Millet growth in windbreak-shielded fields in the Sahel : experiment and model

In the Sahelian zone, future food supply is insecure due to increasing land degradation. Wind erosion contributes significantly to impoverishment of the sandy soils, which are often loose and sparsely covered by vegetation for most of the year. At the onset of the growing season (May - July), strong winds often precede rains and may cause damage to the young seedlings by abrasion and burial. A possible control measure is the implementation of windbreaks (WBs) that reduce the wind velocity near the soil surface, thus protecting soil and crops. Moreover, the change in air flow may lead to a more favourable crop microclimate. Higher crop yields and increases in the long-term sustainability of crop production may be the result of effective windbreaks integrated in crops. However, windbreaks also compete with crops for limited resources, which may outweigh their potential benefits.Windbreak experiments, reported in literature, illustrate various benefits as well as negative impacts of windbreaks on crop growth (Chapter 1). In addition to a more favourable microclimate and reduced wind erosion, increased soil fertility, and complementary use of resources by trees and crops are possible agroforestry benefits. On the other hand, competition and allelopathic interactions between species, as well as attraction of pests and diseases by trees may reduce crop production. Whether the overall windbreak effect results in yield increases or losses depends on many interacting factors, e.g. climate, soil properties, crop species and WB-design. The few studies performed in semi-arid regions give insufficient insights to formulate generally applicable rules that allow extrapolation of experimental results to other locations or from one WB-design to another. Understanding of the tree-crop interactions is required for the design of optimum windbreak-cropping systems that can provide an option for sustainable land use in the Sahel.This research aimed at enhancing the understanding of positive and negative influences of windbreaks on crop production by i) collecting field data (Part A) and ii) analyzing crucial tree-crop interactions in terms of dynamic and spatial occurrence by means of a model (Part B).Between 1991 - 1993, experiments were performed at the ICRISAT Sahelian Center (ISC), Niger, to study the effects of low, narrowly spaced windbreaks on microclimate, light, water and nutrient resources at the windbreak-crop interface, and the growth of pearl millet ( Pennisetum glaucum ). Millet, a C4 tropical cereal, is the principal food crop in Niger, since it is particularly adapted to conditions of high temperatures, nutrient-poor soils and low rainfall. The windbreaks were north-south oriented, i.e. perpendicular to storms and prevailing winds and, hence, protecting crops sown on their west side. The agroforestry system under study consisted of various shelter species and plots without windbreaks as control. Measurements were performed along transects across the tree-crop interface. Radiation, wind speed, relative humidity, and air temperatures were measured continuously and soil temperatures occasionally measured in Bauhiniarufescens plots. Bauhinia windbreaks were 2 and 3 m high and had a porosity of 0.9 and 0.2 at the onset and the end of the growing season, respectively.Crop yields adjacent to seven windbreak species and in control plots were determined. In two rather wet years of experiments it was found that the overall yield in windbreak plots was slightly (mostly not significant) higher than in the control plots. Up to a distance of 2.5 H (with H the windbreak height) from windbreaks lower yields were measured than in the middle of the plots.Growth reduction in the vicinity of the windbreak was more than compensated by increased production in the center of the plots when averaged over all plots. The magnitude of yield increases varied among the tree species, e.g. in Bauhina sheltered crops this positive effect was rather small. From the experiments the causes for yield increases remained unclear, since no amelioration in the microclimate was observed. Air temperature and relative humidity (at the top of the sheltered crop) were constant with distance from the Bauhinia windbreak, although the shelter reduced wind velocity up to a minimum of 5 H. Generally, the influence of shade and wind speed on temperature and relative humidity are opposite and, presumably, negated in the present system. Shade, however, lowered soil temperatures, but this had no measurable impact on crop emergence and yields. Protection against wind erosion, enhanced water availability and/or small microclimate modifications (too small to be recorded) may have caused higher yields.The zone of severe yield reduction (0.5 - 1.5 H) corresponded to that of the strongest reduction in radiation and soil moisture, indicating competition effects. To assess competition between crops and windbreaks, soil water and nutrient status were observed in the windbreak-crop interface and in control plots. Nutrient status did not vary significantly across the tree-crop interface or between windbreak and control plots, whereas soil moisture was strongly reduced in the vicinity of trees during dry periods. Then, trees were likely to compete with crops for water since they had no access to groundwater. Competition for water apparently occurred although total rainfall was rather high in the experimental years. From these results it was concluded that in dry years competition for water would be much stronger. That this is not necessarily the case, was shown by a simulation study covering a dry and two wet years (Part B).Competition for water and light, and wind speed reduction within the 0 - 5 H tree-crop zone were considered to be of major importance in the agroforestry experiment, as described in Part A of this thesis. Consequently, to assess the agronomic feasibility of windbreaks in the Sahel, those effects have to be quantified.Assessment of the competitive effects between trees and crop is complicated, due to difficulties in determining below-ground competition and interference with other tree-crop interactions. A model approach is helpful to quantify and integrate individual effects of tree-crop interactions. When starting this study, no windbreak-cropping system model was available. Existing crop models lacked spatial heterogeneity, and temporal detail as required for the description of agroforestry systems. Therefore, a model, WIMISA (WIndbreak MIllet SAhel), has been developed by combining methods from existing models with newly developed concepts, with special attention to resolution in time and space.WIMISA simulates crop growth as influenced by local soil moisture and radiation intensity simultaneously for a number of rows parallel to a windbreak in daily time steps. Soil moisture content and incoming radiation intensities are computed at process level in smaller time steps. Windbreak growth is not simulated, but windbreak characteristics are introduced either as fixed values or as time dependent forcing functions. As knowledge of the WB characteristics is limited, competition for light, tree water uptake, and wind speed reduction are described using a simple model of the windbreak component to restrict the number of assumptions. Reduction in wind speed is incorporated through an empirical factor for wind velocities in the first 5 H from the windbreak.Competition for light is incorporated as light interception by a two-dimensional WB-barrier, as a function of its height and density (both increasing during the cropping season). WIMISA calculates instantaneous fluxes of direct and diffuse radiation for each crop row separately at intervals of 6 minutes. The interval for the computation of CO2 assimilation was set in accordance with the radiation calculations. Soil water flow is simulated in two dimensions to account for horizontal gradients due to different water extraction by trees and crop, and horizontally varying evapotranspiration. Spatial detail in vertical and horizontal direction is introduced through an arrangement of compartments. The soil water module uses a time step that is adjusted automatically to water flow rate and thickness of the soil compartment. Competition for water is expressed by distributing available soil water between trees and crop in proportion to their uptake rates in a non-competitive situation. Water uptake is calculated on the basis of root length density distribution.WIMISA was parameterised for millet, the tree species Bauhinia rufescens and soil characteristics of ISC. Specific crop parameters of the model for a sole millet system were determined throughout the growing season in a separate plot outside the influence of a windbreak. Simulations were confined to the competition zone (i.e. 5 H).Simulation results of the windbreak-cropping system were evaluated with data from Bauhinia plots of the 1992 and 1993 growing seasons. WIMISA gave reasonable estimates of straw and grain yields for millet westward of a windbreak, considering the generally high spatial variability and the fact that nutrient limitations are not included. Grain yield beyond 3 m was underestimated, probably due to windbreak effects not included in the model. This suggests that in addition to light other microclimate factors might have been somewhat modified by the windbreak, although not observed. Since simulated global radiation intensities and soil water contents, as a function of time and distance from the windbreak, and percentage yield reductions normalized to 10 m corresponded well with field observations, it can be assumed, that the processes of competition for water and light are simulated in the correct proportion.In the present agroforestry system, species utilized the resources competitively rather than complementary, resulting in (ca. 20 - 30 %) yield reductions in the 0.5 - 2 H zone adjacent to the windbreak. This extent of competition zone and magnitude of yield losses correspond to field data and results reported for other windbreak-cropping systems in the Sahel.Three scenarios where competition effects for water and for light were separated by turning off one of the effects during simulations, were compared for the 1987, 1992 and 1993 growing seasons. Surprisingly, water and total competition were highest in the wet year 1992, intermediate in the dry year 1987 and lowest in the wet year 1993. In fields with no access to groundwater, competition for water between trees and crops is likely to occur at the beginning of the rainy season, when available soil water is restricted to the upper horizons and transpiration of the windbreak is higher than that of the crop. Neither horizontal water flow nor reduced soil evaporation as a result of shade could compensate for tree water use. Under severe drought, water remained the major factor (1992), otherwise light was of similar or major importance (1987 and 1993). Simulation results indicate that the degree of light and water competition also depends on rainfall distribution and the time laps between the seasonal onset of tree and crop growth.The model is appropriate to quantify the relative importance of competition for water and light in a windbreak-millet system, although some process descriptions need further refinement (i.e. LAI development) and evaluation for application in other environments. In particular, modelling of tree-root water uptake requires more data for input and validation. Once, nutrient, complete microclimate and erosion modules have been incorporated in WIMISA, the model allows quantification of the overall WB effects and screening of different windbreak-cropping systems.