This article appeared in a journal published by Elsevier. The attached copy is furnished to the author for internal non-commercial research and education use, including for instruction at the authors institution and sharing with colleagues. Other uses, including reproduction and distribution, or selling or licensing copies, or posting to personal, institutional or third party websites are prohibited. In most cases authors are permitted to post their version of the article (e.g. in Word or Tex form) to their personal website or institutional repository. Authors requiring further information regarding Elsevier’s archiving and manuscript policies are encouraged to visit: http://www.elsevier.com/copyright

Author's personal copy

Journal of Hydrology 375 (2009) 114–127

Contents lists available at ScienceDirect

Journal of Hydrology journal homepage: www.elsevier.com/locate/jhydrol

Sahelian rangeland response to changes in rainfall over two decades in the Gourma region, Mali Pierre Hiernaux a,*, Eric Mougin a, Lassine Diarra b, Nogmana Soumaguel c, François Lavenu a,z, Yann Tracol d, Mamadou Diawara c a

Centre d’Etudes Spatiales de la BIOsphère (CESBIO), UMR 5126 UPS-CNRS-CNES-IRD, 18 Avenue Edouard Belin b.p.i. 2801, 31401 Toulouse Cedex 9, France Institut d’Economie Rurale (IER) B.P. 258, rue Mohamed V, Bamako, Mali c Institut de Recherche pour le Développement (IRD) B.P.2528, Hippodrome 238 rue 234, Bamako, Mali d CEAZA, casilla 599, colina El Pino s/n La Serena, Chile b

a r t i c l e Keywords: Sahel Mali Gourma Vegetation dynamics Patchiness Yield variability

i n f o

s u m m a r y Twenty-five rangeland sites were monitored over two decades (1984–2006) first to assess the impact of the 1983–1984 droughts on fodder resources, then to better understand ecosystem functioning and dynamics. Sites are sampled along the south–north bioclimatic gradient in Gourma (Mali), within three main edaphic situations: sandy, loamy-clay and shallow soils. In addition, three levels of grazing pressure where systematically sampled within sandy soils. Located at the northern edge of the area reached by the West African monsoon, the Gourma gradient has recorded extremes in inter-annual variations of rainfall and resulting variations in vegetation growth. Following rainfall variability, inter-annual variability of herbaceous yield increases as climate gets dryer with latitudes at least on the sandy soils sites. Local redistribution of rainfall explains the high patchiness of herbaceous vegetation, especially on shallow soils. Yet spatial heterogeneity of the vegetation does not buffer between year yield variability that increases with spatial heterogeneity. At short term, livestock grazing during the wet season affects plant growth and thus yield in direction and proportions that vary with the timing and intensity of grazing. In the longer term, grazing also impinges upon species composition in many ways. Hence, long histories of heavy grazing promote either long cycle annuals refused by livestock or else short cycle good quality feed species. Primary production is maintained or even increased in the case of refusal such as Sida cordifolia, and is lessened in the case of short cycle species such as Zornia glochidiata. These behaviours explain that the yield anomalies calculated for the rangelands on sandy soils relative to the yield of site less grazed under similar climate tend to be negative in northern Sahel where the scenario of short cycle species dominates, while yield anomalies are close to nil in centre Sahel and slightly positive in South Sahel where the refusal scenario is more frequent. Because grazing promotes short cycle species, grazed rangelands respond faster to droughts. Year to year changes in species composition are abrupt as expected from the transient soil seed stock. However, some decadal trends in species composition are identified, with a wave of pioneer species following the 1983–1984 droughts, and a more progressive diversification and return to typical Sahel flora from 1992 onwards. Ó 2008 Published by Elsevier B.V.

Introduction Major droughts occurred through out the Sahel in 1972–1973 and again in 1983–1984. They followed a relatively wet period of 20 years 1948–1968 (Nicholson, 2001) and were included within a 25 years dry period 1968–1993. Since 1994 rainfall vary around the overall average. Both droughts had severe impact on the vegetation, crops, livestock and the population of the Sahel. The distur* Corresponding author. Tel.: +33 (0) 561558537; fax: +33 (0) 561558500. E-mail address: [email protected] (P. Hiernaux). z Deceased 10/02/2007. 0022-1694/$ - see front matter Ó 2008 Published by Elsevier B.V. doi:10.1016/j.jhydrol.2008.11.005

bances on vegetation and soils by the first drought were studied and monitored over a few sites in the Gourma by Boudet (1972, 1977, 1979) and Leprun (1992). First reports (e.g. Boudet, 1972) were alarming and warning on the risks of rapid desertification. The following reports (Boudet, 1977, 1979, 1984; Coulibaly, 1979) were less pessimistic, herbaceous layer had rapidly recovered, at least on sandy soils, and some of the decimated woody plant populations had started to regenerate soon after the drought. However, the desertification trend was confirmed, and responsibility for a larger fraction of vegetation and land degradation attributed to natural resource management, especially to pastoralist through grazing livestock at rates believed unsustainable (Gallais,

Author's personal copy

P. Hiernaux et al. / Journal of Hydrology 375 (2009) 114–127

1975). Then occurred the second drought that peaked in 1984 with repeated crop failure for two or three years in a row, spectacular losses in vegetation cover that triggered wind and run-off soil erosion and massive losses in livestock, famine, impoverishment and emigration of the population (Hiernaux, 1996). In the framework of the impact assessment of that drought, 25 rangeland sites were sampled and described (Hiernaux et al., 1984). Some of the sites were selected at the locations where Boudet and his colleagues had made their observations in order to capitalise on the dynamics already described. Other sites were added to sample the North– South bioclimatic gradient, the three main soil types and the range of grazing intensity. The monitoring of these 25 sites was carried out over ten years (1984–1993) and progressively intensified in 2000 onwards under the AMMA project (Redelsperger et al., 2006; Mougin et al., 2009). Located at the northern edge of the area reached by the West African monsoon, Gourma has recorded extremes in inter-annual variation of rainfall and resulting variations in vegetation growth (de Leeuw et al., 1992). This paper aims at presenting, first, the methods used to monitor range vegetation, climate and environment, second, the resulting data base, and third, the analyses of the herbage yield dynamics over years. Yield trends are discussed and related to rainfall, grazing status and species composition. The interpretation of yield trends also rely on previous surveys on soil seed stocks and germinations (Hiernaux and Diarra, 1993), root masses and grass tillering (Hiernaux et al., 1994), and on the results of controlled burning and grazing experiments conducted at the same sites (Hiernaux and Turner, 1996). The first hypothesis tested in these analyses, is that Sahel vegetation, dominated by annual herbaceous, would linearly respond to rainfall (Le Houérou et al., 1988; Prince, 1991; Olsson et al., 2005). And because variation in rainfall are larger, at least in relative term, toward the north of the gradient, variation in vegetation production would also be greater toward the drier end of the bioclimatic transect (Le Houérou et al., 1988). Moreover, the impact of rainfall being mediated by the redistribution of rain water at the soil surface, a corollary is that vegetation on shallow soil or poorly permeable soil should vary more than in deep permeable soils. A complementary hypothesis is that the herbaceous layer would suffer more and recover less from drought when subjected to heavy grazing by livestock (Boudet, 1972; Breman and de Wit, 1983; Bille, 1992; Hein and de Ridder, 2006). Because the herbaceous layer is dominated by annuals, it is also hypothesised that the duration of the wet and dry periods that succeeded in the second half of the 20th century are long enough to trigger a shift in species composition toward species more adapted to dryer climate, including some perennial grasses (Breman and Cissé, 1977; Hiernaux and Le Houérou, 2006). Material and methods Monitoring sites are sampled along the South–North bioclimatic gradient in four groups (Fig. 1, Table 1). In each or these four sets of sites, three main edaphic situations were sampled: deep sandy soils, deep loamy-clay soils, shallow soils on rock or hard pans. On the sandy soils that extend over about half the landscape (Kammerud, 1996), three levels of grazing pressure where systematically sampled: low, medium and high, in relation to the proximity, size and seasonal duration of neighbouring water points and associated villages and encampments. Climate Unfortunately, the web of rain gauges and meteorological stations managed by the National Meteorological Service (DNM) is

115

quite lax over Gourma. A few manual rain gauges were set by research or development projects following 1984, but it is only since AMMA research project developed, starting in 2000, that automatic meteorological stations were set in a few field sites (Fig. 1) and the web of manual gauges densified especially within the targeted 2500 km2 ‘AMMA super-site’ area close to Hombori (Frappart et al., 2009). Soils The elevation profile is recorded along site transects, together with the line intersection of the soil surface features visually identified following Casenave and Valentin (1989). Soil profiles have been described systematically in each vegetation facies with texture and chemical analysis (organic C, total N and total P concentrations) of soil samples at depth of 0–6, 6–12, 12–25, 25–50, 50–100 and 100–150 cm (Hiernaux et al., 1984). Herbaceous vegetation Plant species are named according to the Flora of West Tropical African second edition (Hutchinson and Dalziel, 1954–1972). In Sahel, vegetation is composed of an herbaceous layer dominated by annual plants, and a scattered population of shrubs and low trees (Hiernaux and Le Houérou, 2006). At landscape scale, vegetation is organised in large units following main topography, soil types and land uses (Breman and de Ridder, 1991). Locally, woody and herbaceous plant density and species composition are organised in facieses following finer topography and soils nuances or differences in land use practices and histories. The pattern of these facieses is often repetitive such as in the succession of dunes and interdunes in ergs, or thicket and impluvium in ‘tiger bush’, or else crop-fallow fields in croplands. In order to reduce both the variance associated to field sampling and field labour, the herbaceous layer is monitored using a two-level stratified random sampling design. Facieses are the first level of stratification and are sampled separately. When there are at least two facieses, the site is considered a mosaic of facieses described by the relative area and the distribution pattern of the component facieses (Fig. 2). Within each facies the herbaceous layer is variable enough at the scale of measure plots (1  1 m) to maintain the standard deviation of the mean attributes high when the number of samples is increased, even to large numbers (Grouzis, 1988). This local heterogeneity justifies the second level of stratification systematically applied. All facieses are stratified into four strata based on the apparent bulk of the herbaceous layer: either nil in bare soil patches, low, medium or high in vegetated patches. The three vegetated strata are empirically defined relative to the status of the herbaceous layer within facies at the date of observation starting by identifying what could be considered a ‘modal state’ based on the apparent bulk of the herbaceous layer, from which ‘low’ (‘high’) states are derived when bulk is ‘obviously’ (at least by 25%) inferior (superior) to the mode. The three vegetated strata are sampled separately. The facies is described by the frequency of the component strata, and by the weighted mean of the strata mean attributes such as plant cover, mass, species composition. In turn the site is described by the frequency of the component facieses, and the weighed mean of the facieses mean attributes. To match with the spatial scale of facies distribution, the stratification in facieses and strata in the Gourma sites is performed along one (or two) 1000 m long transect, with readings every meter within the 1 m wide band (Fig. 2). Total and green vegetation cover (visual estimates in %), standing and litter mass (destructive measure, with harvest, air drying and weighing) and species composition (list with visual estimates of contribution to bulk) are assessed in 1  1 m plots randomly

Author's personal copy

116

P. Hiernaux et al. / Journal of Hydrology 375 (2009) 114–127

Fig. 1. Distribution of vegetation monitoring sites, meteorological stations, rain gauges and soil moisture measuring devices along a North–South transect in central Gourma, Mali. Main soil types are thick sands (Sand), thick loamy-clay (Clay), shallow soils on rock or hard pan outcrops (Rocks).

sampled in each of the vegetated strata along the transect. Site weighed mean attributes are calculated from strata means as j X k X mjk  pjk PP M¼ pjk 1 1

2

S ¼

j X k X 1

! ð1Þ

with M the weighed mean attribute, j number of facies, k number of strata within facies, mjk the attribute mean of samples in strata k and facies j, pjk the relative frequency of strata k in facies j. The variance of the weighed mean this calculated following Cook and Stubbendieck (1986):

1

p2jk



s2jk njk

! þ

1 j P k P 1



X X   pjk  m2jk  M 2

Pjk

1

ð2Þ with S2 the variance of the weighed mean attribute, s2jk the variance of the mean of the samples in strata k and facies j, and P jk the absolute frequency of strata k and facies j recorded along the transect. In order to optimise the error on weighed mean attributes the number of plots sampled per strata is unequal: set to a minimum

Author's personal copy

P. Hiernaux et al. / Journal of Hydrology 375 (2009) 114–127 Table 1 Sampling grid of the rangeland sites monitored in Gourma from 1984 onwards. The sites are figured by their code number. Hydrology

?

Endoreic systems

Structured watersheds

Top soil texture

?

Sands, loamy-sands

Clay, loam

Grazing pressure

?

Low

High

High

Low

Range of annual rainfall (mm)

250–150 350–250 450–350 450–550

1; 5 10; 12 17; 18 25; 30; 38

4; 6 14; 32 19; 31 37

2; 9 15 20; 21 28

8 16; 40 22 36

Rock, gravels

of three for the low and high strata, and a minimum of six for the medium strata, often six low-six high-twelve medium (Hiernaux, 1995). Moreover, to assess vegetation overall heterogeneity alternative calculation of weighed mean and its standard deviation are performed systematically. They are calculated as the mean and standard deviation of a virtual population of 100 plots randomly selected within the normal distribution of each component strata as defined by the mean and variance of strata samples, in number proportional to the strata frequency (Hiernaux, 1995). Because perennial populations are scattered and the mass of individual plants is high, the assessment of their contribution to the vegetation attributes is done indirectly through the record of the population density by species and/or size class, the record of size parameter on individual plants (height, basal and crown diameters, total and green cover within canopy %). Cover, and mass attributes are derived from population/size density records and regressions established between size and cover, volume and mass on sub-sampled plants. As for woody plant populations (Hiernaux et al, 2009b), two methods are concurrently used to assess population density: either a plain inventory of all plants within circular plots centred on points systematically selected along the transect (every 50 or 100 m), or by measuring the distance (d) of the nearest plant in each of the four sectors defined by the transect axe and the perpendicular to the axe at points systematically selected along the transect. The density (#ha1) is calculated as

D¼

10; 000 d2

ð3Þ

with d2 the mean distance in the four sectors (Cottam and Curtis, 1956).

117

Grazing pressure Livestock movements composed of local grazing circuits and regional itineraries are part of the pastoral management system that prevails in Gourma as in most of Sahel (Schlecht et al., 2001; Turner et al., 2005). These movements are adaptations to spatial, seasonal and interannual variations of pastoral resources availability for both water and fodder. Except experiments conducted in fenced pasture in 1991–1992 to assess the impact of controlled stocking rate on herbaceous growth and straws decomposition (Hiernaux and Diarra, 1993) and a few regional surveys (Boudet et al., 1971; RIM, 1987), there are no reliable data on stocking rates. The assessment of the grazing pressure on the monitoring sites is thus limited to a subjective rating based on the observed indirect indicators of the intensity of livestock frequentation: herb height, grass tillering, standing versus litter ratio, soil trampling and dung deposition (Schlecht et al. 2006). Results Seasonal herbaceous growth Being largely dominated by annual plants, herbaceous growth starts by seed germination following the first rains, sometimes between late May (as in year 2005 at Agoufou, Fig. 3) and July (as in 2007, Fig. 3). In a first period, which duration depends on earliness of germination and on soil moisture availability, above ground growth remains modest, below 15–30 kg of Dry Matter (DM) d1, with seedling establishing their root system, while above-ground plant tillers or branches. Rapid growth (45–60 kg DM ha1 d1) only starts from mid July onwards, whenever soil moisture allows (later in 2007 and 2006 than in 2005 and only for a week in late August in 2004 at site 17, Fig. 3), and ends with flowering which date is set by sensitivity to photoperiod, between late August and late September depending on species and on the date of germination (Penning de Vries and Djiteye, 1982). Standing mass is already decreasing when the annuals reach fructification, and then decreases rapidly with seed dispersion and plant wilting in an environment still hot and humid. In absence of grazing further decrease of dry herbaceous standing mass is slow during the first part of the dry season, and accelerate from March on as the temperature and air humidity build on. With no grazing during the wet season, maximum standing mass provides an acceptable estimate of above

Fig. 2. Each facies includes four strata: O = bare soil, B = low, M = medium, H=High apparent herb density. The strata B, M et H are sampled with a minimum of 3 (B, H) or 6 (M) 1  1 m plots randomly selected along the 1000 m transect.

Author's personal copy

118

P. Hiernaux et al. / Journal of Hydrology 375 (2009) 114–127 -1 a Kg DM ha -2

2500

#m

mass

density

2000 Fire 1500 1000 500

b 70

1/6

1/3

1/12

1/9

1/6

1/3

1/12

1/9

1/6

1/3

1/12

1/9

1/6

1/3

1/12

1/9

1/6

0

mm

Daily rainfall

60 408

50

377

40

291

30

167

20 10

1/6/08

1/3/08

1/12/07

1/9/07

1/6/07

1/3/07

1/12/06

1/9/06

1/6/06

1/3/06

1/12/05

1/9/05

1/6/05

1/3/05

1/12/04

1/9/04

1/6/04

0

Fig. 3. Seasonal dynamics of the annual herbaceous vegetation at the rangeland site of Agoufou (# 17): (a) above ground mass (kg DM ha1), and density of annual herbaceous plants (m2) from germination to flowering (b) daily and annual total rainfall (mm) from 2004 to 2007.

a

kg DM m-2 70

green

dry

total

60 50 40 30 20 10

b

1/6

1/4

1/2

1/12

1/10

1/8

1/6

1/4

1/2

1/12

1/10

1/8

1/6

0

mm 50

daily rainfall 40 30

189

251

20 10

1/6/08

1/4/08

1/2/08

1/12/07

1/10/07

1/8/07

1/6/07

1/4/07

1/2/07

1/12/06

1/10/06

1/8/06

1/6/06

0

Fig. 4. Seasonal dynamics of perennial herbaceous at the rangeland site of Zaouati (# 32): (a) green, dry and total mass of Panicum turgidum per unit basal area of tussock (kg DM m2) (b) daily and annual total rainfall (mm) in 2006 and 2007.

Author's personal copy

119

P. Hiernaux et al. / Journal of Hydrology 375 (2009) 114–127

At any one year, large differences in herbaceous standing mass or cover are observed between sites (Fig. 5). Herbaceous yields are systematically low on the shallow soils sites for which a large fraction of the rainfall is lost by run-off as in the ‘tiger bush’ site at Ortondé (Fig. 5). Herbaceous yields on the fine textured soils in the depressions collecting the water that runs off from watershed present extreme values depending on success or failure of herbaceous germination and seedling to stand flood as for years 2000, 2005 (success), 2002, 1993 (failure) in the flooded clay plain site at Kelma (Fig. 5). Yields on sandy soils vary less dramatically and stand in between the extremes of the two previous soil types regardless of grazing pressure as demonstrated by yields in fixed dune sites at Agoufou (heavily grazed) and Timbadior (lightly grazed) (Fig. 5).

plots, assesses the spatial heterogeneity of the herbaceous yield. The mean of these standard deviations at a site for the series of year 1984–2006 is an indicator of the mean spatial heterogeneity of the herbaceous layer at that site (Hiernaux, 1995). These indicators of spatial heterogeneity, calculated for each site over the whole monitoring period, range between 50% and 498% for standing mass and average 124.5% for all sites (Table 3). Part of the large coefficient of variation is due to extends of bare soil patches. The rock or hard pan outcrops and the patchiness of the vegetation cover on the shallow soil sites explains the extremely high coefficients of variation observed for mass yields (295.1%). On loam-clay soils, extends of bare patches are very variable and linked to microtopography and flood regime. Bare soil patches could be restricted to the termite mounts and to the deepest channels or pools, or else spread largely from the margins of the plain depending on the flood regime. As a consequence the average coefficient of variation of mean yield is relatively high at 86.0%. In sandy soils, bare patches are limited to sand deflation and active sand accumulation due to wind erosion (sites #1, 6, 19, 32), gullies (sites #5, 14, 19) and area artificially denudated either by cropping (site 19, or camp settlement (site #5, 6, 12, 14, 18, 31, 37). However, topography, nuances in soil texture or soils surface feature in some of the sandy soils sites (dune sites #5, 6, 10, 14, 17, 25, 31, 32) affect soil moisture regime and thus vegetation yield adding to the heterogeneity. Indeed, successive dune ridges, slopes, and inter-dune depressions, differ in mean yields, cover and species composition as observed in 2005 at Agoufou (Fig. 6) or at Zaouati in 2006 (Table 2). Finally, mean yield coefficients of variation still reach 79.5% in lightly grazed sandy soils. History of heavy grazing slightly reduces these coefficients to 76.6 (Table 3).

Spatial distribution of the annual herbaceous vegetation

Spatial distribution of the perennial herbaceous vegetation

In most sites, herbaceous yield and cover largely vary between patches and facieses. The coefficient of variation of the mean yield calculated for a site and a year from hundred 1 m2 random virtual

Perennial herbaceous are scarce in Gourma as in sahelian rangelands in general (Hiernaux and Le Houérou, 2006). They occur however in upland position at the northern margins of the Sahelian zone (site #1, 6, 32, and monitoring sites located north of Bamba) and in lowland position at the Southern margin (#25, 28, 38). Perennial growing at Sahel northern edge are mostly grasses or sedges such as P. turgidum, Aristida sieberiana, Cymbopogon schoenanthus, Cyperus jeminicus, Cyperus conglomeratus, and dicotyledons such as Aerva lanata, Chrozophora senegalensis, Pergularia tomentosa. Both group of perennials also occur locally and transiently in central Sahel: some P. turgidum and A. sieberiana tussocks settled temporarily in eroded spots in sites 1, 6, 10 and 19, short living perennial dicotyledons such as C. senegalensis, A. javanica also settled temporarily in almost all sandy soil sites following the drought (4347 plants per ha and 4.4% cover for C. senegalensis in site 19 in 1988) or burning (697 plants per ha and 3.7% cover for C. senegalensis in site 30 in 2007). The density and size of P. turgidum tussocks in Zaouati (site 32) largely vary according to dune

ground production (Bille, 1977; Cornet 1981; Ayantunde et al., 1999). The seasonal growth of perennial herbaceous is more widely spread over the year with regrowth often starting early in the wet season, sometimes even preceding the rains, and a peak in green standing mass and flowering occurring early in the dry season as for Panicum turgidum monitored at Zaouati (Fig. 4). The yield of perennials herbaceous at a site results from the densitysize of the perennial plant population (Table 2) and the production of individual plants (Fig. 4). In addition to the lag of peak production with rainfall, the site yield of perennials appears less related to rainfall volume and distribution than for annuals: yield of P. turgidum at Zaouati was much larger in 2006 than in 2007 despite lower rainfall (Fig. 4). Between sites differences in herbaceous yields

Table 2 Spatial heterogeneity of the herbaceous cover in the dune site at Zaouati (#32) in September 2005: density and mass of the perennial grass, Panicum turgidum, and mass of herbaceous annuals, across the three facieses: dune windward and leeward slopes, and inter-dunes. Site mass and density are weighed averages by the relative extend of facies (area%). Facies

Dune windward Dune leeward Inter-dune All sites

Area (%)

Panicum turgidum tussocks

Annuals herbaceous

Total herbaceous

Density (/ha)

Mass (kg/ha)

Mass (kg/ha)

Mass (kg/ha)

66.7 11.5 21.8

1808 180 1433

784 130 686

389 677 491

1173 807 1177

100.0

1321

678

483

1161

4000

kg DM ha -1

"thick sands, heavy grazing"

"thick sands, light grazing"

"Clay plain, heavy grazing'

"shallow soil,light grazing"

Centre Sahel

3000 2000 1000

2006

2005

2004

2003

2002

2001

2000

1999

1998

1997

1996

1995

1994

1993

1992

1991

1990

1989

1988

1987

1986

1985

1984

0

Fig. 5. Inter-annual variations of herbaceous yields in 4 rangeland sites close to Hombori (15.3 °N, 1.7°W), with contrasted topography, soil types and grazing status: shallow loamy soil over hard pan on a gentle slope at Ortondé (# 22), temporarily flooded clay plain depression at Kelma (# 21), fixed sand dunes with heavy grazing at Agoufou (# 17) and moderate grazing at Timbadior (# 18).

Author's personal copy

120

P. Hiernaux et al. / Journal of Hydrology 375 (2009) 114–127

Table 3 Total and site group means (standard-errors) over 1984–2006 of the annual maximum herbage mass (kg DM ha1) in 24 Gourma rangeland sites grouped along soil texture and grazing pressure. Also included, the total and group means of the inter-annual variability and the spatial heterogeneity indices, respectively: the coefficient of variation of the mean mass over years, and the over years mean of the coefficient of variation of the site mean mass calculated from 100 virtual plots. Soil types

Sand Sand Shallow Clay

Grazing intensity

Number of site

Light Heavy Light Heavy

7 8 5 4

All sites

Herbaceous mass Mean (kg DM ha1)

s.d. (kg DM ha1)

Variability (%)

Heterogeneity (%)

1137.2 896.3 291.3 1483.6

738.8 584.9 480.2 957.6

61.8 68.8 99.3 63.8

79.5 76.6 295.1 86.0

938.4

670.1

72.3

124.5

24

250

m 318

200

316

herb mass (g/m²)

Topography: altitude (m)

314

150

312 100

310

50

308 306

0 0

200

400

600

800

1000

1200

Fig. 6. Distribution of standing herbaceous mass at the end of the 2005 growing season along a 1200 m transect across fixed dunes at Agoufou (# 17). In average (± s.e.) herbage mass was 158 ± 23 g m2 in depressions, 119 ± 31 on slopes and 92 ± 15 on the top of dunes.

topography with a dense population on the windward slope which tussock size is bell shaped distributed, a scattered population of mostly large tussocks in the inter-dune, and a very sparse population of small size tussocks on the leeward slope of the dune (Table 2). In September 2005, the standing mass of the P. turgidum population weighed by the size of tussocks and their density in each of the three facieses was 678 kg DM ha1 which compared to 483 kg DM ha1 of annual grasses.

variability on sandy soil is more directly related to rainfall conditions. This is illustrated by the four fold larger yield of herbaceous at Agoufou in 2005 than in 2004 (Fig. 3). Yield variability on sandy soils is increasing as climate gets dryer as indicated by the regression established between the variability indicator and the latitudinal position of the sites (Fig. 7c). These relations poorly apply in shallow soils and do not apply in clay soils. Rainfall and grazing impact on yield variability

Between years variations of the herbaceous yields The coefficient of variation of the mean yield over years at a site is an indicator of the inter-annual variability of the vegetation at that site. Calculated for each site over the 1984–2006 period the coefficient of variation range from 42% to 139% depending on site, and equal 72.3% in average for 24 sites (Table 2). Inter-annual variability of site yield is thus lower than spatial heterogeneity at the plot (1 m2) scale. However, the large values of spatial heterogeneity are provided by the shallow soil sites (Fig. 7a). When these sites are set aside, temporal variability and spatial heterogeneity are of same magnitude (Fig. 7b). In any cases, temporal variability increases as mean spatial heterogeneity gets larger following a power function (Fig. 7a and b). The vegetation of shallow soil sites is thus both extremely heterogeneous spatially and variable over years. The high yields and low spatial heterogeneity reached by herbaceous on clayed soils in lowlands when vegetation growth is successful compensate large inter-annual variations and strong heterogeneity when flood regime is not favourable to growth. As a result, the average variability and heterogeneity indices of clay soil sites are close to same indices on sandy soils (Fig. 7b). In both shallow soils and lowlands, the extent of bare soil patches is subject to large changes over years and contributes chiefly to heterogeneity and variability. On sandy soils, bare patches are generally restricted, but they extended markedly from 1983 to 1985 due to wind erosion triggered by the drought in 1983–1984. The patches of deflation and micro-dunes healed over progressively from 1988 onwards resulting in a decrease of the area of bare soil and therefore an attenuation of the herbaceous spatial heterogeneity. Yield

The yield relative anomalies to mean yield over the study period for the sandy soil sites monitored from 1984 to 2006 are linearly related to the rainfall relative anomalies to mean rainfall over the study period calculated for the closest meteorological station (Fig. 8). Regressions are weak but improve (steeper slope, higher r2) from the northern, drier, group of site to the southern, wetter, group of site. Maximum standing mass measured at the end of the growing season is also affected by grazing. To assess grazing effects on seasonal yields within each of the four latitudinal groups on sandy soils, yield anomaly was calculated per year for each grazed site relative to the yield of the less grazed site. The means of these anomalies over years for the sites grouped by grazing pressure classes (moderate, high, intense) within each bioclimatic zone range between 32% and +8% with a trend from negative in northern to slightly positive in southern sites (Fig. 9). Thus only northern sites yields appear negatively affected by grazing pressure, while centre and southern sites appears less affected. Species composition of the herbaceous vegetation At all sites herbaceous vegetation is dominated by annual plants among which grasses generally come first, tallying 74.5 ± 17.4% of plant cover in average for 25 sites over 1984–2006 (Table 4). However, the species and their relative contribution to cover differ with soil type. Species composition of shallow soil sites is relatively diverse with 17 species contributing more than 1% (81.6% all together) and two dominant grasses Schoenefeldia gracilis and Panicum laetum contributing 18.8% and 10.8%, respectively. Species

Author's personal copy

121

P. Hiernaux et al. / Journal of Hydrology 375 (2009) 114–127

a

"Sand, grazing" "Sand, ligthligth grazing"

"Sand, heavy grazing" "Sand, heavy grazing"

"Shallow soils" "Shallow soils"

"Clayed soils" "Clayed soils"

175

0,394 11,3 x0.394 y = 11.3 2 0,65 R = 0.65

Temporal Temporal Variability Variability index % index %

150 125 100 75 50 25

SpatialSpatial heterogeneity index %% heterogeneity index

0 0

100

b150

200

300

Tem poral variability index %

125

400

500

600

y = 4.3 x 0.619 R2 = 0.58

100 75 50 Spatial heterogeneity index % 25 25

c

50

75

100

125

150

Latitude North

Regression for all sandy soils

'sand,light'

Iv = 0.032 Lat + 13.525 R2 = 0.67

17

'sand, heavy' 16

'Shallow, light' 'Clay, heavy' Shallow soils (except #2)

15

Iv = 0.021 Lat + 13.261 R2 = 0.99 Temporal variation index, Iv % 14 0

25

50

75

100

125

150

175

Fig. 7. Temporal variability of herbaceous yield versus mean spatial heterogeneity of the standing herbaceous mass calculated over 1984–2006 for (a) all soil types (24 sites), (b) sandy and clayed soils (19 sites), (c) temporal variability of herbaceous yield plotted against the latitude position of sites sorted by soil type and levels of grazing pressure. Linear regression between temporal variability and latitude for sandy sites and for shallow soil sites (site # 2 excluded).

composition of clayed soil sites in lowlands is less diverse with only 10 species contributing to 1% at least (91.4% together), and more largely dominated by grasses adapted to temporary flood: Echinochloa colona and P. laetum contributing to 42.0% and 33.6%, respectively. Sedges and dicotyledons, including legumes such as Aeschynomene ssp and Sesbania ssp are locally abundant. Species composition on sandy soil sites is almost as diverse as on shallow soils with 16 species contributing more than 1% (82.4% together), but the grass species ranking first, Cenchrus biflorus has a large share with 31.8%. Annual grasses dominate largely with 15.7% only left to dicotyledons among which legumes cover 10.7% with two highly palatable species Zornia glochidiata and Alysicarpus ovalifolius. The large dominance of annual grasses (Table 4) does not impede large inter-annual fluctuations in species composition. Some

of these changes contribute to trends over several years while the majority is just abrupt and does not last more than a year or two. Examples of these abrupt changes in species composition are noticed on the graphs of the inter-annual changes of the contribution of grasses to total vegetation cover in the sandy sites grouped by bioclimatic zones and sorted by level of grazing pressure (Fig. 10): abrupt drop in grass contribution was observed in site 4 in 1987 (Fig. 10a), in site 12 in 1985 and 2000 (Fig. 10b), in site 31 in 1990 (Fig. 10c), in site 30 and 38 in 1986, 37 in 2005 (Fig. 10d). The drops in the contribution of grass that followed the 1983–1984 droughts often benefited to dicotyledons which seeds are either dispersed by wind such as Farsetia ramosissima, C. senegalensis, A. lanata or have long dormancies in soils such as Colocynthis vulgaris. The drop observed at site 12 in 2001 followed the exhaustion of the seed stock due a local infestation by grain

Author's personal copy

122

P. Hiernaux et al. / Journal of Hydrology 375 (2009) 114–127

Rharous Rh

Gossi Go

Hombori Ho

y = 1.23 x + 91 R 2 = 0.17

y = 1.15 x + 54 R2 = 0.14

Boni Bo y = 4.23 x + 256 R 2 = 0.41

y = 1.70 x + 159 R 2 = 0.25

500 Herb mass anomalies % 400 300 200 100 0 -100 Rainfall anomalies % -200 -100

-80

-60

-40

0

-20

20

40

60

80

100

120

Fig. 8. Annual maximum standing mass relative anomaly (i.e. deviation from the mean) for sandy sites with moderate grazing plotted against annual rainfall anomaly for the meteorological stations of each group of site from north to south: Rharous, Gossi, Hombori and Boni respectively.

Yield anomaly %

Grazing pressure

20

moderate

high

intense

10 0

y = 8.21 x - 27.5 R2 = 0.97

-10 -20 -30 -40

Climate aridity north

centreN

centreS

South

Fig. 9. Mean impact of grazing intensity on herbaceous mass: mean yield anomaly in relation to the yield under light grazing over 1984–2006 % depending on position along the Sahel bioclimatic gradient in Gourma.

Table 4 Relative contribution (%) to annual maximum herbage bulk of the perennials, short cycle, legumes, grasses (+sedges) and plants with C4 type of photosynthesis: means (m) of site means and coefficient of variation (cv) over 1984–2006 calculated for group of sites defined by soil type and grazing intensity. Soil types

Sand Sand Shallow Clay All sites

Grazing intensity

Light Heavy Light Heavy

Number sites

7 8 6 4 25

Perennials (%)

Short cycle (%)

Legumes (%)

Grasses (%)

C4 (%)

m

cv

m

cv

m

cv

m

cv

m

cv

4.7 4.6 5.4 10.9

174.8 163.0 179.2 195.7

16.5 21.9 16.6 3.6

92.3 77.7 85.7 179.6

9.2 14.1 6.6 5.2

105.6 100.3 128.4 125.8

72.3 72.4 75.8 86.4

20.6 17.4 17.6 11.4

82.1 80.6 83.5 83.0

18.3 19.3 22.5 18.5

5.8

175.2

16.2

96.7

9.5

112.6

75.4

17.4

82.1

19.7

feeding sparrows, while that of site 25 in 2005 followed a wild fire. Among the progressive trends, is the increase of annual grasses that developed progressively after the first reaction to the major drought of 1983–1984 and favour short cycle grasses such as C. biflorus (as in site 17, Fig. 11) and T. berteronianus, progressively replaced by longer cycle grasses such as Aristida mutabilis (as in site 17, Fig. 11), S. gracilis or Diheteropogon hagerupii depending on bioclimatic zone. Grazing pressure during the wet season also has an impact on species composition. Dicotyledons such as the palatable Tribulus terrestris and Z. glochidiata (as in site 17, Fig. 11) or the refused Cassia tora and Sida cordifolia, are more frequently dominant under heavy grazing (Table 4). However, unless heavy grazing is accompanied by elevated manure deposition as in livestock resting or camping places, this dominance is temporary. Abrupt and short lasting changes in species composition occurs

whatever the grazing pressure as indicated by the changes in the contribution of grasses at site 37 (Fig. 10c) in spite of permanent heavy grazing pressure.

Discussion Is the yield response to rainfall of rangeland herbaceous linear? Although statistic relationships have been established between herbaceous yield and total annual rainfall in Sahel (e.g. Boudet, 1984; Breman and de Wit, 1983; Le Houérou et al., 1988) and are classically used to get yields estimates, the relations established at one site between herbaceous yield and rainfall are often poor (Tracol et al., 2006; Hiernaux et al., 2009a). Linear regressions be-

Author's personal copy

123

P. Hiernaux et al. / Journal of Hydrology 375 (2009) 114–127

a 100

Grass %

80 60 40 Sandy soils, northern Sahel 20 1

4

5

6

0

b 100

Grass %

80 60 40

Sandy soils, centre-north Sahel 20

10

12

14

0

c

100

Grass %

80 60 40 Sandy soils, centre-south Sahel 20 17

18

19

31

0

d

100

Grass %

80 60 40 Sandy soils, southern Sahel

20

25

30

37

38 2006

2005

2004

2003

2002

2001

2000

1999

1998

1995

1994

1993

1992

1991

1990

1989

1988

1987

1986

1985

1984

0

Fig. 10. Variation over years of the contribution (%) of grasses and sedges species to the herbaceous bulk in sandy soil rangelands (site code number in legends) along the Sahel bioclimatic gradient depending on pastoral status: light grazing (circle), moderate (triangle), high (square), intense (star): (a) northern, (b) centre-north, (c) centresouth, d) southern Sahel.

'Cenchrus biflorus'

80

Sandy soils

Contribution 'Aristida mutabilis'

to cover %

centre sahel

'Zornia glochidiata'

60

40

20

2006

2005

2004

2003

2002

2001

2000

1999

1998

1995

1994

1993

1992

1991

1990

1989

1988

1987

1986

1985

1984

0

Fig. 11. Inter annual dynamics the mean contribution to herbage bulk (%) of Cenchrus biflorus, Aristida mutabilis and Zornia glochidiata in all sandy soils sites in Gourma from 1984 to 2006.

Author's personal copy

124

P. Hiernaux et al. / Journal of Hydrology 375 (2009) 114–127

tween yield relative anomalies at a site and the corresponding rainfall relative anomalies at the closest meteorological station are weak, and just slightly better in the southern Sahel sandy soils (Fig. 8). There are several causes that may contribute to blur the relationship between rainfall and herbaceous production. Because the herbaceous layer is dominated by photoperiodic annual plants, yields depend very much on the temporal distribution of the rains that drive the regime of soil moisture. Indeed, at the onset of the rainy season, only a small fraction of rain water is used for annual plant production. Seed first germinate and then, the small size of the seedling leaves and the time required to build up the root system, limits above ground growth rates during the first part of the rainy season, till grasses start culm elongation (Grouzis, 1992). On the other end, rains falling later than fructification are not used for production by most annuals that wilt soon after fructification (case of the 45 mm rainfall on the 27th September 2005 at Agoufou, Fig. 3). Rapid growth, also named ‘linear growth’ (Penning de Vries and Djiteye, 1982), only lasts a few weeks from the start of stem elongation to fructification. The productive efficiency of rain water thus depends on the date of rainfall events. And the mean productive efficiency of the rain water over a season will depend on the rainfall distribution pattern. This would explain part of the poor relationship observed between yield and year total rainfall at a given site, as observed for instance in similar rangelands in Niger (Hiernaux et al., 2009a). Productivity of the herbaceous layer, especially during the ‘linear growth’ could also be limited by low plant density. The density of annual plants that results from germination and seedling development does generally not limit growth and production in Sahel (Cissé, 1986). However, it may happen, at least locally, that the number of seedling falls below a few hundred of plants per square meter and thus became a limitation to production (Hiernaux, 2000). These situations result from impoverishment of the seed stock either due to a failure of seed production on the previous year(s) as in several Gourma sites in 1984, or to the erosion of the top soil carrying the seed stock away though wind deflation or run-off, or else deep burying the seed stock through wind deposition or sedimentation. It can else be due to heavy consumption of the soil seed stock by birds (Gaston, 1976), rodents or insects, and or burning of the seeds before dispersion by early bushfire. Finally, low plant density could also result from exhaustion of the seed stock by a succession of aborted germination and seedling development events triggered by successive rain interruptions at the onset of the rainy season (Elberse and Breman, 1989; Elberse and Breman, 1990), for example the first cohort of germinations triggered by the early 10 mm rainfall the 4th May 2005 at Agoufou collapsed but did not exhausted the soil seed stock (Fig. 3). The productivity of the herbaceous layer may also be influenced by species composition, especially by the relative proportions of species of C3 and C4 photosynthesis types and relative proportion of short versus long cycle species (Penning de Vries and Djiteye, 1982). In the Sahel, C4 type largely dominates, reaching 82.1% of herbaceous bulk in average (Table 4), as most grasses and several dicotyledons are C4 plants that are more efficient producers at high temperature and sun illumination. C3 plants, mostly planophilous dicotyledons, close stomata and reduce photosynthetic activity at high temperature, and are more common either at shade of woody plants where they are protected from direct illumination or else in wetlands (Hanan et al., 1997). The influence of the C4/C3 composition on productivity is not conspicuous, because of this niche adaptation, and also interference with other plant physiological traits such as cycle length (Breman and de Ridder, 1991). The productivity of short cycle plants such as T. berteronianus (C4) or Z. glochidiata (C3) is constrained by the short duration of the ‘rapid growth’ period. However, short cycle plants only contribute to 16.2% of the herbaceous bulk in average (Table 4).

Finally, the productivity of annual herbaceous is also controlled by soil fertility, i.e., the provision of nutrients required by plants to grow, especially nitrogen, phosphorus and potassium (Penning de Vries and Djiteye, 1982). This provision is conditioned by soil clay and organic matter contents, soil pH, and is function of soil moisture regime (Breman and de Wit, 1983). In the Sahel, nitrogen and phosphorus are generally limiting yields by controlling root development and photosynthesis and thus the rate of growth during the ‘rapid growth’ as demonstrated by fertilisation trials (Hiernaux et al., 1995). When high grazing is not affecting too much the estimate herbaceous production in sites heavily manured by elevated livestock frequentation because species composition is dominated by poorly palatable species such as S. cordifolia in Diankabou (site 37), high yields are recorded resulting in the levelling of mean yield anomaly relative to ungrazed site (Fig. 9). Are herbaceous yield more variable towards the dry end of Sahel gradient? Because inter-annual variations in rainfall are larger, in relative terms, towards the drier end of the Sahel gradient, higher variations of the vegetation yields could be expected. A later start of the rainy season and an earlier end as observed in average when going north in latitude along the gradient (Frappart et al., 2009) could also contribute to increase the fraction of the rains poorly used for growth by herbaceous annuals to the benefit of perennials. Indeed, perennial herbaceous, especially grasses such as P. turgidum, A. sieberiana, Stipagrostis ssp, increase frequencies below the 200 mm isohyet, first settling in facieses that benefit less from run-on (such windward dune slope at Zaouati site 32, Table 3). Yield variability does increase as climate gets drier along the latitudinal gradient, at least on sandy soils (Fig. 7c). Yield variability also increases with latitude on shallow soil sites except the lower variability at Tiouaz site (#2), which ecology is somehow intergrade between shallow soils (sand deposit) and lowland clay soils (diatomite on fossil lacustrin flat). Yet, the yield variability on clay soil sites in low lands does not increase as climate get dryer as expected from sites in which the growth of herbaceous is driven by run-on and flood regime. Are herbaceous yields more variable as soils get shallow or less permeable? On shallow and poorly permeable soils, a large fraction of rain water runs off, either locally to the benefit of small flats or depressions, or to low lands or pools further away. The soil moisture regimes that result from this spatial redistribution of rain water are extremely contrasted and explain the patchiness of vegetation, extreme in the case of the ‘tiger bush’ where large extend of bare soil serve as impluvium for the narrow but dense linear thickets that develop perpendicular to the very gentle slope (Hiernaux and Gerard, 1999). The hypothesis along which spatial heterogeneity would buffer temporal variability is not verified. Indeed yield variability over years increases with the spatial heterogeneity of the herbaceous yield (Fig. 7a). This is verified for the shallow soils, with low yields but also applies, on a narrower range though, on temporarily flooded clay soils and permeable sandy soils independently of the grazing pressure (Fig. 7b). Is herbaceous layer suffering more and recovering less from drought when grazed? In sandy soils, yield anomalies of grazed sites relative to the yield of neighbour ungrazed or lightly grazed site only indicate losses in yield in the northern drier sites, while centre and southern Sahel sites appear less affected by grazing (Fig. 9). Short term

Author's personal copy

P. Hiernaux et al. / Journal of Hydrology 375 (2009) 114–127

effects of livestock grazing during wet season on herbaceous growth, mid term effect of grazing during wet and dry season on litter and soil surface, and long term effect of grazing history on species composition all contribute to this outcome. Grazing during the wet season affects plant growth and thus yield in direction and proportions that vary with the timing and intensity of grazing (Hiernaux and Turner, 1996). Early during growth, herbaceous plants respond to grazing by resprouting with enhanced tillering of grasses and branching of dicotyledons (Hiernaux et al., 1994). As a consequence, total yield at the end of the growing season may be slightly enhanced by early grazing. However, the ability to resprout declines as plants mature, and drop very fast when grazing cuts are repeated, resulting in a negative impact of repetitive cuts on total yields that could be finally reduced by up to half of the production compared to in ungrazed condition (Ayantunde et al., 1999). Yet, mid-term effect of grazing by soil trampling, accelerating conversion of standing straws into litter, enhancing litter decomposition, seed burying in top soil (Hiernaux et al., 1999) should globally enhance seedling density and herbage growth irrespective of the position along the bioclimate gradient (Hiernaux, 1998). Moreover, heavy grazing by reducing the risks of wild fire could indirectly enhance yield, more to the south where burning are more common than to the north. On the longer term, livestock grazing also alters yields indirectly through its influence on species composition. Indeed, selective grazing affects seed production, dispersion, conservation (burying in the soil) and germination (impact on seed dormancy, seed burying depth, litter cover, soil roughness and crusting). And long history of heavy grazing promotes two types of species either livestock refusal that tends to be long cycle, or else good quality feed species that tend to be short cycle (Table 4). The refusals such as S. cordifolia, dominant at Diankabou (site 37) are more common in South Sahel growing on soils which fertility is enriched by large amounts of dung deposition. The short cycle good quality feed such as the legumes Z. glochidiata, A. ovalifolius, other dicotyledons T. terrestris, Boehrravia repens or grasses C. biflorus, Dactyloctenium aegyptiacum, and T. berteronianus are common in grazed sites. Being short cycle these species tends to yield less explaining the negative anomalies observed with grazing, especially toward the arid end of the gradient (Fig. 9) while in the south of the gradient, long cycle unpalatable cohabit and alternate dominance with short cycle palatable species as observed on the Diankabou site (37). Livestock also contribute to seed dispersion, either by transport through fixation to the hair as observed with C. biflorus, T. terrestris Z. glochidiata, or by dispersion through faeces of hard coated seeds that once ingested, resist to digestion and passage through the animal guts (Danthu et al., 1996) as observed with Z. glochidiata and A. ovalifolius. The facilitation of seed dispersion or particular species by livestock and the homogenisation of the soil surface and the germination bed by trampling, contribute to homogenize species composition to the benefit of short cycle species (Miehe, 1998). Grazing thus scales down the patchiness of the herbaceous layer attenuating spatial heterogeneity (Table 2), but only lessens plant diversity at very high stocking rates (Hiernaux, 1998). Because the species favoured by grazing tend to better resist dry spells or droughts due to their short cycle and ability to tiller or branch profusely, grazed pasture tend to respond more rapidly to droughts than ungrazed pasture (Hiernaux and Diarra, 1993). The response to drought is however conditioned by the grazing pressure exerted while and just after the range is exposed to drought. If heavy grazing is maintained during the drought stress or just after, trampling may accelerate wind erosion processes, expending the setting of deflation patches and microdunes. As grazing, burning tends to homogenise species composition but favour species with small seeds that better germinate on hardened crusts such as the long cycle grasses S. gracilis and D.

125

hagerupii, or the short cycle dicotyledons Heliotropium strigosum and Portulaca foliosa. Is there a shift in species composition towards species more adapted to dryer climate? The soil seed stock of Sahel rangelands is largely transient, i.e., most of the seeds produced at the end of the growing season will germinate on the next year (Hérault and Hiernaux, 2004). As a consequence the similitude in species composition decreases strongly with the numbers of years between observations and explains fast successions of species composition observed in Sahel (Hiernaux, 1995). On another hand, the density of plants required for a species to be part of the dominant species at a site is relatively small (from a few to a few hundred per m2 depending on individual plant size). Together with the fast succession, this limited density requirement explains for the frequent and abrupt changes in dominant herbaceous species often observed in the Sahel, even when no obvious environmental change occurred (Breman and Cissé, 1977). The magnitude and pattern of changes in contribution of group of species (Fig. 10) or individual species (Fig. 11) recorded on the monitored sites are consistent with frequent and abrupt changes predicted by the non-equilibrium behaviour theory (Ellis and Swift, 1988). However, there are also overall trends such as the strong increase in C. biflorus that first followed the drought from 1984 to 1992 (Fig. 11), followed by a progressive decline while legumes such as Z. glochidiata, or longer cycle grasses such as A. mutabilis increased progressively their contribution. The first wave of pioneer species composed of dicotyledons such as C. senegalensis, C. vulgare, F. ramosissima and short cycle grasses such as C. biflorus, T. berteronianus, could be interpreted as a shift towards more arid tolerant flora because most of these pioneers are common species in the transition zone with Sahara. However, the second wave could be interpreted as a slow return to typical sahelian flora. Conclusion Soil texture and topography largely determine the average yield of the herbaceous layer of Sahel rangelands. They also determine its spatial heterogeneity assessed by the coefficient of variation of the mean yield calculated from hundred 1 m2 random virtual plots. The means of these standard deviations calculated over 1984–2006 range from 295.1% in shallow soil, to 86.0% in clay soils, 79.5% in lightly grazed sandy soils and 76.6% in heavily grazed sandy soils, with an overall mean of 124.5%. Inter-annual yield variability is of same magnitude than spatial heterogeneity, except for shallow soils where spatial heterogeneity is much larger. In any cases, temporal variability increases as mean spatial heterogeneity gets larger following a power function. Yet, the hypothesis along which spatial heterogeneity would buffer temporal variability is not verified. Moreover on sandy soils, yield variability increases as climate gets dryer as indicated by the regression established between the variability indicator and the latitudinal position of the sites. Similar regressions poorly apply in shallow soils and do not apply in clay soils. Indeed, herbage growth in sandy soils is driven by rainfall volume and distribution. However, the linear regressions between yield relative anomalies at a site and the corresponding rainfall relative anomalies are weak and just slightly better in southern Sahel. Among the causes for these poor relationship the impact of grazing under which negative yield anomalies were found in northern and drier Sahel, while yield anomalies were nil in centre and slightly positive in southern sites. This selective impact is in turn explained by grazing favouring either short-cycle good quality feed species or else long cycle refusal, the promotion of these highly productive refusal being more common in southern Sahel. In agreement with the non-equilibrium

Author's personal copy

126

P. Hiernaux et al. / Journal of Hydrology 375 (2009) 114–127

theory, changes in species composition monitored in Gourma rangelands are often abrupt and short term, but some are more progressive and define trends that could be interpreted as a shift of more arid tolerant species following the 1983–1984 droughts, till 1992, followed by a slow return to typical Sahelian flora. Acknowledgments This research was funded by INSU and CESBIO in the framework of the AMMA program. AMMA is an international research program funded by a large number of agencies, especially from France, UK, USA and Africa. It benefits from a major financial contribution from the European Community’ Sixth Framework Research Program (AMMA-EU). Detailed information on scientific coordination and funding is available on the AMMA International web site: http://www.amma-international.org. Special thanks to INSU for the grant allocated to the first author of this paper. We would like to acknowledge the major contribution of the IRD and ILCA staff in Mali in the data collection in Gourma over many years and warmly thanks our hosts in Hombori, Gossi and Gourma Rharous. References Ayantunde, A.A., Hiernaux, P., Fernandez-Rivera, S., van Keulen, H., Udo, H.M.J., 1999. Selective grazing by cattle on spatially and seasonally heterogeneous rangelands in the Sahel. J. Arid. Envir. 42, 261–279. Bille, J.C. (Ed.), 1977. Etude de la production primaire nette d’un écosystème sahélien. Orstom Editions, Trav. et Doc, Paris. vol. 65, 82 p. Bille, J.C., 1992. Tendances évolutives comparées des parcours d’Afrique de l’Ouest et d’Afrique de l’Est. In: Le Floc’h, E., Grouzis, M., Cornet, A., Bille, J.C. (Eds.), L’aridité une contrainte au développement. Orstom Editions, Paris, pp. 179–195. Boudet, G., 1972. Désertification de l’Afrique tropicale sèche. Adansonia, ser. 2 12 (4), 505–524. Boudet, G., 1977. Désertification ou remontée biologique au Sahel. Cahiers Orstom, série biologie 12, 293–300. Boudet, G., 1979. Quelques observations sur les fluctuations du couvert végétal sahélien au Gourma Malien et leurs conséquences pour une stratégie de gestion sylvo-pastorale. Bois et Forêts des Trop. 184, 31–44. Boudet, G., 1984. L’exploitation des parcours et la conduite des troupeaux dans les systèmes d’élevage. Cahiers de la Recherche Développement, 97–101. Boudet, G., Cortin, A., Macher, H., 1971. Esquisse pastorale et esquisse de la transhumance de la région du Gourma (Mali). DIWI, Essen, Germany. 120 p. Breman, H., Cissé, A.M., 1977. Dynamics of sahelian pastures in relation to drought and grazing. Oecologia 28, 301–315. Breman, H., de Ridder, N., 1991. Manuel sur les pâturages des pays sahéliens. Karthala, Paris. 485 p. Breman, H., De Wit, C.T., 1983. Rangeland productivity and exploitation in the Sahel. Science 221, 1341–1347. Casenave, A., Valentin, C., 1989. Les états de surface de la zone sahélienne. Influence sur l’infiltration. Orstom Editions, coll. Didactiques, Paris. 229 p. Cissé, A.M., 1986. Dynamique de la strate herbacée des pâturages de la zone sudsahélienne. Ph.D. Thesis, Agriculture Univ., Wageningen, 211 p. Cook, C.W., Stubbendieck, J. (Eds.), 1986. Range research: basic problems and techniques. Soc. For Range Management. Denver, Colorado, p. 317 p. Cornet, A., 1981. Mesure de biomasse et détermination de la production nette aérienne de la strate herbacée dans trois groupements végétaux de la zone sahélienne au Sénégal. Acta Oecologica, Oecol. Plant. 2 (3), 251–266. Cottam, G., Curtis, J.T., 1956. The use of distance measures in phytosociological sampling. Ecology 37, 451–460. Coulibaly, A., 1979. Approche phyto-écologique et phytosociologique des pâturages sahéliens au Mali (Région du Gourma). Ph.D. Thesis, Univ. of Nice, 180 p. Danthu, P., Ickowicz, A., Friot, D., Manga, D., Sarr, A., 1996. Effet du passage par le tractus digestif des ruminants domestiques sur la germination des graines de légumineuses ligneuses des zones tropicales sèches. Rev. Elev. Méd. vét. Pays trop. 2 (3), 235–242. De Leeuw, P.N., Diarra, L., Hiernaux, P., 1992. An analysis of feed demand and supply for pastoral livestock in the Gourma region of Mali. In: Behnke, R.H., Jr., Scoones, I., Kerven, C. (Eds.), Range Ecology at Disequilibrium. ODI, London, pp. 136–152. Elberse, W.T., Breman, H., 1989. Germination and establishment of Sahelian rangeland species. I seed properties. Oecologia 80, 477–484. Elberse, W.T., Breman, H., 1990. Germination and establishment of Sahelian rangeland species. II effects of water availability. Oecologia 85, 32–40. Ellis, J.E., Swift, D.M., 1988. Stability of pastoral ecosystems: alternate paradigms and implications for development. J. Range Manage. 41, 450–459. Frappart, F., Mougin, E., Guichard, F., Kergoat, L., Hiernaux, P., Arjounin, M., Lavenu, F., Lebel, T., 2009. Rainfall regime over the Sahelian climate gradient in the Gourma. J. Hydrol. 375 (1–2), 128–142. Gallais, J., 1975. Paysans et pasteurs du Gourma. La condition sahélienne. CNRS, Paris. 239 p.

Gaston, A., 1976. Travaux écologiques en relation avec la lutte contre Quela Quela, Inst. d’Elev. et de Méd. Vét. des pays Trop. Maisons-Alfort. 203 p. Grouzis, M., 1988. Structure, productivité, et dynamique des systèmes écologiques Sahéliens (Mare d’Oursi, Burkina Faso). Orstom Editions, Paris (Coll. Etudes et Thèses, 318 p.). Grouzis, M. (Ed.), 1992. Germination et établissement des plantes sahéliennesLe Floc’h, E., Grouzis, M., Cornet, A., Bille, J.C. (Eds.). L’aridité, une contrainte au développement. Orstom Editions, coll. Didactiques, Paris, pp. 267–282. Hanan, N.P., Prince, S.D., Bégué, A., 1997. Modelling vegetation primary production during HAPEX-Sahel using production efficiency and canopy conductance model formulation. J. Hydrol. 188–189, 651–675. Hein, L., de Ridder, N., 2006. Desertification in the Sahel: a reinterpretation. Global Change Biol. 12, 751–758. Hérault, 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, 1–9. Hiernaux, P., 1995. Spatial heterogeneity in Sahelian rangelands and resilience to drought and grazing. In: West, N.E. (Ed.), Rangelands in a Sustainable Biosphere. Proceedings of the Fifth International Rangeland Congress, vol. 2, pp. 232–233. Hiernaux, P., 1996. The crisis of sahelian pastoralism: ecological or economic? Overseas Develop. Inst., Pastoral Development Network, 39a, 20 p. Hiernaux, P., 1998. Effects of grazing on plant species composition and spatial distribution in rangelands of the Sahel. Plant Ecol 138, 191–202. Hiernaux, P., 2000. Implications of the ‘new rangeland paradigm’ for natural resource management. In: Adriansen, H., Reenberg, A., Nielsen, I. (Eds.), The Sahel: energy supply, economic pillars of rural Sahelian communities, need for revised development strategies. Serein, Copenhagen, pp. 113–142. Occ. Paper 11. Hiernaux, P., Diarra, L., 1993. Dynamique de la végétation des terres de parcours sahéliennes. Un bilan du suivi des sites pastoraux du Gourma en 1992. ILCA, Addis Ababa (Doc. Progr. 1/93, 46 p.). Hiernaux, P., Gerard, B., 1999. The influence of vegetation pattern on the productivity, diversity and stability of vegetation: the case of ‘brousse tigrée’ in the Sahel. Acta ecologica 20, 147–158. Hiernaux, P., Le Houérou, H.N., 2006. Les parcours du Sahel. Sècheresse 17 (1–2), 51–71. Hiernaux, P., Turner, M.D., 1996. The effect of clipping on growth and nutrient uptake of Sahelian annual rangelands. J. Appl. Ecol. 33, 387–399. Hiernaux, P., Cissé, M.I., Diarra, L., 1984. Bilan d’une saison des pluies 1984 très déficitaire dans le Gourma (Sahel malien). Première campagne de suivi et télédétection expérimentale. ILCA, Addis Ababa (Doc. Prog. 140, 44 p.). Hiernaux, P., Diarra, L., de Leeuw, P.N., 1994. Modelling tillering of annual grasses as a function of plant density: application to sahelian rangelands productivity and dynamics. Agric. Syst. 46, 121–139. Hiernaux, P., de Leeuw, P.N., Diarra, L., 1995. The interactive effects of rainfall, nutrient supply and defoliation on the herbage yields of sahelian rangelands in north–east Mali. In: Powell, J.M., Fernandez-Rivera S., Williams, T.O., Renard, C. (Eds.), Livestock and Sustainable Nutrient Cycling in Mixed Farming Systems of Sub-Saharan Africa. Proceedings of an International Conference on Livestock and Sustainable Nutrient Cycling in Mixed Farming Systems of Sub-Saharan Africa held in Addis Ababa, Ethiopia, 22–26.11.93. ILCA, Addis Ababa, Ethiopia, vol. 2, pp. 337–352. Hiernaux, P., Bielders, C.L., Valentin, C., Bationo, A., Fernández-Rivera, S., 1999. Effects of livestock grazing on physical and chemical properties of sandy soils in Sahelian rangelands. J. Arid Envir. 41 (3), 231–245. Hiernaux, P., Ayantunde, A.A., Gérard, B., Grippa, M., Mougin, E., 2009a. Resilience and productivity trends of crops, fallows and rangelands in Southwest Niger: impact of land use, management and climate changes. J. Hydrol. 375 (1–2), 65–77. Hiernaux, P., Diarra, L., Trichon, V., Mougin, E., Soumaguel, N., Baup, F., 2009b. Woody plant population dynamics in response to climate changes from 1984 to 2006 in Sahel (Gourma, Mali). J. Hydrol. 375 (1–2), 103–113. Hutchinson, J., Dalziel, J.M., 1954–1972. Flora of West Tropical Africa, second ed. In: Keay, R.W.J., Hepper, F.N. (Eds.), Crown Agents for Overseas Gov. and Admin., London, vol. 1–3, 828, 544, 574 p. Kammerud, T.A., 1996. Soil impact on satellite based vegetation monitoring in Sahelian Mali. Geogr. Ann 78 (A4), 247–259. Le Houérou, H.N., Bingham, R.L., Skerbek, W., 1988. Relationship between the variability of primary production and the variability of annual precipitation in world arid lands. J. Arid Envir. 15, 1–18. Leprun, J.C., 1992. Etude de quelques brousses trigrées sahéliennes: structure, dynamique, écologie. In: Le Floc’h, E., Grouzis, M., Cornet, A., Bille, J.C. (Eds.), L’aridité une contrainte au développement. Orstom Editions, coll. Didactiques, Paris, pp. 221–244. Miehe, S., 1998. Inventaire et suivi de la végétation dans le périmètre expérimental à Windou Thiengoly dans le cadre du Projet Sénégalo-Allemand d’autopromotion pastorale dans le Ferlo (PAPF), Saint-Louis (Sénégal), G.T.Z., 49 p. Mougin, E., Hiernaux, P., Kergoat, L., De Rosnay, P., Grippa, M., Timouk, F., Arjounin, M., Le Dantec, V., Demarez, V., Ceschia, E., Mougenot, B., Baup, F., Frappart, F., Gruhier, C., Guichard, F., Trichon, V., Soumaguel, N., Diarra, L., Dembélé, F., Soumaré, A., Kouyaté, M., Lloyd, C., Seghieri, J., Hanan, N.P., Galy-Lacaux, C., Damsin, C., Epron, D., Delon, C., Serça, D., Mazzega, P., Jarlan, L., Frison, P.L., Tracol, Y., 2009. The AMMA Gourma observatory site: overview of research activities. J. Hydrol. 375 (1–2), 14–33.

Author's personal copy

P. Hiernaux et al. / Journal of Hydrology 375 (2009) 114–127 Nicholson, S.E., 2001. Climate and environmental change in Africa during the last two centuries. Clim. Res. 14, 123–144. Olsson, L., Eklundh, L., Ardö, J., 2005. A recent greening of the Sahel –trends, patterns and potential causes. J. Arid Envir. 63, 556–566. Penning de Vries, F.W.T., Djiteye, M.A. (Eds.), 1982. La productivité des pâturages sahéliens, une étude des sols, des végétations et de l’exploitation de cette ressource naturelle. Pudoc, Wageningen (Agric. Res. Rep, 525 p.). Prince, S.D., 1991. Satellite remote sensing of primary production: comparison of results for Sahelian grasslands 1981–1988. Int. J. Remote Sensing 12 (6), 1301– 1311. Redelsperger, J.L., Thorncroft, C., Diedhiou, A., Lebel, T., 2006. African Monsoon Multidisciplinary Analysis (AMMA): an international research project and field campaign. Bul. Amer. Meteo. Soc. 87 (12), 1739–1746. RIM, 1987. Refuge in the Sahel. Resource Inventory and Management Limited, St Helier, Jersey. 115 p.

127

Schlecht, E., Hiernaux, P., Turner, M.D., 2001. Mobilité régionale du bétail: nécessité et alternatives? Elevage et gestion de parcours au Sahel, implications pour le développement. In: Tielkes, E., Schlecht, E., Hiernaux, P. (Eds.). Verlag Grauer, Beuren-Stuttgart, pp. 65–77. Schlecht, E., Hiernaux, P., Kadaouré, I., Hülsebusch, C., Mahler, F., 2006. A spatiotemporal analysis of forage availability and grazing and excretion behaviour of herded and free grazing cattle, sheep and goats in Western Niger. Agric., Ecosys. Envir. 113, 226–242. Tracol, Y., Mougin, E., Jarlan, L., Hiernaux, P., 2006. Testing a sahelian grassland functioning model against herbage mass measurements. Ecol. Model. 193, 437– 446. Turner, M.D., Hiernaux, P., Schlecht, E., 2005. The distribution of grazing pressure in relation to vegetation resources in semi-arid West-Africa: the role of herding. Ecosys. 8 (6), 668–681.