SUMMARYThis research aimed to develop a typology of agriculture in Timor-Leste using national census data at the village level. Although Timor-Leste is a relatively small nation, its varied topography contains a rich diversity in agricultural livelihoods, from coffee covered mountains, to dryland-swidden agriculture. Each of the livelihoods are very complex, with a single household often managing more than 10 crop and 4-5 animal species in very small holdings. Using census data from each village only, statistical clustering analysis was used to group villages with similar levels of participation in crop and livestock production. The clustered village groups were then mapped, and it was seen that villages in each cluster, occupied particular locations. Using expert knowledge about the locations of each cluster, livelihood zones based on a small number of rules were defined to mimic the output of the clustering. Seven livelihood zones were identified from mapping the livelihood systems. These included three zones with irrigation (rice-based), two highland zones (coffee-based) and two lowland zones based on rain-fed agriculture. Government and development agencies have endorsed the typology of livelihood zones, which is now in use for planning and decision-making. The technique of using national census data to define agricultural zones through statistical clustering can be replicated wherever there is reliable village-level census data.

Climate change is predicted to decrease crop yields in semi-arid and subtropical regions of the world and this could negatively affect smallholder farmers in the developing world. Previous analysis has suggested that with low fertiliser input, yields of sorghum increased as temperatures increased. We used the wide range of tropical environments in the small mountainous island country of Timor-Leste to evaluate the impact of global warming on maize (Zea mays) yields with (i) no fertiliser input and (ii) increased nitrogen (N) supply. We calibrated the well-tested APSIM-Maize model for the cultivar of maize grown throughout Timor-Leste. We simulated maize yields at four locations with 8 years of reliable weather records, at present temperatures, +1.5°C and +3.0°C, with 0, 40 and 80 kg/ha of added N, with 1.2, 1.9 and 3.8% soil organic carbon (SOC), and with increased duration of the vegetative phase. With no added N, higher temperatures increased yields at the cooler, higher elevation sites and decreased yields at the warmest site near the coast. With fertiliser application, warming temperatures decreased yields or induced no change in simulated yield at all locations. Simulations with three levels of N supply for the four sites, which differed in temperature, showed a strong temperature × N supply interaction on yield. At maximum growing-season temperatures >31°C, yields decreased with increasing temperature at all levels of fertilisation. At maximum growing-season temperatures of 23−31°C, yields increased with increasing temperature with no added fertiliser, were unchanged with the application of 40 kg N/ha, and decreased with increasing temperatures with application of 80 kg/ha N. The changes in yield with temperature and N supply were associated with N uptake by the maize, which showed the same interaction with maximum temperature and N added. SOC acted as a source of N, so that changes in yield induced by temperature and N were similar whether the N was from an organic or an inorganic source. Increasing the duration of the vegetative phase resulted in lower or no change in yields at all sites. We conclude that global warming will increase the yield for low-input smallholder farmers growing maize at maximum growing-season temperatures <31°C, but that micro-dosing with N will increase yields at all locations with mean maximum temperatures of 20−35°C.