Already prone to erratic rainfall, Malawi may suffer longer dry spells and a shorter growing season in the near future. Eighty percent of Malawians are subsistence farmers dependent on rainfed agriculture, and also face challenges of soil erosion, soil degradation, and lack of access to agricultural inputs. Malawian agriculture must adapt to climate change within the constraints imposed by poverty. Agroforestry may provide part of the solution. Fast-growing, nitrogen-fixing "fertilizer trees" such as Gliricidia sepium and Tephrosia candida, intercropped with maize (the region's staple crop), have been shown to provide great improvements in grain yield and soil quality. Anecdotal evidence suggests that these agroforestry systems may be more drought-resistant than monoculture maize, but this claim has not yet been tested. Our study investigated the drought resilience of these agroforestry systems using a two-pronged approach: (1) direct manipulation of rainfall in the field, using rain exclusion shelters to achieve 100% interception for the last twelve weeks of the growing season (~40% of annual total); (2) comparison of system performance in years of normal (2008-09) versus poor (2009-10) rainfall. We examined a Gliricidia system established in 1992, as well as newly established Gliricidia and Tephrosia systems, to compare drought effects on seedlings versus mature trees.

Results/Conclusions

The shelters achieved their goal of rain interception with reasonable labor requirements and minimal microclimatic artefacts; a drawback was PAR reduction, but this likely did not affect plant growth. The economical design (US$300 per shelter) has many potential applications for climate adaptation research in developing nations. Rain shelter results corroborated results from natural drought. In newly-established systems, survival and growth of Gliricidia seedlings (nursery-raised, field-transplanted) was excellent under normal and drought conditions, with early and late planting. However, Tephrosia (sown in the field) exhibited poor survival and growth under drought conditions and with late planting. Gliricidia appears a safer investment under such conditions. In the long-established Gliricidia system, trees did not compete with maize for water even during drought. Rather, agroforestry plots performed significantly better than maize monoculture under drought conditions, indicating the trees' benefits endure despite drought. We identified several mechanisms, independent of nitrogen addition, by which Gliricidia could confer drought resistance (microclimate, soil carbon, infiltration); however, we observed no effect on soil moisture. We conclude "fertilizer tree" agroforestry systems can improve food production even in a drought, and in some cases may confer drought resilience. We hope our work will inform future experiments on agroforestry and climate adaptation.