Journal of Hydrology 375 (2009) 190–203

Contents lists available at ScienceDirect

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

Towards an understanding of coupled physical and biological processes in the cultivated Sahel – 2. Vegetation and carbon dynamics N. Boulain a, B. Cappelaere a,*, D. Ramier a, H.B.A. Issoufou b, O. Halilou b, J. Seghieri a, F. Guillemin a, M. Oï a, J. Gignoux c, F. Timouk d a

IRD/HydroSciences, BP 64501, 34394 Montpellier Cedex 5, France Université Abdou Moumouni, B.P. 11040 Niamey, Niger c CNRS/Bioemco, Ecole Normale Supérieure, 46 rue d’Ulm, 75230 Paris Cedex 05, France d IRD/CESBIO, 18 Avenue, Edouard Belin, bpi 2801, 31401 Toulouse Cedex 9, France b

a r t i c l e Keywords: Assimilation Respiration Fallow savanna Millet West Africa Carbon dioxide

i n f o

s u m m a r y This paper analyses the dynamics of vegetation and carbon during the West African monsoon season, for millet crop and fallow vegetation covers in the cultivated area of the Sahel. Comparing these two dominant land cover types informs on the impact of cultivation on productivity and carbon fluxes. Biomass, leaf area index (LAI) and carbon fluxes were monitored over a 2-year period for these two vegetation systems in the Wankama catchment of the AMMA (African monsoon multidisciplinary analyses) experimental super-site in West Niger. Carbon fluxes and water use efficiency observed at the field scale are confronted with ecophysiological measurements (photosynthetic response to light, and relation of water use efficiency to air humidity) made at the leaf scale for the dominant plant species in the two vegetation systems. The two rainy seasons monitored were dissimilar with respect to rain patterns, reflecting some of the interannual variability. Distinct responses in vegetation development and in carbon dynamics were observed between the two vegetation systems. Vegetation development in the fallow was found to depend more on rainfall distribution along the season than on its starting date. A quite opposite behaviour was observed for the crop vegetation: the date of first rain appears as a principal factor of millet growth. Carbon flux exchanges were well correlated to vegetation development. High responses of photosynthesis to light were observed for the dominant herbaceous and shrub species of the fallow at the leaf and field scales. Millet showed high response at the leaf scale, but a much lesser response at the field scale. This pattern, also observed for water use efficiency, is to be related to the low density of the millet cover. A simple LAI-based model for scaling up the photosynthetic response from leaf to field scale was found quite successful for the fallow, but was less conclusive for the crop, due to spatial variability of LAI. Time/space variations in leaf distribution for the dominant species are key to scale transition of carbon dynamics. Results obtained for the two vegetation covers are important in light of the major land use/ cover change experienced in the Sahel region due to extensive savanna clearing for food production. Ó 2008 Elsevier B.V. All rights reserved.

Introduction In the context of global change, field quantification of carbon fluxes is important to identify areas of carbon emission or storage at the global scale, and to calibrate/validate models of carbon dynamics from local to global scale. During the last 15 years, different programs have been set up to monitor carbon exchanges for the major ecosystems. In particular, the Fluxnet observation network (Gu and Baldocchi, 2002) aims at the worldwide monitoring of water vapour and carbon fluxes. Yet this network includes very few stations in Africa and none in West Africa. It is crucial to obtain substantial observation series of carbon and water fluxes in this * Corresponding author. Tel.: +33 4 67 14 90 17; fax: +33 4 67 14 47 74. E-mail address: [email protected] (B. Cappelaere). 0022-1694/$ - see front matter Ó 2008 Elsevier B.V. All rights reserved. doi:10.1016/j.jhydrol.2008.11.045

vast region, together with the other environmental variables related to vegetation and climate, governed by the West African monsoon system. Especially in the semi-arid Sahel, rainfall distribution is characterized by marked spatial and temporal variability (Le Barbe and Lebel, 1997; Lebel et al., 1997), resulting in strong heterogeneity in the response of the vegetation cover, particularly that of crops (Boulain et al., 2006). These regions are currently undergoing a considerable increase in population (around 3% per year, Raynaut, 2001) which, combined with low crop productivity, induces generalized clearing of the natural vegetation (Cappelaere et al., 2009; Leblanc et al., 2008). Basic food crops (millet, sorghum) are strongly limited by the scarcity of soil nutrients and the distribution of available water (Badejo, 1998; de Rouw, 2004). Biological activity is closely dependent on the rainy season, which lasts – at best – for only one third of the year (June–September). Climatic

N. Boulain et al. / Journal of Hydrology 375 (2009) 190–203

conditions are extremely unfavorable for growth during another third of the year (February–May). Finally, from October to January, biological activity depends on the characteristics of the preceding rainy season. The Sahelian land surface mostly consists of a patchwork of land units such as crop fields, fallow bush (last cultivated a few months to several years back) or tiger bush (on plateaus). Accounting for this heterogeneity in the surface response and feedbacks to regional climate requires that individual land unit responses to climate variability be fully understood at the local scale. Furthermore, estimating the differences in carbon assimilation or water cycling capabilities between land cover types is key to assessing the impacts of the considerable expansion of cropland. Hence, the objective of the local-scale study presented in this paper, together with the companion paper (Ramier et al., 2009), was to analyse the functioning of a crop system and of a semi-natural, fallow system, submitted to strong climatic and human pressure, based on plantscale and field-scale observations. As part of the AMMA-Catch programme (Lebel et al., 2009), millet and fallow plots were monitored in a small endoreic catchment typical of the Sahelian region of West Niger (Cappelaere et al., 2009), over two contrasted rainy seasons (2005 and 2006). For each land cover type (fallow and millet), carbon and water fluxes were measured by the eddy covariance (EC) method, and then analyzed in relation to seasonal changes in plant biomass and leaf area index (LAI) and in micro-meteorological variables (air humidity, downwelling radiation). The HAPEX–Sahel experiment (Goutorbe et al., 1997) provided initial results relative to carbon and water fluxes linked to vegetation and climate in the same region (Friborg et al., 1997; Hanan et al., 1998; Levy et al., 1997; Lloyd et al., 1997; Moncrieff et al., 1997; Monteny et al., 1997). However, these results only covered a fraction of a single growing season. Our experiment covered two complete rainy seasons for the two major land cover types, and used more accurate instruments (e.g., an openpath gas analyzer instead of a krypton analyzer). Ramier et al. (2009) discuss the differences between the two contrasted rainy seasons and the two main land cover types from the point of view of meteorological variables and of the energy and water cycles. They show that evapotranspiration measured in the fallow was higher than in the millet field. In this paper, the responses of the two cover types to variable rainfall distribution, as regards vegetation development and carbon fluxes, are analyzed. The two cover types react differently to a late rainy season beginning or to prolonged dry spells during the rainy season. Response curves of assimilation to light, and of water use efficiency to air humidity deficit, are presented at leaf scale for the three dominant species in the grass, shrub, and crop layers, respectively, and are compared to responses at the cover scale. Despite a larger assimilation capability at leaf scale, millet crop shows lower assimilation than fallow at the cover scale, largely due to sowing density. The effectiveness of leaf density in accounting for the scale transition is investigated for each cover type, via a simple LAI-based model.

Materials and methods All plant names used in this paper are taken from the International Plant Name Index (http://www.ipni.org/index.html). Study area The study area is located in the Wankama catchment (Cappelaere et al., 2009), 60 km east of Niamey, Niger. The Wankama catchment is part of the AMMA-Niger observatory, one of three mesosites along the latitudinal West-African eco-climate gradient (Lebel

191

et al., 2009). Mean annual rainfall was 479 mm/year for the 1990– 2006 period (Ramier et al., 2009). The soil is sandy (>90% sand) and poor in nutrients (Rockström and de Rouw, 1997). In 2005, Millet (Pennisetum glaucum (L.) R. Br.) and fallow were covering 58% and 23% of the surface area, respectively; the remainder corresponding to degraded vegetation or bare soil (Boulain et al., accepted for publication). Millet is well adapted to the Sahelian climate, in particular thanks to its resistance to periods of drought (Do et al., 1996; Hamon, 1993; Winkel and Do, 1992). Millet is sowed after the first rainfall that is able to moisten the top 5 cm of soil (i.e., 5–10 mm rainfall). It may then need to be sowed several times again up to mid-July depending on rainfall distribution. Non-photosensitive traditional varieties are used at the site. Millet seeds are sowed in pockets (handfuls) with a density of around 10,000 pockets per hectare. Millet is cultivated in the traditional way with very little fertilizer (urea pellets may be added when sowing) and no irrigation. Shrubs and weeds are manually eliminated before the rain season, with no fire, some trees being preserved along field edges. Hand-hoe weeding may be repeated once or twice in the season, but some grass often persists. Mature millet is 2–3 m high at the study site. Harvest is done shortly after the end of the rain season, in September or October, removing the whole plant (ear and stem) from the field, and leaving bare soil through the dry season. Most fallow fields are no more than 5-year old, which is now typical of the Niamey region, and generally include a sparse shrub layer and an herbaceous layer. The average height of the shrub layer is 2 m, against 0.6 m for the grass layer. The shrub layer is essentially composed of Guiera senegalensis Lam, in almost singlespecies population, and of some Combretum micranthum G. Don. Shrub density averages around 700 individuals per hectare. The herbaceous layer is interspersed with bare soil patches, and is mainly composed of annual plants, namely C3 species at the beginning of the rain season, and a mixture of C3 and C4 species at the end of the season. The dominant species in the herbaceous layer was Zornia glochidiata Rchb. ex DC (C3) in 2005 and 2006, mainly accompanied by Mitracarpus scaber Zucc. (C3) and Cenchrus biflorus Roxb. (C4). Other species were present but less widespread, for example Aristida mutabilis Trin. & Rupr, Triumfetta pentandra A. Rich. and Andropogon gayanus Kunth. Small herds of zebus and flocks of sheep occasionally cross and graze the fallow. A few trees are growing within or at the boundaries of the fallow fields, mostly Combretum glutinosum Engl, Piliostigma reticulatum (DC.) Hochst. and Faidherbia albida (Delile) A. Chev. Wood from fallows now represents the main energy supply for the local population. Experimental design Two EC flux stations (black and white circles in Fig. 1a) were installed in June 2005 in a 5-year old fallow (Fig. 1b) and in a millet field (Fig. 1c), both of about 15 ha, at a distance of 570 m. For each cover type, vegetation variables (biomass and leaf area index) were monitored in four 50  50 m plots scattered over the catchment, denoted as M1–M4 for millet and as F1–F4 for fallow (Fig. 1a; M2 and F2 since 2006 only). The plots were chosen to be representative of the spatial heterogeneity found in millet and fallow fields, in terms of bare soil fraction and shrub density. For the fallow, analysis of vegetation variables in this paper is focused on the annual species-dominated herbaceous layer, the dynamics of which is more pronounced and better related to the rainy season’s characteristics than that of perennial shrubs, which reflects previous rain seasons as well as the current one. Biomass monitoring To monitor total aboveground biomass of the herbaceous layer in the fallow plots, samples were taken once every two weeks at a

192

N. Boulain et al. / Journal of Hydrology 375 (2009) 190–203

Fig. 1. Site description. (a): Map of Wankama watershed, including location of study plots (white squares M1–M4 correspond to 50  50 m millet plots, black squares F1–F4 to 50  50 m fallow plots, and white and black circles correspond to millet and fallow EC stations, respectively); (b) and (c): aerial views of fallow and millet fields, end of September 2006.

rate of three samples per plot, each sample being defined by a randomly-thrown 1  1 m metal quadrat. The plant cover at one location could not be sampled twice in a row. The entire aboveground herbaceous biomass was harvested. Aboveground millet biomass was monitored in the millet plots by sampling five pockets per plot, chosen randomly once every two weeks. In the laboratory, the samples were dried at 70 °C for at least 3 days. After drying, below-100 g samples were weighed on an Ohaus Scout Pro balance (1 mg precision) and above-100 g samples were weighed on a Stratorius balance (0.1 g precision). Leaf area index (LAI) The protocol used to monitor LAI was that defined in the VALERI project (validation of land European remote sensing instruments, http://www.avignon.inra.fr/valeri/). It was performed every two weeks through the 2005 and 2006 seasons in the 50  50 m fallow and millet plots. For each vegetation plot, the protocol consisted in taking a series of hemispherical photographs at 13 locations in a 20  20 m square. The hemispherical photographs were taken with a 35 mm film camera (Canon EOS 300v, Kodak Elitechrome 400 slide film) equipped with a fish-eye lens (Fisheye sigma 8 mm F4). The photographs taken in 2005 were then digitized with a scanner (Canon’s CanoScan 8600F). In 2006, the photographs were digitized by the laboratory that developed the slide films. The digital images were analyzed using the CAN-EYE (version 5) software developed by INRA, Avignon, France. A description of the analysis,

performance and limits of the CAN-EYE software can be found in Demarez et al. (2008). Carbon flux, soil water content and evapotranspiration Turbulent CO2 and water fluxes were measured at the field scale by the eddy covariance method, using an open-path infra-red gas analyzer (Licor 7500) coupled with a 3D sonic anemometer (Campbell CSAT-3), at 4.95 m and 5.10 m above the ground for the fallow and millet EC stations, respectively. Fourcomponent radiation is measured by a Kipp–Zonen CNR1 radiometer. The photosynthetically-active radiation (PAR) was calculated as a fraction (0.48) of the incoming solar radiation (Bégué et al., 1991; Frouin and Pinker, 1995). Soil moisture content was measured by time-domain reflectometry (six Campbell CS616 probes per site) to a depth of 275 cm. EC data was processed, at 30-min timestep, with the EdiRe software (Version 1.4.3.1167, R. Clement, University of Edinburgh), based on CarboEurope recommendations (Mauder and Foken, 2004), including despiking, double rotation, cross-correlation for derivation of time lag between the sonic anemometer and the gas analyser, spectral corrections, WPL correction, and atmospheric stability test. Data during rainstorms were removed, no gap-filling was performed, and u*-filtering (u* > 0.10) was applied at nighttime (Merbold et al., 2008). The WPL correction had a very significant effect on estimated CO2 fluxes at the beginning of the monsoon season. No correction was made for analyzer temperature, as the

N. Boulain et al. / Journal of Hydrology 375 (2009) 190–203

high air temperature at the site makes it of lesser importance. A weekly cleaning frequency allowed to keep the problem of optical contamination of flux estimates to a minimum. More information on the experimental setup and data processing can be found in Ramier et al. (2009). The algebraic sum of day and night fluxes yields the net carbon exchange of the system (Baldocchi, 1994). Nighttime fluxes correspond to the respiration of the system (soil and vegetation), and daytime fluxes correspond to the net assimilation flux (photosynthesis minus plant and soil respiration). In agreement with the meteorologists’ convention, fluxes are taken positive when oriented from the surface to the atmosphere, and vice-versa. Photosynthesis thus results in a negative flux, and respiration (plant and soil respiration) in a positive flux. Daytime fluxes are negative (photosynthesis dominates) or close to zero, and nighttime fluxes are positive (respiration). Using half-hourly data, we calculated the water use efficiency (WUE) of evapotranspired water at the scale of the cover (hence denoted WUEcover), as the ratio of carbon to evapotranspiration fluxes (denoted Fc and E, respectively, in lmol/m2/s) (Scanlon and Albertson, 2004)

WUEcover ¼ F c =E

ð1Þ

As WUE is sensitive to low radiation (Verhoef et al., 1996), we were only able to calculate it for values of downwelling short wave radiation above 300 W/m2. As shown by Monteny et al. (1997) for a Sahelian fallow savanna, there is a link between the Priestley–Taylor coefficient a (where a = E/E0, E is evapotranspiration and E0 is the equilibrium evaporation, defined as the limit reached when unsaturated air is in contact with a wet surface over a long fetch, see Slatyer and McIlroy, 1961), and the S/Smax ratio (where S is the volumetric soil moisture and Smax is the maximum observed S after free drainage). Above a certain threshold of S/Smax, which Monteny et al. (1997) found to be 0.7, the coefficient a tends to be constant, implying that soil water is not limiting and that evapotranspiration is thus climatecontrolled. Below the threshold, a decreases with decreasing S/ Smax, indicating that soil water controls evapotranspiration. Using a fixed soil depth of 275 cm, to account for all millet roots (Rockström et al., 1998) and for most of the fallow’s herbaceous and shrubs roots, and an Smax value equal to the maximum moisture recorded in this layer over the 2-year period (rainy days excluded) at a given site, we similarly found (not shown here) that the S/Smax threshold of 0.7 discriminated best the two above E/E0 domains. Hence this soil moisture threshold is used in this study as a simple indicator of the existence or not of root water stress of the vegetation, homogeneously for the two cover types and the two seasons. The Smax values, of 7.7% and 11.3% for the fallow and the millet field, respectively, are in agreement with the observations of Cuenca et al. (1997) for these land cover types in the same environment. Response curve at leaf scale Leaf gas exchanges were measured with a LI-COR 6400 analyser (LI-COR Biosciences, Inc.). This instrument enables field measurements of gas exchanges and photosynthesis while controlling parameters such as CO2 concentration, light intensity, temperature, or air humidity. The results presented in this paper come from nine P. glaucum leaves sampled in a field near Niamey (with sandy soil similar to the Wankama catchment) in 2001, 12 G. senegalensis leaves sampled in the Wankama catchment in 2001, and five Z. glochidiata leaves sampled in the catchment in 2006. For each species, samples mixed sun-exposed and shaded leaves (shading by another leaf of the same plant or by the neighbouring plant cover), to obtain a range of responses that are representative of prevailing environmental conditions. Samples also mixed young and mature leaves.

193

In this paper we present the response of assimilation to light (PAR) and the response of WUE to the deficit in air vapour pressure (VPD), for the three dominant species in the two systems studied. Like for the whole vegetation cover, net assimilation and transpiration data (A and tr, in lmol/m2/s, respectively) enable calculation of water use efficiency at the leaf scale (denoted WUEleaf)

WUEleaf ¼ A=tr

ð2Þ

Scaling the response of assimilation to light Based on the data obtained at the leaf and the cover scales, an attempt was made at scaling the light-to-assimilation relationship via LAI, using a simple, linear model. For the fallow, the scaling model was written as

F c ¼ LAIH  AH þ LAIS  AS  Rsoil

ð3Þ

where Fc is the assimilation at the cover scale; AH and AS are the leaf-scale assimilation for the dominant species in the herbaceous and shrub layers, respectively (namely Z. glochidiata and G. senegalensis); LAIH and LAIS are the LAI in the herbaceous and shrub layers; Rsoil is the soil respiration. Similarly, for a millet field, the model writes

F c ¼ LAIM  AM  Rsoil

ð4Þ

where AM and LAIM are the leaf-scale assimilation of millet and the LAI in the millet field, respectively. These models were applied at the half-hourly time step through the 2006 rain season, for conditions of no water stress defined as: S/Smax > 0.7 and VPD < 1.5 hPa. The observed PAR was transformed into a leaf-scale assimilation A for a given species using a regression curve fitted to the measured leaf-scale light-assimilation data (see ‘‘Response curve at leaf scale”). To estimate LAI at a given date, a time curve was fitted to the LAI values measured in the 50  50 m plots located south-west of the millet and fallow EC stations (corresponding to the main wind direction in the rainy season), i.e., plots F1–F3 for the fallow and M3 for the millet (Fig. 1a). For millet, the mean LAI from the four plots M1–M4 was also tested. LAI for the fallow’s shrub layer was approximated as a linear rise from 0 on June 1st to 0.3 m2 m2 at the end of October. Maximum LAI for G. senegalensis was found to be 0.3 m2 m2 in conditions similar to those of the Wankama catchment (Levy et al., 1997; Verhoef et al., 1996). Soil respiration Rsoil was estimated from the soil water content and soil temperature at 10 cm, using the relationship proposed by Hanan et al. (1998) for similar soils

Rsoil ¼ R0soil ½expð0:059  ðTs  20ÞÞ=ð1 þ expð0:507ðTs  35:8ÞÞÞ½ðS10  S10min Þ=ðS10max  S10min Þ

ð5Þ

where R0soil = 1.194 lmol m2 s1 is the intrinsic respiration rate, Ts is soil temperature, S10 is soil water content, S10min and S10max are the minimum and maximum soil water contents. Details on this equation and the constants used can be found in Hanan et al. (1998). Cover-scale assimilation values Fc estimated with Eqs. (3) and (4) were compared to corresponding values of carbon flux measured by the EC stations. Results The development of vegetation is conditioned by environmental variables such as solar radiation, in particular PAR, soil water content, and the conditions in the surface atmospheric layer (wind speed, vapour pressure, temperature), which determine whether

194

N. Boulain et al. / Journal of Hydrology 375 (2009) 190–203

Table 1 Characteristics of 2005 and 2006 seasons for the millet and fallow plots. Mean daytime PAR, maximum and mean daytime temperatures are calculated from June, 15 to October, 31 in each season. Rain is the season total, and the length of the season is calculated as the number of days between the first and the last rainfall. Biomass and LAI values are sample-averaged seasonal maximums. 2005

2006

Fallow Total rain (mm) Date of first rain Length of season (days) Mean PAR (lmol/m2/s) Mean diurnal temperature (°C) Maximum diurnal temperature (°C) Maximum biomass (g m2) Maximum LAI (m2 m2)

Millet

Fallow

495 May 1st 165 1057 31 41.9 70.8 0.69

the plant is stressed or not. Table 1 compares the climatic conditions that prevailed in the two seasons (total rainfall, number of days between the first and last rainfall events, PAR, mean and maximum daytime temperature), together with maximum biomass and LAI for the two systems. The two seasons showed contrasting conditions with respect to rainfall distribution (Fig. 2a): 2006 had more rainfall (572 mm) over a shorter period than 2005 (495 mm); the other parameters did not exhibit any significant differences between years or systems.

Biomass and LAI Fallow For the fallow, measured LAI values varied between 0 and 1.7 m2 m2 for individual samples over the two seasons. The maximum sample-averaged value for the 2005 season was 0.7 m2 m2

Millet 572 June 1st 115

1066 31 40 200.8 0.26

1031 32 41.1 131.6 0.86

1039 31 40.3 130.1 0.24

while it was 0.9 m2 m2 in 2006 (Fig. 2b). The one-month delay of the 2006 rain season start relative to the 2005 season (Table 1) meant a lag of over one month in the LAI development. Indeed, the LAI was above 0.2 m2 m2 at the beginning of July 2005, whereas it was below 0.1 m2 m2 at the beginning of August 2006. However, in both years the LAI peak occurred between the end of August and the beginning of September. After peaking at the end of the 2005 season, the LAI dropped rapidly to reach 0.2 m2 m2 at the end of September, whereas in the 2006 season, the LAI only decreased from 0.9 m2 m2 to 0.5 m2 m2 in mid-September, and then remained more or less unchanged until October. Peak in aboveground biomass (Fig. 3b) was reached by the end of August in 2005 (at 71 g m2), against mid-September in 2006 (132 g m2). In the same way as with the LAI, the season beginning was quite different for the 2 years. In 2005, biomass was above 8 g m2 at the beginning of July; whereas in 2006 it did not reach 5 g m2 until after July 20. In both seasons, growth took place over

Fig. 2. Rainfall and variations in leaf area index (LAI) of fallow and millet over study period. (a) 2005–2006 rain patterns: event (bars) and cumulated (lines) rainfall (stars: dry spells of at least 10 days); (b) mean herbaceous LAI in fallow; and (c) mean millet LAI; (bars in (b) and (c): unbiased standard deviations).

N. Boulain et al. / Journal of Hydrology 375 (2009) 190–203

195

Fig. 3. Rainfall and variations in aboveground biomass of fallow and millet over study period. (a) Rainfall (same as Fig. 2a); (b) mean aboveground herbaceous biomass in fallow and (c) mean aboveground millet biomass; (bars in (b) and (c): unbiased standard deviations).

Fig. 4. Variations in relative soil water content S/Smax of the 0–275 cm layer, with S the soil water content, Smax the maximum soil water content observed over the study period, for (a) the fallow and (b) the millet, over the two rain seasons 2005 (black) and 2006 (grey). The horizontal dotted line (70% of Smax) represents a limit between the atmospheric control and the soil control of evapotranspiration. Curves on top of each graph represent cumulated rainfall for 2005 (black) and 2006 (grey).

196

N. Boulain et al. / Journal of Hydrology 375 (2009) 190–203

two months. But despite the delayed start of the 2006 season, final biomass and maximum LAI were higher in 2006 than in 2005. Variations in the S/Smax ratio (Fig. 4) inform on periods of stress due to soil water content. In 2006, rainfall was rather regular between mid-July and mid-September with no dry spell longer than 10 days. Conversely, in 2005, there were two long dry spells of over 10 days right in the middle of the growing season (from July 19 to July 30, and from August 19 to August 29). Fig. 4a shows that the first long dry spell in 2005 (from July 19 to July 30) dried the soil, from a high moisture content (80% of Smax) to a value close to that at the beginning of the season (60% of Smax). In that period of the season, the 2006 soil moisture trajectory started to differ from the 2005 trajectory. The second long dry spell in August of 2005 dried the soil again (from 80% down to 60% of Smax). From this second long dry spell until the end of the 2005 season, soil moisture rose only once above 70% of Smax, during the 40-mm rain event of September 20. Oppositely in 2006, through the same period of peak vegetation, soil moisture was high and reached Smax.

Millet fields Millet LAI values varied from 0 to 0.65 m2 m2 for individual samples, i.e., in the same range as those observed by Bégué et al. (1996) in a millet field with no fertilization or irrigation. In 2005, the maximum sample-averaged LAI was 0.26 m2 m2 (Fig. 2c), at the end of August, whereas in 2006 the maximum was 0.24 m2 m2, at the middle of September. In 2005, there was a rapid increase in the LAI after germination: in 15 days it rose from 0 to 0.17 m2 m2 (i.e., more than 50% of the maximum LAI), continued to increase to 0.26 m2 m2 at the end of August, and finally remained at around 0.26 m2 m2 until end of September. In 2006, LAI values were less than 0.1 m2 m2 at the beginning of August, remained at around 0.1 m2 m2 until end of August, rose to 0.24 at the middle of September, and finally decreased to 0.2 m2 m2 at the beginning of October.

Throughout the 2005 season, the S/Smax ratio of the millet field remained below 70% (Fig. 4b), except during two rainfall events (September 8 and September 20). After July 19 (beginning of the first dry period in 2005), the courses of the two seasons diverged. In 2006, the S/Smax ratio was systematically above 2005 values. In 2006, after August 8, the S/Smax ratio was above 70% and remained so until October 1. These results suggest that the millet field was under stress condition more or less through the whole 2005 growing season, but only at the season beginning in 2006. Despite more water being available in 2006 than in 2005 (Fig. 4b), biomass was greater in 2005 than in 2006 (Fig. 3c). In 2005, the maximum of 201 g m2 was reached at the beginning of September, whereas in 2006 it was as late as mid-October (130 g m2), well after the last rains. Carbon fluxes Fallow Fluxes at the scale of the fallow system include those within the herbaceous layer (whose development was described in the previous sections) and those for the shrub and tree layers. Fig. 5a and b show an increase in assimilation fluxes up to the end of August in both seasons, followed by a decrease until the end of October, when they reached values close to those at the beginning of the season. At the beginning of the season, daytime values show very small net assimilation (in 2005, Fig. 5a) or respiration (positive values in 2006, close to nighttime levels, see Fig. 5b). The peak of net assimilation, reached at the end of August in both years, was 10.7 lmol m2 s1 in 2005 against 9.5 lmol m2 s1 in 2006. In both years, net assimilation tended to decrease after the first week of September, and was back to near-zero values at the end of the rainy season. The seasonal pattern of nighttime respiration was similar in the two seasons (Fig. 5a and b), with a maximum of +4 lmol m2 s1 reached by the end of August, lasting till mid-September.

Fig. 5. Daily carbon fluxes (Fc, lmol m2 s1) between June 15 and October 31 of each year, for fallow ((a) = 2005; (b) = 2006) and millet ((c) = 2005; (d) = 2006). Bars on top of each graph represent rainfalls. Daytime fluxes (black dots) are taken from 6 am to 4:30 pm, for gap-free days only. Nighttime fluxes (empty circles) are taken between 5 pm and 5:30 am.

N. Boulain et al. / Journal of Hydrology 375 (2009) 190–203

The most significant difference between the two seasons can be seen at their beginning, due to the late start of the 2006 season. Indeed, assimilation did not start until July 15 in 2006, whereas there was some net assimilation before the July 9 rain in 2005 (ever since June 15 when measurements started), and a sharp increase after that event. A more detailed picture of the first part of the 2006 season is provided by Fig. 6, zooming on the period of June 1–August 31, and showing half-hourly daytime carbon fluxes together with rainfall, aboveground herbaceous biomass, and evapotranspiration. It can be seen that in June/early July, every rainfall event caused an immediate response with an increase in respiration and evapotranspiration followed by a gradual decrease. The same type of behaviour was observed for the millet system. Millet The seasonal pattern of net assimilation differed strongly between years (Fig. 5c and d). In 2005, the maximum net assimilation

197

was reached by mid-August, whereas in 2006 it was not until the end of September. These patterns are consistent with the observed seasonal dynamics of biomass (Fig. 3c). In both seasons, maximum net assimilation was around 5 lmol m2 s1. After the peak, net assimilation did not fall as fast as for the fallow, especially in 2005 when high levels occurred until the beginning of October. Maximum nighttime respiration was around 3 lmol m2 s1 in both seasons. In contrast to net assimilation, maximum respiration was reached at the end of September in both seasons. Comparison of vegetation response curves at leaf and cover scales Fallow Fig. 7 shows the response of half-hourly net assimilation to PAR at the scale of the fallow cover in 2005 and 2006 (Fig. 7a), and at the leaf scale for the two main species: Z. glochidiata (Fig. 7b) for the herbaceous layer and G. senegalensis (Fig. 7c) for the shrub

Fig. 6. Early rain season dynamics of fallow (June 1–August 31, 2006): (a) daily rainfall, (b) mean aboveground herbaceous biomass, (c) half-hourly daytime carbon flux (Fc), and (d) evapotranspiration (LE).

198

N. Boulain et al. / Journal of Hydrology 375 (2009) 190–203

Fig. 7. Photosynthetic response of fallow vegetation to light (PAR in lmol m2 s1), with no water stress (VPD < 1.5 and S/Smax > 0.7), for: (a) the fallow plot scale (black dots for 2005, grey circles for 2006; half-hourly data); (b) the herbaceous leaf scale and (c) the shrub leaf scale.

Fig. 8. Water use efficiency (WUE) response of fallow vegetation to vapour pressure deficit (VPD), with no root water stress (S/Smax > 0.7), for: (a) the fallow plot scale (black dots for 2005, grey dots for 2006; half-hourly data); (b) the herbaceous leaf scale; and (c) the shrub leaf scale.

N. Boulain et al. / Journal of Hydrology 375 (2009) 190–203

layer. At the scale of the cover (Fig. 7a) there was an increase in net assimilation with an increase in PAR. However the response was very scattered. Maximum net assimilation was 17 lmol m2 s1. At the scale of the leaf, a classical response of photosynthesis to PAR was also observed for both species, with photosynthesis increasing with PAR and levelling off at high PAR values. For both species the maximum measured photosynthesis (Amax) was about 25 lmol m2 s1. The values for G. senegalensis were more scattered than for Z. glochidiata. Fig. 8 shows the response of WUE to vapour pressure deficit (VPD) at the scale of the fallow cover in 2005 and 2006 (Fig. 8a), and at the leaf scale for the two species mentioned above (Fig. 8b and c). It can be seen that at the cover scale like at the leaf scale, the WUE decreases with an increase in VPD. Note however that the relationships are much more consistent at leaf scale, for both species. Millet Fig. 9 shows the response of half-hourly net assimilation to PAR at the scale of the millet cover in 2006 (Fig. 9a), and at the scale of the millet leaf (Fig. 9b). Because of the filtering applied to the data (VPD < 1.5 hPa and S/Smax > 0.7) there were no values for the 2005 season. At the leaf scale, observed values showed an extremely high Amax (between 40 and 60 lmol m2 s1), whereas at the cover scale, maximum net assimilation was between 5 and 10 lmol m2 s1. The values observed for net assimilation at leaf scale are in agreement with those reported by Bois (1993), and values observed at the cover scale compare favourably with those measured during the HAPEX–Sahel experiment (Moncrieff et al., 1997). Fig. 10 shows the response of WUE to VPD at the scale of the millet cover in 2006 (Fig. 10a), and at the scale of the leaf (Fig. 10b). While a clear relationship is observed at leaf scale, with

199

WUE decreasing with increasing VPD, no relationship is apparent at the cover scale. Modelling the scale relationship The modelled light-to-assimilation relationships at leaf scale are displayed in Figs. 7b,c and 9b for Z. glochidiata, G. senegalensis and P. glaucum, respectively. Modelled seasonal courses of LAI are shown as insets in Fig. 11a (for the herbaceous and shrub layers of fallow plots) and Fig. 11b (for millet plots). Figs. 11a and b compare, for the fallow and millet fields respectively, the EC-measured half-hourly carbon fluxes with the cover-scale assimilation values obtained, under no water stress condition, with the linear model of Eqs. (3) and (4). Model results for fallow are in good agreement with observations (Fig. 11a). For millet, the model underestimates both respiration and assimilation fluxes compared to EC measurements when only the LAI from the M3 plot is used, but tends to overestimates when using the mean LAI from the four plots M1 to M4 (Fig. 11b). Discussion Responses of vegetation covers to seasonal heterogeneity Our results for biomass and LAI of millet and of the herbaceous layer of fallow are in agreement, with respect to the dynamics and orders of magnitude, with previous observations made in the same region (Levy et al., 1997; Moncrieff et al., 1997; Hanan et al., 1998). During most of the 2005 growing season, unlike 2006, the S/Smax ratio remained below 0.7 for both systems, resulting in water stress at the level of the roots. In the case of the studied fallow, the dry spells in the 2005 season perturbed herbaceous development, whereas in 2006, even if the first rains started later, rainfall distribution was subsequently more regular and abundant, en-

Fig. 9. Photosynthesis response of millet to light (PAR in lmol m2 s1), with no water stress (VPD < 1.5 and S/Smax > 0.7), for: (a) the millet plot scale (half-hourly data from the 2006 rain season (no plot-scale data from the 2005 season satisfied the S/Smax > 0.7 condition)) and (b) the millet leaf scale.

200

N. Boulain et al. / Journal of Hydrology 375 (2009) 190–203

Fig. 10. Water use efficiency (WUE) response of millet to vapour pressure deficit (VPD), with no root water stress (S/Smax > 0.7), for: (a) the millet plot scale (half-hourly data from the 2006 rain season (no plot-scale data from the 2005 season satisfied the S/Smax > 0.7 condition)) and (b) the millet leaf scale.

abling better development of the herbaceous cover (Fig. 3b). On the contrary, the dry spells of 2005 turned out less detrimental to millet development than the late beginning of sustained rainfall in 2006, despite its higher season total (Fig. 3c). Millet is able to withstand dry spells during its growth stage: a water stress period of several days may reduce photosynthesis by more than half, but photosynthesis can recover its pre-stress level as soon as the stress ends (Bois, 1993). On the other hand, a late season can be very harmful (Boulain et al., 2006; Hamon, 1993). In 2006, the farmers sowed several times until mid-July. It is usually not advisable to sow this late (Eldin, 1993). However subsequent regular and abundant rainfall till mid-September 2006 ensured enough soil water reserve to enable millet to continue growing until the beginning of October, and make a harvest possible. Nevertheless, the harvest was small, i.e., smaller than might have been expected given the quantity of rain. Conversely, for the grass production on fallow land, which provides grazing for livestock, what appears important is the volume and, even more so, the frequency of rainfall events during the growing season. Dry spells during the season have a serious adverse effect on the development of herbaceous species, and consequently on total production. Marked differences in the two systems’ responses to variable rainfall distribution can also be traced in the carbon flux patterns. The one-month lag in the time of rain arrival between the 2 years did not translate for the fallow into any substantial delay in peak net assimilation in 2006 (Fig. 5a and b). Hence, a shorter time was needed in 2006 to reach about the same level of peak net assimilation, suggesting a high plasticity of the native species that compose the herbaceous layer of the fallow. Net assimilation then decreased more slowly in 2006, leading to larger biomass (Fig. 3b). By contrast, for the cultivated system, the delay in peak net assimilation from 2005 to 2006 was in the same order as that in the start of rainfall (Fig. 5c and d), pointing to the absence of photo-sensitiv-

ity in the millet varieties used at the study site, unlike some other varieties found in the Sahel (Diop et al., 2005). This did not leave the crop enough time to catch up for this delay, in terms of biomass (Fig. 3c). Seasonal carbon fluxes The two systems show clear differences in rain-season carbon fluxes (Fig. 5). Assimilation over the growing season appears much higher in the fallow than in the cultivated system, both for a dry and a wet year. Contrasting with the leaf-scale assimilation observations (Figs. 7 and 9), this result can be explained by the much higher vegetation cover fraction (as evidenced by Fig. 1b and c) and LAI (Fig. 2b and c) in the fallow system. Before the start of vegetation growth, the behaviour of the two systems is essentially the same, with carbon fluxes being small and mainly determined by soil respiration (Fig. 5). As can be seen in Fig. 6, soil respiration reacts to the first rainfalls. While microorganisms activity is very low in the dry soil of the dry season, soil moistening by these initial rainfalls produces bursts of activity (also reported, e.g., by Brümmer et al., 2008 for a Sudanian savanna), which vanish rapidly as the soil dries out again between rain events. When rainfall becomes regular and abundant enough to maintain some soil moisture, as of the beginning of July generally, soil respiration becomes more continuous and stable, and remains quite similar in the two systems, as evidenced by the nighttime fluxes of Fig. 5. From then on, variations in net assimilation flux are mainly due to vegetation development, particularly in the herbaceous layer of the fallow. Overall, for the rainy period, the fallow system appears as a more significant carbon sink than the cultivated system. This point is important, in light of the major shift in land use/land cover undergone by the West Niger area over the past decades,

N. Boulain et al. / Journal of Hydrology 375 (2009) 190–203

201

Fig. 11. Comparison of modelled and measured assimilation fluxes at cover scale, under no water stress condition (VPD < 1.5 and S/Smax > 0.7), for: (a) fallow and (b) millet (black and grey dots in (b) correspond to modelling with LAI from the M3 millet plot and with LAI from the M1–M4 plots, respectively). Insets represent modelled seasonal (2006) courses of LAI: (a) herbaceous (black line) and shrub (grey line) layers of fallow and (b) M3 millet plot (black line) and M1–M4 millet plots averages (grey line).

with large scale clearing of natural vegetation for crop expansion, as well as a reduction in surface areas and durations of fallows (Cappelaere et al., 2009; Leblanc et al., 2008). A substantial impact on the regional carbon budget could thus result from this land cover change, experienced over much of the Sahelian belt.

Carbon fluxes from leaf to cover Our results at leaf scale showed an Amax of 25 lmol m2 s1 for G. senegalensis, whereas Meir et al. (2007) reported a maximum assimilation of 11 lmol m2 s1 for the same species in the same region, but in a tiger bush system which is quite different from the fallow system. Fallows are generally found on sandy valley slopes with G. senegalensis dominating the shrub layer (almost mono-specific), whereas tiger bush occurs only on the lateritecapped plateaus of ‘‘Continental Terminal” standstone (Seghieri and Galle, 1999), and G. senegalensis is dominant only in the upslope colonization front of the vegetation bands. At the scale of the cover like at the leaf scale, carbon fluxes were found to correlate well with the quantity of PAR for the two systems, under conditions of no water stress at root and leaf levels (Figs. 7 and 9). In the studied fallow, net assimilation values at cover scale were in the same range as those measured at leaf scale for the species dominating the grass and shrub layers (Fig. 7). On the

opposite, for millet, net assimilation at the cover scale was markedly lower than at leaf scale (Fig. 9). When the above experimental relationships of leaf-scale assimilation to PAR were transferred to the cover scale through the use of a simple linear model, using LAI as the scaling variable, a good agreement was found with EC measurements for the fallow system, in the absence of water stress (Fig. 11a). This suggests a high consistency of the measurements performed at the two scales. These results showed low sensitivity to the hypothesized seasonal curve for shrub LAI. The same analysis performed for the millet system (Fig. 11b) was less conclusive, with underestimation by the model when only the M3 monitoring plot, located upwind of the EC system, was used for modelling the seasonal course of LAI, and overestimation when the four M1–M4 plots were used. In relation to a strong relative space variability of millet LAI, this points to the need for better sampling of the millet cover, in order to better address the scaling issue for this system. However our results do allow to conclude that the low cover-scale assimilation by the crop system, despite a high photosynthetic capability of the millet leaf, largely stems from the low LAI, caused in turn by a low density of millet stands, as a result of agricultural practice. Finally, it must be emphasized that under water stress conditions, other variables than just LAI would of course need to be considered in order to attempt transferring to the cover scale the leaflevel responses of assimilation to light.

202

N. Boulain et al. / Journal of Hydrology 375 (2009) 190–203

Water use efficiency

Acknowledgements

Air humidity influences the opening/closing of the stomata, as a high VPD tends to force the stomata to close and to reduce stomatal conductance (Levy et al., 1997; Verhoef et al., 1996). All three species investigated showed a strong inverse relationship at leaf scale between VPD and WUE (Figs. 8b,c and 10b), but with large, systematic differences between them in orders of magnitude of WUE. Over the whole VPD range, WUE was in the following increasing order: Z. glochidiata, then G. senegalensis, and finally P. glaucum. At the fallow cover scale, the WUE-to-VPD relationship (Fig. 8a) was similar but weaker than at the leaf scale, with WUE values in the same orders of magnitude as those of leaf scale Z. glochidiata (Fig. 8b). These values are lower than those observed during the Hapex–Sahel experiment (Verhoef et al., 1996). In the case of millet, no significant relationship was found between WUE and VPD at the scale of the cover (Fig. 10a). Bare soil evaporation probably affects estimation of vegetation efficiency at the millet cover scale, as suggest the studies by Moncrieff et al. (1997): the low plant-covered fraction of soil surface (as indicated by the low LAI values in Fig. 2c, and illustrated by Fig. 1c), compared to fallow (Figs. 1b and 2b), means that the share of soil evaporation in total, cover-scale evapotranspiration is much higher for the millet system.

This work was supported by the ‘‘Eau et végétation au Niger” project of the French ECCO-PNRH programme, by the AMMA programme1 (including the AMMA-Catch O.R.E.), and by IRD and CNRS. We warmly acknowledge all members of AMMA-Catch Niger who helped in field work and data processing. We are grateful to E. Mougin, to N. Hanan, and to the two anonymous reviewers for their very helpful reviews.

Conclusion The new field data collected for the two dominant types of vegetation systems in the West Niger region of the cultivated Sahel, namely fallow and millet fields, revealed differences in behaviour over two years with markedly different rainfall distribution. Fallow appeared to be less sensitive than millet to the dates of the first significant rainfalls. Conversely, prolonged dry spells (more than 10 days) during the rain season had more serious consequences for fallow than for millet. Consistent variations were found in vegetation variables (LAI, aboveground biomass) and in carbon fluxes measured by EC systems. Response curves of assimilation to light were established at leaf scale for the three main species in the studied systems, namely Z. glochidiata and G. senegalensis for the grass and shrub layers of fallow, and P. glaucum for millet fields. A simple upscaling rule based on cover LAI was quite successful, especially for the fallow, in reproducing the cover-scale assimilation fluxes as recorded by the EC stations, under conditions of no water stress of leaves and roots. For the millet cover, more uncertainty on the upscaling arose from sampling of the strong LAI spatial variability. Of course only a first approach to flux upscaling was considered here, more investigation being necessary to account for the effect of the strong heterogeneity, in spatial structure and species composition, and thereby improve our capacity to model the influence of land surfaces on the global atmosphere. Finally, our results show that, at the cover scale, assimilation is larger for the fallow than for the millet fields during the rain season. The considerable land use change, from an essentially natural to a largely cultivated ecosystem with reduced fallowing, which has been occurring in the Sahelian region for the last decades (Reenberg et al., 1998; Mortimore et al., 2005) could thus have reduced the capacity of Sahelian ecosystems to store carbon in the monsoon season. Lower cycling of water back to the atmosphere by rain season evapotranspiration from the cultivated system was evidenced by Ramier et al. (2009). The impact of land use change on carbon and water budgets at catchment scale is presently being investigated.

References Badejo, M., 1998. Agroecological restoration of savanna ecosystems. Ecological Engineering 10, 209–219. Baldocchi, D., 1994. A comparative-study of mass and energy-exchange rates over a closed c-3 (wheat) and an open c-4 (corn) crop. 2. CO2 Exchange and water-use efficiency. Agricultural and Forest Meteorology 67, 291–321. Bégué, A., Desprat, J.F., Imbernon, J., Baret, F., 1991. Radiation use efficiency of pearl millet in the Sahelian zone. Agricultural and Forest Meteorology 56, 93–110. Bégué, A., Roujean, J., Hanan, N., Prince, S., Thawley, M., Huete, A., Tanré, D., 1996. Shortwave radiation budget of Sahelian vegetation. 1. Techniques of measurement and results during HAPEX–Sahel. Agricultural and Forest Meteorology 79, 79–96. Bois, J.F., 1993. Effets de la contrainte hydrique sur la photosynthèse du mil. In: Hamon, S. (Ed.), Le mil en afrique, diversité génétique et agro-physiologique: potentialités et contraintes pour l’amélioration génétique et l’agriculture. ORSTOM, pp. 191–198. Boulain, N., Cappelaere, B., Seguis, L., Gignoux, J., Peugeot, C., 2006. Hydrologic and land use impacts on vegetation growth and NPP at the watershed scale in a semi-arid environment. Regional Environmental Change 6, 147–156. Boulain, N., Cappelaere, B., Seguis, L., Favreau, G., Gignoux, J., accepted for publication. Water balance and vegetation change in the Sahel: a case study at the watershed scale with an eco-hydrological model. Journal of Arid Environments. Brümmer, C., Falk, U., Papen, H., Szarzynski, J., Wassmann, R., Bruggemann, N., 2008. Diurnal seasonal, and interannual variation in carbon dioxide and energy exchange in shrub savanna in Burkina Faso (West Africa). Journal of Geophysical Research-Biogeosciences 113, G02030. doi:10.1029/ 2007JG000583. Cappelaere, B. et al., 2009. The AMMA-CATCH experiment in the cultivated Sahelian area of south-west Niger – strategy, implementation, site description, main results. Journal of Hydrology (AMMA-Catch special issue) 375 (1–2), 34–51. Cuenca, R.H., Brouwer, J., Chanzy, A., Droogers, P., Galle, S., Gaze, S.R., Sicot, M., Stricker, H., Angulo-Jaramillo, R., Boyle, S.A., 1997. Journal of Hydrology 188 (189), 224–266. de Rouw, A., 2004. Improving yields and reducing risks in pearl millet farming in the African Sahel. Agricultural Systems 81, 73–93. Demarez, V., Duthoit, S., Baret, F., Weiss, M., Dedieu, G., 2008. Estimation of leaf area and clumping indexes of crops with hemispherical photographs. Agricultural and Forest Meteorology 148, 644–655. Diop, M., Reyniers, F.N., Sarr, B., 2005. Apport du photopériodisme à l’adaptation du mil à la sécheresse en milieu soudano-sahélien. Sécheresse 16 (1), 35–40. Do, F., Winkel, T., Cournac, L., Louguet, P., 1996. Impact of late-season drought on water relations in a sparse canopy of millet (Pennisetum glaucum (L) R Br). Field Crops Research 48, 103–113. Eldin, M., 1993. Analyse de l’effet des déficits hydriques sur la récolte du mil au Niger conséquences agronomiques. In: Hamon, S. (Ed.), Le mil en afrique, diversité génétique et agro-physiologique: potentialités et contraintes pour l’amélioration génétique et l’agriculture. ORSTOM, pp. 149–160. Friborg, T., Boegh, E., Soegaard, H., 1997. Carbon dioxide flux, transpiration and light response of millet in the Sahel. Journal of Hydrology 189, 633–650. Frouin, R., Pinker, R.T., 1995. Estimating photosynthetically active radiation (PAR) of the Earth’s surface from satellite observations. Remote Sensing Environment 51, 98–107. Goutorbe, J.P., Noilhan, J., Lacarrere, P., Braud, I., 1997. Modelling of the atmospheric column over the central sites during HAPEX–Sahel. Journal of Hydrology 189, 1017–1039. Gu, L., Baldocchi, D., 2002. Fluxnet 2000 synthesis – foreword. Agricultural and Forest Meteorology 113, 1–2. Hamon, S., 1993. Le mil en afrique, diversité génétique et agro-physiologique: potentialités et contraintes pour liamélioration et la culture. I’ORSTOM.

1 Based on a French initiative, AMMA (African monsoon multidisciplinary analyses) was built by an international scientific group and is currently funded by a large number of agencies, especially from France, the UK, the US and Africa. It has been the beneficiary of a major financial contribution from the European Community’s Sixth Framework Research Programme. Detailed information on scientific coordination and funding is available on the AMMA International website http:// www.ammainternational.org.

N. Boulain et al. / Journal of Hydrology 375 (2009) 190–203 Hanan, N.P., Kabat, P., Dolman, A.J., Elbers, J.A., 1998. Photosynthesis and carbon balance of a Sahelian fallow savanna. Global Change Biology 4, 523–538. Le Barbe, L., Lebel, T., 1997. Rainfall climatology of the HAPEX–Sahel region during the years 1950–1990. Journal of Hydrology 189, 43–73. Lebel, T., Taupin, J.D., Damato, N., 1997. Rainfall monitoring during HAPEX–Sahel. 1. General rainfall conditions and climatology. Journal of Hydrology 189, 74– 96. Lebel, T., Cappelaere, B., Hanan, N., Levis, S., Descroix, L., Galle, S., Kergoat, L., Mougin, E., Peugeot, C., Séguis, L., Vieux, B., 2009. AMMA-CATCH studies in the Sahelian region of West-Africa: an overview. Journal of Hydrology (AMMACatch special issue) 375 (1–2), 3–13. Leblanc, M.J., Favreau, G., Massuel, S., Tweed, S.O., Loireau, M., Cappelaere, B., 2008. Land clearance and hydrological change in the Sahel: SW Niger. Global and Planetary Change 61, 135–150. Levy, P., Moncrieff, J., Massheder, J., Jarvis, P., Scott, S., et al., 1997. CO2 fluxes at leaf and canopy scale in millet, fallow and tiger bush vegetation at the HAPEX–Sahel southern super-site. Journal of Hydrology 189, 612–632. Lloyd, C.R., Bessemoulin, P., Cropley, F.D., Culf, A.D., Dolman, A.J., Elbers, J., Heusinkveld, B., Moncrieff, J.B., Monteny, B., Verhoef, A., 1997. A comparison of surface fluxes at the HAPEX–Sahel fallow bush sites. Journal of Hydrology 189, 400–425. Mauder, M., Foken, T., 2004. Documentation and instruction manual of the eddy covariance software package TK2. Report No. 26, U. Bayreuth – Abt. Mikrometeorologie. Available from: . Meir, P., Levy, P., Grace, J., Jarvis, P., 2007. Photosynthetic parameters from two contrasting woody vegetation types in West Africa. Plant Ecology 192, 277– 287. Merbold, L., Ardö, J., Arneth, A., Scholes, R.J., Nouvellon, Y., de Grandcourt, A., Archibald, S., Bonnefond, J.M., Boulain, N., Bruemmer, C., Cappelaere, B., Ceschia, E., El-Khidir, H.A.M., El-Tahir, B.A., Falk, U., Kergoat, L., Le Dantec, V., Mougin, E., Muchinda, M., Mukelabai, M.M., Ramier, D., Roupsard, O., Timouk, F., Kutsch, W.L., 2008. Precipitation as driver of carbon fluxes in 11 African ecosystems. Biogeosciences Discussion 5, 4071–4105. Moncrieff, J., Monteny, B., Verhoef, A., Friborg, T., Elbers, J., et al., 1997. Spatial and temporal variations in net carbon flux during HAPEX–Sahel. Journal of Hydrology 189, 563–588.

203

Monteny, B., Lhomme, J., Chehbouni, A., Troufleau, D., Amadou, M., et al., 1997. The role of the Sahelian biosphere on the water and the CO2 cycle during the Hapex– Sahel experiment. Journal of hydrology 188–189, 516–535. Mortimore, M., Ba, M., Mahamane, A., Rostom, R.S., Serra del Pozo, P., Turner, B., 2005. Changing systems and changing landscapes: measuring and interpreting land use transformation in African drylands. Geografisk Tidsskrift, Danish Journal of Geographic 105, 45–62. Ramier, D., Boulain, N., Cappelaere, B., Timouk, F., Rabanit, M., Lloyd, C.R., Boukraoui, S., Métayer, F., Descroix, L., Wawrzyniak, V., 2009. Towards an understanding of coupled physical and biological processes in the cultivated Sahel – 1. Energy and water. Journal of Hydrology (AMMA-Catch special issue) 375 (1–2), 204– 216. Raynaut, C., 2001. Societies and nature in the Sahel: ecological diversity and social dynamics. Global Environmental Change 11, 9–18. Reenberg, A., Louring-Nielsen, T., Rasmussen, K., 1998. Field expansion and reallocation in the Sahel – land-use pattern dynamics in a fluctuating biophysical and socio-economic environment. Global Environmental Change 8, 309–327. Rockström, J., de Rouw, A., 1997. Water, nutrients and slope position in on-farm pearl millet cultivation in the Sahel. Plant and Soil 195, 311–327. Rockström, J., Jansson, P.E., Barron, J., 1998. Seasonal rainfall partitioning under runon and runoff conditions on sandy soil in Niger. On-farm measurements and water balance modeling. Journal of Hydrology 210, 68–92. Scanlon, T., Albertson, J., 2004. Canopy scale measurements of CO2 and water vapour exchange along a precipitation gradient in southern Africa. Global Change Biology 10, 329–341. Seghieri, J., Galle, S., 1999. Run-on contribution to a Sahelian two-phase mosaic system: soil water regime and vegetation life cycles. Acta Oecologica 20 (3), 209–217. Slatyer, R.O., McIlroy, I.C., 1961. Practical Micrometeorology. CSIRO, Melbourne, Australia. 310pp. Verhoef, A., Allen, S., DeBruin, H., Jacobs, C., Heusinkveld, B., 1996. Fluxes of carbon dioxide and water vapour from a Sahelian savanna. Agricultural and Forest Meteorology 80, 231–248. Winkel, T., Do, F., 1992. Caractères morphologiques et physiologiques de résistance du mil (Pennisetum glaucum (L.) R. Br.) à la sécheresse. L’agronomie Tropicale 46, 339–352.