Machine learning for smarter seed selection to reduce risks for Mexican maize farmers
The International Maize and Wheat Improvement Center (CIMMYT) and BioSense Institute jointly won the CGIAR Platform for Big Data in Agriculture Inspire Challenge in 2018 for machine learning for smarter seed selection. This project, which is piloted with maize farmers in Mexico, will help ensure that farmers are getting the best seed variety possible for their farm, allowing them to reduce risk, save money and improve their yields and food security.
The Inspire Challenge is an initiative to challenge partners, universities, and others to use CGIAR data to create innovative pilot projects that will scale. The challenge awards projects with novel approaches that democratize data-driven insights to inform local, national, regional, and global policies and applications in agriculture and food security in real time; helping people–especially smallholder farmers and producers–to lead happier and healthier lives.
Using machine learning, researchers can predict both yields and risks associated with different seeds at a specific location and select the optimal varieties, taking into account climate and geographical data. Using CIMMYT’s data from hundreds of on-farm as well as experimental station sites and a network of seed companies producing varieties for diverse agro ecologies, BioSense will develop machine learning models that predict the performance of seed varieties in particular conditions in order to advise maize farmers in Mexico on what to plant.
We recently interviewed Kai Sonder, GIS laboratory manager and foresight specialist at CIMMYT, and Alberto Chassaigne, maize seed systems specialist at CIMMYT, members of the winning team to discuss this initiative and what they hope to achieve.
Q: How did this project come about?
A: (Kai Sonder) BioSense is a multi-disciplinary research organization affiliated with the University of Novi Sad in Serbia that has been receiving wheat materials from CIMMYT for many years. They won the Syngenta Crop Challenge in Analytics in 2017 with “Portfolio Optimization for Seed Selection in Diverse Weather Scenarios” using machine learning and analytics for predicting farmer seed selection with soy beans. The leader of the winning team, Oskar Marko, contacted us and asked CIMMYT to collaborate on this machine learning project to help farmers get the varieties that are optimal for their geographies, one of the categories of the INSPIRE challenge. We discussed it and I thought about the MasAgro project at CIMMYT working on developing improved maize for Mexico, as I knew they had a lot of sites where they test materials across different geographies, so I suggested to Alberto Chassaigne to use his breeding sites as a trial. Also the MasAgro component “take it to the farmer” were we work on sustainable crop management solutions with thousands of farmers seemed like a good data rich environment to use for the seed user component in all maize growing environments in Mexico.
Q: How does machine learning for smarter seed selection work?
A: (Kai Sonder) Biosense offers the machine learning algorithms and the knowledge of how to create and apply them and CIMMYT will provide info on locations, climate, soils, varieties and certain characteristics, to see where they would work well. This will allow us to predict where specific varieties will work best before they are released, helping seed companies to market the seed to the correct area.
In the future, we will also be able to use this to help farmers better adapt to climate change, by using past and future climate data that will be provided to us thanks to arrangements with the big data platform.
Q: What is the main benefit of this initiative?
A: (Alberto Chassaigne) Machine learning for smarter seed selection will help seed companies ensure that they are selling the best variety possible for a particular region, reducing the risk of marketing the wrong type of seed in their target area. This, in turn, reduces the farmers risk of buying and planting a seed that is not right for their geography, allowing both the seed company and farmers to save money. It can serve as a decision support tool for seed companies, and can also help them to identify future markets using future climate prediction data, therefore helping seed companies to better support farmers as they adapt to climate change.
A: (Kai Sonder) This is extremely beneficial because lot of seed companies we work with are very small and don’t have the capital to invest in these services, but this will be offered free.
In addition to market optimization for companies and risk minimization for farmers, this project will help reduce risk for government programs that subsidize maize seed to support smallholder farmers, as it will ensure the seed they are purchasing is correct. Overall this will support Mexico’s self sufficiency in maize, as smallholder farmers will be able to produce much more by planting the correct varieties in the correct geographies.
Q: What do you hope to see in the future for this project?
A: (Alberto Chassaigne) This project has enormous potential for impact on maize in Mexico, as it is a major crop here and is grown and consumed everywhere in the country. This will allow us to see immediate impact, both with seed companies and the government, as they will be able to make better, more efficient decisions on which seeds to offer to farmers.
In many cases, variety recommendations for an entire state or region may be based on testing done in just one location. With machine learning for smarter seed selection, seed companies and government programs will be able to make more specialized selections that are individualized to farmers needs.
A: (Kai Sonder) This will help seed companies and government programs better serve smaller farmers, a key focus of the MasAgro project. The technology will be able to be used to make recommendations on both improved maize varieties and landraces, or even specialty varieties such as popcorn. While this project is still in very initial stages, it has huge potential to be used in other regions and/or crops, especially with the multiple testing sites and vast network that CIMMYT offers.
Trackback from your site.