Both approaches are doubly reductionist in that they use reductionism not only to isolate and understand parts but to insist that everything must derive from that most-reduced element. For wilderness, this demand involves a historical reductionism, and for fire behavior, a dynamic reductionism. Accept this reasoning and we remove ourselves from history as we have from geography. If that doesn’t appear to be the case, it is because we have also removed ourselves from models of fire behavior. But since fire science has been sponsored by public agencies interested in wildlands, these formulations are the ones that dominate discourse.

It is entirely possible, however, to invert both propositions. Begin with the inhabited landscape and with anthropogenic fire. These conceptual complexes determine the units of analysis, each of which can be analyzed through reductionist techniques. Where people have chosen to remove their agency, subtract them from the equations. Wilderness fire behavior is thus not a core model, to which we can overlay people, but a special case in which one variable, humanity, nature’s species monopolist over fire, has been restricted as a factor or set to zero. In this conception there remains ample room for the physical sciences, but the whole, not a hierarchy of parts, determines their significance. This is in effect the argument advanced by those scrutinizing the presence of fire in the Ozarks. 

They have no choice, really. But their necessity suggests a larger opportunity to reform fire theory. It says that you don’t begin with wild fire and somehow, someday, maybe, incorporate people. It says you begin with anthropogenic fire – its behavior, its regimes, its landscapes, its history. The old fire-behavior troika of fuel, weather, and terrain can’t by itself describe real-world fire on the landscape, and enshrining it perpetuates a sense that if we only remove people we can solve (or at least understand) fire. We can make the formulas true by positing the historic equivalent of ideal frictionless surfaces. Rather than attempt to derive fire history from physics-based fire behavior formulas, we ought to consider such formulas as a special case of general fire history.[i] 

      This is a bold bid, for it expands the meaning of fire behavior to include humans. The intervening variable is the topographic texture of the Ozarks, what the group has sought to quantify as “roughness.” People determine when and where fires will start, and terrain determines how they will spread. The issue cannot be dodged in the Ozarks, a deeply etched landscape full of rocky bluffs, hollows spared the prevailing winds, creeks and rivers, vegetation shaped by aspect, with north slopes more damp and shaded than south. Those terrain features define the geographic matrix of fire propagation – the landscape’s fire behavior pixels, as it were. Hardly can a fire get a head of steam before it runs against a shift in fuels or an outright fuelbreak. The broken landscape makes for broken fire spread. What decides which pixels get burned and what propagates fires across the pixels is people.

      This argument can be generalized. A kindled fire will fill up its “natural” landscape matrix pixels, whose dimensions are set by the normal fire-behavior troika. On the High Plains this might extend for hundreds of miles; in the sky islands of the Southwest, it might consist of a side canyon. The size of the pixels and roughness of the matrix determine how many fires are needed to burn over a landscape. There are places where this dynamic plays out under wholly natural circumstances: lightning sparks a fire that rides the passage of a cold front over a vast estate, or so much lightning pummels a corrugated mountain that the burned patches eventually merge. Moreover, it is possible for a fire to creep between old burns, in effect, restarting a blaze within a new unit, sweeping out each in turn. But such processes don’t work in the Ozarks where lightning-kindled fire is rare and where the pixel matrix is as intricate as the etched and rocky uplands. Fires occur in the Ozarks because people have chosen directly to ignite them or indirectly to let them happen by not attempting to stop them or by shaping the larger landscape.

Closer view of landscape texture. Fire records indicate that fires, set by people, tended to conform to topographic pixels, the natural boundaries of fire behavior units. Photo: Stephen Pyne  View Middle Ground slideshow

The truly revolutionary corollary is that people are essential to propagation: they carry fire from one pixel to another, they determine how the landscape will fill with flame. Where the roughness index is low, a few people can set fires adequate to flood the landscape. Where the index is high, an equivalent burning demands more people or people with other leverage by which to overcome the resistance the landscape offers. Thus fires and the area they burn vary with human population but in accordance to the character of the matrix of fire-behavior pixels, and this is why the fire history in the Ozarks tracks human migration so tightly.

But it is possible to go further with this insight. Richard Guyette, for example, has restated the Arrhenius equation for wildland fire. In his model people provide the critical “activation energy,” just as in a lab setting they kindle a Bunsen burner. Across a landscape it is people who start fires and then keep them going. Without someone to ignite that burner, the fuel, oxygen, and other physical parameters are meaningless. So it is with many landscapes capable of burning but that don’t burn under wholly natural conditions. In the Ozarks some human has to supply the activation energy to begin the reaction. 

In truth, this happens all the time in wildland fire management as fire crews set prescribed fires and backfires, and burn out along control lines or the borders of pre-determined patches. (How much of the 1988 Yellowstone fire complex was the result of failed burnouts?) Historically, giant fires got big not from purely conditions but from protective burning as settlers rushed to fire around their valued fields and structures; each new smoke column inspired others to burn. A thousand people setting backfires in an afternoon can pump up a thousand-acre wildfire into a million-acre one quickly. Here is a mode of fire propagation far more powerful than radiant heating or even spotting. The firebrands do not have to ride the wind, only reside in nervous, willful hands.

To link fire behavior with fire history, extend this reasoning to account for the movement of fire over time, or the persistence of fire on the land across epochs defined by features other than fire. Consider the lumpiness of history as another roughness index. In this way we can address the chronological patchiness of a place, the sense in which the fire practices of one era can propagate (or not) into another. The Ozarks again offer an exemplary study site. They are a contained landscape feature that has, over the past 600 years, undergone a series of depopulations and repopulations. What this historical model shares with the geographic model is the vital role of people. The comings and goings of people drive the rise and fall of fires. They carry fire across the pixels of history.

In many landscapes people compete against natural fire, and in most, they compete with other species of anthropogenic fire. The Ozarks offer a special case where natural fire is absent and the land has experienced more or less complete cycles of human migration. The historical roughness index is high. What research says is that for history, as for geography, people carry fire from patch to patch. By converting the fire saga of the Ozarks into quantitative date – by tracking fires through the scars they have left on shortleaf pine and bur oak, by translating the converted landscapes into fuel loads through time – researchers including Rich Guyette, Dan Dey, Rose-Marie Muzika, and Mike Stambaugh – have sketched the basis for a general model of anthropogenic fire and have made a case that fire’s behavior on the land cannot be understood apart from human behavior.