Modeling Agroforestry Systems to Improve the Lives of Farmers

Including the above and below ground interactions between plants, trees, soils, water and light, agroforestry systems are complex. Influenced by multiple variables and evolving over time, they are hard to study. Supporting commercial and subsistence agriculture, they are vitally important to the economic and social welfare of farmers in developing countries.

There were many crop models and tree models but not one that looked at the above and below ground interactions between crops and trees in one single model

It’s their complexity and importance that make agroforestry systems perfect modeling candidates and, in fact, many models are available. Developed by the Ecological Modeling Unit of World Agroforestry Center in Bogor, Indonesia using STELLA, WaNuLCAS provides the most holistic view of systems allowing researchers to fully understand interactions between multiple components and test a wide variety of agricultural improvement strategies.

“Before we developed WaNuLCAS, there were many crop models and tree models but not one that looked at the above and below ground interactions between crops and trees in one single model,” said Dr. Betha Lusiana, Associate Scientist and Head of the Ecological Modeling Unit. “WaNuLCAS models the above ground architecture of crops and trees and allows users to see the water, light, and nutrient processes through four soil layers.”

The model allows researchers to take an evidence-based approach without the time and expense of real-time experiments

“The model allows researchers to take an evidence-based approach without the time and expense of real-time experiments,” says Lusiana. With crop/tree combinations and initial stocks (water, nutrients, soil properties) and other inputs (gradient, planting/harvest seasons, fertilizer, organic matter, plant spacing, etc.) specified, the model shows how water finds its way to roots and how nutrients are absorbed and released from the soil.

Determining the best tree and crop (which might include livestock) combinations in order to maximize agricultural productivity while maintaining forestation, is particularly important to farmers and communities in developing countries. “Farmers in developing nations need to plant productive crops and want to save trees but can’t afford expensive agricultural management approaches,” says Lusiana.

Researchers supply WaNuLCAS with regional data to explore the productivity and impact of crop/tree combinations. Researchers gain insight into agricultural methods – should trees be shallow or deep rooted, should canopies shade crops, will irrigation or fertilizer be needed, is mixed cultivation better than monoculture? They also explore larger economic and social issues: how is climate change impacting regional agriculture, where can agroforestry be most productively applied, how can agroforestry be used to improve the livelihoods of small farmers?

The requirement to develop a model of a complex, variable system for use by researchers who were not experienced in modeling, led Lusiana and her team to STELLA

Agroforestry systems and questions about them vary across geographic region. Lusiana explains that while WaNuLCAS is credible and generic enough to apply to all regions, users can apply data and make modifications to accommodate their specific research area and focus. The requirement to develop a model of a complex, variable system for use by researchers who were not experienced in modeling, led Lusiana and her team to STELLA.

“While, the majority of WaNuLCAS users are PhD level students who are conducting scientific work, very few of them have system dynamics, modeling, or development experience,” says Lusiana. “We wanted them to see the processes in the model and we wanted them to be able to modify WaNuLCAS without being scared. STELLA is very good software for those requirements.”

Thomas Hilger was a PostDoc and new to modeling when he first used WaNuLCAS. Now a Senior Scientist at the University of Hohenheim in Stuttgart, Germany, Hilger studies soil conservation and erosion control, measures that are critical to sustainable farming.

We gained insight on why crop management strategies like reduced tillage and the addition of organic matter improved the soil structure

“Soil erosion research is usually very time-consuming,” says Hilger. “You have to find and measure all the possible variables over time in one place so you can test a range of strategies. It’s hard to collect all necessary variables and PhD students might have two years in the field before they have to start analyzing their results and writing their dissertation. Two years often isn’t enough time to collect all the data and test all the scenarios.”

WaNuLCAS gave Hilger and his research partners a way to test crop management strategies in a reasonable time frame. They started by calibrating and validating WaNuLCAS for conditions in Thailand. Then, using local data sets, they modeled a variety of run-off mitigation strategies. “When water infiltrates the soil, there is less run-off and less erosion,” says Hilger. “The better the soil structure, the better the infiltration. We gained insight on why crop management strategies like reduced tillage and the addition of organic matter improved the soil structure.”

We used the model to create a scenario with normal rainfall in the first two years and higher rainfall in the third. Run-off still decreased showing the crop management continued to work

The research and findings prompted many questions from reviewers. “It was difficult to unravel and find answers to those questions,” says Hilger. “We thought ‘Why not turn those questions into scenarios and use WaNuLCAS and available data to find answers?”

WaNuLCAS was particularly helpful in answering questions about run-off and erosion variance year-to-year. “Using actual data, our study showed that in the third year of improved crop management, run-off and associated erosion continued to decline,” says Hilger. “In that third year, rainfall also decreased. Reviewers asked how we could be sure that reduced erosion was due to crop management and not lower rainfall. We used the model to create a scenario with normal rainfall in the first two years and higher rainfall in the third. Run-off still decreased showing the crop management continued to work.”

Hilger and others at the University of Hohenheim have gained confidence in WaNuLCAS by applying “goodness of fit” testing. “We’ve tested the model against multiple parameters – model efficiency, modeling error, coefficient of determination, and correlation between predicted and observed results,” says Hilger. “The more of those parameters achieved, the better the model. WaNuLCAS often achieves all four, particularly under sub-humid and humid conditions of South East Asia.”

We’re currently collaborating with the University of Adelaide to model the impact of agroforestry and community policy on the livelihood and security of farmers

Using realistic assumptions and allowing researches to change a wide range of variables, WaNuLCAS is being applied to soil erosion research in Vietnam and Thailand. Lusiana and others at The World Agroforestry Center are continuing to improve and extend the model through modules that focus on tree phrenology, parasites, greenhouse gas emission, and latex production as well as soil structure.

Having spent many years training researchers to use WaNuLCAS, Lusiana and her colleagues are now working with partners around the world to find solutions to agroforestry challenges. “We’re currently collaborating with the University of Adelaide to model the impact of agroforestry and community policy on the livelihood and security of farmers,” says Lusiana. “We’re also working on a new project that looks at peat soils and assisting researchers who are looking at cropping systems across 38 regions in the Philippines as a way to consider climate change. We’re still focused on helping people.”

Teaching, Researching and Protecting Estuarine Ecosystems with STELLA and isee NetSim

The Virginia Institute of Marine Science’s (VIMS) legal mandate to deliver graduate-level education, conduct research and provide advisory services makes it unique among marine science institutes. The models it builds and uses to teach, research, and protect estuarine ecosystems, areas where rivers meet the sea, are another distinguishing feature. “We are placing models of coastal ecosystems directly into the hands of students, other researchers, and local stakeholders,” says Dr. Mark J. Brush, Associate Professor of Marine Science, Department of Biological Sciences.

They can be new to modeling, they don’t have to download software, and they don’t have to know or learn a programming language

“There are a lot of marine science models built with complex programming languages like Fortran that run on expensive, high performance computers,” says Brush. “I wanted to create an adaptable model that would reduce complexity. STELLA models can do much of what programmed models do and, with isee NetSim™, students and local decision-makers can explore their own systems. They can be new to modeling, they don’t have to download software, and they don’t have to know or learn a programming language.”

Estuarine ecology and modeling

The models Brush, his students, and clients use to study estuarine ecology are focused on eutrophication, a process characterized by degraded water and habitat quality and typically fueled by excessive nutrient loads. “Nutrients are essential up to a point, but too many nutrients are harmful to coastal waters,” says Brush. “Nutrient loads from human activity, like sewage treatment plants and agricultural run-off, lead to algae blooms, a build-up of seaweed, low oxygen ‘dead zones’, and fish kills.”

They might test the effectiveness of reducing sewage treatment plant inputs or increasing the population of shellfish or amount of sea grass

“Our models start with nutrient loads – what and how much is going into the water – and predict what happens when levels of those nutrients are increased or reduced,” says Brush. Students, researchers or local decision-makers can also determine the impact of water quality restoration methods. They might test the effectiveness of reducing sewage treatment plant inputs or increasing the population of shellfish or amount of sea grass, both natural filters.

Teaching estuarine ecology with STELLA

STELLA models are one of the primary teaching tools in Brush’s graduate modeling class. “Together, we’ll look at a published model, take it apart, and then rebuild it step by step,” he says. “Assigned projects require students to build their own models which helps them understand Systems Thinking and how it can be applied to ecological systems.”

He models the exponential growth equation to both explain how it works and demonstrate how it applies to coastal ecosystems

Brush also uses models to teach essential concepts. For example, he models the exponential growth equation to both explain how it works and demonstrate how it applies to coastal ecosystems.

His annual undergraduate guest lecture also features estuarine modeling. “I give an overview of modeling and then, as a group, we review a NetSim model of a small Chesapeake Bay tributary. Students break up into small groups and I challenge them to run the model to determine the best way to restore the system.”

Conducting research

Brush was an undergraduate at Cornell when he was first exposed to STELLA as a teaching assistant for an oceanography lab. At the University of Rhode Island, a professor who had a grant to develop a model of nutrient-fueled eutrophication learned about Brush’s STELLA experience and brought him onto the project. Now Brush coaches his own graduate students through modeling that supports their research.

Researchers can add submodels like eelgrass or shellfish that filter nutrients and simulate their impact

“Most of my graduate students are new to STELLA and most of them use it for their research,” says Brush. “They begin with my core model and adapt it for their own work.”

The core model includes the cycling of basic nutrients - nitrogen, phosphorous, carbon, oxygen and chlorophyll – through an estuarine ecosystem. It also includes levels of phytoplankton in the water and benthic algae on the bottom. “The core model includes all the major components of any temperate estuary or coastal system and the key features that respond to eutrophication,” says Brush. “Researchers can add submodels like eelgrass or shellfish that filter nutrients and simulate their impact.”

Providing advisory services

That same core model gives Brush and his students a starting place in providing service to local, state, and regional managers. A model of the Delmarva Peninsula coastal bays, for example, adds water temperature and sea level rise to show local decision makers how a changing climate will influence eutrophication. “The Delmarva bays are shallow systems, maybe 1 to 2 feet deep,” says Brush. “When the sea level rises, less sunlight reaches the bottom. That results in less benthic algae to filter nutrients so nutrient loads increase.”

The Narragansett Bay model allows decision-makers at the state level to determine acceptable nutrient loads for that ecosystem

Customizing the core model enables local stakeholders to consider their own issues with their own data. Existing models support West-Rhode River Estuary restoration, land use planning on military bases as part of the Defense Coastal Estuarine Research Program (DCERP), oyster restoration in Harris Creek, Delmarva coastal bay management, and Narragansett Bay decision-making.

“The Narragansett Bay model allows decision-makers at the state level to determine acceptable nutrient loads for that ecosystem,” says Brush. “For example, they can use the model to reduce sewage treatment plant loading by 30%. When they run the model, they’ll see if water quality meets their criteria. If it does, they know that 30% is a good target. Or they might want to test other reduction levels.”

Improving the models

Models used by municipal, state, and federal decision-makers are considered Beta versions until the projects they support have been completed. “During projects, the models constantly evolve as we work directly with stakeholders, get their feedback, and make them easy to use,” says Brush. “The DCERP model, for example, is part of a five year project. When we finally hand it over we’ll have finished improving and calibrating it and users will have developed confidence to use it for their work.”

The ability to rapidly develop and serve new models online for direct use in the classroom or by stakeholders - that’s the advantage we have with STELLA and NetSim

Climate change has consequences for all coastal systems and is currently being built into the core model. “The model will allow users to ramp up water temperatures, precipitation, and sea level rise using local forecast data,” says Brush. “They’ll see, for example, that increased water temperatures may require them to further reduce nutrient loads to keep water quality in line with established criteria. And, they’ll be able to plan into the future.”

STELLA’s adaptability is also helping Brush and his team get involved in projects outside of their own mid-Atlantic neighborhood. “We’re about to start a project on the Gulf coast and are continuing to expand our work along the East coast,” says Brush. With the core model in hand, Brush and his team are able to investigate, or enable local scientists and policy makers to investigate, most temperate estuarine systems. “Most local communities and management agencies don’t have the time or money to develop and use a custom-programmed model,” says Brush. “The ability to rapidly develop and serve new models online for direct use in the classroom or by stakeholders - that’s the advantage we have with STELLA and NetSim.”

Dr. Brush’s use of STELLA does not constitute an institutional endorsement of the model.

Story of the Month: Re-Issue

Since isee systems (formerly High Performance Systems) initially released the Story of the Month series in May of 2001 events and settings have changed but many of the issues addressed remain relevant. From corporate corruption, to terrorism, to greenhouse gasses, each story analyzed a past headline news topic from a systems thinking perspective. These topics were modeled using STELLA or iThink with the hopes to gain perspective on the system, stimulate discussion and perhaps adjust our assumptions.

Follow along as we re-issue a Story of the Month from our archives; see how much or how little has changed in each topic area

isee systems is excited to be re-releasing each Story of the Month as an easy-to-access NetSim model. Follow along as we re-issue a Story of the Month from our archives; see how much or how little has changed in each topic area. These models were preserved in their original state and have not been updated to include current information. Are the behaviors of each model consistent with what is currently happening or has the behavior of the system changed?

Our first Story of the Month reissue looks at the escalating violence in the Israeli/Palestinian conflict and how Systems Thinking might prove useful in building greater insight into both the causes of the conflict and what would need to happen to achieve a lasting peace. We hope you enjoy this peak in to history and we would be delighted to hear your thoughts on each Story of the Month.

On the Road

isee systems is headed off to warmer weather this December! From December 7th to the 10th Co-President Bob Eberlein and Sarah Davie will be participating in this year’s Winter Simulation Conference in Savanah, Georgia. This will be our first time at the Winter Simulation Conference and we are thrilled to be exhibiting our software at this congregation of top-notch system simulators. Stop by our booth to learn more about our software and some upcoming new features, and don’t miss Bob’s talk Simulating Queues, Conveyors, and Ovens with Stella Professional as he discusses these discrete elements in STELLA Professional!

Coinciding with the Winter Simulation Conference, Co-president Karim Chichakly will be attending this year’s Congreso Latinoamericano de Dinamica de Sistemas (CLADS), the annual conference held by the Latin American chapter of the System Dynamics Society from December 9th to the 11th in warm Costa Rica! Karim will be showcasing our software as well as giving a workshop! This will also be our first time attending the CLADS and we look forward to meeting Systems Thinking practitioners in this chapter of the Systems Dynamic Society!