united states department wilderness science in a …wilderness science in a time of change...

United StatesDepartmentof Agriculture

Forest Service

Rocky MountainResearch Station

ProceedingsRMRS-P-15-VOL-5

September 2000

Wilderness Science in aTime of Change ConferenceVolume 5: Wilderness Ecosystems,Threats, and ManagementMissoula, MontanaMay 23–27, 1999

AbstractCole, David N.; McCool, Stephen F.; Borrie, William T.; O’Loughlin, Jennifer, comps. 2000.

Wilderness science in a time of change conference—Volume 5: Wilderness ecosystems, threats,and management; 1999 May 23–27; Missoula, MT. Proceedings RMRS-P-15-VOL-5. Ogden,UT: U.S. Department of Agriculture, Forest Service, Rocky Mountain Research Station. 381 p.

Forty-six papers are presented on the nature and management of threats to wildernessecosystems. Five overview papers synthesize knowledge and research on wilderness fire,recreation impacts, livestock in wilderness, nonnative invasive plants, and wilderness air quality.Other papers contain the results of specific research projects on wilderness recreation impactsand management, wilderness restoration, fire and its management, and issues related to air,water, and exotic species.

Keywords: air quality, campsites, fire, fish stocking, invasive species, livestock, recreation impact,restoration, trails

RMRS-P-15-VOL-1. Wilderness science in a time of change conference—Volume 1:Changing perspectives and future directions.

RMRS-P-15-VOL-2. Wilderness science in a time of change conference—Volume 2:Wilderness within the context of larger systems.

RMRS-P-15-VOL-3. Wilderness science in a time of change conference—Volume 3:Wilderness as a place for scientific inquiry.

RMRS-P-15-VOL-4. Wilderness science in a time of change conference—Volume 4:Wilderness visitors, experiences, and visitor management.

RMRS-P-15-VOL-5. Wilderness science in a time of change conference—Volume 5:Wilderness ecosystems, threats, and management.

Cover art by Joyce VanDeWater, Rocky Mountain Research StationConference symbol designed by Neal Wiegert, University of Montana

You may order additional copies of this publication by sending yourmailing information in label form through one of the following media.Please specify the publication title and number.

Telephone (970) 498-1392FAX (970) 498-1396

E-mail [email protected]

Mailing Address Publications DistributionRocky Mountain Research Station240 West Prospect RoadFort Collins, CO 80526

Compilers

David N. ColeStephen F. McCoolWilliam T. BorrieJennifer O’Loughlin

Wilderness Science in a Timeof Change Conference

Volume 5: Wilderness Ecosystems,Threats, and Management

Missoula, MontanaMay 23-27, 1999

ii

CompilersDavid N. Cole is Research Biologist with the Aldo Leopold Wilderness Research Institute,Rocky Mountain Research Station, located on The University of Montana campus in Missoula,MT. Dr. Cole has A.B. and Ph.D. degrees in geography from the University of California,Berkeley, and the University of Oregon. He has been conducting research on wilderness andits management since the mid-1970’s.

Stephen F. McCool is Professor, Wildland Recreation Management, at the School of Forestry,The University of Montana in Missoula, MT. He holds a B.S. degree in forestry from the Universityof Idaho and M.S. and Ph.D. degrees from the University of Minnesota. His research andapplications projects concern wilderness and protected area management and planning, focusingon management systems, applications of social science to management, public participation, andsustainability questions.

William (Bill) T. Borrie is Associate Professor and Program Coordinator, Outdoor RecreationManagement, in the School of Forestry, The University of Montana. Dr. Borrie received a Ph.D.from the College of Natural Resources, Virginia Tech, and has masters and bachelors degreesfrom the School of Forestry, University of Melbourne, Australia. His research interests are focusedon the outdoor recreation experience and on the meanings of parks and wilderness.

Jennifer O’Loughlin holds a B.A. in journalism and history and an M.S. in environmental studiesfrom the University of Montana. After serving for 10 years as editor of the natural resource journalWestern Wildlands, she turned to a life of free-lance writing and editing.

The use of trade or firm names in this publication is for reader information and does notimply endorsement by the U.S. Department of Agriculture of any product or service.

The USDA Forest Service is not responsible for statements and opinions advanced in thispublication. Authors are responsible for the content and quality of their papers.

CAUTION:PESTICIDES

Pesticide Precautionary Statement

This publication reports research involving pesticides. It does notcontain recommendations for their use, nor does it imply that the usesdiscussed here have been registered. All uses of pesticides must beregistered by appropriate State and/or Federal agencies before they canbe recommended.

CAUTION: Pesticides can be injurious to humans, domestic animals,desirable plants, and fish or other wildlife—if they are not handled orapplied properly. Use all pesticides selectively and carefully. Followrecommended practices for the disposal of surplus pesticides andpesticide containers.

iii

The Wilderness Science in a Time of Change Confer-ence was held in Missoula, Montana, May 23 through27, 1999. The conference was conceived to be both afollowup and an expansion of the first National Wil-derness Research Conference, held in Fort Collins,Colorado, in 1985. That conference brought togethermost of the scientists in the world who are working onissues related to the management of wilderness andresulted in literature reviews and compilations ofresearch that remain critical references today (Lucas1986, 1987). Our intent was to bring scientists to-gether again, along with wilderness managers, toproduce an updated compendium of the current state-of-knowledge and current research. In addition, wesought to increase the array of scientific disciplinesrepresented at the conference and to expand the rangeof topics beyond the challenges of managing wilder-ness. Finally, we hoped to use plenary talks to high-light controversy, divergent viewpoints, and manage-ment dilemmas—to challenge participants’ beliefsystems—in the hopes that this would stimulate inter-action and personal growth.

Well over 400 people participated in the conference.Conference attendees included a roughly equal mix ofpeople from federal land managing agencies and fromacademia. There were also several representativesfrom state, local, and tribal governments. There weremore than 30 attendees from 16 different nongovern-mental organizations, as well as a number of privateindividuals, consultants, and members of the press.About 20 participants were from Canada, with about20 more participants from other countries. We suc-ceeded in attracting people from diverse disciplines,united in their interest in wilderness. As usually is thecase, a large proportion of the researchers who at-tended specialize in the social science aspects of out-door recreation. However, attendees also includedother types of social scientists, philosophers, paleon-tologists, and life scientists interested in all scales ofanalysis from cells to the globe.

The conference consisted of plenary talks to bepresented before the entire conference, as well as morenarrowly focused presentations organized around threeconference themes and presented in concurrent ses-sions. The conference’s plenary talks were organizedinto four sessions: (1) global trends and their influenceon wilderness, (2) contemporary criticisms and cel-ebrations of the idea of wilderness, (3) the capacity ofscience to meet the challenges that wilderness facesand to realize the opportunities that wilderness pre-sents, and (4) concluding talks related to conferencethemes.

The bulk of the conference was organized aroundthree themes. The first theme was “Science for Under-standing Wilderness in the Context of Larger Sys-tems.” The emphasis of this theme was better under-standing of the linkages between wilderness and thesocial and ecological systems (regional, national, andinternational) in which wilderness is situated. Theemphasis of the second theme, “Wilderness for Sci-ence: A Place for Inquiry,” was better understanding ofwhat we have learned from studies that have utilizedwilderness as a laboratory. The third and most tradi-tional theme was “Science for Wilderness: ImprovingManagement.” The emphasis of this theme was betterunderstanding of wilderness visitors, threats to wil-derness values, and means of planning for and manag-ing wilderness.

We organized three types of sessions under each ofthese three themes. We invited 18 speakers to presentoverview papers on specific topical areas under eachtheme. Many of these speakers developed comprehen-sive state-of-knowledge reviews of the literature fortheir assigned topic, while others developed moreselective discussions of issues and research they judgedto be particularly significant. In addition, conferenceparticipants were given the opportunity to contributeeither a traditional research paper or to organize adialogue session. Most of the research papers (131papers) were presented orally, but 23 additional pa-pers were presented in a poster session. The 14 dia-logue sessions were intended to promote group discus-sion and learning.

The proceedings of the conference is organized intofive separate volumes. The first volume is devoted tothe papers presented during the plenary sessions.Subsequent volumes are devoted to each of the threeconference themes, with two volumes devoted to wil-derness management, the theme with the most pa-pers. Within each theme, papers are organized intooverview papers, research papers, and papers fromthe dialogue sessions. The format of dialogue sessionpapers varies with the different approaches taken tocapture the significant outcomes of the sessions. Re-search papers include papers presented orally and onposters. Within each theme, research papers are orga-nized into broad topical areas. Although the initialdraft of each proceedings paper was reviewed andedited, final submissions were published as submit-ted. Therefore, the final content of these papers re-mains the responsibility of the authors.

We thank the many individuals and institutions onthe lists of committee members and sponsors that

Preface

iv

follow. They all contributed to the success of theconference.

Planning Committee: Joan Brehm, Perry Brown,David Cole, Wayne Freimund, Stephen McCool, ConnieMyers, and David Parsons.

Program Committee: David Cole (Co-chair), StephenMcCool (Co-chair), Dorothy Anderson, William Borrie,David Graber, Rebecca Johnson, Martha Lee, ReedNoss, Jan van Wagtendonk, and Alan Watson.

Sponsors: Aldo Leopold Wilderness Research Insti-tute; Arthur Carhart National Wilderness TrainingCenter; Bureau of Land Management; Forest Service,Research; Forest Service, Rocky Mountain ResearchStation; Humboldt State University, College of Natu-ral Resources; National Outdoor Leadership School;National Park Service; Parks Canada; State Univer-sity of New York, Syracuse, College of EnvironmentalScience and Forestry; The University of Minnesota,Department of Forest Resources; The University ofMontana, School of Forestry, Wilderness Institute;U.S. Fish & Wildlife Service; and U.S. Geological Sur-vey, Biological Resources Division.

Steering Committee Members: Perry Brown (Co-Chair), David Parsons (Co-Chair), NormanChristensen, Rick Coleman, Chip Dennerlein, DennisFenn, Denis Galvin, David Harmon, John Hendee,Jeff Jarvis, Kenneth Kimball, Luna Leopold, RobertLewis, David Lime, Nik Lopoukhine, JamesMacMahon, Michael Manfredo, William Meadows,III, Chris Monz, Margaret Shannon, Jack WardThomas, and Hank Tyler.

References__________________________Lucas, Robert C., comp. 1986. Proceedings, national

wilderness research conference: current research; 1985July 23-26; Fort Collins, CO. Gen. Tech. Rep. INT-212.Ogden, UT: Intermountain Research Station. 553 p.

Lucas, Robert C., comp. 1987. Proceedings, nationalwilderness research conference: issues, state-of-knowl-edge, future directions; 1985 July 23-26; Fort Collins,CO. Gen. Tech. Rep. INT-220. Ogden, UT: Intermoun-tain Research Station. 369 p.

—The Compilers

v

ContentsPage

David N. Cole Wilderness Ecosystems, Threats, and Management ......................................1Stephen F. McCool

1. Overviews .................................................................................................................................................3

James K. Agee Wilderness Fire Science: A State of Knowledge Review ................................5

Yu-Fai Leung Recreation Impacts and Management in Wilderness:Jeffrey L. Marion A State-of-Knowledge Review .................................................................23

Mitchel P. McClaran Improving Livestock Management in Wilderness ..........................................49

John M. Randall Improving Management of Nonnative Invasive Plants inWilderness and Other Natural Areas .......................................................64

K. A. Tonnessen Protecting Wilderness Air Quality in the United States ..................................74

2. Recreation Impacts and Management ..................................................................................................97

L. Alessa Effects of Soil Compaction on Root and Root Hair Morphology:C. G. Earnhart Implications for Campsite Rehabilitation ..................................................99

Laurel Boyers Twenty-Eight Years of Wilderness Campsite Monitoring inMark Fincher Yosemite National Park .........................................................................105Jan van Wagtendonk

Tracy A. Farrell Camping Impact Management at Isle Royale National Park:Jeffrey L. Marion An Evaluation of Visitor Activity Containment Policies

From the Perspective of Social Conditions ............................................110

Anna M. T. Gajda Managing Coastal Recreation Impacts and Visitor ExperienceJudson Brown Using GIS ..............................................................................................115Grant PeregoodoffPatrick Bartier

Ernest Hartley Thirty-Year Monitoring of Subalpine Meadow VegetationFollowing a 1967 Trampling Experiment at Logan Pass,Glacier National Park, Montana .............................................................124

Mark C. Jewell Assessing Soil Erosion on Trails: A Comparison of Techniques .................133William E. Hammitt

Paul R. Lachapelle Sanitation in Wilderness: Balancing Minimum Tool Policiesand Wilderness Values ..........................................................................141

Yu-Fai Leung Wilderness Campsite Conditions Under an UnregulatedJeffrey L. Marion Camping Policy: An Eastern Example ...................................................148

Christopher A. Monz The Consequences of Trampling Disturbance in Two VegetationTami Pokorny Types at the Wyoming Nature Conservancy’s Sweetwater RiverJerry Freilich Project Area ...........................................................................................153Sharon KehoeDayna Ayers-Baumeister

vi

P. E. Moore Meadow Response to Pack Stock Grazing in the YosemiteD. N. Cole Wilderness: Integrating Research and Management .............................160J. W. van WagtendonkM. P. McClaranN. McDougald

Regina M. Rochefort Human Impact Surveys in Mount Rainier National Park: Past,Darin D. Swinney Present and Future ................................................................................165

Akemi Yoda Erosion of Mountain Hiking Trail Over a Seven-year Period inTeiji Watanabe Daisetsuzan National Park, Central Hokkaido, Japan ...........................172

3. Wilderness Restoration .......................................................................................................................179

David N. Cole Soil Amendments and Planting Techniques: Campsite RestorationDavid R. Spildie in the Eagle Cap Wilderness, Oregon ....................................................181

Sean Eagan Restoration of Multiple-Rut Trails in the Tuolumne Meadows ofPeter Newman Yosemite National Park .........................................................................188Susan FritzkeLouise Johnson

Joseph P. Flood The Influence of Wilderness Restoration Programs on VisitorLeo H. McAvoy Experience and Visitor Opinions of Managers .......................................193

David R. Spildie Effectiveness of a Confinement Strategy in Reducing Pack StockDavid N. Cole Impacts at Campsites in the Selway-Bitterroot Wilderness, Idaho .........199Sarah C. Walker

Charisse A. Sydoriak Would Ecological Landscape Restoration Make the BandelierCraig D. Allen Wilderness More or Less of a Wilderness?............................................209Brian F. Jacobs

Catherine Zabinski Understanding the Factors That Limit Restoration Success onDavid Cole a Recreation-Impacted Subalpine Site ..................................................216

4. Wilderness Fire and Management ......................................................................................................223

Stephen F. Arno Mixed-Severity Fire Regimes in the Northern Rocky Mountains:David J. Parsons Consequences of Fire Exclusion and Options for the Future .................225Robert E. Keane

Anthony C. Caprio Returning Fire to the Mountains: Can We Successfully RestoreDavid M. Graber the Ecological Role of Pre-Euroamerican Fire Regimes to

the Sierra Nevada? ................................................................................233

Peter Z. Fulé Continuing Fire Regimes in Remote Forests of Grand CanyonThomas A. Heinlein National Park .........................................................................................242W. Wallace CovingtonMargaret M. Moore

Thomas A. Heinlein Development of Ecological Restoration Experiments in FireW. Wallace Covington Adapted Forests at Grand Canyon National Park ..................................249Peter Z. FuléMargaret M. MooreHiram B. Smith

vii

Jon E. Keeley Restoring Natural Fire Regimes to the Sierra Nevada in an Era ofNathan L. Stephenson Global Change .......................................................................................255

MaryBeth Keifer Prescribed Fire as the Minimum Tool for Wilderness Forest andNathan L. Stephenson Fire Regime Restoration: A Case Study From the SierraJeff Manley Nevada, California .................................................................................266

Kurt F. Kipfmueller Fire-Climate Interactions in the Selway-Bitterroot Wilderness Area ............270Thomas W. Swetnam

David J. Parsons The Challenge of Restoring Natural Fire to Wilderness ..............................276

Matthew Rollins Twentieth-Century Fire Patterns in the Selway-BitterrootTom Swetnam Wilderness Area, Idaho/Montana and the Gila/Aldo LeopoldPenelope Morgan Wilderness Complex, New Mexico ........................................................283

G.Thomas Zimmerman The Federal Wildland Fire Policy: Opportunities for WildernessDavid L. Bunnell Fire Management ...................................................................................288

5. Air, Water, and Exotic Species ............................................................................................................299

Paul Stephen Corn Fish Stocking in Protected Areas: Summary of a Workshop .......................301Roland A. Knapp

L. Bruce Hill Visitor Perceptions and Valuation of Visibility in the Great GulfWendy Harper Wilderness, New Hampshire ..................................................................304John M. HalsteadThomas H. StevensIna PorrasKenneth D. Kimball

R. A. Knapp Effects of Nonnative Fishes on Wilderness Lake Ecosystems inK. R. Matthews the Sierra Nevada and Recommendations for Reducing Impacts .........312

Marilyn Marler A Survey of Exotic Plants in Federal Wilderness Areas ..............................318

David S. Pilliod Evaluating Effects of Fish Stocking on Amphibian PopulationsCharles R. Peterson in Wilderness Lakes ...............................................................................328

Ellen M. Porter Air Quality Management in U.S. Fish and Wildlife ServiceWilderness Areas ...................................................................................336

6. Wilderness Management .....................................................................................................................341

Shannon S. Meyer Legislative Interpretation as a Guiding Tool for WildernessManagement ..........................................................................................343

S. Thomas Olliff Seeking a Scientific Approach to Backcountry Management inSue Consolo Murphy Yellowstone National Park .....................................................................348

Derek Petersen Grizzly Bears as a Filter for Human Use Management inCanadian Rocky Mountain National Parks .............................................354

Nicholas Sawyer The Development of the 1999 Management Plan for theTasmanian Wilderness World Heritage Area (Australia) ........................362

viii

Michael J. Tranel Wilderness Management Planning in an Alaskan National Park:Last Chance to Do It Right? ...................................................................369

7. Dialogue Session Summary ................................................................................................................375

Peter B. Landres Naturalness and Wildness: The Dilemma and Irony of ManagingMark W. Brunson Wilderness .............................................................................................377Linda MeriglianoCharisse SydoriakSteve Morton

1. Overviews

2. Recreation Impactsand Management

3. Wilderness Restoration

4. Wilderness Fire andManagement

5. Air, Water, andExotic Species

6. Wilderness Management

7. Dialogue SessionSummary

USDA Forest Service Proceedings RMRS-P-15-VOL-5. 2000 1

In: Cole, David N.; McCool, Stephen F.; Borrie, William T.; O’Loughlin,Jennifer, comps. 2000. Wilderness science in a time of change conference—Volume 5: Wilderness ecosystems, threats, and management; 1999 May 23–27; Missoula, MT. Proceedings RMRS-P-15-VOL-5. Ogden, UT: U.S. Depart-ment of Agriculture, Forest Service, Rocky Mountain Research Station.

David N. Cole is Research Biologist, Aldo Leopold Wilderness ResearchInstitute, Missoula, MT 59807 U.S.A., e-mail: [email protected]. Stephen F.McCool is Professor, School of Forestry, The University of Montana, Missoula,MT 59812 U.S.A., e-mail: [email protected]

Wilderness Ecosystems, Threats, andManagementDavid N. ColeStephen F. McCool

The Wilderness Act of 1964 gave wilderness managers adifficult and challenging mandate. Wilderness areas are tobe kept in a wild and natural state—relatively free of humaninfluence and human control. Their value is dependent onthe degree to which they remain unmodified—a contrast tothe highly modified world in which most of us live. However,even the ecosystems in these most protected of public landsare threatened by human activities both internal and exter-nal to wilderness (Cole and Landres 1996). Impacts fromthese activities vary in both intensity and areal extent.Recreation use, often heavy and highly concentrated, hasturned many sites into compacted, erosion-prone places,stripped of vegetation and topsoil. Livestock grazing im-pacts, while absent in a majority of wilderness areas, havebeen profound where they occur, with impacts from currentgrazing practices often less pronounced than those of thepast (Vankat and Major 1978). The impacts of fire suppres-sion, while less intense, are widespread. Huge acreages ofwilderness have already experienced profound changes invegetation structure as a result of this activity. Air pollutioneffects may be even more pervasive and problems with exoticinvasions are increasing all the time.

As recognition of the prevalence and severity of humanimpact in wilderness increases, pressure to restore wilder-ness ecosystems—to compensate for human influence—mounts. Some managers are advocating intentional ma-nipulation of wilderness ecosystems—from thinning ofvegetation and management-ignited fire to liming of waterbodies and genetic manipulation. This raises the seriousdilemma of whether it is best to emphasize naturalness orwildness in wilderness—whether to minimize human influ-ence or human control (Cole 1996).

Science is needed to provide a foundation for appropriatemanagement of wilderness ecosystems. Rich research tradi-tions in the fields of wilderness recreation impact and firehave contributed to relatively firm scientific bases for deal-ing with these threats. Air quality programs, strengthenedby the mandates of the Clean Air Act, are also relatively welldeveloped. Most other threats to wilderness ecosystemshave received even less attention. This problem is aggra-vated, moreover, by the fact that many scientists who work

on large undisturbed ecosystems make little attempt toapply their knowledge to wilderness management.

Managers need research on the nature and significance ofa wide variety of anthropogenic impacts, as well as anunderstanding of factors that influence impact characteris-tics. They need an improved understanding of natural con-ditions and processes and the extent to which existingconditions deviate from natural conditions. They need prac-tical indicators and techniques for assessing conditions andmonitoring deviation from natural or acceptable conditions.Armed with this knowledge, managers should be in animproved position when deciding where and what manage-ment is appropriate.

This volume is devoted to research on human activitiesthat threaten the integrity of wilderness ecosystems, im-pacts of those activities, and management approaches thatminimize these impacts. It is organized into seven sections.The first section provides five overview papers, one on eachof five major threats. Yu-Fai Leung and Jeff Marion providea comprehensive overview of the field of recreation ecologyand update the synthesis of recreation impact researchprovided in the proceedings of the first wilderness scienceconference (Cole 1987). Jim Agee synthesizes the rich re-search tradition on fire and its management in wilderness,again updating a review developed for the first scienceconference (Kilgore 1987). Research on air quality issuesand their management in wilderness, another topic coveredin the first science conference (Schreiber and Newman1987), is reviewed by Kathy Tonnessen. The final two over-view papers provide research syntheses and perspectives onthreats that were not addressed at the first science confer-ence. Mitch McClaran examines livestock management inwilderness, while John Randall covers management of alienplants.

The second section consists of research papers on recre-ation impacts and their management. While some of thesepapers improve our understanding of the fundamental na-ture of recreation impacts, many are devoted to assessmentand management of impacts. Papers in the third sectiondeal with wilderness restoration. Most of these papers areconcerned with restoration of sites damaged by recreationuse. Papers on restoration of fire in wilderness are in-cluded in the fourth section, along with other researchpapers on fire regimes, impacts associated with suppres-sion of fires, and appropriate fire management in wilder-ness. The few research papers presented on air, water, andexotic species issues are collected in the fifth section. Broadpapers on wilderness management and planning are col-lected in the sixth section. The final section consists of theone dialogue session included in this volume, a sessiondevoted to the dilemma of manipulative restoration of wil-derness ecosystems.

2 USDA Forest Service Proceedings RMRS-P-15-VOL-5. 2000

References _____________________Cole, David N. 1987. Research on soil and vegetation in wilderness:

a state-of-knowledge review. In: Lucas, Robert C., comp. Pro-ceedings—National wilderness research conference: issues, state-of-knowledge, future directions; 1985 July 23-26; Fort Collins,CO. Gen. Tech. Rep. INT-220. Ogden, UT: U.S. Department ofAgriculture, Forest Service, Intermountain Research Station:135-177.

Cole, David N. 1996. Ecological manipulation in wilderness: anemerging dilemma. International Journal of Wilderness. 2: 12-16.

Cole, David N.; Landres, Peter B. 1996. Threats to wildernessecosystems: impacts and research needs. Ecological Applications.6: 168-184.

Kilgore, Bruce M. 1987. The role of fire in wilderness: a state-of-knowledge review. In: Lucas, Robert C., comp. Proceedings—National wilderness research conference: issues, state-of-knowl-edge, future directions; 1985 July 23-26; Fort Collins, CO. Gen.Tech. Rep. INT-220. Ogden, UT: U.S. Department of Agriculture,Forest Service, Intermountain Research Station: 70-103.

Schreiber, R. Kent; Newman, James R. 1987. Air quality in wilder-ness: a state-of-knowledge review. In: Lucas, Robert C., comp.Proceedings—National wilderness research conference: issues,state-of-knowledge, future directions; 1985 July 23-26; FortCollins, CO. Gen. Tech. Rep. INT-220. Ogden, UT: U.S. Depart-ment of Agriculture, Forest Service, Intermountain ResearchStation: 104-134.

Vankat, John L.; Major, Jack. 1978. Vegetation changes in SequoiaNational Park, California. Journal of Biogeography. 5: 377-402.

1. Overviews

3


In: Cole, David N.; McCool, Stephen F.; Borrie, William T.; O’Loughlin,Jennifer, comps. 2000. Wilderness science in a time of change conference—Volume 5: Wilderness ecosystems, threats, and management; 1999 May 23–27; Missoula, MT. Proceedings RMRS-P-15-VOL-5. Ogden, UT: U.S. Depart-ment of Agriculture, Forest Service, Rocky Mountain Research Station.

James K. Agee is Professor of Forest Ecology, College of Forest Resources,University of Washington, P.O. Box 352100, Seattle, WA 98195-2100 U.S.A.

Wilderness Fire Science: A State-of-Knowledge ReviewJames K. Agee

Abstract—Wilderness fire science has progressed since the lastmajor review of the topic, but it was significantly affected by thelarge fire events of 1988. Strides have been made in both firebehavior and fire effects, and in the issues of scaling, yet much of theprogress has not been specifically tied to wilderness areas orfunding. Although the management of fire in wilderness has beenslow to recover from the fires of 1988, science has progressed mostsignificantly in its ability to deal with fire at a landscape level. Majorchallenges include better understanding of the regional context andfunction of wilderness areas, as well as understanding and incorpo-rating fire patchiness, variability and synergistic disturbance fac-tors into predictive models. If more precise models are to be appliedaccurately in wilderness, better weather databases are essential.

Wilderness fire has presented both managers and scien-tists with considerable challenges over the 30 years thatwilderness fire programs have been operational. Wildernessfire, in its purest form, should be “wild” fire: unfettered bythe constraints of humans. We have never prescribed a “let-it-blow” policy for tornadoes and hurricanes, a “let-it-erupt”policy for volcanoes or a “let-it-grind” policy for glaciers.Why, then, did we need a “let-it-burn” policy for fires, orsurrogate strategies like prescribed fire? Humans and firehave an inseparable history (Pyne 1995). We have been ableto control fire for human purposes for thousands of years andfind it very difficult to “let wild fire loose” (Pyne 1989). Thereare some good reasons for this reluctance, including theissues of safety to humans and damage to resources andproperty. As much as we have tried, we have not been ableto find areas large enough to “let wild fire loose,” and this hasbeen at the root of the challenges to research and manage-ment over three decades. It remains a primary challengetoday.

The literature on fire in wildlands is immense. As inevery field, some of it is hardly worth printing, while someis insightful and informing. In this review, I cannot covereven the entire latter category, and do not attempt acomplete literature review by any means. My objective is tosummarize the major trends in wilderness fire sciencesince its inception, with a focus on recent times, and todefine scientific challenges for the future. Fortunately,there are a number of major conference proceedings that

have synthesized fire research over the past decades andallow somewhat cursory coverage in this review. In chrono-logical order, they include: Fire Regimes and EcosystemProperties: Proceedings of the Conference (Mooney andothers 1981); Proceedings – Symposium and Workshop onWilderness Fire (Lotan and others 1985); National Wilder-ness Research Conference: Issues, State-of-Knowledge,Future Directions (Lucas 1986a) and National WildernessResearch Conference: Current Research (Lucas 1986b);Fire and the Environment: Ecological and Cultural Per-spectives (Nodvin and Waldrop 1991); and Proceedings:Symposium on Fire in Wilderness and Park Management(Brown and others 1995a). In addition, there is the once-annual and now-periodic Tall Timbers Fire Ecology Con-ference proceedings which contain significant materialrelated to wilderness areas. Several books are availablethat provide specific geographic or disciplinary informa-tion about fire: Fire and Ecosystems (Kozlowski and Ahlgren1974); Fire Ecology of the United States and SouthernCanada (Wright and Bailey 1982); and Fire Effects onEcosystems (DeBano and others 1998). Some regional treat-ments have been possible where information is abundant:Fire and Vegetation Dynamics: Studies from the BorealForest (Johnson 1992); and Fire Ecology of Pacific North-west Forests (Agee 1993).

Definitions of fires have changed over the past decades,most recently in 1997. I have attempted to be faithful to thenew terminology where possible, but doing so is awkward.The first natural fires allowed to burn were called “let-burn”fires, but that phrase conjured up an impression of nomanagement at all. It was changed to “prescribed naturalfires” in the 1970s as part of a tripartite division of fires:wildfires, which were unwanted fires of any origin, pre-scribed fires, which were manager-ignited fires, and pre-scribed natural fires. All of these fires were called wildlandfires, as they occurred in wildlands, in contrast to structuralfires. In the mid-1990s, Federal fire policy was reviewed, anda new terminology was created. Prescribed fire remained aseparate category, and all other fires were classed as “wild-land fires,” which was somewhat confusing as that phrasereferred previously to all fires in wildlands. The wildlandfire category was subdivided into (1) wildfires (unwantedwildland fires) and (2) wildland fires that might be managed(those of natural origin burning within a predeterminedzone and within prescription limits of some type): the oldprescribed natural fire. Unfortunately, there has been noformal phrase adopted for these fires: Prescribed naturalfire is now defined by what it is not (not a prescribed orunwanted wildland fire). A logical name such as “managedwildland fire” is not very descriptive or formally used, so Iwill continue to call these fires “prescribed natural fires,” or“pnf.”


Historical Evolution of Fire ScienceApplied to Wilderness ___________

The recognition of ecological process as a major manage-ment objective for parks and wilderness came of age in the1960s. Before then, of course, there were national parks andmonuments managed by the National Park Service (NPS)and designated wild areas and primitive areas, as well asconsiderable unroaded but unclassified lands, managed bythe Forest Service. Fire was suppressed in all of these units,except for experimental burning in Everglades NationalPark (Robertson 1962). Three major public policy shiftsoccurred in one decade: the Leopold Report (1963), theWilderness Act (1964) and Department of the Interior firepolicy (1968) that recognized natural processes, includingfire, as valid objectives of management. The Leopold Reportwas generated by a wildlife controversy in YellowstoneNational Park, but its chair, A. Starker Leopold, broadenedthe report to a grand vision of the purposes of national parks(Leopold and others 1963).

The report recognized that the primitive landscapes ofAmerica were, in large part, products of disturbance, includ-ing fire, and that in the long run, management would only besuccessful if it was to manage these disturbances, ratherthan just suppress them. The authors were somewhat pes-simistic that this could ever occur, but dreamed of recreatingthe “…vignette of primitive America...at least on a localscale.” The report was very radical for its time and wascirculated by the DOI for a year before Secretary of theInterior Udall accepted it. That year, 1964, was the sameyear the Wilderness Act passed and was signed into law byPresident Johnson. It defined wilderness as an area “…un-trammeled [unaffected] by man,” “…affected primarily bythe forces of nature...” and “…managed to preserve itsnatural conditions.” The Leopold Report and the WildernessAct provided similar guidance to scientists and managers.Clearly, the natural force missing from almost every parkand wilderness area was fire: How could it be reintroducedto these systems? There was no regulatory guidance for anoperational application of fire management until 1968, whenthe DOI released its new fire policy, based on the concepts ofthe Leopold Report. This new policy not only recognizedprescribed fire as a legitimate action, but also sanctioned theuse of natural fires where appropriate.

Within the same year, a fire management program wasinstituted at Sequoia and Kings Canyon National Parks,accompanied by a research program that investigated theeffect of these programs on fuels, flora and fauna (Kilgoreand Briggs 1972). Yosemite National Park followed in 1970.These parks had primarily low- and moderate-severity fireregimes (c.f. Agee 1993), where fire historically was fairlyfrequent and few of the fires were of stand-replacementintensities over large areas (mixed-conifer/pine, red fir).Higher-severity chaparral areas were avoided in the initialyears. The broad granite terrain of these parks also helpedcontain fires to individual valleys: Long wind-driven intensefire runs were uncommon there. The early research there(Agee 1973; Biswell 1961, 1967; Hartesveldt 1964; Kilgore1971a,b, 1972, 1973; Parsons 1976, 1978; van Wagtendonk1972, 1974, 1978) clearly showed that prescribed fire couldbe valuable in moving ecosystems back to more naturalconditions, without unacceptable resource damage, and that

prescribed natural fire could be successfully managed (Kilgoreand Briggs 1972). Although Forest Service research hadbeen helpful to the NPS scientists and managers, in bothresearch and application the NPS was a leader by the early1970s (van Wagtendonk 1991a).

Yellowstone National Park began a prescribed naturalfire program in 1972 (Romme and Despain 1989). Researchand monitoring there found two seemingly apparent pat-terns: (1) Fires tended to burn primarily in old-growth forest(Sweaney 1985) and naturally extinguished themselves atthe boundary of younger forest; and (2) very large fires werecharacteristic of the distant past (Romme 1982). Romme’swork was somewhat consistent with the monitoring, in thathe found older forest to be more flammable than youngerforest. But his reconstruction of the Yellowstone landscapesince the early 1700s suggested an ecosystem never inequilibrium or stability at any park scale, due to large eventsat infrequent intervals. The implications of these findingswere never addressed by the fire management plan forYellowstone, although they were available almost a decadebefore the fires of 1988.

The Forest Service began a similar wilderness fire pro-gram in the Selway-Bitterroot Wilderness in northern Idahoin 1972. This area contained forest types in moderate- andhigh-severity fire regimes (Brown and others 1995b), andthe second fire that was allowed to burn (Fritz Creek 1973)escaped, burning about 500 ha outside of the managementunit (Daniels 1974). The fire had been monitored during theburn, and research work was initiated after the smoke hadcleared (Mutch 1974). The program was continued, althoughit was later described by the agency as meeting with “mod-erate” success (Towle 1985). In 1978 the Forest Serviceadopted a nationwide “appropriate response” suppressionstrategy that more clearly allowed this type of integratedfire management. A naturally occurring ignition, under thispolicy, could be declared a wildfire, but limited resourcesmight be directed to suppress it. Manager-ignited prescribedfire was not allowed in designated Forest Service wildernessthrough the mid-1980s.

The adoption of wilderness fire management plans thatincorporated prescribed fire or prescribed natural fire blos-somed in the 1980s. Associated with this increase wereextensions of plans into primarily high-severity fire regimesand the increase in both prescribed fire and prescribednatural fire (Botti and Nichols 1995). Management wasclearly moving faster than research, partly because of lim-ited funding for park and wilderness research, and thelimitations of science to address operational concerns.

Limitations of the Science Throughthe Mid-1980’s __________________

The primary limitation posed by science for wildernessuntil the fires of 1988 was the dissolving paradigm ofsuccessional theory. The fading of a firm theoretical model(classical Clementsian climax theory) to apply to distur-bance in natural ecosystems allowed managers to viewreintroduction of fire as a “good” thing without much atten-tion to either what fire was doing or where it might go.Ecological problems with some fire programs were difficultto solve because of a lack of records on where burns occurred


and a lack of monitoring of the fires’ effects on resources(Thomas and Agee 1986).

The classical view of shifting paradigms (Kuhn 1970) wasthat after an accepted model of science (a paradigm) wascreated, evolving research would accumulate evidence sug-gesting the current paradigm was too simple or just wrong.Eventually, a relatively rapid shift towards a more robustmodel would occur, and that new paradigm, in turn, wouldeventually be rejected in favor another, more robust para-digm. In plant ecology, the major paradigms of the centurythemselves underwent a succession similar to initial floristics(Egler 1954; Agee 1993), where many of the species (theo-ries) represented in the successional sequence are present inearly succession but display differential dominance overtime. The major plant ecology theories were all proposedwithin a decade early in the 20th century, but exhibiteddifferential dominance over time.

The classical view of plant succession (the theory thatattained initial dominance) persisted much of the 20th cen-tury: the Clementsian view of regional convergence towardsa vegetation life-form created by autogenic succession in thepresence of stable climate (Christensen 1988, 1991). Al-though competing models were proposed early (Gleason 1917,Tansley 1924), the Clementsian model was not seriouslychallenged until Odum (1969) proposed an ecosystem modelthat had a number of tautological premises. Among themwere assumptions that diversity and stability increased withecosystem development (time since disturbance). Odum’spaper generated a number of rebuttals (such as Drury andNisbet 1973) that suggested that ecosystems did not haveemergent properties, that various forms of diversity mightpeak in early succession and that stability might in somecases be maintained by disturbance. Rather than producinga more robust paradigm, these challenges to the existingorder recognized that ecology is a science of place and time.Grand unified theories are unlikely to apply (Christensen1988). Much of the new theory was developed by ecologistswho had worked in disturbance-prone ecosystems, and theyrecognized the multiple pathways that succession might takeafter disturbance, a function of both the disturbance and the“players” or organisms at the site. Disturbance, rather than abinary presence-absence variable, became a complex combi-nation of characteristics (White and Pickett 1985).

Wilderness fire scientists welcomed these challenges tothe classical theory. The incorporation of disturbance intonew theory provided a scientific niche for the presence offire in wilderness: Disturbance had a place in naturallandscapes (White 1979). It was now possible to moreclearly explain the previously baffling myriad of succes-sional trajectories after disturbance. But as the challengeswere comforting in one sense, they were discomforting inanother. To what the new theory added in recognizing fireas a natural factor, it removed in discarding the notion ofconvergence toward stable ecosystem states (Christensen1991). This created two managerial challenges: (1) Theissue of what to preserve became much more complex, asecosystem classification resulted in much less convergenceof community types; and (2) The stable end point towardwhich we should manage suddenly disappeared, leavingmanagers groping for a definition of a natural ecosystemstate or states. This latter point had crucial significance forwilderness fire.

This question took form in 1980s wilderness as a debatebetween structure and process as appropriate goals for parkand wilderness management. In a somewhat simple synop-sis, the process argument stated that every past landscapewas a snapshot of a variable ecosystem, and that ecosystemwould vary into the future. Reintroducing the process of firewould eventually restore an uncertain but natural future setof ecosystem states (Parsons and others 1986). This viewwas supported by some of the early interpreters of theWilderness Act (Worf 1985a,b). The structure argument(Bonnicksen and Stone 1982) stated that in any ecosystemswhere an unnatural structure had developed, reintroducingfire without attention to current structure could not result ina restored natural ecosystem. To some extent, the debatedepended on where one was (Agee and Huff 1986): after all,ecology is a science of place. But the question remained evenwhere scientists were viewing the same place. The argumentbecame most heated in the Sierra Nevada/Cascades low-severity fire regimes, where almost everyone agreed on thedegree of ecological change but differed on the need forstructural approaches to restoration (Bancroft and others1985; Bonnicksen 1985).

Added to the uncertainty of a desired future condition wasthe uncertainty of the disturbance regime. In the 1960s, therecognition of fire as a natural factor was sufficient toencourage management implementation. In the 1970s and1980s, more information began to emerge about fire re-gimes. White and Pickett (1985) defined a number of char-acteristics important for understanding the effects of distur-bance (such as frequency, magnitude, seasonality, extent,etc.), but for fire regimes, the primary one investigated wasfrequency, and primarily for low-severity fire regimes.Kilgore’s review of wilderness fire (1986) for the first confer-ence on wilderness focused primarily on frequency withinbroad fire regime types. More than 40 references to firefrequency were made by generalized fire regime types. Thefire regime types did carry implications for fire intensity, butlittle was known about extent, season or synergism withother disturbances. Variability and patchiness, now knownto be very important, were largely unquantified. Someinformation on variability in fire frequency was presented interms of ranges of fire frequency. Complex fire regimes in themoderate severity fire regimes had little information avail-able on patch size, proportions of different severity or otheraspects of the fire regime.

Standards for monitoring were largely lacking during thisperiod. Success was often gauged by area burned by pre-scribed fire and/or prescribed natural fire. Even thoughuncertainty about the operational goals of fire management(fuel reduction, ecological effects, etc.) persisted, there waslittle information that could be used to track progress to-wards any goal. Concerns about visual effects of prescribedfire in giant sequoia groves led to establishment of anindependent committee to review the fire program at Se-quoia and Kings Canyon National Parks (Christensen andothers 1987; Cotton and McBride 1987). The committeerecommended development of a detailed monitoring systemfor fires by the National Park Service.

Stand-level dynamic models incorporating disturbance be-gan to emerge in the 1970s, but they suffered from the absenceof established subroutines for stand growth, fire effects, orfire behavior. Most were derived from the JABOWA-type


gap models that grew stands on a small area (Botkin andothers 1972). The first model, FYRCYCL, was developed atYosemite (van Wagtendonk 1972) and was far ahead of itstime in using historical fire weather to drive the fire portionof the model. Another early model was SILVA (Kercher andAxelrod 1984), which was an improvement on FYRCYCL inthe stand growth routine but less elegant in its fire behaviorand fire weather. Fire effects on trees were estimated fromscorch height (a function of fireline intensity) and treediameter. However, many of the weather inputs were heldconstant, so a crude simulation at best of the fire regime waspossible. CLIMACS (Dale and Hemstrom 1984) was anotherfire model parameterized for the Pacific Northwest. Itsstand growth subroutines were robust but it treated distur-bance as an external effect that required the user to defineexactly which size classes and species were removed from aparticular disturbance. It was verified for only one foresttype in the region.

Two models linking fire behavior and fire effects weredeveloped during this period. Peterson and Ryan (1986)developed an algorithm that integrated stand-level charac-ters and fire behavior (including estimated flame residencetime) into a probability of mortality that was a function ofvolume of crown kill and the ability of a given bark thicknessto withstand lethal heat. The model requires estimation ofburning time in order to compare time of lethal heat tocritical time for cambial kill (based on bark thickness), andburning time was not commonly available to users. Ryanand Reinhardt (1988) used empirical data to develop asimilar mortality function based on crown scorch volumeand bark thickness.

One of the major developments useful in fire behavioranalysis was adaptation of the Rothermel spread model(1972) to a variety of stylized fuel models (Albini 1976),including those applicable to wilderness. A PC-version knownas BEHAVE was made available in 1984 (Burgan andRothermel 1984), with later improvements in several areas(Andrews 1986). This model allows prediction of surface firebehavior for given fuel, weather and topographic predic-tions. At high levels of input variables, fire behavior ex-pressed as fireline intensity or flame length can be inter-preted as leading to erratic fire behavior, but crown firemodels during this period were limited to empirical studiesin boreal forests (Van Wagner 1977).

Most of the growth in operational fire management plansin the 1980s was in parks and wilderness areas withmoderate- to high-severity fire regimes, suggesting thatthese plans contained sufficient research information oneffects and behavior of fire to indeed make these “pre-scribed” natural fire plans. In most cases, this informationwas very generalized. Boundaries of prescribed natural firezones were rather arbitrarily drawn inside the boundariesof the preserves, with little attention to the main directionof spread for intense fires or their historical or projectedeventual size. Historical size could be estimated from firehistory research, but technology to project fire behaviordays or weeks in advance was not available. In other areas,such as the chaparral of California, research in high-severity fire regimes did occur but focused on ecologicaleffects of fire (Baker and others 1982; Parsons 1976; Rundeland Parsons 1979, 1980) and much less on behavioralaspects. Limited research in the Pinnacles Wilderness

(Agee and others 1980) focused more on behavior thanecology.

Social science research was encouraged during this pe-riod, focusing on visitor perceptions and acceptance of wil-derness fire. Visitors who understood the role of fire inwilderness generally supported the policies (Cortner andothers 1984; Rauw 1980; Stankey 1976; Taylor and Daniel1984; Taylor and Mutch 1986). The economics of fire inwilderness remained clouded due to the blending of firemanagement activities outside and inside wilderness whichmade separation of costs difficult, and the different waysthat agencies accounted for prescribed natural fire versuswildfire in the pre-Yellowstone fires era. The Forest Serviceand some regions of the National Park Service requiredupfront budgeting for monitoring activities; when that bud-get was expended, the fire was reclassed as a wildfire (Agee1985, Daniels 1991). Another complication is the contrastbetween classical “least-cost-plus-loss” approaches, whichassumes all resource change is a loss, and evaluation ofresource change when fire could be viewed either as a cost orbenefit. Mills (1985) defined the major obstacle to appropri-ate economic analysis of fire in wilderness as understandingthe “natural state” objective of wilderness which would thenallow resource change to be viewed as cost or benefit.Ecologists, as noted above, had been little help in agreeingon a consensus definition useful for economic analysis.

The Wilderness Fire workshop held in Missoula in 1983(Brown and others 1985) defined the major issues apparentat that time. Over 100 papers and posters were presented atthe conference, and five major issues were addressed: (1) the“natural fire” issue—what is natural; (2) the “Indian fire”issue; (3) the “lightning (prescribed natural fire) versushuman (prescribed fire)” issue; (4) the “fire size and inten-sity” issue; and (5) the “unnatural fuel buildup” issue. Therewere no resolutions of these issues at that time, but consid-erable discussion of each. Clearly, the issue of “naturalness”was paramount in the first three topics. Are the origins oreffects of fire the basis for “natural?” Native Americansburned many of the landscapes of their day, often repeat-edly, and these effects had a large influence on vegetation asfar back as we can reconstruct it (Arno 1985; Gruell 1985;Kilgore 1985; Lewis 1985). How should this be incorporatedinto current fire planning for wilderness? The lightningversus human ignition issue is tied to the previous questionsand to the last question as well. Arguments about how closea prescribed fire can mimic a natural ignition (Despain1985), the need for caution in using prescribed fire inwilderness (Daniels and Mason 1985), the need to focus onfire effects (Van Wagner 1985) and the need to keep humanhands off wilderness (Worf 1985a) all surfaced in this discus-sion. The management-caused fuel buildup in some ecosys-tems was suggested to be reason enough for prescribed fireprograms to restore more natural conditions (Brown 1985;van Wagtendonk 1985).

Yellowstone: The Revolutionof 1988 ________________________

A revolution is defined as a drastic change of any kind, andthat describes the events of the summer of 1988. Yellowstone’sfires were at the center of the controversy because of their


visibility, but other fire events occurred that same yearunder similar circumstances.

Yellowstone’s Fire ProgramYellowstone’s prescribed natural fire program began in

1972 and was considered by the Park to be a successfulprogram before 1988. An average of 30 fires per year burnedbetween 1972 and 1987 (Despain and Romme 1991), andabout half were monitored. The monitoring of the firesduring this time indicated that fuels were a major determi-nant of where fires burned, with weather influencing thebehavior of the fires. Most fire starts and fire spread oc-curred in older lodgepole pine (Pinus contorta) stands, andfires appeared to naturally extinguish themselves at theedges of younger stands. The monitoring results might havebeen interpreted to mean that as more natural fires burned,the Park would be buffered from extreme events by the patchmosaic of fuels (Sweaney 1985). However, work by Romme(1982) had suggested that a very large event had occurred inthe early 1700s over at least part of the Park.

The summer of 1988 brought many fires and little precipi-tation compared to the 1972-1987 record, a very short periodof comparison for a high-severity fire regime of hundreds ofyears. It is not surprising that conditions of the extremeevent were not forecast, and two-thirds of the 1972-1987period July and August precipitation was well above long-term averages (Despain and Romme 1991). When the firesof 1988 began to spread, they were pushed by a series of coldfronts, which resulted in substantial increases in fire area inshort periods of time, capped by the runs of early Septemberthat resulted in fire area growth of tens of thousands of haper day.

By the end of the summer, over 300,000 ha (750,000 ac) ofthe Park, and similar areas around it, had burned in aspectacular series of fire runs. Roughly half of the areaburned was from direct or indirect human causes (camper,firewood, power line), reviving the argument of whethernature cared who started the fire (Van Wagner 1985). Parkresearchers defended that area as “natural” by claiming thatnatural fire starts in each area occurred later in the sameyear and, under the extreme conditions of 1988, would haveresulted in similar spread patterns (Despain and Romme1991). Yet that argument remains a weak ex post factoattempt to justify the argument that we were witnessing a“natural” event of unparalleled magnitude in recent history.Certainly the scale had precedent (Pyne 1982), but humanactivities altered the pattern and extent of the fires of 1988(Christensen and others 1989).

Canyon CreekThe Canyon Creek fire burned in the Bob Marshall Wil-

derness. Ignited by lightning on June 25, 1988, it wasdesignated a prescribed natural fire and was allowed to burn(Daniels 1991). It stayed at less than 1 ha (2.5 ac) for 26 days,but in late July grew to 4,000 ha (10,000 ac) in three days,burning in a mosaic pattern so that about a third of theencompassed area actually burned. After 65 days of activemanagement, the fire escaped the wilderness boundary andgrew from about 25,000 ha (60,000+ ac) to almost 100,000 ha(250,000 ac) in 16 hours, at the same time the Yellowstone

fires were rapidly expanding. Full suppression action wasordered for the fire.

Prophecy FireThe Prophecy fire burned at Crater Lake National Park,

Oregon, in August 1988. It began in the eastern boundaryarea of the Park, but was within the approved natural firezone. Crater Lake had managed natural fires for a decade inthe moderate-severity red fir type, and these burns hadremained in prescription. The Prophecy fire was pushed bystrong westerly winds and moved out of the Park to coverabout 400 ha of Forest Service land to the east. These windsmay not have been unusual, but the absence of weatherstations in the area meant that this fire weather, and theassociated fire behavior, would not be predicted. The firecrowned through a sparsely vegetated climax lodgepole pinetype that was thought to rarely support such behavior (Agee1981, Gara and others 1985).

Sifting Through the AshesBy late summer of 1988, the political climate of an election

year, combined with the perceived multi-regional, multi-agency failure of the natural fire program, resulted in thesuspension of all such programs until completion of a reviewand implementation of any review recommendations. Localpolicy reviews of the Yellowstone situation (Christensen andothers 1989) and a major national fire policy review (Philpotand Leonard 1989) were completed before the end of theyear. The local review focused on ecological issues andproposed both research and management recommendationsfor Yellowstone. For research, the review recommended anecosystems approach, a landscape or geographic context forindividual projects and provision for long-term studies(Christensen and others 1989). For management, the localreview recommended that an ecological blueprint evolve ona wilderness-specific basis, to articulate clearly the range oflandscape configurations locally acceptable and to guide firemanagement planning. The national review (Philpot andLeonard 1989) suggested that the natural fire policy was ingeneral a sound policy, but that it had been implementedwithout sufficient prescription criteria. Most of the plansthat did not meet current policy were in national parks(Wakimoto 1989).

The Flame Flickers: Politics andPhilosophy After Yellowstone _____

The political landscape has been as important as thenatural landscape in directing wilderness fire science. Theevents of 1988 essentially shut out wilderness fire, and therecovery of management programs over the past decade hasbeen relatively slow. No one wanted to be the supervisor ofthe next Yellowstone event. Some wildernesses, such asYosemite and Sequoia-Kings Canyon, which pioneered bothprescribed fire and prescribed natural fire, had their pro-grams reinstated almost immediately, as they met thecriteria of the 1988 national fire policy review even before1988. Other suspended programs have never been rein-stated. The result was a significant and immediate decline


in numbers of fires and area burned (fig. 1; Parsons andLandres 1998). Although area contained with prescribednatural fire zones increased by seven percent between 1988-92, area burned by prescribed natural fires decreased by 94percent (Botti and Nichols 1995), largely due to conservativemanagement criteria, including funding. At the same time,prescribed fire activity doubled over its pre-1988 levels(Botti and Nichols 1995), but this is largely due to increasesfor one unit (Big Cypress National Preserve).

The conservative management criteria were all based oncontrol (flame length) or external issues (smoke, availabilityof regional forces). Not a single criterion was based onmeeting objectives for wilderness management. Given thatplanning context, major reductions in numbers of programsand fires allowed to burn are not at all surprising. Yet theoperational management plans were not to blame. Withoutan ecological blueprint for what was desired in wilderness,it was not only much easier but more defensible to defineconditions where fire was not wanted than to define condi-tions where it was.

The consolidation of research scientists in the Depart-ment of the Interior also affected wilderness fire science. Themanagement agencies (such as the NPS) lost their ability tofund research, because that function was now in the newlycreated National Biological Survey. The brief life of both theNational Biological Survey and its replacement, the Na-tional Biological Service, resulted in financial chaos forresearch scientists, and funding for fire research has contin-ued to be problematic in the Geological Survey, where thesescientists now reside.

The political developments and problems of wildernessfire management began to erode the “era” of wilderness fire(Pyne and others 1996). Pyne correctly foresaw the 1990s asa new era of urban intermix fires, and it was ushered in withthe horrific Oakland fire of 1991 (Ewell 1995). Pyne’s decla-ration was rooted in the belief that the philosophical ques-tions posed by the marriage of fire and wilderness had neverbeen resolved and that technical approaches could not re-solve them. Yet in the end, technical approaches must beemployed to foster operational fire management programs,even if the philosophical issues remain unresolved.

Science Since Yellowstone _______The science of wilderness fire has progressed remarkably

in the past decade, withstanding the political issues and alargely fragmented research approach. There have been fewlarge research programs directed specifically toward wilder-ness fire, partly because of the fragmented, multi-agencymanagement of wilderness and a lack of research focus thatis characteristic of many other large, national-scope projects(Long Term Ecological Research, International BiologicalProgram, NASA’s space program, etc.). The NPS GlobalChange program is one larger program that has producedsome substantial implications for wilderness fire. Yet manyof the technical developments have resulted from locallyfunded projects, or from research done for other purposes.

Drivers of Wilderness FireThat fuel, weather and topography drive the behavior of

an individual fire has long been known (Barrows 1951,Brown and Davis 1973). Yet the factors driving wildernessfire regimes continue to be debated: Are fuels or weathermore important? Our research of the past decade suggeststhat the answer not only differs by fire regime, but to someextent on the interaction of fuels and weather. Swetnam andBetancourt (1990) linked a set of regional cross-dated firehistories in ponderosa pine (Pinus ponderosa) forests to high(La Nina) and low (El Nino) phases of the Southern Oscilla-tion. During the El Nino phases, precipitation in the South-west is much higher and fire activity is much less. At thesame time, tropical and subtropical areas receive less pre-cipitation as those storms are moving further north. Largeareas burned in Borneo (Davis 1984) and Australia (Rawsonand others 1983) during a large El Nino event in the early1980s. This link between global climate and local variabilityin fire regime shows a trend that links wilderness to the restof the world.

In high-severity fire regimes, arguments about the rela-tive influence of fuels and weather continue (Weir andothers 1995, Wierzchowski and others 1995). In Canadianboreal and subalpine forests, prescribed fire has been usedoperationally under the assumption that decades of fireexclusion have changed these forest types, that youngerstands have not been created during that period and thatolder forests were more flammable. Bessie and Johnson(1995) concluded that weather was the primary drivingfactor in large fire behavior; and since large fires constitutealmost all the area burned, fuel conditions are relativelyunimportant. They generalized these conclusions to all for-est types, a conclusion rebutted by Agee (1997). He sug-gested that under extreme weather in low-severity fireregimes, fire size may well have increased, but that fireseverity may not have been markedly increased. Fuel condi-tions have been shown to affect fire behavior and extent inlow- (Wright 1996) and moderate-severity (van Wagtendonk1995) fire regimes (fig. 2).

In some high-severity fire regimes, fire return intervalsmay be so long that very unusual synergistic influences mayoccur and mask more simple correlations of fire with flam-mability-stand age or weather-climate patterns. In the Olym-pic Mountains, Henderson and others (1989) mapped a verylarge forest fire event (fig. 3) circa 1700 A.D. that had been

Figure 1—Trends of numbers of fires and area burned since theinception of prescribed natural fire programs in 1968. Note the pro-nounced drop after 1988 (Parsons and Landres 1998).


Figure 2—A. Reconstructed fires of 1775-1778 in mixed-conifer forests of eastern Washington (Wright 1996). Firesoccurring with 1-2 years of one another in this low-severity fire regime appear to be extinguished when they enter recentlyburned areas. B. Monitored fires 1974-1991 in Yosemite National Park show similar mosaics (van Wagtendonk 1995).These appear to be more stable patterns than in high-severity fire regimes where process overwhelms pattern under severeweather (Romme and Turner 1991).

TeanawayButte

Swauk Prairie


partially identified by Fonda and Bliss (1969). Fire cyclemodels based on climate (Agee and Flewelling 1983) wereunable to reproduce similar fire events, and it was thoughtthat very unusual patterns of lightning frequency or foehnwinds may have occurred in the past. Recently, a very largehistoric earthquake along the Washington coast was recon-structed from tree ring records of trees buried beneath sealevel by submergence of coastal lands at the time of thequake (Yamaguchi and others 1997). This date was consis-tent with records of a major tsunami that hit the east coastof Japan on January 26, 1700. At a time when soils aresaturated, this earthquake likely felled many stands of treesaround the peninsula, and this additional dead fuel mayhave driven the large fire activity that apparently occurred.Lightning frequency, drought or foehn winds, the usualcombinations of factors associated with large fires, may haveremained quite average during this period.

Where fire return intervals are quite long, these “surprises”may be a major factor in the disturbance dynamics. Not onlymay extreme events be driving the system, but they may beevents that we have not yet uncovered. Lertzman and others(1998) showed through simulating fire regime parameters thatsubstantial variability may result, even in the absence of anunderlying physical or ecological pattern. They recommendcaution in attributing causality of fire regime drivers that arenot motivated by independently generated hypotheses.

Fine-Tuning the Fire RegimeWhen early fire management programs began in wilder-

ness, general knowledge of the fire regime was consideredadequate. Research inside and out of wilderness has led to a

more precise understanding of the fire regime, but it is stillnot possible to generate many parameters of a fire regime bysimply knowing, for example, what forest type is beingconsidered. Where more precise information has been gener-ated, it usually shows variability in frequency, intensity orextent. Synergistic effects are known to be more importantthat previously considered, although our ability to predictthem is still poor. And the general implications for manage-ment have been clouded by the complexity of these emergingfire regimes. Faced with considerable à·nges in variability,which combination is appropriate for a certain place now?Research on fire regimes has allowed us to place bounds onuncertainty, but it has also generally driven us away fromrelying on simple statistics like the mean. Programs haveevolved from rather uniform burns to those incorporatingconsiderable variability (Bancroft and others 1985; Parsonsand Nichols 1986).

Fire frequency has always been a primary parameter of thefire regime. Kilgore’s wilderness fire review (1986) has over 40citations on fire frequency in selected wilderness ecosystems,and he recognized that more examples could be cited. Butinformation on other fire regime parameters was lacking.Since that time, we know even more about fire frequency inwilderness. These new data have allowed us to understandthe distribution of fire frequency, not just its central tendency.A remarkable achievement was the reconstruction of giantsequoia (Sequoidendron giganteum) fire regimes back overmillennia (Swetnam 1993). The mean fire-return intervalshifted significantly for this low-severity forest type overperiods of centuries, and inferences about fire intensity weremade from correlations of tree-ring growth with fire occur-rences and percentages of sample trees scarred from anindividual fire. Landscape juxtaposition of forest types wasfound to be important in determining fire frequency. In thenorth Cascades, where wet, west Cascades forest types aremixed with dry, east Cascades types due to a rainshadoweffect west of the Cascade crest, the wet types had fire-returnintervals well below those measured elsewhere in the Cas-cades for those types. The dry, eastside forest types had firereturn intervals well above those measured in the easternCascades (Agee and others 1990).

Fire intensity remains difficult to reconstruct from his-toric fire regimes. Reconstruction of growth on trees experi-encing fire, and defining age classes of trees likely to estab-lish in fire-generated gaps, have been used to infer historicintensities. In giant sequoia groves where the history ofprescribed fire includes some fairly hot burns, reconstruc-tion of tree-ring growth showed that fire generally increasedgrowth, but some variable response was evident (Mutch andSwetnam 1995). A delayed growth response was foundwhere very intense fires had occurred and scorched thefoliage of the sequoias. Sequoia regeneration was tied to fire-generated gaps where sunlight could penetrate to the forestfloor. These data were used infer past fire intensities. Forexample, a fire in 1297 A.D. was inferred to be relativelyintense due to the increase in tree growth on giant sequoias(fig. 4), suggesting a release from competition and substan-tial regeneration that occurred locally (Stephenson andothers 1991). A recent article suggests that high-intensityfire also was characteristic of ponderosa pine stands(Shinneman and Baker 1997). However, these stands in theBlack Hills are transitional to boreal forest; white spruce

Figure 3—The large fire of ca. 1700 in the Olympic Mountains(Henderson and others 1989) appears to have occurred after a largeearthquake in 1700. This quake, occurring in January, may have beenassociated with considerable treefall and copious dead fuel, needingonly an ignition source to become very large.


Synergism, or the interaction of fire with other distur-bances, was recognized by White and Pickett (1985) as animportant parameter of disturbance regimes. Very littlequantification of this effect was evident for fire regimesbefore the late 1980s. Interaction with insects has long beenrecognized as a major second-order fire effect (Fischer 1980),but defining the degree of interaction is difficult, as manyother factors are important (Amman and Ryan 1991). Afterthe Yellowstone fires of 1988, the major tree species in thearea (lodgepole pine, Douglas-fir [Pseudotsuga menziesii],Engelmann spruce [Picea engelmannii], and subalpine fir[Abies lasiocarpa]) were attacked by a variety of insects;between 28-65% of the trees living after the fire wereinfested and killed (Amman 1991). Most of the bark beetle-attacked trees had basal damage from the 1988 fires.

At Crater Lake National Park, Swezy and Agee (1991)found that low-intensity but long-duration fires, caused byforest floor buildup due to fire exclusion, killed many of thefine roots after late spring burns. Low vigor, old-growthpine trees had an increased level of insect attack andmortality after these fires, and fall burning was recom-mended as a better season, based on surveys of trees burnedin spring and fall.

Disease can also be an important synergistic factor. Inthe western United States, perhaps the most importantsynergism between fire and disease is the introducedwhite pine blister rust (Kendall and Arno 1990). Thisdisease causes cankers on the stems of young pines andkills them. When fire kills older trees, recolonization ofwhitebark pine (Pinus albicaulis), often mediated by Clark’snutcrackers (Nucifraga columbiana) (Tomback 1982) maybe short circuited. In mountainous terrain, snow ava-lanches can create persistent snow avalanche paths andalter other processes such as landsliding and future firespread (Butler and others 1991).

Models ________________________The past decade has witnessed an explosion in personal

computing power and with that growth, an accompanyingexpansion of models attempting to explain fire behavior andeffects. These models have particular relevance to wilder-ness fire because they allow forecast of spatially explicit firesizes, as well as fire effects.

One of the more important models for fire effects has beenthe individual tree model FOFEM (First Order Fire EffectsModel; Reinhardt and others 1997). It scales mortality to thestand level by aggregating individual tree effects to thestand level based on the Ryan and Reinhardt (1988) mortal-ity algorithm. This model has gone through four iterations inthe past decade and will continue to be updated periodically.It is national in scope and provides information in additionto tree mortality on fuel consumption, mineral soil exposureand smoke. Synergistic effects, which tend to be difficult topredict as second-order interactions, are not predicted byFOFEM. Nevertheless, it has served as the basis for treemortality prediction in several important models.

A variety of individual-based gap models have been devel-oped since the 1970s (Hinckley and others 1996; Urban andothers 1991), but few have concentrated on incorporatingfire. FIRESUM (Keane and others 1989) was an improvedgap model that incorporated stand growth and disturbance

Figure 4—Patterns of tree-ring growth from giant sequoias (Stephensonand others 1991) show a pronounced growth effect after a recon-structed fire of 1297 A.D. Unlike many previous fires, this one must havebeen severe and reduced competition, as all trees show a growthrelease. The pattern of unusual growth continues for a century, andsome fires are associated with decreases in growth for sample trees.This suggests severe fires did occur in sequoia groves, and reminds usof the variability in fire regime for very long-lived organisms like giantsequoias.

(Picea glauca) is a common understory species, and a com-plex mix of fire regimes (e.g., Agee and others 1990) shouldbe expected where types are in transition. Tree regenerationis closely linked to fire severity; in moderate-severity fireregimes, severity will have significant effects on tree specieslikely to establish (Chappell and Agee 1996).

Quantifying season of burning has been important be-cause of the opportunity to ignite prescribed fires over abroad seasonal range. What is most natural? Historicalseasonality has been evaluated primarily for low-severityfire regimes by defining the placement of the fire scar for aparticular year in the earlywood to latewood of the annualring. In Southwest ponderosa pine stands, most scars are inthe earlywood, defining spring as the most common seasonfor fires (Baisan and Swetnam 1990), although some areasexhibit more even distribution of fires across the growthseason (Grissino-Mayer and Swetnam 1995). In the PacificNorthwest, the same species exhibits mostly late-seasonfires (Wright 1996). Heyerdahl (1997) showed that there wasconsiderable seasonal variation in the Blue Mountains ofOregon and Washington. Southerly Blue Mountain standshad a longer snow-free season and more scars within thegrowing portion of the annual ring than stands of the samespecies composition in the northern Blue Mountains, whichhad a shorter growing season and a concentration of scarsafter growth for the year had ceased.


for inland Northwest conifers. The fire algorithms werecomplex, but the stand-level results were greatly influencedby the initializing stand condition; an individual tree dyingof old age, for example, had a large influence on the basalarea output over the simulation period.

While the science of gap modeling grew, the ability torepresent wilderness landscapes in geographically refer-enced form also increased. Geographic information systems(GIS) represented a way to evaluate often inaccessible land-scapes in digital form. The development of better softwarepackages and more powerful personal computers allowedrobust analyses to occur at relatively low expense. Fireapplications, such as analysis of historic fire incidence byvegetation type, fuel inventories, prescribed burn units,lightning strike incidence analysis and fire regime analysis,were done (van Wagtendonk 1991b). Links of these types ofanalyses to fire growth simulators were beginning (Bevinsand Andrews 1989). Development of accurate input layersfor the current generation of fire area growth models re-mains relatively poor (Keane and others 1998).

FIRE-BGC (Keane and others 1996a) was developed bymarrying some of the algorithms of FIRESUM with FOR-EST-BGC, a physiologically based model (Running andGower 1991) that has been scaled up to a landscape ap-proach. As applied to wilderness ecosystems in GlacierNational Park (Keane and others 1996a) and the BobMarshall Wilderness (Keane and others 1996b), the modellinks many across-scale interactions, but it has the univer-sal problem of marrying not only diverse spatial scales, butthose of time as well (Keane and others 1996a). Temporalinformation at scales from annual (stand growth equations)to hourly (fire growth equations) complicate current model-ing efforts.

Disturbance propagation across landscapes has beenmodeled in two general ways: percolation-type models anddeterministic models. The percolation models suffer fromthe fact that fire does not move across a landscape with equalprobabilities of spread in all directions. The deterministicmodels suffer from data deficiencies (Van Wagner 1987).Both have increased our knowledge of fire effects and behav-ior at broader scales.

The percolation models have increased our knowledgeabout the influence of landscape pattern on process (fire)(Turner 1989). Most of the percolation work has been inhigh-severity fire regimes, where the binary process of acell being occupied or not by disturbance fits the high-severity nature of the disturbance. Work in the 1980ssuggested that disturbance in heterogeneous landscapeswas dependent on the structure of the landscape, as well asdisturbance frequency and intensity (Turner and others1989). This evolved to a more complex view that distur-bance probability affecting percolation can change overtime, particularly where fire weather becomes extreme(Turner and Romme 1994). Under extreme conditions,process is relatively independent of pattern (Agee 1998;Romme and Despain 1989). Nonequilibrium systems willbe the result (Baker 1989, Turner and Romme 1994);scientific advances in landscape theory have resulted in atougher job for managers by increasing the envelope ofuncertainty. Percolation-type models have suggested thatlandscapes altered by past intervention in fire regimes, orthose subject to climate change in the past (for example,

Clark 1988) or the future, will take 0.5 to 2 rotations of thenew disturbance regime for the landscape to adjust to thatnew regime (Baker 1989, 1994).

In contrast to the ecological gap and disturbance models,fire behavior models received less attention over the sameperiod, yet our inability to predict fire spread and intensityhas had much more effect on wilderness fire programs thanimprecision in predicting ecological effects. Fortunately,substantial progress has been made in landscape modelingof fire behavior. A nonspatial model (RERAP) was developedto determine probabilities that a prescribed natural firewould exceed an acceptable size (predetermined by the user)before a fire ending event (precipitation) would halt spread(Carlton and Wittala, no date). However, it has not beenwidely used in wilderness fire management. A fire growthsimulator (Bevins and Andrews 1989) was developed by theForest Service, and a similar model was being developed bythe National Park Service (Finney 1995). These effortsmerged in the mid-1990s at the Missoula Fire SciencesLaboratory.

The model currently holding most promise for wildernessfire behavior is FARSITE, a spatially and temporally ex-plicit fire growth model (Finney 1998). The model wasinitially developed to help predict spread of wilderness fires,but it has shown great applicability to wildlands in general.The landscape “themes” or data layers require informationon elevation, aspect, slope, fuel model and canopy cover, withoptional themes for crown fire behavior: crown height, crownbase height and crown bulk density. Daily and hourlyweather streams are required over the simulation period.Surface fire, spotting and crown fire behavior are simulated,subject to the limitations of models that currently exist forthose types of fire behavior. Fires spread in the model usingHuygens’ principle, where the fire front is expanding basedon elliptical wavelets, the shape of which depends on the fuelmodel and local wind-slope vectors (fig. 5). Backing andflanking fire spread is estimated from the forward rate ofspread, as the current fire spread model (Rothermel 1972)only predicts the forward rate of spread. Finney (1998)discusses the limitations of FARSITE.

Figure 5—The fire growth algorithm of FARSITE uses a series ofellipses (Finney 1998). A. Under constant weather and fuels, these“wavelets” are of constant shape and size. B. Non-uniform conditionsshow the dependency of wavelet size on the local fuel type but waveletshape and orientation on the local wind-slope vector.


Given accurate input data, the model is consistent withexpectations for fire growth of surface fires. Spotting andcrown fire spread are not possible to verify, although simu-lations do produce patterns that resemble phenomena ob-served on real fires. Outputs for FARSITE are geographi-cally referenced, and flame length or fireline intensity percell can be exported to fire effects and stand growth modelsto simulate landscapes over time (for example, Keane andothers 1996 a,b). For wilderness applications, FARSITEcould be applied to generate behavior under worst-caseconditions to evaluate possible escape scenarios over asummer for a prescribed natural fire, and could be linked toecological effects. If adjacent fuelbreaks are proposed adja-cent to wilderness as a rationale for loosening prescriptionsfor fire within wilderness (Agee 1995), FARSITE can be usedto evaluate effectiveness of the fuelbreak (van Wagtendonk1996) and spatial effects on fire control efficiency (Finneyand others, in press).

Few wilderness areas have databases that allow applica-tion of FARSITE. Yosemite National Park was on-line earlydue to the presence of an advanced geographic informationsystem (J. van Wagtendonk, personal communication). WhereFARSITE data layers (elevation, aspect, slope, fuel model,canopy cover, height to crown base, crown bulk density andcanopy height) have been generated, accuracy levels aresometimes so low (Keane and others 1998) that applicationof the FARSITE model is bound to produce uncertain re-sults, even if weather variables were perfectly predicted.

One of the major lessons learned in the 1988 fires wasthat the Rothermel fire spread model was not particularlyrobust in predicting the behavior of fires that contained alarge degree of crown fire activity (Thomas 1989). Most ofthe quantification of conditions where crown fire occurredwas derived from boreal forests of Canada (Van Wagner1977). Crown fire assessments were possible (Alexander1988) but not routinely employed by wilderness fire man-agers. After the 1988 fire season, it was apparent thatbetter understanding of crown fire behavior was needed.Rothermel (1991) evaluated crown fire potential in north-ern Rocky Mountain forests, and his derivation of crownfire spread was empirically derived as 3.34 times thesurface fire rate of spread of NFFL fuel model 10. Links offorest structure (Agee 1996) and weather conditions (Scottand Reinhardt, in press), using the Van Wagner and/orRothermel approaches, have been made and are i

united states department wilderness science in a …wilderness science in a time of change...

Documents