The scientists undertook comprehensive monitoring of crop growth and development, yield and yield components, and plant nutrient content for all three sites in Bangladesh. They found yields of all component crops were generally much lower than potential or maximum recorded yields in the region. The reasons for low yields with farmer practice (FP) rates and soil test-based fertiliser (STB) rates are not fully known, however major contributors were undoubtedly:
1) for rice, lodging (where plants fall due to wind, rain, flooding or pest damage, or because the stems are too tall or too weak to stand erect during the grain-filling stage) often caused yield loss during the reproductive stage of every crop; it was exacerbated by heavy rain and wind damage, and was most serious in higher nitrogen rate treatments, and to some extent with phosphorus and potassium deficiencies.
2) for wheat, high temperatures during flowering and grain filling; phosphorus deficiency in all treatments, years, and sites; potassium deficiency in all treatments/years at Nashipur, in 1998-99 and 1999-00 at Joydebpur, and in 1997-8 and 1998-9 at Ishwordi.
3) for mungbean, lodging and physical flower and pod damage and waterlogging due to very heavy rain and winds during the reproductive stages
4) for maize, lodging, cob damage and waterlogging due to very heavy rain and winds; phosphorus deficiency, especially at Nashipur
The modelling studies suggested that all crops were sown later than the optimum dates desirable to achieve yield potential. The results also highlight the importance of achieving timely sowing in intensive cropping systems where a delay in one crop affects the performance of all crops in the sequence because of the narrow window for establishment. The biggest window is between the harvest of maize/mungbean and establishment of rice. Therefore, techniques for earlier establishment of rice, shorter duration rice varieties, direct drilling or relay sowing of wheat, and relay sowing of mungbean warrant further investigation.
Maize and wheat yields were significantly higher using STB fertiliser rates than FP rates. Total system productivity (TSP) was highest for sequences that included maize, due to the much higher grain yields of maize compared with mungbean. Mean TSP in the absence of nitrogen fertiliser was about half that with nitrogen fertiliser; it was about 2 t/ha higher for STB than FP, and 1-2 t/ha higher at Nashipur and Ishwordi than at Joydebpur.
For STB(+nitrogen) and FP(+nitrogen) treatments the nitrogen balance calculations suggested there was an excess of nitrogen input over extraction by crops at all sites - assuming non-symbiotic biological nitrogen fixation (BNF) in the rice of 40 kg N/ha, symbiotic fixation of 70 per cent of the total nitrogen uptake by mungbean, and no nitrogen losses. Where it was assumed there was no nitrogen fixation, only the maize and mungbean/residues-retained sequences remained in positive balance. The major nitrogen input was from fertiliser, and in the absence of fertiliser N the balance was negative in all sequences.
However, after 3-4 years there were large and significant decreases in total nitrogen (%N) in the topsoil (0-15 cm) of the mungbean-residues removed and maize sequences, and of the control (-N) at Ishwordi. There were smaller but generally significant declines in the same treatments at Nashipur, and generally similar trends at Joydebpur (but the differences were much smaller and never significant). The consistent trend and often significant decrease in soil %N (loss of soil nitrogen from the system) suggest that BNF was overestimated and/or that there were substantial losses of nitrogen (e.g. by volatilisation, denitrification, leaching, runoff) in the nitrogen balance calculations.
All cropping systems at all three sites were profitable. Returns were consistently lower in all treatments at Joydebpur than at the other two sites, consistent with the lower TSP at Joydebpur. At all sites, returns with maize in the rotation were highest, and the difference was greatest at Ishwordi. At each site, net returns of the other five treatments were similar.
Results of the 'validation' of the crop models (CERES and CROPGRO) against the field data for Bangladesh were generally poor, especially for rice and wheat. Factors such as lodging, rain, wind, disease and nutrient deficiency (other than nitrogen) greatly reduced yield potential, and these factors are not considered in the crop models.
Results of the validation of the crop models (CERES and SWAGMAN Destiny) against the Australian wheat data from field experiments and lysimeters were good for a range of parameters including: biomass production over time, nitrogen uptake over time, grain yield, volumetric soil water content at different layers in the profile, root length density, depth to the watertable and evapotranspiration.