The challenge of HIV/AIDS: Where does agroforestry fit in? - World ...

104World Agroforestry into the FutureAgroforestry can make solid contributionsto resolving the apparent trade-off betweenmaintenance of watershed functions andproductive agriculture, if it addresses theissues in a way that links patch, field, farmand landscape scales.In this brief description of current approachesto agroforestry solutions to watershedproblems, we will consider the followingfour basic steps, and discuss the conceptsand tools required for each:1. Diagnosis of problems at watershedscale.2. Comparing land-use options on the basisof buffering functions.3. Modelling physical degradation andrehabilitation processes in the analysisof trade-offs between profitability andwatershed functions.4. Negotiations between stakeholders ofsolutions on the basis of trade-offs.Diagnosis of problems atwatershed scaleBecause there are many potential solutionsto problems with watershed function, weneed to be clear and specific about whatthe problem is, and this requires a commonperception (criteria and indicators; Figure 1).A list of criteria for the contribution of watershedsto water quantity (the capacity totransmit water, buffer peak flows and releasewater gradually), water quality (reduce sedimentloads and other pollutants and maintainaquatic biodiversity) and integrity of theland surface (control landslides and reduceloss of fertile topsoil through erosion), needsto be combined with criteria that relate tobiodiversity conservation and to the socialand economic welfare of the people livingin watershed areas.The relationship between full (as providedby a forest) and partial (agroforestry) treecover and hydrological functions in termsof the five watershed functions listed inFigure 1 involves different time scales andtrade-offs between total water yield and thedegree of buffering of peak river flows relativeto peak rainfall events. The role of landuse can be analysed in terms of changes inevapotranspiration, linked to the presenceof trees; infiltration, linked to conditions ofthe soil; and the rate of drainage linked tothe drain network in the landscape.van Noordwijk et al. (2003) completed adetailed analysis of both the 4000 km 2 MaeChaem catchment in northern Thailand(mean annual rainfall 1500 mm, populationdensity 20 km –2 ; mean annual river flow20–30 m 3 s –1 ) and the 500 km 2 Way Besaicatchment in Lampung, Indonesia (meanannual rainfall 2500 mm, population density160 km –2 ; mean annual river flow 15–20m 3 s –1 ). Daily rainfall and river flows forthese two watersheds are shown in Figure 2.The two rivers have very different patterns:the largely forested Mae Chaem shows avery strong seasonal pattern, falling nearlydry for a few months of the year; the WayBesai (only 15 percent forest) has approximatelycontinual flow. These differences,Function/criteriaMain indicator1. Transmit water • Q/P=1–(E/P)2. Buffer peak rain events • ? Q abAvg /? P abAvg3. Release gradually • Q slow /P = (P inf –E S+V )/P4. Maintain quality • Qual out /Qual in5. Reduce mass wasting • ? riskof course, primarily relate to the rainfallpattern. They show, however, that commonlyused indicators such as the ratio ofmaximum and minimum flow of the river,Q max/Q mincannot be used to analyse thecondition of watersheds, without regards torainfall.The indicators of Figure 1 are all expressedin dimensionless form, relating river flow(discharge) to rainfall. For the analysis ofthe Mae Chaem and Way Besai situations,a new ‘buffering indicator’ was developed(van Noordwijk et al. 2003) that relates thefrequency distribution of daily river flowto the frequency distribution of point-levelrainfall. It can be used to test perceptionsof increased flooding and peak flows. TheWay Besai data relate to a 23-year periodwhere forest cover was reduced from almost30% to less than 10% in 2002. Themain effect of this land-cover change wasto increase the total water yield as a fractionof total rainfall. The total discharge inthe month with the lowest flow, expressedas a fraction of annual rainfall, showedconsiderable variation between years butdid not change along with total water yield.The buffering indicator was negatively correlatedwith the total water yield, but forScaledependentFigure 1. Indicators for the five criteria. The quantitative properties of river discharge changealong the river course, and lead to scale dependence of three out of the five criteria. Q =river flow; P = precipitation; E S+V= total evapotranspiration minus evaporation of canopyintercepted water; abAvg = sum of all above average values; inf = infiltration.