Simulating Management with Models

By Jeff Hardesty, Jonathan Adams, Doria Gordon, & Louis Provencher

The colonel was not a happy man. As a high-ranking officer at Florida’s Eglin Air Force Base, he had just been forced to reject a multi-million dollar test of the Star Wars missile defense system because he could not estimate how the proposed test would affect the base’s population of the federally endangered red-cockaded woodpecker (Picoides borealis). To comply with federal law, the Colonel should have had a good understanding of the endangered bird’s needs, since the base contains some of the best remaining examples of the woodpecker’s critical longleaf pine (Pinus palustris) habitat. Eglin, however, had never invested in basic inventory and monitoring. The U.S. Fish and Wildlife Service, therefore, issued a Jeopardy Opinion for the red-cockaded woodpecker under the Endangered Species Act, forcing the Colonel to take responsibility for the lost test mission.

Natural resource managers at Eglin had long known about the woodpecker — it has been a federally listed endangered species since 1970 and the subject of considerable research. But the Department of Defense (DOD), like most other federal agencies charged with managing large, complex ecosystems at that time, lacked data on both what species and communities they had and how those elements should be managed. Although highly skilled in traditional natural resource management and with years of hands-on experience working in longleaf pine forests, Eglin’s managers and partners lacked important knowledge and understanding of the composition, structure, and function of those ecosystems. Managers of the air force base were quite literally “flying blind.”

To make matters worse, the Colonel and Eglin’s civilian natural resource managers now faced two compelling and, on the surface, mutually exclusive mandates: fulfill a key military mission and conserve an endangered species, as required by federal law. Not only did they need more information about red-cockaded woodpeckers and their habitat, they needed to learn how to use that information to change the way one of the nation’s largest and most important air bases conducted its business.

That was 1988. Since then, Eglin has become one of the most sophisticated federal facilities in the country in terms of integrating science and management. Lying on the Gulf of Mexico in Florida’s panhandle, Eglin has a vital conservation mission: the base contains about 1,500 square kilometers of longleaf pine forest, the largest remaining example of an ecosystem that once covered roughly 250,000 square kilometers and stretched across the southeastern United States from Texas to Virginia. The Eglin population of red-cockaded woodpeckers was critical to species recovery. Moreover, despite covering less than 2% of the East Gulf Coastal Plain Ecoregion, Eglin contains viable examples of 33% of the ecoregion’s imperiled species and natural communities, making it one of the most important conservation sites in the southeastern U.S.

The Colonel and the rest of Eglin’s resource managers had the courage to admit that management, more than the military mission, was probably the leading cause of ecological degradation and the decline of red-cockaded woodpeckers. “We came to realize that the most important way we could support the military mission of the base — our primary program objective — was to restore the integrity and resilience of ecosystems on Eglin, thus providing the greatest flexibility and options for future management,” says Rick McWhite, Chief of the Natural Resources Branch. “Traditional resources management was having the opposite effect.”

Resource management at Eglin, as elsewhere, had focused on forestry and game management, though Eglin’s managers used prescribed fire as well. Combined with a systemic lack of funding for ecological monitoring and inventory, this focus had led to significant degradation in the integrity of the longleaf pine ecosystem and declines in many sensitive species including red-cockaded woodpeckers.

Once they began probing, Eglin managers realized they had more questions than answers. They were reasonably confident that more prescribed fires, particularly during lightning season, would be necessary to reverse the trend toward hardwood and sand pine (Pinus clausa) dominance at the landscape scale. But fundamental questions remained. Beyond recovery of the endangered woodpecker, what are the goals of landscape-scale restoration? How do we set priorities? What is “the natural range of variation” in fire frequency and other natural processes, and how can we manage for them? What other natural communities and species are of conservation concern, and what is their current distribution and condition? Given that fire has been partially suppressed for 50 years, would increased and likely more intense fires harm native plants, animals, and ecosystems? What other management tools can be used in addition to fire, and what might be their impact on native species and ecosystems? Given the new mandate from the DOD that ecosystem management should now be practiced, how can we nest traditional game, forestry and recreational programs within in an overall ecosystem management framework?


The Eglin command realized that they could not postpone managing red-cockaded woodpeckers and longleaf pine until scientists had found answers to these questions: the decline of both the woodpecker and its habitat demanded action. Since the Colonel’s immediate need was more information on red-cockaded woodpeckers, he contracted in 1991 with The Nature Conservancy and the University of Florida to inventory and monitor threatened and endangered species and natural communities. In 1992, Eglin’s natural resource managers and The Nature Conservancy convened a suite of ten scientific meetings and workshops, each focusing on an information gap or management conflict. This effort opened Eglin’s management programs to direct review and participation by more than 100 external scientists and managers representing more than 30 agencies and organizations. The result was a five-year natural resource management plan written by Eglin managers and completed in 1993. This plan set the standard for integrated, ecosystem-based management in the DOD and was the first such plan developed for any federal landholding.

Eglin’s management plan described a continuing need for information resulting in two key insights. First, treating management as scientific experiments was essential if Eglin’s managers were both to take action and learn more about longleaf pine and other systems. Second, models based on manager and expert opinion and data could provide an important additional tool for evaluating the probable long-term and large-scale effects of different management scenarios.

“This adaptive management approach enabled us to conduct the critical management and restoration that is necessary now, while continuing to refine our understanding and approaches to the system,” says Carl Petrick, Chief of Eglin’s Wildlife Section. “We now have a framework that enables us to see how the pieces of the puzzle fit together. The workshops generated a lot of information — enough to move our management forward, build simple conceptual models of how these ecosystems work, and sometimes approach our management experimentally. Management experiments and models allowed us to test our basic assumptions.”

In 1993, Eglin managers and The Nature Conservancy, along with George Tanner from the University of Florida and Leonard Brennan of Tall Timbers Research Station, launched the first such large-scale management experiment, focusing on restoration of longleaf pine tracts that had been invaded by hardwood and sand pine. Managers asked the collaborating researchers to determine the most efficient way to reduce hardwood encroachment while protecting longleaf pine seedlings and trees, protecting and stimulating herbaceous species and arthropods (important food sources for wildlife), and increasing wildlife populations. At the same time, this experiment sought to illuminate the underlying structure, composition, and function of natural and degraded longleaf pine systems on Eglin, how they are changing over time and space, and how they are responding to alternative management treatments.

The research team compared four management methods that consisted of maintaining fire-suppression or applying three hardwood reduction techniques in 30 large (200 acre) test plots: spring burning, mechanical felling/girdling, and herbicide application. Herbicide and felling/girdling plots were burned for fuel reduction two years later, a standard forestry practice. These methods were evaluated against more natural, fire-maintained longleaf pine stands at Eglin and followed for five years.

Although individual species responded differently to the three hardwood reduction treatments, all treatments increased the number of red-cockaded woodpeckers compared to controls. Additionally, all treatments promoted the growth of understory plants that provide food, cover, and nesting sites for key game and non-game bird species. Most importantly, the large-scale study showed managers that they could significantly improve degraded sandhills using fire alone (i.e., without mechanical or herbicide approaches). Using mechanical or herbicide applications followed by fire may have sped the restoration effort, but they increased the cost more than eight-fold and harmed some important non-target species. Moreover, the data collected dramatically increased understanding of system dynamics and the distribution of species therein. The research, for example, identified important characteristics of healthy longleaf pine sandhills, and these indicators are now being incorporated into a base-wide monitoring program.

In a second large-scale experiment, initiated in 1995, Virginia Tech researchers Jeff Walters, Carola Haas, and Kathy Gault began a base-wide adaptive management effort aimed at red-cockaded woodpeckers. This experiment sought to determine how quickly woodpecker populations would respond to ecological management of forest ecosystems compared to intensive and expensive single-species management.

Management techniques ranged from least intensive (maintaining longleaf pine through burning) to more intensive (constructing nest cavities, removing hardwoods, and translocating woodpeckers from other parts of the base). Researchers were able to document a significant increase in population size and rate of growth in response to habitat management alone, something that had been rarely documented despite extensive study. Managers also were able to estimate the maximum rate of growth that could be achieved with intensive management, albeit at greater cost. For the first time, these results allowed managers from Eglin and elsewhere to evaluate the trade-offs between these two recovery approaches in terms of time and cost.

These experiments provided enormously valuable management information in and of themselves, and also generated data for a detailed model of the population dynamics of red-cockaded woodpeckers at Eglin and two nearby forest reserves. Jeff Walters, along with Larry Crowder and Jeffrey Priddy from Duke University, sought to predict the future abundance of red-cockaded woodpeckers by tracking the fate of individual birds and their progeny. The model, which used data from numerous population studies covering many years, was also spatially explicit, placing each bird within a territory the coordinates of which could be located on an actual landscape.

The value of this population model became apparent to the managers when it challenged an assumption from Eglin’s management plan. The plan calls for building a demographic bridge between the Eglin’s more robust western sub-population of red-cockaded woodpeckers and the smaller, more threatened eastern sub-population. Planners assumed that such a bridge would stabilize the eastern sub-population.

The red-cockaded woodpecker model suggested this might not be an effective strategy, either biologically or economically. According to the model, the eastern sub-population would grow more quickly if artificial clusters and translocated birds were used to buffer the core and edges of the eastern population, essentially managing the eastern sub-population in isolation from the western one. The clusters of birds that would form the bridge were generally not predicted to persist over time. If borne out by management, this will mean shifting the priority of habitat restoration, including intensive control of hardwood and sand pine, from the bridge area to several large areas that were thought to be of lower priority. The model is a boon to managers, as its predictions allow for better allocation of scarce resources by indicating which sub-populations and clusters have a higher probability of persistence over the long-term.

The Nature Conservancy also facilitated workshops to construct a series of simple conceptual models of vegetation dynamics in longleaf pine ecosystems. The first model focused on understanding how past management had influenced the composition and structure of longleaf pine systems. At Eglin, as elsewhere, landscape-level fire suppression, fragmentation, intensive soil disturbance, and past longleaf pine harvest had pushed longleaf pine sandhill ecosystems on a path toward other stable states dominated by fire-intolerant sand pine and fire-resistant hardwood communities. The conceptual model hypothesized that some kinds of disturbance may send ecosystems on a trajectory of nearly irreversible change (Figure 1).

To test this hypothesis and others, Eglin managers and scientists from The Nature Conservancy then worked with Garry Peterson, a graduate student at the University of Florida, to develop a simple spatial forest simulation model to address the complex interactions among fire frequency, landscape geometry and vegetation structure that appeared to be hindering the efforts of Eglin managers to deploy fire effectively. The model allowed managers to explore how much fire was needed to reduce the dominance of hardwoods and to control the spread of sand pine. The purpose of the spatial model was not to predict the exact future of the Eglin landscape — an impossible task — but to explore the consequences of different sets of assumptions and to allow Eglin mangers and their partners to question the ideas underlying past, current, and proposed management actions.

A key hypothesis tested at Eglin was whether fragmentation and widespread hardwood encroachment and sand pine invasion into a landscape once dominated by longleaf pine had so fundamentally changed ecosystem dynamics that fire would now behave differently and have different ecological impacts. If so, then restoration would require different ways of approaching fire, including much greater consideration of spatial and temporal factors.

This simple model produced compelling and surprising results for Eglin’s land managers and scientific partners, and also provided insights into the limits of biological intuition and individual experience. For example, a traditional approach for managing large landscapes with fire is to divide the landscape into burn blocks and burn each in rotation. This approach, however, implies that an optimal fire frequency exists, and ignores landscape geometry and variation. The model revealed that a simple rotation strategy, unless applied at higher frequencies than are feasible, would lead to long-term landscape-scale ecological degradation. The failure to adapt fire management to change and variability in the landscape produced a stable but largely transformed landscape. Worse, the model predicted that degradation might not appear for two or more decades, by which time reversing it would be difficult and expensive. This essentially describes what had happened at Eglin over the past fifty years (Figure 2).

“Adaptive” fire strategies prioritize specific areas for different fire frequencies based on the vegetation present and restoration or maintenance needs. The fire model clearly indicates that managers have to expect and account for significant variation in prescribed fire effectiveness. Yet the reality is that limitations of staff, the annual number of suitable burning days, and conflicts with the military mission mean that Eglin managers rarely if ever meet all of their prescribed fire objectives. This creates a backlog of unburned areas. The fires themselves also are variable, depending on the interaction of such factors as weather and vegetation patchiness. Modeling showed that these sources of variation interact and rapidly accumulate in the landscape and, along with the starting landscape geometry, dictate which fire management strategies will succeed or fail. Landscape-scale fire strategies have to estimate and account for inevitable variation. Eglin managers were able to use the model to develop a simple, rule-based adaptive fire management strategy that would likely allow them to achieve landscape-level ecological objectives, and at a lower cost.

Even using the adaptive fire strategy, however, meant dramatically increasing the frequency and size of prescribed fires, given the degradation that had already occurred. Some military commanders resisted that conclusion. The model, however, generated “movies:” graphic simulations that capture in a few minutes what can happen to a landscape over fifty years. The movies were so compelling that they convinced even the most reluctant military commanders of the need to double the amount of prescribed fire applied annually.

The Eglin managers’ experience with ecosystem management, scientific partners, experiments, and models was successful in large part because early on they had accepted that they had to use management as an opportunity for learning. Adopting this approach required humility and challenged organizational culture and priorities, as well as individual training and values.

This learning was not limited to the managers. Both researchers and managers learned that even simple experiments, if conducted at management level, can provide information relevant to management. For example, the base-wide RCW management experiment took advantage of already planned management and monitoring but placed them in an experimental framework, thus gaining important insights. Similarly, even the most simple of Eglin’s “throw-away” conceptual models rapidly advanced learning and helped management teams reach consensus about the functioning of ecosystems and management goals. Conceptual models and experimental designs are cheap; management and monitoring are expensive.

The Eglin experience also highlighted the limits of “trial and error management” and biological intuition (often confused with “adaptive management”). Both Eglin managers and partner scientists needed to re-examine their assumptions. The fire, woodpecker, and longleaf pine models and experiments clearly showed that population and ecological dynamics often were occurring at one or two spatial and temporal scales above and below normal human perception. Expert opinion alone would have steered Eglin in some ecologically damaging and costly wrong directions.

More than a decade after the Jeopardy Opinion, red-cockaded woodpeckers have increased by nearly 40%, and Eglin’s managers and partners have a much deeper understanding of ecosystem dynamics. Longleaf pine forests, and the habitat they provide for a host of other species, are under active restoration. Now, because of their 10-year investment in inventory, monitoring, modeling, and management experiments, Eglin managers are engaged in developing a second generation ecosystem management plan that promises to be deeper and much better informed than their celebrated first effort. And the example set by this approach is being duplicated at other DOD and agency properties. The risk Eglin managers took in opening their work to the scrutiny and input of the management, research, and conservation communities is paying dividends in local partnerships and in conservation across a broad landscape.

Further Information:
Jeff Hardesty, Director, Ecological Management and Restoration, The Nature Conservancy, University of Florida, P.O. Box 118526, Gainesville, FL 32611-8526;