Review of Agroecozones for Use in Yield Gap Analysis.

Increasing demand for food expected during coming decades will require a substantial increase in crop production. Given limitations for massive expansion of cropland area, it is of critical importance to know where and how to increase crop productivity per unit of land (i.e., crop yields). This can be quantified through yield gap analysis, an examination of the difference between current farm yields and crop yield potential, the maximum attainable yield per unit land area that can be achieved when pests and diseases are effectively controlled and nutrients are non-limiting. ‘Point-based’ estimates of yield gaps, either derived from research plots or simulation models, are available only for a limited number of sites due to economic and logistic constraints and lack of weather, crop, and soils data. Therefore, it is necessary to understand how a point-based estimate of yield gap from a specific location can be scaled up to a wider extrapolation domain. To define the extrapolation domain of a yield gap estimate, one can make use of agroecological zones (AEZ), defined as geographic regions having similar climate and soils relevant for agriculture. In this study, 2 AEZ methodologies and four existing AEZ schemes are reviewed and analyzed, focusing on their applicability to scale up point-based estimates of yield gaps to a larger geographic area. Preliminary analysis indicates that AEZ schemes used by groups such as the Food and Agricultural Organization of the United Nations and the International Institute for Applied Systems Analysis are too coarse and climate variability within zones too high for use as extrapolation domains. Clusters based AEZs show promise as extrapolation domains, but require several key, subjective input for their creation.  Statistically derived matrix AEZs hold the most promise as transparent, robust, global extrapolation domains.