The African Cassava Agronomy Initiative (ACAI) project has developed and deployed AKILIMO in southern Nigeria and Tanzania. AKILIMO is a suite of decision support tools (DSTs, Fig. 1) that provides tailored recommendations to farmers and extension service providers based on digital soil and weather data combined with market and price information, and farmers’ cropping objectives and risk attitude. They are available in diverse formats: digital to conventional paper-based.
This innovation has been validated for function, confirming that using the tool effectively results in net revenue increases for at least 75% of users; and for architecture, ensuring a positive user experience with easy interpretation of recommendations.
AKILIMO’s core components include: (i) a database holding primary georeferenced data on crop yield responses along with essential covariates; (ii) a suite of models and algorithms to predict yield response and yield gaps to provide tailored advice optimized for maximum return on investment; (iii) a user-friendly mobile app (available on Google Play Store) for extension agents; (iv) simplified paper-based tools to provide tailored recommendations directly to farmers; (v) a detailed user guide, training modules, and farmer-friendly videos that explain the principles of the recommendations; and (vi) Arifu’s chatbot service, providing tailored recommendations directly through simple mobile phones as well as Viamo’s 321 service.
Site-specific recommendations are available for fertilizer use (FR), fertilizer blending (FB), cassava weed management and the best planting practices (WM/BPP), intercropping, with maize in Nigeria and sweet potato in Zanzibar (IC), and increased starch content in the cassava roots and scheduled planting and harvesting (HS/SP) to ensure a continuous supply of cassava roots to the processing industry.
Figure 1. Schematic presentation of the AKILIMO framework, stretching from field experimentation to understand agronomy by environment interactions taken to scale through crop and geospatial modeling (left), to a cloud-based prediction engine and database that feed diverse formats (printable guides, smartphone app or USSD, and IVR based services) of the individual decision support tools (center) to the delivery to the farmer, via extension agents or direct as in the case of USSD and IVR-based services (right).
ACAI is funded by the Bill & Melinda Gates Foundation and is implemented by IITA together with partners from strategic international, regional, and national research institutions, and partners from the private and public sectors referred to as primary or development partners.
During the inception workshop in 2015, the primary partners identified main issues that if addressed, would help resolve drawbacks preventing the cassava sector from achieving its full potential.
The requests from development partners formed the foundation of the ACAI demand-driven approach used in implementing the project: to develop solutions in response to specific needs of partners. The issues were grouped into six use cases: site-specific recommendations for fertilizer use (FR), site-specific fertilizer recommendations for fertilizer blending (FB), six steps to weed management and best planting practices (BPP), intercropping (IC), increased starch content in roots (HS, and scheduled planting (SP) to ensure a continuous supply of roots for processing.
The critical roles of national agricultural research system (NARS) partners include implementation of field trials to collect the data to calibrate and validate AKILIMO tools, while strategic international research partners contribute to the back-end prediction framework. IITA coordinated the trials implemented by the NARS and primary partners in the field.
The National Root Crops Research Institute (NRCRI) Umudike, the Federal University of Agriculture in Abeokuta (FUNAAB) in Nigeria, and the Tanzania Agricultural Research Institute (TARI) collaborated in setting up and running thousands of on-farm trials for several use cases generating ground truth data to test the factors influencing production under real conditions in farmers’ fields. Soil and plant samples and other field
crop management data were collected and used in the prediction engine generating the recommendations. ACAI applied a coupled modeling approach combining crop growth models such as the Light Interception and Utilization model (LINTUL) and the Quantitative Evaluation of the Fertility of Tropical Soils model (QUEFTS) with geospatial statistics and economic optimizer algorithms to generate recommendations optimized for maximum net return on investment for FR, FB, and SH/SP. At the same time, through their extension networks, primary partners mobilized farmers so that most of these research trials could be conducted as on-farm but researcher-managed trials. At the close of 2019, for the cassava intercropping use case, ACAI had run 926 field trials covering 5 states in the south of Nigeria, and 4 ecological and administrative zones in Tanzania.
Successful evaluation (Fig. 2) and approval of innovations is a crucial step towards the completion of the development process. In late 2018 and early 2019, ACAI released the earliest version of the AKILIMO decision support tools that provided recommendations based on the data collected from the earliest research phase. Primary partners led the validation process with backstopping from NARS and IITA researchers. This way, ACAI and partners conducted validation trials in the 2018/2019 cassava growing season involving 4,084 households in Nigeria and Tanzania.
The project is currently running a second series of AKILIMO validation trials with partners to evaluate the accuracy of the advice provided and further recalibrate the models to address possible discrepancies in the recommendations. Given the highly variable soil and agroecological factors in the regions, ACAI aims to provide recommendations that result in increased returns-on-investment for over 75% of users, relative to standard recommendations or current practice.
Figure 2. Example of validation results for the cassava-maize intercropping decision support tool in southern Nigeria. In about 75% of the cases the application of fertilizer (fixed rate of 90 N – 20 P – 37 K kg/ha) plus increased maize density (40,000 plants/ha) was more profitable than the control of no fertilizer application and low maize density (20.000 plants/ha) even when this was not the recommended practice. On average, fertilizer application increased revenue by US$1,200/ha, and losses were small. Increasing only maize plant density had little effect.
The validation trials also provided opportunities to not only technically validate the performance of the prediction engine set in place, but also improve the ‘look and feel’ and user experience and create ownership and trust in the tools. Positive results from validation trials allowed the project to move forward in the development process and look at suitable formats of packaging and delivering the tools to the end-user. Through a series of participatory workshops, ACAI together with researchers, farmers, extension agents, representatives from project partners and consulting with experts, considered the demographics of the end-users of AKILIMO, socioeconomic factors, skills, literacy levels, and ease of access to available technologies. ACAI also evaluated possible formats for the tools, including Interactive Voice Response (IVR) system, Unstructured Supplementary Service Data (USSD), a smartphone application, and simple printable guides. All these formats have advantages and disadvantages, and none of the stakeholders expressed concerns about any of these formats, but rather expressly requested that all these be made available for testing and use.
The project’s monitoring, evaluation and learning team were instrumental in crafting tools to capture and effect partner and end-user feedback, conduct feedback sessions, and data collection.
The final versions of the ACAI Decision Support tools were released in July 2019. During the pre-launch function, partners and the ACAI team branded the agronomy advice tools AKILIMO, with a tag line ‘we know cassava’ as the official trademark identities. AKILIMO is a portmanteau of two Swahili words: akili which loosely translates to intelligence or smart and kilimo which means agriculture. A logo was designed that combines elements of digital agriculture and a human face to reflect the importance of combining science-based advisory with user-centred design to successfully deliver agronomic recommendations to smallholders.
The focus for ACAI is to expand the reach and influence the adoption of the AKILIMO tools through strategic scaling and dissemination. ACAI has already brought on board extensive and diverse talents with experience in scaling new technologies.
In the same spirit of strong partnerships and sustainability, ACAI seeks to expand its secondary partnership to reach more farmers outside the project while intensifying efforts to influence the adoption of the tools within the primary partnerships. Secondary partners that are linked with existing primary partners have been identified. This helps long-term use of AKILIMO tools within their networks of smallholder farmers.
At the end of the project in 2020, ACAI would have impacted at least 120,000 households in Nigeria and Tanzania.
Authors: David Ngome, Meklit Chernet, Christine Kreye, Stefan Hauser, Thompson Ogunsanmi, Frederick Baijukya, Theresa Ampadu-Boakye, Friday Ekeleme, Godwin Atser, and Pieter Pypers