The study was conducted in the northern Highlands Zone (NHZ) of Tanzania. We used a combination of farm surveys and on-farm trials with crop growth modelling on small-holder farms in northern Tanzania. A farm survey was first conducted covering three regions of NHZ (Arusha, Kilimanjaro and Manyara), which lies between 3o and 5o South latitude, 35o and 37o East longitude and is at an elevation of 685 to 1920 m above sea level. Generally in NHZ, two rainy seasons are recognised: one from March-July and the other from October-December locally referred to as “long rains” or “Masika” and “short rains” or “Vuli”, respectively. The rainfall is more intense and less variable during the long rains than during the short rains (Zorita and Tilya 2002). Predominant soil types range from humic nitosols that are well drained, to soils of volcanic origin (sandy loams to clay alluvial soils) (Mowo et al. 2006).
For the farm survey, farm households interviewed were randomly selected within the major agricultural areas of northern Tanzania. Following the stratified random approach developed by Africa Soil Information Service (AfSIS) (Huising 2012), twelve districts were selected in NHZ (3 in Arusha, 5 in Kilimanjaro and 4 in Manyara), within which a randomly selected 10 × 10 km grid was established per district. Within each of these 10 × 10 km grids, three 1 × 1 km clusters (wards) were randomly selected, where all farm households involved in farming were listed. Thereafter, within each cluster, eight farm households were randomly chosen. This translated into a sample size of 24 farm households per district (3 clusters × 8 farm households). Comprehensive data was collected using a semi-structured questionnaire in the form of an open data kit (ODK)-based survey instrument with the help of well-trained agricultural officers. Data collected included information on farm-household at plot level, reflecting the 2016/17 main agricultural season.
Following the farm survey, field experiments were conducted in three sites (Riroda, Arri and Dareda) within Babati district, Manyara region, an area with high diversity of agro-ecological zones (AEZs) and lies within 04o 25’S, 35o 60’E, at an altitude of 1300-1700 m above sea level. From each of the three AEZs, three farms were randomly selected to host experimental trials for three consecutive seasons (2017/2018- 2019/2020). Fields were ploughed and plots measuring 10 m long x 5 m wide marked just before planting. A path measuring 1m was left in between adjacent plots. Test crops used included: seed co. 513 maize (Zea mays L.) hybrid variety, lablab (Lablab purpureus (L.) Sweet) “Selian-Rongai” variety and pigeonpea (Cajanus cajan (L.) Millsp.) long- duration ICEAP 00040 and medium-duration ICEAP 00557 varieties. Pure stands of maize, pigeonpea and lablab were planted at a spacing of 0.90 m × 0.50 m inter and intra-row respectively. Intercrops followed an additive design, with legumes planted in between maize hills, maintaining same plant population for each crop in sole and intercrops. Thus, the total plant population in intercrops was double that of sole crops. An intercrop model was applied to further analyse the data and ascertain to what extent the productivity of maize– grain legume systems in northern Tanzania is limited by water availability.