ARTICLE IN PRESS

Perspectivesin Plant Ecology,Evolution andSystematics

1433-8319/$ - se

doi:10.1016/j.pp

�Correspondfax: +4690 786

E-mail addr

Savadogo.Patri

(D. Tiveau), sa

Mulualem.Tiga

Perspectives in Plant Ecology, Evolution and Systematics 10 (2008) 179–195

www.elsevier.de/ppees

Herbaceous species responses to long-term effects of prescribed fire, grazing

and selective tree cutting in the savanna-woodlands of West Africa

Patrice Savadogoa,�, Daniel Tiveaub, Louis Sawadogoc, Mulualem Tigabua

aDepartment of Forest Genetics and Plant Physiology, Tropical Silviculture and Seed Laboratory,

Swedish University of Agricultural Sciences SLU, SE-90183 Ume (a, SwedenbCenter for International Forestry Research (CIFOR) 06 BP 9478, Ouagadougou 06, Burkina FasocCentre National de Recherche Scientifique et Technologique, Institut de l’Environnement et de Recherches Agricoles,

Departement Production Forestiere, BP 10, Koudougou, Burkina Faso

Received 27 November 2006; received in revised form 27 March 2008; accepted 27 March 2008

Abstract

Grazing, fire and selective tree cutting are major disturbances that shape species diversity in savanna ecosystems, yettheir effects are highly variable. We carried out a factorial experiment with two levels to examine the effects of grazing,fire and selective tree cutting on herbaceous species richness, abundance and diversity on two sites in the Sudaniansavanna-woodlands of Burkina Faso for 10 years (1994–2003). The results showed significant inter-annual variation inspecies richness, abundance and diversity at both sites (po0.001), while main or combined effects of fire, grazing andselective cutting were very limited and varied between life forms and sites. Grazing tended to favour the diversity ofperennial grasses; fire tended to influence the richness of annual grasses and abundance and diversity of perennialgrasses while selective tree cutting had no effect on any of the vegetation attributes assessed. The combined effect ofgrazing, fire and selective cutting tended to increase the diversity of forbs. In many cases, the responses of herbaceousspecies to treatments were clearer on the site with deeper soils than the one with shallow soils. Depending on the siteand treatments, the inter-annual variation in vegetation attributes was partly related to amount and/or frequency ofrainfall and partly to inter-annual variation in grazing or fire intensity. It can be concluded that both disturbances andclimatic condition influence the structure and diversity of herbaceous flora in the Sudanian savanna-woodlandecosystem. The responses were site-specific, which accentuates the importance of landscape-scale approaches tounderstand the impacts of disturbances on composition, structure and diversity of savanna ecosystems.r 2008 Rubel Foundation, ETH Zurich. Published by Elsevier GmbH. All rights reserved.

Keywords: Burkina Faso; Conservation; Disturbance; Diversity; Herbivory; Species richness

e front matter r 2008 Rubel Foundation, ETH Zurich. Pub

ees.2008.03.002

ing author. Tel.: +4690 786 83 32;

83 14.

esses: Patrice.Savadogo@genfys.slu.se,

ce@gmail.com (P. Savadogo), d.tiveau@cgiar.org

wadogo_ls@hotmail.com (L. Sawadogo),

bu@genfys.slu.se (M. Tigabu).

Introduction

Savanna ecosystems are often subjected to multipledisturbances, such as grazing, fire and selective treecutting (Breman and Kessler, 1995), which are majorfactors that shape species diversity (McNaughton, 1983;van Langevelde et al., 2003). Generally, local species

lished by Elsevier GmbH. All rights reserved.

ARTICLE IN PRESSP. Savadogo et al. / Perspectives in Plant Ecology, Evolution and Systematics 10 (2008) 179–195180

richness and diversity of savanna ecosystems aremaintained by a dynamic interaction between localcolonisation processes from species pools at largerspatial scales and local extinction due to competitiveexclusion processes, which in turn are influenced bydisturbances (Gibson and Brown, 1991). Herbivoresaffect plant diversity through each or both of theseprocesses (Olff and Ritchie, 1998). For example,herbivores enhance local colonisation processes throughenhanced propagule dispersal and increasing availabilityof light and improving soil conditions while reducinglocal extinction rates by preferentially consumingcompetitively dominant plants and by allowing morefunctionally different plant species to coexist. Conver-sely, herbivores reduce the colonisation process throughconsumption of seeds and reproductive structures. Theycan enhance the extinction processes by preferentiallygrazing on rare, high-quality plant species, such asforbs, and favour the dominance of few species that cantolerate high grazing pressure.

Fire is a prominent feature of most tropical savannaecosystems and between 25% and 50% of the Sudanianzone burns annually primarily due to anthropogeniccauses (Menaut et al., 1991). Its effect on herbaceousvegetation could be positive or negative depending onthe intensity and severity of burning (Hoffmann, 1999;Garnier and Dajoz, 2001). Fire burning early in the dryseason tends to be of low intensity as the predominantlyherbaceous fuel still holds moisture from the wet season(Liedloff et al., 2001). It thus enhances the colonisationprocesses by inducing a flush of germination andflowering, a transient increase in overall productivitydue to removal of litter that enhances the availability ofnutrients, space and light, as well as maintainingtussocks and increasing their cover by favouring tillering(Whelan, 1995; Williams et al., 1999; Garnier andDajoz, 2001). On the other hand, high-intensity firesmay create pH and osmotic conditions unfavourable forgermination of some species. Post-fire gaps may bedrought-prone, as increased exposure may lead toelevated evaporation and thus reduced moisture avail-ability at shallow depths where germination occurs; andhence contributing to enhanced extinction processes(Elberse and Breman, 1990).

Canopy gaps created by tree removal is expected toresult in increased diversity and abundance due toreduced competition for water and nutrients as well asincreased availability of light (Frost et al., 1986; Casadoet al., 2004), thus contributing to enhanced localcolonisation processes, although the opposite has some-times been found (Akpo et al., 1999), especially forisolated trees (Belsky, 1992). In arid and semi-arid areas,the canopy gaps created by selective removal of treesmay create unfavourable thermal conditions and favourthe growth of drought-tolerant species, thereby con-tributing to competitive exclusion processes. In addi-

tion, as most savanna trees regenerate vegetatively post-cutting disturbance (Sawadogo et al., 2002; Ky-Dembeleet al., 2007), the compensatory response of trees maylead to increased competition for water and nutrients tothe detriment of the herbaceous vegetation.

These disturbances can act independently or addi-tively (Belsky, 1992; Drewa and Havstad, 2001; Valoneet al., 2002; Valone, 2003). Generally, the co-occurrenceof fire and grazing has a synergistic effect on plantcommunities both in time and space. Many grazers areattracted to recently burnt ground to feed on the post-fire regrowth of grasses. Grazers, in turn, reduce the fuelload by consumption and trampling and therefore lowerthe intensity and frequency of fire. Selective cuttingprovides more space and resources for the growth ofherbaceous species temporarily, which in turn attractsmore grazers and the relatively high grazing intensityreduces fire-severity by reducing the fuel load. Selectivecutting may favour the dominance of drought-tolerantspecies (e.g. perennial grasses) while grazing and firereduce their abundance and the combined effect of thesethree factors eventually reduces the overall diversity ofherbaceous flora. Nonetheless, it is still difficult tomake generalisations as effects of these disturbancesare still highly variable depending on the types ofplant community and their interaction with sitespecific ecological factors, such as soil and precipitation(e.g. Belsky, 1992; Harrison et al., 2003).

In West African savanna-woodland, the focus ofmodels and experimental studies investigating the rolesof biotic and abiotic factors has mainly been on treepopulation dynamics (Gignoux et al., 1997; Sawadogoet al., 2002) or on plant community responses to singletreatment of either grazing (Hiernaux, 1998) or burning(Garnier and Dajoz, 2001). Few studies conducted ingrassland communities elsewhere have shown interac-tion effects (Collins, 1987; Belsky, 1992; Valone andKelt, 1999; Drewa and Havstad, 2001; Valone, 2003).Additionally, most studies have been of relatively shortduration rather than as time series investigations, andtherefore unable to evaluate long-term responses. Long-term multi-factor studies are essential to get a completepicture of the responses of herbaceous vegetation toseveral disturbance regimes and their interaction withother ecological factors such as rainfall, which is a keygrowth-limiting factor in dry savanna ecosystems. Inthis paper, we presented 10 years data (1994–2003) onthe effects of annual burning, grazing and selective treecutting on species richness, abundance and diversity ofherbaceous vegetation in the Sudanian savanna-wood-lands of Burkina Faso. The current managementregimes in these savanna-woodlands entail prohibitionof grazing, application of annual early fire and selectivetree cutting of 50% of the merchantable standingvolume on a 20-year rotation period (Bellefontaineet al., 2000), which are not based on scientific evidence.

ARTICLE IN PRESS

200

400

600

800

1000

1200

Ann

ual r

ainf

all (

mm

)

50

60

70

80

90

Num

ber

of r

ainy

day

s

Laba

P. Savadogo et al. / Perspectives in Plant Ecology, Evolution and Systematics 10 (2008) 179–195 181

Thus, results from the present study will provideknowledge for an informed discussion on the effects ofthese management regimes on the herbaceous vegeta-tion. The specific questions we addressed were: (1) howdoes herbaceous vegetation respond to fire, grazing andselective tree cutting as well as treatment combinationsthereof? (2) Is there inter-annual variation in theresponse of herbaceous vegetation? If so, how does thisvariation interact with disturbance regimes? (3) Couldthe inter-annual variation be explained by ecologicalfactors; mainly amount and distribution of rainfall?

0 40

0

200

400

600

800

1000

1200

Rainfall per annum Rainy days per annum

Ann

ual r

ainf

all (

mm

)

40

50

60

70

80

90Tiogo

Num

ber

of r

ainy

day

s

1992 1994 1996 1998 2000 2002

1992 1994 1996 1998 2000 2002

Fig. 1. Total annual rainfall (mm/year) and number of rainy

days during the study period at Laba and Tiogo.

Materials and methods

Site description

The experimental sites are located on flat areas inLaba (111400N, 21500W) and Tiogo (121130N, 21420W)State forests (forets classees), both at an altitude of300m a.s.l. in Burkina Faso, West Africa. The Laba andTiogo State forests were delimited by the colonialFrench administration in 1936 and 1940 and cover17,000 and 30,000 ha, respectively. Both forests arelocated along the only permanent river (Mouhoun,formerly known as Black Volta) in the country. Theunimodal rainy season lasts for about 6 months, fromMay to October. Based on data collected from in situ

mini-weather station at each site, the mean annualrainfall during the study period (1994–2003) was8837147mm for Laba and 8567209mm for Tiogo,and the number of rainy days per annum was 75716and 7079 for Laba and Tiogo, respectively (Fig. 1).Mean daily minimum and maximum temperaturesranged from 16 to 32 1C in January (the coldest month)and from 26 to 40 1C in April (the hottest month). Mostfrequently encountered soils are Lixisols (Driessen et al.,2001), and the soil at Laba is shallow (o45 cm depth)silty-sand while it is mainly deep (475 cm) silty-clay atTiogo. These soils are representative of large tracts ofthe Sudanian Zone in Burkina Faso (Pallo, 1998).

Phyto-geographically, the study sites are situated inthe Sudanian regional centre of endemism in thetransition from the north to south Sudanian Zone(Fontes and Guinko, 1995). The Sudanian savanna is anarea stretching across the African continent fromSenegal in the west to the Ethiopian highlands in theeast and is characterised by 6–7 months of dry seasonand a mean annual rainfall between 700 and 1200mm(Breman and Kessler, 1995; Menaut et al., 1995). Thevegetation type at both sites is a tree/bush savanna witha grass layer dominated by the annual grasses Andro-

pogon pseudapricus Stapf. and Loudetia togoensis

(Pilger) C.E. Hubbard as well as the perennial grassesAndropogon gayanus Kunth. (dominant in Tiogo) and

Andropogon ascinodis C.B.Cl. (dominant in Laba). Themain forb species are Cochlospermum planchonii Hook.F., Borreria stachydea (DC.) Hutch. and Dalz., Borreria

radiata DC. and Wissadula amplissima Linn. Species ofthe families Mimosaceae and Combretaceae dominatethe woody vegetation component at both sites. In termsof basal area, the main woody species are Detarium

microcarpum Guill. & Perr., Combretum nigricans Lepr.ex Guill. & Perr., Acacia macrostachya Reichenb. exBenth., Entada africana Guill. & Perr., Lannea acida

A. Rich., Anogeissus leiocarpus (DC.) Guill. & Perr. andVitellaria paradoxa C.F. Gaertn. Prior to the start of theexperiments, the mean basal area of woody species atLaba was 10.7 and 6.3m2 ha�1 at stump level (20 cm)and breast height (130 cm), respectively, with the standdensity of 582 individuals ha�1 having at least one stemX10 cm GBH (girth at breast height). At Tiogo, theequivalent values were 10.9m2 ha�1 at stump level,6.1m2 ha�1 at breast height and 542 individuals ha�1.Both sites were frequently grazed by livestock and wildanimals and burnt almost every year during the dryseason (November–May) long before the start of theexperiment. The presence of livestock in the two Stateforests varies spatially and temporally, occurring mainly

ARTICLE IN PRESSP. Savadogo et al. / Perspectives in Plant Ecology, Evolution and Systematics 10 (2008) 179–195182

during the rainy season (June–October) when the grassis green and the surrounding areas are cultivated.During the dry season, they graze on straws in the bushclumps that have escaped the fire as well as the youngshoots of perennial grass species and young woodyfoliage induced by the fire.

Experimental design

A factorial experiment was established in each of thetwo State forests to examine the effects of grazing, earlyfire, selective cutting and their interaction on composi-tion, abundance and diversity of herbaceous vegetation.Each experimental site (18 ha) was divided into eightblocks (2.25 ha); four of which were fenced to excludelivestock (hereafter referred to as non-grazed plots) andthe other four were open for grazing (hereafter referredto as grazed plots). Each block was further divided intofour plots of 0.25 ha (50� 50m2), separated from eachother by 20–30m fire-breaks (Fig. 2). To the four plotswithin each block, the following treatments wererandomly assigned: no cutting – no fire, no cutting –early fire, cutting – no fire, and cutting – early fire. Theselective cutting was done in December 1993 at Tiogoand a month later in January 1994 at Laba by removing50% of the merchantable standing volume. Prior tocutting, all species were categorised according to theirlocal uses as protected species, timber, poles andfuelwood, and others (Hagberg et al., 1996; Sawadogo,1996). Except protected species, individuals of othercategories were cut according to the following sizecriteria: 430 cm butt diameter for timber species,414 cm diameter at stump level for poles and fuelwoodspecies and 48 cm diameter at stump level for others(Sawadogo et al., 2002). The prescribed early fire wasapplied at the end of the rainy season (October–November) each year beginning 1993 when the grasslayer humidity was approximately 40%. The grazingplots at both study sites were open for grazing bylivestock (a mixed herd of cattle, sheep and goats). Thelivestock carrying capacity in Laba forest was 1.0tropical livestock unit ha–1 (T.L.U. ha–1) and that ofTiogo was 1.4 T.L.U. ha–1 (Sawadogo, 1996) and thegrazing pressure at both sites was about half of thiscapacity (Sawadogo et al., 2005).

Data collection and analyses

The assessment was carried out every year from 1994to 2003 at the end of the rainy season (September–October) when most of the species are flowering andfruiting, which allows for easy species identification. Thepoint-intercept sampling procedure (Levy and Madden,1933) was used to gather species-cover data. Thepresence of species was recorded along a 20m perma-

nent line laid in each plot at an interval of 0.20m. Atevery 0.20m, a pin (6mm in diameter) taller than themaximum height of the vegetation was projected fromabove; and a species was considered as present if the pinhit any of its live parts. Identification of species andfamilies of plants follows Hutchinson et al. (1954).

Abundance, richness and diversity of herbaceousspecies were computed for each replicate in eachtreatment. Abundance was expressed as percentagecover and species richness as total number of species.Species diversity was assessed using Shannon–Weiner’sdiversity index (H0), calculated using the equation asfollows:

H 0 ¼ �X

pi ln pi,

where pi is the relative abundance of species i in a plot(Magurran, 2004). Abundance, species richness anddiversity were also computed by life form. We alsocomputed the abundance of some selected species(A. ascinodis and Diheteropogon hagerupii at Laba andA. gayanus and L. togoensis) that were encountered in alltreatment plots. Prior to statistical analyses, abundanceand species richness data were checked for normality.We used repeated-measures analysis of variance (Davis,2002) to determine whether abundance, species richnessand diversity were affected by year, grazing, fire,selective cutting and their interactions.

The analysis of variance was performed for eachexperimental site separately following the general linearmodel (GLM) for repeated measures:

Y ijk ¼ mþ bi þ lj þ ðblÞij þ �jðiÞ þ �jðkÞ,

where Yijk was the response variable for the herbaceousvegetation, m was the overall mean, bi was the effect ofthe between-subject factors, i (fire, grazing, selectivecutting and their interactions), lj was the effect of thewithin-subject factor, j, year, (bl)ij was the interaction ofthe between- and within-subject factors. The parametersej(i) and ej(k) are random error of the between- and thewithin-subjects factor, respectively with k number ofreplicates. When the homogeneity of variance assump-tion was violated, according to Mauchly’s test ofsphericity, the degrees of freedom for testing thesignificance of the within-subject factors were adjustedusing Huynh–Feldt correction factor, which is lessbiased than other correction factors (Davis, 2002). TheBonferonni adjustment for multiple comparisons wasemployed to control the inflation of a (Quinn andKeough, 2002). With four treatments in our study, fire,grazing, selective cutting and time, we have six possiblepair-wise comparisons, thus the Bonferonni adjusted p-value was 0.0083 (0.05/6). Results of the statisticalanalyses were considered significant if po0.0083 and toshow tendencies if 0.0083opp0.05.

To examine whether the inter-annual variability inspecies richness, abundance and diversity was related to

ARTICLE IN PRESS

104101 106

211108701

114 Block 1 116 Block 1

602102

012802702

214 Block 2 216 Block 2 218

DNEGEL503303203

310 311

Annual Fire + Cutting

Block 3 314 315 316 Block 3

No Fire + No Cutting

403

No fire + Cutting

114804

414Block 4Block 4

Livestock exclosure

FIRE BREAK

FIRE BREAK

FIRE BREAK

Annual Fire + No Cutting

415 416 417

404

Fig. 2. Experimental set-up at Tiogo and Laba.

P. Savadogo et al. / Perspectives in Plant Ecology, Evolution and Systematics 10 (2008) 179–195 183

amount and frequency of rainfall during the studyperiod, we performed linear regression analyses, wheretotal annual rainfall (mm/year) and number of rainydays were used as predictor variables, and vegetationattributes as responses. In a semi-arid area (as in thecase of our study sites), it is not only the amount ofrainfall but also the frequency of rainfall that influencesthe soil moisture status, which, in turn, is essential foremergence and establishment of seedlings (Elberse and

Breman, 1990). During fitting the regressions, thepredictor values were standardised by unit variance tonarrow down their initial variability. At both sites,amount and frequency of rainfall were correlated with atolerance value 40.1 (t ¼ 0.526 and 0.694, respectivelyat Laba and Tiogo) suggesting lack of collinearityproblem (Quinn and Keough, 2002), and temporalautocorrelation was checked and no such problem wasfound. We run linear regressions for each life form, each

ARTICLE IN PRESSP. Savadogo et al. / Perspectives in Plant Ecology, Evolution and Systematics 10 (2008) 179–195184

site, and each treatment separately (e.g. fire, grazing,selective cutting) and their interactions. The overallsignificance of each model and the significance ofeach predictor variable were assessed. Non-signifi-cant predictor variables were eliminated until allremaining predictor variables were significant. Themaximum likelihood method was used to selectthe models that would best approximate the vegeta-tion attributes based on Akaike’s InformationCriterion (Burnham and Anderson, 2002). All statisticalanalyses were done with SPSS 15 software package(Copyright SPSS for Windows, Release 2006 Chicago:SPSS Inc.).

Results

Main effects of treatments

The total number of herbaceous species recordedduring the study period was 176 and 152 at Laba andTiogo forests, respectively (data not shown). These taxawere made up of 43 families and 112 genera at Laba and47 families and 117 genera at Tiogo site. Prior toapplying the treatments in 1993, the pool of herbaceousspecies at plot level was 8, 22, and 38 for perennialgrasses, annual grasses and forbs at Laba, respectivelyand the corresponding values at Tiogo were 5, 23 and43. The main effects of grazing on total species richness,abundance and diversity of herbaceous flora were notsignificant at either of the experimental sites, except totalabundance of the herbaceous flora, which tended to behigher on ungrazed than grazed plots at Tiogo (Table 1,between-subject factors). Among life forms, perennialgrasses tended to be more abundant on ungrazed thangrazed plots at Laba (F1,24 ¼ 4.206, p ¼ 0.051) whilethey tended to be more diverse on grazed than ungrazedplots at Tiogo (F1,24 ¼ 4.909, p ¼ 0.036) (Fig. 3). Themain effects of fire on total species richness, abundanceand diversity of herbaceous vegetation were notsignificant at either Laba or Tiogo (Table 1); howeverfire tended to influence the richness of annual grassesat Tiogo (F1,24 ¼ 4.336, p ¼ 0.048), and abundance(F1,24 ¼ 4.787, p ¼ 0.039) and diversity (F1,24 ¼ 4.218,p ¼ 0.051) of perennial grasses at Laba. The numberof annual grasses tended to be higher on burnt thanunburnt plots at Tiogo while abundance and diversityof perennial grasses tended to be higher on unburntthan burnt plots at Laba (Fig. 4). Selective treecutting affected neither the total species richnessnor abundance and diversity of herbaceous flora ateither of the experimental site (Table 1). Further,selective cutting had any effect neither on speciesrichness, abundance nor diversity of different life forms(Fig. 5).

Combined effects of treatments

None of the treatment combinations had significanteffects at both experimental sites, except fire and cuttingtreatment that tended to influence the total speciesrichness and abundance of herbaceous flora at Laba(Table 1). At all levels of grazing, the richness ofherbaceous flora at Laba was slightly lower on plotssubjected to fire and selective cutting treatment thanplots subjected to either fire or cutting treatments whilethe reverse held true for total abundance (Table 2). Atlife form level, the combined treatment of fire andgrazing tended to reduce the richness of forbs at Tiogo(F1,24 ¼ 5.427, p ¼ 0.029) while it tended to increasetheir diversity (F1,24 ¼ 5.363, p ¼ 0.029) than eithergrazing or fire (Table 2). The diversity of forbs tendedto be higher (F1,24 ¼ 4.941, p ¼ 0.036) on plots subjectedto fire combined with grazing and selective cuttingtreatments than on plots subjected to neither of thesetreatments (Table 2).

Inter-annual variation

Total species richness, abundance and diversity ofherbaceous flora exhibited a highly significant inter-annual variation at both experimental sites (Table 1,within-subject factor). Some of the treatments interactedsignificantly with the within-subject factor, year. Therewere significant interaction effects of year, fire andselective cutting treatment on total species diversity atLaba, and year and grazing on total species richness atTiogo (Table 1). The diversity of herbaceous flora atLaba, increased during the first and last 3 years of thestudy period, while at Tiogo, a clear increasing trend inspecies richness was observed from 2000 to 2003compared to other years (Fig. 6). On plots subjectedto fire and grazing treatment at Laba, the abundance ofherbaceous flora tended to be more in some years whilethe inter-annual variation in abundance was remarkableon grazed plots at Tiogo.

The richness, abundance and diversity of different lifeforms also varied significantly between years, exceptrichness of perennial grasses at Laba and abundance offorbs at both Laba and Tiogo (Fig. 6). Significant yearby treatment interaction effects were also observed atlife form level. On plots subjected to grazing at Tiogo,annual grass species richness varied considerably be-tween years (F9,216 ¼ 2.930, p ¼ 0.003). At the same site,the effects of fire on richness (F9,216 ¼ 2.607, p ¼ 0.007)and diversity (F6,150 ¼ 3.089, p ¼ 0.006) of annualgrasses also varied substantially between years. Theinfluence of other treatments and their combinationsalso tended to vary between years. At Laba, theinteraction effect of year and fire combined withselective cutting treatment on richness of annual grasses,

ARTICLE IN PRESS

Table 1. Summary of repeated measures ANOVA for testing the significance of the between- and within-subject effects on total

species richness, abundance and diversity of herbaceous vegetation in the Sudanian savanna woodlands at Laba and Tiogo

Source Richness Abundance Diversity

d.f. F p d.f. F p d.f. F p

(A) Laba

Between-subject factors

Fire 1 0.726 0.403 1 0.617 0.440 1 0.412 0.527

Grazing 1 2.750 0.110 1 3.590 0.070 1 2.097 0.097

Cutting 1 0.213 0.648 1 0.016 0.899 1 0.122 0.729

Fire� grazing 1 1.447 0.241 1 0.580 0.454 1 0.112 0.740

Fire� cutting 1 4.357 0.048 1 5.238 0.031 1 1.973 0.173

Grazing� cutting 1 0.251 0.621 1 0.017 0.896 1 0.024 0.879

Fire� grazing� cutting 1 0.147 0.705 1 1.707 0.204 1 0.259 0.615

Error 24 24 24

Within-subject factors

Year 9 6.227 0.000 7 5.375 0.000 9 5.963 0.000Year� fire 9 0.631 0.770 7 0.457 0.853 9 1.079 0.379

Year� grazing 9 0.595 0.800 7 0.704 0.659 9 0.529 0.852

Year� cutting 9 0.587 0.807 7 0.601 0.743 9 1.729 0.084

Year� fire� grazing 9 1.062 0.392 7 2.155 0.045 9 0.965 0.470

Year� fire� cutting 9 1.887 0.055 7 1.985 0.065 9 2.971 0.002Year� cutting� grazing 9 0.974 0.463 7 1.154 0.333 9 0.751 0.662

Year� fire� cutting� grazing 9 0.692 0.716 7 0.855 0.537 9 0.671 0.735

Error 216 157 216

(B) Tiogo

Between-subject factors

Fire 1 0.937 0.343 1 2.093 0.161 1 2.275 0.144

Grazing 1 0.916 0.348 1 4.474 0.045 1 0.731 0.401

Cutting 1 0.281 0.601 1 0.814 0.376 1 0.008 0.927

Fire� grazing 1 2.261 0.146 1 3.180 0.087 1 2.227 0.149

Fire� cutting 1 2.294 0.143 1 2.892 0.102 1 1.291 0.267

Grazing� cutting 1 0.611 0.442 1 0.508 0.483 1 0.787 0.384

Fire� grazing� cutting 1 3.854 0.061 1 3.766 0.064 1 1.636 0.213

Error 24 24 24

Within-subject factors

Year 9 16.473 0.000 5 9.323 0.000 7 13.135 0.000Year� fire 9 1.407 0.186 5 1.472 0.199 7 1.656 0.118

Year� grazing 9 3.585 0.000 5 2.694 0.021 7 1.782 0.089

Year� cutting 9 0.897 0.529 5 0.744 0.603 7 1.091 0.371

Year� fire� grazing 9 0.962 0.473 5 0.671 0.658 7 0.717 0.665

Year� fire� cutting 9 0.918 0.510 5 0.797 0.563 7 0.487 0.852

Year� cutting� grazing 9 0.639 0.763 5 0.909 0.483 7 0.683 0.694

Year� fire� cutting� grazing 9 1.141 0.335 5 1.483 0.195 7 0.699 0.681

Error 216 130 178

Note that the degrees of freedom for the within-subject factor for testing abundance were Huynh–Feldt adjusted.

P. Savadogo et al. / Perspectives in Plant Ecology, Evolution and Systematics 10 (2008) 179–195 185

year and fire and grazing treatment on their abundance,and year and grazing and year and fire combined withgrazing and selective cutting on diversity of perennialgrasses tended to vary between years. These inter-annualvariations in vegetation attributes were partly explainedby the amount and/or frequency of rainfall, whichshowed significant relationships with some vegetation

attributes, depending on the site, life form andtreatments (Table 3).

Dominant species

Four species, A. ascinodis and D. hagerupii at Labaand A. gayanus and L. togoensis at Tiogo were the most

ARTICLE IN PRESS

AN0

2

4

6

8

10

12

14

16

18

20GrazingNo Grazing

GrazingNo Grazing

Ric

hnes

s

Laba

0

2

4

6

8

10

12

14

16

18

20

Ric

hnes

s

Tiogo

0

4

8

12

16

20

24

28

32

36

LabaGrazingNo grazing

GrazingNo grazing

Abu

ndan

ce

0

4

8

12

16

20

24

28

32

36

Tiogo

Abu

ndan

ce

0

2

4

LabaGrazingNo grazing

GrazingNo grazing

Div

ersi

ty

0

2

4

Tiogo

Div

ersi

ty

PE FB Total AN PE FB Total

AN PE FB Total AN PE FB Total

AN PE FB Total AN PE FB Total

Fig. 3. The main effects of grazing on species richness, abundance and diversity of herbaceous vegetation in the Sudanian savanna

woodlands at Laba and Tiogo (AN: annuals; PE: perennials; FB: forbs).

P. Savadogo et al. / Perspectives in Plant Ecology, Evolution and Systematics 10 (2008) 179–195186

common species on nearly all experimental plots, thusmade analysis at individual species level possible.Among these species, the abundance of L. togoensis

tended to be influenced by fire (12.5073.08% on burntversus 5.8572.05% on unburnt plots; p ¼ 0.039), bygrazing (5.2173.08% on grazed versus 13.1573.21%on non-grazed plots, p ¼ 0.043) and by grazing com-

bined with selective cutting (p ¼ 0.041) treatments. Theannual variation in abundance was significant(po0.0083) for all species except A. ascinodis. ForL. togoensis, the annual variation in abundance wasmore pronounced on plots subjected to grazing(p ¼ 0.006) and grazing combined with selective cutting(p ¼ 0.006).

ARTICLE IN PRESS

AN0

2

4

6

8

10

12

14

16

18

20

Ric

hnes

s

0

2

4

6

8

10

12

14

16

18

20

Ric

hnes

s

0

4

8

12

16

20

24

28

32

36

Abu

ndan

ce

0

4

8

12

16

20

24

28

32

36

Abu

ndan

ce

0

2

4

Div

ersi

ty

0

2

4

Div

ersi

ty

PE FB Total AN PE FB Total

AN PE FB Total AN PE FB Total

AN PE FB Total AN PE FB Total

FireNo Fire

FireNo FireLaba

Tiogo

LabaFireNo Fire

FireNo Fire Tiogo

LabaFireNo Fire

FireNo Fire

Tiogo

Fig. 4. The main effects of fire on species richness, abundance and diversity of herbaceous vegetation in the Sudanian savanna

woodlands at Laba and Tiogo (AN: annuals; PE: perennials; FB: forbs).

P. Savadogo et al. / Perspectives in Plant Ecology, Evolution and Systematics 10 (2008) 179–195 187

Discussion

Temporal variation

A conspicuous inter-annual variation in herbaceousspecies richness, abundance and diversity was foundduring the study period. These could be partly explainedby the difference in precipitation patterns between and

within years. The change in diversity and herbaceous coveris, in many places, found to be related to weatherconditions and in particular spatial and temporal distribu-tion of rainfall (Frost et al., 1986; Seghieri et al., 1994).Other factors that might account for inter-annual variationin species richness, abundance and diversity of herbaceousflora are variation in grazing intensity, stocking rate andfeeding preferences, fire intensity, and competition.

ARTICLE IN PRESS

AN0

2

4

6

8

10

12

14

16

18

20

Ric

hnes

s

0

2

4

6

8

10

12

14

16

18

20

Ric

hnes

s

0

4

8

12

16

20

24

28

32

36

Abu

ndan

ce

0

4

8

12

16

20

24

28

32

36

Abu

ndan

ce

0

2

4

Div

ersi

ty

0

2

4

Div

ersi

ty

PE FB Total AN PE FB Total

AN PE FB Total AN PE FB Total

AN PE FB Total AN PE FB Total

CuttingNo Cutting

CuttingNo CuttingLaba

Tiogo

LabaCuttingNo Cutting

CuttingNo Cutting Tiogo

LabaCuttingNo Cutting

CuttingNo Cutting

Tiogo

Fig. 5. The main effects of selective cutting tree cutting on species richness, abundance and diversity of herbaceous vegetation in the

Sudanian savanna woodlands at Laba and Tiogo (AN: annuals; PE: perennials; FB: forbs).

P. Savadogo et al. / Perspectives in Plant Ecology, Evolution and Systematics 10 (2008) 179–195188

Effects of grazing

Results from the present study show that grazing hadno effect on total species richness and diversity ofherbaceous vegetation at either of the experimental sites.There are three possible reasons why grazing had nonegative impact on total species richness and diversity of

herbaceous species. First, the grazing intensity on ourplots was half the carrying capacity, thus many speciescould survive intermediate levels of grazing, whichallows succession to proceed but limit the ability offew highly competitive species to dominate the commu-nity (Olff and Ritchie, 1998). This is further proven bythe dominance of A. gayanus (47.1% cover) on ungrazed

ARTIC

LEIN

PRES

S

Table 2. The interaction effects of grazing, fire and selective cutting on species richness, abundance and diversity of herbaceous vegetation in the Sudanian savanna woodlands

Life form Variables Laba Tiogo

Grazed Non-grazed Grazed Non-grazed

Burnt Unburnt Burnt Unburnt Burnt Unburnt Burnt Unburnt

Annuals Richness Cutting 7.071.0 8.071.0 7.071.0 7.071.0 7.071.0 7.071.0 7.071.0 7.071.0

No cutting 8.071.0 7.071.0 8.071.0 5.070.3 7.071.0 6.072.0 7.071.0 4.071.0

Abundance Cutting 9.5871.31 8.8771.87 8.1770.25 6.4270.25 6.4771.03 5.7070.77 11.3372.33 10.2470.56

No cutting 8.2071.09 6.2071.51 7.6771.74 9.1771.43 7.4971.69 10.3773.89 8.0871.43 10.9272.23

Diversity Cutting 1.7270.06 1.6870.25 1.4270.33 1.3370.21 1.3370.10 1.2270.13 1.5570.19 1.5570.31

No cutting 1.7370.09 1.4370.28 1.6770.25 1.1070.13 1.4470.16 1.3470.37 1.6670.28 0.9370.11

Perennials Richness Cutting 2.071.0 2.071.0 2.071.0 2.070.2 3.070.3 3.070.2 2.071.0 2.070.4

No cutting 2.070.4 2.070.4 2.071.0 2.070.4 3.070.2 2.070.4 2.071.0 2.070.2

Abundance Cutting 6.2871.90 8.3572.73 14.0377.11 19.2475.90 19.0372.85 17.8373.24 7.0672.70 15.9379.85

No cutting 7.4972.93 17.6075.73 11.0474.71 21.6872.06 16.2273.90 17.8074.50 10.8973.67 27.2076.45

Diversity Cutting 0.3970.18 0.4870.18 0.4870.19 0.7270.10 0.8470.10 0.8870.12 0.5970.19 0.3670.09

No cutting 0.3970.11 0.7470.20 0.5670.22 0.8770.13 0.9070.08 0.5670.15 0.5870.21 0.7570.14

Forbs Richness Cutting 6.070.3 8.071.0 6.071.0 7.071.0 4.071.0 5.071.0 4.071.0 6.071.0

No cutting 7.071.0 7.070.2 6.071.0 5.071.0 4.070.3 7,071.0 6.072.0 3.070.2

Abundance Cutting 3.1570.68 2.3970.53 4.7471.36 2.6270.44 2.6370.44 2.5170.46 3.9771.07 2.5770.44

No cutting 3.0270.75 2.2670.25 2.3470.36 2.6070.51 3.4171.01 2.9570.82 3.0570.68 2.9270.85

Diversity Cutting 0.8570.10 0.8770.20 0.8370.10 0.8070.17 0.5070.10 0.6470.17 0.5370.15 0.6870.15

No cutting 0.9370.20 0.7370.06 0.6870.12 0.5770.10 0.4470.05 0.9070.19 0.7370.17 0.3370.03

Total Richness Cutting 16.071.0 17.072.0 15.072.0 17.072.0 14.071.0 15.071.0 13.072.0 14.072.0

No cutting 17.071.0 16.071.0 17.071.0 12.071.0 14.071.0 15.072.0 16.072.0 9.071.0

Abundance Cutting 6.9170.61 6.3170.76 8.3471.54 7.0771.02 7.8370.41 7.5270.64 9.2771.32 8.8572.00

No cutting 6.2970.17 6.9270.39 6.2770.32 9.4671.23 8.1570.50 7.6970.99 7.4470.91 13.5671.83

Diversity Cutting 2.9670.12 3.0370.21 2.7470.25 2.8570.31 2.6770.11 2.7470.28 2.6670.30 2.5970.41

No cutting 3.0571.01 2.8970.12 2.9170.10 2.5570.11 2.7870.12 2.8170.21 2.9870.22 2.0070.24

P.Savadogoet

al./Persp

ectives

inPlantEcology,Evolutio

nandSystem

atics

10(2008)179–195

189

ARTICLE IN PRESS

0

2

4

6

8

10

12

14

16

18

20 Laba

Ric

hnes

s

Ric

hnes

s

0

2

4

6

8

10

12

14

16

18

20 Tiogo

Year Year

0

5

10

15

20

25

30

35

40 Laba

Abu

ndan

ce (%

)

0

5

10

15

20

25

30

35

40 Tiogo

Abu

ndan

ce (%

)

0.0

0.5

1.0

1.5

2.0

2.5

3.0

3.5 Laba

Total; Annual; Perennial; Forbs

Div

ersi

ty in

dex

0.0

0.5

1.0

1.5

2.0

2.5

3.0

3.5 Tiogo

Div

ersi

ty in

dex

Year Year1994 1996 1998 2000 2002 2004 1994 1996 1998 2000 2002 2004

Year Year1994 1996 1998 2000 2002 2004 1994 1996 1998 2000 2002 2004

1994 1996 1998 2000 2002 2004 1994 1996 1998 2000 2002 2004

Fig. 6. Inter-annual variation in herbaceous richness, abundance and diversity at Laba and Tiogo.

P. Savadogo et al. / Perspectives in Plant Ecology, Evolution and Systematics 10 (2008) 179–195190

plots and from the slightly lower abundance ofherbaceous species on grazed than ungrazed plots,particularly at Tiogo. Second, the experimental siteshave been subjected to various disturbances, such asbush fire and grazing by domestic and wild animals formany years prior to the establishment of the experiment,

thus the herbaceous species might be adapted toherbivory. Third, the effect of grazing on total speciesrichness interacted significantly with year, especially atTiogo, suggesting spatial and temporal variations interms of grazing intensity, stocking rate and feedingbehaviour, which could average out the overall effect of

ARTIC

LEIN

PRES

S

Table 3. Coefficient of determination (r2), together with F-statistic and p-values for significant linear regressions and lowest Delta AIC computed to examine the relationship

between herbaceous vegetation attributes and amount (X1) and frequency (X2) of rainfall at Laba and Tiogo (n ¼ 10)

Site Treatment Model Symbol Value b-Coefficient t p-Value

(predictor)

r2 (%) F p-Value

(model)

Delta

AIC

Tiogo Fire TD ¼ b0+b2X2 b0 4.941 11.8 0.000 74.4 23.3 0.001 0.00

b2 �0.246 �0.863 �4.8 0.001

AD ¼ b0+b2X2 b0 3.803 5.6 0.001 56.0 10.2 0.013 0.00

b2 �0.267 �0.748 �3.2 0.013

TA ¼ b0+b1X1+b2X2 b0 �9.801 �0.9 0.003 59.3 5.1 0.043 0.00

b1 �0.532 �0.210 �0.4 0.022

b2 4.423 0.780 2.9 0.022

FA ¼ b0+b2X2 b0 �3.765 �2.2 0.003 64.9 14.8 0.005 0.00

b2 0.081 0.805 3.8 0.005

Grazing FR ¼ b0+b1X1+b2X2 b1 13.193 6.0 0.001 64.6 6.4 0.026 0.00

b2 0.732 0.643 2.5 0.044

b0 �1.095 �0.913 �3.5 0.010

Cutting FD ¼ b0+b1X1+b2X2 b0 1.547 6.0 0.001 78.9 13.1 0.005 0.00

b1 0.141 0.820 4.1 0.005

b2 �0.174 �0.959 �4.8 0.002

Fire� grazing TD ¼ b0+b1X1+b2X2 b0 4.761 4.9 0.002 60.3 5.3 0.040 0.00

b1 0.345 0.732 2.6 0.033

b2 �0.411 �0.828 �3.0 0.020

Fire� grazing FR ¼ b0+b1X1+b2X2 b1 8.444 2.8 0.027 59.8 5.2 0.041 0.00

b2 1.183 0.810 2.9 0.023

b0 �1.157 �0.753 �2.7 0.031

Fire� cutting� grazing TD ¼ b0+b1X1+b2X2 b0 4.596 6.9 0.000 68.2 7.5 0.016 0.00

b1 0.236 0.657 2.8 0.033

b2 �0.354 �0.938 �3.9 0.007

Fire� cutting� grazing TR ¼ b0+b1X1+b2X2 b1 28.576 4.1 0.004 59.8 5.2 0.041 0.00

b2 2.354 0.707 2.5 0.039

b0 �2.944 �0.841 �3.0 0.020

AR ¼ b0+b1X1+b2X2 b1 15.596 3.9 0.006 62.6 5.9 0.032 0.00

b2 1.513 0.751 2.8 0.027

b0 �1.787 �0.843 �3.1 0.017

Laba Fire PD ¼ b0+b1X1+b2X2 b0 0.721 3.7 0.008 56.7 4.6 0.050 0.00

b1 0.091 0.648 1.8 0.021

b2 �0.147 �1.085 �3.0 0.021

Response variables were: TD, total diversity; AD, annual grass diversity; TA, total abundance; FA, forb abundance; FR, forb richness; FD, forb diversity; TR, total richness; AR, annual grass

richness; PD, perennial grass diversity.

P.Savadogoet

al./Persp

ectives

inPlantEcology,Evolutio

nandSystem

atics

10(2008)179–195

191

ARTICLE IN PRESSP. Savadogo et al. / Perspectives in Plant Ecology, Evolution and Systematics 10 (2008) 179–195192

grazing. Temporal variation in grazing intensity, stock-ing rate and feeding behaviour is common in the Sahel(Hiernaux, 1998) where free grazing by mixed herds –cattle, sheep and goats, is practiced (which is also thecase on our sites). Grazing has, however, tended toreduce the abundance of herbaceous flora, particularlythe abundance of perennial grasses at Laba. Thiscould be related to trampling effect, which in turn isrelated to the species ability to resist trampling-inducedchanges, their tolerance to a cycle of disturbanceand their resilience following cessation of trampling(Cole, 1995).

The effects of grazing varied significantly betweenyears, particularly at Tiogo. Total species richness washigher on grazed plots in some years. Although soilfertility is generally low at both experimental sites,the soils at Tiogo are mainly deep (475 cm) silty-claywhile they are shallow (o45 cm) silty-sand at Laba.Clay soils are more productive than sandy soils duringyears of above average rainfall (e.g. 1998, Fig. 1)and maintain greater quantities of available soil waterover a longer period (Seghieri et al., 1994), therebycontributing to enhanced local colonisation rates.Evidence shows that the effects of herbivores is some-what predictable and varies across environmentalgradients, notably soil fertility and precipitation gradi-ents (Belsky, 1992; Fuhlendorf et al., 2001; Harrisonet al., 2003). Grazing is expected to have a strongpositive effect on diversity in systems with higherproductivity (Milchunas and Lauenroth, 1993; Proulxand Mazumder, 1998; Osem et al., 2002; Bakker et al.,2006), which explains the fairly good correlationbetween rainfall parameters and vegetation attributesat Tiogo compared to Laba.

We also found some species, namely B. stachydea,Zornia glochidiata and Pandiaka heudelotii, on grazedplots during 1998 assessment that were not recordedearlier on the same plots. Herbivores play an importantrole for seed dispersal, germination, soil seed bankcomposition and amelioration of soil conditions byenhancing water infiltration and localised fertilisationeffects of dung and urine (O’Connor and Pickett, 1992;Herault and Hiernaux, 2004); thereby, enabling speciesto colonise grazed sites. The new species that appearedon grazed plots in our study sites have been previouslydescribed as indicators of grazing disturbance in theSudanian savanna (Cesar, 1992).

Effects of prescribed fire

The main effects of early fire on total species richness,abundance and diversity were not significant at either ofthe study sites, but fire tended to have a positive effecton species richness of annual at Tiogo, which isconsistent with a previous study in the Serengeti

grasslands (Belsky, 1992). Inhibition of emergence andestablishment of herbaceous species in savannas isrelated to high fire intensity (Jensen et al., 2001) thatresults in increased seed mortality, and eventuallydecreases species richness (Jensen and Friis, 2001).However, early fire is often of low intensity and severitydue to high moisture in the fuels and thus has a minorimpact on seed viability while creating conditionsfavourable for germination and favouring tillering byremoval of litter cover (Whelan, 1995; Garnier andDajoz, 2001). Germination of several herbaceous speciesis also stimulated by smoke released during burning(Adkins and Peters, 2001).

The effect of fire on species richness and diversity ofannual grasses varied significantly between years,especially at Tiogo, i.e., increased these vegetationattributes in some years more than other years. Thiscould be related to inter-annual variation in fireintensity and severity, which in turn depends on the lifeform, quantity of fuels, weather conditions and moisturecontent of fuels (Coughenour, 1991; Cheney et al.,1993), and the ability of some grasses to fully recoverafter fire, which may require 2–4 years or more (Bockand Bock, 1992). The relatively large number of speciesfound on burnt plots in our study is consistent withprevious studies on arid plant communities (Valone andKelt, 1999; Valone, 2003). At Laba, fire has tended toreduce the abundance of perennial grasses (Fig. 4).Perhaps, the intensity of fire could be sufficiently high atplot level to hamper resprouting of some perennial grassfrom buried vegetative structures. Uniform burn isdifficult to achieve in savanna-woodlands due to theirmosaic architecture with bush clumps and open areas, sofire generally skirts around the edges of bush clumps,leaving the centre unburnt. Also, germination of somespecies is inhibited by unfavourable pH and osmoticconditions on post fire seedbeds and due to reducedmoisture availability on drought-prone postfire gaps(Elberse and Breman, 1990).

Effects of selective tree cutting

Selective removal of trees is thought to enhance thecolonisation process by reducing competition for waterand nutrients, opening up more growing space and byincreasing the availability of light at the ground level(Frost et al., 1986; Guo, 1998). Our results show that theoverall main effect of selective cutting of trees was notsignificant. It should be noted that the selective cuttingtreatment was applied once by extracting 50% of thebasal area of all trees; therefore, the competition forlight, water and nutrients might be determined by initialdensity of trees. For instance, if tree density was highbefore cutting, selective cutting of trees might reduce thecompetition for resources and thus have a clear effect.

ARTICLE IN PRESSP. Savadogo et al. / Perspectives in Plant Ecology, Evolution and Systematics 10 (2008) 179–195 193

However, the distribution of woody component of mostsavanna-woodlands is scattered and further opening ofthe canopy can create unfavourable thermal condition inthe understory that increases soil evaporation and plantevapotranspiration, thereby counterbalancing the initialpositive effect of increased water and nutrient avail-ability due to selective removal of trees. There is alsoevidence from the same sites (Sawadogo et al., 2002;Nyg(ard et al., 2004) that shows vigorous resproutingand suckering ability of most savanna-woodland speciesfollowing cutting disturbance. Since the sprouts andsuckers depend on the already established root system ofthe original trees for accessing water and soil nutrients,they can easily out-compete the herbaceous vegetation.

Combined effects of treatments

There is insufficient evidence to suggest combinedeffects of annual early fire, grazing and selective treecutting on richness, abundance and diversity of herbac-eous flora in the Sudanian savanna-woodland. Some ofthe combined effects found were life form and sitespecific. Furthermore, similar additive effects involvinggrazing, fire and selective cutting were not detected forspecies individually regardless of the site. Species such asA. gayanus or A. pseudapricus responded differently tosimilar manipulation at different sites, conforming withthe concept that grassland species cannot be divided intofunctional groups in which all species respond in asimilar manner to all major environmental variables(Belsky, 1992). As a whole, the combined effects of fire,selective cutting and grazing tended to vary across yearsfor diversity of forbs at Tiogo while the combined effectsof grazing and fire slightly enhanced the richness anddiversity of forbs at Tiogo, and that of cutting and fireslightly reduced the total species richness and abundanceat Laba. These results contrast with earlier reports inmesic grasslands where grazing combined with burningincreases plant species richness and diversity to unu-sually higher level (Collins and Barber, 1985; Collins,1987). In mesic grasslands, burning opens space forcolonists and grazing limits the ability of competitivelydominant C4 grass species to outcompete other plants,resulting in opportunities for colonisation by C3 grassesand forbs (Collins, 1987; Collins et al., 1998). In semi-arid savannas, plant communities tend to have lessground cover and presumably less competition for space(Valone and Kelt, 1999). It is, therefore, the resourceavailability that limits local diversity (Collins et al.,1998). Such limited additive effects of grazing and firehave been observed in other semi-arid plant commu-nities (Belsky, 1992; Valone and Kelt, 1999; Drewa andHavstad, 2001; Valone, 2003). Furthermore, the slightcombined effects of fire and grazing could have resultedfrom different sets of species responding positively or

negatively to each disturbance or from the fact that firesimply duplicated the effect of grazing by reducingabove ground biomass. The removal of trees createsevaporative stress and limits the growth of annualgrasses, as can be seen from the relatively lowerabundance values during the first 3 years of post-treatment at Laba. Selective cutting possibly favouredthe dominance of drought-tolerant perennial grasses,which in turn serve as good fuel and result in high-intensity fire that eventually had tendency to reduce theoverall abundance of herbaceous species. It should benoted that treatment combinations also tended to showsignificant interactions with year, suggesting temporalvariability in grazing and fire intensities, as well ascompetition with profoundly regenerating individualsfollowing cutting disturbance, which could average outthe overall interaction effects.

Conclusions

The study provides evidence that both disturbances andclimatic condition influence the species composition,diversity and abundance of herbaceous flora in theSudanian savanna-woodlands. The responses of herbaceousflora to treatments, particularly fire and grazing was site-specific, suggesting the importance of landscape-scaleapproaches to understanding the impact of disturbanceson composition, structure and diversity of savanna-wood-land ecosystems. From a management perspective, grazing,which is currently prohibited in State forests, could beallowed in savanna-woodland reserves provided that it iskept at a moderate level, as livestock husbandry is the mainsources of livelihood and source of revenue for the localpeople in Burkina Faso. The application of early fire couldcontinue to be used as a management tool with due cautionto the timing of burning, weather conditions and otherpossible factors that may increase fire intensity. The currentlevel of selective cutting (50% of the basal area of all trees)seems to have less beneficial effect on herbaceous vegeta-tion, thus further research is required to determine theappropriate cutting intensity that balances the compositionand diversity of the herbaceous and woody components ofthe Sudanian savanna ecosystem.

Acknowledgements

Funding for this study was provided by SwedishInternational Development Cooperation Agency (Sida).We thank the fieldworkers for their invaluable assistancein carrying out the inventory over the entire studyperiod, particularly Bama Theophile, Meda Modeste,Bako Lambin and Yaro Badema.

ARTICLE IN PRESSP. Savadogo et al. / Perspectives in Plant Ecology, Evolution and Systematics 10 (2008) 179–195194

References

Adkins, S.W., Peters, N.C.B., 2001. Smoke derived from burnt

vegetation stimulates germination of arable weeds. Seed

Sci. Res. 11, 213–222.

Akpo, L.E., Grouzis, M., Bada, F., Pontanier, R., Floret, C.,

1999. Effect of fallow tree cover on the species structure of

herbaceous vegetation in a Sudanian savanna. Secheresse

10, 253–261.

Bakker, E.S., Ritchie, M.E., Olff, H., Milchunas, D.G.,

Knops, J.M.H., 2006. Herbivore impact on grassland plant

diversity depends on habitat productivity and herbivore

size. Ecol. Lett. 9, 780–788.

Bellefontaine, R., Gaston, A., Petrucci, Y., 2000. Management

of Natural Forests of Dry Tropical Zones. FAO Conserva-

tion Guide, vol. 32. Food and Agriculture Organization of

the United Nations, Rome.

Belsky, J.A., 1992. Effects of grazing, competition, disturbance

and fire on species composition and diversity in grassland

communities. J. Veg. Sci. 3, 187–200.

Bock, J.H., Bock, C.E., 1992. Short-term reductions in plant

densities following prescribed fire in an ungrazed semidesert

shrub-grassland. Southw. Natural. 37, 49–53.

Breman, H., Kessler, J.J., 1995. The role of woody plants in

agro-ecosystems of semi-arid regions: with an emphasis of

the Sahelian countries. Advanced Series in Agricultural

Sciences, vol. 23. Springer-Verlag, Berlin.

Burnham, K.P., Anderson, D.R., 2002. Model Selection and

Multimodel Inference: A Practical Information – Theoretic

Approach. Springer, New York.

Casado, M.A., Castro, I., Ramirez-Sanz, L., Costa-Tenorio,

M., de Miguel, J.M., Pineda, F.D., 2004. Herbaceous plant

richness and vegetation cover in Mediterranean grasslands

and shrublands. Plant Ecol. 170, 83–91.

Cesar, J., 1992. La production biologique des savanes de Cote

d’Ivoire et son utilisation par l’homme. Biomasse, valeur

pastorale et valeur fourragere. These de doctorat, Uni-

versite Paris VI, France.

Cheney, N.P., Gould, J.S., Catchpole, W.R., 1993. The

Influence of fuel, weather and fire shape variables on fire-

spread in grasslands. Int. J. Wildland Fire 3, 31–44.

Cole, D.N., 1995. Experimental trampling of vegetation. 2.

Predictors of resistance and resilience. J. Appl. Ecol. 32,

215–224.

Collins, S.L., 1987. Interaction of disturbances in tallgrass

prairie: an experiment. Ecology 68, 1243–1250.

Collins, S.L., Barber, S.C., 1985. Effects of disturbance on

diversity in mixed-grass prairie. Vegetatio 64, 87–94.

Collins, S.L., Knapp, A.K., Briggs, J.M., Blair, J.M.,

Steinauer, E.M., 1998. Modulation of diversity by grazing

and mowing in native tallgrass prairie. Science 280,

745–747.

Coughenour, M.B., 1991. Spatial components of plant–herbi-

vore interactions in pastoral, ranching, and native ungulate

ecosystems. J. Range Manage. 44, 530–542.

Davis, C.S., 2002. Statistical Methods for the Analysis of

Repeated Measurements. Springer, London.

Drewa, P.B., Havstad, K.M., 2001. Effects of fire, grazing, and

the presence of shrubs on Chihuahuan desert grasslands.

J. Arid Environ. 48, 429–443.

Driessen, P., Deckers, J., Spaargaren, O., 2001. Lecture notes

on the major soils of the world. FAO World Soil Resources

Reports-94. Food and Agriculture Organization of the

United Nations, Rome.

Elberse, W.T., Breman, H., 1990. Germination and establish-

ment of Sahelian rangeland species. II. Effects of water

availability. Oecologia 85, 32–40.

Fontes, J., Guinko, S., 1995. Carte de la vegetation et de

l’occupation du sol du Burkina Faso. Ministere de la

Cooperation Francaise: projet campus (88 313 101).

Frost, P., Menaut, J.C., Walker, B., Medina, E., Solbrig, O.T.,

Swift, M., 1986. Responses of savannas to stress and

disturbance. A proposal for a collaborative programme of

research. In: IUBS-UNESCO-MAB, Biology International

Special issue 10.

Fuhlendorf, S.D., Briske, D.D., Smeins, F.E., 2001. Herbac-

eous vegetation change in variable rangeland environments:

the relative contribution of grazing and climatic variability.

Appl. Veg. Sci. 4, 177–188.

Garnier, L.K.M., Dajoz, I., 2001. The influence of fire on the

demography of a dominant grass species of West African

savannas, Hyparrhenia diplandra. J. Ecol. 89, 200–208.

Gibson, C.W.D., Brown, V.K., 1991. The effects of grazing on

local colonization and extinction during early succession.

J. Veg. Sci. 2, 291–300.

Gignoux, J., Clobert, J., Menaut, J.C., 1997. Alternative fire

resistance strategies in savanna trees. Oecologia 110,

576–583.

Guo, Q.F., 1998. Microhabitat differentiation in Chihuahuan

Desert plant communities. Plant Ecol. 139, 71–80.

Hagberg, S., Gomgnimbou, M., Some, D.B., 1996. Forets

classees et terres des ancetres au Burkina Faso. Working

papers in cultural anthropology no 3. Department of

cultural anthropology, Uppsala University.

Harrison, S., Inouye, B.D., Safford, H.D., 2003. Ecological

heterogeneity in the effects of grazing and fire on grassland

diversity. Conserv. Biol. 17, 837–845.

Herault, B., Hiernaux, P., 2004. Soil seed bank and vegetation

dynamics in Sahelian fallows; the impact of past cropping

and current grazing treatments. J. Trop. Ecol. 20, 683–691.

Hiernaux, P., 1998. Effects of grazing on plant species

composition and spatial distribution in rangelands of the

Sahel. Plant Ecol. 138, 191–202.

Hoffmann, W.A., 1999. Fire and population dynamics of

woody plants in a neotropical savanna: matrix model

projections. Ecology 80, 1354–1369.

Hutchinson, J., Dalziel, J.M., Hepper, F.N., Keay, R.W.J.,

1954. Flora of west tropical Africa: all territories in West

Africa south of latitude 181N and to the west of Lake Chad,

and Fernando Po. Crown Agents for Oversea Governments

and Administrations, London.

Jensen, M., Friis, I., 2001. Fire regimes, floristics, diversity, life

forms and biomass in wooded grassland, woodland and dry

forest at Gambella, Western Ethiopia. Biologiske skrifter/

Det Kongelige Danske Videnskabernes Selskab 54,

349–387.

Jensen, M., Michelsen, A., Gashaw, M., 2001. Responses in

plant, soil inorganic and microbial nutrient pools to

experimental fire, ash and biomass addition in a woodland

savanna. Oecologia 128, 85–93.

ARTICLE IN PRESSP. Savadogo et al. / Perspectives in Plant Ecology, Evolution and Systematics 10 (2008) 179–195 195

Ky-Dembele, C., Tigabu, M., Bayala, J., Ouedraogo, S.J.,

Oden, P.C., 2007. The relative importance of different

regeneration mechanisms in a selectively cut savanna-

woodland in Burkina Faso, West Africa. For. Ecol.

Manage. 243, 28–38.

Levy, E.B., Madden, E.A., 1933. The point method of pasture

analysis. N.Z. J. Agric. 46, 267–279.

Liedloff, A.C., Coughenour, M.B., Ludwig, J.A., Dyer, R.,

2001. Modelling the trade-off between fire and grazing in a

tropical savanna landscape, northern Australia. Environ.

Int. 27, 173–180.

Magurran, A.E., 2004. Measuring Biological Diversity. Black-

well Publishing, Malden, Oxford and Victoria.

McNaughton, S.J., 1983. Serengeti grassland ecology: the role

of composite environmental factors and contingency in

community organization. Ecol. Monogr. 53, 291–320.

Menaut, J.C., Abbadie, L., Lavenu, F., Loudjani, P., Podaire,

A., 1991. Biomass burning in West Africa Savannahs. In:

Levine, S.J. (Ed.), Global Biomass Burning. Atmospheric,

Climatic, and Biospheric Implications. The MIT Press,

Cambridge, Massachusetts, USA, pp. 133–142.

Menaut, J.C., Lepage, M., Abbadie, L., 1995. Savannas,

woodlands and dry forests in Africa. In: Bullock, S.H.,

Mooney, H.A., Medina, E.E. (Eds.), Seasonally Dry

Tropical Forests. Cambridge University Press, London,

pp. 64–92.

Milchunas, D.G., Lauenroth, W.K., 1993. Quantitative effects

of grazing on vegetation and soils over a global range of

environments. Ecol. Monogr. 63, 327–366.

Nyg(ard, R., Sawadogo, L., Elfving, B., 2004. Wood-fuel yields

in short-rotation coppice growth in the north Sudan

savanna in Burkina Faso. For. Ecol. Manage. 189, 77–85.

O’Connor, T.G., Pickett, G.A., 1992. The Influence of grazing

on seed production and seed banks of some African

savanna grasslands. J. Appl. Ecol. 29, 247–260.

Olff, H., Ritchie, M.E., 1998. Effects of herbivores on

grassland plant diversity. Trends Ecol. Evol. 13, 261–265.

Osem, Y., Perevolotsky, A., Kigel, J., 2002. Grazing effect on

diversity of annual plant communities in a semi-arid

rangeland: interactions with small-scale spatial and tem-

poral variation in primary productivity. J. Ecol. 90,

936–946.

Pallo, F., 1998. Etude des feux sur la matiere organique des

sols des forets naturelles dans la region centre-ouest du

Burkina Faso. In Seminaire International sur l’Amenage-

ment Integre des forets Naturelles des Zones Tropicales

Seches en Afrique de l’Ouest. 16–20 Novembre 1998

Ouagadougou, Burkina Faso: CNRST, SLU Uppsala,

pp. 187–198.

Proulx, M., Mazumder, A., 1998. Reversal of grazing impact

on plant species richness in nutrient-poor vs nutrient-rich

ecosystems. Ecology 79, 2581–2592.

Quinn, G.P., Keough, M.J., 2002. Experimental design

and analysis for biologists. Cambridge University Press,

Cambridge.

Sawadogo, L., 1996. Evaluation des potentialites pastorales

d’une foret classee soudanienne du Burkina Faso. (Cas de

la foret classee de Tiogo). These Doctorat 3eme Cycle.

Thesis, Universite de Ouagadougou.

Sawadogo, L., Nyg(ard, R., Pallo, F., 2002. Effects of livestock

and prescribed fire on coppice growth after selective cutting

of Sudanian savannah in Burkina Faso. Ann. For. Sci. 59,

185–195.

Sawadogo, L., Tiveau, D., Nyg(ard, R., 2005. Influence of

selective tree cutting, livestock and prescribed fire

on herbaceous biomass in the savannah woodlands of

Burkina Faso, West Africa. Agric. Ecosyst. Environ. 105,

335–345.

Seghieri, J., Floret, C., Pontanier, R., 1994. Development of an

herbaceous cover in a Sudano-Sahelian savanna in North

Cameroon in relation to available soil–water. Vegetation

114, 175–184.

Valone, T.J., 2003. Examination of interaction effects of

multiple disturbances on an arid plant community. Southw.

Natural. 48, 481–490.

Valone, T.J., Kelt, D.A., 1999. Fire and grazing in a shrub-

invaded arid grassland community: independent or inter-

active ecological effects? J. Arid Environ. 42, 15–28.

Valone, T.J., Nordell, S.E., Ernest, S.K.M., 2002. Effects of

fire and grazing on an arid grassland ecosystem. Southw.

Natural. 47, 557–565.

van Langevelde, F., van de Vijver, C.A.D.M., Kumar, L., van

de Koppel, J., de Ridder, N., van Andel, J., Skidmore,

A.K., Hearne, J.W., Stroosnijder, L., Bond, W.J., Prins,

H.H.T., Rietkerk, M., 2003. Effects of fire and herbivory

on the stability of savanna ecosystems. Ecology 84,

337–350.

Whelan, R.J., 1995. The Ecology of Fire. Cambridge

University Press, New York.

Williams, R.J., Cook, G.D., Gill, A.M., Moore, P.H.R., 1999.

Fire regime, fire intensity and tree survival in a tropical

savanna in northern Australia. Aust. J. Ecol. 24, 50–59.