Potential crop yields are sometimes tens of percent higher than expected

Food security policy for governments, companies and organizations such as the United Nations depends in part on global models that determine current and potential crop yields around the world. Scientists from Wageningen University & Research (WUR) and the University of Nebraska-Lincoln in the US warn of the shortcomings of these “top-down” global models.

According to them, the models rely too much on rough data about weather, soil and crops and are not adequately fed and validated with local data. In an article in Nature Food, they argued in favor of improving estimates through structural use of locally collected data and testing models more often against local trials.

In their article, the researchers compared the performance of two commonly used top-down models (Global Agro-Ecozones and the Agricultural Model Comparison and Improvement Project) with the performance of their bottom-up approach, the Global Yield Gap. atlas.

Methodically underestimated
Top-down global model estimates, on average for a large country like the US or a continent, often but certainly not always, still give a reasonable picture, but if you look at specific regions or smaller countries, their results say co-author Professor Prof. Martin van Atterzoom of the WUR Plant Production Systems Chair Group:

As an example, he cites the results of global models for rice in Asia and maize in sub-Saharan Africa: “For rice in Asia, potential yields were systematically underestimated by top-down models, whereas for maize in sub-regions—these models show Sub-Saharan Africa has very few differences between high and low yield potential countries.”

Approximate data and lack of tests
The shortcomings of regressive models are due to the fact that databases are generally poor and depend on generated weather data or assumptions about crop calendars. For example, the times when crops are planted and harvested in an area are not always accurately estimated. In global studies, the same model is also used for different crops and for the world at large, while the models have not been tested locally against well-performing trials.

“The potential yield of a crop in a given area can be tens of percent higher than what the ‘top-down’ models assume,” says van Atterzoom. As investors, seed producers, and other parties base their decisions in part on these models, the consequences can be far-reaching. “We cannot make informed decisions about how to enhance food security in Africa or other parts of the world and how to deal with scarce resources like land and water.”

Local Data Integration
According to the authors, the problem can be solved by using structurally local data in global studies. These local data (on weather, soil and crop management) and simulations are already available because they have been systematically mapped since 2011 at the Global Yield Gap Atlas (GYGA) project, which Van Ittersum co-leads.

“We started this project with the University of Nebraska-Lincoln because we found that global models in specific countries and territories were systematically wrong. With the help of local experts, we now have high-quality, locally relevant data from about 70 countries. We now know the yield gap for some of the world’s largest agricultural crops. 80% of the Earth’s surface Such a “bottom-up” approach requires a lot of work, but generates very valuable information for policy makers and researchers who are concerned with the question of whether different countries and continents can feed themselves in the future and how countries and continents can feed themselves in the future and where there are more or less opportunities.”

Source: WUR