Edible biomass production from some important browse species in the Sahelian zone of West Africa

ARTICLE IN PRESS

Journal of Arid Environments 71 (2007) 376–392

Journal of Arid Environments www.elsevier.com/locate/jaridenv

Edible biomass production from some important browse species in the Sahelian zone of West Africa H.O. Sanona, C. Kabore´-Zoungranab, I. Ledinc, a

Institut de l’Environnement et de Recherche Agricole, CRREA de l’Ouest, Station de Farako-ba`, Bobo-Dioulasso, Burkina Faso b Institut du De´veloppement Rural, Universite´ Polytechnique de Bobo-Dioulasso, Burkina Faso c Department of Animal Nutrition and Management, Swedish University of Agricultural Sciences, P.O. Box 7024, SE-750 07 Uppsala, Sweden Received 25 October 2006; received in revised form 28 February 2007; accepted 23 March 2007 Available online 18 May 2007

Abstract Acacia senegal, Guiera senegalensis and Pterocarpus lucens, browse species important in the Sahelian zone of Burkina Faso were studied by the estimation of their phenological variation over time and the evaluation of edible biomass production, total and accessible directly to animals. Biomass production was also estimated using dendrometric parameters. All the three species started the foliation phase as soon as the rains started. A. senegal and P. lucens flowered before G. senegalensis and A. senegal lost leaves earlier. The fruiting phase lasted 6–7 months for all species. Accessible edible biomass varied according to the animal species, the plant species and the height of plants. G. senegalensis showed the highest proportion of accessible biomass, but P. lucens had higher total edible biomass. Goats browsing at higher height had more edible biomass at their disposal. The accessible edible biomass was weakly correlated with tree parameters, while crown diameter was the best parameter to predict total edible biomass production, with R2 varying from 90% (G. senegalensis) to 98% (P. lucens) in log10 transformation of dependent and independent variables. The single species models developed could be applied in similar agro-ecological zones, taking into account the height stratification of plants. Further investigations on others species are needed to be able to estimate total biomass available for browsing. r 2007 Elsevier Ltd. All rights reserved. Keywords: Acacia senegal; Guiera senegalensis; Phenology; Pterocarpus lucens

Corresponding author. Tel.: +46 18 671646; fax: +48 18 672995.

E-mail address: [email protected] (I. Ledin). 0140-1963/$ - see front matter r 2007 Elsevier Ltd. All rights reserved. doi:10.1016/j.jaridenv.2007.03.019

ARTICLE IN PRESS H.O. Sanon et al. / Journal of Arid Environments 71 (2007) 376–392

377

1. Introduction The rapid increase in human population and the occurrence of extensive droughts have resulted in overexploitation of land and land degradation in the dry lands of Sub-Saharan Africa (Breman and Kessler, 1995). The increase in livestock numbers due to eradication of the major cattle diseases has aggravated the pressure on rangelands, which constitute the main feed resources of ruminant animals (de Leeuw and Tothill, 1990). Pasture productivity based on herbaceous cover is decreasing with concomitant loss of tree cover. This has affected the productivity of livestock, which constitute an important part of the economy in the area. Browse species have always played a significant role in animal feeding in dry areas, and according to Baumer (1992) the use of browse is recorded as long ago as Roman times. le Houerou (2000) reported that the utilization of fodder trees and shrubs in the Mediterranean arid and semi-arid zones was initiated between World Wars I and II, then later expanded and diversified. A similar slow expansion has taken place in other arid zones in the USA, South Africa, South America and Australia. The increasing importance of trees and shrubs as animal feed is reported especially from harsh environments, hence most of the research has focused on arid and semi-arid rangelands (von Carlowitz, 1989). Although the herbaceous production can decrease due to competition with the woody cover (Scholes and Walker, 1993), this could at least partly be compensated for by increased browse production, depending on the species composition and degree of woody cover (de Leeuw and Tothill, 1990). Ligneous material in arid areas in Africa is estimated to constitute 40–50% of the total available feed (Speedy and Pugliese, 1992) and Baumer (1992) reported that up to 80% of the dietary protein was provided by plants of the Capparaceae family during the 3 driest months of the year. Behaviour studies have shown that animals spend most of their time browsing in the dry season (Ngwa et al., 2000; Sanon et al., 2007) and the leaves and fallen fruits/pods constitute the main component of the diet. Supplementation with browse is also known to improve productivity and growth rate of animals (Fall-Toure´ et al., 1997). The diversity of browse plants and the variability in their phenology extend the period of availability (Hiernaux et al., 1994), and the pods and fruits which fall in the dry season are readily consumed by animals (Sanon et al., 2005). The assessment of browse production can be direct by destructive techniques but indirect methods using regression equations based on more easily measured variables of the trees are frequently utilized for predicting biomass. These approaches are widely used in forestry, either for the estimation of above-ground biomass (wood, branches, leaves) (Ter-Mikaelian and Korzukhin, 1997; Salis et al., 2006), or to assess the available browse for wildlife or domestic animals (Paton et al., 2002; Saı¨ d et al., 2005). In the Sahelian area single regression models for a few browse species have been developed (Bille, 1980; Cisse, 1980). Most models are focused on total foliage production and do not give any information on the proportion available to animals. The accessibility has been expressed according to height reached by the animals by Breman and de Ridder (1991). The direct accessibility depends not only on the animal species but also on tree parameters such as shape of crown, height and system of branches of the tree, morphology of the leaves, the presence or not of spines and structure of branches, that can be open or closed (Ickowicz, 1995). While research efforts have focused on introduced species such as Leucaena, Gliricidia and Sesbania, of known nutritive value and productivity, the full potential of several local

ARTICLE IN PRESS 378

H.O. Sanon et al. / Journal of Arid Environments 71 (2007) 376–392

species has not been well exploited. The knowledge of browse production is important for the management and for a sustainable exploitation of rangelands. A study on feeding behaviour of domestic ruminants (Sanon et al., 2007), showed that in the Sahelian zone of Burkina Faso Acacia senegal (Mimosaceae) and Pterocarpus lucens (Papilionaceae) were browsed preferentially by goats and sheep, while Guiera senegalensis (Combretaceae) was often selected by cattle and sheep. These species were also well appreciated by farmers in the study area and were well represented in the pasture. This study aims to evaluate the biomass available for browsing from Acacia senegal (L.) Willd, Guiera senegalensis J.F. Gmel and Pterocarpus lucens Lepr. ex Guill. & Perrott. for ruminants (sheep, cattle and goats). Regression models were tested to predict the production and estimate the accessibility to ruminants in the area. 2. Materials and methods 2.1. Location and climate The study was undertaken in the Sahelian area of Burkina Faso, at the village of Tongomayel, located in the province of Soum (131440 –141500 N and 01320 –21070 W). The area is characterized by a dry climate with low rainfall (from June to September) and a long dry season of 8 months from October to May. The period of active vegetation lasts from the end of June to the second half of September. The dry season is characterized by the dry wind ‘‘Harmattan’’, which blows from the North-East to the South-west. The mean monthly temperature ranges from 21 1C in January to 40 1C in April. The mean rainfall of the last 10 years recorded at the weather station of Djibo is 480 mm. During the study in 2004 and 2005, the rainfall was 391 and 551 mm, respectively. The vegetation is of the steppe type, with shrubs and trees, and forest galleries are found along riversides and in some parts ligneous species may form more or less penetrable bush (e.g. tiger bush). The most common ligneous species found in the area are: Acacia nilotica, Acacia senegal, Balanites aegyptiaca, Boscia senegalensis, Commiphora africana, Dalbergia melanoxylon, Pterocarpus lucens and Grewia flavescens. The grass cover is sparse and dominated by annual grasses such as Aristida mutabilis, Cenchrus biflorus and Schoenefeldia gracilis. 2.2. Data collection A. senegal, family of Leguminosae (Mimosoideae), Guiera senegalensis, family of Combretaceae, and Pterocarpus lucens, family of Leguminosae (Papilionaceae), three species commonly browsed in the study area as reported by Sanon et al. (2007) were investigated. Each species was selected on the pasture type where it is abundant. Previously, the pastures in the study area were classified according to an interpretation of available aerial photographs in the area, which allowed distinguishing four types of pasture: shrubby steppe, sparse woody steppe, lowland pasture (valleys of temporary rivers and flooded basins) and tiger bush. An inventory of the woody flora was done in three random plots of 1 ha in each pasture type, by counting all individual trees and shrubs according to their height divided into five classes: [o1 m], [1–3 m], [3–5 m], [5–7 m] and [47 m]. A. senegal was selected in sparse woody steppe, G. senegalensis in lowland pasture and P. lucens in tiger bush.

ARTICLE IN PRESS H.O. Sanon et al. / Journal of Arid Environments 71 (2007) 376–392

379

The parameters measured were the phenological phases and the edible biomass production, total and at browsing height for sheep, goats and cattle. The two parameters were measured on different sets of plants. The plants were selected within or close to the 1 ha plots. 2.2.1. Phenology study Four individual plants were chosen per height class, giving 16 individuals of A. senegal, 16 of G. senegalensis and 20 of P. lucens, since no trees were found in the highest class of the first two species. The trees were studied from June 2003 to March 2004 and the phenology of all the individuals selected was determined by scoring of the development of leaf, flower and fruit every 2nd week. The development phase of the leaves was divided into four stages: leafless, the beginning (leaf buds and o50% of open buds), fully leafed or optimum foliation (50–100% of leaves) and end of foliation (leaves drying and falling). The three stages recorded in the flowering phase were the beginning (floral buds and o50% of open flowers), full flowering (50–100% of open flowers) and the end of flowering (dry flowers and shedding of floral elements). For fruiting the three stages were the beginning (o50% of fruits), full fruiting (50–100% mature fruits), and end of fruiting (fruits dried and falling). A stage was recorded as occurring when at least three of the four individuals in the height class considered were in this stage, and the results are presented in a phenogram. 2.2.2. Total and accessible edible biomass production The study of accessible edible biomass production was undertaken during 2 consecutive years, 2004 and 2005, with different plants of the same species. The accessible edible biomass production refers to leaves growing on the trees or shrubs between the soil level and the height reached by animals on pasture. The mean browsing height of sheep, goats and cattle was determined in a former study (Sanon et al., 2007) as being 0.87 m for sheep, 1.47 m for cattle and 1.65 m for goats. For each plant species, 16 individuals of A. senegal and G. senegalensis and 20 of P. lucens from different height classes (as described previously) were protected from browsing in the dry season by wire enclosure. The edible biomass production was estimated in September, corresponding to the optimum stage of biomass production in the area. The browse biomass was harvested manually (leaves, fruits and small branches less than 1 cm diameter) up to the mean height reached by the different animal species, starting from the lower height. The edible biomass out of reach of animals was also evaluated by cutting leafy branches and harvesting the browse components to determine total edible biomass. The edible biomass collected was weighed and air dried and samples were taken for dry matter (DM) determination. Before harvesting, the dendrometric parameters of each individual were measured to determine their correlations with the browse production. The trunk diameter was taken at the base (for G. senegalensis with multiple stems a sum of individual stem basal diameters was made). The diameter of the crown was measured by projecting the edges of the crown to the ground and measuring the length along one axis from edge to edge through the crown centre. Then two perpendicular directions were averaged and the total height of plant was measured with a clinometer (Suunto). The potential browse production in the pastures, with regards to the species studied was estimated by multiplying the edible biomass production per species by their relative proportion (number of plants/ha and according to height class) in the pasture.

ARTICLE IN PRESS 380

H.O. Sanon et al. / Journal of Arid Environments 71 (2007) 376–392

2.3. Statistical analysis The GLM procedure of SAS software (SAS, 1998) was used to analyse the data. The treatment means which showed significant differences at the probability level of Po0.05 were compared using Tukey-Kramers’s pairwise comparison procedures. The following statistical models were used: For the density of plants per pasture type: yij ¼ m+ai+eij, where yij is the dependent variable; m: general mean; ai: effect of pasture type i and eij: residual term. For the density per height class: ln(yijk) ¼ m+ai+mij+bk+(ab)ik+eijk, where m is the general mean, ai is the main effect of pasture type i, mij is the main-plot effect corresponding to plots of trees within pasture types, bk is the effect of height class k, (ab)ik is the interaction between pasture type and height class and eijk is the residual term. For the edible biomass production: ln(yijkn) ¼ m+ai+bj+(ab)ij+tk+(at)ik+(bt)jk+eijkn, where m is the general mean, ai the main effect of species i; bj is the effect of height class j; tk is the effect of period (year); (ab)ij, (at)ik and (bt)jk are the interaction effects and eijk the error term. The interaction of the three factors has been removed from the model as it was not significant in a preliminary analysis. A regression analysis was performed to test the relationship between the dendrometric parameters of trees and the biomass production. Linear regression, polynomial regressions and log10 transformation were tested and the significant regressions showing the highest regression coefficients were retained. 3. Results 3.1. Pasture vegetation and the contribution of the species studied The mapping of the study area distinguished four vegetation types, whose characteristics are presented in Table 1. The woody flora recorded consisted of in total 42 species, and lowland pasture had the highest number of species (35) and shrubby steppe the lowest (15). The density of plants was highest in tiger bush and lowest in shrubby steppe, 1602 and 222 plants/ha, respectively. Most of the woody species found were browsed (more than 80% of species recorded) in all pasture types. Table 2 shows the distribution of species studied per height class in the pastures. A total of 65 plants/ha of A. senegal was recorded in sparse woody steppe, and most of the plants (68%) were smaller than 1 m. In sparse woody steppe 78% of all plants were less than 1 m high. A. senegal is also represented in tiger bush, but is rare in lowland pasture. G. senegalensis is a species dominant in lowland pasture in the study area and the individuals of height class 2 represented the majority (77%), while individuals taller than 5 m (class 4) were seldom found. In tiger bush pasture, all the classes of height of P. lucens were found, and the individual plants o1 m constituted 73% of this species. Overall, A. senegal represented 13% of the woody plants in sparse woody steppe, G. senegalensis 45% in lowland and P. lucens 22% in tiger bush. 3.2. Phenology Fig. 1 shows the phenology of A. senegal, G. senegalensis and P. lucens.

ARTICLE IN PRESS H.O. Sanon et al. / Journal of Arid Environments 71 (2007) 376–392

381

Table 1 General characteristics of the pasture types in the study area Pasture types

No of species

Density plants/ha

Browsea species, %

Dominant species

Sparse woody steppe

27

517 (186)b

87.2 (2.7)

Acacia senegal Boscia senegalensis Balanites aegyptiaca Pterocarpus lucens

Lowland pasture

35

1051 (66)

83.0 (3.7)

Guiera senegalensis Combretum micranthum Balanites aegyptiaca Combretum aculeatum

Tiger bush pasture

25

1590 (283)

86.3 (2.7)

Pterocarpus lucens Combretum micranthum Boscia senegalensis Grewia flavescens

Shrubby steppe

15

222 (67)

95.8 (4.2)

Balanites aegyptiaca Acacia nilotica Guiera senegalensis

a

Species being browsed by animals in an earlier study (Sanon et al., 2007). Means and standard deviation.

b

Table 2 Density of species in each height class in the different pasture typesa Class 1, o1 m

Class 2, 1–3 m

Class 3, 3–5 m

Class 4, 5–7 m

Class 5, 47 m

A. senegal Sparse woody steppe Lowland Tiger bush

44724.1 170.3 35714.3a

572.6 170.2 371.3ab

1075.2 170.1 672.4ab

673.5 0 271.0b

— — —

G. senegalensis Sparse woody steppe Lowland Tiger bush

24713.4a 71726.4ab 2379.3a

572.6ab 3647135.7a 32712.9a

170.3b 35713.2b 572.1ab

0b 170.3c 0b

— — —

P. lucens Sparse woody steppe Lowland Tiger bush

974.7 170.1 2677109.0a

271.1 170.3 2479.8ab

371.5 170.1 37715.1ab

271.0 0 27710.8bc

170.4 0 1074.0c

Density of all species Sparse woody steppe Lowland Tiger bush

3987220.4a 244791.1a 9117371.5a

63734.7ab 6407238.8a 4437180.5a

36719.7ab 136750.7a 225791.6ab

1176.3bc 18710.8b 37715.2bc

170.5c 774.0b 1375.2c

a Mean and standard error of the mean. a,b,c,d: means in the same row with different letters are significantly different (Po0.05).

ARTICLE IN PRESS H.O. Sanon et al. / Journal of Arid Environments 71 (2007) 376–392

382

Falling of leaves

Start foliation

Falling of fruits Flowering

Flowers and fruits

May June July Acacia senegal

Aug

Sept

Oct

Nov

Dec

Jan

Febr

March April

Aug

Sept

Oct

Nov

Dec

Jan

Febr

March April

Guiera senegalensis

Pterocarpus lucens

May

June

July

Fig. 1. Phenogram of Acacia senegal, Guiera senegalensis and Pterocarpus lucens.

All three species started the foliation immediately after the first rain in June. The time between the onset of flushing and full foliation was approximately 1–1.5 months. A. senegal was the first to loose its leaves by the end of the rains in October, and this stage of shedding leaves was so fast that at the end of December, no A. senegal trees bore leaves. G. senegalensis and P. lucens kept their leaves longer, and at the end of March the defoliation was total in all species, except for some individuals of G. senegalensis. On average, the foliation phase lasted 6–7 months for A. senegal, 7–8 months for P. lucens and 8–9 months for G. senegalensis. A. senegal and P. lucens flowered before G. senegalensis, in which flowers appeared in full foliation phase in September. The fruiting lasted on average 6–7 months for all species. Some individuals of A. senegal and P. lucens did not lose all pods and a few pods were observed late in the dry season. For all species, individuals o1 m did not bear flowers or fruits, nor did class 2 of P. lucens. 3.3. Total and accessible edible biomass production Table 3 shows the mean edible biomass production accessible to sheep, cattle and goats for A. senegal, G. senegalensis and P. lucens according to the height classes. The edible

ARTICLE IN PRESS H.O. Sanon et al. / Journal of Arid Environments 71 (2007) 376–392

383

Table 3 Foliage production available (g DM)a to sheep, cattle, goat and total biomass production per species and per height class Class 1, o1 m

Class 2, 1–3 m

Class 3, 3–5 m

Class 4, 5–7 m

Class 5, 47 m

A. senegal Sheep Cattle Goat Totalb

2179a 2177a 21710a 2176a

572ab 119740bc 204766bc 14237380b

672ab 204769c 3787123c 716571911c

170.2b 28710ab 56718ab 1163273102c

— — — —

G. senegalensis Sheep Cattle Goat Total

90738 90730a 90729a 90723a

155764 4197143ab 4887159b 10307275b

114747 4687159b 604 7196b 34567922bc

67728 3247110ab 4477145b 611871632c

— — — —

P. lucens Sheep Cattle Goat Total

1676 1675a 1675a 1674a

572 57719ab 98732b 21047561b

170.4 73725ab 163732b 616771645bc

271 139747b 300798b 1127873008cd

271 110 738b 289794b 261757 6980d

a,b,c,d: means in the same row with different letters are significantly different (Po0.05). a Mean and standard error of the mean. b All edible biomass on the tree (leaves and fruits).

biomass available for sheep decreased significantly from class 1 for all species. The part available to cattle and goats showed a peak in class 3 of A. senegal and G. senegalensis and class 4 of P. lucens. Goats browsing at higher height had more edible biomass at their disposal. Considering the total edible biomass production, P. lucens was the most productive species with a maximum production recorded with individuals taller than 7 m, 26 kg, followed by A. senegal (11.6 kg mean of individuals from 5 to 7 m high). In general, all the leaves from individual plants o1 m are accessible to all the animal species studied. The amount of accessible edible biomass in percent of total edible biomass decreased with increasing plant height for all three species (Fig. 2). If goats are used as the reference, the accessible edible biomass of P. lucens is less than 5% of total edible biomass in all height classes. For A. senegal the edible biomass accessible to goats was 14.3%, 5.3% and 0.5%, respectively, of total biomass, for class 2, 3 and 4. The corresponding values for G. senegalensis were 47.4%, 17.5% and 7.3%, respectively, for the same height classes. 3.4. Relation between edible biomass production and dendrometric parameters In general, simple linear regression models gave low but significant correlations between edible biomass and dendrometric parameters. Linear regression with log10 transformed data gave the best estimation of total edible biomass production for all species and all parameters. Table 4 gives the matrix of regression coefficients with related equations between dendrometric parameters of trees (trunk diameter, crown diameter, height) and total edible

ARTICLE IN PRESS 384

H.O. Sanon et al. / Journal of Arid Environments 71 (2007) 376–392

50 sheep cattle goats

45

% edible biomass

40 35 30 25 20 15 10 5 0 1-3m 3-5m 5-7m A. senegal

1-3m 3-5m 5-7m G. senegalensis

1-3m 3-5m 5-7m >7m P. lucens

Fig. 2. Percentage of edible biomass available to sheep, cattle or goats according to different heights of the trees.

Table 4 Correlations between total foliage production and dendrometric parameters

A. senegal Trunk diameter Crown diameter Height G. senegalensis Trunk diameter Crown diameter Height P. lucens Trunk diameter Crown diameter Height

Regression equation

R2

Log(FP) ¼ 0.12+2.64 Log(TD) Log(FP) ¼ 3.48+2.49 Log(CD) Log(FP) ¼ 2.18+2.22 Log(H)

0.80*** 0.96*** 0.87***

Log(FP) ¼ 0.55+1.89 Log(TD) Log(FP) ¼ 3.02+2.46 Log(CD) Log(FP) ¼ 2.06+2.12 Log(H)

0.68*** 0.90*** 0.87***

Log(FP) ¼ 0.40+2.86 Log(TD) Log(FP) ¼ 2.82+2.35 Log(CD) Log(FP) ¼ 2.51+2.30 Log(H)

0.75*** 0.98*** 0.96***

*** Po 0.001; FP ¼ foliage production, g; TD ¼ trunk diameter, cm; CD ¼ crown diameter, cm; H ¼ height, m.

biomass production of the species. Since the edible biomass accessible to animals was only a part of total edible biomass and also varied with the height of the plants, the correlations between dendrometric parameters and accessible edible biomass were lower. The highest regression coefficients for total edible biomass were obtained with crown diameter for all three species, R2 ¼ 0.96, 0.90, 0.98 for A. senegal, G. senegalensis and P. lucens, respectively (Figs. 3–5). The regression coefficients for accessible edible biomass varied with animal species and plant species, from R2 ¼ 0.36–0.73 with crown diameter, R2 ¼ 0.31–0.71 with trunk diameter and R2 ¼ 0.37–0.62 with height. In general the regressions tested had low R2 values for the edible biomass accessible to sheep. The correlation improved as the accessible proportion increased. For instance, for goats regression coefficients above 50% for all parameters in each plant species were found.

ARTICLE IN PRESS H.O. Sanon et al. / Journal of Arid Environments 71 (2007) 376–392

385

log10(total edible biomass) = - 3.48 + 2.49 log10(crown diameter) 35000 R2 = 0.96

Total edible biomass (g)

30000

Regression

25000

95% CI

20000 15000 10000 5000 0 0

200

400

600 800 1000 Crown diameter (cm)

1200

1400

Fig. 3. Relationship between total edible biomass and crown diameter for Acacia senegal.

log10(total edible biomass) = - 3.02 + 2.46 log10(crown diameter) 16000 R2 = 0.90

14000

Regression Total edible biomass (g)

12000

95% CI

10000 8000 6000 4000 2000 0 0

100

200

300 400 500 Crown diameter (cm)

600

700

800

Fig. 4. Relationship between total edible biomass and crown diameter for Guiera senegalensis.

3.5. Potential of browse production in pastures Table 5 shows the contribution of A. senegal, G. senegalensis and P. lucens to the edible biomass production of different pasture types. In sparse woody steppe, the major part of the edible biomass production is provided by the individuals of classes 3 and 4 of A. senegal, with 64.5 and 69.8 kg DM/ha, respectively. In total this species produced

ARTICLE IN PRESS H.O. Sanon et al. / Journal of Arid Environments 71 (2007) 376–392

386

log10(total edible biomass) = - 2.82 + 2.35 log10(crown diameter) 50000 R2 = 0.98

Total edible biomass (g)

40000

Regression 95% CI

30000

20000

10000

0 0

200

400

600 800 Crown diameter (cm)

1000

1200

1400

Fig. 5. Relationship between total edible biomass and crown diameter for Pterocarpus lucens.

Table 5 Estimation of the production of browse fodder (kg DM/ha) from the three species studied in the different pasture types Class 1, o1 m

Class 2, 1–3 m

Class 3, 3–5 m

Class 4, 5–7 m

Class 5, 4 7m

Total/ species

Lowland pasture A. senegal G. senegalensis P. lucens Total/class

0.02 6.39 0.02 6.43

1.42 374.92 2.10 378.45

7.16 120.96 6.17 134.29

— 6.12 — 6.12

— — — —

8.61 508.39 8.29 525.28

Sparse woody steppe A. senegal G. senegalensis P. lucens Total/class

0.92 2.16 0.14 3.23

7.12 5.15 4.21 16.47

64.48 3.46 18.50 86.44

69.79 — 22.56 92.35

— — 26.17 26.17

142.32 10.77 71.58 224.67

Tiger bush pasture A. senegal G. senegalensis P. lucens Total/class

0.73 2.07 4.27 7.08

5.69 32.96 50.50 89.15

28.66 17.28 228.18 274.12

34.90 — 304.51 339.40

— — 261.75 261.75

69.98 52.31 849.20 971.50

142 kg DM/ha. P. lucens also had a substantial contribution, reaching 72 kg DM/ha. The tiger bush pasture recorded the highest edible biomass production (972 kg DM/ha) with P. lucens providing 849 kg DM/ha. Browse production in lowland pasture is mainly

ARTICLE IN PRESS H.O. Sanon et al. / Journal of Arid Environments 71 (2007) 376–392

387

represented by that of G. senegalensis, which accounted for 508 kg DM/ha. The individuals of 3–5 m height contributed 375 kg DM/ha. 4. Discussion 4.1. Phenology The phenological phases varied both between species and within species according to height class. In general, all three species bore leaves in the beginning of the rainy season and from July to September–October all were in full foliation and fruiting, indicating a good availability of browse fodder. However, in the middle of the dry season (February–March) this availability decreased considerably as the trees entered a period of vegetative rest. The phenograms of A. senegal and G. senegalensis are similar to those shown by Poupon (1979), who did a long-term phenological study (1971–1977) on 17 spp. in the Sahelian zone in Northern Senegal and reported that the leaves of G. senegalensis yellow quickly in the dry season, but remain on the branches almost 2 months before falling. This ability of G. senegalensis to keep its leaves longer seems to be related to its deep root system, which is extensive, enabling the plant to access soil water several metres away both horizontally and vertically (Breman and Kessler, 1995; Seghieri and Simier, 2002). G. senegalensis has also been shown to have a high plasticity (Devineau, 1999; Seghieri and Laloe¨, 2004) that allowed the plant to survive a large range of water status. Ickowicz et al. (2005) noted that the foliation phase of A. senegal occurred from before the rainy season (April) until the beginning of the dry season. Poupon (1979) found that the starting date of foliation appears earlier during years with higher precipitation, but the end of foliation does not differ significantly from year to year. Thus the annual amount of rain, and its distribution, influences the development of the vegetation. Many factors are shown to influence the phenological pattern of woody species, but the most important is water availability, as determined by soil moisture (Borcherf, 1994; Pavon and Briones, 2001). Relative air humidity, soil type, water storage capacity of the plant and temperature are also involved in the changes between the different phases. Do et al. (2005) also gave a new insight into the additional role of evaporative demand in canopy phenology of dry tropics species. However, flowering seems to be less predicable and fruiting is shown to depend on a certain amount of water available (Fredel et al., 1994; Poupon, 1979). Depending on the duration of the foliation phase, Breman and Kessler (1995) and De Bie et al. (1998) classified A. senegal and P. lucens as deciduous species and G. senegalensis as semi-deciduous. The difference between the two groups was shown in their strategy for avoiding drought damage. Deciduous species lost their leaves while semi-deciduous species shed leaves in short periods, having the ability to use water reservoirs in deep layers or close to a river system, and have a strategy for reducing water loss in the dry season by the scleromorphic feature of their leaves (De Bie et al., 1998). Although limited to an observation period of 1 year, this study gives an idea about the period of browse availability, which lasted almost 10 months per year. The variability of phenological phases coupled with the diversity of browse species stressed the fundamental role of this component in animal feeding in the arid zone. If the number of species studied were to be increased, one could probably note as Poupon (1979), that there were species

ARTICLE IN PRESS 388

H.O. Sanon et al. / Journal of Arid Environments 71 (2007) 376–392

with leaves in all seasons. Hiernaux et al. (1994) distinguished five phenological types of Sahelian trees and shrubs, varying in foliation length and the period of renewal of the leaves, when describing the seasonal availability of browse. On the other hand, lopping the trees (P. lucens and G. senegalensis) could prolong the foliation phase. This feature is well exploited by herders at the end of rainy season with P. lucens and during the dry season with G. senegalensis (personal communication). However, tree species react differently to lopping and Hiernaux et al. (1994) noticed that pruning, if not accompanied by the death of the tree, delayed the fall of the leaves. 4.2. Edible biomass production The total edible biomass production depends not only on the species (physiological characteristics), but also on the site parameters (soil, water availability, etc.), whereas accessible edible biomass depends mainly on tree and canopy structure. G. senegalensis is a shrub of 3–4 m height but in the study site most of the plants ranged between 1 and 3 m and had multi-stems, allowing the sprout of leaves in the lowest part of the plants. A. senegal is a shrub or small tree (3–6 m) with thorns. The taller individuals reached 7 m in the sparse woody steppe investigated. The top of the tree is in general flattened and the crown has many branches and erect twigs spreading within the upright part. For P. lucens a thornless shrub or tree of 3–12 m, a maximum height of 15 m was recorded in the tiger bush where it was investigated. The plant has a straight smooth stem in general, but often with ramification at the basal level and with a spreading canopy. Thus the morphology of the species, determined by the crown shape, the branch systems and the height is an important factor of variation in accessible edible biomass. A. senegal and P. lucens have an umbrella-shaped crown, while the crown of G. senegalensis can be described as conical with more biomass in the lower part, which could influence the amount of accessible foliage. In addition the density of the leaves and their distribution on the branches could play a role in the amount of biomass obtained. There was a decrease in accessible edible biomass for all species with increasing height class. For the highest browsing height (1.65 m), apart from the smallest individuals (o1 m) accessible edible biomass accounted for less than 5% of total edible biomass for P. lucens, varied from 0.5% to 14.3% for A. senegal and from 7.3% to 47.4% for G. senegalensis. This decrease in availability with height has been described by many authors. Hiernaux (1980) reported a considerable variability in accessible edible biomass according to the height of the tree, and also according to the species, e.g. with 2 m as the browsing limit, 100% of P. lucens was accessible up to 2 m, decreasing gradually to 7% at 8 m. These values are higher than the results obtained in this study. For A. senegal the result is similar to that reported by Walker (1980), who noted 85.4% of browse taken from the 0–l m layer, 10.4% from the 1–2.5 m layer and 4.2% from the 2.5–5 m layer. Ickowicz and Mbaye (2001) reported also that for plants o2 m 98% of the edible biomass was accessible compared to 0% for plants 47 m. Kalen and Bergquist (2004) studying forage availability for moose of young silver birch and Scots pine noted that edible biomass accessible for browsing increased with tree height to the point at which the defined physical height limit (2.5 m) reduced the accessibility of edible biomass. The proportions of accessible edible biomass in this study for A. senegal and P. lucens are low compared to the estimation of Breman and de Ridder (1991), who suggested that

ARTICLE IN PRESS H.O. Sanon et al. / Journal of Arid Environments 71 (2007) 376–392

389

25% of total biomass could be reached directly by animals. Pellew (1980) considered that 20% of the total browse edible biomass could be found below 2 m. Although the browsing limit proposed by these authors was 2 m, higher than in this study, this broad estimation somewhat overestimate the accessible biomass as it does not take into account the stratification in height of the trees. The present study highlights the need to take the height of the plants into consideration. This is very important when determining accessibility, but the crown shape of the species can also have an influence. The largest part of the edible biomass of trees is not directly accessible to animals. This can lead herders to cut branches for their animals, but this practice is prejudicial in longterm production and also for the survival of the tree if not managed appropriately. With regard to the peak of accessible production at class 3 (3–5 m) for A. senegal and G. senegalensis, and class 4 (5–7 m) for P. lucens, an appropriate pruning for individuals in the higher classes could be suggested. The browsing of the accessible part of the other individuals (except the ones o1 m) could also be used as a range management tool insofar as the leaves in the low strata are shown to be less involved in photosynthesis (Kalen and Bergquist, 2004) and their removal would have only a marginal effect on plant growth. The management of individuals of o1 m that are browsed in all crown area is difficult in the open pastures, which characterize the grazing land in the area. A. senegal and P. lucens showed signs of over-browsing in the areas close to water points and rest areas near camps. Thus the high number of individuals of o1 m recorded with A. senegal and P. lucens (68% and 73%, respectively) may not only be accounted for by good regeneration but also by the browsing effect, that may have slowed down the growth rate of some of these individuals. The total edible biomass production was higher in this study than recorded by Bille (1980), who found a total browse production of 5180 g per tree for A. senegal in an area in Senegal with 250 mm rainfall, 900 g for G. senegalensis in Burkina Faso (440 mm rainfall) and 6450 g for P. lucens in Mali (600 mm rainfall). In the present study, the edible biomass included green fruits, which are normally not well consumed until they have dried. There were no differences in the accessible edible biomass between the years of measurement. For the total edible biomass a higher amount was recorded in the 2nd year, especially for G. senegalensis (70%) and A. senegal (8%), compared to the 1st year. The rainfall in 2005 (551 mm) was above the mean (480 mm) over the last 10 years, and higher compared to 2004 (391 mm). The amount of water available probably influences the edible biomass production. This variation in edible biomass production of browse species is, however, weak compared to that of herbaceous species, as stressed by many authors in the arid zone (le Houerou, 1980). The observations in this study were limited to 2 years, which may be too short a period of time to see significant differences. Care should therefore be taken not to draw any consistent conclusions about the effect of year. The potential browse production in the pastures investigated with regard to the species studied was 225, 525 and 971 kg DM/ha, respectively, in sparse woody steppe, lowland pasture and tiger bush. The three species together represented 22%, 46% and 29% of the browse plants in sparse woody steppe, lowland pasture and tiger bush, respectively. Investigating a broad number of species could improve substantially the estimation of browse production, since most of the species found in these pastures were browsed (480% of species).

ARTICLE IN PRESS 390

H.O. Sanon et al. / Journal of Arid Environments 71 (2007) 376–392

4.3. Estimation of total edible biomass production The strong relation between the crown and total edible biomass can be related to the fact that the crown displays the leaves which capture the radiation energy for photosynthesis. The measurement of crown is easy to undertake in the Sahelian area as the trees are sparse and the crowns seldom jointed. The crown can be estimated if aerial photographs of an appropriate scale are available, although field observation will be required to identify different species. Breman and Kessler (1995) related the canopy area to the edible biomass, arguing for the limited use of trunk diameter at body height and height as dominant parameters in the arid zones. Good predictions of edible biomass by crown measurements, such as crown volume (Bryant and Kothmann, 1979; Paton et al., 2002), crown width (Hughes et al., 1987; Paton et al., 2002) or canopy/sapwood area (Kalen and Bergquist, 2004; Northup et al., 2005) have been reported. The log–log transformation of dependent and independent variables is a regression model widely used in allometric studies (Hughes et al., 1987; Paton et al., 1999; Salis et al., 2006). Many advantages of this model were stressed: it can be derived by means of linear regression and can be extrapolated easily; it has a good ability to stabilize the variance and seems to give reliable results for many types of edible biomass production. It is also shown to reduce the effects of large dependant values relative to that of small values on the prediction equation. Thus, Paton et al. (2002) suggested that this model could be generalized as a standard in vegetation assessment. However, the log transformation is shown to produce systematic errors in estimates when transformed back to the original scale, hence the use of a correction factor which will be multiplied by the predicted values (Ter-Mikaelian and Korzukhin, 1997; Onyekwelu, 2004; Northup et al., 2005). The model developed in this study (single species basis) could be applied for these species within the same site condition and for plants exempted from lopping and protected from current year browsing. 5. Conclusions The availability of browse biomass and its distribution in time in terms of edible biomass was good from June to February. The relations found between the biomass production and different tree parameters demonstrated the possibility of estimating browse biomass production of pasture. The models developed are species-specific and for extrapolation the heterogeneous distribution of plant height should be considered as well as the degree of accessibility, thus the models could be applied to similar agro-ecological zones. It is necessary to extend this study to more species for better assessment of total browse potential. The method used is labour intensive, but it has the advantage of avoiding the destruction of the plants. The results from this study can be used as pasture management tools, but further research is needed on the spatial and temporal variation in browse production including a proper evaluation of production of fruits from various browse species. Acknowledgements The authors would like to acknowledge the Swedish International Development Authority (Sida), Department for Research Cooperation (SAREC), for financial support for this research.

ARTICLE IN PRESS H.O. Sanon et al. / Journal of Arid Environments 71 (2007) 376–392

391

References Baumer, M., 1992. Trees as browse and to support animal production. In: Speedy, A., Pugliese, P.L. (Eds.), Legume Trees and Other Fodder Trees as Protein Sources for Livestock. FAO, Kuala Lumpur, pp. 1–10. Bille, J.C., 1980. Measuring the primary palatable production of browse plants. In: le Houerou, H.N. (Ed.), Browse in Africa. The Current State of Knowledge. ILCA, Addis Ababa, pp. 185–196. Borcherf, R., 1994. Soil and water storage determine phenology and distribution of tropical dry forest trees. Ecology 75, 1437–1447. Breman, H., de Ridder, N., 1991. Manuel sur les paˆturages des pays sahe´liens. Karthala, ACCT, CABO-DLO, CTA, Paris. Breman, H., Kessler, J.J., 1995. Le roˆle des ligneux dans les agro-e´cosyste`mes des re´gions semi-arides (avec un accent particulier sur les pays sahe´liens). AB-DLO, Wageningen. Bryant, F.C., Kothmann, M.M., 1979. Variation in predicting edible browse from crown volume. J. Range Manage. 32, 144–146. Cisse, M.I., 1980. The browse production of some trees of the Sahel: relationships between maximum foliage biomass and various physical parameters. In: Le Houerou, H.N. (Ed.), Browse in Africa. The Current State of Knowledge. ILCA, Addis Ababa, pp. 204–208. De Bie, S., Ketner, P., Paasse, M., Geerling, C., 1998. Woody plant phenology in the West Africa savanna. J. Biogeogr. 25, 883–900. de Leeuw, P.N., Tothill, J.C., 1990. The concept of rangeland carrying capacity in Sub-Saharan Africa—myth or reality. In: Pastoral Development Network 29b, Overseas Development Institute, London, pp. 1–20. Devineau, J.L., 1999. Seasonal rhythms and phenological plasticity of savanna woody species in a fallow farming system (South-west Burkina Faso). Ecology 15, 497–513. Do, F.C., Goudiaby, V.A., Gimenez, O., Diagne, A.L., Diouf, M., Rocheteau, A., Akpo, L.E., 2005. Environmental influence on canopy phenology in the dry tropics. For. Ecol. Manage. 215, 319–328. Fall-Toure´, S., Traore´, E., N’Diaye, N.S., Se`ye, M., 1997. Utilisation des fruits de Faidherbia albida pour l’alimentation des bovins d’embouche paysanne dans le bassin arachidier au Senegal. Livest. Res. Rural. Dev. 9 (5), 1–14. Fredel, M.H., Nelson, D.J., Sparrow, A.D., Kinloch, J.E., Maconochis, J.R., 1994. Flowering and fruiting of arid zone species of Acacia in Central Australia. J. Arid Environ. 27, 221–239. Hiernaux, P., 1980. Inventory of the browse potential of bushes, trees and shrubs in an area of the Sahel in Mali: method and initial results. In: le Houerou, H.N. (Ed.), Browse in Africa. The Current State of Knowledge. ILCA, Addis Ababa, pp. 197–203. Hiernaux, P.H.Y., Cisse´, M.I., Diarra, L., de Leeuw, P.N., 1994. Fluctuations saisonnie`res de la feuillaison des arbres et des buissons sahe´liens. Conse´quences pour la quantification des ressources fourrage`res. Rev. Elev. Med. Vet. Pays Trop. 47, 117–125. Hughes, H.G., Varner, L.W., Blankenship, L.H., 1987. Estimating shrub production from plant dimensions. J. Range Manage. 40, 367–369. Ickowicz, A., 1995. Approche dynamique du bilan fourrager applique´ a` des formations pastorales du Sahel tchadien (Dynamic approach of fodder balance applied at pastoral formations in the Sahel of Tchad) The`se de Doct. Univ Paris XII, Val de Marne-Creteil, France. Ickowicz, A., Friot, D., Guerin, H., 2005. Acacia senegal, arbre fourrager sahe´lien. Bois Foreˆt des tropiques 284, 59–70. Ickowicz, A., Mbaye, M., 2001. Foreˆts soudaniennes et alimentation des bovin au Senegal: potentiel et limites (Sudanian forests and cattle feeding in Senegal: potential and limits). Bois Foreˆt des tropiques 270, 47–61. Kalen, C., Bergquist, J., 2004. Forage availability for moose of young silver birch and scots pine. For. Ecol. Manage. 187, 149–158. le Houerou, H.N., 1980. The role of browse in Sahelian and Sudanian zones. In: le Houerou, H.N. (Ed.), Browse in Africa. The Current State of Knowledge. ILCA, Addis Ababa, pp. 83–102. le Houerou, H.N., 2000. Utilization of fodder trees and shrubs in the arid and semiarid zones of West Asia and North Africa. Arid Soil Res. Rehab. 14, 101–135. Ngwa, A.T., Pone, D.K., Mafeni, J.M., 2000. Feed selection and dietary preference of forages by small ruminants grazing natural pasture in the Sahelian zone of Cameroun. Anim. Feed Sci. Technol. 88, 253–266. Northup, B.K., Zitzer, S.F., Archer, S., McMurtry, C.R., Boutton, T.W., 2005. Above-ground biomass and carbon and nitrogen content of woody species in a subtropical thornscrub parkland. J. Arid Environ. 62, 23–43.

ARTICLE IN PRESS 392

H.O. Sanon et al. / Journal of Arid Environments 71 (2007) 376–392

Onyekwelu, J.C., 2004. Above-ground biomass production and biomass equations for even-aged Gmelina arborea (ROXB) plantations in south-western Nigeria. Biomass Bioenergy 26, 39–46. Paton, D., Nunez, J., Bao, D., Munoz, A., 2002. Forage biomass of 22 shrub species from Monfragu¨e Natural Park (SW Spain) assessed by log–log transformation models. J. Arid Environ. 52, 223–231. Paton, D., Nunez-Trujillo, J., Diaz, M.A., Munoz, A., 1999. Assessment of browsing biomass, nutritive value and carrying capacity of shrublands for red deer (Cervus elaphus L.) management in Monfragu¨e Natural Park (SW Spain). J. Arid Environ. 42, 137–147. Pavon, N.P., Briones, O., 2001. Phenological patterns of nine perennial plants in an intertropical semi-arid Mexican schrub. J. Arid Environ. 49, 265–277. Pellew, R.A., 1980. The production and consumption of acacia browse and its potential for animal protein production. In: le Houerou, H.N. (Ed.), Browse in Africa. The Current State of Knowledge. ILCA, Addis Ababa, pp. 223–232. Poupon, H., 1979. Etude de la phenologie de la strate ligneuse a` Fe´te´-Ole´ (Senegal septentrional) de 1971 a` 1977. Bull. I.F.A.N. 41 (Ser. A), 44–91. Saı¨ d, S., Pellerin, M., Guillon, F., De´bias, F., Fritz, H., 2005. Assessment of forage availability in ecological studies. Eur. J. Wildl. Res. 51, 242–247. Salis, S.M., Assis, M.A., Mattos, P.P., Piao, A.C.S., 2006. Estimating the aboveground biomass and wood volume of savanna woodlands in Brazil’s Pantanal wetlands based on allometric correlations. For. Ecol. Manage. 228, 61–68. Sanon, H.O., Kabore´-Zoungrana, C., Ledin, I., 2007. Behaviour of goats, sheep and cattle and their selection of browse species on natural pasture in a Sahelian area. Small Rumin. Res. 67, 64–74. Sanon, H.O., Nianogo, J.A., Nanglem, N.S., Sanou, S., 2005. Amount of fallen fodder components from Acacia raddiana, Balanites aegyptiaca and Ziziphus mauritiana available to ruminants in selected Sahelian pastures. Livest. Res. Rural. Dev. 17 (12), 1–7. SAS, 1998. Statistical Analysis System. SAS/STAT User’s Guide, version 8.02. Statistical System Inc., Cary, NC. Scholes, R.J., Walker, B.H., 1993. An African Savanna. Synthesis of the Nylsvley Study, Cambridge Studies in Applied Ecology and Resource Management. Cambridge University Press, Cambridge. Seghieri, J., Laloe¨, F., 2004. Characterisation of the variability of the daily course of leaf water potential in the dominant shrub species within Sahelian fallows in south-west Niger. Ecol. Model. 173, 271–281. Seghieri, J., Simier, M., 2002. Variations in phenology of a residual invasive shrub species in Sahelian fallow savannas, south-west Niger. J. Trop. Ecol. 18, 897–912. Speedy, A., Pugliese, P.L., 1992. Legume trees and others trees as protein sources for livestock. In: Speedy, A., Pugliese, P.L. (Eds.), Proceedings of FAO expert Consultation held at MARDI, 14–18 October 1991. FAO, Kuala Lumpur, p. 339. Ter-Mikaelian, M.T., Korzukhin, M.D., 1997. Biomass equations for sixty-five North American tree species. For. Ecol. Manage. 97, 1–24. von Carlowitz, P.G., 1989. Agroforestry technologies and fodder production—concepts and examples. Agrofor. Syst. 9, 1–16. Walker, B.H., 1980. Review of browse and its role in livestock production in southern Africa. In: le Houerou, H.N. (Ed.), Browse in Africa. The Current State of Knowledge. ILCA, Addis Ababa, Ethiopia, pp. 7–24.