Conservation Agriculture Research Updates - April 2026

This paper uses machine learning (ML) approaches as a promising artificial intelligence alternative and complimentary tools to the commonly used crop production models.The study was designed to answer the following questions: (a) Can machine learning techniques predict maize grain yields under conservation agriculture (CA)? (b) How close can ML algorithms predict maize grain yields under CA-based cropping systems in the highlands and lowlands of Eastern and Southern Africa (ESA)? Linear algorithms (LDA and LR) predicted maize yield more closely to the observed yields compared with nonlinear tools (NB, KNN, CART and SVM) under the conditions of the reported study. Overall, the LDA algorithm was the best tool, and SVM was the worst algorithm in maize yield prediction. 

This paper was included because of the debate about use of residues in crop-livestock systems in Africa and their impact on adoption of CA. The area in Zimbabwe for this study is a semi-arid area where yields are low and residue biomass is limited. The paper looks at the economic tradeoffs and profitability of using residues for feeding the livestock or livestock using household surveys. The results show that a maize-macuna rotation can reduce tradeoffs of residues for mulch or feed. The results also show that The poverty effects of all considered alternative biomass options are limited; they do not raise income sufficiently to lift farmers out of poverty. Further research is needed to establish the competitiveness of alternative biomass enhancing technologies and the socio-economic processes that can facilitate sustainable intensification of mixed crop-livestock systems, particularly in semi-arid environments.

Mixed groups of scientists including soil scientists, agronomists, agricultural economists and other environmental scientists, facilitated by experienced senior researchers, were presented with multiple subsets each of three states, and asked to rank the states in each subset with respect to expected yield improvement under CA in South Africa. The results revealed two contrasting groups of conceptual assumptions. One group broadly expected larger absolute yield improvements from conservation agriculture in settings where water is most likely to be limiting and the carbon status of the soil is poor. By contrast, the other group expected larger improvements where water was less likely to be limiting. Modelling the ranking process, could be of more general interest for the elicitation of expert opinion about complex soil, crop and environmental systems.

The effects of contrasted cropping systems have been studied on upland rice yield and N uptake in rainfed conditions: conventional tillage (CT) and CA with a mulch of maize or a legume (Stylosanthes or velvet bean). This study used the "Decision Support Systems for Agrotechnology Transfer" (DSSAT) crop growth model to quantify the soil N balance according to the season and the cropping system in rainfed, upland rice in the hillsides of the Malagasy highlands. The model gave interesting results but the challenge is now to evaluate the model in less contrasted experimental conditions in order to validate its use for N uptake and yield prediction in support to the optimization and design of new cropping systems.

This is an interesting paper that looks at the conflicts between private interests and communal use of resources, for example the free grazing of crop residues. The objective was to assess the impacts of crop residue management on crop productivity in Burkina Faso. They use the AMBAWA model to simulate the flows of biomass and nutrients between crop and livestock systems at the village level scale for 4 types of farmers: subsistence-oriented crop farmers, market-oriented crop farmers, agro-pastoralists and pastoralists. The paper presents some interesting conclusions and suggests that the AMBAWA model can support discussion amongst stakeholders in order to co-design effective arrangements and practices for their sustainable use.

This analysis applied machine-learning techniques to a wealth of published data to create a predictive model of the agronomic outcome of CA relative to conventional practice (CP) based on 21 variables. Results showed that over-yielding of CA relative to CP was driven by a complex of climate, soil, geographic and management variables, and cannot be predicted accurately from precipitation amount or aridity index alone. Success of CA greatly increases with mean air temperature from 20°C and with duration of CA for up to 13years. They suggest that predictive models can be used as an important tool by policy-makers and funding organizations to target financial resources to those regions where CA adoption will have the greatest impact on productivity.

This paper explores the effectiveness of high-resolution topography in characterizing no-tillage (NT) versus conventional tillage (T) surface morphology in order to better understand the hydro-geomorphic processes associated with these crop systems in a clay loam soil in Italy. Surfaces in the NT plots were rougher, had more pronounced slopes and curvatures, sediments with a widespread connection to the plot boundaries, had more irregular flow paths, and had a higher water storage potential due to surface concavities. The NT surface morphology significantly reduces surface runoff, sediment transport, and the off-site movement of agricultural chemicals.

This paper used the ARMOSA process-based crop model to simulate the contribution of different CA components to soil organic carbon stocks sequestration at 0-30 cm depth compared to conventional in three different regions (Central Asia, N and S Europe). Simulations looked at current and future climate conditions.  Five cropping systems were used. The paper concludes the simultaneous adoption of all the three CA principles becomes more and more relevant in order to accomplish soil C sequestration as an urgent action to combat climate change and to ensure food security.