ADVANCES IN RESEARCH ON RHIZOBIOLOGY AND AGRONOMY
Understanding variability in yield
Photography: Taskscape Media
Some N2Africa farmers genuinely benefited from applying inoculants and fertilizer, whereas the other farmers’ yields remained the same. The results of our attempts to understand the underlying causes of variation in yields have been disappointing. Predicting the likelihood of success remains difficult.
A key benefit of large-scale projects such as N2Africa is the opportunity to test yield-improving technologies in a large number of farmers’ fields. During the project, we monitored farmers who planted a small try-out, consisting of a simple comparison of the best local variety of the grain legume with and without inoculation and phosphorus fertilizer, and with inoculation and P-fertilizer combined, to give four distinct plots in each farmer’s field.
We collected data from these try-outs on the type of field, the management practices that the farmer used (such as weeding, applying manure, planting in rows or not), as well as a number of household characteristics. The yield of the try-outs was measured by field assistants at the end of the season.
Surprisingly similar patterns
Although the farmers’ try-outs in the various countries differed in terms of climate and soils, we observed surprisingly similar patterns in yields and responses to the inputs (inoculant and P-fertilizer). The yields that farmers got without inputs varied widely, and so did the response to the improved practices – often ranging crop failure to yields close to the genetic potential of the legume at around 4 tons per ha (see for instance results from try-outs with soyabeans in northern Nigeria in Figure 1). While some farmers gained real benefits from applying the inputs, others risk investing in inputs while their yields remain the same.
Limited climbing bean yields in Uganda, caused by a diversity of variety, environmental and crop managment factors.
Photo: Esther Ronner
Understanding the variation in crop response is key to being able to predict the likelihood of success of the technology for farmers. N2Africa therefore studied the variation in legume yields and the response to inputs in order to find out which practices worked best, under which circumstances (climate, soils) and for which types of farmers (poorer or wealthier, men or women, producing for home consumption or for the market).
Little predictive value
The attempts to understand the variation gave disappointing results. For instance, differences in year, farm size, plant establishment, total rainfall and soil pH explained 40–60 per cent of the variation in soyabean yield response to inputs in northern Nigeria. These variables had little predictive value across locations or seasons, though, which means we cannot predict which farmers would benefit from applying inoculant or P-fertilizer in a new district in the next season.
Video: Ken Giller explains variability in yields of climbing beans. Produced by Taskscape Associates
It is unclear why it is so difficult to predict yield responses to inputs. It is possible that we do not have sufficiently accurate information about the crop management (such as planting densities, sowing dates and weeding) and environmental variables (soil and climate). Collecting more detailed information from larger numbers of try-outs (we had about 200 per country per season) would require an enormous investment in well-trained field staff.
The research explains why farmers are reluctant to invest: the risk may simply be too high
Moreover, it is questionable whether more data would give better predictability for the yields and responses. A variable such as rainfall for the next season is hard to predict and analysing soils from every field is not feasible.
The finding that there is wide variation in yields and responses is valuable in itself. It makes clear that all farmers may not benefit to the same extent and it explains why farmers may be reluctant to invest in a certain technology: the risk may simply be too high. Testing a technology on a large number of farmers’ fields is therefore certainly a valuable approach for any future project.
Soyabean plants grown in problem soils in pots and fed with different nutrient solutions to identify which nutrients are critically deficient. Photo: Samson Foli
Studying specific factors in detail
A better starting point for understanding yield variability may be to study specific factors in detail: what the effect of late planting is on the response to fertilizer, for instance, or whether inoculation still works well in sandy soils. Such factors can be studied in replicated trials on a small number of representative farmers’ fields. This information will help develop decision support tools that give recommendations on which practices to apply under which circumstances.
