Data-driven Advisories AgWise for maize in Kenya The AgWise Development Team 2023 Pieter: The idea is to organise this in a way that there can be discussion and interaction, so we opted to have the strategic research topics presented in different rooms and allow the participants to move around between them. There is 75’ foreseen for the session, and we will make participants move every 25’ unless we see people moving around faster on their own initiative.   In terms of objectives, first is to get everyone informed of the strategic work planned, but we also want people to understand how the strategic research will strengthen the initiative, and we want people to reflect on the linkages between the research in INNOVATE and the activities in the other work packages. The initiative is complex, and many people are familiar with the activities they are involved in, but not always the bigger picture. So we want to explore how the strategic work can be better linked to the other activities across the initiative, and identify opportunities for new research. 1 AgWise overview Data analytics framework to develop tailored agronomic recommendations Joint effort across centers & experts Enriched by diverse expertise, experiences Continual learning AKILIMO, NextGen, EDACAP, GAYA, … Legacy + current data Carob, partners (IFDC, OAF, Kilimo, CoW…) Carob: http://carob-data.org Co-development High level of partner engagement is encouraged Modular design Advance & maintain independently,  adds flexibility, adoptability  Target crops Maize, rice, wheat, potato, soybean, cassava Key elements of AgWise   Tailored fertilizer advice Demand mapping Climate forecast Good agronomic practices Optimal planting dates and cultivars Support to advisory platforms 3 Generalized AgWise workflow Optimised fertilizer advice for increased agricultural productivity in Kenya Purpose: Develop evidence-based recommendations on fertilizer requirements and fertilizer types to achieve crop production targets at sub-county level Requirements: Consider smallholder farming practices Map soil nutrient constraints Adjust based on seasonal weather forecasts Compare different fertilizer formulations Permit subcounty-specific production targets 1. For the county governments: Calculate total fertilizer quantities to achieve crop production targets (subcounty level) Decide on best-suited fertilizer types by compare total requirements and associated costs to achieve production targets using different fertilizer formulations 2. For the e-voucher fertilizer subsidy programme: Determine maximal fertilizer quantities to subsidise for individual farmers based on landholding and other profile information in the national farmer database Reduce investment risks by adjusting subsidies based on seasonal forecasts 3. For extension services: Disseminate optimised fertilizer advice to farmers, providing recommendations on best fertilizer types and rates for individual crops and production targets Inform subsequent interventions by tracking fertilizer use, productivity and profitability as part of the national farmer database Kenya maize: Purpose of the fertilizer recommendation Kenya maize: Data sources used for current functionality On-farm fertilizer response Current production levels Data from the State Department for Crop Development (kilimo.go.ke) Kenya Maize Production by Counties 2012-2018 Data from digitised on-farm fertilizer experiments from published scientific articles using the carob workflow (https://github.com/reagro/carob): 4191 observations from 568 unique trial-seasons across East Africa Yields under smallholder management Maize grain yield, current fertiliser use and crop management from >16,000 smallholder farm surveys used to estimate nutrient-limited (lower bound) and attainable yields (upper bound) by sub-county for fertilizer response Digital soil information Open-access gridded soil information to predict soil nutrient supply capacity using machine learning algorithms, trained using on-farm fertilizer trial data AgWise workflow implemented to generate fertilizer advice that: Accounts for soil nutrient deficiencies using on-farm data and digital soil maps; Considers fertilizer response under smallholder management (survey data); Considers multi-year data and rainfall, but permits refining based on seasonal forecasts; Allows customised yield targets at (sub) county level relative to current production; Can generate recommendations for different fertilizer formulations. Example: recommended application rates (kg ha-1) for NPK17:17:17 as basal and CAN as top-dress fertilizer  to achieve a 30% increase in maize production above current mean annual production. Kenya maize: Evidence-based fertilizer recommendations (subcounty) 8 AgWise – A cross-CGIAR team effort  Meklit Chernet, Pieter Pypers, Amsalu Tilaye, Degefie Tibebe, Eduardo Bendito, Siyabusa Mkuhlani, Amit Srivastava, Payel Ghosh, Vimbayi Chimonyo, Arturo Gonzalez, Louise Leroux, Lizeth Llanos, Jane Mugo, Andrew Sila, Camilo Perez, Kristin Persson, Patricia Moreno, Julian Ramirez, Ani Ghosh, Joao Silva, Robert Hijmans, Anton Urfels, Ahmed Kheir, Wuletawu Abera, Sammy Barasa, Bester Mudereri, Camila Bonilla, Tewodros Mesfin, Diego Agudelo, Mukund Patil, Medha Devare image1.png image12.png image13.png image14.png image15.png image16.png image17.png image18.jpeg image19.png image20.png image21.gif image22.jpeg image23.png image24.png image25.png image26.png image27.png image28.png image29.png image30.png image31.png image32.png