BC1 FRONT COVER REV4.indd

East Africa highland banana (Musa spp., AAA-EAHB) or ‘matooke’ is a major staple crop grown for food and sale by smallholder farmers in Uganda. Banana yields in farmers’ fi elds average 15 t/ha fresh weight (FW), and have remained small compared with yields of 60 to 70 t/ha/yr achieved at a research station with fertilizer application (Tushemereirwe et al., 2001; Smithson et al., 2001). Low banana yields achieved by smallholder farmers are attributed to poor soil fertility, low fertilizer use and increasing pest pressure (especially the banana weevil Cosmopolites sordidus) and moisture stress (NARO, 2000). Very few (<5%) banana farmers in Uganda use mineral fertilizers due to perceived high cost, poor availability, and lack of knowledge related to its use. Past research has highlighted the large extent of soil fertility decline. For most soils in Uganda, soil pH, extractable P, Ca and K are below critical concentrations for most crops (Ssali, 2002). Increased agricultural productivity, mainly through increased fertilizer use, is recognized as key to alleviating poverty and ensuring food security in rural parts of Uganda. To increase banana productivity in a profi table way, there is a need to develop fertilizer recommendations for balanced application of nutrients. Two nutrient omission trials were established at Kawanda (near Kampala) in central Uganda and Ntungamo in southwest Uganda to: (i) identify limiting nutrients and nutrient interactions in banana production; (ii) quantify banana yield responses to mineral fertilizers; and (iii) assess agronomic and economic effi ciency of fertilizer use in banana production. Soils at Kawanda are Haplic Ferralsols, while soils at Ntungamo are Lixic Ferralsols. Soil pH was 5.5 at Kawanda and 4.8 at Ntungamo. Average soil organic matter and total N values were higher at Kawanda (2.6 and 0.1%, respectively) than at Ntungamo (0.7 and 0.07%, respectively). Average Mehlich-3 extractable P was higher at Ntungamo (3.5 mg/kg), but exchangeable K and Mg were low at both sites. Rainfall at both sites averages 1,258 mm/yr and follows a bimodal pattern with dry periods from June to July and January to March. The trials were laid out in a completely randomized block design with four replicates. Treatments consisted of the following nutrient applications (kg/ha/yr): (1) 0N-0P-0K; (2) 0N-50P-600K; (3) 150N-50P-600K; (4) 400N-0P-600K; (5) 400N-50P-0K; (6) 400N-50P-250K, and (7) 400N-50P-600K. With the exception of the control, all plots received 60Mg6Zn-0.5Mo-1B kg/ha/yr. Nitrogen and K were applied in 8 splits each year, while P, Mg, Zn, Mo and B were applied in two splits at the start of each rainy season. All fertilizers were applied in a circle at 0.4 to 0.5 m from the base of the plant. A plant spacing of 3 x 3 m was used resulting in a density of 1,111 plants/ha. Under good management planting, one banana corm results in 3 production cycles. Yields from crop cycles 2 and 3 better represent a stable state. The development rate for banana at Kawanda was faster than at Ntungamo probably because of the difference in average temperatures (22 vs. 20°C, respectively). It was assumed that cycle 3 at Kawanda is reaching a stable state and would be comparable with cycle 2 from Ntungamo. The nutrient conversion effi ciencies [CE; kg fi nger (dry matter)/kg nutrient in plant] for individual banana plants at harvest were calculated. Bunch yields (t/ha/yr ) were calculated based on the duration from planting to harvest, but yields of successive crop cycles were based on the duration between consecutive harvests.