By
J.I. House & D.O. Hall
Division of Life Sciences, King's College London

Water availability
Nutrient availability

Fires
Herbivory

As noted earlier, the amount of moisture and nutrients available to savanna plants determines the structure of savannas and their NPP. The primary determinants of both of these factors are climate and soil type; rainfall and soil texture being particularly important. In low rainfall areas, deep, sandy soils allow more water to infiltrate to greater depths, reducing run-off and evaporation and increasing moisture availability. However, where rainfall is high, sandy soils can lose water through drainage and run-off often leaching nutrients. Clay soils have a higher water-holding capacity and nutrient retention leading to a higher NPP in moist and mesic systems but are less advantageous in arid areas due to low infiltration (Dodd & Lauenroth, 1997).

Moisture and nutrient availability in turn influence the occurrence and intensity of fire and herbivory which are secondary determinants modifying savanna vegetation (Walker & Noy-Meir, 1982; Frost et al, 1986; Solbrig, 1990; Scholes & Hall, 1996). At a more local scale differences in topography, geomorphology and management lead to further differences in structure and floristic composition (Solbrig, 1996). Many of the factors controlling savannas are interdependent, interacting positively or negatively on each other either simultaneously or sequentially. The relative importance of individual factors shifts with time and location leading to the dynamic mosaic of savanna types and NPP. Light is generally sufficient in the tropical savannas except when canopies become dense.

Water availability is widely considered the most important controlling factor. Rainfall determines the supply of water, but the amount that is subsequently available to plants (it’s effectiveness) depends on drainage and storage (soil texture, depth and topography) and losses due to evaporation and evapotranspiration (climatic conditions, vegetation cover, etc.). Annual potential evaporation substantially exceeds annual rainfall, therefore rainfall is, in a sense, less effective than in temperate climates. Rainfall and evapotranspiration are highly seasonal and therefore annual means can be misleading.

Rainfall mostly occurs as short-duration, high-intensity convectional storms. Variability is high, increasing with aridity, and is a primary cause of compositional change. Rainfall may be bimodally distributed and the timing affects the species mix, survivability, extent and overall growth (Pandey & Singh, 1992; Veenendaal; et al, 1996). Many areas regularly experience drought and have developed a range of adapted and tolerant species. Plants close their stomata to reduce water loss during dry periods, which also restricts CO₂ uptake and hence productivity. In some savanna areas flooding is common and can lead to a long-term reduction in production due to soil erosion (Rutherford, 1978).

Many studies in arid areas have shown a high correlation between productivity and rainfall. However, NPP can often be more closely related to the length of the wet and dry seasons than annual rainfall as this is a better indicator of water availability. Scholes & Hall (1996) calculated that semi-arid savannas, where the duration of the growing season is about 100 days, have a total (above- and belowground tree and grass) NPP of around 500 g (DM) /m²/y, while very moist savannas (300 growing days per year) have an NPP of around 3000 g (DM) /m²/y, see Figure 3.

Savanna soils tend to have low cation exchange capacity (CEC), very low phosphorous and nitrogen contents, and high aluminium and iron contents (Solbrig, 1996). More than half of the tropical savanna soils are derived from old, highly-weathered acid crystalline igneous rock leading to leached sandy soils with low fertility and CEC. There are also extensive areas of basic igneous rock forming base-rich clays with more favourable nutrient content and retention (Solbrig, 1996). Local conditions, age and history lead to a complex regional distribution. Soil catenas are common with dense woodlands on sandy upper areas, through mixed scrub on shallow mid-slope soils, to open tall-tree savanna on deep fine-textured soils at slope bottoms (Walker, 1985). Iron pans often occur in arid areas (Menaut et al, 1985).

Nutrient mineralisation, transport and root uptake are all dependent on soil water content. Thus it has been suggested that water availability may actually be controlling plant growth by its influence on nutrient availability, and that, certainly in some areas, nutrients rather than water may be the major production constraint. This is borne out by the high carbon to nutrient ratio of many savanna plants, various fertilisation experiments, and comparisons of NPP on adjacent soils with different fertility status (Scholes & Hall, 1996).

High rainfall can leach soils causing a gradient from arid/eutrophic to moist/dystrophic savannas seen clearly in Africa (Huntley,1982). Arid, sandy savannas have a slow release, low nutrient stock favouring perennials with low growth rate over a long period. Clay soils release nutrients in pulses after rain favouring annuals with rapid regrowth for short periods. In arid savannas, litter production plays a major role in nutrient cycling while moist savannas tend to have a short, closed nutrient cycle bound closely to root decomposition so the amount of nutrients in the soil at any instant may be misleading (Abbadie et al, 1992). Nutrients are released steadily throughout the year in humid areas favouring perennials, although there is still a nutrient pulse at the beginning of the rainy season as rain washes the nutrients from litter decomposition and fire ash into the soil (Menaut et al, 1985).

Low nitrogen and phosphorous availability constrain many savanna ecosystems, yet little is known in detail of the nutrient dynamics of tropical savannas (Breman & De Wit, 1983; Medina,1987; Menaut et al, 1985; Solbrig et al, 1996). Most nitrogen is lost through pyrodenitrification with frequent fires and an accumulation of decomposition-resistant charcoal. Leguminous plants are common, but may fix little nitrogen on low-fertility soils, possibly due to a deficiency of co-factors (Zietsman et al, 1988). Nitrogen reallocation from leaves can be high, thereby retaining nitrogen stocks before leaf fall and fire.

Earthworms can process up to 70% of soil organic matter in upper horizons (Lavelle, 1983; Menaut et al, 1985). Termites are more dominant in drier climates and are very efficient secondary consumers, probably more important than the herbivores, generally consuming around 30% of litter biomass, but this can reach 70%. Termite biomass often averages 10 g/m² of fresh weight with maximums up to 50 g/m² which is comparable to large herbivores (Wood & Sands, 1978). Nutrients (and seeds) become concentrated in large termite mounds. Herbivory also transforms and concentrates nutrients (Scholes & Walker, 1993).

Fires are inevitable with the build up of dead grass during the dry season. "Natural" (lightning-induced) fires occur every 1 to 3 years in humid savannas and perhaps 1 in 20 years in arid savannas (Frost, 1985; Walker, 1985). Fires are deliberately set or restricted by humans as part of vegetation management which has a history of at least 50,000 years in Africa (Menaut et al, 1985). In Africa, 25 to 50% of the total land surface is burned annually in the "Sudan Zone" (arid) and 60 to 80% in the "Guinea Zone" (humid) (Menaut et al, 1991). Fire is considered necessary to the functioning of savannas which have evolved with it, preventing bush encroachment, removing dead material and recycling nutrients (Troloppe, 1982; Walker & Noy-Meir, 1982). Fire prevention leads to organic matter and litter accumulation, and increased tree density, often resulting in the long term in a forested area without grasses. While fire leads to local and short-term heterogeneity, creating a mosaic structure of open grassland which carries fire, and clumps of trees which lack the herbaceous fuel load to do so, it also maintains regional and long-term homogeneity and stability (e.g. maintaining a sharp boundary between savanna and forest) (Menaut et al, 1990; Menaut et al, 1985).

The incidence, intensity and impact of fire depends on the amount of fuel (grass) present, the prevailing environmental conditions and thus the time (season) of burning (Trollope 1982, 1984; Menaut et al, 1993, 1995). Fire removes aboveground herbaceous vegetation, small/immature woody plants and changes the microclimate and nutrient status. This creates opportunities for seedling development, enhanced reproduction and fresh growth. Mature trees are not usually affected but total woody biomass is controlled in the long term by reducing recruitment. Underground plant parts are only affected by very hot fires, although root:shoot ratio increases in regularly burnt areas. (Troloppe, 1984). Humid plants show many fire adaptations such as withdrawal of plant parts underground, increased (fire-resistant) seed production, and fire-induced phenology.

Fire-induced mortality of plant populations is often extremely low (0-10%, Frost, 1985). Early season fires burn 20 - 25% of the aboveground herbaceous biomass, and late fires burn 60 - 95% (Menaut et al, 1991). Since much of the total system NPP is belowground or in tree leaves, less than 20% of the total NPP is lost in annually burnt savannas (Scholes & Walker, 1993), and only 5% in triennially burned savannas (Scholes & Hall, 1996). In the long-term, frequent burning, tends to increase herbaceous production in humid regions is by perhaps 20 - 50% while decreasing it in arid regions (San Jose & Medina, 1975; Menaut et al, 1993). In fact, the NPP of annually burnt humid savannas is comparable to that of rain forests under similar climatic and edaphic conditions (Scholes & Hall, 1996).

Fire releases carbon, nitrogen and other elements that are partly returned to the soil as ash. The long-term effect on nutrient budgets is unclear. It is suggested that savannas which have experienced frequent fires for thousands of years may have been driven to a low-equilibrium nutrient status, but on an annual basis losses of N and C are balanced by annual inputs. Incorporation of charcoal residues in the soil may lead to carbon accumulation in the long-term (Menaut et al, 1993).

Savannas support large numbers of herbivores, both grazers and browsers. Browsers consume various parts of woody plants and grazers prefer herbaceous biomass, although diets are often mixed and depend on food availability. Insects account for an equal, or even greater, proportion of herbivory than mammals, yet they are infrequently studied (Lamotte, 1982; Anderson & Lonsdale, 1990), while mammals have greater impact on savanna structure as they trample, urinate, defecate and trash plants (Skarpe, 1991). Large herbivores such as elephants kill trees thus opening up woodlands and turning savanna into grasslands, while giraffe keep trees at a lower, fire-sensitive height (Pellew, 1983). Large concentrations of grazers can cause degradation and erosion of grasslands by severe, long-term removal of the herbaceous cover. Grazing favours tree seedlings removing competition from grasses and reducing fuel available for fire (Skarpe, 1990).

At the species level, herbivory changes composition and physiognomy, favours annual grasses, and increases annual and perennial forbs (Illius et al, 1996). At the individual plant level, changes in quantity, chemical composition and physiognomy either promote or deter different herbivores. Infertile soils support vegetation with a nitrogen content less than 1% which is below the threshold for ruminant consumption for most of the year and less than 10% of NPP is typically consumed compared to 80% on fertile soils (Drent & Prins, 1987; Scholes & Walker, 1993). Some trees can be entirely defoliated by browsing and produce a new crop of leaves in the same season, almost doubling leaf production (Rutherford, 1978).

Grazing has contradictory effects depending on intensity and local conditions. Moderate grazing removes dead plant material, reduces shade, recycles nutrients, and improves seed dispersal and germination promoting palatable species, NPP and high vegetation cover. For example, Pandey & Singh (1992) found that prevention of grazing in a dry savanna resulted in a decline of NPP (and species diversity), while moderate grazing stimulated aboveground NPP by 4 - 45% and reduced belowground NPP by 25 - 65%. Intense grazing and trampling results in low plant cover, high mortality, low soil nutrients, low infiltration, decline in water availability, and higher erosion reducing NPP, particularly in arid and infertile areas (McNaughton, 1983). As the number of humans living in savanna areas has increased, so has the number of domestic livestock, which putting pressure on savanna systems (Skarpe, 1991).

Many thanks to Bob Scholes (CSIR, South Africa), Xavier Le Roux (INRA, Clermont-Ferrand, France), Jonathan Scurlock (ORNL, USA) and Joe Scanlan (Department of Natural Resources, Queensland, Australia) for providing information and making corrections to the manuscript. Dale Kaiser & Sonja Jones (ORNL, USA) for calculating tropical % of the Olson et al (1983) "grasslands" category.

Sadly, David Hall passed away in August 1999 before this chapter was published. His knowledge and love of savannas was only surpassed by his eagerness to learn and teach.

     If you have entered this site from UNEP.NET or SCOPE, then feel free to use the link below to view the full site.