The impact of conservation agriculture on smallholder agricultural yields: A scoping review of the evidence

• Existing reviews were qualitative or used overly simplistic approaches to synthesis.
• Renewed rigor is needed in reporting results from CA and agronomic research.
• Required minimum datasets for future studies will enable reviews with meta-analysis.
• Immediately following conversion to zero tillage yields were often reduced.
• Lower zero tillage yields may ameliorate with time but long-term studies are sparse.

Widespread implementation of conservation agriculture (CA) in North and South America and Australia suggests significant farmer profitability achieved through some combination of sustained or increased agronomic productivity and reduced input costs. Many believe similar agronomic benefits can accrue to smallholder farmers in sub-Saharan Africa (SSA) and South Asia (SA) for a broad array of crops and farming systems despite marked differences in biophysical and socio-economic environments across these regions. Our objectives were to characterize (1) the quality of existing research including an assessment of the relevance of previously published reviews and surveys to SSA and SA, and (2) the empirical evidence from SSA and SA for agronomic benefits derived from implementing zero tillage (ZT) including the identification of knowledge gaps. Mulching and rotation were considered as associated practices within systems. Among surveys and reviews, most syntheses of multiple, independent studies were either entirely qualitative or used overly simplistic approaches to data aggregation. Few reviews used meta-analysis or other rigorous statistics that permit assessment of outcome sensitivity to influential observations; in general, review protocol descriptions were not sufficient to ensure transparency and appropriate handling of common biases. A search and screening of peer-reviewed literature identified empirical studies on conservation tillage in SSA and SA for maize (22), rice (16), cowpea (10) and sorghum (8). In attempting to extract data for an unbiased, systematic review of CA and maize, we found few studies fully reported critical data or meta-data; most common omissions were the univariate statistics required for study use in meta-analyses and critical supporting or explanatory data on soil type, prevailing weather, and management practices including handling of crop residues. In the short-term, ZT generally resulted in lower yields than with conventional tillage (CT). Occasionally these reductions could be linked to direct effects (e.g. increased soil compaction in rice), but failure to adapt other managements (e.g. weed control) to the CA system was a common and confounding indirect effect. Sufficient maize data existed to demonstrate that negative impacts on yield ameliorated with time in some cases accompanied by higher soil water infiltration and soil organic matter, particularly when mulch was added. However, the low number of studies, the missing supporting data and the large variation in treatments made it difficult to infer general direct effects due to mulching or rotation.Well-designed long-term experiments on CA featuring sound agronomic practice and comprehensive documentation are largely missing from the literature. Future systematic reviews addressing agronomic impacts of CA interventions will require appropriate handling of within and between study variance as well as sensitivity analyses and quantitative assessments of publication bias; on-going and future empirical studies must report a minimum dataset encompassing valid statistical measures and comprehensive intervention descriptions that enable standardization and systematic approaches in syntheses. We propose a minimum dataset that is generic to competent agronomy with measurements that are increasingly low-cost and easy to achieve and should therefore be routine in field experiments quantifying and explaining crop and cropping system performance. Until a larger number of field studies provide such quantifying and explanatory data from key crops and representative cropping systems, it is not possible to make strong general conclusions about benefits of CA and ZT on yields and resource use efficiency of smallholder farmers.