A reassessment of the fire-tolerance of some miombo woody

species in the Central Province, Zambia

A. E. Cauldwell* and U. Zieger
Centre for Wildlife Management, University of Pretoria, Pretoria, 0002, South Africa

Abstract
The fire-tolerance of 21 woody species in a dry miombo
woodland in the Central Province, Zambia was evaluated Introduction
and compared to an existing classification. The evalu-
Fire is an important ecological factor in the miombo
ation is based on differences in the vertical distribution
woodlands of southern and central Africa (Trapnell,
of canopy volumes over a range of height intervals after
1959; Thompson, 1975; Moss, 1976; Trapnell et al.,
repeated exposure of woodlands to fire. Woody species
1976; Lawton, 1978; Chidumayo, 1988). To evaluate
that showed an inverse correlation between canopy vol-
the impact of fire on such woodlands, the then Northern
ume and tree height were classified as fire-sensitive,
Rhodesia Forestry Department initiated long-term burn-
whilst those that showed a positive correlation were
ing trials in 1933 in the wet miombo woodland close to
classified as fire-tolerant. A new method of a rapid and
Ndola. Evaluation of these trials has clearly shown that
simple field evaluation is described, and further appli-
miombo woodlands are susceptible to fire, and that they
cations are discussed.
can be destroyed by repeated late-season, intense fires
(Trapnell, 1959; Chidumayo, 1988). Late-season burn-
Key words: canopy volume, fire ecology, miombo, ing inhibits the regeneration of trees, leading over time
structure to a loss of the tall canopy-forming trees of the genera
Brachystegia, Julbernardia and Isoberlinia. Not all woody
species, however, are equally sensitive to fire. Trapnell
(1959) and Lawton (1978) proposed a successional path-
Résumé way that is driven by fire in wet miombo woodland. In
On a évalué la tolérance au feu de 21 espèces ligneuses the absence of fire, the wet miombo woodland increases
dans une forêt sèche à Miombo, dans la Province du in density and becomes dominated by evergreen species
Centre, en Zambie, et on l’a comparée à une classification to eventually form a tall, dry evergreen forest, which is
existante. L’évaluation est basée sur des différences de la locally known as mateshi. On the other hand, repeated
distribution verticale des volumes de la canopée, sur une exposure to fire will convert the miombo woodland to an
gamme de hauteurs suite à l’exposition répétée des forêts open woodland of fire-tolerant trees that is dominated by
aux incendies. Les espèces d’arbres qui présentent une tall grasses, known locally as chipya. Trapnell (1959)
corrélation inverse entre le volume de la canopée et la and Lawton (1978) proposed a fire-tolerance classi-
hauteur des arbres ont été classées comme sensibles au fication in which woody species are classified into one of
feu, et celles qui présentaient une corrélation positive four fire-tolerance classes based upon abundance data
étaient classées comme tolérantes au feu. On présente collected over a range of fire treatments. The aim of the
une nouvelle méthode rapide et simple d’évaluation sur present study was to test these fire-tolerance classi-
le terrain et on discute d’autres applications. fications by evaluating differences in the structure of
woody plants after repeated exposure to fire in central
Zambia.
*Corresponding author Several methods for the quantification of the structure

138 © 2000 East African Wild Life Society, Afr. J. Ecol., 38, 138–146
 Fire-tolerance assessment of miombo tree species 139

of woody vegetation have been developed to date (e.g. cool fires early in the cool dry season to remove the excess
Coetzee & Gertenbach 1977; Smit 1989a; Teague 1989). fuel load.
A new method is proposed here, in which the field data
collection is simple and rapid. Data can be analysed on a
spreadsheet to calculate the canopy volume to any speci- Methods
fied height above ground level.
Field data collection
Transects of 2 × 100 m that were representative of their
surroundings in respect of both floristic composition and
Study area
structure were subjectively laid out within homogeneous
The study area is located in the Chibombo District of the vegetation units (Werger, 1974). Every living woody
Central Province of Zambia. The altitude of this region plant rooted within the transect was identified and the
varies from 1100 to 1200 m above sea level, and the following data were recorded:
annual rainfall from 800 to 900 mm (Hutchinson, 1 Species
1974). Three distinct seasons occur. They are: the warm 2 Shape: Fig. 1 illustrates the seven basic shapes that
rainy season from December to April, the cool dry season were used. These give a good approximation of the shape
from May to August, and the hot dry season from Sep- of most woody plants. The assumption was made that all
tember to November. The topography consists of gently woody plants can be adequately represented as circular
undulating terrain. The Chibombo District lies on the objects in the horizontal plane (Smit, 1989a).
watershed between the Zambezi and Kafue Rivers, and 3 The measurements X = total tree height, Y = height of
there are no perennial streams within the study area. the lowest leaves and 2R = widest diameter, as illustrated
The area is drained by dambos, which are seasonally in Fig. 1, were measured to the nearest 50 mm and rec-
waterlogged drainage lines with no clear water channels orded for all shapes.
(Bolnick, 1995). The vegetation in this region is domi- When measuring the dimensions of a tree, small stems
nated by central, dry miombo woodland (Chidumayo, protruding in odd directions were ignored (Smit, 1989a).
1987). The tree canopy ranges from 8 to 20 m in height All trees were measured as if circular. However, where
and the vegetation is classified as a tall, open woodland trees were noticeably ellipsoid in the horizontal section
(Edwards, 1983). Dry miombo woodland is separated the mean diameter was recorded. No tree can be perfectly
from wet miombo by the 1000 mm mean annual rainfall represented by any geometric shape and only the best
isohyet (White, 1983). approximation can be chosen.
Studies were conducted on Mtendere Game Ranch and
the nearby Mwomboshi State Forest. The fire man-
agement policy for the miombo woodlands of Mtendere Calculation of total canopy volume
Game Ranch from 1990 to 1996 was aimed at the pre- The following equations were used to calculate the total
vention of fires. When fires did occur, an attempt was canopy volume of the shapes A–G illustrated in Fig. 1:
made to extinguish them as quickly as possible. However,
a number of fires did occur and these woodlands can Shape A: total volume = pR2(X – Y)
now be classified into burnt and fire-protected variations Shapes B and C: total volume = 0.333pR2(X – Y)
based upon their floristic composition. The burnt miombo
Shape D: total volume = 1.333pR3
woodland has been exposed to three uncontrolled fires
in November 1991, July 1994 and October 1996. The Shapes E and F: total volume = 0.666pR3
fire-protected miombo woodland has not been burned
Shape G: This shape can be represented by the following
since 1990. The miombo woodlands of the Mwomboshi
two components:
State Forest support a similar plant species composition
a base cylinder with volume = pR2(X – Y – R)
to those of Mtendere Game Ranch. This state forest is
a half sphere with volume = 0.666pR53
managed to optimize its tree resources. As a safeguard
against hot fires that are potentially damaging to the These are summed to give total volume = pR2(X – Y –
larger trees, these woodlands are burned annually with 0.333R).

© 2000 East African Wild Life Society, Afr. J. Ecol., 38, 138–146
140 A. E. Cauldwell and U. Zieger

Fig 1 Seven basic shapes for structural

measurements of trees showing
measurements X, Y and 2R required per
tree where X = tree height, Y = height
of the lowest leaves and 2R = widest
diameter. H is a variable height used in
data analysis and is not measured in
the field. The measurements (mm) are
illustrated for shape G only, but are
applicable to all shapes.

Calculation of canopy volumes to a specified height H Shape C: the new shape becomes an incomplete cone
with radii R and Q. The equation for calculating volume
The following derivations give the equations that were
to H consists of the total volume minus the shape of the
used for calculating the canopy volume of a tree to a
small cone represented above height H with radius Q.
specified height denoted here as H (Fig. 1). When setting
a variable height H, the following situations could arise Volume to H = 0.333pR2(X – Y) – 0.333pQ2(X–H)
when calculating the canopy volume to height H:
Similarly as in Fig. 2, Q = R(X – H)(X – Y)−1
H greater than the tree height (X), then volume to H = to-
tal volume as described above Therefore volume to H = 0.333pR2[(X – Y) – (X –
H)3(X – Y)−2]
H less than the lowest leaves (Y), then volume to H = 0
Shapes D to G involve spheres or parts thereof. To
H less than X but greater than Y, then the volume to
derive the equation used to calculate the volume of a
H is defined in the equations described for each shape
partial segment of a sphere, a general volume equation
below:
was integrated where r2 = x2 + y2
Shape A: the resulting shape remains a cylinder, there-
V = rÐapy2dx
fore volume to H = pR2(H – Y)
Shape B: the resulting shape is represented by an = rÐap(r2 – x2)dx
inverted cone to height H with a radius denoted here as = p(r2x–0.333x3)ra
Q (Fig. 2)
= p(r3 – 0.333r3) – p(ar2–0.333a3)
Volume to H = 0.333pQ2(H – Y) as Q/(H – Y) = R/(X – Y)
or Q = R(H – Y)(X – Y)−1 = p[(r3 – ar2)–0.333(r3 – a3)]

Therefore: volume to H = 0.333pR2(H – Y)3(X – Y)−2 = p[r2(r – a) – 0.333(r – a)(r2 + ar + a2)]

Fig 2 Relationship between Q and R for

shape B, and between r, a and P when
calculating the volume of a partial
segment of a sphere as used for the
calculation of canopy volume of a
woodland.

© 2000 East African Wild Life Society, Afr. J. Ecol., 38, 138–146
 Fire-tolerance assessment of miombo tree species 141

Fig 3 The mean canopy volume (m3 per

200 m2) per height interval for twelve
transects exposed to regular fires, and
similarly for twelve fire-protected
transects in miombo woodlands in the
Central Province, Zambia; where the
height intervals were: A = 0.00–
0.20 m; B = 0.21–0.40 m; C = 0.41–
0.60 m; D = 0.61–0.80 m; E = 0.81–
1.00 m; F = 1.01–1.20 m; G = 1.21–
1.40 m; H = 1.41–1.60 m; I = 1.61–
1.80 m; J = 1.81–2.00 m.

Let P be the height of a partial segment of a sphere The total canopy volume of a transect was calculated
where P = r – a or a = r – P (Fig. 2) as the sum of the total volumes for all specimens recorded
within it. The canopy volume up to any specified height
then: volume = p[PR2–0.333P(R2 + R(R – P) + (R – P)2)] for the transect similarly represents the sum of canopy
= p[PR2–0.333P(3R2–3PR + P2)] volumes to the same specified height for all specimens.
Multiple heights were set at a series of equal intervals and
= p(P2R–0.333P3) the canopy volume per height interval for the transect
determined.
= 0.333p(3P2R – P3)

= 0.333pP2(3R – P) Determination of fire-tolerance

Shape D: As P = H – Y then the volume to a variable To determine the fire-tolerance of individual woody spec-
height H can be expressed as: ies, the general patterns of canopy volume per height
interval in the burnt and the fire-protected woodlands
Volume to H = 0.333p(H – Y)2(3R – H + Y)
were first determined (Fig. 3). Two different patterns were
Shape E: As P = H – Y then: revealed. In the burnt woodland, the canopy volume up
to a maximum of 2 m is inversely related to height above
Volume to H = 0.333p(H – Y)2(3R – H + Y)

Shape F: As P = X – H then:

Volume to H = 0.666pR3 – 0.333p(X–H)2(3R–X + H)

= 0.333p[2R3 – (X–H)2(3R–X + H)]

Shape G: If X – H − R then the volume is represented

by a cylinder, therefore:

Volume to H = pR2(H – Y)

If X – H ³ R then the volume is represented by a cylinder Fig 4 Hypothetical relationships between the mean canopy volumes
plus the segment of a half sphere with P = X – H therefore: of a woodland over a range of height intervals above ground level
for a woody species after repeated exposure to fire for a (A) fire-
Volume to H = pR2(X – R – Y) + 0.333p[2R3 – (X – tolerant woody species (B) semifire-tolerant woody species and (C)
2
H) (3R – X + H)] fire-sensitive woody species.

© 2000 East African Wild Life Society, Afr. J. Ecol., 38, 138–146
142 A. E. Cauldwell and U. Zieger

© 2000 East African Wild Life Society, Afr. J. Ecol., 38, 138–146
© 2000 East African Wild Life Society, Afr. J. Ecol., 38, 138–146
Fig 5 Patterns of canopy volume (m3 per 200 m2) distribution of a woodland over a range of height intervals (m) for 21 miombo woody species in the burnt miombo
woodlands of Mtendere Game Ranch and the Mwomboshi State Forest in the Central Province, Zambia. n = number of plants measured and r2 = correlation
coefficient for the regression.
Fire-tolerance assessment of miombo tree species
143
144 A. E. Cauldwell and U. Zieger

ground level. Many saplings are killed by fire, while plotted against the height intervals and a regression
others have their above ground plant material burned curve was fitted. The general trend of the regression line
off and coppice from the base. This happens each time was then compared with the three hypothetical patterns
that they are burned, which leads to an accumulation of (Fig. 4). A species in which the canopy volume was pro-
small stunted plants, but a lack of plant material at higher portionately related to height was classified as fire-toler-
levels. In the fire-protected woodland, a steady increase ant. A species in which the canopy volume was inversely
in canopy volume in relation to height was observed up related to height was classified as fire-sensitive. A species
to a maximum of 2 m. Fewer seedlings are present here, in which the canopy volumes remained constant over
but recruitment and growth appear to be undisturbed. the range of height intervals, or showed no correlation
Trapnell (1959), Chidumayo (1988) and Martin (1996) to height was classified as semifire-tolerant.
have recorded similar observations in a wet miombo
woodland.
Results and discussion
When individual plants of a fire-sensitive woody spec-
ies are exposed to repeated burning, they are either killed Regression curves were obtained for 21 woody species
or their growth is reversed by fire, forcing them to coppice (Fig. 5), and their fire-tolerance was evaluated. The
from ground level (Dublin, Sinclair & McGlade, 1990). results are summarized in Table 1, which also shows the
When plant growth is reversed, the distribution pattern fire-tolerance classes of Trapnell (1959), Lawton (1978)
of the canopy volumes of such plants over a range of and Chidumayo (1994).
height intervals should resemble the general pattern The majority of woody species that were evaluated in
obtained for the burned woodlands (Fig. 3). Fire-tolerant this study are fire-sensitive. This agrees with the results
species should withstand the effects of fire, and not be of the experimental burning trials evaluated by Trapnell
susceptible to reversal by fire. Their vertical distribution (1959) and Chidumayo (1988), who found that the
pattern of canopy volumes over a range of height inter- miombo woodlands are generally fire-sensitive. The
vals would therefore resemble that found in the fire- important canopy-forming species in these woodlands
protected woodland where the mean canopy volume is are Brachystegia spiciformis and Julbernardia globiflora.
directly related to height above ground level. Semifire- These were classified as fire-sensitive in the present study
tolerant woody species would be partly affected by fire, and also by Trapnell (1959) and Lawton (1978). The
and their canopy volume would neither increase nor fire-tolerant species are short-growing and do not con-
decrease over a range of height intervals. The three pat- tribute to the upper canopy of a miombo woodland. None
terns of canopy distribution based on this hypothesis of the fire-sensitive species of Lawton (1978) occurred in
appear in Fig. 4. the burnt or the fire-protected woodlands of the present
Data obtained from the transects in the burnt wood- study. This indicates a species difference between the wet
lands of Mtendere Game Ranch and the nearby Mwom- miombo woodlands in which Lawton (1978) conducted
boshi State Forest were used. All the measurements for his study and the dry miombo woodlands here.
each woody species were pooled. The data were then Trapnell (1959) and Lawton (1978) allocated woody
analysed using the above equations entered into a spre- plants to four fire-tolerance classes based upon the rela-
adsheet. The following ten equal height intervals were tive abundance of species in miombo and chipya wood-
used for the analysis: 0.00–0.20 m; 0.21–0.40 m; 0.41– lands exposed to various burning regimes. Many of the
0.60 m; 0.61–0.80 m; 0.81–1.00 m; 1.01–1.20 m; species classified by Trapnell (1959) and Lawton (1978)
1.21–1.40 m; 1.41–1.60 m; 1.61–1.80 m and 1.81– as fire-tolerant or semifire-tolerant are classified here as
2.00 m. Catunaregam spinosa, Friesodielsia obovata and fire-sensitive (Table 1). The classification used is the pre-
Multidentia crassa are short growing woody species that sent study is based upon woody structure rather than
do not reach a height of 2 m. These were analysed using abundance values. When a species belongs to the fire-
the following ten height intervals: 0.00–0.10 m; 0.11– sensitive class in the present study, this does not imply
0.20 m; 0.21–0.30 m; 0.31–0.40 m; 0.41–0.50 m; that it will disappear or be reduced in abundance after
0.51–0.60 m; 0.61–0.70 m; 0.71–0.80 m; 0.81–0.90 m exposure to fire. The fire-sensitive class proposed here
and 0.91–1.00 m. suggests that such species would not establish themselves
The mean canopy volumes for each species were then as mature trees in an environment that is repeatedly

© 2000 East African Wild Life Society, Afr. J. Ecol., 38, 138–146
 Fire-tolerance assessment of miombo tree species 145

Table 1 Fire-tolerance classification for some miombo woody species of Mtendere Game Ranch and the Mwomboshi State Forest in
the Central Province, Zambia as determined from their pattern of canopy distribution over a range of height intervals up to
2 m after repeated exposure to fire. Fire-tolerance classes of the following authors are presented: 1 = Trapnell (1959); 2 = Lawton
(1978); 3 = Chidumayo (1994)

Fire-tolerance classification

Woody species Present Study Other Classification

Albizia antunesiana Harms Fire-tolerant Wide ecological range 1

Brachystegia spiciformis Benth. Fire-sensitive Semifire-tolerant 1
Burkea africana Hook. Semifire-tolerant Fire-tolerant 2
Catunaregam spinosa (Thunb.) Tirveng. Fire-sensitive Wide ecological range 2
Combretum adenogonium Steud. ex A.Rich. Fire-sensitive Fire-tolerant 2
Combretum molle R.Br. ex G.Don Fire-sensitive Fire-tolerant 1
Combretum mossambicense (Klotzsch) Engl. Fire-sensitive –
Dalbergiella nyassae Bak.f. Fire-sensitive –
Dichrostachys cinerea (L.) Wight & Arn. Fire-sensitive Fire-sensitive 3
Diplorhynchus condylocarpon (Müll.Arg.) Pichon Semifire-tolerant Fire-tolerant 1
Friesodielsia obovata (Benth.) Verdc. Fire-sensitive –
Hexalobus monopetalus (A.Rich.) Engl. & Diels Fire-sensitive Fire-tolerant 1
Julbernardia globiflora (Benth.) Troupin Fire-sensitive Fire-sensitive 2
Monotes glaber Sprague Fire-tolerant Semifire-tolerant 1
Multidentia crassa (Hiern) Bridson & Verdc. Semifire-tolerant –
Ochna schweinfurthiana F.Hoffm. Fire-sensitive Fire-tolerant 1
Pseudolachnostylis maprouneifolia Pax Fire-tolerant Fire-tolerant 1
Strychnos spinosa Lam. Semifire-tolerant Fire-tolerant 1
Vangueria infausta Burch. Fire-sensitive –
Vitex madiensis Oliv. Fire-sensitive Fire-sensitive 2
Xeroderrus stuhlmannii (Taub.) Mendonça & E.C. Sousa Fire-sensitive –

exposed to fire. However, woody plants, which are highly woody species. However, according to the results of the
susceptible to fire, are likely to be killed or become sev- present study, only Julbernardia globiflora is a fire-sensitive
erely reduced in abundance in regularly burnt miombo species, while the other three species are fire-tolerant or
woodlands. Few specimens of such species would there- semifire-tolerant. Factors other than bark thickness must
fore be available for measurements. The approach used therefore also play a role in determining the fire-tolerance
in the present study therefore probably under-represents of a species.
the spectrum of fire-sensitive trees. However, although
only two specimens of Dichrostachys cinerea were present
Conclusion
in the burnt miombo woodlands in the present study,
results indicated that this species is susceptible to fire. The fire-tolerance classification of woody plants has use-
Results for species with low abundance in the burnt ful applications in the fields of forestry and protected area
miombo woodlands can therefore still be conclusive. management, and is useful to assess and monitor the fire
Dichrostachys cinerea was recorded to have a 100% mor- regime status of woodlands. The method introduced here
tality rate in dry miombo woodlands after repeated has also other applications, such as in the structural
exposure to fire (Chidumayo, 1994). descriptions of vegetation, comparison of the impact of
Chidumayo (1994) speculated that saplings of Burkea treatments on woody vegetation structure (fire or brows-
africana, Julbernardia globiflora, Monotes species and ing intensity), monitoring of woody vegetation, or the
Pseudolachnostylis maprouneifolia would be fire-sensitive determination of the horizontal visibility at various
because their bark is thinner than that of most miombo height levels. The method can further be used as a step

© 2000 East African Wild Life Society, Afr. J. Ecol., 38, 138–146
146 A. E. Cauldwell and U. Zieger

towards correlating canopy volume to foliage availability HUTCHINSON, P. (1974) The Climate of Zambia (Occasional Study
in the assessment of browse resources (Smit, 1989b). no. 7). Zambia Geographical Association, Lusaka.
LAWTON, R.M. (1978) A study of the dynamic ecology of Zambian
vegetation. J. Ecol. 66, 175–198.
MARTIN, S.C. (1996) The effects of burning on the regeneration
of miombo woodland in Kasanka National Park, Zambia. In:
Acknowledgements Biological Research in Kasanka National Park, Zambia (Eds D.
We express sincere thanks to the following persons for BURNHAM and P. RIORDAN). Manchester Metropolitan
University, Manchester.
assistance in this study: Prof. J. du P. Bothma, Prof. G. J.
MOSS, P.F.N. DE V. (1976) Kafue National Park – a Management
Bredenkamp, Mr and Mrs M. O’Donnell, Mr D. Manda Plan. National Parks and Wildlife Service, Chilanga.
and Mr M. G. Bingham. SMIT, G.N. (1989a) Quantitative description of woody plant
communities: Part I. An approach. J. Grassl. Soc. S. Afr. 6, 186–
191.
SMIT, G.N. (1989b) Quantitative description of woody plant
References communities: Part II. Computerised calculation procedures. J.
Grassl. Soc. S. Afr. 6, 192–194.
BOLNICK, D. (1995). A Guide to the Common Wildflowers of Zambia. TEAGUE, W.R. (1989) Monitoring vegetation change and
Macmillan Publishers, Lusaka. assessing veld condition. Monitoring vegetation in thorn/grass
CHIDUMAYO, E.N. (1987) Species structure in Zambian miombo communities. In: Veld and Pasture Management in the Eastern
woodland. J. trop. Ecol. 3, 109–118. Cape (Eds J. E. DANCKWERTS and W. R. TEAGUE). Pasture
CHIDUMAYO, E.N. (1988) A reassessment of the effects of fire on Research Section, Eastern Cape Region, Stutterheim.
miombo regeneration in the Zambian Copper belt. J. trop. Ecol. THOMPSON, P.J. (1975) The role of elephants, fire and other agents
4, 361–372. in the decline of a Brachystegia boehmii woodland. J. S. A. Wildl.
CHIDUMAYO, E.N. (1994) Handbook of Miombo Ecology and Man. Assoc. 5, 11–18.
Management. Draft edition. Stockholm Environment Institute, TRAPNELL, C.G. (1959) Ecological results of woodland burning
Stockholm. experiments in Northern Rhodesia. J. Ecol. 47, 129–168.
COETZEE, B.J. & GERTENBACH, W.P.D. (1977) Technique for TRAPNELL, C.G., FRIEND, M.T., CHAMBERLAIN, G.T. & BIRCH, H.F.
describing woody vegetation composition and structure in (1976) The effects of fire and termites on a Zambian woodland
inventory type classification, ordination and animal habitat soil. J. Ecol. 64, 577–588.
surveys. Koedoe 20, 67–75. WERGER, M.J.A. (1974) On concepts and techniques applied in
DUBLIN, H.T., SINCLAIR, A.R.E. & MCGLADE, J. (1990) Elephants the Zürich-Montpellier method of vegetation survey. Bothalia
and fire as causes of multiple stable states in the Serengeti- 11, 309–323.
Mara woodlands. J. Anim. Ecol. 59, 1147–1164. WHITE, F. (1983) The Vegetation of Africa. Unesco, Paris.
EDWARDS, D. (1983) A broad-scale structural classification of
vegetation for practical purposes. Bothalia 14, 705–712. (Manuscript accepted 25 September 1999)

© 2000 East African Wild Life Society, Afr. J. Ecol., 38, 138–146