12 Dec 2007, 1:28pm
by admin

Historical Fire Cycles in the Canadian Rocky Mountain Parks

Van Wagner, Charles E., Mark A. Finney, and Mark Heathcott. Historical Fire Cycles in the Canadian Rocky Mountain Parks. Forest Science 52(6) 2006, (704-717).

Charles E. Van Wagner, Canadian Forest Service (ret.), Mark A. Finney, USDA Forest Service, Rocky Mountain Research Station, Fire Sciences Laboratory, and Mark Heathcott, Parks Canada.

Full text [here]

Review by Mike Dubrasich

A remarkable and historic forest science paper was published last December in Forest Science, the leading US scientific journal about forest science. The paper is remarkable for a half dozen or more reasons, and in this essay we (attempt to) sort them out and explain them.

First, the authors are the cream of the crop. Charles E. Van Wagner is the unofficial dean of Canadian fire science. No one has advanced the science more, up there and few down here, during his lifetime. Finney and Heathcott are equally experienced grey beards of fire ecology. Hidden from direct view are dozens of field and laboratory researchers who contributed to the data collection for this paper, over a period of more than two decades.

Previous accounts of data collection and fire history have been published for all seven parks, some more than once. Jasper fire scar data were studied by Tande (1979a, b) but no formal reference exists for the whole-park age-class survey of 1987 to 1990. These data are on file at Jasper National Park; sampling work was begun by B. Wallace and G. Fenton, completed and mapped by S. Cornelsen, and finally compiled by R. Kubian. Fire history and age-class data for Banff National Park were reported by White (1985), Rogeau and Gilbride (1994), and Rogeau (1996); for Kootenay National Park by Masters (1990); for Yoho National Park by Tymstra (1991); for Peter Loughheed (formerly Kananaskis) Provincial Park of Alberta by Hawkes (1979, 1980), Johnson (1987), Johnson and Fryer (1987), and Johnson and Larsen (1991); for Mount Assiniboine and Spray Lakes Provincial Parks of Alberta by Rogeau (1994 a, b).

In addition, the authors acknowledge “Parks Canada for providing the data, and Ian Pengelly, Cliff White, and Stephen Woodley, all of Parks Canada, for helpful comment and interest.”

Historical Fire Cycles in the Canadian Rocky Mountain Parks is about the fire history of seven contiguous national and provincial parks in the Canadian Rockies. They include Banff, Jasper, Kootenay, and others. Their combined total area is 21,900 sq. km., or 13,600 sq. miles. Of those, 6,300 sq. miles are forest, and the rest are rock, ice, water, or treeless vegetation.

No matter what units are used, that is a very large chunk of forest for a study, and one of the paper’s remarkable features. Another is that the study took over twenty years to complete and is the work of dozens of researchers. Another is that the fire dates they discovered go back to 1280 AD. To my knowledge, no other fire study has ever come close to the breadth of acreage and time comparable to Historical Fire Cycles in the Canadian Rocky Mountain Parks.

Another remarkable feature is the principal finding of the study. Historically, forest fires in the Canadian Rockies have not been controlled by climate or random chance. Van Wagner et al. disproved those hypotheses, with an intensive yet elegant work of deductive science.

What the authors did not disprove, and further postulate as a heuristic hypothesis in the Discussion and Interpretation section, is that human agency has been responsible for fire ignitions for a very long time in the Canadian Rockies. Furthermore, the study provides strong evidence for anthropogenic fire, dating back millennia, in some of the (arguably) most remote forests on the continent.

If people controlled the fire cycles way the heckandgone out there in Jasper NP Canada, then surely people controlled the fire cycles everywhere south of Jasper, that being mostly everywhere.

The data that the study produced included the age-class mapping of the entire area. Every forest stand in 6,300 sq. miles of forest had to be aerially photographed, mapped, walked, measured, and analyzed. From that massive data set Van Wagner et al. computed the fire cycle over eight centuries.

Van Wagner utilized the concept that the fire cycle (or frequency) is the reciprocal of the burning rate, the acreage burned per year. For instance, if the fire cycle is 50 years, then 2 percent of the total area in question burns up every year. Reciprocally, if 2 percent of the area burns each year, the average fire cycle on any acre is 50 years. This is a tricky concept at first encounter, but if you think about, it makes perfect sense.

The main parameter sought from these age-class data is an estimate of the fire cycle, namely the length of time required to burn an area equal to the area-at-large (Heinselman 1973, Van Wagner 1978, Suffling et al. 1982); it is also equal to the average interval between fires at any given point (Johnson and Van Wagner 1985, Johnson and Gutsell 1994). The fire cycle is, in fact, the reciprocal of a more basic parameter, the mean annual burning rate (i.e., the proportion of area burned per year). These parameters rest on the assumptions that the fire climate and the ignition regime are reasonably stable over the time period of interest. Likewise, the forest and its flammability are assumed to be more or less constant and uniform over the area at large.

Because they had the age-class acreage data, Van Wagner et al. were able to calculate the fire cycles (east and west of the Continental Divide). It was more than a nifty science trick; it was an exercise in the fundamental logic of fire science (a concept and a discipline that Charles Van Wagner is in large part responsible for creating). The authors were also able to track the fire cycles (frequency) over time, to see if they had changed at all over the centuries.

And sure enough, the fire cycles had changed:

The principal conclusions were: The seven parks as a unit supported a fire cycle of 60-70 yr for nearly five centuries before 1760. The burning rate then dropped sharply and a longer cycle of about 175 yr prevailed until 1940. Fire history in the combined large parks east side was similar to above. By contrast, the smaller west side parks group sustained a fire cycle of about 90 yr for at least five centuries before 1840; the burning rate then decreased irregularly to 2006.

For five centuries the forests of the Canadian Rockies burned at a rate of 1 to 2 percent per year. Then in 1760 the burning rate on the eastside went down to half that, and a similar drop-off occurred in 1840 on the westside.

No climatic shift or change has been found that might explain the sudden decrease in burning:

A number of such studies [tree ring climate histories] have been published, e.g. Case and MacDonald (1995), Luckman (1998), Watson and Luckman (2001, 2004). The general impression from all these sources is of frequent variations in rainfall and temperature about long-term means, on scales of several years to about three decades; nor does “cool” always pair with “wet” nor “warm” with “dry”. Nowhere, however, during five centuries of record, is there any evidence of a long-term climate shift that could account for a marked semi-permanent reduction in burning rate in the Rocky Mountain Parks.

Both the eastside and westside forests experienced another decrease in burning rate around 1940. This change is attributed to organized, modern fire suppression. Nobody accuses climate change of being the causal factor in 1940, because it is rather obvious that humanity was the agency. Furthermore, there is ample evidence that lighting could not have been the principal forest fire ignition source historically.

There is plain evidence that lightning fires east of the divide have in recent times been comparatively rare. White (1985) compiled the fire record for Banff from 1880 to 1980, and found only seven of 43 fires greater than 40 hectares were of lightning origin; large human-caused fires account for the great bulk of the burned area. Furthermore, almost all these large fires predated the era of intensive fire control beginning about 1940.

Van Wagner et al. also noted that in the modern era lightning fires are more frequent on the westside, yet the westside had a longer fire cycle historically than the eastside. Unless lightning frequency and general location made radical shifts in centuries past, lightning cannot have been the cause of most of the historical fires in the Canadian Rockies.

Whether lightning fire occurrence has been more or less constant over the centuries is naturally open to question. If it has, then some other ignition source must explain the short 60-yr cycle prevailing before 1740 in the east side parks.

And the leading suspect is humanity. There are no other credible hypotheses that can explain the anomalous behavior of the fire cycles.

The argument for human agency supposes that the aboriginal people on either or both sides of the divide had, for perhaps a very long time, periodically burned the forest to improve access and hunting; other possible reasons might be listed. Christensen (1971), for example, provides evidence of human habitation within the mountains extending back for several millennia.

Why then would the burning rate diminish in the mid-1700’s?

This argument further supposes that the traditional aboriginal land-use pattern was first disrupted by the arrival of horses and guns after about the year 1700 (Jenness 1963). Serious depopulation by smallpox then followed throughout the 18th century. The presence of the mountain barrier may have resulted in different time sequences on the west side.

This same pattern of frequent, regular fire for millennia, followed by sharply diminished fire concurrent with destruction of Native American populations, is the same pattern seen in forests across the western US, and was the pattern once seen in the East and Midwest, too. The pattern leads to one conclusion: anthropogenic fire was widespread across the continent for millennia.

Anthropogenic fire shaped our forests. It wasn’t Mother Nature alone. After the last glacial stadial, people arrived here before the forests did. People have been shaping our forests since the beginning of the Holocene.

People have been the generators and stewards of our forests for millennia. Kind of weird if you think about it, probably not what you’ve been taught and not what you read most places, but those are the facts, or at least the best hypothesis so far. Human beings are the only animals that can make and spread fire, and we have been doing so since our species arose in Africa some 150,000 years ago.

Van Wagner et al. went to some lengths to apply different statistical methods to the data. The authors used non-parametric (data driven) survival analysis techniques. Survival analysis is a method of statistically evaluating lifetimes, or data in the form of durations of time, like the intervals between fires.

The authors also used a parametric (model driven) approach, the negative exponential distribution. The negative exponential distribution, which is the same as the exponential distribution (EXP), is a mathematically limiting (continuous) form of the (discrete) geometric distribution (GEO).

Here’s an example of the GEO. Suppose in a blackjack game you have a probability p of being dealt a winning a hand. Then the probability you will need three hands before you get a winner is distributed GEO (3,p).

Another example: if the probability that an acre will burn in any year is p, then the probability that the acre will require X years before burning is distributed GEO (X,p).

Both the geometric and the exponential distributions have a “no-memory” property. This is related to the assumption of a constant failure intensity, implying no wearout or change in the probability p that a stand will burn in any given year. In Historical Fire Cycles in the Canadian Rocky Mountain Parks the assumptions that ignitions and burns are independent of stand age and species are important requirements of a “no memory” constant failure process. That is, those assumptions are necessary for the chosen statistical model.

Both of these parametric distributions are based on the laws of probability. They are used to detect deviations from random chance. That is their purpose in the scientific method.

As it turned out, the data fit the exponential distribution, but not exactly in the way they should have. Instead, three distinct patterns emerged: one before 1760 (or 1840 on the westside), one between 1760 (or 1840 on the westside) and 1940, and one after 1940. That meant the burning rates weren’t constant over time. If there was no change in climate and lightning frequency (eastside and westside), then some deterministic agent was playing with fire-loaded dice. Most probably it was the human residents.

This is how science is done, people. This is the real thing. Pay attention.

Van Wagner et al. began their study with a testable hypothesis: that the fire cycles in the Canadian Rockies were natural. They opened their paper thusly:

Most of the Canadian forest is, in a state of nature, recycled and renewed by random periodic fire.

Then they tested that hypothesis and found very strong evidence against it, enough to logically disprove it. The “randomness” of periodic fire changed over the centuries. Therefore something had affected the fire frequency besides random chance. That’s the Scientific Method. Heavyweights Sir Karl Popper and Thomas Kuhn would both be proud.

The Mother Nature Alone Hypothesis is a big one. It is a widely held belief popularly, and within the scientific community. Van Wagner et al. chose a fundamental hypothesis to test. And found it to be flawed. The fires were not random periodic. The fire cycles were not entirely “natural.”

Van Wagner et al. in effect scientifically disproved the Old Paradigm. The Old Paradigm holds that Mother Nature has been in charge, that natural forces shaped our landscapes, that people were absent from the natural “wilderness,” and that people can only do harm to nature.

The empirical data, carefully analyzed, says otherwise.

There are implications to the findings of this study. The authors are well aware of this:

The purpose of Canada’s national parks, and of some provincial ones, is to reserve representative samples of landscape along with their plant and animal life in a natural state. But such preservation is obviously dynamic—continuation of the normal vegetation renewal process is implied. The historical rate of renewal in parks with fire-oriented forest is therefore of great interest in the formulation of policy within the parks mandate (Van Wagner and Methven 1980).

The normal vegetation renewal process is and has been human-mediated. People have been tending the landscape for millennia, even in places as remote as the Canadian Rockies. Our forests and other landscapes are cultural and historical. People have been the dynamic force. As the agents and masters of fire, we have been the artists and actors who have helped sculpt and care for forests and nature, here and around the world, ever since we first evolved, or were ourselves created, on this planet.

That’s a useful thing for Parks Canada to know. Scholars will be citing Historical Fire Cycles in the Canadian Rocky Mountain Parks for a very long time. It is an historic paper, as well as an historical one.

This review, in a slightly different form, was posted at SOS Forests (original version) in March, 2007. Charles Van Wagner graciously responded:

Dear Mr. Dubrasich,

This is in reply to your message of March 11, with your review of the paper “Historical fire cycles in the Canadian Rocky Mountain parks.” My colleagues and I thank you for your positive reaction and complimentary remarks. If I comment at all it is to clarify our purpose and to reorder the emphasis in your review.

Our primary goal was to analyse the available data and to measure the fire cycles at work in the parks forest over several centuries past, as background for present park fire management policy. To do this legitimately, we were obliged to draw in as many mathematical methods as possible. Their common theoretical framework, as we described, was the negative exponential age distribution, which is based on the primary assumption that fire strikes and burns in this forest without regard for age or forest type. The Methods section lays this out in some detail, in part because the subject is quite controversial. The fact that the data line themselves up in fairly acceptable straight lines justifies this primary assumption. Changes in slope, then, signify changes in average burning rate from period to period.

Following a comparison of the results of the various methods, our main numerical conclusions appear toward the end of Discussion and Interpretation. These basically tell the story of the fire cycles prevalent during the period of study as we found them.

The discussion of Prior Causes of Change in Fire Regime is a distinctly secondary feature of the paper. We did not at any time set up a hypothesis about the role of the native people in times past. Rather, statistical philosophy obliged us to discuss all possible reasons for any apparent fire regime changes and the timing of the transitions. The recent literature on tree-ring chronology seemed to us to disqualify the possibility of distinct climate changes throughout the whole history. For the early transitions, the other main possible cause is, of course, human ignition. You will notice, however, that this argument is delivered in the impersonal sense, not from our direct knowledge or conviction. The evidence available to us is circumstantial, marking human agency as a distinct possibility, but requiring firmer evidence for certainty. Our key sentence, then, is “All the above mixture of evidence and speculation cannot be resolved here.” Please note also that this story cannot be extended eastward over the bulk of the Canadian boreal forest, where quite different considerations would apply.

Once a paper is in print, however, any reader may take from it any further implications he deems justified. Finally I should also emphasize that this paper was a truly cooperative effort that took all three of us to bring to a successful conclusion. Once again, we are grateful for the interest you have shown in this fascinating subject and our venture into it.


Charles Van Wagner

  • Colloquia

  • Commentary and News

  • Contact

  • Topics

  • Archives

  • Recent Posts

  • Meta