Development and Validation of Nutrient Expert for Maize in Bangladesh

The project was initiated to develop and validate the Nutrient Expert for Maize, a site specific fertilizer recommendation tool for maize, in different maize growing areas of Bangladesh. The results from the validation trial is expected to improve the output of the tool and help the extension agencies to provide improved nutrient recommendation to farmers.

IPNI-2013-BGD-7

29 Apr 2016

2015 Annual Interpretive Summary


The association between agro-ecological zones (AEZs), indigenous soil fertility (ISF) and maize yield responses to fertilizer application was insufficiently characterized in the eastern Indo-Gangetic Plain (IGP) of South Asia. This has constrained the development of precision nutrient management recommendations in maize that aim to optimize nutrient use, increase yield and farm profits, while simultaneously reducing environmental footprints of agricultural nutrient use in the region. The factors affecting ‘rabi’ (winter season) maize yield responses to N, P, K, and Zn in the eastern IGP were studied to support the development of precision nutrient management strategies across Bangladesh’s predominant AEZs. Nutrient omission plot trials (NOPTs), with sequential omission of fertilizer N, P, K, and Zn from a nutrient sufficiency plot in which all four nutrients were applied at adequate rates for high yield targets, were conducted in 324 farmers’ fields across 10 AEZs.

Ranges for yield response to N, P, K, and Zn fertilization across AEZs were 2.90 to 6.35, 0.50 to 3.60, (-)0.10 to 2.40, and (-)0.40 to 1.00 t/ha, respectively. The Additive Main Effect and Multiplicative Interaction (AMMI) model was used to assess the effect of NOPT treatment on maize yields resulting from ISF, and to determine yield response association to AEZ. Nutrient omission treatments explained ~60% of yield variation, AEZ explained ~32%, while the AEZ x NOPT interaction explained ~8% of the yield variation. Mean yields in the ample nutrient plots were highest in the Active Ganges Floodplain (AEZ-10) and Young Brahmaputra and Jamuna Floodplain (AEZ-8), and were consistently lower in the Old Meghna Estuarine Floodplain (AEZ-19). Across all AEZs, iterative random forest analysis indicated the accuracy of MSE prediction of maize yield nutrient response as K (30%), N (27%), P (23%), and Zn (5%). There were variable N and K responses across regions. The N response was associated with heterogeneity in soil organic carbon (SOC) and total N, while the K response was largely influenced by soil texture. The indigenous nutrient supply in all AEZs, determined from omission plots, were poorly correlated with soil test values, highlighting limitations of existing analytical methodologies for tropical soil testing to provide nutrient management recommendations.

More precise estimation of variability in land types, soil characteristics, and ISF, along with information on farmers’ crop management practices, and robust attainable yield estimates will be required to develop precision nutrient management guidelines for maize in the Eastern IGP.