Douglas Briggs, Jim Westervelt, Shaun Levi, Steve Harper

A Desert Tortoise Spatially Explicit Population Model

Introduction

Across the nation, land management offices responsible for the management of natural resources, endangered species, water quality, aesthetics, and economic productivity are turning toward dynamic spatial ecological modeling. Military land managers, continuously facing the challenge of accommodating both the legislated conservation measures such as the Environmental Protection Act, designed to protect ecosystems on military installations, and the demands of rigorous training schedules, can use dynamic spatial ecological models to investigate the impact that various training schedules and other human impacts have on the environment in both the short term and the long term. With the use of the modeling tools generated by this research, military land managers can develop a training schedule that both satisfies training requirements and preserves the ecological systems within which the training will occur.

A dynamic spatial ecological model is a computer model that evolves in simulated time. Research teams create and develop the complex system of mathematical, logical, and stochastic processes that govern the progress of the computer simulation. A set of initial conditions (culled from raster images, vector, or survey data, for example) seeds the system and provides initial conditions for the simulation to develop in time.

This research project developed a prototype dynamic spatial ecological model of the endangered desert tortoise (Gopherus agassizii) population on the grounds of the Fort Irwin Army Training Center in the Central Mojave Desert of California. A 57-by-57 grid of 1 kilometer square areas divided a sample portion of the landscape of Fort Irwin into "cells." Each cell is assigned an identical computer model that simulated pertinent environmental variables such as elevation, soil type and moisture, precipitation, air and surface temperature, tortoise mortality, and percentage of vegetation cover. The model itself, developed in four parts by independent, interdisciplinary research teams, was written with the STELLA modeling software and captures the hydrology, vegetation cover, tortoise population, and tortoise migration dynamics as they evolve in monthly time-steps. This poster describes the process by which the model was developed and executed as a part of the larger simulation of the entire Fort Irwin ecological landscape.

The Spatial Modeling Environment

After the STELLA model is complete, it is converted into the computer simulation which will execute a copy of the model in each of the cells simultaneously and in parallel. We used Dr. Thomas Maxwell's Spatial Modeling Environment (SME) software tool for this conversion process.

First, the STELLA model is saved as a text file which summarizes the difference equations governing the model's stochastic processes. This text file is then translated in two stages: first into Modular Modeling Language (MML), and then into C++. Raster GIS maps for ground elevation, slope, aspect, vegetative ground cover, ground compaction, and soil water content are linked to the SME model for initialization data, output preferences are registered in a configuration file, and the C++ code is linked and compiled. We then run the resulting executable binary on a UNIX workstation and both observe the output as the simulation runs, and collect it for further analysis later.

Future Directions

The Desert Tortoise Model demonstrates how dynamic, spatial, ecological modeling tools can be designed for effective use in land management. Though similar in some ways to the Sage Grouse model (Westervelt, 1995) the Desert Tortoise model pushes the frontiers of these research efforts by implementing a land management tool, as opposed to a demonstration device. The Desert Tortoise model presented here is a broad-based picture of the relevant dynamics of tortoise habitats at Fort Irwin. In all cases, a model is a simplification of the system that it represents and for that reason this model provides most accurately predictive results when run for no longer than 100 years. For this first stage of development the critical components of development included the multidisciplinary research team; a coherent, logical.modeling process; and a collection of modeling software tools capable of managing the type and volume of data needed to effectively model a multi-cellular landscape. The software tools we used were:

GRASS -- The Geographical Resources Analysis Support System, which helped initialize the model.
STELLA -- A graphical user interface oriented dynamic programming language that enabled all researchers regardless of modeling experience to participate in the simulation model design process.
SME -- The Spatial Modeling Environment tool designed by Dr. Thomas Maxwell of the Univeristy of Maryland's International Institute of Ecological Economics converted STELLA models into a form that could be at a multi-cellular level on desktop workstations.

The Desert Tortoise Model research efforts to implement dynamic, spatial, ecological models as effective landscape management tools are not yet finished. The first phase comprising the development of the spatial model is complete. We now look to develop a battery of powerful yet comprehensible statistical tests for performing sensitivity analyses to validate the model and its output. We plan also to continue to integrate into the STELLA model our expanding empirical knowledge of tortoises and their habitats, in order to improve the predictive power of the model.

Authors and Acknowledgements

Douglas R. Briggs, Dept. of Computer Science, University of Illinois at Urbana-Champaign. (drbriggs@cs.uiuc.edu)

Dr. James Westervelt, Dept. of Urban and Regional Planning, University of Illinois at Urbana-Champaign. (westerve@gis.uiuc.edu)

Shaun Levi, Dept. of Geography, University of Illinois at Urbana-Champaign. (s-levi@uiuc.edu)

Steve Harper, Dept. of Ecology, Ethology, and Evolution, University of Illinois at Urbana-Champaign. (sjharper@uiuc.edu)

The authors gratefully acknowledge the work of Dr. Thomas Maxwell of the Chesapeake Biological Laboratory at the University of Maryland for his development of the SME software. This study was conducted for Headquarters, US Army Corps of Engineers under Project 4A1102AT25, "Environmental Restoration", Work Unit IA4, "Fundamentals in Dynamic Ecological Modeling".

Further Information

More information about this and related projects at the Geographic Modeling Systems Laboratory at the University of Illinois at Urbana-Champaign/U.S. Army Construction Engineering Research Laboratory is available here.