Abstract
No-till is an agricultural practice widely promoted by governments, development agencies, and agricultural organisations worldwide. However, the costs and benefits to farmers adopting no-till are hotly debated 1–4. Using a meta-analysis of unprecedented study size, Pittelkow et al.5 reported that adopting no-till results in average yield losses of -5.7%, but that these losses can be limited with the added implementation of two additional conservation agriculture practices - crop rotation and crop residue retention, and in dry environments. They claimed that, as a result, resource limited smallholder farmers, that are unable to implement the whole suite of conservation agriculture practices are likely to experience yield losses under no-till. In a re-evaluation of their analysis, we found that they overly biased their results toward showing that no-till negatively impacts yields, and overlooked the practical significance of their findings. Strikingly, we find that all of the variables they used in their analysis (e.g. crop residue management, rotation, site aridity and study duration) are not much better than random for explaining the effect of no-till on crop yields. Our results suggest that their meta-analysis cannot be used as the basis for evidence-based decision-making in the agricultural community.