ICTworks

Average cocoa yields in Ghana by smallholder farmers are currently 400 kg/ha, far below their production potential of over 1,500 kg/ha. Dominant factors contributing to this current yield gap include climate, cultural practices, and the soil. Among these, long-term soil degradation and, in particular, soil infertility is recognized as one of the main factors limiting cocoa yields

There is huge potential for precision agriculture services to help farmers identify soil practices that lead to higher productivity and profitability, lower input costs and optimize input use, particularly fertilizer use. However, a major limitation in assessing farm-scale soil suitability and limitations lies is acquiring accurate soil property data.

In Map to the Future (M2F): Integrating soil mapping into cocoa farm development plans in Ghana, Grameen Foundation USA, in collaboration with The Sustainability Innovation Lab at the University of Colorado, Boulder, and the USDA Agricultural Research Service, evaluated the use cases for integrating site-specific soil data with traditional regional-to-global soil maps and agronomic-support applications in an effort to provide location-specific soil information that smallholder cocoa farmers can use to improve soil health and overall farm productivity.

They studied LandPKS, a site-specific soil data mobile application and FarmGrow, which is an agronomic advisory application that monitors the adoption of good agricultural practices and provides an 8-10 year financial plan for the cocoa farmer.

Comparing LandPKS vs. Soil Map Data

The first analysis sought to compare site-specific soil data using LandPKS with the relative accuracy of different soil map products like HWSD and WISE which are traditional soil maps and SoilGrids which is digital soil map data.

This initial comparison of site-specific soil properties collected with the LandPKS app showed that all of the soil maps underestimated soil variability in Northern Ghana, and often mis-identified the soil. SoilGrids was found to be the most similar to LandPKS soil data in Northern Ghana; however, HWSD and WISE were found to be the most accurate in Ghana’s cocoa growing regions (Ashanti and Western regions) compared to LandPKS, illustrating the variable accuracy of soil map products.

Combining LandPKS and FarmGrow Data

The second analysis compared LandPKS with soil data integrated from FarmGrow data among 225 sites from 75 farms in the Ashanti and Western regions. The analysis found there was significant improvement in understanding soil limitations at cocoa farms with the integration of LandPKS and FarmGrow data.

For example, the FarmGrow assessment of soil physical condition only rated 1 farm as ‘Bad’ and 74 as ‘Good’, whereas the more detailed LandPKS soil assessment rated 47 farms as ‘Bad’ and 28 as ‘Good’.

Using LandPKS, FarmGrow, and HWSD/WISE

The third analysis explored how site-specific (LandPKS) and soil map data (HWSD/WISE) theoretically could be used to provide site-specific agronomic advice at the 225 FarmGrow sites.

The team downscaled the FAO’s Agro-Ecological Zones (AEZ) soil suitability modelling framework, which uses HWSD soil data to calculate crop-specific soil quality ratings for 54 different crop types at three different production input levels (low, intermediate, high) to translate site-specific soil information into crop-specific soil suitability ratings at the 225 locations.

This assessment found that relying on soil map data alone may lead to an under or overestimation of land suitability and thus fail to identify the soil management actions needed to improve cocoa yields.

Site-specific data is therefore needed for individualized farmer agronomic recommendations. When smallholder farmers have limited resources (financial, human, etc.), these differences could mean success or failure or limited impact of the investments they are making.

Gigi Gatti of Grameen Foundation noted that extension agents always talk about fertilizer and good agricultural practices, but what you can do with soil is limited. This constraint is an ah-ha moment for agents and farmers. We can only push soil so far.

For the smallholder cocoa farmer in Ghana, every single cedi counts. We have to be careful when giving recommendations for farm investments to make sure they are always optimized for that farmer.

Recommendation: Farmers Use All 3 Approaches

In conclusion, while existing soil map data would provide very limited benefit to products such as FarmGrow that provide individualized coaching, integrating site-specific data such as that from LandPKS and leveraging the AEZ framework would be beneficial for two reasons:

• Some observational soil data currently collected by FarmGrow agronomists and farmers could be replaced with data collected through the LandPKS app, therefore reducing agronomist or farmer error, and
• Site-specific soil recommendations could go beyond the focus on one crop, in this case cocoa, and identify crops most suited for intercropping with cocoa based on their soil’s condition, in an effort to improve soil fertility through increased biodiversity on cocoa farms.

As Jeff Herrick of USDA and a soil scientist for LandPKS pointed out, the real challenge is how to integrate and support ag-tech solutions so that they are simple enough, site-specific enough, and also cheap enough to sustain within the smallholder farmer value chain.

Peter van Grinsven of Brightlife Farming also added that, while FarmGrow and LandPKS can be seen as part of an ecosystem of apps, it’s important we aim for an application that does try to do most things for a farmer in a simple way, even if multiple applications are integrated to build off their complementarities.

By Bobbi Gray, Research Director, Grameen Foundation