Computational simulations to explore sustainable pathways in agriculture

Agriculture is one of the sectors that is putting the environment under pressure, leading to biodiversity loss, greenhouse gas emissions and soil degradation. Despite the known high environmental impact, it is necessary to continue to feed ourselves today and in the future. Therefore, many different disciplines are focused on how to ensure a future that guarantees food security while reducing the environmental impacts of agricultural activity. In short, we are looking at how to make agriculture sustainable.

So what exactly does it mean for something to be "sustainable"?

Broadly speaking, the word "sustainability" is defined as development that meets the needs of the present without compromising the ability of future generations to meet their own needs. Thus, practices that ensure the conservation of resources could be called sustainable. For example, increasing the number and diversity of natural enemies of pests in crops could be a sustainable practice, as this reduces dependence on plant protection products, a limited resource that harms environmental health in the long term. Therefore, we cannot have sustainable socioeconomic systems without taking the environment into account.

But how can we know if a system is sustainable? A correct understanding of the various dimensions of sustainability (economic, social and ecological), as well as a complete understanding of all the elements that make up complex systems, such as agricultural systems, is one of the main challenges today.

Doughnut graphic on the different dimensions of sustainability and planetary boundaries. Author: Kate Raworth.

What is currently being done to make agriculture more sustainable?

Today, different strategies are being employed to find ways to make agriculture more sustainable without harming food production. To this end, different strategies are emerging that reduce environmental impact while maintaining or even increasing production, such as the production of genetically improved crops or the optimised and high-precision use of technologies in agriculture. There is a wide range of strategies in this sector!

However, these are not the only strategies that can improve sustainability in farming systems. These systems, also called agroecological systems, are complex, as they are made up of several interconnected parts with very specific linkages between them. In addition, it is also important to take into account consumers and other actors involved in these systems. A holistic view allows looking at complex systems from various dimensions, such as the ecological, social and economic dimension to assess the sustainability of a system. You may be thinking: "And with today's technological advances, wouldn't it be possible to do a computer simulation to see how to be more sustainable? Yes, it is possible! For several decades, computer simulations have been our allies in evaluating systems and estimating their performance. Although there are many other models that can be used (generally more focused on the economic aspect, such as econometric models), today I am going to explain in more detail the agent-based models.

Agent-based models

Agent-based models are a relatively new tool that integrates different dimensions and can simulate these complex systems. Agent-based models (also known as ABMs) are computational models that simulate the behaviour and interactions of different agents in the same environment. This computational resource has been used to model and simulate reality. In the field of agriculture (among others) they are mainly used to explore different scenarios such as the implementation of a particular agricultural practice, how farmers and consumers make decisions, and the effect of agricultural policies on the sector. These simulations of scenarios and uncertain futures can help to make decisions that lead to more sustainable agricultural systems. 

ABMs are a representation of reality, of the main agents that make up a system and of the interactions between them. Source: Izquierdo et al., 2008.

The main unit of these models are called, as you might expect, agents. These are the main entities of the model that possess certain attributes and interact with each other in time and space. Agents are a collectivised representation of individuals with similar characteristics and that simulate what we observe in reality. In this way, we can carry out numerous simulations that allow us to observe emergent phenomena derived from the joint interaction of these agents. In the diagram above we see how the people (entities) we observe in reality are characterised as agents in the computational model on the right. Each agent has its own attributes, behaviours and interactions, representing reality.

Since a picture is worth a thousand words, let me give a more visual example: Have you ever observed flocks of starlings, or schools of fish, moving in a characteristic way when they form groups? This phenomenon cannot be observed simply through a single agent (in this case, a single starling or fish moving alone), but requires the interaction of a large group of agents as well as an understanding of how they behave. 

A flock of starlings at dusk and their peculiar way of flying in a group.

Not only agents and their interactions are relevant, but also parameters need to be added to the model in order to define these scenarios. Following the example of sustainable agriculture, these parameters can be the area under cultivation, agricultural practices that determine production, the use of fertilisers, etc. To build these parameters, the model must be calibrated with data obtained from reality, from various case studies or databases that allow us to represent the reality we observe. For this, quantitative or qualitative data, or both, can be used. In turn, these can easily integrate geographic information systems (GIS), i.e. a spatial reference taken by satellite that contributes to calibrate the model with observed data. Furthermore, it is not only important to add data to characterise the model, but the behaviour of the agents must also be defined through rules or equations, depending on the case. This will shed light on how the dynamics of the model works, which will help us to understand the system we are studying and how we can intervene in it to obtain the expected results in reality.

This will define the scenarios we want to test: What would happen if I reduce the use of fertilisers? What if I change my production method? What would be the result of varying these parameters in the system? Which scenario causes more sustainability in my farming system? To answer these questions, hundreds of simulations of these scenarios and their future projections are carried out and then analysed to find out which parameters will radically change my model, how these parameters relate to each other and what the consequences might be in reality. However, these models represent complex systems with a multitude of interactions and data that may not have been observed. This implies some uncertainty in the models, so they cannot predict with complete accuracy the effects of varying certain parameters. Even so, they are a very practical tool for decision making or for exploring what our study system is like.

The application of ABMs allows us to explore future scenarios in order to choose the strategy that best suits our objective. Thanks to these models that integrate interactions between agents in space and time, it is possible to study these alternatives in order to achieve more sustainable production.

Bibliography:

Luis R. Izquierdo, José M. Galán, José I. Santos, Ricardo del Olmo, 2008. Modelado de sistemas complejos mediante simulación basada en agentes y mediante dinámica de sistemas. Empiria. Revista De metodología De Ciencias Sociales, (16), 85–112. https://doi.org/10.5944/empiria.16.2008.1391

-

This article was sent by Alba Alonso Adame. Alba studied at the University of Pablo Olavide in Seville (Spain), where she also studied the master of Biodiversity and Biology of Conservation. In 2019 she moved to Gent (Belgium) to start her professional career. In 2021 she started her PhD in Environmental Sciences at the Flanders Research Institute for Agriculture Fisheries and Food (ILVO) in collaboration with the University of Antwerp. She investigates sustainable transitions in agriculture by means of agent-based models.

You can find her on LinkedIn and Twitter.