An R package for boundary line analysis

BLA: An R package for boundary line analysis

Posted on July 30, 2024

Boundary line stakeholder workshops in Nairobi and Harare.

As part of a joint PhD under the University of Nottingham and Rothamsted Research Program in International Agricultural Development, Chawezi Miti and his supervisors Professor Murray Lark (Nottingham), Dr Alice Milne (Rothamsted) and Professor Ken Giller (Wageningen University) have developed a user-friendly library of functions for the open-source platform R. This library, called BLA, will facilitate the reproducible application of boundary line analysis, including data exploration, model fitting, and post hoc analysis. You can access and download the package through this linked page.

According to the Food and Agriculture Organization of the United Nations, a 60% increase in food production will be needed to meet the needs of the world population projected for 2050. This could be met in various ways, but it is widely agreed that ensuring that the land currently in production meets its yield potential is more feasible and sustainable than taking more land into production. This is sometimes called closing the yield gap, i.e. exceeding yield limitations. In addition, it is also agreed that yield differences are poverty traps for rural communities (Tittonell and Giller, 2013)(1); failure to realize potential yield represents a loss of income for small producers.

To close the yield gap in a given context, we need to understand what the limiting factors are. If we were to obtain data on crop yield in a region and an associated variable that is a potential limiting factor, (such as available soil water capacity), then one way to interpret the data is to model the yield upper bound for different waters. contents. The figure below shows a graph for a hypothetical example, the red line is the modeled limit and for a given site, A, the difference between the limit value and the observed yield, represents the production gap.

Boundary line methods have been used primarily in agronomy, but have also been applied to biophysical modeling problems in plant physiology, microbiology, ecology, plant/atmosphere interactions, and forestry. In any of these areas, the BLA library provides a number of methods to help the researcher extract meaningful information from the data. The boundary line approach is suitable for the analysis of biological response datasets from surveys, i.e. cases where multiple potential limiting factors of the biological response occur but are not controlled, as is done under experimental conditions. In this case, the highest expected biological response (also called limit value, eg, crop yield) for a given value of a factor of interest (eg, soil N) can be determined. In addition, the most limiting factor for the biological response can be modeled from the observed potential limiting factors. The 1.0.1 BLA for the R platform has been published on CRAN and is available for download.

This tool was developed in consultation with potential users from the centers of the Consortium of International Agricultural Research Centers (CGIAR) (CIMMYT, IITA). Workshops attended by CGIAR researchers in Nairobi and Harare in 2023 provided valuable feedback on the BLA library and the methods it encodes, which has been incorporated into the published package to improve its functionality and user experience.

Figure 2: Stakeholder workshops on delineation methods in Nairobi and Harare

By providing a practical tool for implementing boundary line analysis, the BLA R package represents a significant advance in agricultural research. It promises to play a critical role in bridging the yield gap by enabling researchers to understand and optimize crop yield potential under real-world conditions. For more information about the BLA package visit https://chawezimiti.github.io/BLA/.

Additional references and links

(1) https://doi.org/10.1016/j.fcr.2012.10.007

(2) https://doi.org/10.1080/00221589.1972.11514472.