Modeling Ecosystem Dynamics at Moderate Complexity

The L-Range family of models are moderate resolution tools that simulate ecosystem dynamics. The first in the group was G-Range, built to simulate global rangelands. A survey of rangeland modeling tools done prior to creating G-Range identified models that were: very complex, for example, simulating leaf angle to the sun and a suite of species; very simple, such as adding fixed amounts of biomass to elements each year; or non-spatial, representing a point in space. Complex tools are powerful but can many months to learn, simple tools can limit the potential for emergent responses and flexibility in addressing scenarios, and point-based models require many simulations across landscapes and complex summation of results to represent regional responses. We created G-Range to represent ecosystems in a spatially explicit way with results for regions in a single process and at moderate resolution, representing herbs, shrubs, and trees in various layers and pools. A simplified representation of the Century model is repeated for each landscape cell simulated, with new components to represent plant populations. A user may understand the basics of the model in a few hours of study and be adept in its use quickly. That said, skill in the ecosystem science informing the assignment of parameters is helpful. The model and its use is described in Boone et al. (2018) and described in detail in Sircely et al. (2019).

We modified G-Range to allow simulation for any area and at any resolution, in a tool called L-Range, reflecting its localized focus. L-Range may be applied to an area by preparing spatial surfaces for the area and assigning parameters to describe how plants grow in different areas. The results are assessed by comparing them to real-world patterns. The tool has been modified to connect with agent-based models that represent human decision-making. L-Range may then be used to address an number of scenarios involving concepts such as production dynamics, stocking density, animal mobility, drought responses, and climate change.

Our most recent tool is Af-Range, which is still in development. This is an application of L-Range to Africa, with each landscape cell 10 km x 10 km resolution. Af-Range that will be assessed carefully. Herbivores are represented in Af-Range directly as 10 herds in each landscape cell. Users may parameterize those herds as they wish, such as for four domestic livestock species and several wildlife guilds such as browsers and grazers.

Acknowledgements

Original funding for G-Range was through the International Livestock Research Institute, Nairobi, Kenya, with Dr. Philip Thornton and the Commonwealth Scientific and Industrial Research Organisation (CSIRO) in Australia, with Dr. Mario Herrero. Both encouraged the development of a moderate complexity rangeland simulation tool. NASA funding allowed for continued model development. Dr. Greg Kiker through the University of Florida is providing continued funding.

Our thanks to Dr. William Parton and Cindy Keough for endorsing our use of Century in the L-Range family of models to represent biogeochemistry. Our thanks too to Dr. Michael Coughenour, the author of the Savanna Modeling System. Dr. Boone’s experience with that tool informed aspects of G-Range, and the decomposition component of the model is based on that in Savanna and Century.