See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/247848261

The Ecological Effects of Fire in Savannas

Article · January 1987

CITATIONS READS

238 7,459

2 authors, including:

Peter Frost

125 PUBLICATIONS 7,449 CITATIONS

SEE PROFILE

All content following this page was uploaded by Peter Frost on 28 January 2016.

The user has requested enhancement of the downloaded file.

 CHAPTER 5 - FIRE
THE ECOLOGICAL EFFECTS OF FIRE IN SAVANNAS
P.G.H. Frost and F. Robertson

DETERMINANTS 1. INTRODUCTION
Regular fires are one of the characteristic features
OF of tropical savannas. While some are still caused by
lightning (Komarek, 1972; West, 1972) the main source of
TROPICAL SAVANNAS ignition for the past tens of thousands of years has been
man, for hunting, preparing land for cultivation, improving
the quality of grazing for livestock, and controlling the
spread of woody plants (West, 1972). This use has been so
pervasive that some authors suggest that most savannas are
anthropogenic systems derived from deforestation and repeated
Presentations made by savanna researchers burning (Rawitscher, 1948; Budowski, 1956; Clayton, 1961;
at a workshop in Harare, Zimbabwe, Singh et al., 1985). Although such 'derived' savannas are
widespread in the tropics, particularly in the higher
December 1985 rainfall zones, fire is now generally recognized to be only
one of a number of interacting factors affecting savanna
dynamics (Huntley and Walker, 1982; Bourliere, 1983;
Sarmiento, 1984; Tothill and Mott, 1985). Nevertheless, it is
one of the few determinants which can be readily manipulated
and, as such, is an important variable in any management
programme.
In this chapter we emphasize the composite nature of
fire as an ecological factor. Fire behaviour, timing,
intensity and frequency of occurrence all vary somewhat
independently of each other and affect both the environment
and the biota in a number of direct and indirect ways. Fires
reduce plant biomass and litter, thereby altering energy,
nutrient and water fluxes between the soil, plants and
atmosphere. These changes in turn may affect the long-term
nutrient status and productivity of the system.
Fires also kill individual organisms, damage or
destroy unprotected living tissues, modify growth and
reproductive rates, change the availability and use of
resources and alter competitive and other relationships
between organisms. The effects of these impacts depend
Edited by largely on the recent history of a site, the physiological
and developmental state of an organism at the time of burning
Professor Brian H. Walker
and the occurrence of future events such as rainfall, drought
Chief, Division of Wildlife and or herbivory. In the longer term, these effects may result in
Rangeland Research changes to the productivity and population structure of a
CSIRO, P.O. Box 84 species, the composition of communities and, ultimately, the
Lyneham ACT 2602, Australia probability and characteristics of future fires (see recent
reviews by
Published 1987 by IRL Press, Oxford
© The International Union of Biological Sciences
IUBS Monograph Series, No. 3 93
94 P.G.H.Frost and F.Robertson Effects of fire in savannas 95

Coutinho, 1982; Trollope, 1982, 1984a; Gillon, 1983; normally ignited, though it can be severely scorched.
Hodgkinson et al., 1984). The rate of spread of savanna fires is highly
The effective use of fire as a management tool variable, even in a single fire, reflecting differences in
requires a thorough understanding of this complexity of wind speed, topography and the amount and moisture content of
interaction and response. Even though considerable practical the fuel. Recorded mean values for rates of spread range from
knowledge is available on the application of fire in savanna 0.02 - 0.67 m/s (Gillon, 1972; Trollope, 1978; Gandar, 1982;
management, much still needs to be learnt about its different Griffin and Friedel, 1984a). Head fires, burning with the
effects and how these interact with other ecological wind, move faster than fires burning against the wind, and
processes to influence savanna dynamics. In this chapter we this affects both the temperature and duration of fire at a
review some of the ecological effects of fire in savannas, in point and thereby fire intensity (Trollope, 1978).
the context of the following questions: Fire intensity also depends on the amount and type of
1. What are the ecologically important fuel, its moisture content, and prevailing climatic
characteristics of savanna fire regimes? conditions, principally air temperature and relative
2. To what extent does fire influence the environment humidity. Since savanna fires are fuelled largely by grass,
of savannas and, in particular, the fertility of savanna fire intensities vary considerably between seasons, landscape
soils? units and vegetation types. Communities in which there is a
3. How does fire affect the composition, structure, high biomass of grass, such as in valley bottoms or on
and productivity of savanna plant communities? floodplains, generally experience the most intense and
4. To what extent are these patterns influenced by uniform fires. In contrast, in woodlands and shrublands,
interactions between fire and other ecosystem components? where there is generally a lower grass biomass and more
uneven distribution of fuel, fires tend to be less intense
2. SAVANNA FIRE REGIMES and burn more patchily.
The type and intensity of fire, its seasonal Measured fire line intensities, calculated as the
occurrence and periodicity make up the fire regime of an product of the amount of fuel present (or, more correctly,
area. These vary considerably across the range of savanna the amount of fuel actually consumed), its heat yield, and
types. In the moist savannas, fires generally occur during the rate of spread of the fire front, range from 104 - 770
the dry season or early wet season, at intervals of 1 - 5 kJ/s/m in Acacia aneura shrublands and open woodlands in arid
years (Huntley, 1982; Trollope, 1982, 1984a; Hodgkinson et central Australia (Griffin and Friedel, 1984a) to more than
al., 1984). Most dry-season fires are ignited by man but 5000 kJ/s/m in mesic, open woodlands on porous sands in South
lightning becomes important at the beginning of the rains Africa (Trollope and Potgieter, 1983). Even higher
(Komarek, 1972; West, 1972). In the more arid savannas, the intensities can be expected in the tallgrass savannas of the
interval between successive fires is much longer, 5 - 50 tropics.
years, depending on fuel loads and therefore on the Classifying fires in terms of the rate of release of
occurrence of periods of above-average rainfall (Siegfried, heat energy does not provide sufficient information on the
1981; Hodgkinson et al., 1984; van der Walt and le Riche, duration of the heat pulse at different points in the system.
1984). Because of the general aridity of these savannas, Yet both the magnitude and duration of the heat pulse affect
fires can occur at almost any time, provided that there is a the survival of organisms and their propagules, as well as
source of ignition. However, most occur at the beginning of the amounts of nutrient release and volatilization, and
the wet season when the frequency of lightning is highest changes in soil properties. More biologically meaningful
(Siegfried, 1981; Hodgkinson et al., 1984). indices of fire intensity need to be developed.
The majority of savanna fires are surface fires,
burning through the herbaceous layer. Flame heights are
generally low, the mean flame-length of experimental head and
back fires in a savanna grassland averaged 2.8 m (1.2 - 5.0
m) and 0.8 m (0.5 - 1.5 m) respectively (Trollope, 1978).
Plant matter occurring above 3 - 4 m height is therefore not
96 P.G.H.Frost and F.Robertson Effects of fire in savannas 97

3. EFFECTS OF FIRE ON THE ENVIRONMENT

The direct effects of fire on the environment centre
on the rise in soil and atmospheric temperatures during the
fire, and on the reduction of organic matter and release of
elements. Fire also affects the environment indirectly by
removing or reducing plant and litter cover, thereby
modifying both the post-fire microclimate and the activity of
the soil biota.

3.1 TEMPERATURES DURING FIRE

There is considerable variation in the temperatures
recorded during savanna fires. The fires are generally
hottest at, or just above, the soil surface, ranging from <
70° - > 800°C at ground level to 200° - 800°C at 1 m (Cook,
1939; Pitot and Masson, 1951; Ramsey and Rose-Innes, 1963;
Hopkins, 1965; Gillon, 1970; Harrison, 1978; Sweet, 1982;
Trollope, 1978, 1984b; Trollope and Potgieter, 1983). Back
fires are generally hotter than head fires at the soil
surface, though there are exceptions. In savanna grasslands,
some head fires exceed 500°C whereas under similar conditions Fig. 5.1. Vertical temperature profiles of some
savanna fires. A. Acacia nigrescens-
back fires seldom exceed 400°C (Trollope, 1978). Above 1 m, Sclerocarya caffra wooded savanna, Kruger
head fires are almost always hotter than back fires. National Park, South Africa (Potgieter
Temperatures though are generally much lower than those 1974, in Trollope, 1984b); B. Acacia
nigrescens-Combretum apiculatum tree
closer to the ground and decrease sharply with increasing savanna, Botswana (Sweet, 1982); C.
height. At 4 m, for example, temperatures hardly reach 100°C various savanna communities, Kruger
and then only for very brief periods (Figure 5.1). National Park, South Africa (Trollope and
The temperature increases are transient, lasting a few Potgieter, 1983).
minutes at most (Cook, 1939). In two experimental fires in
Brazilian cerrados, surface soil temperatures were above 60°C
for about 215 and 75 s, respectively (Coutinho, 1978). In The rate of heat transfer is higher in moist than in dry
another series of experiments, temperatures at ground level soils, though temperatures at a given depth normally cannot
remained above 60°C for 50% longer under back fires (257 s) rise above 100°C until all the moisture there has been
than under head fires (171 s) (Trollope, 1978). The evaporated (Raison, 1979).
differences in residence time at higher temperatures were These patterns of fire behaviour are broadly similar
even more pronounced (e.g. 270% at 120°C). Conversely, at 1 m to those in other biomes, but they come from a very limited
above grass canopy height (0.34 m) air temperatures of back data base. In order to have a better understanding of the
and head fires remained above 60°C for 47 s and 115 s, links between the various components of a fire regime, the
respectively (Trollope, 1978). behaviour of given fires, and their short- and long-term
Soil has a low thermal conductivity. Temperatures effects on the environment and biota, more studies of fire
recorded at 2 cm below the surface seldom exceed 35°C, while regimes and fire behaviour in savannas are needed.
at 5 cm depth there is almost no rise in temperature (Cook,
1939; Pitot and Masson, 1951; Ramsey and Rose Innes, 1963; 3.2 ALBEDO AND THE POST-FIRE ENERGY BALANCE
Gillon, 1970; Coutinho, 1978). Higher soil temperatures than Fire causes a decrease in the reflection coefficient
these can be expected under burning logs and other (albedo) of the soil surface, resulting in greater absorption
smouldering vegetation, but even then only a small fraction of solar radiation (Figure 5.2). This in turn leads to soil
of the heat produced is likely to be transferred to the soil. surface temperatures being higher during the day and lower at
night, and hence to greater daily surface temperature
fluctuations (Raison, 1979; Savage, 1980; Cass
98 P.G.H.Frost and F.Robertson Effects of fire in savannas 99

annually burnt plots when compared with adjacent unburnt

areas. The greatest increases occurred in plots burnt during
the late dry season. Other studies though have shown no
differences (Cass et al., 1984).

3.4 SOIL MOISTURE DYNAMICS

A number of studies have demonstrated lower soil
moisture levels on burnt plots compared with levels on
adjacent unburnt areas (Cook, 1939; San Jose and Medina,
1975; Savage, 1980; Gandar, 1982). The soil moisture balance
on burnt ground is affected by a number of factors including
(i) a reduction in plant and litter cover and subsequent
exposure of the soil surface for a variable period of time
after fire, leading to increased evaporation; (ii) the
physical characteristics of the soil and its susceptibility
to structural collapse; (iii) the timing and intensity of
subsequent rainfall, and (iv) the rate of vegetation
recovery.
Infiltration rates on sites burnt annually for many
Fig. 5.2. Changes in the reflection coefficient years can be much slower than on adjacent unburnt sites,
(albedo) and in herbaceous standing crop largely as a result of changes in soil surface structure
after fire (albedo: solid circle burnt
ground, o unburnt ground; herbaceous
(Cass et al., 1984). These are probably caused by a decrease
standing crop: x - - - x burnt, +-----+ in soil organic matter levels and a consequent reduction in
unburnt). Data from Harrison (1978). soil aggregate stability leading to individual soil particles
becoming detached during rainfall. The particles block soil
et al., 1984). The extent of these changes depends on the pores and form a surface crust which inhibits infiltration
thermal conductivity of the soil and on the partitioning of and increases runoff. The distillation of aliphatic
the additional energy between latent energy used in hydrocarbons from litter and soil organic matter during fire,
evaporating water, energy used by plants during and their subsequent condensation on the soil particles, also
photosynthesis, and sensible heat transfer to both the soil leads to the formation of water-repellent surfaces, even on
and atmosphere (Savage, 1980). sandy soils (Cass et al., 1984).
These changes persist until sufficient plant cover
develops to intercept the incident radiation during the 3.5 EROSION
daytime and reduce radiative heat loss at night (Figure 5.2). The risk of increased soil loss from an area laid
Low rainfall, extreme temperatures and herbivory all tend to bare by fires is a major concern in savanna management. Since
slow down the rate of plant recovery and thereby the increase soil nutrients are concentrated in the surface soil,
in albedo. extensive erosion may result in a significant depletion of
the nutrient status of the soil, as well as in a reduction in
3.3 SOIL PHYSICAL PROPERTIES soil depth and water holding capacity. Despite this concern,
Savanna fires are generally not hot enough to cause there is little quantitative information on the extent of
direct changes in soil physical properties. However, in the soil loss in burnt savannas and on the factors which
longer term, soil bulk density and porosity can be adversely influence this (Cass et al., 1984; O'Connor, 1985). Some of
affected by the reduction in plant and litter cover, changes the factors which increase the risk of erosion include
in microclimate, an increase in the rates of organic matter reductions in soil surface stability, lowered infiltration
mineralization, and changes in soil fauna activity. Increases rates and corresponding increases in runoff, topographic
in surface soil bulk density (Trapnell et al., 1976; Webber, position and soil type. Sites on steep slopes with friable,
1979; Brookman-Amissah et al., 1980), and reductions in clayey soils are particularly susceptible.
moisture holding capacity (Cook, 1939) have been recorded on
100 P.G.H.Frost and F.Robertson Effects of fire in savannas 101

those burnt annually for 23 years (Trapnell et al., 1976). In

3.6 LITTER REDUCTION this last experiment, wood and litter harvesting termites
In African savannas, fires consume around 70-90% of were more active on unburnt plots where they accumulated
the herbaceous standing crop and grass litter, 20-50% of organic matter and nutrients in their mounds, thereby
woody plant leaf litter, 12-58% of twigs, bark and woody reducing the overall levels of soil organic matter (Trapnell
fragments, and up to 20% of standing dead wood (Collins, et al., 1976).
1977; Isichei and Sanford, 1980; Villecourt et al., 1980;
Frost, 1985 and in prep.). The fuel loads in these studies 3.8 CATION EXCHANGE CAPACITY
varied from 4200-9000 kg/ha, with 48-63% of this being Since organic matter contributes greatly to the
consumed by fire on each occasion. Comparable figures have cation exchange capacity of many savanna soils, changes in
been obtained from moist savannas in Central America (62-84% SOM can be expected to result in parallel changes in CEC. In
reduction; fuel loads 6740-13800 kg/ha: Kellman et al., 1985) only one of the studies showing a difference between
and in arid central Australia (17-64% reduction; fuel loads treatments in SOM was cation exchange capacity also measured
2500-7400 kg/ha: Griffin and Friedel, 1984a). and the change paralleled the change in SOM (Moore, 1960). In
Some of the variation in the amount of material that those studies where only minor differences in soil organic
is burnt depends on when litter falls in relation to the time matter content were detected, CEC did not differ
of annual fires. In many West African savannas where the significantly between treatments (Coutts, 1945: Harrington,
rainfall is bimodal, litter falls either during the short dry 1974; Harrington and Ross, 1974; Trapnell et al., 1976;
season or at the end of the long dry season after burning has Griffin and Friedel, 1984a).
taken place (Isichei and Sanford, 1980). Other sources of
variation include the kind and intensity of fire. Back fires 3.9 RELEASE OF NUTRIENTS
generally consume more of the herbaceous standing crop and Burning oxidizes organically bound elements in the
leaf litter than do head fires (Trollope, 1984b). vegetation and litter and releases them in forms available to
plants. The intensity and duration of a fire, the amount of
3.7 SOIL ORGANIC MATTER AND SOIL CARBON material which is consumed, and its nutrient content, all
The recorded effects of savanna fires on soil organic determine the quantities of nutrients released. Some of the
matter and soil carbon levels are inconsistent. Lower levels elements, mainly nitrogen, carbon and sulphur, but including
of organic carbon have been measured in the surface soils of to a lesser extent phosphorus and potassium, are volatilized
annually burnt plots (White and Grossman, 1972; and may be lost to the atmosphere. Material which is not
Brookman-Amissah et al., 1980; Abbadie, 1983), and in plots volatilized is either deposited on the soil surface or, in
burnt during the late dry season compared with early-burn and lesser amounts, is removed as particulate matter in smoke or
unburnt areas (Moore, 1960; Oguntala, 1980). In these latter ash. Material that is deposited on the ground either re-
experiments the levels of soil carbon were higher in the enters the soil, or is redistributed by wind or through
early-burn than in the unburnt plots. This has been surface runoff. The extent of this depends on topography, the
attributed to the more vigorous root growth on the early-burn length of time the soil remains bare after fire, and on the
plots, especially by grasses (Kadeba, 1982). Higher levels of timing, amount and intensity of subsequent rainfall. The loss
soil carbon have also been recorded in annual-, biennial- and of nutrients through volatilization and export in ash during
triennial-burn plots than in unburnt plots or plots burnt savanna fires has seldom been measured satisfactorily. Given
only every 4 and 5 years. This is thought to be due to the amount of organic matter that is burnt, considerable
charcoal accumulation on the more frequently burnt plots losses might be expected. However, in most cases, significant
(Sweet, 1982). differences in nutrient concentration between fire treatments
In other studies, the differences in soil organic have been limited to the upper 0 - 5 cm of the soil.
matter levels between treatments have not been significant The release of nutrients by fire differs markedly
(Cook, 1939; Harrington, 1974; Harrington and Ross, 1974; from the relatively slow, weakly pulsed decay of organic
Griffin and Friedel, 1984a), even between unburnt plots and matter by micro-organisms in unburnt savannas. After fire,
102 P.G.H.Frost and F.Robertson Effects of fire in savannas 103

the concentration of plant available nutrients in the soil LOSSES GAINS

(kg N ha-1) (kg N ha-1 a-1)
solution increases sharply by amounts which broadly SAVANNA MEAN FIRE RAINFALL N- TOTAL REFERENCEa
correspond to those released from the vegetation and litter (LOCALITY) ANNUAL INPUT FIXATION
RAINFALL
(Cavalcanti, 1978 (quoted by Coutinho, 1982); Kellman et al.,
1985). The increases are largely confined to the surface High grass savanna 1500 (28)b (8) (39) (47) 10
(Andropogonae)
layers of the soil and are maintained for up to a month after (Ejura, Ghana)
fire. However, the initial peak in concentration declines Axeropus-Leptocoryphium 1350 (15) (5) (12) (17) 12
rapidly, suggesting that free ions are soon adsorbed onto the seasonal savanna
Bariras, Venezuela)
exchange complex. Despite deep percolation of rainwater,
leaching losses appear to be minimal (Kellman et al., 1985). Trachypogon savanna 1330 11.5 1-5 - - 7, 8
(Calabozo, Venezuela) 9-32

3.10 NITROGEN Loudetia-Andropogon 1230 10-23 19 9-12 23-33 1, 13

"derived” savannas
Significant losses of nitrogen through volatilization (Lamto, Ivory Coast)
during fires occur at temperatures above 400°C (Cass et al.,
Guinea savanna 1000 12-15 3-7 3-9 6-16 2, 5, 6
1984). How significant this is in the overall dynamics of (Kairji, Nigeria)
nitrogen in savanna ecosystems is difficult to assess. Most Monsoonal tall- 950 8-10 1 2-3 3-4 9
of the published estimates of nutrient loss through grass savanna
(Katherine, Australia)
volatilization and up-draught during fires are probably not
accurate (Raison, 1979). However, they serve as a guide. Burkea-Eragrostis 630 33 1-5 30 31-35 3, 4
woodland
The reported losses due to volatilization in savanna (Nylsvley, S. Africa)
fires vary between 4 and 33 kg N ha-1 (Table 5.1 ). Regular
Sahel-Sudanian savanna 580 4 3 4 7 11
losses of this magnitude might be expected to result transition (Nino, Mali)
eventually in lower levels of nitrogen in the soil. The
recorded changes in soil nitrogen due to fire parallel the a) 1. Abbadie (1983, 1984); 2. Adeniyi, quoted by Sanford (1982); 3. Frost (in prep.); 4.
Bate (1981); 5. Isichei (1980); 6. Isichei and Sanford (1980); 7. Medina (1982); 8. Medina
findings for changes in soil organic matter, described et al. (1978); 9. Norman and Wetselaar (1960) quoted by Mott et al. (1985); 10. Nye and
earlier, and are not repeated here. Greenland (1960); 11. Penning de Vries and Djiteye (1982); 12. Sarmiento (1984); 13.
Villecourt et al. (1980).
In addition to the loss caused by volatilization,
losses may also occur as a result of increased mineralization b)Estimated values in parentheses
and the subsequent leaching of nitrates beyond the rooting
Table 5.1. Calculated losses of nitrogen in savanna
zones of plants. Nitrification rates are initially reduced by fires, and measured or estimated gains in
fire but soon increase above levels recorded on adjacent rainfall and through microbial fixation of
unburnt plots, probably as a result of higher soil nitrogen.
temperatures, and remain significantly higher for up to 3
months afterwards (Adedeji, 1983). Nitrate ions are highly those in unburnt savannas. Only if there is a higher rate of
mobile and increased concentrations have been recorded in N-fixation on frequently burnt areas will the losses due to
water draining out of the topsoil of burnt plots (F. fire be counterbalanced. There is little information in this
Meredith, pers. comm.). However, losses of nitrate through regard. It is worth noting though that the incidence of
leaching are likely to be limited to porous, sandy soils in legumes, including those known to be nodulated and therefore
regions of high rainfall. potential N-fixers, is often higher on frequently burnt plots
The limited data available on annual inputs of compared with unburnt ones (see data in Isichei, 1979 [quoted
nitrogen in rainwater and through microbial fixation suggests by Sanford, 1982]; Edroma, 1984; P.G.H. Frost, personal
that, at least in some cases, they are sufficiently large to observation).
balance the estimated losses due to volatilization and
leaching (Table 5.1). However, there is no reason to suppose 3.11 PHOSPHORUS
that the inputs from the atmosphere differ markedly from Phosphorus can be volatilized at temperatures above
500° C though this is only likely to be significant under
complete combustion (Raison, 1979). The low solubility of P
104 P.G.H.Frost and F.Robertson Effects of fire in savannas 105

and its tendency to form complexes with Al and Fe in acid case though, the nutrients are being redistributed spatially
soils, or with Ca in alkaline soils, makes it less vulnerable within the system rather than being lost outright (Kellman et
to leaching. Higher levels of extractable P have been al., 1985). More intensive sampling would probably reveal
recorded in the surface soils of annually burnt savannas increases in exchangeable bases at sites receiving the
compared with unburnt plots (Harrington, 1974; Trapnell et run-off.
al., 1976; Afolayan, 1978; Brookman-Amissah et al., 1980; The effects of long term burning on individual
Sweet, 1982). There is a slight positive relationship with elements are variable. Sodium is very mobile but has not
frequency of burn (Sweet, 1982) and with season (intensity), often been measured. White and Grossman (1972) noted an
since plots which have been subjected to more intense, late apparent decline on plots burnt annually for 38 years, but
dry season fires have higher levels than either early-burn or Kellman et al. (1985) detected no changes in plots burnt a
no-burn plots (Trapnell et al., 1976; Oguntala, 1980). The number of times over a 7-year period. Potassium is also
increase in extractable P levels in the surface soil of highly mobile and is readily leached from the vegetation and
regularly burnt areas suggests that fire promotes a more litter. This may explain in part the variable results which
rapid cycling of phosphorus through the vegetation and soil, have been obtained. Most studies have shown no significant
in contrast to unburnt areas where P is retained for longer differences between treatments (Harrington and Ross, 1974;
by the vegetation. Trapnell et al., 1976; Griffin and Friedel, 1984a; Kellman et
A number of studies have shown no significant al., 1985). In other studies, higher levels of exchangeable-K
difference in the levels of extractable P between different have been found in annually burnt compared with unburnt
burning treatments (Moore, 1960 [cf. Oguntala, 1980]; plots, with early-burn having more than late-burn plots
Harrington and Ross, 1974; Griffin and Friedel, 1984a; (Moore, 1960; Harrington, 1974; Afolayan, 1978;
Kellman et al., 1985). To what extent these results reflect Brookman-Amissah et al., 1980; Oguntala, 1980; Sweet, 1982).
differences in the methodology of extracting phosphorus, as It is not clear to what extent these are site or sampling
opposed to differences in process, is not clear. effects, as opposed to treatment effects, perhaps reflecting
higher concentrations of potassium in plant tissues at the
3.12 SULPHUR start of the dry season. Two studies have recorded lower
Like nitrogen, sulphur is readily volatilized during amounts of exchangeable-K in annually burnt plots, possibly
fire. Frequent burning therefore could result in a decrease as a result of the removal of ash by surface wash (White and
in sulphur levels in the soil. This may be one of the reasons Grossman, 1972; Strang, 1974).
why cerrado soils in Brazil are deficient in sulphur (McClung Calcium and magnesium are less soluble than K or Na
and de Frietas, 1959). In central Australian arid woodlands, and are also not readily volatilized. In most cases, a change
no significant differences between treatments in extractable in one of the elements is paralleled by a change in the
S were recorded (Griffin and Friedel, 1984a). other, though Griffin and Friedel (1984a) noted an increase
in exchangeable Ca in open woodland sites after dry-season
3.13 EXCHANGEABLE BASES fires which was not matched by any increases in the other
Sweet (1982) has recorded an increase in total cations. It may be significant that these were the only sites
exchangeable bases with increasing burning frequency. In where woody plant densities were markedly reduced (Griffin
other cases, a decline in TEB has been noted, usually in and Friedel, 1984b). In general, higher levels of
situations where the vegetation has been exposed to hot fires exchangeable Ca and Mg are found in annually burnt plots
for many years (Moore, 1960; White and Grossman, 1972; compared with unburnt areas (Harrington, 1974; Trapnell et
Strang, 1974). These declines are probably the result of fire al., 1976; Afolayan, 1978; Sweet, 1982; Kellman et al.,
intensity rather than fire frequency. Moore (1960) 1985). In some cases either no difference between treatments
demonstrated higher levels of TEB in plots exposed to cooler, (Harrington and Ross, 1974), or a reduction on annually burnt
early dry-season fires (which had the lowest TEB). However, plots has been recorded (White and Grossman, 1972; Strang,
the lack of replication in these experiments makes it 1974). The response to season and intensity of fire is more
difficult to assess their significance. In cases where variable. Plots subjected to more intense late dry season
declines have been observed, the prime cause of the reduction fires can have higher levels of exchangeable Ca and Mg than
is thought to be the removal of ash by surface wash. In this plots burnt during the early dry
106 P.G.H.Frost and F.Robertson Effects of fire in savannas 107

season (Trapnell et al., 1976), or lower levels (Moore, 1960; soil chemistry (Raison, 1979).
Oguntala, 1980). In some cases, there have been no The change in environmental conditions can also
differences (Harrington, 1974; Afolayan, 1978). affect community composition. Short-term increases in the
numbers of nitrifying bacteria have been recorded in some
3.14 SOIL REACTION (pH) cases (Dommergues, 1954; Adedeji, 1983), though not in all
The input of base-rich ash after fire can cause a (Meiklejohn, 1955). The rise in pH and a higher concentration
short term increase in soil pH. The extent of the increase of soluble sugars in the soil solution may make conditions
depends on the amount and composition of the ash, and on the more favourable for bacteria than for fungi (Raison, 1979).
buffering capacity of the soil (Cass et al., 1984). Longer In the longer term, frequent burning affects the soil biota
term changes arising from exposure to different fire regimes by altering the amount and nature of organic matter inputs to
tend to parallel the changes in exchangeable bases. There is the soil. For example, annual burning apparently reduces the
a slight increase in pH with increased frequency of burning activity of surface foraging, wood- and litter-eating
on soils derived from acid, igneous parent material (Sweet, termites (Trapnell et al., 1976). Similarly, in the monsoonal
1982), between annually burnt and unburnt plots (Cook, 1939; tallgrass savannas of northern Australia, the numbers of
Harrington, 1974; Trapnell et al., 1976; Afolayan, 1978), and arthropod detritivores declined in frequently burnt areas,
between early-burn and both late-burn and unburnt plots whereas other epigaeic invertebrates were not permanently
(Moore, 1960; Brookman-Amissah et al., 1980). In all cases, affected (Greenslade and Mott, 1983).
the changes 'have been less than 1 pH unit, an amount which
is unlikely to greatly affect plant nutrient availability. 4. EFFECTS OF FIRE ON PLANT SPECIES COMPOSITION
Thirty years of annual burning of a grassland Plants differ widely in their tolerance of fire and
firebreak in Zimbabwe resulted in a lower pH in the surface in their capacity to recover afterwards. As a result,
soil compared to that of an adjacent woodland, matching the recurrent fires have considerable potential to influence the
recorded decline in total exchangeable bases (Strang, 1974). structure and composition of vegetation. The extent to which
However, in another case where apparently significant this occurs depends not only on differences in sensitivity
declines in both the concentrations of exchangeable cations between species but also on the type, frequency and intensity
and percentage base saturation occurred, there was no of fire, and on the physiological and developmental states of
corresponding decrease in pH (White and Grossman, 1972). No individuals at the time of burning. Events occurring in the
significant differences in soil pH have been noted in other interval between fires also influence the eventual outcome.
studies (Harrington and Ross, 1974; Oguntala, 1980; Griffin Drought, above-average rainfall and herbivory affect fuel
and Friedel, 1984a), but in these cases the amounts of loads, and thereby fire intensity, as well as the condition
exchangeable bases also did not differ between treatments. of individual plants and their degree of recovery. The
effects of fire regime on species composition therefore
3.15 SOIL BIOTA cannot be seen in isolation from the influence of these other
The effects of burning on the composition and activity factors. This complicates the interpretation of the results
of the soil biota depend upon factors such as the kinds of of most of the experiments on fire in savannas (van Wyk,
organisms involved, fire intensity, and the extent to which 1972; O'Connor, 1985).
burning alters the post-fire environment. Most savanna fires
are not sufficiently intense to have a marked direct effect 4.1 FIRE TYPE AND PATTERN
on the soil biota. The changes that do occur, even under Fire-sensitive woody species are widespread in the
relatively intense fires, are small, temporary, and usually savanna biome (see Trapnell, 1959, Hopkins, 1965; West, 1972;
confined to the top few centimetres of the soil surface (Nye San Jose and Farinas, 1983; Frost, 1984, for examples). In
and Greenland, 1961; Meiklejohn, 1955; Cass et al., 1984). some cases they form the dominant component of the
Populations also tend to recover rapidly, often to levels vegetation, despite recurrent fires (e.g. Brachystegia and
above those in unburnt soil, though this depends largely on Julbernardia spp. in Central African miombo: Trapnell, 1959).
environmental conditions at the site, particularly soil To survive and grow under these conditions individuals need
moisture levels and the extent to which fire has modified to escape the full effects of
108 P.G.H.Frost and F.Robertson Effects of fire in savannas 109

fire for long enough to enable them to pass through the be expected to be more tolerant of frequent fires than plants
vulnerable sapling stage and reach the more fire-resistant growing in less favourable environments.
mature tree stage. Irregular fires in space and time produce Species which are tolerant of fire and regenerate
the necessary circumstances for this to happen. vegetatively generally recover their pre-fire status in the
Most savanna fires burn patchily as a result of community more rapidly than species which only regenerate
varying wind speeds, topography and fuel loads. The uneven from seed. For populations of these latter species to survive
distribution of herbaceous biomass in particular, resulting recurrent fires, they must be able to establish, grow and
from differences between sites, suppressed grass production reproduce in the interval between successive fires. The high
by trees, or the localized impact of herbivores, affects the fire frequency in moist savannas is one of the main factors
likelihood of a plant being damaged or killed. The regular selecting against such species, particularly among woody
occurrence of fire-sensitive woody species in savannas plants.
suggests not only that patchiness is a characteristic of most Frequent fires reduce woody plant densities in moist
savanna fires but also that the spatial pattern is relatively savannas, primarily by killing or suppressing individuals in
predictable. However, apart from the studies by Hopkins the smaller size classes (Trapnell, 1959; Charter and Keay,
(1965) and Braithwaite and Esthergs (1985), the pattern of 1960; Hopkins, 1965; Geldenhuys, 1977; Brookman-Amissah et
savanna fires has not been analysed. It remains to be seen al., 1980; San Jose and Farinas, 1983). Fire is relatively
how consistent the pattern is from one fire to the next. selective in this respect, with some species disappearing at
The spatial scale and uniformity of burning also tend a faster rate than others. Lianes appear to be particularly
to define the nature and extent of certain post-fire fire-tender (Trapnell, 1959; Malaisse, 1978). Protection from
interactions. For example, large herbivores are attracted to fire results in an increase in tree density, particularly of
burnt areas by the flush of vegetation after fire and tend to these fire-sensitive species. In the high rainfall savannas
concentrate at higher densities in areas which are patchily of the savanna/forest transition-zone, the species which
burnt or comprise only a relatively small proportion of the establish are often evergreen forest species (Charter and
total. This can lead to heavy, localized use of the Keay, 1960; Lawton, 1978) but elsewhere they are typical
vegetation and possibly to changes in species composition, as savanna species (Menaut, 1977; Trapnell, 1959;
well as to a lower probability of the area supporting another Brookman-Amissah et al., 1980; San Jose and Farinas, 1983).
fire in the short-term. This interaction between the size of The dominant woody plants in arid savannas appear to
areas burnt, herbivore densities and subsequent vegetation be more fire-sensitive than most of the plants occurring in
change has not been thoroughly investigated, though it is moist savannas, despite generally lower fire intensities.
probably an important factor affecting the changes in grass However, fires are infrequent and normally only occur after
species composition on experimental fire plots in the Kruger periods of exceptional rainfall, although when they do occur
National Park (van Wyk, 1972). they can cause considerable mortality. Under these
circumstances the affected plant populations rely almost
4.2 FIRE FREQUENCY entirely on post-fire regeneration from seeds for re-
Fire frequency determines the length of time that a establishment (e.g. Acacia aneura, A. deanei, Callitris
plant has to recover before the next fire occurs. The slower columellaris, Cassia nemophila, Eremophila latrobei and
the rate of recovery, the more likely it is that the others in central Australia (Hodgkinson, 1979; Walker et al.,
structure and composition of the vegetation will be altered, 1981; Griffin and Friedel, 1984b); Acacia erioloba in
particularly where fires occur frequently. The rate of southern Africa (van der Walt and le Riche, 1984)).
recovery depends on (i) the extent of damage sustained by the This relationship between seedling establishment and
plants; (ii) the method of regeneration, and (iii) the adult mortality means that these arid savanna communities are
favourableness of the post-fire environment for establishment particularly susceptible to changes in species composition.
and growth, particularly the amount and temporal distribution Events such as intense herbivory, heavy rains or drought,
of rainfall, and the extent and intensity of herbivory. occurring after a fire during the period of seedling
Plants growing on moist, nutrient rich soils could therefore establishment, can differentially affect the
110 P.G.H.Frost and F.Robertson Effects of fire in savannas 111

survival of seedlings of the different species. This can lead these species may be responding more to changes in tree
to sudden, unexpected shifts in the composition of the density and associated microclimate than directly to the
post-fire community (Griffin and Friedel, 1985). absence of fire, since the removal of woody plants is known
Where fires in arid savannas occur relatively to influence herbaceous composition and production (Dye and
frequently (usually under experimental conditions) they are Spear, 1982; O'Connor, 1985).
seldom intense enough to kill established woody plants, Fire may also mediate competitive interactions between
though canopy cover is often sharply reduced. Woody plant species. For example, in South America, frequent burning
density may even increase as a result of increased favours species such as Trachypogon plumosus and Axenopus
recruitment from seeds stimulated to germinate by fire canescens, which are subdominant in unburnt savannas, whereas
(Sweet, 1982; Hodgkinson et al., 1984; Griffin and Friedel, protection favours their competitor, Trachypogon montufari
1985). Establishment is often aided by the adverse effects of (San Jose and Medina, 1975).
fire on the herbaceous layer, particularly in years of low
rainfall (Gertenbach and Potgieter, 1979). 4.3 SEASON AND INTENSITY OF FIRES
Savanna herbaceous communities are less affected by The time of burning interacts with plant phenology and
fire than are the woody communities. The most obvious changes post-fire weather conditions to affect plant survival and
occur at the extremes of fire frequency (annual burning vs no reproduction. In most savannas, the season of burn is a
burning). For example, in south and east African savannas, relatively predictable component of fire regime since most
annual burning generally favours species such as Themeda fires occur during the dry season. It is this within-year
triandra, Digitaria pentzii, Pogonathria squarrosa and predictability which enables many species to avoid fire by
Heteropogon contortus whereas less frequent burning favours being dormant during the main fire season.
species such as Cymbopogon plurinodis, Sporobolus fimbriatus Fire intensity is a function of fuel type, fuel load,
and forbs (Davidson, 1953; van Wyk, 1972; Harrington and moisture content, and atmospheric conditions at the time of
Ross, 1974; Robinson et al., 1979; Edroma, 1984). fire, and therefore is linked to both the seasonality and
However, rainfall variability is a confounding factor frequency of burning. The longer the interval between fires
(Kennan, 1972; Gertenbach and Potgeiter, 1979; O'Connor, the greater the fuel load and therefore the more intense the
1985). For example, Themeda triandra is susceptible to fire. Fire intensity exerts differential effects on the
drought, and although it increases under frequent burning and survival of plants and their propagules, and stimulates the
decreases if fire is excluded (Davidson, 1953; Harrington and germination of seeds of different species to various degrees.
Ross, 1974; Robinson et al., 1979; Edroma, 1984), this only Intensity also variously affects the physical and biological
happens under average or above-average rainfall. The environments of each species, making conditions sometimes
combination of frequent burning and low rainfall causes T. more, sometimes less, suitable for establishment and growth.
triandra to decline. In semi-arid regions therefore, the The late dry season in Africa is a period of nutrient
species is most successful under less frequent burning or translocation and rapid growth by woody plants, and fire at
even under complete protection (Theron, 1937 [quoted by this time is considered to be very damaging (West, 1972;
Tainton and Mentis, 1984] ; Kennan, 1972). Annual burning in Kennan, 1972). Woody plant density is lowest and grass
these areas leads to an increase in annual and pioneer production highest under a late dry season fire regime. Fires
species (e.g. Aristida barbicollis, A. scabrivalvis, Urochloa during the late dry season, when the grass is driest and the
bulbodes, U. mosambicensis: Kennan, 1972; Harrington and conditions are hottest, are more intense than early dry
Ross, 1974; Gertenbach and Potgieter, 1979; R.I. Yeaton, season fires (Sanford, 1982; Edroma, 1984).
personal communication). In moist savannas, frequent late dry season fires
Protection from fire results in an increase in mesic, destroy young trees and shrubs, or their aboveground parts,
shade-loving grasses such as Panicum maximum (Harrington and so preventing the development of taller, more fire-resistant
Ross, 1974), Andropogon tectorum (Brookman-Amissah et al., size classes. Although woody plant growth may be further
1980), Crasspedorhachis africanus (Strang, 1974), and forest inhibited by browsing ungulates, fire alone is sufficient to
species (e.g. Oplismenus and Lasiacis: San Jose and Farinas, promote a lower woody plant density. Conversely, with longer
1983; P.G.H. Frost, personal observation). However, some of intervals between fires, woody plant density
112 P.G.H.Frost and F.Robertson Effects of fire in savannas 113

increases to a point where there is insufficient grass to Differential growth and mortality of grass species
fuel a fire intense enough to affect woody plants. Beyond depends partly on the degree of protection of tiller
this point, other factors such as soil texture and depth, and primordia. Species such as Themeda triandra elevate their
soil water and nutrient availability become key determinants tillers during the late dry season and so become very
of grass:woody biomass ratios. sensitive to damage by fire and herbivory. In other species,
Alternatively, fires in the middle of the wet season such as Heteropogon contortus in Africa and Australia, and
in Africa are seldom damaging to woody plants because the Hyparrhenia filipendula and H. hirta in Africa, the tillers
burns are cool and patchy (Trollope, 1984a). The majority of remain protected throughout the dry season and are not
trees and shrubs in South American humid savannas are elevated until well into the wet season (Booysen et al.,
evergreen, but grow and change their leaves during the dry 1963; Stocker and Mott, 1981; Mott et al., 1985). This
season (Medina, 1982). In contrast to Africa, Sarmiento and accounts for the differences in productivity of these species
Monasterio (1983) consider that fires during the season of and for the shifts in species composition which can occur
growth in South America are not particularly damaging, under early, and late dry season fire regimes.
because they simply stimulate new buds. It is fires during In central Australian rangelands, species such as
the wet season in South America that cause most damage to Enneapogon polyphyllus, E. avenaceus and Aristida contorta
woody plants because this is not a growth period and no new are maintained under a regime of dry season fires, but are
leaves are produced. In Australia midsummer (wet season) significantly reduced by wet season fires and replaced by
fires generally decreased the density of Acacia harpophylla forbs such as Calocephalus platycephalus, Helipterum
suckers, while dry season fires increased sucker density tietkensi, H. charsleyae, Brachycome ciliaris and Senecio
(Johnson and Purdie, 1981 ). magnificus (Griffin and Friedel, 1984a).
Head fires are most damaging to woody plants because Late dry season fires promote the growth and
maximum heat release occurs well above the soil surface, development of perennial grasses while early burning reduces
nearer the terminal buds, while back fires are most damaging perennial grasses and encourages annuals (Afolayan, 1978),
to grass because fire intensity is highest at ground level Since most of the annuals have already set seed by that time,
(Trollope, 1984a). Most savanna grasses are capable of they are less adversely affected by fires occurring then than
surviving burning during the dry season by being dormant and are perennials, which are generally still physiologically
having their tiller initials protected by persistent leaf active. Consequently, a regime of regular early dry season
bases. However, the plants are much less tolerant of burning fires tends to favour the development of an herbaceous
during the wet season even though fire intensity is lower, community dominated by annual grasses. Sanford (1982),
because they are physiologically active at the time, (West, however, states that there is little evidence of any
1965; Trollope, 1984). influence of time of burning on the relative success of
There is very little quantitative information annuals and perennials. This is an issue which requires more
available on the effect of season of burn on herbaceous layer research.
composition. Those effects that have been noted are
frequently confounded by grazing and rainfall variability 4.4 REGENERATION OF WOODY VEGETATION
(van Wyk, 1972). Species such as Themeda triandra become Fire has much more of an effect on woody plant
dominant in swards burnt during the late dry season composition, though isolating the effects of season from
(Davidson, 1953; Trollope, 1982) whereas species such as those of intensity is difficult because the two variables are
Cymbopogon plurinodis decrease. interrelated. Woody plants tend to be more susceptible in the
Conversely, Themeda declines in swards burnt during late dry season when fires are generally more intense because
the late wet season and in unburnt areas. Burning during the at that time the initial temperature of plant tissue is
middle of the wet season, at a time when plants are higher and closer to lethal limits. The moisture content of
physiologically active, can have a marked adverse effect the plant is also higher, resulting in a higher thermal
(West, 1965; Trollope, 1984). Burning during a mid-wet season conductivity and more rapid transfer of heat to the interior
drought indicates that this effect depends on the of the plant. Consequently, the proportion of woody plants in
physiological state of the plant rather than on the season the community declines as fire regimes become
per se.
114 P.G.H.Frost and F.Robertson Effects of fire in savannas 115

progressively hotter. In addition, many woody plants produce Australian Acacia savannas because scarification by fire
new leaves before the start of the rainy season and so have permits imbibition and extensive regeneration of woody
depleted reserves at that time. Since new leaves are also species (Leigh and Noble, 1981). In the same region, isolated
more susceptible to damage by fire, burning at the start of large Eucalyptus trees may release large quantities of seed
the wet season forces the plant to draw on already depleted in response to heat stimulation by fire (Johnson and Purdie,
reserves in order to replace those consumed in the fire 1981).
(West, 1965; Kennan, 1972). Fire tends to favour those woody Regeneration in the humid savannas at the
species having protected meristems and other fire-resistant forest/savanna boundary is dependent both on sprouting and
above-ground structures (for example, thick bark); the suckering and on seedling establishment. A favourable
capacity to resprout from below-ground meristems on microhabitat provided by standing trees may be necessary for
rootstocks, lignotubers etc.; or seeds that can survive fire the germination and establishment of woody seedlings and in
and in which heat triggers germination. fact few seedlings establish in the grass patches in West
The development of adventitious buds from vascular African Guinea savanna even in the absence of burning (Menaut
cambium is a feature of savanna trees and shrubs (Sarmiento & and Cesar, 1982). The canopy species of Brachystegia woodland
Monasterio, 1983). This adaptation permits both the regrowth seldom establish in open grassland probably because of the
of branches after canopy damage and sprouting from alternation between cool moist weather which stimulates
underground parts if the aboveground shoots are removed. Most germination and hot dry conditions which kill emergent
woody regrowth in the humid and subhumid savannas of seedlings (Strang, 1966). Seedlings are more common under
Australia (Mott et al., 1985), South America (Eiten, 1972; savanna trees than in open grassland in woodland savanna in
Medina, 1982) and Africa (Lawson et al., 1968; Boaler 1966; Belize (Kellman and Miyanishi, 1982) and Trachypogon savanna
Jackson, 1974) arises vegetatively from material remaining in in Venezuela (San Jose and Farinas, 1983), perhaps through
the soil after disturbance. The most vigorous shoots are local nutrient enrichment or suppression by grass
produced by plants with large belowground organs (Trapnell, competition. Grass competition has been shown to prevent
1953; Fanshawe, 1959). establishment of Acacia spp. seedlings in a well developed
Even where disturbance is sufficiently severe to Cenchrus ciliaris sward in South Africa (Knoop and Walker,
remove large rootstocks, suppressed seedlings may remain in 1985). Regeneration of fire-sensitive Callitris columellaris
large numbers. After 15 years of hoe cultivation, 3,500 to in Australia is restricted to clumps of adult trees where
11,500 small woody plants per hectare remained in fields grass growth is suppressed and fires infrequent, although
derived from Combretum spp. woodland in Malawi (Robertson, regeneration will extend beyond the clumps if protected from
1984). These suppressed seedlings, or seedling suffrutices fire (Stocker and Mott, 1981).
(Boaler, 1966), whose belowground parts may be years older In the derived savanna zone of West Africa forest may
than the shoots which are regularly destroyed by fire, are regrow if the fire-sensitive forest species regenerate
also a feature of Brachystegia woodland (Boaler and Sciwale, rapidly from sprouts and seed (Hopkins, 1983). However, if
1966; Strang, 1974). Spontaneous dieback to ground level as savanna grasses invade before the woody vegetation is
reported by Trapnell (1959) and Boaler (1966) is probably established hot fires are likely and the grassland may then
rare unless the shoots are killed by fire, although partial be invaded by the woody species characteristic of savanna.
dieback during the dry season is common (personal The dry dipterocarp forests of South-East Asia share a common
observation). The root:shoot ratio of young plants is high flora with the derived savannas (Blasco, 1983). Here there is
(Rutherford, 1982; Menaut and Cesar, 1982) and although the no invasion but an impoverishment of the original forest
above ground parts may be killed, suffrutices persist despite flora brought about by felling, cultivation and fire.
burning. Burning becomes less effective in preventing The effect of fire on tree regrowth is strongly
regeneration in the drier subhumid savanna types with their influenced by stem diameter and height, but not in any
lower grass biomass and greater number of fire-resistant consistent way. Seedlings and saplings are generally at
species. Regular burning may reduce the number of suffrutices greatest risk (Rutherford, 1981; Pellew, 1983; Griffin and
in humid savannas, but many still remained after thirty-six Friedel, 1984b), but in some species, mortality appears to be
years of annual late burns in Zambian Brachystegia woodland highest in intermediate height classes (e.g. Acacia karroo:
(Mansfield et al., 1976). Burning is not recommended in some Trollope, 1974). In contrast, in Acacia erioloba,
116 P.G.H.Frost and F.Robertson Effects of fire in savannas 117

mortality of mature trees was more than 7 times higher than In the humid Guinea savanna zone at Olokemeji in
that of saplings (van der Walt and le Riche, 1984) (Figure Nigeria, the chances of stem survival increased considerably
5.3). with increased size and basal area (Hopkins, 1965). Less than
25% of trees under 3 m in height survived 5 years of severe
annual fires. Greater size was associated with apical buds
borne above flame height and thicker bark which protected the
cambium. Early cool fires in subhumid Brachystegia woodland
kill stems below 0.5 cm in diameter while hot fires kill
stems up to 5.0 cm (Bell, 1981a). The critical height for
fire mortality among Acacia tortilis regrowth in the
Serengeti woodlands is 3.0m (Pellew, 1983).
Although survival is usually high, fire often
severely damages the aboveground parts of plants,
particularly in individuals smaller than 2-3 m (Figure 5.4).
The percentage topkill of Acacia karoo (as opposed to
mortality shown in Fig. 5.3) was strongly associated with
fire intensity in stems between 0.5 to 2.0 m in height
(Trollope, 1984a). Stems less than 0.5 m tall were killed
irrespective of fire intensity, while few stems exceeding 2 m
in height were killed even if fires were very intense. This
reverses the normal progression from the smaller to taller
height classes as individuals age. Frequent reversals of this
sort eventually result in distorted size distributions in
which most of the population is confined to the smaller
height classes (Figure 5.5).
Fire probably cannot prevent woodland regeneration
after clearing in areas where conditions are particularly
favourable for woody plant growth, such as areas of high
annual rainfall with permeable soils and high infiltration
rates. Annual burning was shown to hold regrowth at the open
tree savanna stage following clear-felling in the humid
Southern Guinea savanna zone, but under an annual early
burning regime woody regrowth was almost as fast as that
under complete protection, and the plots developed into
forest (Rose Innes, 1971). In the subhumid Northern Guinea
savanna zone even early burning prevented satisfactory tree
growth (Brookman-Amissah et al., 1980). Clear-felled plots in
Fig. 5.3. Percent mortality caused by fire in various
savanna woody plants. A. Acacia tortilis
Brachystegia woodland at Ndola in Zambia developed only a few
(Pellew, 1983); B. Acacia karoo (Trollope, small trees and shrubs after 11 years of late burning, but
1974); C. Acacia erioloba (van der Walt and early burned plots showed clumped coppice regrowth 4.5 m to 6
le Riche, 1984); D. various tree species, m tall (Trapnell, 1959). At a slightly lower rainfall, even
and E. Ochna pulchra (shaded columns
represent plants killed by a slow moving annual late burns were insufficient to prevent the regrowth
fire; clear columns, plants killed during a of clear-felled Brachystegia woodland in Zimbabwe (Barnes,
fast moving fire: Rutherford, 1981); F. 1965). Woodland species regenerate under an early burning
various tree and shrub species (lines
calculated from regression equations linking
regime whereas frequent late dry season fires destroy a
mortality to fire line intensity (I) and woodland canopy and reduce it to coppice (Trapnell, 1959;
plant height class: a. I = 770 kJ/s/m; b. = Charter and Keay, 1960; Brookman-Amissah et al., 1980).
104 kJ/s/m (data from Griffin and Friedel,
1984b).
118 P.G.H.Frost and F.Robertson Effects of fire in savannas 119

Fig. 5.4. Percentage reduction in tree canopy volume

caused by savanna fires of differing
intensities. A. Acacia karroo: open and half-
open circles low intensity fires, solid
circle high intensity fire (Trollope, 1984a);
Acacia davyii, open square low intensity
fire, solid square high intensity fire
(Macdonald. 1982, in Frost, 1984). B. Acacia
tortilis: open triangle (Pellew, 1983),
solid triangle (Sweet and Tacheba, 1985). C.
Ochna pulchra: -.-.-.-.- ‘slow’ burn, - - - -
‘fast' burn (calculated from regressions
given by Rutherford (1981)); various species:
+ (Herlocker, in Norton-Griffiths, 1979), x
(Sweet and Tacheba, 1985). Recalculated from
original data sources using midpoints of tree
size and damage class intervals.

Complete protection from fire results in an increase

in tree density, particularly of fire-sensitive species,
leading to the establishment of woodland in many savanna
Fig. 5.5. Size distributions of trees in savannas
areas, and to the development of forest in the high rainfall burnt at different frequencies. A.
areas of the transition zone from savannas to forest Brachystegia spiciformis-Julbernardia
(Trapnell, 1959; Eiten, 1972; Menaut, 1977; Brookman-Amissah globiflora woodland, Zimbabwe (Boultwood
et al., 1980; San Jose and Farinas, 1983). and Rodel, 1981; B. Acacia
nigrescens-Combretum apiculatum tree
Fire exclusion is the most effective method of savanna, Botswana (Sweet, 1982). The
encouraging woody regrowth. However, complete protection from smallest size class is shaded for
fire may be unattainable because of the high incidence of emphasis.
man-made fires and the probability of lightning fires, which
are common in Africa (West, 1972) and Australia, although not burning is used in Zimbabwean Brachystegia woodland to
reported in South America (Coutinho, 1982). The intensity of prevent fuel accumulation after canopy removal by elephant,
accidental fires will increase with each year of protection and also in forestry areas (Fanshawe, 1959). Fire intensity
owing to the accumulation of moribund grass. Annual early (determined by air
120 P.G.H.Frost and F.Robertson Effects of fire in savannas 121

temperature, relative humidity, windspeed and grass fuel heavy grazing, which might limit seed set (Mott and Andrew,
load) must be such that the woody plant biomass lost does not in press). For example, in northern Australia, the seeds of
exceed the increment made during the previous wet season. the annual grass Sorghum intrans germinate soon after the
Complete protection for some years is necessary if the woody first rains. Burning during dry intervals in the early wet
plants are very slow-growing or unusually fire-sensitive, season causes a marked change in the composition of
e.g. the canopy species of Baikaea woodland in Africa communities dominated by this species because the developing
(Geldenhuys, 1977) and Callitris spp. in Australia (Leigh and plants are killed before they ran mature and set seed (Smith,
Noble, 1981). 1960; Stocker and Mott, 1981). In contrast, dry season fires
Woody plant regeneration may be encouraged by fire have little effect on composition, although plant density
protection in humid savannas, allowing the germination and decreases because the harsher physical conditions of burnt
subsequent establishment of fire-sensitive forest species. In sites result in higher seed mortality and lower seedling
subhumid savannas regeneration is plentiful unless the establishment.
disturbance has been very severe, but may be improved if a
few canopy trees are left and stumps are not removed. Late 5. PLANT PRODUCTIVITY AND QUALITY
hot fires in Brachystegia woodland where trees have been
felled by elephant burn the trunks of even the largest fallen 5.1 PLANT PRODUCTION
trees to 30 cm or more below ground level. This destroys the A range of conceptual and methodological problems
root-collar junction where the majority of adventitious buds surround the measurement of plant production (Sarmiento,
arise and the trees seldom sprout again. In arid savannas 1984). The most commonly used methods involve destructive
fire may encourage regeneration through seed, but only if sampling of plant biomass at regular intervals during the
followed by favourable circumstances. Where grass growth is growing season, with production being calculated either as
very high, such as in valley situations, fires will be very the sum of all positive biomass increments or as the
intense and so be able to trap woody plants in an early difference between the seasonal peak and trough of biomass.
growth stage. All other things being equal, the ratio of Differences in the timing of growth between species are
grass to woody plants will be highest under a regime of usually ignored, as is growth after the time of peak biomass.
predominantly late-dry season fires and least under a regime Contemporaneous flows to decomposers and to herbivores are
of late-wet season or early-dry season fires. often neglected. In most cases, only the biomass of
aboveground parts is measured; belowground production and the
4.5 SEED AND SEEDLING DYNAMICS transfer of material to and from roots and underground
Fire creates opportunities for enhanced reproduction storage organs is seldom taken into account, even though such
by removing plant cover and reducing competition from transfers are an essential part of plant functioning. Since
established plants. The increase in nutrients is also a defoliation usually results in a temporary cessation or
potential factor. Fire directly and indirectly stimulates reduction in root growth, any apparent increase in
germination in a number of species (Heteropogon contortus aboveground production may be more than offset by lowered
(Shaw, 1957; Tothill, 1969, 1977); Themeda triandra (Lock and belowground production. In view of these deficiencies in
Milburn, 1971; Trollope, 1984a); Cassia nemophila and Acacia method, the following discussion is focused primarily on
aneura (Moore, 1973; Hodgkinson, 1979). This effect can come changes in aboveground yield rather than in net production.
about both by the direct effect of heat on the seeds, causing Most of the applicable studies have been concerned with
abrasion in hard-seeded species, and increased soil changes in grass yield.
temperatures resulting from the reduction in plant and litter In perennial grasses, fire removes moribund material
cover (Tothill, 1977). and old leaves, thereby allowing more light to reach the
In some annual grass communities, particularly in younger, photosynthetically more efficient tillers at the
areas of high and reliable rainfall, germination at the start base of the plant. The number of tillers often also increases
of the wet season exhausts the seed store in the soil. Since as a result of the removal of apical dominance. The higher
these communities depend on a continued input of viable seed daytime temperatures and increased net radiant flux density
in order to persist at a site, they are particularly of burnt areas generally produce more
vulnerable to any factors, such as early wet season fires or
122 P.G.H.Frost and F.Robertson Effects of fire in savannas 123

favourable conditions for photosynthesis than on unburnt by the type of fire. Critical temperatures are maintained for
areas, while the cooler temperatures at night result in lower longer during back fires than during head fires, with the
respiration rates (Savage, 1980). Finally, an increase in result that back fires adversely affect grass yields more
nutrient supply apparently also has positive effects; if ash than head fires do (Trollope, 1978).
is removed after a fire both the rate of biomass Fire also has longer term effects on grass biomass and
-accumulation and final yield are reduced (Cavalcanti, 1968 yield. Protection from fire generally leads to a marked
(quoted by Coutinho, 1982)). reduction in both, particularly in moist savannas where the
Although these changes enhance the photosynthetic establishment of fire-sensitive woody species results in the
capacity of a plant, thereby potentially increasing its formation of a closed canopy woodland or forest which shades
productivity, the overall effect on production depends out the grass (Menaut, 1977; Brockman-Amissah et al., 1980;
ultimately on whether the regrowth can be sustained. This in San Jose and Patinas, 1983). Basal cover also declines in
turn is influenced by the time of burning, the prevailing mesic savannas where woody plants do not form a complete
soil moisture levels, and the length and severity of the dry canopy when fire is excluded (Boultwood and Rodel, 1981). In
season. Most fires occur during the dry season when soil this case, the accumulation of grass and litter gradually
moisture levels are low and the plants are desiccated and inhibits seedling establishment and causes developing tillers
dormant. Because defoliated plants have higher root:shoot to lodge in the grass canopy where they are more susceptible
ratios than unburnt plants, they can rehydrate and develop to drought (Egunjobi, 1974). Eventually, the sward becomes
water potentials above those in the soil. This enables them moribund and basal cover declines. In contrast, in semi-arid
to extract some of the residual moisture and grow (Fisher, savannas, where plant production is constrained by low
1978). If plants are burnt during the early dry season and rainfall and the short growing season, tillering and seedling
there in no recharge of the already depleted soil moisture establishment are seldom limited by the growth of the sward.
store during the period of regrowth, plant water potentials Consequently, basal cover tends to increase with decreasing
will gradually decline to wilting point as the increasing fire frequency (Kennan, 1972; Gertenbach and Potgieter, 1979;
transpirational demand eventually exceeds uptake. In this van Rooyen and Theron, 1982; Yeaton et al., in prep.). In
case, the regrowth will not be sustained through to the start general, therefore, regular dry season fires promote grass
of the wet season and there will be no increase in yield production in the wetter savannas, where the dry season is
(Afolayan and Fafunsho, 1978; Batmanian and Haridasan, 1985). short, while in the more arid areas production is depressed
In contrast, in areas where the dry season is relatively relative to adjacent unburnt areas (West, 1965; San Jose and
short, or where the plants are burnt only a few weeks prior Medina, 1975).
to the first rains, regrowth is able to continue into the wet There is a paucity of information on the effects of
season, giving the burnt plants the advantage of a longer fire on woody plant production. In mature woodlands, where
growth period. The resulting yields are higher than from damage to adult trees is low, moderate intensity fires have
unburnt plants (San Jose and Medina, 1975). no effect on stem radial growth (Geldenhuys, 1977). For
The advantage of early growth is lost if plants are smaller-sized plants, particularly individuals less than 3 m
burnt immediately prior to or just after the first spring in height, burning drastically reduces canopy volume and
rains. If the regrowth of the plants is delayed total yield therefore leaf biomass (Figure 5.4). Any short-term decline
may be less than that of unburnt plants (Tainton et al., in plant production resulting from this is partially offset
1977; Trollope, 1984a), even though the rate of biomass by an exponential increase in basal shoot production with
accumulation is higher (Grossman et al., 1981). Burning in increasing reduction in canopy leaf mass, a pattern which is
the middle of the wet season, when the grasses are most marked in multistemmed shrubs (Rutherford, 1981).
physiologically active and therefore more susceptible to Rejuvenation of this kind, while probably lowering total
damage, usually results in a significant drop in yield during production, could result in greater production per unit leaf
the subsequent growing season (Trollope, 1984a). area, depending on the extent to which fire alters the ratio
Grass yields are also affected by fire temperature. of assimilatory to support tissues, in the same way that
Significant reductions in yield occur where ground level pruning does. Clearly there is scope for more research in
temperatures exceed 95°C. Temperatures in turn are determined this area.
124 P.G.H.Frost and F.Robertson Effects of fire in savannas 125

5.2 PLANT QUALITY thereby lowering the probability of another fire, and this
Plants regrowing after fire often have significantly can lead to the eventual encroachment of the woody plants.
higher concentrations of nitrogen and minerals compared with In contrast, in moist savannas, sufficient grass is
unburnt or mechanically defoliated plants (Mes, 1958; Falvey, usually produced each year to fuel annual fires. Most of the
1977; Tainton et al., 1977; Abbadie, 1983; Griffin and grasses are tall, fast growing and fibrous, with the result
Friedel, 1984a; Batmanian and Haridasan, 1985). This increase that the level of herbivory is generally low. Under these
in plant quality is one of the main reasons for the frequent circumstances, fire alone is capable of restricting woody
burning of moist savanna grasslands by man during the dry plant recruitment and maintaining the balance between trees
season. The increase though is transitory, lasting only for and grass.
the early part of the growing season. Although there is some Grasses, trees, fire, and herbivores interact most
evidence for rapid uptake of nutrients released during the extensively in mesic savannas where year to year variations
fire or soon afterwards (Coutinho, 1982), most of the in rainfall and soil nutrient supply significantly affect the
difference arises from differences in the age and composition outcomes of the interactions. The key elements of these
of the regrowing tissues. The initial regrowth of burnt interactions are (i) the competition between trees and grass,
grasses, for example, is from young tillers in which there is with trees shading out grasses and competing with them for
a high proportion of leaf to stem. In contrast, in unburnt or water and nutrients, and the grasses outcompeting woody plant
mown plants, much of the initial regrowth comes from old seedlings; (ii) the suppression of tree recruitment by fire
tillers, while a high proportion of the residual material and browsers, acting separately or together; (iii) the
consists of low quality, fibrous, stems and leaf sheaths. inverse relationship between grazing and fire, with heavy
grazing reducing the dry-season standing crop of grass to
6. INTERACTIONS OF RAINFALL, PLANT PRODUCTION, FIRE AND levels below that needed to sustain fire; and (iv) the
HERBIVORES influence of grass production and quality on the kinds,
In many savannas, fire and herbivores interact to numbers and movements of grazers.
affect the dynamics of the vegetation, particularly the For example, in African savannas, elephants Loxodonta
grass:woody ratio. The interaction is not uniform across the africana act synergistically with fire in reducing woodland
range of savanna types but depends largely on the patterns of and thicket to grassland. By opening up the woody canopy,
plant production and quality, themselves the products of soil elephants enable more grass to grow and this provides fuel
water availability and nutrient supply as determined by a for fires (Norton-Griffiths, 1979). Fire inhibits tree
given rainfall regime and soil type (Bell, 1981). recruitment by killing seedlings and stunting the growth of
In semi-arid savannas, fire is an uncommon event; the young trees. Fire also keeps woody plants at a height and in
amount of grass produced is usually insufficient to support a a state that is highly acceptable to browsers. This further
dry season fire. In addition, because most of the grasses are limits recruitment (Trollope, 1980; Pellew, 1983). For
palatable, grazing pressure is often high and this further example, sustained heavy browsing by domestic goats reduced
reduces the standing crop of grass. Sufficient grass to fuel the numbers of Acacia karroo by 90% over five years following
a fire usually only accumulates after one or more seasons of a single late dry season fire (Trollope, 1984).
above-average rainfall. These conditions also favour the Increased grass production also favours grazers,
recruitment of woody plants and can trigger marked changes in provided that the grass is sufficiently nutritious. Frequent
composition, depending on subsequent events. An open fires can lead to an increase in the numbers of more
grass-dominated community can be maintained by a second fire, palatable grasses such as Themeda triandra (Norton-Griffiths,
occurring soon after the first, which kills the young plants 1979; Tainton and Mentis, 1984). The removal of moribund
before they have the chance to mature and reproduce material and the stimulus given to the growth of young
(Hodgkinson et al., 1983; Griffin and Friedel, 1985). This tillers keeps the grass layer in a more productive and
situation is rare and usually depends on the coincidence of palatable growth phase. This benefits short-grass,
several consecutive years of above-average rainfall and light area-selective grazers such as wildebeest Connochaetes
grazing. Heavy grazing can prevent the build-up of fuel,
126 P.G.H.Frost and F.Robertson Effects of fire in savannas 127

taurinus which congregate on recently burnt areas once the fire on single species, particularly the ways in which fire
grass begins to grow. (In contrast, dry season fires can affects individuals and how this varies with the age, size
reduce the availability of browse and cause browsing animals and physiological condition of the individual and with the
to move elsewhere [Bell and Jachmann, 1984]. This may type, timing and intensity of the fire. In this respect, we
increase browsing pressure on trees in unburnt areas need to know if plants which are drought-stressed, or
(Harrington and Ross, 1974).) recovering from herbivory, are likely to be more or less
Heavy grazing, particularly if the interval between affected by fire. More consideration needs to be given to the
the start of regrowth and subsequent defoliation is short, possible contingency of effects, not only on the intensity,
can limit the rate of recovery and reduce the standing crop timing and frequency of fire, but also on the state of the
of grass carried over into the dry season. This decreases the organisms at the time, as well as on subsequent interactions
likelihood of the area sustaining fire the following year. with herbivores, rainfall, drought, and other fires.
Fire can even be excluded in the longer term, provided that The following hypothesis is relevant in this regard:
the grazing pressure remains high. In this case, woody plants the main determinant of the effect of fire on population
are able to invade, especially when the competitiveness of structure and community composition in savannas is the
the grass layer is reduced by heavy grazing, drought, or interaction with future events, such as rainfall, drought, or
both. If this encroachment is not checked subsequently by herbivory, occurring during the recovery phase. Obviously,
fire or browsing, thickets and woodland may eventually the more severe a fire is, in terms of mortality or degree of
re-establish (Norton-Griffiths, 1979; Pellew, 1983). topkill, the longer the recovery phase and the greater the
Fire-maintained grasslands occur mainly in areas where potential for interaction with other events. Populations
drainage is restricted. In these areas, the grasses are which have to recover from seeds will be more susceptible to
usually too fibrous to support anything other than change than those whose members recover by resprouting. The
large-bodied grazers such as buffalo Syncerus caffer and rate of recovery is also affected by the favourableness of
elephant. As a result, the grass biomass is seldom reduced conditions for regrowth, especially the adequacy of water and
sufficiently to exclude fire (Bell, 1981). Invasion by woody nutrient supply. Therefore, as a corollary, we suggest that
plants under these conditions is difficult and only likely to the less favourable the environment, the slower the recovery
occur if the area is protected for long enough from fire (San phase will be, and that this increases the likelihood of
José and Farinas, 1983). subsequent events affecting the eventual outcome. We also
predict that the effects of fire will be less pronounced on
7. DISCUSSION sandy than on clayey soils because (i) the effect of exposure
Much of the current knowledge of fire and its effects of the soil surface is less severe on sandy soils (there is
on savanna structure and functioning has come from general less erosion, less compaction, and better infiltration
observation supplemented by information derived from a resulting from single-grain structure of these soils), and
limited number of experiments, most of which have been in (ii) the higher infiltration rates on sandy soils promote a
existence for many years. This is unusual, for most more rapid recovery of the plants because more of the
ecological experiments tend to be of relatively short rainfall becomes available for their growth.
duration. Consequently, one might expect that over the years
a significant body of understanding has been built up from
the results of these experiments. Is current understanding
sufficient to allow the effects of fires to be predicted?
This review suggests not.
There are still a number of areas where our
understanding of the ecological effects of fire is poor.
These include information on the spatial aspects of fire
behaviour, the degree of heterogeneity that results, what
causes this, and what the consequences are for plant
recruitment. Few studies have investigated the effects of
128 P.G.H.Frost and F.Robertson Effects of fire in savannas 129

REFERENCES
Abbadie, L. (1983). Contribution a l'etude de la production Blasco, F. (1983). The transition from open forest to
primaire et du cycle de l'azote dans les savanes de savanna in continental Southeast Asia. In: Bourliere,
Lamto (Cote-d'Ivoire). Oecologica Plantarum, 5 (19): F. (Ed), Ecosystems of the World, 13, Tropical
321-334. Savannas, pp. 167-181. Elsevier, Amsterdam.
Abbadie, L. (1984). Evolution saisonniere du stock d'azote Boaler, S.B. (1966). Ecology of Pterocarpus angolensis D.C.
dans la strate herbacee d'une savane soumise au feu en in Tanzania. Ministry of Overseas Development,
Cote d'Ivoire. Oecologica Plantarum, 5 (19): 321-334. London.
Abbadie, L. (in press). Evolution saisonniere du stock Boaler, S.B. and Sciwale, K.C. (1966). Ecology of a miombo
d'azote dans la strate herbacee d'une savane de Cote site, Lupa North Forest Reserve, Tanzania. III.
d'Ivoire protegee des feux de brousse. Oecologica Effects on the vegetation of local cultivation
Plantarum 6 (20). practices. Journal of Ecology 54: 577-587.
Abedeji, F.O. (1983). Effect of fire on soil microbial Booysen, P. de V., Tainton, N.M. and Scott, J.D. (1963).
activity in Nigerian southern Guinea savanna. Revue Shoot apex development in grasses and its importance
d'Ecologie et de Biologie du Sol, 20: 483-492. in grassland management. Herbage Abstracts, 33:
Afolayan, T.A. (1978). Effects of fire on the vegetation and 209-213.
soils in Kainki Lake National Park, Nigeria. In: Boultwood, J.N. and Rodel, M.G.W. (1981). Effects of
Hyder, D.N. (Ed.) Proceedings of the First stocking rate and burning frequency on
International Rangeland Congress, pp. 55-59. Brachystegia/Julbernardia veld in Zimbabwe.
Afolayan, T.A. and Fafunsho, M. (1978). Seasonal variation in Proceedings of the Grassland Society of southern
the protein content and the grazing of some tropical Africa 16: 111-115.
savanna grasses. East African Wildlife Journal, 16: Bourliere, F. (1983). Tropical Savannas. Elsevier,
97-104. Amsterdam.
Barnes, D.L. (1965). The effects of frequency of burning and Braithwaite, R.L. and Estbergs, J.A. (1985). Fire patterns
mattocking in the control of coppice in Marandellas. and woody vegetation trends in the Alligator Rivers
Rhodesian Journal of Agricultural Research 3: 55-56. region of North Australia. In: Tothill, J.C. and
Bate, G.C. (1981). Nitrogen cycling in savanna ecosystems. Mott, J.J. (Eds.), Ecology and Management of the
In: Clarke, F.E. and Rosswall, T. (Eds). Terrestrial World's Savannas, pp. 359-364. Australian Academy of
Nitrogen Cycles, pp. 463-475. Ecological Bulletin 33, Science, Canberra.
Stockholm. Brookman-Ammissah, J., Hall, J.N., Swaine, M.D. and
Batmanian, G.J. and Haridasan, M. (1985). Primary production Attakorah, J. Y. (1980). A re-assessment of a fire
and accumulation of nutrients by the ground layer protection experiment in North-Eastern Ghana savanna.
community of cerrado vegetation of central Brazil. Journal of Applied Ecology, 17: 85-99.
Plant and Soil, 88: 437-440. Budowski, G. (1956). Tropical savannas, a sequence of forest
Bell, R.H.V. (1981a). An outline of a management plan for felling and repeated burnings. Turrialba 6, 23-33.
Kasungu National Park, Malawi. In: Jewell, P.A. and Cass, A., Savage, M.J. and Wallis, F.M. (1984). The effect
Holt, S. (Eds), Problems in Management of Locally of fire on soil and microclimate. In: Booysen, P. de
Abundant Wild Mammals, pp. 69-89. Academic Press, New V. and Tainton, N.M. (Eds.), Ecological Effects of
York. Fire in South African Ecosystems, pp. 313-325.
Bell, R.H.V. and Jachmann, H. (1984). Influence of fire on Springer-Verlag, Berlin.
the use of Brachystegia woodland by elephants. African Cavalcanti, L.H. (1978). Efeito das cinzas resultantes de
Journal of Ecology, 22: 157-163. queimada sobre a productividade do estrato herbaceo
subarbustivo do cerrado de Emas. DSc thesis,
University of Sao Paulo, Brazil.
130 P.G.H.Frost and F.Robertson Effects of fire in savannas 131

Charter, J.R. and Keay, R.W.J. (1960). Assessment of the Fanshawe, D.B. (1959). Silviculture and management of miombo
Olokemeji fire-control experiment (Investigation 254) woodland. C.C.T.A./C.S.A. Publication No. 52.
28 years after institution. Nigerian Forestry Fisher, K.J. (1978). The recovery of leaf water potential
Information Bulletin (New Series), 3: 1-32. following burning of two droughted tropical pasture
Clayton, W.D.(1961). Derived savanna in Kabba province, species. Australia Journal of Experimental
Nigeria. Journal of Ecology, 49: 595-604. Agriculture and Animal Husbandry, 18; 423-425.
Collins, N.M. (1977). The population ecology and energetics Frost, P.G.H. (1984). The responses and survival of organisms
of Macrotermes bellicosus (Smeathman) Isoptera. PhD in fire-prone environments. In: Booysen, P. de V. and
thesis, Imperial College, London University, 339 pp. Tainton, N.M. (Eds.), Ecological Effects of Fire in
Cook, L. (1939). A contribution to our information on grass South African Ecosystems, pp. 273-309.
burning. South African Journal of Science, 36: Springer-Verlag, Berlin.
270-282. Frost, P.G.H. (1985). Organic matter and nutrient dynamics in
Coutinho, L.M. (1978). Aspectos ecologicos do fogo no a broadleafed African savanna. In: Tothill, J.C. and
cerrado. I - A temperatura do solo durante as Mott, J.J. (Eds.), Ecology and Management of the
queimadas. Revista Brasiliera de Botanica, 1: 93-97. World's Savannas, pp. 206-206. Australian Academy of
Coutinho, L.M. (1982). Ecological effects of fire in Science, Canberra.
Brazilian cerrado. In: Huntley, B.J. and Walker, B.H. Gandar, M.V. (1982). Description of a fire and its ecological
(Eds), Ecology of Tropical Savannas, pp. 273-291. effects in the Nylsvley Nature Reserve: a synthesis
Springer-Verlag, Berlin. report. South African National Scientific Programmes
Coutts, J.R.H. (1945). Effects of veld burning on the base Report, 63: 1-39.
exchange capacity of a soil. South African Journal of Geldenhuys, C.J. (1977). The effect of different regimes of
Science, 41: 218-224. annual burning on two woodland communities in Kavango.
Davidson, R.L. (1953). A seasonal burn experiment in sour South African Forestry Journal, 103: 32-42.
mixed bushveld. Annual Report of the University Field Gertenbach, W.P.D. and Potgieter, A.L.F. (1979).
Research Station, Frankenwald. pp. 29-32. University Veldbrandnavorsing in die struikmopanieveld van die
of the Witwatersrand, Johannesburg. Nasionale Krugerwildtuin. Koedoe, 22: 1-28.
Dommergues, Y. (1954). Action du feu sur la microflore des Gillon, D. (1970). Recherches ecologiques dans la savane de
sols de prairie. Memoire d'Institut Scientifique de Lamto (Cote d’Ivoire): les effets du feu sur les
Madagascar, D, 6: 149-158. arthropodes de la savane. La Terre et la Vie, 24:
Dye, P.J. and P.T. Spear. (1982). The effect of bushclearing 80-93.
and rainfall variability on grass yield and Gillon, D. (1983). The fire problem in tropical savannas. In:
composition in south-west Zimbabwe. Zimbabwe Journal Bourliere, F. (Ed.), Tropical Savannas, pp. 617-641.
of Agricultural Research, 20: 103-117. Elsevier, Amsterdam.
Edroma, E.L. (1984). Effects of burning and grazing on the Gillon, Y. (1972). The effect of bush fire on the principal
productivity and number of plants in Queen Elizabeth acridid species of an Ivory Coast savanna. Proceedings
National Park, Uganda. African Journal of Ecology, of the Tall Timbers Fire Ecology Conference, 11;
22: 165-174. 419-471.
Egunjobi, J.K. (1974). Dry matter, nitrogen and mineral Greenslade, P. and Mott, J.J. (1983). Effect of fire on
element distribution in an unburnt savanna during the invertebrates in an australian tropical grassland. In:
year. Oecologica Plantarum, 9: 1-10. Lebrun, P., Andre, H.M. Medts, A. de, Gregoire-Wibo C.
Eiten, G. (1972). The cerrado vegetation of Brazil. The and Wantky, G. (Eds.), New Trends in Soil Biology, pp.
Botanical Review 38(2): 201-341. 635-637. Pien Brichert, Louvain-la-Neuve (Belgium).
Falvey, J.L. (1977). Dry season regrowth of six forage
species following wildfire. Journal of Range
Management, 30. 37-39.
132 P.G.H.Frost and F.Robertson Effects of fire in savannas 133

Griffin, G.F. and Friedel, M.H. (1984a). Effects of fire on Huntley, B.J. (1982). Southern African savannas. In: Huntley,
central Australian rangelands. I. Fire and fuel B.J. and Walker, B.H. (Eds.), Ecology of Tropical
characteristics and changes in herbage and nutrients. Savannas, pp. 101-119. Springer-Verlag, Berlin.
Australian Journal of Ecology, 9: 381-393. Huntley, B.J. and Walker, B.H. (eds.). (1982). Ecology of
Griffin, G.F. and Friedel, M.H. (1984b). Effects of fire on Tropical Savannas. Springer-Verlag, Berlin.
central Australian rangelands. II. Changes in tree and Isichei, A.O. (1979). Elucidation of stocks and flows of
shrub populations. Australian Journal of Ecology, 9: nitrogen in some Nigerian savanna ecosystems. PhD
395-403. thesis, University of Ife, Ile-Ife, Nigeria.
Griffin, G.F. and Friedel, M.H. (1985). Discontinuous change Isichei, A.O. (1980). Nitrogen fixation by blue-green algal
in central Australia: some implications of major crusts in Nigerian savanna. In: Rosswall, T. (Ed.),
ecological events for land management. Journal of Arid Nitrogen Cycling in West African Ecosystems, pp.
Environments, 9: 63-80. 191-198. Royal Swedish Academy of Sciences,
Grossman, D., Grunow, J.O. and Theron, G.K. (1981). The Stockholm.
effect of fire, with and without subsequent Isichei, A.O. and Sanford, W.W. (1980). Nitrogen lost by
defoliation, on the herbaceous layer of Burkea burning from Nigerian grassland ecosystems. In:
africana savanna. Proceedings of the Grassland Society Rosswall, T. (Ed.), Nitrogen Cycling in West African
of southern Africa, 16: 117-120. Ecosystems, pp. 325-331. Royal Swedish Academy of
Harrington, G.N. (1974). Fire effects on a Ugandan grassland. Sciences, Stockholm.
Tropical Grasslands 8: 87-101. Jackson, G. (1974). Cryptogeal germination and other seedling
Harrington, G.N. and Ross, I.C. (1974). The savanna ecology adaptations to the burning of vegetation in savanna
of Kidepo Valley National Park. Part I. The effects regions: the origin of the pyrophytic habit. New
of burning and browsing on the vegetation. East Phytologist 73: 771-780.
African Wildlife Journal, 12: 93-106. Johnson, R.W. and Purdie, R.W. (1981). The role of fire in
Harrison, T.D. Report on maximum temperature measurements the establishment and maintenance of agricultural
during the Nylsvley veld fire of 5 September 1978 and systems. In: Gill, A.M., Groves, R.H. and Noble, I.R.
on postfire micrometeorological measurements. (Eds.), Fire and the Australian Biota, pp. 471-495.
Unpublished report to the National Programme for Australian Academy of Science: Canberra.
Environmental Sciences, CSIR, Pretoria, 19 pp. Kadeba, O. (1982). Studies on soil organic matter and
Hodgkinson, K.C. (1979). The shrubs of poplar box (Eucalyptus nitrogen under savanna forest reserves. In: Sanford,
populnea) lands and their biology. Australian W.W., Yefusu, H.M. and Ayeni, J.S.O. (Eds.) Nigerian
Rangeland Journal, 1: 280-293. Savanna, pp. 160-188. Kainji Lake Research Institute,
Hodgkinson, K.C., Harrington, G.N., Griffin, G.F., Noble, New Bussa.
J.C. and Young, M.D. (1984). Management of vegetation Kellman, M. and Miyanishi, K. (1982). Forest seedling
with fire. In: Harrington, G.N., Wilson, A.D. and establishment in Neotropical savannas: observations
Young, M.D. (Eds.), Management of Australia's and experiments in the Mountain Pine Ridge Savanna,
Rangelands, pp. 141-156. CSIRO, Melbourne. Belize. Journal of Biogeography 9: 193-206.
Hopkins, B. (1965). Observations on savanna burning in the Kellman, M., Miyanishi, K. and Hiebert, P. (1985). Nutrient
Olokemeji Forest Reserve, Nigeria. Journal of Applied retention by savanna ecosystems. II. Retention after
Ecology, 2: 367-381. fire. Journal of Ecology, 73: 953-962.
Hopkins, B. (1983). Successional processes. In: Bourliere, F. Kennan, T.C.D. (1972). The effects of fire on two vegetation
(Ed.), Ecosystems of the World. 13. Tropical types of Matopos, Rhodesia. Proceedings of the Tall
Savannas. Elsevier: Amsterdam. Timbers Fire Ecology Conference, 11: 53-98.
134 P.G.H.Frost and F.Robertson Effects of fire in savannas 135

Knoop, W.T. and Walker, B.H. (1985). Interactions of woody Menaut, J.C. (1977). Evolution of plots protected from fire
and herbaceous vegetation in a southern African since 13 years in a Guinea savanna of Ivory Coast.
savanna. Journal of Ecology 73: 235-253. Actas del IV Simposium Internacional de Ecologia
Komarek, E.V. (1972). Lightning the fire ecology in Africa. Tropical (Panama), 2: 541-558.
Proceedings of the Tall Timbers Fire Ecology Menaut, J.C. and Cesar, J. (1982). The structure and dynamics
Conference, 11: 473-509. of a West African savanna. In: Huntley, B.J. &
Lawson, G.W., Jenik, J. and Armstrong-Mensah, K.O. (1968). A Walker, B.H. (Eds.), Ecology of Tropical Savannas,
study of a vegetation catena in Guinea savanna at pp. 80-100. Springer-Verlag: Berlin.
Mole Game Reserve (Ghana). Journal of Ecology 66:. Mes, M.G. (1958). The influence of veld burning or mowing on
505-522. the water, nitrogen and ash content of grasses. South
Lawton, R.M. (1978). A study of the dynamic ecology of African Journal of Agricultural Science, 54: 83-86.
Zambian vegetation. Journal of Ecology, 66: 175-198. Moore, A.W. (1960). The influence of annual burning on a soil
Leigh, J.H. and Noble, I.R. (1981). The role of fire in the in the derived savanna zone of Nigeria. Transactions
management of rangelands in Australia. In: Gill, of 7th International Congress of soil Science, 4,
A.M., Groves, R.H. and Noble, I.R. (Eds.), Fire and 257-26
the Australian Biota, pp. 471-495. Australian Academy Moore, C.W.E. (1973). Some observations on ecology and
Science, Canberra. control of woody weeds on mulga lands in northwestern
Lock, J.M. and Milburn, T.R. (1971). The seed biology of New South Wales. Tropical Grasslands, 7: 79-88.
Themeda triandra Forsk. in relation to fire. Mott, J.J. and Andrew, M.H. (in press). The effect of fire on
Symposium of the British Ecological Society, 11: the population dynamics of native grasses in tropical
337-349. savannas of North-west Australia. In: Ridpath, M.G.
Malaisse, F. (1978). The Miombo Ecosystem. In: and Corbett, L.K. (Eds.), Ecology of Wet-Dry Tropics.
Unesco/UNEP/FAO, Tropical Forest Ecosystems pp. Proceedings of the Ecological Society of Australia.
589-606. Unesco, Paris. Mott, J.J., Williams, J., Andrew, M.H. and Gillison, A.N.
Mansfield, J.E., Bennett, J.G., King, R.B., Lang, D.M. and (1985). Australian savanna ecosystems. In: Tothill,
Lawton, R.M. (1976). Land resources of the Northern J.C. and Mott, J.J. (Eds.), Ecology and Management of
and Luapula provinces, Zambia - a reconnaissance the World's Savannas, pp. 56-82. Australian Academy
assessment. Volume 4. Ministry of Overseas of Science, Canberra.
Development, Surbiton. Norman, M.J.T. and Wetselaar, R. (1960). Losses of nitrogen
McClung, A.C. and de Freitas, L.M.M. (1959). Sulphur on burning native pasture at Katherine, N.T. Journal
deficiency in soils from Brazilian campos. Ecology of the Australian Institute of Agricultural Science,
40: 315-317. 26: 272-273.
Medina, E. (1982). Physiological Ecology of Neotropical Norton-Griffiths, M. (1979). The influence of grazing,
savanna plants. In: Huntley, B.J. and Walker, B.H. browsing and fire on the vegetation dynamics of the
(Eds.). Ecology of Tropical Savannas, pp. 308-335. Serengeti. In: Sinclair, A.R.E. and Norton-Griffiths,
Springer-Verlag, Berlin. M. (Eds.), Serengeti: dynamics of an Ecosystem, pp.
Medina, E., Mendoza, A. and Montes, R. (1978). Nutrient 310-352. University of Chicago Press, Chicago.
balance and organic matter production in the Nye, P.H. and Greenland, D.J. (1960). The soil under shifting
Trachypogon savannas of Venezuela. Tropical cultivation. Commonwealth Agricultural Bureaux
Agriculture (Trinidad), 55: 243-253. Technical Communication, 51: 1-156.
Meiklejohn, J. (1955). The effect of brush burning on the O'Connor, T.G. (1985). A synthesis of field experiments
microflora of a Kenya upland soil. Journal of Soil concerning the grass layer in the savanna regions of
Science, 6: 111-118. southern Africa. South African National Scientific
Programmes Report, 114: 1-126.
136 P.G.H.Frost and F.Robertson Effects of fire in savannas 137

Oguntala, A.B. (1980). The effects of fire on aspects of San Jose, J.J. and Farinas, M.R. (1983). Changes in tree
nitrogen cycling in the Olokomeji Forest Reserve, density and species composition in a protected
Nigeria. In: Rosswall, T. (Ed.), Nitrogen Cycling in Trachypogon savanna, Venezuela. Ecology, 64: 447-453.
West African Ecosystems. Royal Swedish Academy of San Jose, J.J. and E. Medina. (1975). Effect of fire on
Sciences, Stockholm. organic matter production and water balance in a
Pellew, R.A.P. (1983). The impacts of elephant, giraffe and tropical savanna. In: Golley, F.B. and Medina, E.
fire upon the Acacia tortilis woodlands of the (Eds.), Tropical Ecological Systems, pp. 251-264.
Serengeti. African Journal of Ecology, 1: 41-74. Sprinqer-Verlag, New York.
Penning de Vries, F.W.T. and Djiteye, M.A. (1982). La Sanford, W.W. (1982). The effects of seasonal burning: a
Productivite des Paturages Saheliens: Une Etude des review. In: Sanford, W.W., Yefusu, H.M. and Ayeni,
Sols, des Vegetations et de l'Exploitation de Cette J.S.O. (Eds.), Nigerian Savanna, pp. 160-188. Kainki
Resource Naturelle. Pudoc, Wageningen. Lake Research Institute, New Bussa.
Pitot, A. and Masson, H. (1951). Quelques donnees sur la Sarmiento, G. and Monasterio, M. (1983). Life forms and
temperature du cours des feux de brousse aux environs phenology. In: Bourliere, F. (Ed.), Ecosystems of the
de Dakar. Bulletin Institut Africain Noir, 13: World. 13. Tropical Savannas, pp. 79-108, Elsevier:
711-732. Amsterdam.
Raison, R.J. (1979). Modification of the soil environment by Sarmiento, G. (1984). The Ecology of Neotropical Savannas.
vegetation fires, with particular reference to Harvard University Press, Cambridge, Mass.
nitrogen transformations: a review. Plant and Soil, Savage, M.J. (1980). The effect of fire on the grassland
51: 73-108. microclimate. Herbage Abstracts, 50: 589-603.
Ramsey, J.M. and Rose-Innes, R. (1963). Some quantitative Shaw, N.H. (1957). Bunch speargrass dominance in burnt
observations on the effects of fire on the Guinea pastures in south-eastern Queensland, Australia.
Savanna vegetation of Northern Ghana over a period of Journal of Agricultural Research, 8: 325-334.
eleven years. African Soils, 8: 41-85. Siegfried, W.R. (1981). The incidence of veld-fire in the
Rawitscher, F. (1948). The water economy of the vegetation of Etosha National Park, 1970-1979. Madoqua, 12:
the "campos cerrados" in southern Brazil. Journal of 225-230.
Ecology, 36: 238-268. Singh, J.S., Hanxi, Y. and Sajise, P.E. (1985). Structural
Robertson, E.F. (1984). Regrowth of two African woodland and functional aspects of Indian and southeast Asian
types after shifting cultivation. PhD thesis, savanna ecosystems. In: Tothill, J.C. and Mott, J.J.
University of Aberdeen. (Eds.), Ecology and Management of the World's
Robinson, E.T., Gibbs-Russel, G.E., Trollope, W.S.W. and Savannas, pp. 34-51. Australian Academy of Science,
Downing, B.H. (1979). Assessment of short-term Canberra.
burning treatments on the herb layer of False Smith, E.L. (1960). Effects of burning and clipping at
Thornveld of the eastern Cape. Proceedings of the different times during the wet season on tropical
Grassland Society of southern Africa, 13: 79-83. tall grass range. Northern Australia Journal of
Rose Innes, R. (1971). Fire in West African vegetation. Rangeland Management, 13: 197-203.
Proceedings of the Tall Timbers Fire Ecology Stocker, G.C. and Mott, J.J. (1981). Fire in the tropical
Conference 11: 147-173. forests and woodlands of northern Australia. In:
Rutherford, M.C. (1981). Survival, regeneration and leaf Gill, A.M., Groves, R.H. and Noble, I.R. (Eds.), Fire
biomass changes in woody plants following spring and the Australia Biota, pp. 425-439. Australian
burns in Burkea africana – Ochna pulchra savanna. Academy of Science, Canberra.
Bothalia, 13: 531-552. Strang, R.M. (1966). The spread of Brachystegia spiciformis
Rutherford, M.C. (1982). Woody plant biomass distribution in Benth. and Julbernardia globiflora (Benth.) Troupin
Burkea savannas. In: Huntley, B.J. & Walker, B.H. in the Rhodesian highveld. Commonwealth Forestry
(Eds.), Ecology of Tropical Savannas, pp. 120-141. Review. 45: 253-256.
Springer-Verlag: Berlin.
138 P.G.H.Frost and F.Robertson Effects of fire in savannas 139

Strang, R.M. (1974). Some man-made changes in successional Trollope, W.S.W. (1980). Controlling bush encroachment with
trends on the Rhodesian highveld. Journal of Applied fire in the savanna area of South Africa. Proceedings
Ecology 11: 249-263. of the Grassland Society of Southern Africa, 15:
Sweet, R.J. (1982). Bush control with fire in Acacia 173-177.
nigrescens - Combretum apiculatum savanna in Trollope, W.S.W. (1982). The ecological effects of fire in
Botswana. Proceedings of the Grassland Society of South African savannas. In: Huntley, B.J. and Walker,
southern Africa, 17: 25-28. B.H. (Eds.), Ecology of Tropical Savannas, pp.
Tainton, N.M. and Mentis, M.T. (1984). Fire in grassland. In: 292-306. Springer-Verlag, Berlin.
Booysen, P. de V. and Tainton, N.M. (Eds.) Ecological Trollope, W.S.W. (1984a). Fire in savanna. In. Booysen, P. de
Effects of Fire in South African Ecosystems, pp. V. and Tainton, N.M. (Eds.), Ecological Effects of
113-147. Springer-Verlag, Berlin. Fire in South African Ecosystems, pp. 149-175.
Tainton, N.M., Groves, R.H. and Nash, R.C. (1977). Time of Springer-Verlag, Berlin.
mowing and burning of veld: short-term effects on Trollope, W.S.W. (1984b). Fire behaviour. In: Booysen, P. de
production and tiller development. Proceedings of the V. and Tainton, N.M. (Eds.), Ecological effects of
Grassland Society of southern Africa, 12: 59-64. Fire in South African Ecosystems, pp. 199-217.
Theron, G.C. (1937). Veld management investigations at the Springer-Verlag; Berlin.
School of Agriculture, Potchefstroom. Preliminary Trollope, W.S.W. and Potgieter, A.L.F. (1983).
Report. Bulletin 166, Department of Agriculture, Characteristics of fire behaviour in the Kruger
Union of South Africa, 23 pp. National Park. Unpublished Report, National Parks
Tothill, J.C. (1969). Soil temperatures and seed burial in Board of Trustees, Skukuza, South Africa, 38 pp.
relation to the performance of Heteropogon contortus van Rooyen, N. and Theron, G.K. (1982). ‘n Kwantitatiewe
and Themeda australis in burnt pastures in eastern analise van die Kruidstratum van die Eragrostis
Queensland. Australian Journal of Botany, 17: pallens - Burkea africana boomsavanne op die Nylsvley
269-275. natuurreservaat. South African Journal of Science,
Tothill, J.C. (1977). Seed germination studies with 78: 116-121.
Heteropogon contortus. Australian Journal of Ecology, van der Walt, P.T. and le Riche, E.A.N. (1984). The influence
2: 477-484. of veld fire on an Acacia erioloba community in the
Tothill, J.C. and Mott, J.J. (Eds.). (1985). Ecology and Kalahari Gemsbok National. Park. Koedoe (Supplement),
Management of the World's Savannas. Australian 103-106.
Academy of Science, Canberra. van Wyk, P. (1972). Veld burning in the Kruger National Park,
Trapnell, C.G. (1953). The soils, vegetation and agriculture an interim report of some aspects of research.
of North-Eastern Rhodesia. Revised Edition. Proceedings of the Tall Timbers Fire Ecology
Government Printer: Lusaka. Conference, 11: 9-31.
Trapnell, C.G. (1959). Ecological results of woodland burning Villecourt, P., Schmidt, W. and Cesar, J. (1980). Pertes d'un
experiments in Northern Rhodesia. Journal of Ecology, ecosysteme a l'occasion du feu de brousse (Savane
47: 129-168. tropicale de lamto, Cote d'Ivoire). Revue d'Ecologie
Trapnell, C.G., Friend, M.T., Chamberlain, G.T. and Birch, et de Biologie du Sol, 17: 7-12.
H.F. (1976). The effects of fire and termites on a Walker, J., Condon, R.W., Hodgkinson, K.C. and Harrington,
Zambian woodland soil. Journal of Ecology, 64: G.N. (1981). Fire in pastoral areas of poplar box
577-588. (Eucalyptus populnea) lands. Australian Rangelands
Trollope, W.S.W. (1974). The role of fire in preventing bush Journal, 3: 12-23.
encroachment in the eastern Cape. Proceedings of the Webber, C.N. (1979). The effects of fire on soil-plant
Grassland Society of southern Africa, 9: 67-72. ecological relationships in the southern part of the
Trollope, W.S.W. (1978). Fire behaviour - a preliminary Kruger National Park. MSc thesis, University of South
study. Proceedings of the Grassland Society of Africa, Pretoria.
Southern Africa, 13: 123-128.
140 P.G.H.Frost and F.Robertson

West, O. (1965). Fire in vegetation and its use in pasture

management with special reference to tropical and
subtropical Africa. Report of Commonwealth Bureau of
Pastures and Field Crops, Berkshire 1: 1-53.
West, O. (1972). Fire, man and wildlife as interacting
factors limiting the development of climax vegetation
in Rhodesia. Proceedings of the Tall Timbers Fire
Ecology Conference, 11: 121-145.
White, R.E. and Grossman, D. (1972). The effect of prolonged
seasonal burning on soil fertility under Trachypogon
- other species grassland at Frankenwald. South
African Journal of Science, 68: 234-239.

View publication stats