Global patterns of nutritional stability
We lack straightforward ways to measure the ability of agriculture to produce nutritious food through space and time in the face of chronic disturbance and acute shocks. Considering crops and their constituent nutrients as an ecological network we quantify nutritional stability at a global scale
Agricultural systems are now evaluated not only in terms of production, but by the nutrient content of crops produced. This re-focus is important; food security policy is pivoting from evaluating outcomes solely in terms of total yield towards ensuring the production of a variety of affordable, nutritious, culturally appropriate foods that meet people’s dietary requirements. These nutrition-sensitive approaches promise to overcome malnourishment and micronutrient deficiencies by re-focusing crop diversity and nutritionally rich foods as production objectives. Yet an open question is whether diverse crop mixes can also provide sufficient nutrients when confronted with shocks or stressors, such as drought or crop failure
In our paper we develop an approach to measure nutritional stability – which we define as the capacity of a food system to provide nutrients despite disturbance. Using 55 years of FAO production data across 184 countries, we assemble 22,000 crop-nutrient networks to quantify nutritional stability by simulating individual countries’ crop and nutrient loss. We find that despite increases in crop diversity, nutritional stability has remained stagnant or even declined in many regions. This is in part due to a saturating relationship between crop diversity and nutritional stability (Fig. 1). However, we also observed that over this time period crop degree – the average number of nutrients provided by each crop in a network – declined, indicating that crops being added into networks appear to provide fewer links to nutrients not already present in the food system.
This work was inspired by an earlier study by Stephen Wood of The Nature Conservancy that considers individual nutrients as crop functional traits. Based on methods common in functional ecology, this study creates species-by-trait – that is to say crop-by-nutrient – matrices and calculates food system functional diversity. In Dr. Wood’s words, “Plant traits—such as leaf area and photosynthetic rate—are important traits to ecosystem processes; functional trait diversity metrics connect the diversity of these traits as the mechanism by which plant diversity impacts ecosystem functioning. Analogously, I use crop nutrient content as a functional trait that would be the mechanism by which crop diversity impacts nutrition”. This work was foundational to our study because it transplants methods from community ecology to food systems in order to gain novel insights about the links between biodiversity and human nutrition.
I encountered Stephen’s paper at a time when I had been considering how to measure the robustness of ecological networks, such as plant-pollinator communities. The commonly used method of removing network nodes (e.g. plant species) and tracking the resulting ‘secondary extinctions’ (e.g. loss of pollinator species) has an elegant simplicity. It was an exciting ‘ah ha’ moment when I realized that crops and their constituent nutrients could be considered an ecological network to which functional ecology methods (by way of information science) could be applied (Fig. 2). As an interdisciplinary scientist with training in landscape and community ecology I was excited to adapt this approach and start exploring diversity-function relationships in food systems.
This interdisciplinary work is the result of a similarly diverse team. Meredith Niles, a professor of Food Systems at the University of Vermont, brought expertise and expansive vision to the project. When I pitched the idea of nutritional stability to her, she immediately identified new angles (considering imports, comparing nutritional stability between regions and economic status). Ben Emery joined from UVM’s Complex Systems Center and brought key insights on how to parameterize and optimize the removal algorithm. It’s been a fun team to work with and this research would not have happened without them.
As an early career researcher, there is also a lesson in perseverance here for me. Originally, I pitched the idea of nutritional stability as part of a 2018 postdoc fellowship proposal. I was not awarded that fellowship, but nonetheless, I couldn’t get the approach out of my head. A few months later, I showed up to Meredith’s office and pitched the idea to her; she loved it! Her enthusiasm and mentorship helped put wind in the sails of this work. As an early-career scientist, this project has taught me that sometimes rejected ideas are worth rejuvenating.
To me, this paper represents the ‘proof of concept’ – that the method is valuable, replicable, and scalable. The next steps for us are to apply this method more broadly and with greater complexity. In the present work we use simple extinction sequences (random and ranked by node degree). In the future though, we are excited about exploring the impacts of different extinction sequences of crops. For example, how does nutritional stability change if we order crop removal based on within-country production? Or based on the climate-sensitivity of crops? We can imagine that these more realistic removal scenarios could generate very different stability outcomes.
Target 2 of the Sustainable Development Goals stresses connecting crop diversity, nutritious diets and resilient farming systems as essential components of food security. Our manuscript is timely because it provides a straightforward approach to evaluate a food system’s nutritional stability that is generalizable across different levels of aggregation (e.g. household, community, national food system). We add to the toolbox of nutrition-sensitive agriculture because our approach could be used to identify resilient food systems, detect weak links in nutrient availability, and evaluate if stability-focused interventions actually work. We think there is a lot of work to be done applying this method across scales and testing for links between actual nutritional outcomes. Stay tuned!