an increasing population, forest losses and gains were in a kind of steady-state equilibrium (Amanor 1993, Afikorah-Danquah 1997, both cited in Leach & Fairhead 2000).
Increase in population pressure as well as transformation of farming systems toward cash crop production has completely altered the intensity of human impact on vegetation during the last decades. Comparing aerial photographs in the north of the V-Baoulé in the years 1952, 1969, and 1975, Spichiger and Lassailly (1981) indicate that during the first time interval, vegetation changes expressed its "internal dynamics" with a tendency towards afforestation. During the second interval, the area of cultivated land increased at the expense of both savanna and forest. Especially, the duration of the fallow in the farming system is of paramount importance. If the fallow period is reduced from c. 10 years to c. 5 years, as is experienced in many places in the forest-savanna transition zone, then afforestation processes can no longer take place.
Large-scale deforestation in the northern part of the forest zone through logging, charcoal production and slash and burn agriculture can dramatically enhance fire hazard, especially when one takes into account the subsequent alteration of microclimatic conditions. Colonisation by grasses is likely to occur, further increasing fire hazard. Such processes have been studied in the Amazon basin (Uhl & Kauffmann 1990). It is uncertain if such processes take place in West Africa. Attempts to correlate occurrence of fires with broad-scale land cover changes using remote sensing data did not give conclusive results (Ehrlich et al.
As pointed out by Leach and Fairhead (2000), "population-forest relationships [are] variable, non-linear and unpredictable", especially taking into account changes in population pressure, the evolving strategies and environmental global changes.
It makes no sense to try to explain present distribution of forest and savanna by one of the above factors only. All published works point at the importance of the combination of factors. There are even secondary effects of vegetation on soil at the local scale or on climate at the global scale that are very difficult to take into account. Distribution of vegetation is also of primary importance on human choice of where to set a field or where to build a village (Adjanohoun 1964). One can only get an idea of the effect of a single factor on the forest-savanna boundary, as all factors are intimately intricated in a way that their respective weight depends on the others.
Furthermore, it has been widely recognised that historical factors must be taken into account to explain the present distribution of forest and savanna. Climate and human pressure have not been constant in the late quaternary and vegetation tend to show considerable resilience towards environmental changes.
Was this article helpful?
You Might Start Missing Your Termites After Kickin'em Out. After All, They Have Been Your Roommates For Quite A While. Enraged With How The Termites Have Eaten Up Your Antique Furniture? Can't Wait To Have Them Exterminated Completely From The Face Of The Earth? Fret Not. We Will Tell You How To Get Rid Of Them From Your House At Least. If Not From The Face The Earth.