DNA fingerprinting for crop varietal identification: fit-for-purpose protocols, their costs and analytical implications

Abstract

In an effort to understand the diffusion of improved agricultural technologies in the developing world, researchers have long sought to measure farmers’ adoption of improved crop varieties. A number of different approaches have been used: eliciting the opinions of informed experts, collecting self-reported data from farmers as part of household surveys, or imputing varietal areas from data on seed sales. Little information has been available, however, about the validity of these approaches, which undoubtedly lend themselves to various biases and multiple sources of measurement error. In the past 15 years, as a result of several technological breakthroughs in the laboratory, the cost of genotyping has fallen significantly. It is now possible to mainstream the use of DNA fingerprinting for estimating varietal adoption. In experimental studies, data on varietal adoption can be collected using all three methods (expert opinion, farmer self-reports, and DNA fingerprinting of tissue collected from farmers’ fields), allowing us to use the latter as an objective benchmark against which the earlier methods can be judged. In most experimental studies that have used this benchmarking, we are finding significant differences between the estimates from DNA fingerprinting and those established using earlier methods. In some cases the prior estimates underestimated the true extent of adoption, but in many cases older methods overestimated adoption by farmers. However, we are only scratching the surface of the insights we stand to gain from scaling up DNA fingerprinting. The method can be applied to a host of second-order questions, such as varietal turnover, the age of varieties in farmers’ fields, and the efficacy of the seed system in providing high-quality seed to farmers or of the extension system in promoting new varieties. To inform this process of scaling up, the CGIAR Standing Panel on Impact Assessment (SPIA) commissioned this report. Researchers who want to use DNA fingerprinting to analyze the adoption of improved crop varieties in farmers’ fields face multiple methodological options. They must make careful decisions to match protocols for sampling and analysis to their specific analytical needs. This document synthesies what we have learned about the state of the art regarding that process. This is a field that is rapidly shifting— the technology is changing, and with it the questions we can ask, thus this document should be seen as a set of best practices as of today. The interdisciplinary team of authors assembled for this study (from genetics, data science, and economics) used evidence from multiple empirical studies, supplemented with their own research and consultations with experts. Much of the evidence was generated by studies carried out in the context of the five-year SPIA program “Strengthening Impact Assessment in the CGIAR” (SIAC), which ran from 2013 to 2017. Other fingerprinting studies were run independently by individual CGIAR centers, and these too played a significant role in informing the material presented in this report. The Bill and Melinda Gates Foundation, and program officers Greg Traxler, Marianna Kim, and Richard Caldwell in particular, helped convene discussions of the methodological issues related to DNA fingerprinting, and the foundation provided significant grant funding to SIAC and other fingerprinting studies. Indeed, the foundation offices in Seattle hosted two methodological workshops on DNA fingerprinting—in August 2014 and again in January 2018—and many of the perspectives presented in this document were first debated by participants in those two events. We thank them all. SPIA is grateful for the work that the author team carried out in producing this report. They have gone well beyond the original vision for this document, taking the initiative to get updated cost estimates and methodological details from alternative providers of genotyping services, as well as exhaustively reviewing the scientific literature. Given the speed at which the technology and commercial landscape for genotyping services is changing, we will likely need to revisit this document in a few short years to update it. The case for doing so will be all the stronger if it is widely used in the interim. We hope that CGIAR researchers and the broader agricultural research community will find this document useful and aspire to contribute the empirical evidence to inform the next edition.