United StatesDepartment ofAgriculture

Forest Service

Pacific NorthwestResearch Station

General TechnicalReportPNW-GTR-596January 2004

Proceedings: Views From theRidge—Considerations forPlanning at the Landscape Scale

Compilers

Hermann Gucinski (retired) was a physical scientist, Forestry Sciences Laboratory, 3200 SW JeffersonWay, Corvallis, OR 97331; Cynthia Miner is a general biologist and Becky Bittner is a forester, PacificNorthwest Research Station, 333 SW First Avenue, Portland, OR 97208-3890.

Papers were presented at a conference by the authors, who are therefore responsible for the content andaccuracy. Opinions expressed may not necessarily reflect the position of the U.S. Department ofAgriculture.

The use of trade, firm, or corporation names in this publication is for the information and convenience ofthe reader. Such use does not constitute an official endorsement or approval by the U.S. Department ofAgriculture of any product or service to the exclusion of others that may be suitable.

Photo credits

Top photo taken by Roger Clark, bottom left to right Kevin Burke, Roger Clark

Proceedings: Views From the Ridge—Considerations for Planning at theLandscape Scale

Hermann Gucinski, Cynthia Miner, and Becky Bittner

Editors

Views From the Ridge—Considerations for Planning at the Landscape Scale,Vancouver, WA, November 2-4, 1999

Sponsored by:Pacific Northwest Research Station, USDA Forest ServiceWestern Forestry and Conservation Association

U.S. Department of AgricultureForest ServicePacific Northwest Research StationPortland, OregonGeneral Technical Report PNW-GTR-596January 2004

Abstract

Gucinski, Hermann; Miner, Cynthia; Bittner, Becky, eds. 2004. Proceedings: views from the ridge—considerations for planning at the landscape scale. Gen. Tech. Rep. PNW-GTR-596. Portland, OR:U.S. Department of Agriculture, Forest Service, Pacific Northwest Research Station. 133 p.

When resource managers, researchers, and policymakers approach landscape management, they bringperspectives that reflect their disciplines, the decisions they make, and their objectives. In working at alandscape level, they need to begin developing some common scales of perspective across the variety offorest ownerships and usages. This proceedings is a compilation of 22 papers presented at a conferencethat addressed divergent views on landscape management. The conference was a forum for exchangingconcepts and knowledge from research and management experiences about managing landscapes. Theprogram addressed the issues of managing landscapes when everyone has a different perspective;approaching landscape management from aquatic, terrestrial, and socioeconomic viewpoints; andcharacterizing landscape management.

Keywords: Landscape management, forest policy, forest management, aquatic, terrestrial,socioeconomic.

Contents

1 IntroductionHermann Gucinski

Conference Overview

5 From Red Queens to Mad Hatters—A Wonderland of Natural Resource ManagementLance Gunderson

11 What Is a Landscape and How Is One Studied?Virginia H. Dale and Barry R. Noon

How Do We Determine Scale?

17 Landscapes: The Dilemma of Scale and the Role of TheoryLee Benda and Daniel Miller

22 Scale Considerations for Linking Hillslopes to Aquatic HabitatsRobert R. Ziemer

33 Landscape Assessment in a Multiownership ProvinceThomas A. Spies

38 Implications of Scale on Viability Assessments of Terrestrial Wildlife SpeciesMartin G. Raphael

42 Projecting Potential Landscape Dynamics: Issues and ChallengesRonald P. Neilson

48 Scaling and Landscape Dynamics in Northern EcosystemsMarilyn Walker

57 Landscape PerspectivesRoger N. Clark, Linda E. Kruger, Stephen F. McCool, and George H. Stankey

61 Landscape-Level Management, It’s All About ContextBruce Shindler

62 Human Adaptation to Environmental Change: Implications at Multiple ScalesEllen M. Donoghue, Terry L. Raettig, and Harriet H. Christensen

67 Land Use and Land Cover Dynamics: Changes in Landscapes Across Space and TimeRalph J. Alig

Worldviews Panel

75 Private Landowner Perspective on Landscape ManagementCarl F. Ehlen

78 Back to the Future: The Value of History in Understanding and Managing DynamicLandscapesPenelope Morgan

Case Studies

87 Northwest Forest PlanNancy M. Diaz

93 Recent Changes (1930s-1990s) in Spatial Patterns of Interior Northwest Forests, USAP.F. Hessburg, B.G. Smith, R.B. Salter, R.D. Ottmar, and E. Alvarado

Landscape Methodology

95 Analysis and Communication of Monitoring Data With Knowledge-Based SystemsKeith M. Reynolds and Gordon H. Reeves

105 Seeing the Forest and the Trees: Visualizing Stand and Landscape ConditionsRobert J. McGaughey

110 A Method to Simulate the Volume and Quality of Wood Produced Under an EcologicallySustainable Landscape Management PlanGlenn Christensen, R. James Barbour, and Stuart Johnston

Transforming Information From Many Perspectives to Action on the Ground

119 A Social Science Perspective on the Importance of ScaleLinda E. Kruger

Conclusion

127 Summary Remarks on Views From the RidgeT.F.H. Allen

130 The Future of Landscape-Level Research at the Pacific Northwest Research StationThomas J. Mills

1

Fred Swanson, a scientist who has devoted muchof his career to the study of the processes thataffect the character of a landscape, was askedwhile leading a tour of the H.J. Andrews Experi-mental Forest in Oregon with the group poisedatop a commanding ridge, “Well, just what is agood landscape?”

As a scientist committed to expanding knowledgerather than to just citing known facts, he took thisquestion as a serious challenge, admitted to find-ing it perplexing, and used it in turn to engenderdeep discussion, and some introspection, aroundthe campfire that evening.

Before we can use the same question as a frame-work for the contributions in this volume, it mightbe good to ask: “What is a landscape?” and, be-cause this is a scientific endeavor, “What is land-scape ecology, and how might it contribute towardthe larger question?” For it was indeed the objec-tive of the symposium, and the proceedings thatresulted, to take the “View From the Ridge” as ascientific challenge, rather than a subjective one,and to try to learn if a scientifically framed answercan also help with the subjective part of theneeded answer.

Webster’s unabridged dictionary and the OxfordEnglish Dictionary have similar definitions: “a por-tion of land or territory, which the eye can compre-hend in a single view, including all the objects soseen” for the former, and “a view or prospect ofnatural in-land scenery, such as can be taken inat a glance from one point of view; a piece ofcountry scenery” for the latter. Both offer someadditional definitions: “a tract of land with its dis-tinguishing characteristics and features, esp.considered as a product of modifying or shapingprocesses and agents (usually natural)”—“a view,prospect of something”—“a distant prospect: avista,”—“a bird’s-eye view; a plan, sketch, map.”Thus, our title, “Views From the Ridge” is appro-priate, although one contributor, Bob Ziemer,astutely notes that a “view from space” would

provide a yet larger prospect (“Vogelschau”[bird’s-eye view] may be the most apropos Ger-man word for this), whereas “a view from the val-ley” would help discern the delivery of results fromupland processes when taking an aquatic or ripar-ian perspective.

These processes, the delivery of mass and en-ergy, among others, bring us to the scientific partof the inquiry—the appropriate subject of “land-scape ecology.” Richard T.T. Forman and MichaelGodron (1986) used this definition in their seminalwork:

Landscape ecology is the study of struc-ture, function and change in a heteroge-neous land area composed of interactingecosystems. It therefore focuses on:

• structure, the spatial patterns of land-scape elements and ecological objects(such as animals, biomass and mineralnutrients);

• function, the flows of objects betweenlandscape elements; and

• change, alterations in the mosaic throughtime.

Monica Turner (1989) defines it thus:

Landscape ecology emphasizes the inter-action between spatial pattern and eco-logical process—that is, the causes andconsequences of spatial heterogeneityacross a range of scales. Two importantaspects of landscape ecology distinguish itfrom other subdisciplines within ecology.First, landscape ecology explicitly ad-dresses the importance of spatial configu-ration for ecological processes. Not only islandscape ecology concerned with howmuch there is of a particular componentbut also with how it is arranged. Second,landscape ecology often focuses uponspatial extents that are much larger thanthose traditionally studied in ecology.

1Physical scientist (retired) Forestry Sciences Laboratory,3200 SW Jefferson Way, Corvallis, OR 97331.

Introduction

Hermann Gucinski1

2

The following papers show the interest focused onthe importance of spatial configuration for an eco-logical process and will demonstrate the astonish-ing variety of both viewpoints and scales used toaddress problems. This shows the science to beevolving and is a sign of the value placed on prob-lems in this area. A brief look at the history of for-estry science shows its earliest endeavors, silvi-culture and entomology, to be largely driven bystand-scale processes. Problems were consid-ered solvable at that scale. Not only that, but thenature of the problem and suggested solutionswere seen as largely separable. You isolated eachvariable, and treatments of that variable werebased on the discipline within which it fell.Although this is oversimplified, earlier scientificendeavors engendered the separation of the dis-ciplines and slowed understanding of the interde-pendence of ecological processes. “If all you haveis a hammer, every problem will begin to re-semble a nail.” This approach worked for a longtime.

Many phenomena contributed to the expandingview of ecological processes. We recognized thatthe accumulation and translocation of long-livedpollutants required a comprehensive look at bio-logical, hydrological, and atmospheric processes.So did the need to develop concepts of “carryingcapacity” as the pressures of nonsustainable re-source use—hastened by population growth—brought conditions that could no longer bedescribed or understood at stand levels. Thisdevelopment challenged resource managers toabandon linear approaches to management—ecological processes are cyclical in nature, andevolutionary development has allowed highly inte-grated system functioning that produces nowaste. Superimposing solutions based on linearthinking caused new problems to pop up in placesoutside the scientific discipline where the originalproblem first appeared. Addressing the conse-quences resulted in yet other problems, leading toa vicious circle of “stimulus-response.”

That dynamic can be broken by moving up inscale from the domain of the initial problem, seek-ing to understand the connections while standingon the ridge—or, alternatively, in the valley. Thiscan help us regain perspective, and find pathwaysto solutions. Dale and Noon tackle this by relating

the view from the ridge to a quite specific ridge,namely Oak Ridge, site of the Oak Ridge NationalLaboratory and its rich history of contributions tolandscape ecology. Today’s efforts are enhancedby better tools and methods, as Spies notes inhis contribution, and they also require lookingbeyond political boundaries such as ownerships.McGaughey introduces us to valuable visualiza-tion tools now available to permit examination oflandscape “treatment” scenarios. Raphael arguesthat being able to hold onto all scales is a neces-sity when considering the problem of “viability” ofa species that is threatened. Clark et al. wrestlewith the subjective components of a landscape asthey relate to the values that people assign to it.And, as Shindler reminds us, right up to thepresent, our decisions have almost always beenmade at the stand level. Morgan expands ourvista to not only view a landscape in space, butbecome aware of it over time such as a knowl-edge of history permits. Here the concept of “his-torical range of variation” can be a vital tool.Benda and Miller take landscape processes underthe lens, focusing on the differences betweenstochastic and deterministic processes. Confus-ing these can have disastrous consequences.Neilson reminds us of the nature of “emergentproperties” when going to a larger scale, and usesclimate response as a useful integrator for theseprocesses.

The intent here is not to give away the principaltheses and conclusions of each contributor. In-stead, I want to whet the appetite of the reader,convey the breadth and depth that are containedin this volume, reflect on the variability of the ap-proaches outlined, and weigh their implications.Finally, I hope the information given and the chal-lenges raised will encourage the reader to rethinkthe question, “What is a good landscape?” in lightof available science as it illuminates the subjectiveneeds we have.

Literature Cited

Forman, R.T.T.; Godron, M. 1986. Landscapeecology. New York: John Wiley and Sons.

Turner, M.G. 1989. Landscape ecology: the effectof pattern on process. Annual Review of Ecol-ogy and Systematics. 20: 171-191.

3

4

Conference Overview

5

Lewis Carroll (1899) created a number of memo-rable characters who appear and reappearthroughout the adventures of Alice in Wonderland.A few of these characters highlight and resonatewith my sense and experience of the kinds ofissues faced by resource managers. Just as Aliceis bewildered by the complexities and uncertain-ties of Wonderland, managers often face similarissues of sensemaking and coping with the inevi-table surprises of complex systems of humansand nature. In the following sections, I use theserich characters from Wonderland as an entreeinto a set of topics that will hopefully stimulatesome ideas related to the paradigms that underlieour management of natural resources at the land-scape scale, beginning with the Red Queen.

Red Queens and Other Caricatures

“Well, in our country,” said Alice, still pant-ing a little, “you’d generally get to some-where else—if you ran very fast for a longtime, as we’ve been doing.”

“A slow sort of country!” said the Queen.“Now, here, you see, it takes all the run-ning you can do, to keep in the sameplace. If you want to get somewhere else,you must run at least twice as fast asthat!”……..Lewis Carroll

The Red Queen’s action depicted in this scene isoften used as a metaphor by evolutionary biolo-gists to describe whether species can adapt torapidly changing environments. I see similar “RedQueen dilemmas” as practitioners work as fastas they can to cope with rapidly and unpredictablychanging systems. This metaphor can be ex-tended to include a wider set of assumptions ormyths that help to explain how nature operates.Each of these myths has a different supposition ofhow stable systems are (in the sense of respond-ing to perturbations over time), the types of pro-

cesses that support that stability domain, conse-quences of those assumptions, and explanationsfor policy and action. At least five such caricaturesare described in the following paragraphs (modi-fied from Holling and Gunderson 2002).

Nature Flat

Although this perception is perhaps so simplisticas to be silly, it is included here as a starting pointfrom which subsequent contrasts with other viewscan be made. In this view, the word “flat” is usedto describe a system in which there is little or nostability of structure over time. The hands ofhumans, given sufficient resources can changenature, and there are no feedbacks or conse-quences of those actions—it is much like rolling aball around on a cookie sheet. The processes thataffect the state of nature are random or stochas-tic. In such a view of nature, policies and politicsare random as well, often described as garbagecan dynamics (March and Olsen 1989). It is anature that is infinitely malleable and amenable tohuman domination. As such, the issues of re-source use, development, or control are identifiedas issues of people and resolved by activism orcommunity organization.

Nature Balanced

The second is a view of nature at or near an equi-librium condition, which can be static or dynamic.Hence if nature is disturbed, it will return to anequilibrium through (in systems terms) negativefeedback. As such, nature seems to be infinitelyforgiving. This is the view of nature that underpinslogistic growth where the issue is how to navigatea looming and turbulent transition—demographic,economic, social, and environmental—to a sus-tained plateau. This is the view of several institu-tions with a mandate for reforming global re-source and environmental policy: the BrundtlandCommission (Brundtland and Khalid 1987), theWorld Resources Institute, the International Insti-tute of Applied Systems Analysis (Clark and Munn1986), and the United Nations (Munasinghe andMcNeely 1995) for example, who are contributing

From Red Queens to Mad Hatters—A Wonderland of NaturalResource ManagementLance Gunderson1

1Associate professor and Chair, Department of EnvironmentalStudies, Emory University, Atlanta, GA 30322; Tel: 404-727-2429; e-mail: lgunder@emory.edu

6

skillful scholarship and policy innovation. They areamong some of the most effective forces forchange.

Nature Anarchic

If the previous myth is one where the system sta-bility is defined as a ball at the bottom of a cup,this myth is one of a ball at the top of a hill. It isglobally unstable. It is a view dominated by hyper-bolic processes of growth and collapse, or whereincrease is inevitably followed by decrease. It is aview of fundamental instability where persistenceis only possible in a decentralized system withminimal demands on nature. It is the view ofsome extreme environmentalists. If the previousview assumes that infinitely ingenious humans donot need to learn anything different, this view as-sumes that humans are incapable of learning.This is implicit in the writings of Tenner (1996),where he argues that all technology that is un-leashed will eventually “bite back.” This view pre-sumes that small is beautiful because of theinevitable catastrophe of any policy. It is a viewwhere the precautionary principle dominatespolicy, and social activity is focussed on mainte-nance of the status quo.

Nature Resilient

The fourth view is of nested cycles organized byfundamentally discontinuous events and pro-cesses. That is, there are periods of exponentialchange, periods of growing stasis and brittleness,periods of readjustment or collapse, and periodsof reorganization for renewal. Instabilities organizethe behaviors as much as do stabilities. This hasrecently been the focus of fruitful scholarship in awide range of fields—ecological, social, eco-nomic, and technical. These dynamics have simi-larities in Harvey Brook’s view of technology(1986), Brian Arthur’s and Kenneth Arrow’s recentviews of the economics of innovation and compe-tition (Waldrop 1992), Mary Douglas’ (1978) andMike Thompson’s (1983) views of cultures, DonMichael’s view of human psychology (1984), andBarbara Tuchman’s (1978) and William McNeill’s(1979) views of history. The “nature resilient” viewis a view of multiple stable states in ecosystems,

and management approaches that are adaptive.But both of these presume a stationary land-scape. In this case, our cookie sheet has beenmolded and curved in three dimensions, but it isfixed over time.

Nature Evolving

The emerging fifth view is evolutionary and adap-tive. It has been given recent impetus by the para-doxes that have emerged in successfully applyingthe previous more limited views. Complex sys-tems behavior, discontinuous change, chaos andorder, self-organization, nonlinear system behav-ior, and adaptive evolving systems are all thepresent code words characterizing the more re-cent activities. Such thinking is leading to integra-tive studies that combine insights and people fromdevelopmental biology and genetics, evolutionarybiology, physics, economics, ecology, and com-puter science. It is a view of an actively shiftinglandscape with self-organization (where the stabil-ity of the landscape affects behavior of the vari-ables, and the variables, plus exogenous events,affect the stability of the landscape). It is a view ofcross-scale interactions of processes–dubbedpanarchy in previous writings (Gunderson et. al.1995, Holling and Gunderson 2002). It is a viewthat requires a focus on active policy probes of ashifting domain and a focus on institutional andpolitical flexibility for learning.

Comparing and contrasting these underlying cari-catures or worldviews, provides some insight intohow we create our wonderlands in order to makeprescriptions for action. It is also useful to notehow these partial myths are adopted and rein-forced and prescribed by a variety of disciplines,as described in the next section, organizedaround Mad Hatters of various disciplines.

Disciplinarily Mad Hatters

Alice had been looking over his shoulderwith some curiosity. “What a funny watch!”she remarked. “It tells the day of themonth, and doesn’t tell what o’clock it is!”

“Why should it?” muttered the Hatter.“Does YOUR watch tell you what year itis?”

7

“Of course not,” Alice replied very readily,“but that’s because it stays the same yearfor such a long time together.”

“Which is just the case with MINE,” saidthe Hatter.

Alice felt dreadfully puzzled. The Hatter’sremark seemed to have no sort of mean-ing in it, and yet it was certainly English. ”Idon’t quite understand you,” she said, aspolitely as she could.……..L. Carroll

The Mad Hatter utters seemingly nonsensicalquestions and answers in his dialogues with Aliceat the tea party. He reminds me of those of uswho are technically or scholarly oriented, the edu-cated skeptics who pay attention to different indi-cators, have different frames of reference anddifferent interpretations and understandings. Aswith the Mad Hatter, in attempting to communi-cate our technical understanding to policymakersor decisionmakers, more often than not, we arepolitely misunderstood. And all of us are like theHatter in that we wear our disciplinary hatsproudly—whether as ecologist, biologist, politicalscientist, economist, or any other scholarly cat-egorization. Yet there is a growing sense thatthese discipline-based organizations of inquiryand understanding are problematic.

Management of global and regional resources isnot an ecological problem, nor an economic one,nor a social one. It is a combination of all three.And yet actions to integrate all three inevitablyshortchange one or more. Sustainable designsdriven by conservation interests ignore the needsfor an adaptive form of economic developmentthat emphasizes enterprise and flexibility. Thosedriven by economic and industrial interests act asif the uncertainty of nature can be replaced withhuman engineering and management controls, orignored altogether. Those driven by social inter-ests act as if community development and em-powerment of individuals hold the key and thereare no limits to the imagination and initiative oflocal groups. As investments fail, the policies ofgovernment, private foundations, internationalagencies, and nongovernmental organizations flipfrom emphasizing one kind of myopic solution toanother. Over the last three decades, such poli-cies have flipped from large investment schemes,to narrow conservation ones to, at present,

equally narrow community development ones.

Each one builds their efforts on theory, althoughmany would deny anything but the most pragmaticand nontheoretical foundations. The conservation-ists depend on theories of ecology and evolution,the developers on variants of free market models,the community activists on theories of communityand social organization. All these theories arecorrect in the sense of being partially tested andcredible representations of one part of reality. Theproblem is that they are partial. That partiality ofconcepts leads to the search for theories ofchange that bridge disciplines in rich ways.

One such integrated theory is Holling’s (1992)adaptive cycle. The heuristic combines ecologicaltheories of succession and ecosystem develop-ment with other concepts of stability and resil-ience. It is a rich framework to indicate hownatural capital and connectivity of systems in-crease slowly over time. But those properties leadto an increasing vulnerability and inevitable peri-ods of destruction and reorganization. Authorshave used this framework to explain co-evolutionof resources and management through time(Light et al. 1995), business and organizationaldynamics (Westley 1995), and political systems(Holling and Sanderson 1996).

The above excerpt from the tea party, in whichAlice and the Mad Hatter discuss differences intheir watches, also suggests to me that one of thekey challenges that resource managers face isovercoming obstacles of scale. Those obstaclesare both theoretical and practical. How managersattempt to analyze and learn from their actionsare both related to issues of scale. Most modelsand modes of inquiry are scale bound and depen-dent. Walters (1997) cites the cross-scale prob-lem as a severe obstacle in most assessment/modeling activities. Development of new theoriesis needed to help address ecosystem and naturalresource dynamics across space and time scales.Over the last 40 years, time and space have beenseparated for analytical purposes. Most field eco-logic investigations either freeze space and ex-periment over time or freeze time and look atspatial patterns (witness the explosion and ubiq-uity of geographic information system technologyin resource management agencies). Perhapsthere are practical reasons for this pattern, but italso can be explained in part because of underly-ing theoretical frameworks. There is a growing

8

sense that this separable-dimension frameworkresults in assessments with different outputs, sug-gesting the need for integration and reconciliation.

Through The Looking Glass and WhatAlice Found There

For some minutes Alice stood withoutspeaking, looking out in all directions overthe country—and a most curious country itwas. There were a number of little brooksrunning across from side to side, and theground between was divided up intosquares by a number of hedges, thatreached from brook to brook.

“I declare it’s marked out just like a largechessboard!” Alice said at last. “Thereought to be some men moving aboutsomewhere—and so there are!” sheadded in a tone of delight, and her heartbegan to beat quick with excitement asshe went on. “It’s a great game of chessthat’s being played all over the world—ifthis is the world at all, you know. Oh, whatfun it is!……..L. Carroll

One of the enduring and endearing aspects of theAlice in Wonderland stories is that she faces anuncertain world (from the nightmarish to the sub-lime) with hope and wonder. For some who dealwith issues of natural resource management,there is only doubt and gloom. However, I tend toagree with Alice, and I suggest that there is rea-son to hope. But that hope is founded on develop-ing and creating new ways to think about andmanage issues of the environment.

Perhaps it is time to rethink the paradigms orfoundations of resource management institutions,and place more emphasis on development ofsustaining foundations for dealing with complexresource issues. Learning is a long-term proposi-tion that requires a ballast against short-term poli-tics and objectives. Another shift likely will requirea change in the focus of actions away from man-agement by objectives and determination ofoptimum policies toward new ways to define, un-derstand, and manage these systems in an ever-changing world. That focus should not be solelyon variables of the moment (water levels, popula-tion numbers) and their correlative rates, but

rather on more enduring system properties suchas resilience, adaptive capacity, and renewal ca-pability. Indirectly, these system properties havebeen explored in large complex ecosystemssuch as the Grand Canyon (Walters 1997), theColumbia River basin (Volkmann and McConnaha1993), and the Everglades (Gunderson 1999). Aresilience or adaptive capacity framework involvesboth the human components of the system (op-erations, rules, policies, and laws) and the bio-physical components of the landscape and itsecosystems. The shift of focus to a learning basisis likely to require flexible linkages with a broaderset of actors, or network. Another way of sayingthis bluntly is, until management institutions areable and willing to embrace uncertainty and sys-tematically learn from their actions and respond tothat learning, adaptive management will not con-tinue in its original context, but rather be redefinedin a weak context of “flexibility in decisionmaking”(Gunderson 1999).

But what does it take to be hopeful in a world thatis perhaps becoming much more unforgiving? Asthe degree of human impact continues to increasein scale, a key unanswered question is whetherthe adaptive capacity of both ecologic and socialsystems can keep pace with this expanding hu-man footprint. Under those conditions, the pre-scription for facilitating constructive changeappears to be:

• Identify and reduce destructive constraints andinhibitions on ecological change, such as per-verse subsidies (e.g., sugar farming in theEverglades).

• Protect and preserve the accumulated experi-ence on which change will be based (such asmanagers in land management agencies withmultiple decades of experience).

• Stimulate innovation in a variety of fail-safeexperiments that probe possible directions inways that are low in costs for people’s careersand organizations’ budgets (such as adaptivepolicies in the Grand Canyon).

• Encourage new foundations for renewal thatbuild and sustain the capacity of people,economies, and nature for dealing withchange.

9

These suggestions are founded on the premisethat we must learn our way into an uncertain fu-ture. That learning should help guide deliberationstoward workable and sustainable futures. Thosedeliberations will not solve all social or distribu-tional issues, but rather might help frame ways towork through this wonderland of resource man-agement—we don’t have the luxury of awakeningand realizing that it may have been a dream.

Acknowledgments

This is a contribution from the Resilience Net-work, funded by a grant from the John D. andCatherine T. MacArthur Foundation. Buzz Hollingcontinues to inspire creative and novel ap-proaches to resource management.

Literature Cited

Brooks, H. 1986. The typology of surprises intechnology, institutions and development. In:Clark, W.C.; Munn, R.E., eds. Sustainabledevelopment of the biosphere. Cambridge,United Kingdom: Cambridge University Press:325-348.

Brundtland, G.H.; Khalid, M., eds. 1987. Ourcommon future. Oxford, United Kingdom: Ox-ford University Press.

Carroll, L. 1899. Through the looking glass andwhat Alice found there. New York; London:Macmillan and Co., Ltd.

Clark, W.C.; Munn, R.E., eds. 1986. Sustainabledevelopment of the biosphere. Cambridge,United Kingdom: Cambridge University Press.

Douglas, M. 1978. Cultural bias. Royal Anthropo-logical Institute. [Publisher and location un-known.]

Gunderson, L. 1999. Resilience, flexibility andadaptive management—antidotes for spuriouscertitude? Conservation Ecology. 3(1): 7.http://www.consecol.org/vol3/iss1/art7.(10 November 2003).

Gunderson, L.; Holling, C.S.; Light, S.S., eds.1995. Barriers and bridges for the renewal ofecosystems and institutions. New York: Colum-bia University Press.

Holling, C.S. 1992. Cross-scale morphology,geometry and dynamics of ecosystems. Eco-logical Monographs. 62(4): 447-502.

Holling, C.S.; Gunderson, L.H. 2002. Panarchy,understanding transformations in systems ofhumans and nature. Washington, DC: IslandPress.

Holling, C.S.; Sanderson, S. 1996. Dynamics of(dis)harmony in ecological and social systems.In: Hanna, S., ed. Rights to nature. Washing-ton DC: Island Press: 57-85.

Light, S.S.; Gunderson, L.H.; Holling, C.S.1995. The Everglades: evolution of manage-ment in a turbulent ecosystem. In: Light, S.S.;Gunderson, L.H; Holling, C.S. eds. Barriersand bridges to the renewal of ecosystems andinstitutions. New York: Columbia UniversityPress: 103-168.

March, J.G.; Olsen, J.P. 1989. Rediscoveringinstitutions. New York: Free Press.

McNeill, W.H. 1979. The human condition: anecological and historical view. Princeton, NJ:Princeton University Press.

Michael, D.N. 1984. Reason’s shadow: notes onthe psychodynamics of obstruction. TechnicalForecasting and Social Change. 26(2): 149-153.

Munasinghe, M.; McNeely, J. 1995. Key con-cepts and terminology of sustainable develop-ment. In: Munasinghe, M; Shearer, W., eds.Defining and measuring sustainability: thebiogeophysical foundations. Washington DC:World Bank: 19-56.

Tenner, E. 1996. Why things bite back: technol-ogy and the revenge of unintended conse-quences. New York: Knopf.

10

Thompson, M. 1983. A cultural bias for compari-son. In: Kunreuther, H.C.; Linnerooth, J. Riskanalysis and decison processes: the siting ofliquid energy facilities in four countries. Berlin:Springer: 232-262.

Tuchman, B.W. 1978. A distant mirror. New York:Ballantine.

Volkman, J.M.; McConnaha, W.E. 1993.Through a glass, darkly: Columbia Riversalmon, the Endangered Species Act, andadapative management. Environmental Law.23: 1249-1272.

Waldrop, M. 1992. Complexity: the emergingscience at the edge of order and chaos. NewYork: Simon and Schuster.

Walters, C. 1997. Challenges in adaptive man-agement of riparian and coastal ecosystems.Conservation Ecology [online]. 1(2): 1. http://www.consecol.org/vol1/iss2/art1.(10 November 2003).

Westley, F. 1995. Governing design. In: Light,S.S.; Gunderson, L.H.; Holling, C.S., eds.Barriers and bridges to the renewal of ecosys-tems and institutions. New York: ColumbiaUniversity Press: 391-427.

11

Introduction

Because this workshop is entitled “Views Fromthe Ridge,” I interpret my task in reviewing contri-butions to resource management from landscapeecology to be to provide you with the view fromOak Ridge. The Environmental Sciences Divisionat Oak Ridge National Laboratory (ORNL) wasformed in 1955 as a response to the concern ofhow radiation was affecting the environment. TheDivision’s objectives rapidly expanded to includeanalysis of environmental effects of all aspects ofenergy use. This topic clearly encompasses allaspects of environmental sciences, and now ourDivision houses 96 scientists with degrees from18 fields of study.

The history of advancements in environmentalsciences at ORNL parallels developments in landmanagement and ultimately leads to a landscapeperspective. Therefore, I thought it would be use-ful to highlight those scientific achievements as aprecursor to discussing landscapes. Over theyears, ORNL scientists have participated in thedevelopment of key research areas that formedthe basis for landscape ecology (Hagen 1992).Supported by the International Biological Pro-gram, the field of systems ecology was created.Systems ecology views ecological interactionsfrom a holistic perspective and seeks to quantifyinteractions of varied components. Global changeresearch was initiated shortly after views of theEarth from space led to a realization that the eco-logical system was global. But this view wastainted with clouds of pollution. Recognition thathuman impacts occur on a global as well as localscale led to remediation science as a way to usethe scientific process to learn about ways to rem-edy pollution problems. Yet this application ofecological science to real-world problems wasdifficult because of the intricacies of ecological

systems. To deal with this complexity, first hierar-chy theory (O’Neill et al. 1986) and then riskanalysis (Bartell et al. 1994, Suter 1992) arose.However, neither of these advances recognizedthe spatial relationships inherent to ecologicalinteractions. Landscape ecology has recentlycome to be a new field of study that explicitly fo-cuses on spatial interactions (Turner 1989, Turnerand Gardner 1991).

Need for a Landscape Perspective

From an ecological viewpoint, a landscape is aspatial extent over which ecological processestake place (King 1997). More simply, it refers to aspatially heterogenous area that has a similargeomorphology and disturbance regime (Turnerand Gardner 1991).

A landscape perspective is needed to addresstoday’s land management problems for severalreasons. It is now recognized that the spatialscale of environmental problems is large and thatall ecological processes (and management ac-tions) occur in a spatial context and are con-strained by spatial location. As an example,Fraser fir (Abies fraseri (Pursh) Poir.) are dying inthe southern Appalachians as a result of herbivoryand population dynamics of the introduced woollyadelgid (Adelges piceae Ratzeburg), but the in-sects’ distribution, and thus fir mortality, is influ-enced by the topographic conditions that restrictthe fir to the highest peaks (Dale et al. 1991).Furthermore, ecological systems can be viewedas spatially and temporally hierarchical. That is,processes observed at one level of organizationarise from lower level behaviors and are con-strained by higher level processes. Therefore,solutions for contemporary environmental prob-lems need to be provided within a spatial context.For example, natural areas that provide essentialecological services are limited in extent, and theircontributions must be interpreted within the land-scape matrix in which they reside and with theunderstanding that environmental conditions maychange over space as well as time (as with globalwarming). Thus spatially optimal solutions to landmanagement options should be considered.

What Is a Landscape and How Is One Studied?

Virginia H. Dale1 and Barry R. Noon2

1 Corporate fellow, Environmental Sciences Division, OakRidge National Laboratory, Oak Ridge, TN 37831-6036; Tel:865-576-8043; e-mail: dalevh@ornl.gov

2 Department of Fishery and Wildlife Biology, Colorado StateUniversity, Fort Collins, CO 80523-1474; Tel: 970-491-7905;e-mail: brnoon@cnr.colostate.edu

12

A Landscape Approach to LandManagement

The defining attributes of an ecological landscapeare structure, composition, function, and change(fig. 1). Structure deals with the physical relation-ships of landscape elements to each other: theirshape, patchiness, juxtaposition, etc. Compositionrefers to the variety of elements that make up alandscape (e.g., cover types, land forms, etc.).Function indicates the ecological processes thatoccur on the landscape (e.g., persistence ofpatches, rates of nutrients and energy flow, ero-sion, etc.). These attributes lead to the key focusof landscape ecology: estimating the reciprocalrelationships between landscape structure, func-tion, and composition and how they might changeover time.

Linking landscape ecology to application takesseveral steps. First, analyses must move beyonda description of the attributes of structure, func-tion, and composition to analyze interactionsamong these attributes. Second, structure, func-tion, and composition should be considered atmultiple spatial scales (see fig. 1). For example, itis critical to know how structure, function, andcomposition of the landscape are reflected inpopulation or ecosystem features. Finally, knowl-edge of temporal dynamics of landscape changeis essential. Natural disturbance regimes can becharacterized by their frequency, spatial extent,intensity, and duration. System resilience hasevolved in the context of these disturbances, andthus it is important to compare human-causeddisturbance regimes to natural disturbance re-gimes to determine if human impacts lie within thebounds of system resilience. The concept of his-torical range of variability has value as a bench-mark for human-induced changes. It is importantto understand reciprocal relationships betweendisturbance and landscape pattern (e.g., distur-bances both respond to and create landscapepattern).

Key goals of responsible land management are toprovide for societal needs without usurping theresources of future human generations and tomaintain ecological integrity and thus sustainableecological systems. The concept of ecologicalintegrity refers to system wholeness, including the

presence of appropriate species, populations, andcommunities as well as the occurrence of proc-esses at appropriate scales (Angermeier and Karr1994, Karr 1991). To maintain integrity, it is neces-sary to perpetuate the “characteristic” structure,composition, and processes of ecological sys-tems, preserve those key elements of landscapegeometry that facilitate essential processes, retainthe productive capabilities of the land, and main-tain the evolutionary capabilities of ecologicalsystems. Often resource extraction or use com-promises these features of integrity, and thusmanagement actions seek ways to reinstate thesefeatures across the landscape.

The Land Use Committee of the Ecological Soci-ety of America recommends several guidelines toassist managers in decisions about the use ofland (Dale et al. 2000). The guidelines are pre-sented in full awareness that all of these rules ofthumb cannot be implemented in every (or evenmost) situations. These guidelines suggest that,when possible, land managers should:

• Examine impacts of local decisions in a re-gional context.

• Plan for long-term change and unexpectedevents.

• Preserve rare landscape elements and associ-ated species.

• Avoid land uses that deplete natural re-sources.

• Retain large contiguous or connected areasthat contain critical habitats.

• Minimize the introduction and spread of non-native species.

• Avoid or compensate for the effects of devel-opment on ecological processes.

• Implement land use and management prac-tices that are compatible with the natural po-tential of the area.

These guidelines are based on ecological prin-ciples such as the idea that the size, shape, andspatial relationships of habitat patches on thelandscape affect the structure and function ofecosystems (Dale et al. 2000). This landscapeprinciple has several corollaries:

13

• Landscape elements vary in spatial distribu-tion and quality in time and space.

• The structure and attributes of landscape ele-ments and patches affect movement (flows ofmatter and energy), which in turn affects spa-tial distributions.

• The nature of boundaries between patchesand matrix elements controls horizontal flowsacross landscapes.

• The spatial context (neighborhood) of a patchaffects its properties and dynamics.

Several new insights to management arise from alandscape perspective. The concept of ecologicalintegrity needs further interpretation within thelandscape perspective so that it becomes meas-urable and thus a practical management tool at allscales. Methods are needed to reliably estimatean expected range of natural variation for specificecological systems (see Parsons et al. 1999).

Improved procedures are needed for selectingecological indicators to assess the status andtrend of ecological systems (allowing interpreta-tion of indicators for large spatial and temporalscales).

Acknowledgments

Comments on the abstract by Tony King andtwo anonymous reviews are greatly appreciated.Some of the ideas reported here were developedon a project funded by the Conservation Programof the Strategic Environmental Research andDevelopment Program through the U.S. Depart-ment of Defense Military Interagency PurchaseRequisition No. 74RDV53549127 and the Officeof Biological and Environmental Research, U.S.Department of Energy (DOE). Oak Ridge NationalLaboratory is managed by Lockheed Martin En-ergy Research Corporation for DOE under con-tract DE-AC05-96OR22464.

Figure 1—The structural, compositional, and functional components of ecological systems as viewed at various levels ofecological organization: the landscape, ecosystem, and population (from Dale and Beyeler 2001).

14

Literature Cited

Angermeier, P.L.; Karr, J.R. 1994. Biologicalintegrity versus biological diversity as policydirectives: protecting biotic resources.BioScience. 44: 690-697.

Bartell, S.M.; Gardner, R.H.; O’Neill, R.V. 1994.Forecasting toxicological risk in aquatic eco-systems. New York: Columbia UniversityPress.

Dale, V.H.; Beyeler, S.C. 2001. Challenges in thedevelopment and use of ecological indicators.Ecological Indicators. 1: 3-10.

Dale, V.H.; Brown, S.; Haeuber, R.A. [et al.].2000. Ecological principles and guidelines formanaging the use of land. Ecological Applica-tions. 10: 639-670.

Dale, V.H.; Gardner, R.H.; DeAngelis, D.L. [etal.]. 1991. Elevation-mediated effects of bal-sam woolly adelgid on southern Appalachianspruce-fir forests. Canadian Journal of ForestResearch. 21: 1639-1648.

Hagen, J.B. 1992. An entangled bank: the originsof ecosystems ecology. New Brunswick, NJ:Rutgers University Press.

Karr, J.R. 1991. Biological integrity: a long-neglected aspect of water resource manage-ment. Ecological Applications. 1: 66-84.

King, A.W. 1997. Hierarchy theory: a guide tosystem structure for wildlife biologists. In:Bissonette, J.A., ed. Wildlife and landscapeecology. New York: Springer-Verlag: 185-212.

O’Neill, R.V.; DeAngelis, D.L.; Waide, J.B.;Allen, T.F.H. 1986. A hierarchical concept ofecosystems. Princeton, NJ: Princeton Univer-sity Press. 254 p.

Parsons, D.J.; Swetnam, T.W.; Christensen,N.L. 1999. Uses and limitations of historicalvariability concepts in managing ecosystems.Ecological Applications. 9: 1177-1178.

Suter, G.W., II. 1992. Ecological risk assessment.Chelsea, MI: Lewis Publishers.

Turner, M.G. 1989. Landscape ecology: the effectof pattern on process. Annual Review of Ecol-ogy and Systematics. 20: 171-197.

Turner, M.G.; Gardner, R.H., eds. 1991. Quanti-tative methods in landscape ecology: theanalysis and interpretation of landscape het-erogeneity. New York: Springer-Verlag. 536 p.

15

16

How Do We Determine Scale?

17

Introduction: The Dilemma of Scale

Landscapes fall into a class of objects that fit theadage: “you know it when you see it, but it’s hardto define.” As one looks from a ridge into a for-ested valley, words such as ecosystems, interac-tions, disturbance, and connectivity come readilyto mind. But, it has proven to be another matterentirely to scientifically define a landscape as awhole system of interacting parts that can bereadily integrated into management of naturalresources, environmental assessments, water-shed restoration, and regulation. This is the di-lemma of scale. Landscapes comprise thousandsof components and processes that are difficult tocharacterize at any one time. Furthermore, inter-actions among landscape components occur overdecades to centuries, making it difficult to analyzethem over the short duration of research studies.The study of landscapes, therefore, poses a com-plicated problem.

Major components of riverine habitats depend onthe supply, routing, and storage of inorganic andorganic materials that originate from terrestrialsources. A stochastic climate exerts a degree ofrandomness in the supply of sediment and or-ganic debris to channel networks over 10 to 100years; topography and channel-network geometryimpose a spatially determined organization in therouting and storage of those materials. Hence,aquatic and riparian habitats have both stochas-tic and deterministic origins. Studies that haveincorporated stochastic effects have been re-ferred to as disturbance ecology (Pickett andWhite 1985), temporal hierarchies (Frissel et al.1986), pulses (Junk et al. 1989), and landscapedynamics (Benda et al. 1998). Studies focusingon deterministic aspects are described in termsof continuums (Vannote et al. 1980), spatial hier-archies (Frissel et al. 1986), ecotones (Naiman etal. 1988), and classification systems (Montgomeryand Buffington 1997, Rosgen 1995). Despite asustained interest in ecological processes over

a range of scales (Naiman and Bilby 1998,Swanson et al. 1988), it has proven difficult todevelop general principles on how stochastic anddeterministic landscape factors, in combination,govern habitat development. One example of thislimitation is the continuing inability to define natu-ral disturbance regimes, including the range ofvariability in aquatic and riparian environments(Benda et al. 1998, Naiman et al. 1995). In prac-tice, this has often led to a preference for singleenvironmental states and single-value regulatorythresholds by many scientists, managers, andregulators.

The lack of quantitative and predictive theory onlandscape-scale processes has created a depen-dence on case studies that often have focused onunique and detailed aspects of landscapes, andon classification systems that emphasized spatialdeterminism (i.e., stochastic effects ignored).Moreover, theory absence discourages hypoth-esis testing and commensurable research efforts(similar things measured in similar ways). Theseproblems in combination increase the perceiveddifficulty of landscape-scale and environmentalproblems. In sum, the lack of theory hinders thescientific analysis of landscapes and thereforemanagement planning at the landscape scale.

Role of Theory in the Study ofLandscapes

“Theory” refers to an explicit set of rules and pa-rameters that are used to describe observed phe-nomena in a quantitative manner and that accordwith the empirical record (Gell-Mann 1994, Pop-per 1972). Theories should make testable predic-tions and hence be in a continual state of evalua-tion, rejection, and modification (Popper 1972). Inthe study of landscapes, theories are generallyapplied at small spatial and temporal scales (i.e.,slope stability and sediment transport theories).The term “concept” in the aquatic sciences gener-ally refers to new and innovative ideas, and it isthat class of knowledge where the greatest strideshave been made in articulating the multivariateattributes of landscapes. Concepts, however, donot make testable predictions in the same way

Landscapes: The Dilemma of Scale and the Role of Theory

Lee Benda and Daniel Miller1

1 Earth Systems Institute, 1314 NE 43rd St., Suite 206-207,Seattle, WA 98105-5832; Tel: 425-787-1960/206-633-1792;e-mail: leebenda@aol.com/dan@ohana.org

18

theories do because they contain componentsthat are not fully or explicitly defined. Conceptsmay play a key role in the development of a disci-pline because they act as verbal precursors to thedevelopment of quantifiable and testable theories(Haines-Young and Petch 1986). A third class ofknowledge, “classification systems,” provides anorganizing framework for grouping items into simi-lar categories. Classification is a powerful tech-nique for developing a common vocabulary andfor arraying physical and biological properties ofcertain watershed elements. Typically, classifica-tion is a precursor to development of theories andlaws (Hempel 1966).

Developing theoretical understanding pertinent tolarge scales is a critical step in the coordinatedand organized study of landscapes. First, land-scape-scale theory would encourage the meas-urements of landscape attributes in similar ways,thereby contributing to a regional pursuit of gen-eral principles, similar to theories at smallerscales. This would tend to counter the notion thatevery place is unique and has to be studieduniquely on its own merits. Second, because land-scape study is fundamentally interdisciplinary,theory would encourage and guide how differentscientific disciplines would need to converge ormerge in studies of various phenomena. Third,theory makes it easier and more defensible toextrapolate findings from one landscape to an-other, thereby obviating the need for re-creatingthe wheel at every location. Fourth, theory allows“bridges” to be built among incomplete data (ei-ther temporally or spatially), a strategy that couldbe cast in terms of “hypotheses,” but that wouldallow for more comprehensive understanding.This would make explicit the gaps in data andunderstanding and would aid in targeting futureresearch priorities. Fifth, general theoretical prin-ciples would create a hierarchy of scientific under-standing in which case studies of processes orconditions obtained over small spatial and tempo-ral scales would be evaluated in the context of thelarger scales that characterize landscapes. Theseadvantages would apply to research in the veg-etative, geomorphic, hydrologic, and biotic sci-ences; in natural resource management; and inrestoration and conservation biology.

The Study of Large Numbers ofInteracting Landscape Processes

Clues for developing inductive or deductive theo-ries at large scales pertinent to landscapes arefound in the scientific disciplines that have alreadytackled problems involving large numbers of de-terministic and stochastic elements. One suc-cessful strategy has been to represent the inter-actions of many small-scale processes by largerscale parameters. For example, application ofanalytical mechanics to the problem of landslideprediction views soil as a “continuum,” eventhough soil is composed of many individualgrains. For soil, continuum mechanics representsthe multitude of millimeter-scale grain-to-graininteractions by meter-scale parameters such assoil cohesion, bulk density, and soil friction angle.(A similar approach has been applied to problemsinvolving turbulent fluid flow [e.g., fluid mechan-ics]). Analytical mechanics is most effective whendealing with problems at relatively small spatialand temporal scales, and it runs into difficultywhen applied at larger scales. For landsliding, theuse of a one-dimensional, infinite-slope model isan example of simplifying the interactions of mul-tiple three-dimensional unit volumes of soil in-volved with failures.

Statistical mechanics provides another tech-nique for predicting the behavior of exceedinglylarge numbers of randomly behaving elements. Apurely statistical approach can describe the be-havior of gases that contain vast numbers of ran-domly colliding molecules (James Clark Maxwell[1831-1879] and Ludwig Boltzmann [1844-1906]).To calculate the macroenergy state of a gas inresponse to applied temperature and pressure,molecules are parameterized by probability distri-butions of energy states. As pointed out by Dooge(1986), however, there are large differences be-tween the statistical mechanical approach thatdepends on concepts of energy equilibrium andaverage conditions, and the nonequilibrium andtransient conditions manifest in hydrologic andgeomorphic processes that are of interest to sci-entists and resource managers.

19

In the context of these methods, landscapes con-tain too many components to be treated strictlydeterministically and too few components to betreated purely statistically. Environmental systemsthat fall between the end members of determin-ism and stochastism but that exhibit both charac-teristics have been referred to as “intermediatenumber systems,” or systems of “organized com-plexity” (Weinberg 1975). This characterizationalso has been extended to ecological and geo-morphological systems (Allen and Starr 1982,Graf 1988). To deal with systems of organizedcomplexity, “systems theory” has been developed(Von Bertalanffy 1968). Application of systemstheory typically relies on building comprehensivemathematical models that are used to scale upanalytical descriptions of processes at smallscales to predict the macrobehavior of a systemof such processes over larger space and timescales. This so-called “upwards approach” hasbeen applied to the study of certain hydrologicalproblems (Rodriguez-Iturbe and Valdes 1979,Roth et al. 1989, Smith and Bretherton 1972).

Pursuit of General Landscape Theory

The concept of “organized complexity” is pro-posed as a framework for unifying random andorganized attributes of landscapes that govern theflux, storage, and routing of mass between andwithin terrestrial and aquatic systems. Random-ness refers to behavior that is not predictable indetail (such as climate), but that can be describedin probabilistic terms. Organization refers to spa-tial regularities or patterns in a landscape and caninclude laws of stream ordering and bifurcation(Horton 1945, Strahler 1952), and systematicvariations in network geometry, such as decreas-ing channel gradient and increasing channel widthwith increasing drainage area (Leopold et al.1964).

Pursuit of landscape theory requires landscape-scale parameters. These include temporal distri-butions of the frequency-magnitude characteris-tics of climatic, hydrologic, and geomorphicevents, and spatial distributions that characterizethe attributes of large numbers of landscape ele-ments (Benda et al. 1998). A general landscapetheory is proposed that predicts how mixtures ofclimate, topography, lithology, and vegetation im-pose overarching constraints on the probability

distributions and spatial patterns of sediment andwood flux to streams. Included is how temporaldistributions of material fluxes evolve along achannel network owing to asynchronous materialsupply, network geometry, attrition, spatial scale(drainage area), and transport and storage re-gimes. The derived long-term probability distribu-tions of material flux and storage indicate thestochastic and deterministic origins of aquatic andriparian landforms. Probability distributions alsodefine the space-time structure of variability or thenatural disturbance regime.

The general theory is a work in progress, andhence there is need for testing and refining gen-eral principles pertaining to climatic and vegeta-tion disturbances, erosion regimes, sedimentrouting, and wood recruitment. In addition, newtheoretical principles covering riparian vegetationand the formation of aquatic and valley floor habi-tats are needed. Many of the overarching interac-tions between stochastic and deterministiclandscape factors can be sketched on the back ofa napkin. Simulation modeling and field studiesare needed to make more quantitative and land-scape-specific predictions (Benda and Dunne1997a, 1997b; Benda and Sias 1998). Refer toGeneral Landscape Theory of Organized Com-plexity (Benda et al. 1999) for a more thoroughdiscussion.

Potential Applications of LandscapeTheories:

1. Guide field studies of landscape-scale pro-cesses.

2. Provide context for studies conducted atsmall spatial and temporal scales.

3. Define natural disturbance regimes throughprobability and frequency distributions.

4. Evaluate environmental change through shiftsin distribution form (in time or space).

5. Promote risk assessments that use a probabi-listic approach.

6. Base environmental analyses, resource man-agement planning, environmental regulation,watershed restoration, and conservation biol-ogy on a theoretical foundation.

20

Literature Citations

Allen, T.F.H.; Starr, T.B. 1982. Hierarchy: per-spectives for ecological complexity. Chicago,IL: University of Chicago Press. 310 p.

Benda, L.; Dunne, T. 1997a. Stochastic forcing ofsediment routing and storage in channel net-works. Water Resources Research. 33(12):2865-2880.

Benda, L.; Dunne, T. 1997b. Stochastic forcingof sediment supply to the channel networkfrom landsliding and debris flow. Water Re-sources Research. 33(12): 2849-2863.

Benda, L.; Miller, D.; Dunne, T. [et al.]. 1998.Dynamic landscape systems. In: Naiman, R.;Bilby, R., eds. River ecology and management:lessons from the Pacific Coastal Ecoregion.Washington, DC: Springer-Verlag: 261-288.Chapter 12.

Benda, L.; Miller, D.; Sias, J. [et al.]. 1999.General landscape theory of organized com-plexity. Seattle, WA: Earth Systems Institute.

Benda, L.; Sias, J. 1998. Wood budgeting: land-scape controls on wood abundance instreams. Seattle, WA: Earth Systems Institute.70 p.

Dooge, J.C. 1986. Looking for hydrologic laws.Water Resources Research. 22: 465-585.

Frissell, C.A.; Liss, W.J.; Warren, C.E.; Hurley,M.D. 1986. A hierarchical framework forstream habitat classification: viewing streamsin a watershed context. Environmental Man-agement. 10(2): 199-214.

Gell-Man, M. 1994. The quark and the jaguar:adventures in the simple and complex. NewYork: W.H. Freeman and Company. 329 p.

Graf, W.L. 1988. Fluvial processes in drylandrivers. Berlin, West Germany: Springer. 346 p.

Haines-Young, R.; Petch, J. 1986. Physical ge-ography: its nature and methods. London,England: Harper and Row. 230 p.

Hempel, C.G. 1966. Philosophy of natural sci-ence. Upper Saddle River, NJ: Prentice Hall.

Horton, R.E. 1945. Erosional development ofstreams and their drainage basins: hydro-physical approach to quantitative morphology.Bulletin of the Geological Society of America.56: 275-370.

Junk, W.J.; Bayley, P.B.; Sparks, R.E. 1989. Theflood-pulse concept in river-floodplain systems.Canadian Special Publication of Fisheries andAquatic Sciences. 106: 110-127.

Leopold, L.B.; Wolman, M.G.; Miller, J.P. 1964.Fluvial processes in geomorphology. San Fran-cisco: Freeman. 522 p.

Montgomery, D.R.; Buffington, J.M. 1997.Channel reach morphology in mountain drain-age basins. Geological Society of AmericaBulletin. 109: 595-611.

Naiman, R.J.; Bilby R.E., eds. 1998. River ecol-ogy and management: lessons from the PacificCoastal Ecoregion. New York: Springer-Verlag.

Naiman, R.J.; DeCamps, H.; Pastor, J.;Johnston, C.A. 1988. The potential impor-tance of boundaries to fluvial ecosystems.Journal of the North American BenthologicalSociety. 7: 289-306.

Naiman, R.J.; Magnuson, J.J.; McKnight, D.M.1995. The freshwater imperative: a researchagenda. Washington, DC: Island Press. 165 p.

Pickett, S.T.A.; White, P.S., eds. 1985. The ecol-ogy of natural disturbance and patch dynam-ics. New York: Academic Press.

Popper, K.R. 1972. The logic of scientific discov-ery. 6th ed., rev. London, England: Hutchinson.

Rodriguez-Iturbe, I.; Valdes, J.B. 1979. Thegeomorphic structure of hydrologic response.Water Resources Research. 15(6): 1409–1420.

Rosgen, D.L. 1995. A stream classification sys-tem. In: Johnson, R.R.; Zeibell, C.D.; Patton,D.R. [et al.], eds. Riparian ecosystems andtheir management: reconciling conflicting uses.Gen. Tech. Rep. RM-120. Fort Collins, CO:U.S. Department of Agriculture, Forest Service,Rocky Mountain Forest and Range ExperimentStation: 91-95.

21

Roth, G.; Siccardi, F.; Rosso, R. 1989. Hydrody-namic description of the erosional develop-ment of drainage patterns. Water ResourcesResearch. 25: 319-332.

Smith, T.R.; Bretherton, F.P. 1972. Stability andthe conservation of mass in drainage basinevolution. Water Resources Research. 8(6):1506-1529.

Strahler, A.N. 1952. Hypsometric (area-altitude)analysis of erosional topography. Bulletin of theGeological Society of America. 63: 1117-1142.

Swanson, F.J.; Kratz, T.K.; Caine, N.;Woodmansee, R.G. 1988. Landform effectson ecosystem patterns and processes.BioScience. 38(2): 92-98.

Vannote, R.L.; Minshall, G.W.; Cummins, K.W.[et al.]. 1980. The river continuum concept.Canadian Journal of Fisheries and AquaticSciences. 37(1): 130-137.

Von Bertalanffy, L. 1968. General systemstheory. New York: Braziller.

Weinberg, G.M. 1975. An introduction to generalsystems thinking. New York: Wiley-Interscience.

22

The title of this conference, “Views From theRidge,” suggests a particular scalar view of is-sues. From the ridge, one obtains a somewhatbroad but restricted view of the landscape. Cer-tainly, “Views From Space” would provide a largerspatial overview in which landscape pattern be-comes a dominant theme. For an aquatic or ripar-ian theme, “Views From the Valley” would suggestlooking upward to the hillslope and ridges, in con-trast to looking down from the ridge. Issues con-cerning appropriate scale have been prevalent inmost of the recent landscape assessments, in-cluding the Northwest Forest Plan (FEMAT 1993),the PACFISH (1994) strategy, and, most recently,the Forest Service Roads Analysis (USDA ForestService 1999) procedure. In all of these efforts,three struggles were common: (1) issue identifica-tion and integration of information across multipledisciplines, (2) appropriate spatial scales, and (3)appropriate temporal scales.

Issue Identification and Integration ofInformation Across MultipleDisciplines

Several decades ago, some of us thought that itwould be a good idea to get a bunch of fisheryresearchers and watershed researchers togetherfor a joint meeting. The joint meeting lasted about4 hours until someone voiced the opinion that wehad nothing in common to speak about and themeeting broke up into two different rooms: oneroom for the biologists and another for the physi-cal scientists. Since that time, interdisciplinarywork has improved. At least now we occasionallycan identify common issues. But still, people con-tinue to struggle with understanding the crosscut-ting complexity within a common issue. Forexample, foresters tend to identify forestry issuesas centered around trees; hydrologists see for-estry issues as related to water quality or quantity;biologists see the same forestry issues as revolv-ing around birds, salamanders, or fish. Seldom

are we successful in dealing with the full complex-ity of the issue across disciplines. Traditional waysof looking at problems are either from the topdown, or from the bottom up.

Top Down

The top-down approach (fig. 1) starts with someland use activity, such as logging, grazing, or ur-banization. The next step is to identify the onsitechanges produced by that activity; that is, howdoes that land use activity modify the site—soil,vegetation, terrain, slope, and so forth. Then, howare these onsite changes translated into alteredwatershed products? Primary products of alteredwatersheds are water, sediment, organics, chemi-cals, and heat. And finally, how are these productstransported away from the site of disturbance tocause some offsite impact? For example, sup-pose there are logging and associated roads in aparticular watershed. These activities compact thesoil, modify the vegetation, and alter the topogra-phy by making the slope steeper at road cuts andfills. These physical changes can modify runofftiming and volume, wood input to streams, sur-face erosion, and landslides. The result can pro-duce changes in peak flow, base flow, watertemperature, channel condition, and sediment.Society is more concerned about the conse-quences of these changes offsite: increasedflooding, increased sedimentation, fewer salmon,and so forth. By looking at the full set of potentialinfluences of a land-disturbing activity, a broaderrange of potential concerns can be identified thanif we simply focused on our favorite impact.

Bottom Up

Another equally useful approach (fig. 2), which isa common engineering exercise, is to identifysome offsite impact and trace the way back up tofind the activity that caused that offsite impact. Forexample, if a bridge was washed out, there couldbe many potential reasons including increasedpeak flow, channel erosion, water diversion, bat-tering by debris, and so forth. Identification of thecorrect process and successive linkages is impor-tant in order to be successful in preventing futurefailures or to identify the guilty party.

Scale Considerations for Linking Hillslopes to Aquatic Habitats

Robert R. Ziemer1

1 Chief research hydrologist, U.S. Department of Agriculture,Forest Service, Pacific Southwest Research Station, 1700Bayview Dr., Arcata, CA 95521; Tel: 707-825-2936; e-mail:rziemer@fs.fed.us

23

Figure 1—The top-down approachstarts with a land-disturbing activity,then describes the onsite changes,the subsequent effects of thesechanges, and finally the con-sequences (from Ziemer and Reid1997).

Figure 2—The bottom-up approach starts with an identified consequence (bridge washed out),then describes the important conditions and linkages that could have produced the problem,then the processes that caused the conditions, and finally links to the land-disturbing activities(from Ziemer and Reid 1997).

24

To be most successful, we should analyze theissues simultaneously from the top down andbottom up by linking the land use activity to poten-tial offsite impacts, and also by linking identifiedoffsite impacts to potential land use activities(fig. 3).

Putting It Together

As an example, let us take “disappearing salmon”as an issue for consideration (fig. 4). If we were tosimply focus on the number of salmon at a par-ticular point as the appropriate metric of success,we may develop some programs of salmon resto-ration that are rather silly when the problem isconsidered within its broader context. For ex-ample, we might try to restore habitat above adam or a culvert where the fish are unable toreach. Or, perhaps the reason the fish numbersare low is because they were caught downstream.By producing diagrams similar to figure 4, we canbegin to visualize and understand the complexityand interactions within the issue of concern. Theprocess of developing the diagram is more impor-tant that the final diagram itself. In building thediagram, individuals with different backgroundsand focus can identify where their knowledge con-tributes to the solution of a single issue. In figure4, there are three major components potentiallyaffecting salmon: land use, human predation,

and ocean conditions. The land use and terrestrialconditions include the traditional issues and link-ages: logging, grazing, agriculture, urbanization,dams, and so forth, with their associated effects.The human predation component addressessport, commercial, and subsistence fishing. Theocean conditions influence a major portion of thesalmon’s life cycle.

The traditional view of the problem (fig. 5) is toignore all of this complexity and other influencesand focus on the parts that we particularly careabout. We select a land use of interest and evalu-ate the linkages and pathways between that landuse (logging) and the target concern (disappear-ing salmon). Commonly, we further narrow thescope to a specific component, for example, towoody debris. We want to demonstrate that achange in woody debris has some effect on disap-pearing salmon. So this becomes our top-downapproach. We only think about how woody debrisis affecting the salmon and we ignore all of theother influences.

It is common to find that an agency only considersthose components for which they are directly re-sponsible and ignores the potential effects ofother land uses. For example, a forestry agencybecomes only concerned with the effects of forestland management on salmon, while the influence

Figure 3—The top-down and bottom-upapproaches can be merged into a single-analysis approach by linking the land useactivity to potential offsite impacts, andalso by linking identified offsite impacts topotential land-use activities.

25

Figure 4—A generalized diagram of some possible important interactions affecting “disappearing salmon” (fromZiemer and Reid 1997).

Figure 5—Example of a typical shortcut that ignores many of the important components in the generalizeddiagram to link a particular component to “disappearing salmon” (from Ziemer and Reid 1997).

26

of agriculture, urbanization, dams, fishing, and soforth are ignored because the forestry agency isonly authorized to regulate logging or managetimberland. This focus is appropriate at a latertime when the agency decides upon a program ofaction. Unfortunately, such a myopic view oftenmisses the context of the agency’s program withinthe larger issue and can lead to uneven regulationor to ineffective management actions.

The end point of problem simplification is to selectsome index that directly links the activity to thetarget issue without regard to other influences(fig. 6). For example, a group working to restoresalmon runs in the South Fork Trinity River innorthwestern California assumed that their favor-ite variable, changes in the volume of large poolsin the mainstem river, was related to the numberof returning salmon. The group decided to meas-ure annual changes in the volume of these largepools and then to correlate these annual poolvolume changes to logging and road building,which were assumed to produce decreased poolvolume, and to the amount of future watershedrehabilitation, which was assumed to result inincreased pool volume. In other words, pool vol-ume was the index that was to tie changes in landuse to fish. None of the other components or influ-ences upon fish numbers were to be evaluated orconsidered. The problem was that the group hadno information about what was happening to fishdownstream and no independent indication thatthere was any relation between fish numbers andpool volume, let alone between land use and poolvolume.

Spatial Scale

Individuals who design projects, such as timbersales, roads, grazing permits, recreation facilities,and so forth, are quite accustomed to and com-fortable in dealing with the project or subwater-shed scale (fig. 7) that ranges from 10 to a fewthousand acres. Project designers are less accus-tomed to evaluating the context of that projectwithin larger scales. The appropriate size of thatlarger scale depends strongly on the issue beingconsidered. If, for example, there is a concernabout the effect of a project on the drinking water

supply for a small community, evaluating thesubwatershed directly above the water supplyintake is the appropriate geography and scale.Areas beyond that direct influence are not rel-evant to the problem. If, however, we are dealingwith the effect of a project on anadromous fish,then we are dealing with a much different geo-graphical and spatial arena. For each of the boxesand linkages in the disappearing salmon diagram(fig. 4), there are sets of scales that are appropri-ate to that box. For the fish population, the scalesrange from the individual stream reach to thePacific Northwest, including the ocean. Salmonstocks from the Columbia River may compete inthe ocean for food and resources with salmonfrom northwestern California. Anything thatchanges the competitive advantage of one stockis important to consider. Further, there may bemigratory wandering of fish from one river systemto another. A depleted stock from one river mayresult in success of stocks from another riverbecause of reduced competition, or vice versa. Itis important to recognize such external forces thatare operating at the large scale outside of the im-mediate frame of reference. Similarly, within agiven river system, it is not possible to evaluatethe value of improving fish habitat quality at thesmall watershed scale without some understand-ing of how habitat along the migratory route influ-ences the population. In the extreme example,improving salmon habitat above a migration bar-rier will have no effect, because the fish will neverbe able to use that habitat.

The appropriate scale or geography depends onthe issue to be addressed. Some issues remainfixed in one location (trees, soil fertility), whereasothers are mobile (animals, water, sediment).Products associated with aquatic issues (water,heat, chemicals, wood, sediment) tend to movedownslope or downstream and are constrainedwithin defined topographic boundaries. Fishmove upstream and downstream, so for them,watershed boundaries are useful geographiclimits. Terrestrial animals (deer, birds) are notconstrained by watershed boundaries, and thewatershed concept is not particularly useful. Forthese animals, movement range is a more usefulscale than topographic boundaries.

27

Figure 6—Occasionally, an index is used to link a land-disturbing activity to the target, ignoring all of the com-plexity and interactions that may also influence the target (from Ziemer and Reid 1997). (ECA = equivalent clear-cut area, TMDL = total maximum daily load.)

Figure 7—A hierarchy of spatial scales.

Vegetation change

High water velocities

Disappearing Salmon

Fishing

Habitat change

More predation

Altered riparian vegetation

Less woody debris

Shallow water

Altered circulation and temperature

Industrialization

Subsistence

Sport

Commercial

Ocean Freshwater and estuary

Migration blockage

High water temperature

Altered spawning gravels

Dams

Logging, grazing, urbanization, etc.

Aggradation

More erosion

Higher peak flows

ECA, TMDL,

etc.

The Index Approach

28

A survey of any geographic area will result in ahigh variance for most parameters. For example,consider a hypothetical survey of 30 streams toevaluate the risk of mortality of some species offish (fig. 8). Some streams have good habitat anda low risk of mortality resulting from some action,whereas others will have a high risk. The level of“acceptable” risk has two components, biologicaland social. If the species is abundant, it may bebiologically and socially acceptable to adopt alevel of regulation that would overprotect somestreams and underprotect others. As the speciesbecomes rarer, a higher level of regulation may beappropriate, depending on the consequences ofmaking a judgment error. The problem with regu-lations that produce or require a generic “designerstream” is that stream systems are dynamic andmay require a wide range of evolving habitat con-ditions to be productive. The stream systemsdescribed by Reeves (this volume) require a sub-stantial amount of perturbation and resulting pro-ductivity changes over time. Designing for theperceived “ideal” condition in all places all of thetime may lead to a poor stream condition in thefuture. Further, a poor condition today may con-tain exactly the components needed for the besthabitat in the future.

Temporal Scale

It is well known that “significant” hydrologic ormeteorologic events occur rarely, and the tempo-ral distribution of these events is not uniform. Thispresents a problem because most monitoringactivities represent only a short snapshot of thetemporal distribution of events. If the long-termdistribution was uniform and well behaved, thesnapshot may be an adequate representation ofthe expected population of future events. How-ever, if the events are not uniformly distributed(fig. 9), then any short period of monitoring canproduce flawed information. For example, as-sume habitat conditions are monitored on astream continuously for 75 years, considered bymost to be an exceedingly long record. If themonitoring period ran from year 1 to 75 (fig. 9),the conditions represented would be greatly differ-ent than if the period was from year 75 to 150.More realistically, most monitoring activities are

much shorter than 75 years, often 10 or feweryears. Any 10-year period in figure 9 could findconditions ranging from no severe storms to mul-tiple storms. In other words, the temporal scaleneeded to adequately represent the significantgeomorphic or ecologic drivers is often orders ofmagnitude longer than our monitoring database.

How does this relate to the level of regulation andrisk of mortality? Suppose that the average of thestreams depicted in figure 8 had a monitoringrecord of 30 years (fig. 10). The maximum riskof mortality, and perhaps the appropriate levelof regulation, could differ substantially based onwhich period is monitored: for example, years1 through 10, years 11 through 20, or the entire30-year record.

What is the appropriate time scale to consider?The answer depends strongly upon the issue.Different folks or the same folks consideringdifferent issues look at the problem differently(table 1). For those in the corporate world of prof-its and losses, a quarter (of a year) is an impor-tant scale. Corporate well-being 150 years fromnow is often not an important consideration to theboard of directors. Politicians like to see programsthat they sponsor put into effect and have someresult during their time in office. For politicians,the election cycle (2, 4, or 6 years) is an importanttime scale. The length of a human life is an impor-tant time scale for people, and sometimes plan-ning includes several generations, that is,planning cycles ranging from 10 to perhaps 100years. For most people, something that happened20 years ago was a long time in the past. Withsome exceptions, such as planning for infrequentbut catastrophic events such as earthquakes andfloods, something that happens once every 20years or so is beyond the immediate concern ofmost people. However, a 20-year time scale isextremely long for an insect species having sev-eral life cycles per year, or extremely short for aredwood or bristlecone pine having a life cycle of1,000 years or longer. An individual storm be-comes very important for the domestic water userwho turns on the water tap and finds the water tobe turbid. Geomorphic events that shape thestream channel may occur only once a decade,century, or millennium.

29

Figure 8—Hypothetical survey of streams for risk of fish mortality and the level of regulation needed to protect fishat two levels of abundance (from Ziemer 1994).

Figure 9—Distribution and magnitude of severe storms during a single 300-year simulation (fromZiemer 1991).

Endangered species

Underprotected

Overprotected

Abundant species

Stream identifier

Ris

k o

f m

ort

alit

y

Lev

els

of

reg

ula

tio

n

30

Table 1—Appropriate time scales

Entity Period Years

Corporations Quarterly profits and losses 0.25

Politicians Election cycles 2, 4, or 6

Humans Memory of significant events 1 to 20

Humans Lifespan 50 to 100

Insects Life cycle 0.2 to 1

Anadromous fish Life cycle 2 to 4

Humans Life cycle 50 to 100

Trees Life cycle 100 to 1,500

Domestic water user Individual storm 0.1 to 5

Channel adjustments Large storm 1 to 1,000

Figure 10—Hypothetical risk of fish mortality based on monitoring streams for different periods oftime and the effect of cycles or unusual events on the perceived level of regulation needed toprotect fish (from Ziemer 1994).

31

Finally, how one views the world depends stronglyon conceptual models about how things operate.For example, our belief about how the level ofwatershed disturbance is related to salmon habi-tat (fig. 11) has a strong influence on land man-agement and restoration strategies. The initialassumption for both curves a and b is that thebest habitat represents that area having the leastwatershed disturbance. In one model (fig. 11,curve a), a watershed can be increasingly dis-turbed with little effect on habitat quality until athreshold is reached, beyond which there is aprecipitous decline in habitat quality. The manage-ment objective would be to allow disturbance ac-tivities to continue until just before the point isreached where habitat quality begins to drop rap-idly. Conversely, curve a suggests that a severelydegraded habitat can be restored with a smallreduction in the amount of watershed disturbance.

The second model (fig. 11, curve b) suggests thata small amount of disturbance in watersheds hav-ing the best habitat can result in a rapid decline inhabitat quality. Once the habitat quality is low,additional disturbance has little incremental effecton habitat quality. Conversely, curve b suggeststhat recovery of habitat quality in heavily disturbedwatersheds will require a huge effort before anyimprovement will result. Many past land manage-ment plans followed assumptions of curve a. TheNorthwest Forest Plan (FEMAT 1993) aquaticconservation strategy follows the assumptions ofcurve b, that is, to identify and protect those wa-tersheds that have the best remaining habitat (keywatersheds), and to concentrate continued har-vesting in those areas having the poorest habitat(matrix). It is important to determine which ofthese models best represents the relationshipbetween watershed disturbance and habitatquality.

Figure 11—Two conceptual models of the relation between watershed disturbance and salmonid habitat:(a) habitat quality is not degraded until substantial watershed disturbance is reached; (b) habitat quality isdegraded most quickly during initial stages of watershed disturbance (from Ziemer 1997).

32

Conclusion

Management and policy strategies to sustain aresource depend on physical and biological hy-potheses that often are untested. The success orfailure of a particular strategy will depend stronglyon how the resource actually responds once thatstrategy is applied. Understanding the responseof the resource, in turn, will depend critically onviewing that resource from the appropriate scalein time and space.

Literature Cited

Forest Ecosystem Management AssessmentTeam [FEMAT]. 1993. Forest ecosystem man-agement: an ecological, economic, and socialassessment. Portland, OR: U.S. Departmentof Agriculture; U.S. Department of the Interior[et al.].

PACFISH. 1994. Environmental assessment forthe implementation of interim strategies formanaging anadromous fish-producing water-sheds in eastern Oregon and Washington,Idaho, and portions of California. Washington,DC: U.S. Department of Agriculture, ForestService; U.S. Department of the Interior,Bureau of Land Management. 68 p. + attach-ments.

U.S. Department of Agriculture, Forest Ser-vice. 1999. Roads analysis: informing deci-sions about managing the national foresttransportation system. Misc. Rep. FS-643.Washington, DC. 222 p.

Ziemer, R.R. 1991. An approach to evaluating thelong-term effects of land use on landslides,erosion, and stream channels. In: Proceed-ings, Japan-U.S. workshop on snow ava-lanche, landslide, debris flow prediction andcontrol. Tsukuba, Japan: Organizing Commit-tee of the Japan-U.S. Workshop on snow ava-lanche, landslide, debris flow prediction andcontrol: 533-542.

Ziemer, R.R. 1994. Cumulative effects assess-ment impact thresholds: myths and realities.In: Kennedy, A.J., ed. Cumulative effects as-sessments in Canada: from concept to prac-tice. Edmonton, AB: Alberta Associa-tion of Professional Biologists: 319-326.

Ziemer, R.R. 1997. Temporal and spatial scales.In: Williams, J.E.; Wood, C.A.; Dombeck. M.P.,eds. Watershed restoration: principles andpractices. Bethesda, MD: American FisheriesSociety: 80-95. Chapter 6.

Ziemer, R.R.; Reid, L.M. 1997. What have welearned, and what is new in watershed sci-ence? In: Sommarstrom, S., ed. What is wa-tershed stability? Proceedings, sixth biennialwatershed management conference. ReportNo. 92. Davis, CA: University of California,Water Resources Center: 43-56.

33

Controversies about forest sustainability, ad-vances in the sciences of landscape ecology andecosystem management, and new tools such asgeographic information systems (GIS) and remotesensing have led natural resource policymakers,planners, and managers from the stand, to thelandscape, and to regional scales. As scientistsand managers have expanded to these broaderscales, they have typically encountered multi-ownership landscapes. The management andscientific challenges posed by multiownershiplandscapes are especially complex. Species andecosystems do not recognize legal boundariesbetween ownerships, and the landscape dynam-ics of individual ownerships are controlled by acomplex of economic, social, political, and bio-physical forces. The aggregate ecological condi-tions of landscapes are controlled by the spatialpattern and dynamics of individual owners andecological interactions among those ownerships.The dominant disturbance regimes in many land-scapes are now directly or indirectly controlled byhuman activities. Consequently, to understandand predict these anthropogenic disturbances, wemust also understand their linkages to economics,policy, and sociology. Solutions to problems ofconservation policy and practices for multiowner-ship landscapes do not lie in isolated owner-by-owner planning and management. Broader scaleapproaches are needed. Work in multiownershiplandscapes also reveals the need for increasedintegration among ecological and social sciences.The challenges to conducting integrated regionalassessments are numerous and frequently notappreciated by scientists who typically have littleexperience in these types of efforts.

A group of Pacific Northwest Research Stationand university scientists is currently involved ina research program that is designed to test andevaluate multiownership issues at provincescales. The Coastal Landscape Analysis and

Modeling Study (CLAMS) is a large interdiscipli-nary effort designed to evaluate aggregate ef-fects of different forest policies on the ecologicaland socioeconomic conditions of the CoastRange province as a whole (Spies et al. 2002;www.fsl.orst.edu/clams). Here, I briefly describeour general approach and present an example ofa simulation of changing forest landscape condi-tions over time. I discuss the potential ecologicalconsequences of the mosaic of different owner-ship policies and conclude by identifying some ofthe challenges associated with building integratedregional models.

The goal of CLAMS is to develop and evaluateconcepts and tools to help understand patternsand dynamics of ecosystems at province scalesand to analyze the aggregate ecological and so-cioeconomic consequences of forest policies fordifferent owners. Our approach is based on theassumption that by knowing landscape structureand dynamics of vegetation we can project conse-quences of different forest policies for biologicaland social responses. The major steps in ourapproach are:

1. Build high-resolution spatial models (grainsize of 0.1 to 10 ha) of current biophysicalconditions (e.g., vegetation, ownership pat-terns, topography, streams) across all ownersby using Landsat satellite imagery, forest in-ventory plots, and GIS layers.

2. Conduct surveys and interviews of forestlandowners to determine their expected man-agement intentions (e.g., rotation ages, thin-ning regimes, riparian management intensity)under current policies and to develop spatialland use change models based on retrospec-tive studies.

3. Project expected successional changes inforest structure and composition under differ-ent management regimes by using standdynamics models.

4. Build a landscape change simulation systembased on forest management intentions andforest stand models to project potential land-scape structure for 100 to 200 years.

Landscape Assessment in a Multiownership Province

Thomas A. Spies1

1 Research ecologist, U.S. Department of Agriculture, ForestService, Pacific Northwest Research Station, ForestrySciences Laboratory, 3200 SW Jefferson Way, Corvallis, OR97331; Tel: 541-750-7354; e-mail: tspies@fs.fed.us

34

5. Develop biophysical response models forhabitat quality for selected terrestrial andaquatic vertebrate species, viability of se-lected vertebrates species, coarse-filtermeasures of community and landscape con-ditions, historical range of natural variation offorest successional stages, and landslide anddebris flow potential.

6. Develop socioeconomic response models formeasures of employment and income by eco-nomic sector, timber value and production,recreational opportunities, and contingentvalue of biological diversity to the public.

7. Estimate potential ecological and socioeco-nomic consequences of current forest policiesby using the landscape simulator and variousresponse models.

8. Evaluate, test, and revise overall simulatorsystem and submodels.

9. Provide policymakers, landowners, and thepublic with results of spatial projections ofconsequences and interact with those groupsof people to help inform debate and facilitatecollaborative learning.

At this point in the project we are simulating onlyforest management-related disturbances (e.g.,clearcutting, partial cutting, thinning) and landslideand debris flow disturbances. We focus on thesebecause they are among the most frequent in theregion, potentially have large impact on measuresof biological diversity, and are of great interest incurrent policy debates. We are not simulatingstochastic disturbances such as wildfire, wind,insects, and disease. Studies in the region indi-cate that wildfire occurs infrequently (150 to 400years) and its spatial pattern is only weakly con-trolled by topography, especially for large fireevents (Impara 1997). Smaller wind and pathogendisturbances are quite frequent, but they are diffi-cult to predict and typically occur at patch sizesbelow our level of spatial resolution for this provin-cial study. Climate change is another process thatwe do have in the current model. These otherecological processes could be incorporated intofuture modeling efforts, either directly in the simu-lation model or as scenarios (e.g., effects of alarge fire) for comparative analysis. These pro-cesses could have profound implications to man-

agement. For example, a large, rare fire eventcould influence ecological and socioeconomicsystems for decades and centuries. Relativelyfine-scale processes such as disease or land-slides could affect biophysical potential acrosslarge areas. We do not mean to imply that theseother processes are not important but our initialinterest is in isolating the effects of managementactions. The model should be viewed not as apredictive tool but rather as a computer-basedexperiment to provide insights into the relativeeffects of different forest policies.

We developed a prototype of our landscapesimulator for the Coast Range province and ranit for a 100-year scenario under current policies(fig. 1). Patterns of current forest condition arenonuniformly distributed across ownerships. Cur-rent vegetation patterns in the province are char-acterized by a predominance of early and mid-sized conifer forests. Forests dominated by treesof the largest size classes (large and very largeconifers) are rare and restricted primarily to publiclands. Broadleaf forests are less common thanconiferous forests and tend to be concentrated inriparian areas. Old-growth forest condition (ap-proximately equivalent to the very large coniferclass) is currently a small percentage of the totalarea, and what is remaining is concentrated onBureau of Land Management (BLM) and ForestService lands in the southwestern portion of theprovince. Little old growth occurs on private land,but some small remnant patches do occur andform the basis of Habitat Conservation Plans forthe northern spotted owl (Strix occidentaliscaurina). Conversely, open (pasturelands, mead-ows, agricultural lands, and recent clearcuts) andearly-successional stages of forest (typically for-ests less than 15 to 20 years old) occupy almost40 percent of the province and are concentratedon private lands.

By 50 years into the simulation of future condi-tions, the pattern of vegetation classes haschanged dramatically. Amounts of large-diameterclasses have increased, especially on federallands, and the spatial pattern of vegetation hasbegun to resemble the underlying ownership pat-tern. Young plantations (10 to 30 years old) onfederal lands have matured and are beginning toblend into the matrix of large conifer size classes.On private lands, intensive forest management

35

(40- to 50-year rotations) keep these landscapescycling between early successional stages andharvest-age timber plantations. By 100 years, thecontrasting patterns of vegetation across owner-ships are even stronger.

Although total amounts of late-successional foresthave increased dramatically in the Coast Rangein this simulation, the spatial pattern of these for-ests creates considerable potential edge effectsand spatial pattern interactions, especially on BLMcheckerboard lands. The simulations suggest thatlarge watersheds of the Coast Range will developinto a mosaic of very different landscape typesbased on the amount and spatial pattern of forestconditions. These landscapes range from water-sheds dominated by late-succesional forest towatersheds dominated by early-successionaland mature-forest plantations. Between these

extremes is a wide range of mixtures of succes-sional dominance and dispersed or blocked spa-tial patterns. Consequently, we hypothesize thata new landscape pattern is emerging in this prov-ince in which ownership patterns, managementstrategies, and boundaries will control patterns ofbiophysical processes more than in the past. Theecological and socioeconomic consequences ofchanging diversity and spatial pattern are the pri-mary focus of our ongoing research efforts.

To summarize, we have learned that recently en-acted forest policies in the Oregon Coast Rangehave the potential to create novel landscape pat-terns of vegetation. We hypothesize that in thisdynamic landscape, the combination of complexownership patterns, contrasting managementregimes, and ecological processes create thespatial interactions that could not be predictedbased on information from individual ownerships

Figure 1—Current conditions (as of 1995) and simulated changes in forest types of the Coast Range at 50 and 100 years intothe future under current policies.

36

in isolation from each other. Although simpleanalysis of ownership patterns indicates a strongpotential for aggregate effects in this province,more detailed analyses are needed to test thedegree and distribution of these effects. The spa-tial interactions that we expect to have the great-est impact on the ecological systems of thisprovince are the following:

• Imbalances and gaps in seral stage distribu-tions across environmental strata, includingsubecoregions, watersheds, and topographicpositions.

• Gaps in distribution of habitat of relativelywide-ranging species (such as the northernspotted owl or salmonids) whose movementpatterns are at similar scales to ownershiptracts and management allocations withinownerships.

• Decline in aquatic habitat quality in streamreaches and watersheds as amounts of largewood are lost because of forest and agricul-tural management practices along streamsand upslope in landslide and debris flowareas that are sources of large wood formany streams.

The building of integrated regional models to as-sess forest policies is a relatively new endeavor,and many scientists have little experience withthis type of integrated research, which also maybe conducted in an unfamiliar policy and publicenvironment. We have learned much about theprocess of building integrated regional models toassess ecological and socioeconomic effects. Ourlessons learned include:

• The importance of problem definition andthe conceptual model. Without adequateproblem definition and conceptual framework,the process can degenerate into separatestudies that may not meet project goals.

• The importance of involving policymakersand identifying policy questions. Withoutincorporating policymakers and specific policyquestions at the beginning, the potential rel-evance of the project will be diminished.

• The difficulty of developing spatial infor-mation about landscapes and regions. Spa-tial information about provinces and regions isinadequate and will always be flawed. Thechallenge is to determine when data quality isgood enough to provide a first approximationat large scales.

• The value of landscape projections. Spatialprojections of possible future landscapes area powerful way to engage policymakers andstakeholders in joint learning efforts.

• The challenge of measuring ecologicaleffects. We lack quantitative measures ofecological response. The challenge is to blendempirical, modeling, and expert judgmentapproaches to provide working hypotheses foruse in model projections and to direct futureresearch.

• The challenge and importance of scale.The spatial and temporal scales of ecological,policy, and socioeconomic processes andmeasures are typically not the same. Continu-ous attention to scale is needed to ensure thatlinkages can be made among components.

• The diversity of ways that integration ei-ther happens or not. Integration across disci-plines is central to the effort. Although not allscientists in the team have the time and inter-est to attend to integration of the project as awhole, one or a few leaders must pay closeattention to this process.

• The challenge of conducting science inpublic policy and private landowner envi-ronment. Applying landscape ecology to largemultiownership areas cannot be done entirelywithin the walls of a research institution. Sci-entists must interact with policymakers andstakeholders in new and sometimes uncom-fortable ways. These interactions can be timeconsuming and disruptive to the “normal” pro-cess of research. Without them, however, therelevance of the effort can be seriously jeopar-dized.

37

These lessons on the process are important forscientists, policymakers, funding agencies, andmanagement agencies. These types of effortsrequire, above all, patience, leadership, long-termfunding, and flexibility to deal with different per-spectives and changing goals.

Beyond these challenges and the lessons fromthe research process, we expect this effort tomake significant contributions to policymakingand the science of landscape assessments. Weexpect that the detailed ecological and spatialstructure of the model will help us understand therelative importance of fine-scale managementdecisions at broad scales and the importance ofbroad-scale policy decisions at fine scales. Weexpect to determine the locations in a regionthat can provide the greatest contribution tobiodiversity goals. We expect to learn if a differentmix of policies could provide greater overall eco-

logical and socioeconomic values than the currentpolicies. Finally, we expect to discover how muchfine-scale information is needed to answer ourquestions and understand the dynamics of land-scapes and ecosystems at broad spatial scales.

Literature Cited

Impara, P.C. 1997. Spatial and temporal patternsof fire in the forests of the central OregonCoast Range. Corvallis, OR: Oregon StateUniversity. 354 p. Ph.D. dissertation.

Spies, T.A.; Reeves, G.H.; Burnett, K.M.[et. al.]. 2002. Assessing the ecological conse-quences of forest policies in a multi-ownershipprovince in Oregon. In: Liu, J.; Taylor, W.W.,eds. Integrating landscape ecology into naturalresource management. Cambridge, MA: Cam-bridge University Press: 179-277.

38

Introduction

A prudent manager desiring to meet objectives ofecological sustainability and society’s economicneeds and social desires must consider a varietyof scales, both temporal and spatial. This is per-haps an obvious point, but the techniques andoperational requirements to manage at multiplescales are not always so obvious. Large-scaleassessments and conservation strategies, suchas the Northwest Forest Plan and the ongoingInterior Columbia Basin Ecosystem ManagementProject, illustrate the challenges and frustrationsof managing at multiple scales. In particular,meeting the legal requirements for species viabil-ity, as embodied in the Endangered Species Actand the National Forest Management Act, hasforced a clear and explicit recognition of the needfor a broader view and has crystalized the difficul-ties managers face. In this paper, I introducesome of these challenges and describe recentattempts to meet them.

The viability requirement of the National ForestManagement Act illustrates the many facets ofspecies conservation. Viability is defined in thecurrent regulations (36 CFR 219.19) as:

...a viable population shall be regarded as onewhich has the estimated numbers and distri-bution of reproductive individuals to insure itscontinued existence in the planning area. Inorder to insure that viable populations will bemaintained, habitat must be provided to sup-port, at least, a minimum number of reproduc-tive individuals and that habitat must be welldistributed so that those individuals can inter-act with others in the planning area.

Key points in this definition include considera-tions of numbers, distribution, interactions(i.e., among individuals within populations), habi-tat, and boundary (i.e., planning area). New draftregulations, as proposed following the report ofthe Committee of Scientists (Federal Register2000 65: 67514-67581), define species viabilityas:

A species consisting of self-sustaining andinteracting populations that are well distributedthrough the species’ range. Self-sustainingpopulations are those that are sufficientlyabundant and have sufficient diversity to dis-play the array of life history strategies andforms to provide for their long-term persis-tence and adaptability over time.

New considerations introduced in these regula-tions include adaptability (a concept that includesgenetic diversity), and time.

All Scales Count

To evaluate whether management will contributeto or detract from species viability, one must con-sider a variety of scales, both spatial and tempo-ral. The reason for this is simple: animals re-spond to their environment at a variety of scales.For example, consider the marbled murrelet(Brachyrhamphus marmoratus). This bird nestson large-diameter branches of large-diameterconiferous trees, usually within patches of olderforest. At the microsite level, a tree must have asufficiently large limb to support the bird’s nest. Ata slightly larger scale, the important elements arethe nest tree itself and the structure of the canopysurrounding that tree. A nesting murrelet feeds incoastal waters and makes daily trips to the nest tocare for its young. The canopy must allow spacefor a flight path into the nest. At the same time,cover must be sufficient to protect the nest frompotential predators. Therefore, the structure of thenest stand is also important as the size and shapeof that stand may influence the abundance ofpredators and the susceptibility of the nest to pre-dation. Over a broader scale, one must consider

Implications of Scale on Viability Assessments of TerrestrialWildlife Species

Martin G. Raphael1

1 Research wildlife biologist, U.S.Department of AgricultureForest Service, Pacific Northwest Research Station, ForestrySciences Laboratory, 3625 93rd Ave. SW, Olympia, WA98512-9193; Tel: 360-753-7662; e-mail: mraphael@fs.fed.us

39

the total extent of suitable nesting habitat, as thisdetermines the size of the murrelet population andits overall distribution over a landscape.

As illustrated in this example, several levels ofhabitat selection exist. The smallest scale iswhere individual animals obtain resources atany point in time. It is where a nest is located orwhere a prey item is captured–a specific micrositeor habitat element. An intermediate scale is ananimal’s home range, the geographic area ananimal traverses to obtain resources during aspecified period. A description of habitat condi-tions within a home range might include the ex-tent of various cover types, their arrangementand pattern, and the composition of each of thetypes of cover. A larger scale encompasses theset of conditions that support a population of ani-mals over time. The size of the area to consideris determined by the extent of the population ormay be an arbitrary definition covering a specifiednumber of individuals from within a larger popula-tion. At an even broader scale, patterns of bio-diversity might be described. This scale coversthe geographic area needed to support a commu-nity of species, each of which is representedby populations. In this case, the area is largeenough to contain a variety of cover types andstructures, each of sufficient extent to supportthese populations.

Research on habitat relationships of the northernspotted owl (Strix occidentalis caurina) can be

used to illustrate the importance of considerationof many scales. As evidenced in figure 1, spottedowls choose nest sites in concentrated patches ofnesting habitat. This is suggested by the greaterdifference between nest sites and random sites atthe smallest circle size. As circles around nestsgrow larger, the differences between amounts ofhabitat in nest sites and random sites growsmaller. At very large sizes, both nesting circlesand random circles would reflect average condi-tions in the broader landscape. This exampleshows that smaller scales reveal important bio-logical relationships. In turn, larger scales arenecessary to describe population phenomena, soboth scales are important in assessing habitatconditions.

Another important scale consideration is the effectof study area boundary on measures of landscapepattern. Using the example in figure 1, note thatthe smallest circular area contains only one patchof habitat and that the boundary defined by thecircle cuts off part of that patch. A measure ofpatch size is biased because the circle is toosmall to contain the whole patch. Similarly, meas-ures of edge can be biased because artificialedges are created around the boundary of thestudy area. This suggests that calculation ofmeasures of landscape pattern should be doneon a landscape that is much larger than the aver-age patch size in that landscape, perhaps on theorder of 10 to 50 times as large as the averagepatch size.

Figure 1—Amount of spotted owl nesting habitat in circular areas of varying size. The figure on the left is acomparison of mean (+ standard error) percentage of habitat in a sample of 89 nest sites and 100 randomlylocated sites, Coast Range, Oregon. The figure on the right is a sample of one site showing habitat (dark) anda nested set of circular areas.

Circular Area (ha)0 200 400 600 800 1000 1200

Per

cen

t H

abit

at

10

15

20

25

30

35

40

45

50

Owl SitesRandom Sites

Circular area (hectares)

Owl sites Random sites

Hab

itat

(p

erce

nta

ge

of

area

)

0 200 400 600 800 1,000 1,200

40

The temporal dimension is also important. Spe-cies viability cannot be assessed at a single pointin time. At any particular time, a species (or popu-lation) exists or it does not. Land managers aregenerally more interested in the future. A typicalquestion is, If I manage the land in some manner,is it likely the species will persist into the future?Time can be assessed relative to generation timeof the species in question. Short-lived species(e.g., most insects or annual plants) experiencemany generations over a decade; longer livedspecies may take a century or more for multiplegenerations to pass. Viability assessments ofvertebrates are often considered over hundredsof years.

Species Viability in the InteriorColumbia River Basin

The interior Columbia River basin presents an-other good illustration of the importance of scalein evaluating viability of terrestrial wildlife (fig. 2).In this case, we2 assessed habitat conditions fora set of individual species at two geographicscales, the subwatershed and the entire basin.At the subwatershed scale, we estimated the totalextent of primary habitat within each of about7,000 subwatersheds that make up the basin.Quality of habitat was adjusted by using modelsthat accounted for more specific habitat elementsthat might affect populations in that watershed.

2 Raphael, M.G.; Wisdom, M.J.; Rowland, M.M. [et al.]. 2001.Status and trend of habitats of terrestrial vertebrates in rela-tion to land management in the interior Columbia River basin.In: Forest ecology and management. Elsevier Science B.V.153(1-3): 63-87.

Figure 2—Trend in habitat of the pygmy nuthatch from historical to current to hypothetical future in the interiorColumbia River basin.

41

We then displayed the geographic distribution ofhabitat quality over a species’ range at three peri-ods in time (fig. 2): historically (circa 100 yearsago), currently (average conditions over the pastdecade), and future (100 years into the futureunder a supplemental draft environmental impactstatement alternative). We used three generalindices to evaluate, at the broad scale, conditionof habitat as it might influence species viability ateach period. We summarized the total amountof habitat (actually a weighted average of habitatquality relative to historical conditions), percent-age of historical range that is occupied (extent),and connectivity of habitat. By using Bayesianmethods and conditional probability tables, thesethree indices were combined (see footnote 2 fordetailed methods) to summarize the likelihood ofconditions that would support the species at eachtime.

As shown in this example, trend in any onesubwatershed is not particularly informative inevaluating viability. It is necessary to take a muchbroader view encompassing the species range togain such insight. But conditions within eachsubwatershed are critical for building an aggre-gate picture of habitat suitability for a given spe-cies over an extensive landscape.

Conclusion

Scale is a vitally important consideration in as-sessing species viability. No one scale is suffi-cient. We need to consider all scales, from localto regional, and each is important in answeringdifferent questions.

42

Natural resource science and management seemto have entered the “age of the landscape.” His-torically, resources have been studied and man-aged from a stand, or site, perspective and havegenerally been approached as single-issue prob-lems. Forests, water, fish, and biodiversity weretreated largely as separate topics. The recent shiftto a landscape perspective is an explicit recogni-tion that the world is far more complex than canbe comprehended from strictly site-based, single-issue studies. Thus, scientists and managers arechallenged to address resource issues at multiplescales: site, landscape, regional, national, interna-tional, and integrated across multiple resources.Examples of the latter include vegetation andwildlife habitat interactions; upland vegetationdynamics affecting water quantity and quality;climate, vegetation, pest, disease, and fire inter-actions; ecosystem productivity and biodiversity;and many others. Natural resource managementhas become so pervasive that it is comparable inscale to most large-scale natural patterns of re-source structure and function.

My intent in this discussion is to point out some ofthe prevailing issues and challenges facing re-source managers and scientists alike in predictingor “forecasting” possible trajectories of such com-plex landscapes. Under the assumption that theclimate and land use patterns of the future will bedifferent from the past, I presume that a process-based modeling approach is required, that is, anapproach that is robust to changing environmentalconditions and that can comfortably incorporateincreasing levels of complexity. The discussionreflects my personal views and is not intended tobe a complete review of the issues and technolo-gies available.

Natural resource issues occur across a range ofspatial and temporal scales. Most processes aresmall scale, for example leaf physiology or water

infiltration. Others are intermediate, or landscapescale, such as metapopulation dynamics or seeddispersal. Few processes are truly large scale.Most large-scale patterns are recognized as“emergent properties,” that is, they arise fromsmall-scale processes and interactions yet revealcoherent large-scale patterns. The distributions ofspecies and ecological zones and the regionalpatterns of species diversity are examples ofemergent properties. Such large-scale emergentproperties often are correlated with large-scaleclimate patterns. However, the regional climatepatterns are themselves emergent properties ofthe small-scale fluid dynamics of the atmosphere,playing out on a rotating sphere and interactingwith oceans, continents, mountains, and othersurface features, including vegetation.

Correlations between large-scale resource pat-terns and climate are useful in calling attention tothe relationship and point to likely causation. If wecould view the climate as a very slowly changingphenomenon, for example, glacial-interglacialcycles, then we could largely ignore mechanismsof interaction between resources and climate andsimply focus on the resource processes. How-ever, we now know that the world is heating upand that perhaps half of the current rate of heatingis human caused (Trenberth 2001). We cannotsay with certainty what the future climate holds,only that we cannot assume the status quo. Thus,we must strive to understand and predict resourceinteractions and future trajectories from first prin-ciples, which take us to the small scale of dy-namic processes. Yet, we must acquire predictivecapabilities of future resource trajectories overlarge spatial extents. It is ultimately the regionalemergent properties of resources for which wemust manage. The challenge of the day, there-fore, is to develop predictive capability over verylarge spatial extents from small-scale processes.

Projecting Potential Landscape Dynamics:

Issues and Challenges

Ronald P. Neilson1

1 Bioclimatologist, U.S. Department of Agriculture, ForestService, Pacific Northwest Research Station, ForestrySciences Laboratory, 3200 SW Jefferson Way, Corvallis, OR97331; Tel: 541-750-7303; e-mail: rneilson@fs.fed.us

43

A Framework for Science andManagement

Scientists appear to be converging on a three-scale structure for developing model-based pre-dictive capabilities of resource trajectories andmultiresource interactions. These are the site, orstand scale, spanning a few tens of meters; thelandscape scale, spanning a few to tens of kilo-meters; and the regional or continental scale,spanning hundreds to thousands of kilometers.Approaches to these three scales are either topdown (empirically based) or bottom up (processbased), being distinguished primarily by the reso-lution of the simulation grid. Stands may be con-sidered as spatially homogeneous at a scale toabout 10 meters and may be treated as pointsimulations, e.g., traditional stand or “gap” models(Shugart and Smith 1996). Stands are also oftensimulated as spatially explicit with a grid resolutionof perhaps 1 meter, but this is less common(Miller and Urban 1999, Peters 2000). Each gridcell contains a full model simulation and may ormay not include cell-to-cell interactions. The spa-tially explicit approach is bottom up; the spatiallyaggregated approach is top down. The stand orpoint model can be placed on a grid for spatiallyexplicit simulations of heterogeneous landscapeswith a grid resolution of perhaps tens of meters.Regional-scale models are of necessity coarsegrid, ranging in grid resolution from about 1 km upto about 50 km or more. Clearly, even the highestresolution regional model must absorb consider-able spatial heterogeneity within each grid cell.

Independent models have been developed at allof these scales; however, there is little coherencyamong those differently scaled models. For ex-ample, traditional “gap” models, which simulatesuccession in a small opening, such as a single-tree blowdown, use empirical curves based ondegree-day sums (temperature sums over time)(Botkin et al. 1972), whereas other models mightuse elaborate physiologically based processes forthe same purpose (Running and Hunt 1994). Inthe extreme, such models can produce oppositeresponses to external forcing, such as climatechange, even though both may be equally wellcalibrated to current conditions (Loehle andLeBlanc 1996).

One approach to building a verifiably accurateregional-scale model is to take advantage of thetop-down and bottom-up approaches across all

scales while relying on a core modeling paradigmthat carries through all scales. Thus, the core toall of the scales is the site-based model, whichmust incorporate such things as leaf physiology;competition among life forms for light, water, andnutrients; soil hydrology; and other processes.At each unique scale, scale-specific phenomenalikely must be included, but the approach mayvary depending on whether it is top down orbottom up. Many forest simulation and manage-ment models do not require the physiological,biogeochemical, and hydrological processes(Crookston and Stage 1999). These are empiri-cally based models that cannot accept a chang-ing climate as input. I thus restrict my commentsto the process-based approach to ecosystemmodeling.

The unique aspect of global ecosystem modelingin comparison to more traditional ecological mod-eling is that the primary points of interest are theemergent, large-scale spatial patterns and theirdynamics. Large-scale biogeographic modelingcan be simulated from small-scale processes(leaf to landscape), but calibrated to large-scalebiogeographic and hydrologic patterns (Neilson1995). The challenge is to find the simplest modelstructure that is sufficient to capture the neces-sary processes at all the appropriate scales. Onemay find that a particular mathematical represen-tation of a process works fine in most locations,but needs to be recalibrated at each. Presumably,however, if one is appropriately representing theprocesses, a single calibration will suffice in alllocations. For example, I used large-scale biogeo-graphic and hydrologic patterns as simultaneousconstraints to help determine both model struc-ture and calibration (Neilson 1995).

In constructing the Mapped Atmosphere-Plant-Soil System (MAPSS) vegetation distributionmodel, I coupled the simulation of leaf area index(LAI, the area of leaves per unit area of ground) tothe simulation of site hydrology (Neilson 1995).The LAI was used, in part, to define boundaries orecotones between different vegetation types, suchas that between forest and savanna, thus provid-ing the biogeographic constraint at the regionaland continental scales. The simulation of LAI wasdependent upon the accurate simulation of sea-sonal soil moisture and its distribution through thesoil profile, as transpiration is a function of leafarea, stomatal conductance, and vertical rootdistribution. The hydrologic module was calibrated

44

against locally observed runoff in contrasting re-gions across the United States and was initiallystructured quite simply. However, the simplestructure required unique calibrations for eachregion of the country. I incrementally increasedthe complexity of the hydrology model until I foundthe minimum complexity that would accuratelysimulate runoff in all the diverse regions of thecountry with a single calibration, and which alsoaccurately simulated the LAI to provide propermapping of vegetation boundaries. Thus, themodel was simultaneously constrained by bothvegetation and hydrologic patterns at the largescale of emergent properties, but was based onthe simulation of small-scale processes.

A robust framework of science can be constructedby combining the top-down and bottom-up ap-proaches. Each spatial scale can be representedfrom both perspectives. At the top of the hierar-chy, grid cells are sufficiently coarse to simulatewall-to-wall coverage but with uncertainties owingto subcell heterogeneity, because each grid cell isusually treated as being spatially homogeneous.Higher resolution grids can be used; however, thefiner scales would only be feasible over smallerspatial extents, that is within a sampling frame-work over specific watersheds with subplots andso on. Each scale can inform the other; processconstraints can be imposed through all levelsfrom a common small-scale theory, and large-scale constraints, such as biogeographic patterns,can be imposed from above. Thus, each level hasthe ability to peer up or down one level, and thestructure can be moved up and down levels for asmany levels as are of interest. Yet, the core pro-cesses are applied at all spatial scales, thus en-suring that the patterns and sensitivities of theapplications at different scales will be fully inter-nally consistent.

Historical management practices can be similarlyapproached, spatially explicit at a scale of focus,statistical below, and deterministic above. Forexample, at the national scale, targets might beset for harvest or fuel reduction practices, buttheir locations are not determined. These targetsmight be further dissected to specific regions,providing a spatially determined component, butwithin each domain, the cutting or thinning repre-sents a statistical target and is not spatially deter-mined. This process can continue iteratively, untilfinal operations are spatially determined at the

finest scale of interest. This is the inverse of thenatural landscape, where the large-scale patternsare emergent from below and constrained fromabove. Management emerges from the top and isimplemented below. However, top-down manage-ment also clearly is constrained by the bottom-upspatial resource availability and condition. Nested-scale management modeling can be verticallyintegrated with the nested-scale ecological model-ing. Some of the issues and challenges in con-structing the ecological side of this dual hierarchyare addressed below.

Subcell Heterogeneity: Representingthe Landscape in Coarse Grids

Much of the challenge of large-scale resourcemanagement revolves around the opposing con-cepts of spatial heterogeneity and homogeneity.For example, as one moves from wet to dry alongan aridity gradient, the density of a forest will thinto a point where a grassy understory just beginsto be supportable because of enhanced under-story light. This is an unstable threshold conditionbecause new processes suddenly enter the sys-tem forcing it away from the point of transition asfollows. Introduction of an understory entrainspositive feedbacks through competition for water,which further thins the canopy overstory. Addi-tional feedbacks through fire can thin the over-story even more, allowing yet more grass andmore fire until equilibrium is reached. Thus, thesystem undergoes a transition from a spatiallyhomogeneous entity to a heterogeneous one.Along this hypothetical aridity gradient, with notopographic complexity, there is an endogenousshift from a homogeneous system (forest) at thewet end to a heterogeneous system (savanna)with increasing aridity and back to a homoge-neous system (grassland) with further increasesin aridity. The system has also gone through scaletransitions, from coarse-scale forest, to fine-scaletree-grass patches, to coarse-scale grassland.Transitions between these physiognomic shifts inheterogeneity are generally termed ecotones andoccur at the point of finest scale interfingeringpattern between the adjacent homogeneoustypes.

If we interject topographic complexity to the abovemoisture gradient, the spatial disposition of eco-tones both horizontally and with elevation canbecome quite complex along both elevational and

45

horizontal temperature and moisture gradients. Atransect along the west slope of the Rocky Moun-tains from southern Idaho to the Mexico borderillustrates the complex shifts in elevational eco-tones along latitudinal temperature and moisturegradients (Neilson 2003). Winter precipitationdecreases from north to south, but summer pre-cipitation increases from north to south in the formof the Arizona monsoon. In combination, the win-ter temperature and summer moisture gradientscreate a latitudinal “wedge” of ecotones, upperelevational ecotones increasing with decreasinglatitude and lower elevational ecotones decreas-ing with decreasing latitude. In the southern partof the transect, the wide elevational separation ofecotones creates the classic ecosystem zonationpatterns described by Whittaker and Niering(1965) on the Santa Catalina Mountains, Arizona,and earlier by Merriam (1890) on the San Fran-cisco Mountains, Arizona. At the northern part ofthe transect, however, the elevational ecotonesconverge, creating a tremendous amount of spa-tial diversity across microhabitats within compara-tively small landscapes. The result is a spatialpattern of complexity through the region that con-tains both vertical and horizontal gradients of di-versity. Peet (1978) described a similar gradientalong the east slope of the Rocky Mountains.

It is well recognized that diversity tends to in-crease at ecotones as the dominant organismsfrom neighboring domains intermingle into com-plex combinations (Hansen et al. 1992). Treesand grass, for example, interdigitate at the prairie-forest ecotone, enhancing diversity. The sametype of interdigitation and spatial diversity gradi-ents occur at elevational ecotones at the southernend of the Rocky Mountain transect, for examplein the Santa Catalina Mountains. At the northernend of the transect, however, with the spatial con-vergence of ecotones, the different vegetationzones sort out on unique topographic slopes, as-pects, and soils, compressing the interdigitation ofvegetation from the macroscale to the microscaleand creating a wholly new elevational zonationpattern.

Thus, attempts to understand the patterns of al-pha, beta, and gamma diversity (local, gradient,and regional) at, for example, only one end of theRocky Mountain transect, would be only partiallyrevealing and would provide little general under-

standing of the landscape patterns. The empiricalanalysis that led to the description of this wedgeof ecotones was based on a set of nested-scaleexperimental seedling transplants along environ-mental gradients at scales of meters (shrub tointershrub), tens of meters (landscape uniquegeomorphic positions), hundreds of meters (el-evation), and hundreds of kilometers (regional)(Neilson and Wullstein 1983). Yet, simulations atthe relatively coarse scale of 10-km resolution(Neilson 1995) were able to elicit the same re-gional gradients in ecotones, providing inferencesto spatial patterns and processes at much smallerthan grid-cell resolution.

Such convergence of ecotones may tend to occurwhere steep airmass gradients converge. I pro-pose that these “nodes” of airmass convergencedrive a rescaling of ecological gradients, which ismost apparent at the landscape scale. At a dis-tance from these nodes, homogeneous vegeta-tion occurs in large patches, but near the nodes,smaller patches of vegetation intermix on micro-sites dictated by topography and soil (Neilson etal. 1992). But, perhaps most interesting, is thepossibility of inferring landscape-scale patternsfrom the coarse-scale patterns simulated by themodels. Such top-down inference of landscape-scale pattern must, of course, be validated withdirect simulation of spatially explicit landscapepatterns across regional gradients.

Much of the art in landscape simulation arisesin the challenge to validate innovative top-downmodeling techniques with direct, bottom-up simu-lations and ground- or satellite-based data. Thequestion becomes one of how to represent andsimulate subcell patterns and dynamics accu-rately, but without having to do the direct simula-tion at all locations. A topographically inducedmosaic of forests and grasslands would appearas a savanna in a large grid cell. The savannamay provide sufficient accuracy for some issues,but for others, it clearly will not. The spatial het-erogeneity can be handled through explicit bot-tom-up simulation of the complex terrain, or asunique simulations of each patch type as anaggregated homogeneous entity. There are alsoinnovative ways to explicitly recognize heteroge-neity in the aboveground components at the land-scape scale, while preserving a more homogene-ous belowground competitive environment.

46

Fires and other disturbances in the landscapeproduce significant problems for large-scale simu-lations with coarse grids. Fires create a mosaic ofuneven-aged patches, with new patches beingcreated as often as each year in some cases.There are numerous differences in the vegetationof an area 1 year after fire and 15 years after fire.In contrast, the differences in vegetation 100years and 115 years after fire are minimal if theclimate and substrate are the same. Thus, oneapproach is to allow creation of new patches eachyear and track them individually, but as they be-come increasingly similar with age, merge themback together. Initially in the course of model de-velopment, these patches, in an otherwise homo-geneous grid cell, might be noninteracting andonly be represented uniquely by their areas andages. In grid cells with complex terrain, thesepatches could be maintained on unique soils andwith unique climates, but again nonspatially. Even-tually, some level of interaction among patchescould be implemented, but still without spatiallyexplicit representation within the cell. Spatiallyexplicit simulations on fine-scale grids, coupledwith ground and satellite data would assist in thevalidation of these top-down approaches.

Topographically induced spatial heterogeneitypresents additional difficulties when integratingacross resources, such as between vegetationand hydrologic processes. Hydrologic processesare strongly coupled to vegetation processes andspan scales from local infiltration processes toregional river routing, yet most of the physics oc-curs at very fine scales. A bottom-up modelingapproach will couple the vegetation and hydro-logic point-scale processes and also account forthe horizontal flows of surface and subsurfacewater. Thus, accurate representation of riparianzones and stream hydrology is possible. Repre-sentation of these interactions in coarse grids alsomay be possible along the lines described above,but would require direct linkage of subgrid patchtypes for routing of water between upland andlowland parcels.

Conclusions

Current modeling approaches for natural resourcemanagement often are organized around threedifferent scales: patch, landscape, and regional to

global. Most large-scale modelers are attemptingto incorporate the important processes that occurat all three scales: patch (competition, gas ex-change); landscape (fire, dynamic heterogeneity);and global (emergent, spatial pattern). It will bevery important for practitioners working within anyone of these three modeling communities to coor-dinate closely with those working at the otherscales. Patch models built around one type ofecosystem or in one region may not be well struc-tured for working in other systems or regions.Consistency of process should be maintainedacross scales. If models are to be nested orlinked across scales, then their processes shouldbe based upon the same theoretical underpin-nings, or they may not translate well acrossscales. An area of research that I believe mayhave some potential, but that remains largelyuntapped, is the possibility of downscaling fromregional to landscape patterns by using coarse-scale information, as in the relation between spa-tial ecotone convergence and landscape diversitypatterns.

The key points of this discussion serve to empha-size the importance of accurate simulation of eco-system constraints at all relevant scales. Under arapidly changing climate and with changing physi-ology under elevated CO2, constraints normallyassumed to be stationary must now be assumedto be dynamic and must be explicitly simulated.Heterogeneous landscapes are among the mostcomplex, yet globally among the most commontypes of ecosystems. Accurate simulation of land-scape patterns and processes under globalchange requires attention to organism-level andlower processes within the constraints of biome-level dynamic biogeography. A carefully crafted,hierarchical framework with dual-scale represen-tation at each level can provide the structure forintegration among natural resources and betweenmanagement and natural processes in a rapidlychanging environment.

English Equivalents

1 meter (m) = 3.28 feet

1 kilometer (km) = 0.62 miles

47

Literature Cited

Botkin, D.B.; Janak, J.F.; Wallis, J.R. 1972.Rationale, limitation, and assumptions of aNortheastern forest growth simulator. IBMJournal of Research and Development. 16(2):101-116.

Crookston, N.L.; Stage, A.R. 1999. Percentcanopy cover and stand structure statisticsfrom the Forest Vegetation Simulator. Gen.Tech. Rep. RMRS-GTR-24. Fort Collins, CO:U.S Department of Agriculture, Forest Service,Rocky Mountain Research Station.

Hansen, A.J.; Risser, P.G.; di Castri, F. 1992.Epilogue: biodiversity and ecological flowsacross ecotones. In: Hansen, A.J.; di Castri, F.,eds. Landscape boundaries: consequences forbiotic diversity and ecological flows. New York:Springer-Verlag: 423-438.

Loehle, C.; LeBlanc, D. 1996. Model-based as-sessments of climate change effects on for-ests: a critical review. Ecological Modelling. 90:1-31.

Merriam, C.H. 1890. Results of a biological sur-vey of the San Francisco Mountains region anddesert of the Little Colorado, Arizona. NorthAmerican Fauna. 3: 1-136.

Miller, C.; Urban, D.L. 1999. A model of surfacefire, climate and forest pattern in the SierraNevada, California. Ecological Modelling. 114:113-135.

Neilson, R. 2003. The importance of precipitationseasonality in controlling vegetation distribu-tion. In: Weltzin, J.; McPherson, G., eds. Pre-cipitation and terrestrial ecosystems. Tucson,AZ: University of Arizona Press: 47-71.

Neilson, R.P. 1995. A model for predicting conti-nental-scale vegetation distribution and waterbalance. Ecological Applications. 5: 362-385.

Neilson, R.P.; King, G.A.; DeVelice, R.L.;Lenihan, J.M. 1992. Regional and local veg-etation patterns: the responses of vegetationdiversity to subcontinental air masses. In:Hansen, A.J.; di Castri, F., eds. Landscapeboundaries. New York: Springer Verlag: 129-149.

Neilson, R.P.; Wullstein, L.H. 1983. Biogeogra-phy of two southwest American oaks in relationto atmospheric dynamics. Journal of Biogeog-raphy. 10: 275-297.

Peet, R.K. 1978. Latitudinal variation in southernRocky Mountain forests. Journal of Biogeogra-phy. 5: 275-289.

Peters, D. 2000. Climatic variation and simulatedpatterns in seedling establishment of two domi-nant grasses at semi-arid—arid grasslandecotone. Journal of Vegetation Science. 11:493-504.

Running, S.W.; Hunt, E.R. 1994. Generalizationof a forest ecosystem process model for otherbiomes, biome-BGC, and an application forglobal-scale models. In: Ehleringer, J.R.; Field,C., eds. Scaling processes between leaf andlandscape levels. San Diego: Academic Press:141-158.

Shugart, H.H.; Smith, T.M. 1996. A review offorest patch models and their application toglobal change research. Climatic Change. 34:131-153.

Trenberth, K.E. 2001. Stronger evidence of hu-man influences on climate: the 2001 IPCCassessment. Environment. 43: 8-19.

Whittaker, R.H.; Niering, W.A. 1965. Vegetationof the Santa Catalina Mountains, Arizona. (II) Agradient analysis of the south slope. Ecology.46: 429-452.

48

This manuscript overviews issues of scale in rela-tion to the core research of the USDA Forest Ser-vice, Pacific Northwest Research Station (PNW)Boreal Ecology Cooperative Research Unit lo-cated in Fairbanks, Alaska. The unit’s main re-search focus is the long-term ecology of borealforests in Alaska’s interior region. Other ongoingand related projects focus on pattern and changein tundra and on effects of silvicultural practiceson white spruce (Picea glauca (Moench) Voss;the major economically important tree species)regrowth following various cutting and clearingtreatments. Alaska’s interior region and NorthSlope have experienced a warming trend overthe past two decades that has been unprec-edented in the last century (Barber et al. 1998,Hammond and Yarie 1996, Juday et al. 1998,Serreze et al. 2000). Understanding how key pat-terns and processes scale in time and space isabsolutely critical to understanding how this sys-tem may respond to changing temperatures.

An alternative title for this paper, “Boreal forest–Tundra with trees?”, makes reference to the simi-larities between tundra vegetation and taiga forestunderstory. The understory vegetation in flat,low-lying areas in northern boreal forests is re-markably similar in both composition and struc-ture to tussock tundra vegetation (Jorgenson et al.1999, 2001; Walker et al. 1994). This type iscodominated by sedges (Eriophorum vaginatumL., Carex lugens Holm) and shrubs (Betula nanaL. or B. glandulosa Michx., Ledum palustre L. ssp.decumbens (Ait.) Hulten or L. groenlandicumOeder, Salix L. spp.) and underlain by completecoverage of mosses and lichens (Sphagnum L.spp., Hylocomium splendens (Hedw.) Schimp.in B.S.G., Pleurozium schreberi (Brid.) Mitt.). Inthe taiga, black spruce (Picea mariana (P. Mill.)

B.S.P.) is also present in open, sparse stands.There is a gradient between full-scale wetlandswith few trees to fully closed forests withoutpermafrost.

The need to explicitly recognize and account forscale in ecosystem studies has become increas-ingly appreciated in the past two decades. Studiesthat examine pattern or process in space mustconsider explicitly the sources of variation that areeither secondary or primary to the research. Thisis not a new idea; it is the foundation of biogeog-raphy and classical plant ecology, both of whichlook at patterns of organisms in space and at-tempt to determine the causal mechanisms be-hind those patterns. What has become increas-ingly recognized is how closely together temporaland spatial patterns are tied in most systems, andalso how traditional ecological hierarchies do notmap directly to more fundamental hierarchies(Allen and Hoekstra 1992, Allen and Starr 1982,O’Neill et al. 1986). Because all landscapes havesome degree of dynamism, the spatial patterns atany point in time are the result of multiple pro-cesses operating at multiple time scales (Delcourtet al. 1983). The National Science Foundation’s(NSF) Long-Term Ecological Research (LTER)Program addresses the problem of temporalscales directly by focusing research resources atselected sites over periods sufficient for explicitstudy of decadal-scale pattern (Franklin 1987).The Bonanza Creek and Caribou-Poker CreeksLTER site in the Fairbanks region is one of sixLTER sites nationally that are cofunded by NSFand USDA, and is one of two such sites withinPNW.

Northern ecosystems represent opportunitiesto examine natural processes at their limits. Byfocusing at these limits, we can better understandthese processes and their roles in complex situa-tions where many factors are operating simulta-neously. What is most distinctive about thesesystems as a whole is that a single factor, tem-perature, controls essentially all other processes(Chernov and Matveyeva 1997). Thereforechanges in this single factor will affect almost all

Scaling and Landscape Dynamics in NorthernEcosystems

Marilyn Walker1

1 Unit leader, U.S. Department of Agriculture, Forest Service,Pacific Northwest Research Station, Boreal EcologyCooperative Research Unit, P.O. Box 756780, University ofAlaska Fairbanks, Fairbanks, AK 99775-6780; Tel: 907-474-2424; e-mail: mdwalker@fs.fed.us

49

processes and patterns. I examine the interac-tions between climate, species abundance anddistribution, and disturbance regimes along agradient from the northern limit of plant growth tothe most northern forests (fig. 1). This gradientcovers key thresholds of biodiversity, including thefirst introduction of woody-dominated ecosystemsin the low arctic to the first dominance of trees inthe northern boreal forests. The introduction ofnew functional types into the system, first shrubsand then trees, has dramatic impacts on temporaland spatial patterns, and on how disturbance andvegetation interact.

Vegetation zonation within the arctic regions hasbeen described by many authors, and variousapproaches have been used to describe thesegeobotanical zones (Walker 2000). Althoughslightly different criteria can be used resulting indifferent outcomes, within Alaska it is useful torecognize three broad zones defined by summertemperature and dominant plant structural types.Table 1 lists characteristics of the three majorarctic zones along with transitional and fully borealzones. There are multiple processes and charac-teristics associated and interacting with thesestructures. Species richness is very tightly corre-lated with temperature at multiple scales in thesesystems (Chapin and Danell 2001, Walker 1995,Walker et al. 2001). The decrease in species rich-ness from south to north is mainly due to loss ofspecies rather than turnover in species, so thatthe most northerly floras represent depauperateversions of the southern ones (that is not whollytrue, but it is a reasonable representation). There-fore, the latitudinal similarities among the vegeta-tion and flora increase from south to north,because there is a smaller pool of species avail-able. Until treeline is encountered, the dominantspecies are the same or closely related through-out the zone; however, the species representingthe northern treeline vary throughout the region(Alexandrova 1980).

Because this is such a major gradient in bothclimate and structure, many other variables aresimultaneously changing and correlated along thegradient including available moisture, energy bal-ance and radiative transfer, distribution and depthof permafrost (permanently frozen ground), bio-mass and production, and others. Changes invegetation structure and composition along this

gradient do not follow a classic life-zone patternof movement up and down an altitudinal gradient.Instead, trees and shrubs first come into the land-scape at midslope positions (fig. 2) (Van Cleveand Viereck 1981). The presence of shallowpermafrost results in a sealed hydrologic systemand drainage conditions unsuitable for tree growthat lower slope positions. The exception is blackspruce, Picea mariana, which grows in openstands in permafrosted lowlands or on north-fac-ing slopes.

The introduction of trees into the landscape hasan overridingly important influence on the spatialand temporal domains of the systems (fig. 3). Themain differences between the two systems are:

1. Insects and herbivores have a more consis-tent and extensive role in the boreal system.

2. The roles of fire and flooding are out of scalerelative to other disturbances in the borealsystem, with fire being the more significant.

3. The formation and melting of ice wedges isa significant disturbance in the Arctic,whereas in the current boreal climate, themain effect is surface melting of permafrost(thermokarst), with less recovery. Ice wedgeformation is on a scale small enough to beconsidered insignificant as a landscape-forming process.

4. The current timber harvest regime is quitesmall relative to other stand-replacing distur-bances. However, new technologies andeconomies in fiber harvest may result in avery different regime in the near future.

5. Patterns of eolian loess distribution in thesystems differ because of an interaction withtopography, vegetation, and wind regime.Arctic rivers follow broad, often anastomosedpathways with large expanses of barren ornearly barren silt and sand bars. Nearly con-stant wind redistributes this loess across thelandscape, with major influence on patternsof vegetation composition and productivity(Walker and Everett 1991, Walker et al.1998). The Alaska boreal forest has little windcompared to the Arctic, and although thereare active sand and silt bars associated withmedium and large rivers, much of what isblown off is effectively trapped by the riparianand flood-plain vegetation.

50

Figure 1—Major ecosystems of Alaska. Source: Markon, C. 1991. US Geological Survey digital map of the major ecosystemsof Alaska. Based on the map unit boundaries delineated by the Joint Federal-State Land Use Planning Commission forAlaska in 1973. Scale 1: 2,500,000.

51

Table 1–Structural characteristics of major arctic-boreal vegetation and ecosystem zones

Mean JulyZone Vegetation structure temperature

Polar desert Mainly herbaceous or nonvascular, discontinuous scattered plants. 1-3 °CMosses, when present, limited to acrocarpous species or smallhummocks of individual clones.

High arctic Mainly herbaceous (Graminae and Cyperaceae); woody species in 4-6 °Cgenera Salix and Dryas with woody sections limited to below groundor 5 cm above ground, areas of semicontinuous cover in protectedareas. Increasing diversity of pleurocarpous mosses in small clustersof plants.

Low arctic Codominated by herbaceous (mainly graminoids of Cyperaceae) and 7-9 °Cwoody species in the Ericaceae family and in genera Salix, Dryas, andBetula. Woody sections to as high as 40 cm above ground. Continuouscover. Development of full moss and lichen carpet underlying vascularspecies with high nonvascular species diversity. Sphagnum spp. dominatemoss carpets in certain regions.

Treeline or Tall shrubs from 1 to 2 m tall representing four to five families, with 10-12 °Cwoody tundra graminoids of Cyperaceae subdominant. Scattered islands of coniferous

trees, occasional stands of poplar (Populus spp.) or tree birches (Betulaspp.). Increased moss biomass and production.

Boreal forest Uplands completely dominated by fully developed forest, mainly of Picea, 12-15 °CBetula, Populus, and Larix. Graminoid-dominated stands of Cyperaceaelimited to areas of permafrost in lowlands. Moss biomass and productionextremely variable depending on forest type and condition, from none indeciduous upland forests to thickest development of carpets in open,permafrost lowlands.

52

Figure 2—Arrangement of major structural-dominance vegetation types along gradients ofmegascale climate and mesotopography (catenas) in Alaska and western Canada. (Latitude isshown for each location.) Symbols: P–polar desert, see table 1; H–heath, ericaceous androsaceous prostrate and dwarf shrubs with lichens; W–fen (wet meadow) dominated by species ofCarex (s) or grasses (g), mainly Dupontia fisheri R.Br.; T–tussock tundra, Eriophorum vaginatumL., Carex bigelowii Torr. ex Schwein. with low and prostrate shrubs; S–shrub tundra, species oferect Salix L., Betula L., and sometimes Alnus P. Mill. with an understory of prostrate shrubs,graminoids, and mosses; O–open woodland/muskeg, combination of sparse black spruce Piceamariana (P. Mill.) B.S.P. with tussock tundra; F–forest, mainly Picea glauca (Moench) Voss as wellas Populus tremuloides Michx. and Betula papyrifera Marsh.

53

The increased importance of fire and animals is adirect result of the presence of trees but also is afunction of climate. Climate sets an ultimate limiton tree growth and in some cases animal distribu-tion (particularly insects; Chapin and Danell2001), but the presence of trees also drives bothinsect and fire regimes. In the Arctic, patterns ofvegetation and major differences in ecosystem

function can be clearly linked to a series of mostlyabiotic controls (Walker et al. 1994, Walker et al.1998). Although the arctic biome in Alaska hasvaried significantly in spatial extent and appear-ance over the past million years, most of its floraformed in major glacial refugia and has sincebeen redistributed to form existing patterns(Hultén 1937). The Alaska boreal ecosystem, in

Figure 3—Spatial and temporal domains of the major landscape-forming natural disturbances inboreal and low arctic ecosystems, respectively (adapted in part from Walker and Walker 1991).Timber harvest is overlain on the boreal system to show its extent relative to other disturbances.

54

contrast, had its dominant species invade onlywithin the past 10,000 years, in a series of“waves” that represented almost instantaneousappearance over large areas (Anderson andBrubaker 1994, Hu et al. 1993). Existing patternswithin the Alaska boreal zone can be explainedalmost completely by a combination of mesotopo-graphy and succession following disturbance(Van Cleve et al. 1983, 1996).

The ability of a few species to play a dominantrole in ecosystems processes has major implica-tions for many management issues as well as forour ability to comprehend the potential for changein the system under changing climate. Becausearctic systems are controlled by fairly fixed setsof regimes, developing comprehensible modelsof landscape change related to climate changeshould be possible (Epstein et al. 2000). In theboreal system, however, the strong interactionbetween biota, climate, and disturbance regimemakes for a much more chaotic system (Rupp etal. 2000, Starfield and Chapin 1996). Getting firemanagement “right” is perhaps the single mostimportant management question in Alaska’s inte-rior region today. Fire has more than major eco-logical consequence in Alaska; fire fighting is amajor source of cash economy in many smallvillages.2 The Alaska Fire Service’s current policyis to control human-caused fires and to let naturalfires (started by lightning) burn unless theythreaten life or property.3 Their goal is to support anatural regime, with the idea that ultimately such aregime will maximize all major economic land-scape uses: wildlife habitat, recreation, and timberharvest. Because fire regime is so closely tied toboth climate and vegetation, it seems unlikely thatany policy other than the current one could bejustified by scientific evidence. A natural fire re-gime will forever be a moving target, however,and thus management goals need to follow theregime rather than dictate it.

Forest harvest in the Alaskan interior represents ascale issue mainly in the sense that current forestharvest is on a scale small enough that it can beconsidered ecologically insignificant. A closer ex-amination indicates this view to be somewhat of afalsehood, however. Most harvest is on privatelands owned by regional corporations. Forestindustries on these lands are not economicallyviable from a market perspective but are subsi-dized as a way of getting local timber and fire-wood and providing employment in small villages.4

Harvest on public lands is primarily within theTanana Valley State Forest and to a lesser degreeon Bureau of Land Management lands. Currentharvest rates on these lands are at about 800acres per year, and many offered sales are notpurchased (Crimp et al. 1997). Forest harvest is apublic concern, however, for the following rea-sons: (1) Sustainability of the harvest resource,mature white spruce stands, is little understoodand highly variable in time and space. There hasbeen no systematic analysis of forest practicesover the last 30 to 50 years to analyze changes incomposition following harvest, but there is someevidence that certain harvest practices may resultin a long-term dominance of deciduous species,an undesirable management outcome if whitespruce harvest is to be sustained. Natural reseed-ing and successful establishment of white spruceare dependent on timing of cone crops and herbi-vore population levels, both of which vary in time,creating an uncertainty that is beyond manage-ment possibility. (2) Most harvests are in highlyvisible areas near roads and along rivers, andthus engender public scrutiny and debate. Al-though the level of harvest is essentially invisiblerelative to large timber industries, it is of greatconcern locally. (3) The Alaska economy is almostcompletely dependent on natural resources, bothfrom extraction and tourism viewpoints. Changingglobal markets for fiber and new technologies thatmake fiber use more practical could potentiallylead to a regional fiber industry that would not beso dependent upon certain vegetation types.Thus, although the current timber industry effectis masked by the effects of fire, fiber extraction onlarge scales could change this. A firm ecological

2 Chapin, S.F., III. 2001. Personal communication. Professor,University of Alaska, Fairbanks, P.O. Box 757520, Fairbanks,AK 99775.

3 Jandt, R. 2001. Personal communication. Ecologist, AlaskaFire Service, Bureau of Land Management, P.O. Box 35005,Fort Wainwright, AK 99703.

4 Wurtz, T. 2001. Personal communication. Researchecologist, U.S. Department of Agriculture, Forest Service,Pacific Northwest Research Station, Boreal Ecology Coopera-tive Research Unit, P.O. Box 756780, University of AlaskaFairbanks, Fairbanks, AK 99775-6780.

55

understanding of how disturbances operate in thissystem in time and space is therefore essential toour ability to provide input on sustainable man-agement practices.

In conclusion, Alaska is considerably differentfrom other states in terms of having extensive,remote landscapes that are mostly undeveloped.Management of Alaska landscapes must considerthe interactions between the spatial and temporalscales of various disturbances. Both long-termdata collection and retrospective analysis areimportant tools for understanding the frequencyand importance of certain types of events. Manag-ing for sustainability must consider that the rangeof variability in key disturbances will rarely matchwith the affected spaces and timeframes requiredfor most management decisions.

Metric Equivalents

1 acre = 0.405 hectares

Degrees Fahrenheit = 1.8 °Celsius + 32

Literature Cited

Alexandrova, V.D. 1980. The Arctic and Antarctic:their division into geobotanical areas. Cam-bridge, United Kingdom: Cambridge UniversityPress. 247 p.

Allen, T.F.H.; Hoekstra, T.W. 1992. Toward aunified ecology. New York: Columbia UniversityPress. 384 p.

Allen, T.F.H.; Starr, T.B. 1982. Hierarchy: per-spectives for ecological complexity. Chicago,IL: University of Chicago Press. 394 p.

Anderson, P.M.; Brubaker, L.B. 1994. Vegeta-tion history of northcentral Alaska: a mappedsummary of late-Quaternary pollen data. Qua-ternary Science Reviews. 13: 71-92.

Barber, V.A.; Juday, G.P.; Finney, B.P. 1998.Tree-ring data for climate reconstruction andindication of twentieth century warming andsummer drying. Paleotimes. 6: 20-22.

Chapin, F.S.; Danell, K. 2001. Scenarios ofbiodiversity change in the boreal forest. In:Chapin, F.S.; Sala, O.; Huber-Sannwald, E.,eds. Global biodiversity in a changing environ-ment: scenarios for the 21st century. New York:Springer-Verlag: 101-120.

Chernov, Yu.I.; Matveyeva, N.V. 1997. Arcticecosystems in Russia. In: Wielgolaski, F.E.,ed. Polar and alpine tundra. Amsterdam, TheNetherlands: Elsevier: 361-507.

Crimp, P.M.; Phillips, S.J.; Worum, G.T. 1997.Timber resources on state forestry lands in theTanana Valley. Fairbanks, AK: Alaska Depart-ment of Natural Resources, Division of For-estry. 80 p.

Delcourt, H.R.; Delcourt, P.A.; Webb, T., III.1983. Dynamic plant ecology: the spectrum ofvegetational change in space and time. Qua-ternary Science Reviews. 1: 153-175.

Epstein, H.A.; Walker, M.D.; Chapin, F.S., III;Starfield, A.M. 2000. A transient, nutrient-based competition model of arctic vegetation.Ecological Applications. 10(3): 824-941.

Franklin, J.F. 1987. Importance and justificationof long-term studies in ecology. In: Likens,G.E., ed. Long-term studies in ecology: ap-proaches and alternatives. New York: Springer-Verlag: 3-19.

Hammond, T.; Yarie, J. 1996. Spatial predictionof climatic state factor regions in Alaska.Ecoscience. 3: 490-501.

Hu, F.S.; Brubaker, L.B.; Anderson, P.M. 1993.A 12,000-year record of vegetation change andsoil development from Wien Lake, centralAlaska. Canadian Journal of Botany. 71: 1133-1142.

Hultén, E. 1937. Outline of the history of arcticand boreal biota during the Quaternary period:their evolution during and after the glacial pe-riod as indicated by the equiformal progressiveareas of present plant species. Lehre, Ger-many: J. Cramer Publishing. 168 p.

Jorgenson, M.T.; Roth, J.; Raynolds, M. [et al.].1999. An ecological land survey for Fort Wain-wright, Alaska. Report 99-9. Hanover, NH: U.S.Army Cold Regions Research and EngineeringLab. 83 p.

Jorgenson, M.T.; Roth, J.; Smith, M.D. [et al.].2001. An ecological land survey for FortGreely, Alaska. Rep. TR-01-4. Hanover, NH:U.S. Army Cold Regions Research and Engi-neering Lab. 85 p.

56

Juday, G.P.; Ott, R.A.; Valentine, D.W.; Barber,V.A. 1998. Forests, climate stress, insects, andfire. In: Weller, G.; Anderson, P., eds. Implica-tions of global change in Alaska and the BeringSea region. Fairbanks, AK: Center for GlobalChange and Arctic System Research: 23-49.

O’Neill, R.V.; Deangelis, D.L.; Waide, J.B.;Allen, T.F.H. 1986. A hierarchical concept ofecosystems. Monographs in Population Biol-ogy 23. Princeton, NJ: Princeton UniversityPress. 262 p.

Rupp, T.S.; Starfield, A.M.; Chapin, F.S., III.2000. A frame-based spatially explicit modelof subarctic vegetation response to climaticchange: comparison with a point model. Land-scape Ecology. 15: 383-400.

Serreze, M.C.; Walsh, J.E.; Chapin, F.S., III[et al.]. 2000. Observational evidence of re-cent change in the northern high-latitude envi-ronment. Climatic Change. 46(1/2): 159-206.

Starfield, A.M.; Chapin, F.S., III. 1996. Model oftransient changes in arctic and boreal vegeta-tion in response to climate and land usechange. Ecological Applications. 6: 842-864.

Van Cleve, K.; Dyrness, C.T.; Viereck, L.A.[et al.]. 1983. Taiga ecosystems in interiorAlaska. Bioscience. 33(1): 39-44.

Van Cleve, K.; Viereck, L.A. 1981. Forest suc-cession in relation to nutrient cycling in theboreal forest of Alaska. In: West, D.; Shugart,H.; Botkin, D., eds. Forest succession: con-cepts and application. New York: Springer-Verlag: 185-211.

Van Cleve, K.; Viereck, L.A.; Dyrness, C.T.1996. State factor control of soils and forestsuccession along the Tanana River in interiorAlaska, USA. Arctic and Alpine Research.28(3): 388-400.

Walker, D.A. 2000. Hierarchical subdivision ofarctic tundra based on vegetation response toclimate, parent material, and topography. Glo-bal Change Biology. 6: 19-34.

Walker, D.A.; Auerbach, N.A.; Bockheim, J.G.[et al.]. 1998. Energy and trace-gas fluxesacross a soil pH boundary in the Arctic. Nature.394: 469-472.

Walker, D.A.; Everett, K.R. 1991. Loess ecosys-tems of northern Alaska: regional gradient andtoposequence at Prudhoe Bay. EcologicalMonographs. 61: 437-464.

Walker, D.A.; Walker, M.D. 1991. History andpattern of disturbance in Alaskan arctic terres-trial ecosystems: a hierarchical approach toanalyzing landscape change. Journal of Ap-plied Ecology. 28: 244-276.

Walker, M.D. 1995. Patterns of arctic plant com-munity diversity. In: Chapin, F.S., III; Korner,C., eds. Arctic and alpine biodiversity: patterns,causes, and ecosystem consequences. Eco-logical Studies. 113: 1-18.

Walker, M.D.; Gould, W.A.; Chapin, F.S., III.2001. Scenarios of biodiversity changes inarctic and alpine tundra. In: Chapin, F.S., III;Sala, O.; Huber-Sannwald, E., eds. Globalbiodiversity in a changing environment: sce-narios for the 21st century. New York: Springer-Verlag: 83-100.

Walker, M.D.; Walker, D.A.; Auerbach, N.A.1994. Plant communities of a tussock tundralandscape in the Brooks Range foothills,Alaska. Journal of Vegetation Science.5: 843-866.

57

In this presentation, we describe perspectivesabout landscapes and scales of analysis appropri-ate to understanding the nature of relationshipsbetween people and the environment of whichthey are a part. We also discuss how to improveintegration of human and biophysical consider-ations in the study and management of land-scapes. Given the purpose of this conference, thepaper is not in the format of a formal review orsynthesis of the literature. This paper reflects theexperience of the authors whose work has collec-tively spanned several decades and a variety oftopics related to people and natural resource in-teractions. Our observations might best be con-sidered as propositions to aid discussion ratherthan scientific conclusions or testable hypotheses.

Landscapes Are in the Eye of theBeholder

Landscapes come in many sizes. There is no oneright definition for what a landscape is (or what itmeans to different people) or the scale(s) appro-priate for understanding relationships betweenhumans and natural resources. Various needsand questions will define the appropriateness oflandscape meanings and scales of analysis.Sometimes these needs are defined by scientists,other times by forest managers, and yet other

times by citizens. Technical definitions are impor-tant for technical analyses but not necessarilyimportant to everyone; they are a means to often-unclear ends. The challenge is to understand theneeds and questions to be addressed before linesare drawn on maps and data collection started.Such problem framing is the most important, yetleast well-done step, particularly if all interestedparties are not included up front.

In a sense, the meanings that landscapes holdare constructed by those viewing the landscape orinteracting with it in other ways. Each meaning isdifferent, not better or worse. Full understandingof the values and meanings landscapes producerequires that analyses be inclusive of the peoplewho interact with the landscape. The implicationof different types of meanings is that conflict islikely to occur where such meanings collide.

People Think and Act at MultipleScales for Many Reasons

No one way to divide time and space will accountfor the multiple values, concerns, and uses thatpeople bring to the understanding of natural re-sources. Some ways to think about how land-scapes can be considered from a social scienceperspective are summarized below.

Many values are important to people. Manyvalues are important to people as they think aboutand use forests and other landscapes. Theseinclude a variety of commodity, public use, amen-ity, environmental quality, spiritual, and healthvalues. Such meanings are attached to land-scapes by different types of people and at differ-ent scales. For example, recreation can bethought of as people using microsites such ascampsites or as driving for pleasure across largerlandscapes. And the array of values often blendsbiophysical, economic, and social domains indifferent combinations across space, people, andtime. This means that to understand the meaningand importance of these values, expertise beyondthe biophysical sciences is required.

Landscape Perspectives

Roger N. Clark,1 Linda E. Kruger,2 Stephen F. McCool,3 and George H. Stankey4

1 Research social scientist, U.S. Department Agriculture,Forest Service, Pacific Northwest Research Station, ForestrySciences Laboratory, 400 N 34th Street, Suite 201, Seattle,WA98103; Tel: 206-732-7833; e-mail: rnclark@fs.fed.us.

2 Research social scientist, U.S. Department Agriculture,Forest Service, Pacific Northwest Research Station, ForestrySciences Laboratory, 400 N 34th Street, Suite 201, Seattle,WA98103; Tel: 206-732-7832; e-mail: lkruger@fs.fed.us.

3 Professor, School of Forestry, University of Montana,Missoula, MT 59812; Tel: 406-243-5406; e-mail:smccool@forestry.umt.edu.

4 Research social scientist, U.S. Department Agriculture,Forest Service, Pacific Northwest Research Station, ForestrySciences Laboratory, Corvallis, OR 97331; Tel: 541-737-1496; e-mail: ghstankey@fs.fed.us.

58

People organize in many ways. There are avariety of ways to think about how people (indi-viduals) are combined at different scales and howa social organizational hierarchy can be de-scribed. These include individuals, family andhousehold groups, neighborhoods, communities,counties/boroughs, states/provinces, nations, andultimately, the globe. The interests (political, eco-nomic, cultural) people hold in the landscape atdifferent scales and the decisions they makeabout how they interact with it may cut acrossthese different levels. Each level or scale is char-acterized by different emergent properties, suchthat the next higher scale is not simply an aggre-gation of units. There are often mismatches be-tween these organizational units and biophysicalscales that need to be reconciled before anyanalysis begins if an integrated solution is de-sired. Putting the pieces together later has provento be difficult.

People act at multiple spatial scales. Theseinclude microsites, areas (e.g., a grove of trees,meadows), drainages, watersheds, landscapes(e.g., “the valley”), regions (e.g., Puget Sound),continents, and the globe. It is important to con-sider such ways of defining scales because differ-ent social, cultural, and institutional propertiesmay emerge at each scale. Appropriate scalesmay be defined by the processes at work, interac-tions within and between components of complexbiophysical and social systems, and policy or sci-

entific needs. Meanings cannot be aggregatedupward; people may define an entire watershedas a suitable place for timber harvesting, yet holdclaims to spiritual, aesthetic, and recreationalmeanings at the site level.

Human lives and activities consider multipletemporal scales. Ways of defining time includethe past, today, tomorrow, weeks, seasons, years,decades, and generations. These may or may notcoincide with how time is considered by special-ists concerned with biophysical phenomena. Dif-ferences between biological and social scales ofsignificance are frequently at the root of conflict—such as when forest plans are considered over a50-year timeframe but budgets are appropriatedannually. Considerations of time often influencepublic acceptance of forest management prac-tices.

We must beware of the ecological fallacywhen drawing conclusions about people.What may be true at a higher scale, such as thecounty level, may not be so at lower scales, suchas the communities in the county; the attributes ofa transportation system may not apply to the indi-vidual roads within; qualities of a dispersed recre-ation area may differ when one looks at specificsites; and the distribution of meanings across alandscape cannot necessarily be summed to ar-rive at an overall assignment of landscape mean-ings. Looking up and down this sort of scale is not

Pho

to b

y R

oger

Cla

rk

59

simply an aggregation or disaggregation issue. Itis likely in many cases that different processeswork at different scales. The perspectives peoplehave when they think at different scales influencejudgments about the appropriateness and accept-ability of change. And what may be acceptable atone scale may not be so at another.

Human habitats are definable. Concepts thathelp explain why some wildlife species behave asthey do also help explain the behavior of people.For example, we have found in studies ofrecreationists that home ranges exist for bothresidents and tourists (a migratory species!).Edges, particularly along water, affect use. Criticalhabitats are definable and include attributes thatrelate to many temporal and spatial scales. Hu-man habitats—for recreation, spiritual, and aes-thetic purposes—are dynamic in time and space.Travel corridors and how they are accessed influ-ence patterns of use. Place attributes and mean-ings (at multiple scales) influence choices peoplemake. However, the meanings people attach to

specific places, and how they define their criticalhabitat needs, often are not correlated with certaintypes of biogeographical features mapped bybiologists. These concepts can help us thinkabout how people relate to landscapes at multiplescales and improve our ability to understand theeffects of policy options and management prac-tices on existing and potential public values anduses.

Corridors channel energy. Natural topographyand features and human-created corridors chan-nel air, water, critters, and people. The intersec-tions of corridors (water crossings, powercorridors, dams) or flows within them often revealconflicts and compatibilities between public valuesand uses and other resource values.

Acceptance of change varies both in time andspace. Judgments about the acceptability ofchanges depend on their nature, extent, cause,and location with respect to the meanings andvalues people attach to specific places and land-scapes.

Pho

to b

y Li

nda

Kru

ger

60

Problem Framing Is the Key toSuccess

It is important not to “start” until one gets theissues and questions right. This usually meanswe need to step back from individual, disciplinarydefinitions and join with other “–ologies” and in-terests. We need to be sure we are not solving“solutions” or solving the “wrong” problem. Wemust learn from one another about how we “see”and define landscapes so that we can jointly con-struct opportunities and redefine problems andissues and then develop explicit questions todrive joint actions. Action in today’s society re-quires multiple actors, and joint agreement onproblems and opportunities is needed beforeeffective action can be implemented. It is impor-tant to determine where questions require “broadand integrated” as well as “narrow and deep” ap-proaches. Picking the wrong boundaries andscales is inevitable unless driven by questions.Because integrated approaches require a fullrange of interests, types of knowledge, and pointsof view, such approaches will challenge the powerand authority normally accorded to experts andspecialists.

Problem Framing and Resolution MustInclude Various Value Systems

Because landscape values and meanings differby time, scale, and individual or group, there is no“correct” definition. Although this suggests thatdiversity may be an obstacle, it may be an oppor-tunity as well. What can unite us is recognition ofthe power of both individual and collective per-spectives. Although we “look at” the same placeswe “see” them differently, so to really understandthem we need to look at them together and de-scribe what we see and why. But this is not aneasy task. Planning processes that are inclusiveraise the possibility of increased understanding,improved representativeness in public participa-tion, an opportunity to learn, and eventually identi-fying ways to get to a desired future.

What’s Getting in the Way?

As we think about and conduct analyses of land-scapes from multiple perspectives, it becomesevident that to develop a more holistic approach isdifficult. A number of things make designing andimplementing integrated approaches challenging.

Ideologies and beliefs (worldviews) condition howwe think. The common focus on getting answersand solutions before clarifying the questions andproblems often leads us in the wrong direction.The language of science and expertise makes itdifficult for interested citizens to easily engage inprocesses and activities that affect them and tocontribute the knowledge they have about thecomplex systems we all value. Technology can beboth a means and an end but often drives us indirections that later prove to be less than effec-tive. For example, more and better geographicinformation system technology can be a hin-drance if the wrong questions are under study.

Conclusions

All landscape definitions are human conventions–they do not “exist” in any objective sense. Wecarry expectations and definitions with us. Andthese expectations and definitions are dynamic.What often divides diverse interests and needsare assertions that there is one best way to de-scribe landscapes and scales.

Integrated multiscale approaches will link humanswith biophysical landscapes. But social and bio-physical considerations often are not in sync atleast initially. Improved problem framing beforeany action to collect data will improve the integra-tion of human consideration, values, and uses intolandscape analyses.

People form strong opinions about places and thecharacteristics of places at multiple scales. Theyalso are concerned about the appropriateness ofresource management uses (in time and space).Legacies on the land affect judgments in a differ-ent way. It is hard for some people to relate tolarge landscapes when they are concerned aboutfavorite places.

So, landscapes are socially defined in multipleways for multiple purposes from multiple perspec-tives. They are complex—socially and biophysi-cally. These definitions are dynamic. Landscapesat multiple scales can be used as a means tounite and understand interactions among bio-physical and social systems.

We need to take a more holistic view when think-ing about landscapes. Boundaries matter—theydefine time and space for many relevant pur-poses. The trick is to look across boundaries onthe map and in the mind.

61

Abstract

Forest managers and researchers are now working at a landscape level, but is our public with us? Formost people, landscape-level management is not a clear concept. To better understand how citizensview landscapes, we need to go beyond attempts to “educate the public” and instead interact with citi-zens to promote learning, find appropriate outreach activities and simulation techniques, and directlyaddress questions about risk and uncertainty. Various scales of analysis will be necessary, including thegeographic, temporal, and normative contexts and their relevance to citizens.

Landscape-Level Management: It’s All About Context1

Bruce Shindler2

1 Shindler, B. 2000. Landscape-level management: it’s allabout context. Journal of Forestry. 98 (12): 10-14. Reprintedwith permission from the Society of American Foresters.

2 Associate professor, Department of Forest Resources,Oregon State University, Corvallis, OR 97331; Tel: 541-737-3299; e-mail: Bruce.Shindler@orst.edu.

62

Introduction

The purpose of this paper is to describe recentand ongoing work that contributes to understand-ing social and economic change at broad scales.In particular, the paper focuses on a recently re-leased atlas on human adaptation to environmen-tal change at the county level. The paper alsodiscusses issues pertaining to data availabilityand geographic and temporal scales.

Issues of scale are important to understandingsocial and economic change. Two important con-siderations affect social and economic analysesat the broad scale. First, there is the wealth ofsecondary data collected by an array of federal,state, and, in some cases, local governmentagencies (Salant and Waller 1995). Frequentlythese data sets provide information for specificvariables across multiple scales and on a rela-tively consistent basis for extended periods oftime. Examples include the decennial censusinformation that is available from individual cen-sus “blocks” or “block groups” and additive to thecounty, state, and national levels and the employ-ment information collected at the firm level byemployment security departments and provided ata range of geographic and industrial sector scalesand levels of detail. The second consideration isthat most of this secondary information is avail-able only at scales defined by political units of

government. Physical and biological effects oftenare relevant only indirectly as they have shapedthe boundaries of government entities such ascities, counties, and states. Understanding therelevant secondary data available is a prerequisiteto primary data collection efforts.

Winnowing this mass of information on social andeconomic variables and selecting the appropriatedata to address the research or policy issues athand is a formidable task. Also, there is the needto provide meaningful tabular, graphical, and geo-graphical displays of this information as an aid tounderstanding and interpretation. It is in this arenathat the work of the Rural Communities andEconomies Team of the USDA Forest Service’sPacific Northwest Research Station has beenconcentrated. A related issue of interest is howdifferent scales of the social and economic vari-ables are hierarchically related and how theserelationships impact the interpretation usefulnessof the variables. Team work has focused on thesethree areas: (1) the collection of data on key so-cial and economic variables for the Pacific North-west; (2) the development of geographical dis-plays (atlases) that include interpretations ofthese variables (Christensen et al. 2000, Raettiget al. 2001); and (3) the exploration of theoreticalissues concerning social and economic variablesand the linkages between different spatial andtemporal scales.4

Recently in the West, several interagency,multidisciplinary, broad-scale assessments wereconducted in response to, or in anticipation of,changes in resource management. These as-sessments include the Forest Ecosystem Man-agement Assessment Team (FEMAT 1993), theSierra Nevada Ecosystem Project (SNEP 1996),

Human Adaptation to Environmental Change: Implications atMultiple Scales

Ellen M. Donoghue,1 Terry L. Raettig,2 and Harriet H. Christensen3

1 Research social scientist, U.S. Department of Agriculture,Forest Service, Pacific Northwest Research Station, ForestrySciences Laboratory, P.O. Box 3890, Portland, OR 97208-3890; Tel: 503-808-2018; e-mail: edonoghue@fs.fed.us.

2 Economist, U.S. Department of Agriculture, Forest Service,Olympic National Forest, 1835 Black Lake Blvd. SW,Olympia, WA 98512-5623; Tel: 360-956-2449; e-mail:traettig@fs.fed.us.

3 Research social scientist (retired), U.S. Department ofAgriculture, Forest Service, Pacific Northwest ResearchStation, Forestry Sciences Laboratory, 400 N 34th, Suite 201,Seattle, WA 98103; Tel: 206-732-7824; e-mail:chchristensen@fs.fed.us.

4 Raettig, T.L. [N.d]. Spatial and temporal scale issues inNorthwest Forest Plan monitoring and evaluation activities.Manuscript under review. On file with: USDA Forest Service,Pacific Northwest Research Station, Forestry SciencesLaboratory, P.O. Box 3890, Portland, OR 97208.

63

and the Interior Columbia Basin EcosystemManagement Project (Quigley and Arbelbide1997). Social assessments at the county andcommunity level were conducted as part of thesebioregional assessments (Doak and Kusel 1997,Harris et al. 2000, Horne and Haynes 1999).Many dimensions of human life were reported onincluding culture, the history of human-naturalresource interactions, socioeconomic conditionsand trends, and measures of community resil-iency and community capacity. This work expandson these efforts, by depicting other variables overdifferent geographic and spatial scales. It alsocontributes to meeting the requirements for socialand economic monitoring and forest planning. Forinstance, in the Northwest Forest Plan (NWFP)region, the record of decision for the NWFP di-rects agencies to conduct three types of monitor-ing—implementation, effectiveness, andvalidation—to detect desirable and undesirablechanges as a result of ecosystem management(USDA and USDI 1994). Planning rules for Na-tional Forest System land also have requirementsfor social and economic sustainability (see for ex-ample, USDA Forest Service 1999). Finally, thework reflects appropriate roles for scientists inassessments, analyses, and monitoring activitiesassociated with forest management.

Data Collection

The atlases focus on county-level information. Formany of the social and economic indicators avail-able from secondary sources (except for the de-cennial census), the county is the smallest unit forwhich information is disseminated or published. Inthe case of the commonly used economic vari-ables such as employment and income, informa-tion cannot be disclosed or published that can betraced to an individual firm or employee (Wash-ington Employment Security Department 1997).For other variables, such as those collected in theCurrent Population Survey (CPS), small samplesizes may preclude the calculation of data forsmaller geographical areas. County-level data canbe easily aggregated into larger scale areas suchas planning regions, states, and even multistateregions. They also provide useful context for so-cioeconomic assessments at smaller scales, al-though in and of themselves may not be adequateproxies for conditions and trends at smaller scales(Beckley 1998).

An important consideration in collecting second-ary data is the specific definition for the variable ofinterest. Employment information, for instance, isbased on place of employment in the employmentsecurity and regional economic information sys-tem, but is based on place of employee residencefor the decennial census and CPS. Temporalscale also is a consideration. Employment Secu-rity Department employment and wage informa-tion is often available on a monthly as well asannual basis. The decennial census collects infor-mation only once per decade with estimates pro-vided for key census variables in interveningyears.

The key economic variables depicted in the atlasinclude employment, income, unemployment, andeconomic diversity. The key social variables in-clude population (and characteristics of the popu-lation, such as race and ethnicity), migration,education, poverty, and crime. These variablesreflect key indicators that are used in social andeconomic assessments and studies on socioeco-nomic well-being. In addition to spatial and tempo-ral scale issues, the atlases focus on absolute,trend, and relative change in the values of thevariables. An example has been the documenta-tion and interpretation of how different sectors ofthe wood products industry have been impactedby changes in timber harvest in the Pacific North-west (Raettig and McGinnis 1996).

For the Pacific Northwest, work also has begunon the collection of community-level data incorpo-rating the secondary data that are available andmay ultimately include the collection of selectedprimary data items (Donoghue 2003). The social,economic, and resource data, such as timberharvests, assembled by the team have providedthe basic information in an analysis of the North-west Economic Adjustment Initiative (Raettig andChristensen 1999). The data and analytical workalso have been resources for continuing effortsrelated to Northwest Forest Plan social and eco-nomic monitoring.

Atlases

The Atlas Of Human Adaptation to EnvironmentalChange, Challenge, and Opportunity for the Pa-cific Northwest (Christensen et al. 2000), and theAtlas of Social and Economic Conditions andChange In Southern California (Raettig et al.2001) provide geographical expression of change

64

in social, economic, and timber-related variablesat the county level. The Pacific Northwest atlaswas designed to provide Northwest Forest Planmanagers and planners with a comprehensivereference depicting the dimensions, location,magnitude, and direction of social and economicchange in the Northwest Forest Plan region. In-cluded are displays that show patterns of the so-cial and economic diversity and health of theregion. This atlas focuses on change during theperiod of timber harvest reduction and initial eco-nomic restructuring in the region. Most maps arebased on changes in variables between 1989 and1994.

The first section of the Pacific Northwest atlas(Christensen et al. 2000) contains five base mapsthat depict the region, the Northwest Forest Planprovinces, counties, population distribution, majortransportation routes, and the public land owner-ships. Six sections follow that address:

1. What kinds of social and economic changeshave taken place in the face of reduced tim-ber harvest? Are Pacific Northwest communi-ties changing? If so, how?

2. What changes have occurred in the timberindustry since 1990? Has timber employmentchanged? Is private harvest increasing?

3. Have changes in federal harvest had a signifi-cant effect on county revenues?

4. Are western Oregon and Washington andnorthern California singularly dependent onnatural resources?

5. What federal assistance has aided cities andrural areas?

6. How have the population characteristicschanged? What are the trends in migration,educational attainment, and changes inethnicity?

7. What have been the changes in selectedsocial issues such as rates of poverty, prop-erty and violent crimes, and alcohol-relatedincidences?

Specific maps displayed in these sections:

• Section 2: Change in timber harvest and woodproducts employment

Change in public timber harvestChange in public and private timberharvest

Change in wood products employmentFederal lands-related payments tocounties

• Section 3: Change in economic conditions:economic performance

Unemployment rate compared toregion

Change in total employmentWage trendsWage level

• Section 4: Structural economic change

Economic diversityFastest growing nonfarm industriesSlowest growing nonfarm industries

• Section 5: Characteristics of the population

Change in county populationChange in unincorporated andincorporated population growth

Migration status and trendsChange in ethnicityEducational attainment

• Section 6: Social issues

Change in income maintenanceChange in poverty rateChanges in violent crimeChanges in property crimeAlcohol-related incidences

• Section 7: Federal assistance

Northwest Economic AdjustmentInitiative

65

Accompanying the maps are text, supportinggraphics, and tabular displays that interpret theinformation and patterns on the maps, explorethe relevant issues in more depth, and providea working tool for managers and planners.

A similar atlas was prepared as part of the socio-economic assessment for the Southern CaliforniaConservation Strategy, a broad-scale planningeffort for the four national forests in southern Cali-fornia (Raettig et al. 2001). This atlas covers 26counties in southern California and includes mapsfor the same variables displayed for the NorthwestForest Plan region, except timber harvest infor-mation was not displayed owing to the relativeinsignificance of timber harvest on the four na-tional forests of that region. Both atlases are avail-able on CD-ROMs that include some interactivefeatures.

Theoretical Issues and the LinkagesBetween Different Spatial andTemporal Scales

If social and economic variables are to be usefulfor managers and planners, a thorough under-standing is necessary of the implications of theinformation displayed and the relationship be-tween decision processes and different scales ofthe variables. Work has been done exploring therelationships between individual and aggregatedata (the “ecological fallacy”) (King 1997, McCooland Troy 1998) in political science and other so-cial sciences. Still, other work has explored theissues of spatial data, unit definition, and scale ingeographic and geographic information systems-related problems (Fotheringham and Wong 1991,Longley and Batty 1997). The Rural Economiesand Community Team has work in progress thatwill produce a synthesis of the theoretical workthat has been done in these areas (see footnote4). Ultimately this theoretical work will provide thebasis for understanding scale issues and howthese issues impact the use and interpretation ofsocial and economic variables in the planning,monitoring, and management settings that arepart of the ecosystem management processesnow being implemented.

Literature Cited

Beckley, T.M. 1998. The nestedness of forestdependence: a conceptual framework andempirical exploration. Society and Natural Re-sources. 11: 101-120.

Christensen, H.H.; McGinnis, W.J.; Raettig,T.L.; Donoghue, E. 2000. Atlas of human ad-aptation to environmental change, challenge,and opportunity: northern California, westernOregon, and western Washington. Gen. Tech.Rep. PNW-GTR-478. Portland, OR: U.S. De-partment of Agriculture, Forest Service, PacificNorthwest Research Station. 66 p.

Doak, S.; Kusel, J. 1997. Well-being assessmentof communities in the Klamath region.Taylorsville, CA: Forest Community Research;contract 43-91W8-6-7077. Report prepared forthe USDA Forest Service, Klamath NationalForest.

Donoghue, E.M. 2003. Delimiting communities inthe Pacific Northwest. Gen. Tech. Rep. PNW-GTR-570. Portland, OR: U.S. Department ofAgriculture, Forest Service, Pacific NorthwestResearch Station. 51 p.

Forest Ecosystem Management AssessmentTeam [FEMAT]. 1993. Forest ecosystem man-agement: an ecological, economic, and socialassessment. Portland, OR: U.S. Departmentof Agriculture; U.S. Department of the Interior[et al.]. [Irregular pagination].

Fotheringham, A.S.; Wong, D.W.S. 1991. Themodifiable areal unit problem in multivariatestatistical analysis. Environment and PlanningA. 23: 1025-1044.

Harris, C.C.; McLaughlin, W.; Brown, G.;Becker, D. 2000. Rural communities in theinland Northwest: an assessment of smallcommunities in the interior and upper Colum-bia River basins. Gen. Tech. Rep. PNW-GTR-477. Portland, OR: U.S. Department of Agri-culture, Forest Service, Pacific NorthwestResearch Station. 120 p.

66

Horne, A.L.; Haynes, R.W. 1999. Developingmeasures of socioeconomic resiliency in theinterior Columbia basin. Gen. Tech. Rep.PNW-GTR-453. Portland, OR: U.S. Depart-ment of Agriculture, Forest Service, PacificNorthwest Research Station. 41 p. (Quigley,T.M., ed.; Interior Columbia Basin EcosystemManagement Project: scientific assessment).

King, G. 1997. A solution to the ecological infer-ence problem: reconstructing individual behav-ior from aggregate data. Princeton, NJ:Princeton University Press. 342 p.

Longley, P.; Batty, M. 1997. Spatial analysis:modeling in a GIS environment. New York:John Wiley and Sons. 392 p.

McCool, S.F.; Troy, L.R. 1998. Estimating effectsof natural resource policy: the impact of theecological fallacy. RJVA 97-8016. Missoula,MT: U.S. Department of Agriculture, ForestService, Rocky Mountain Research Station.34 p.

Quigley, T.M.; Arbelbide, S.J., tech. eds. 1997.An assessment of ecosystem components inthe interior Columbia basin and portions of theKlamath and Great Basins. Gen. Tech. Rep.PNW-GTR-405. Portland, OR: U.S. Depart-ment of Agriculture, Forest Service, PacificNorthwest Research Station. 2066 p.

Raettig, T.L.; Christensen, H.H. 1999. Timberharvesting, processing and employment in theNorthwest Economic Adjustment Initiative re-gion; changes and economic assistance. GenTech. Rep. PNW-GTR-465. Portland, OR: U.S.Department of Agriculture, Forest Service,Pacific Northwest Research Station. 16 p.

Raettig, T.L.; Elmer, D.M.; Christensen, H.H.2001. Atlas of social and economic conditionsand change in southern California. Gen. Tech.Rep. PNW-GTR-516. Portland, OR: U.S. De-partment of Agriculture, Forest Service, PacificNorthwest Research Station. 66 p.

Raettig, T.L.; McGinnis, W.J. 1996. Wood prod-ucts employment in the Pacific Northwest:emerging trends. In: Proceedings of the 30thannual Pacific Northwest regional economicconference. Portland OR: Pacific NorthwestRegional Economic Conference and Center forUrban Studies, Portland State University: 45-50.

Salant, P.; Waller, A.J. 1995. Guide to rural data.Revised ed. Washington, DC: Island Press.140 p.

Sierra Nevada Ecosystem Project [SNEP].1996. SNEP final report to Congress: assess-ments and scientific basis for managementoptions. Davis, CA: University of California,Centers for Water and Wildland Resources.1528 p. Vol. 2.

U.S. Department of Agriculture, Forest Serv-ice. 1999. National Forest System land andresource management planning. Federal Reg-ister. 64(192): 54074-54112.

U.S. Department of Agriculture, Forest Serv-ice; U.S. Department of the Interior, Bureauof Land Management. 1994. Recordof decision for amendments to Forest Serviceand Bureau of Land Management planningdocuments within the range of the northernspotted owl. [Place of publication unknown].74 p. [plus attachment A: standards and guide-lines].

Washington Employment Security Depart-ment. 1997. User’s guide to labor marketinformation. Olympia, WA: Labor Market andEconomic Analysis Branch. 32 p.

67

Introduction

Given interactions among terrestrial, aquatic,and socioeconomic systems, planning at the land-scape level could benefit from including informa-tion from different scales. This paper offers ex-amples of broader systems views that may informnatural resource managers and policymakersdealing with restoring and maintaining healthyecosystems while addressing the demands ofpeople. World growth in human populations andincome has resulted in social, economic, andtechnological changes that profoundly affect theglobal management and use of natural resources.The world population will continue to grow, possi-bly from 5.9 billion in 1998 to 8.9 billion by 2050.The U.S. population also will continue to grow,particularly in the southern and western regions,with a projected increase of more than 120 millionpeople by 2050. Projected increases in populationand income will, in turn, increase demands forrenewable resources. Demographic shifts, suchas the aging of the population and increasing eth-nic and racial diversity, will affect the patterns ofdemand for natural resources.

Growing human populations and associated useand consumption of the Earth’s natural resourcesare placing more pressure on the chemical, bio-logical, and ecological systems of the planet.Human activities are significant drivers of environ-mental changes on the mosaic of land uses andland ownerships (e.g., Cohen et al. 2002). Analy-sis of the ecosystem response to changes inthese human-derived environmental factors iswarranted at multiple scales, depending on thepolicy-relevant questions. Changes in societalvalues are part of the dynamics affecting forestecosystems and influencing land management

strategies (e.g., Franklin 1989), warranting a moreexplicit treatment of the human system in con-junction with ecosystem properties. Integrating thehuman dimension more directly into analysis sys-tems is critical, as choices made by resource andland managers become increasingly important forthe long-term sustainable use of our natural re-sources (e.g., biodiversity, wildlife habitat, carbonsequestration) within the context of suitable hu-man welfare.

Examining the state of the land is a key part ofsustainability and livability analyses. Land usechanges can provide opportunities to help societyadjust to changing demands for and supplies ofrenewable resources from the Nation’s forest andaquatic ecosystems. Interfaces between landuses, such as riparian zones on forest and agri-cultural land, should be considered alongsidegrowing urban areas and increased fragmentationof some forests. In this paper, I examine drivers ofchanges in the large private forest land base, anddiscuss changes in land use and land cover bymajor categories.

Land Base Changes Over Time

From 1800 to 1930, forest land area in the UnitedStates declined by some 300 to 350 million acres(Clawson 1979), primarily in the East. This wasabout a one-third loss in forest cover. Some ofthis land was employed for urban and infrastruc-tural developments, but the largest portion wascleared and converted to agriculture. These landuse changes reflected federal policies of the timeto transfer the original public domain to privatehands and to expand agricultural production. Withthe closing of the public domain, establishment ofpermanent federal forest reserves, conversion ofmost suitable nongovernment forest lands tosome form of cropping or pasture, and dramaticimprovements in agricultural productivity, the netmovement of land between forestry and agricul-ture has become far less marked. Since 1950,

Land Use and Land Cover Dynamics: Changes inLandscapes Across Space and Time

Ralph J. Alig1

1 Team leader and research forester, U.S. Department ofAgriculture, Forest Service, Pacific Northwest ResearchStation, Forestry Sciences Laboratory, 3200 SW JeffersonWay, Corvallis, OR 97331; Tel: 541-750-7267; e-mail:ralig@fs.fed.us; Ralph.Alig@orst.edu.

68

U.S. forest land has declined about 4 percent,with the largest losses being to developed uses(Alig et al. 1999b).

Although the large amount of forest land clearingfor agriculture of the last century is probably aphenomenon of the past, continued expansion ofurban and developed areas is likely (e.g., Alig andHealy 1987). In 1994, forests covered 65 percentof nonfederal land in the Pacific Northwest, andthese forests are under increasing pressure ofconversion. Oregon’s and Washington’s popula-tions have increased faster than the national aver-age over the last decade. Most of the increasewas in the western part of the region, especiallyalong the I-5 corridor. For example, Portlandwas one of the fastest growing cities in the UnitedStates over the last decade. Projections are thatthe Pacific Northwest’s share of total populationwill increase, as the Nation’s population alsoexpands.

Urban sprawl and livability are issues receivingincreased attention around the Nation, as theNation becomes more urban, with more than 75percent of U.S. residents now living in cities(USDC Bureau of the Census 2002). In the Pa-cific Northwest (PNW), approximately 400,000acres of forest were converted to urban and de-veloped uses between 1982 and 1997 (USDANRCS 2001). Such changes can interact withother environmental alterations, such as habitatfragmentation, wetland loss, loss of biodiversity,and water pollution, and the cumulative impactson any particular ecosystem need to be consid-ered carefully.

Socioeconomic Trends

A number of socioeconomic trends affect optionsfor restoring and maintaining ecosystems, andaffect the supply and demand of land-basedgoods and services. Along with the move to cities,more and more people are moving to the Nation’scoasts. Some 53 percent of the U.S. populationlives in the 17 percent of the land in the coastalzone (within 50 miles of a coast). Further, thelargest population increases over the next severaldecades are projected to continue to be in coastalareas. Increases in population and personal in-come tend to lead to reductions in aggregate for-est area (e.g., Alig and Healy 1987). Such trends

need to be factored into landscape analyses,given implications for fish habitat, riparian areas,wetlands, and recreational opportunities.

In addition to an expanding population, the U.S.population is aging. Early in the 20th century, 1 in25 Americans was over age 65. In 2000, 1 in 8was over 65 (USDC Bureau of the Census 2002).Other demographic changes include continuingshifts in ethnic and racial compositions. Thechanging U.S. demographic composition likelywill impact demands for forest and range re-sources and how resources are managed (USDAForest Service 2001).

Americans’ incomes have grown substantially (inreal terms or constant dollars) in recent decades,and average per capita personal incomes in Or-egon have increased more than 75 percent overthe last 30 years. Projected increases in discre-tionary income will increase demands for renew-able resources, but may also lead to further con-version of forests for developed uses.

The United States in the last decade of the 20th

century was in a period of economic growth, asstock markets experienced relatively large gains.In the new millennium, wealthier nations are likelyto have a greater ability to accommodate restora-tion and maintenance of ecosystems than arethose nations with economies under seriousstress. However, some development patterns mayincrease vulnerability of land-based systems inspite of greater wealth. With more assets andinfrastructure located in coastal areas, this maycomplicate rehabilitation of ecosystems home toPacific salmon. Such trends also may increasevulnerability of coastal areas to pollution and toextreme events that are under study for any link-ages to global climate change (e.g., Easterlinget al. 2000).

The United States is also currently in the midst ofrapid changes and advancements in technologycapabilities in many economic sectors. The newinformation technology has affected agriculturalproduction and the forest sector to a lesser de-gree. Socioeconomic conditions will influencehow the technology is disseminated, and this willaffect how well our region and country can effectimprovements in the state of the land. With grow-ing and wealthier populations, development pres-sures and recreational use stresses on forests,wetlands, and riparian areas are likely to increase.

69

Types of Land Use and Land CoverDynamics

Five categories of change involving the land baseare (1) afforestation, (2) deforestation, (3) owner-ship, (4) forest type transitions, and (5) land man-agement intensity (Alig and Butler 2002, Alig et al.2002). Such changes arise from an interaction ofbiophysical, ecological, and socioeconomic pro-cesses and forces, often operating at a varietyof scales. For example, market forces tend tooperate at much larger scales than biophysicalprocesses commonly studied at microlevels suchas stands or reaches of a stream. I will provide abrief overview of the types of land base changes,with illustrations of temporal and spatial consider-ations drawn from empirically based studies. Theexample studies range in geographic scope fromthe Oregon Coast Range province (Spies, thisvolume) to the global level of analyses for currentinvestigations of climate change.

Afforestation

Afforestation is the forestation, either by humanor natural forces, of nonforest land. Two examplesare planting conifers on former cropland andformer Christmas tree plantings that are allowedto revert to “wild” forest. In the Pacific Northwest(Oregon and Washington), most afforested acreswere formerly in pasture and range use. Treeplanting is a major component of afforestation.Approximately 10 percent of the area of U.S.private tree planting in 1994 was in the PacificNorthwest in contrast to more than 80 percent inthe South. Most tree planting in the South was onnonindustrial private lands, whereas in the PNWregion, the majority was on forest industry lands.Nonfederal forest area in both Washington andOregon has steadily declined between 1982 and1997 surveys by the USDA Natural ResourcesConservation Service (2001), indicating that con-versions of forest to other uses, primarily devel-oped uses, have in net been larger than theamount afforested. The regional net loss hasbeen more than 300,000 acres of forest over that15-year period.

Tree planting has been a target of various poli-cies. The United States has a long history of for-est policies that jointly pursue both economic andecological objectives, often involving policy instru-ments designed to affect land use or cover. Publicpolicies aimed at one sector have impacts, either

expected or inadvertent, on another (e.g., forestsector vs. agriculture sector). In recent years poli-cies have increasingly focused on riparian zonesand incentives for promoting more favorableaquatic habitat for salmonid stocks. Many paststudies have examined policy impacts by either(1) ignoring spillovers in other sectors or (2) sim-ply “adding up” impacts across the two sectors,ignoring feedbacks or interactions through themarkets for land. At times, this also has includedignoring linkages between resource conditions onpublic and private lands, and associated intercon-nections via markets.

Deforestation

Deforestation is the conversion from forest tononforest use. Between 1982 and 1992, about250,000 acres were deforested on nonfederalland in the region. The annual rate of conversionof forests to urban/developed uses increased inthe 1990s to 32,000 acres, from the approxi-mately 25,000 acres in the 1980s (USDA NRCS2001), with stronger population growth and moreeconomic activity. In total, between 1982 and1997, the net outcome of land use shifts wasa regional loss in forest area of approximately400,000 acres. The destination of most convertedforested acres was to urban and developed uses.

Ownership Change

The main ownership change between traditionalownership categories has been between forestindustry (FI) and nonindustrial private forest(NIPF) owner classes. The FI land is owned bycompanies for the purpose of timber production(see for example, Zheng and Alig 1999). TheNIPF lands compose the remainder of the privateforest land base. Forest industry owns most of theprivate timberland in the PNW region, and gener-ally manages forest land intensively for timberproduction. Land exchanges among private own-ers have been more common than exchangesbetween private and public owners (Zheng andAlig 1999). For example, Forest Inventory andAnalysis (FIA) surveys in western Oregon indicatea net shift of 750,000 acres from NIPF to FI own-ers for 1961-94. Such changes in ownership canresult in notably different land management, par-ticularly as some private owners respond to mar-ket changes prompted by reductions in federaltimber supply.

70

With mergers, acquisitions, and other institu-tional changes, there also have been significantchanges within the industrial ownership category(e.g., combining Willamette and Weyerhaeusercompanies in 2002). The diverse NIPF ownercategory also continues to change in composi-tion. Along with this, forest fragmentation andparcelization and other factors may result in morenoncorporate individual owners (e.g., Sampson2000), but they are likely to have smaller tractsizes on average. Accompanying changes in tractsize may be changes in the values and percep-tions of forest landowners. New owners some-times bring different land management attitudescompared to traditional owners, with changingvalues often the result of exurbanization, or urbanresidents moving to rural settings.

Forest Type Transitions

Within the land base retained in forests over time,changes in forest cover types are caused by acombination of natural and human-related forces.Earlier literature dealing with forests focused onforest succession and natural disturbances, withemphasis on natural forces prompting forest typechanges. However, human-caused disturbancessuch as timber harvests now dominate in mostPNW forests (Alig et al. 2000, Cohen et al. 2002).Over the last FIA survey cycle, human-causeddisturbances were more frequent, by at least anorder of magnitude, than recorded natural distur-bances. The probability of harvesting on privatelands was about 2 to 10 times more likely thanfull- or partial-replacement disturbances from wild-fire, the primary natural disturbance agent of thepast. In the approximatly 10 years between FIAmeasurements, timber harvests affected about20 percent of private timberland, whereas fireaffected less than 1 percent.

Given this disturbance backdrop, clearcuttingDouglas-fir (Pseudotsuga menziesii (Mirb.)Franco) stands on FI lands resulted in 15 percentbeing replaced by hardwood (primarily red alder[Alnus rubra Bong.]) stands, and another 3 per-cent went to other types. Clearcutting red alderstands on FI lands resulted in 72 percent beingreplaced by other forest types. In contrast, only42 percent of red alder stands on NIPF landschanged forest types after clearcutting, reflectingfewer intentional type-conversion efforts after

harvest. Even for those stands not disturbed,12 percent of Douglas-fir stands on FI lands and23 percent on NIPF lands changed forest type.Such findings suggest that recognition of owner-ship differences can be important, beyond thepublic vs. private classifications sometimes used.

Land Management Intensity

Within a given forest type, investments in forestmanagement can significantly influence optionsfor different mixes of goods and services. Man-agement intensity includes the timing and type ofharvests, which have differed notably over time byownership in this region. Investments in improvedplanting stock on private land have allowed higherforest production per acre in some areas, lessen-ing pressures on other areas looked upon fornontimber goods and services. From a broadperspective, demands for different mixes of land-based goods and services are likely to shift overtime with a growing and wealthier population.Because some forest-based processes are de-cades or centuries in length, planning such invest-ments with adequate lead time is critical.

Assessments of Land Base Changes

The 2000 Resources Planning Act (RPA) assess-ment by the USDA Forest Service (2001) is de-signed to examine the current situation of theNation’s forest and rangeland ecosystems andassess the prospective situation over the next fivedecades. In support of that national assessment,research pertaining to drivers of land use changeat macroscales indicates that major determinantsof land use changes are population, personalincome, and potential profits from land enterprisessuch as forestry or agriculture (Alig et al. 2002).For example, the Forest and Agricultural SectorModel (Adams et al. 1996a, Alig et al. 1998) hasendogenous determination of land use changes.The linked model of the U.S. agriculture and for-est sectors has land demands and supplies inboth sectors that are endogenous and determinedsimultaneously with demands and supplies forproducts from those lands. One possible objectivefunction in the model is economic maximization(Adams et al. 1996b). Examination of land usechanges within a multisector context is importantfor assessing alternative futures of “sustainableforestry,” “sustainable agriculture,’’ “sustainablecommunities,’’ or some composite.

71

Owner Behavior

A long-standing issue in forest resource supplystudies is the likely behavior of private ownersand how behavior may be affected by incentivesand institutional factors. In addition to changingcharacteristics of the owner population over time,we also need to consider how changes in owner-ship objectives may impact the resources. Thelarger, publicly owned companies are affected byshareholder expectations, as well as pursuingacceptable rates of return on investment within achanging institutional environment of incentivesand regulations. The NIPF owners, in particular,possess multiple objectives, causing them to re-spond to socioeconomic forces and policies incomplex and sometimes unpredictable ways.Nontimber services likely could be enhanced ef-fectively by targeting incentive programs towardselect groups of NIPF owners (Kline et al. 2000).Other research has found that financial incentivesare not as critical to owners’ decisions in forestingriparian zones as some have suggested, andother factors include “neighborly concerns’’ andconcerns about restrictions on land managementand loss of flexibility (Kingsbury 1999). Behaviorby the diverse NIPF owner group also is affectedby whether owners live on their forested propertyin contrast to absentee owners. Another factor isincreasing density of people per forested unit,especially in areas such as the Puget Sound area.

Future Directions

The landscape comprises a mosaic of the majorland uses, which reflects and warrants a wideview of forces at work in shaping the landscape,in both temporal and spatial dimensions. Projec-tions indicate that demographic and other forceswill continue to affect that mosaic of land uses.If the U.S. population increases by 126 millionpeople over the next 50 years, along with a risein personal income, that could result in a net lossof 15 million acres of timberland (Alig et al. 2002,2003). Future improvements in policy could befacilitated by acknowledging different ownershipcharacteristics and a broader representation inanalyses and policy deliberations of those withcompeting interests in the land, including publicinformation about actions and possible conse-quences. With increasing populations and pres-sures and conflicts on the land, relationshipsamong major land uses, such as forestry, agricul-

ture, and urban/developed uses, are becomingmore evident. Natural resource policy issues nowoften involve forest and nonforest components inthe overall system. For example, the condition ofriverine systems is affected by a broad range ofland uses, and opportunities for improvementsexist on both wildlands and other lands (e.g.,pastureland in riparian zones). Efforts to betterunderstand how systems function, especially ter-restrial, aquatic, and socioeconomic systems,should bear in mind policy-relevant questions andconsider the marginal costs versus marginal ben-efits of undertakings.

From a long-term conservation perspective, pur-suit of new options for improvements in naturalresource conditions is increasingly faced withcompatibility issues among competing interestsin the land and incentives for owners. Efforts tobetter align commercial uses of the forests withother conservation objectives has led to increasedinterest in what is being called sustainable for-estry, although there are similar efforts tied toother major competing interests in the land,such as sustainable agriculture or sustainablecommunities.

When one looks at managing forested systemsfor multiple uses within ecological limits, thenconsideration of capital markets and other inter-connections between sectors of the economybecome more important. Policy issues associatedwith sustainability and planning information to aidin collective decisions about managing naturalresources are affected by changing product mar-ket and institutional conditions, as in recent de-cades with changes tied to marked reallocationof rural land. Improvements in U.S. agriculturalproduction technology allowed expanded agricul-tural production on a fairly constant land base.Less pronounced and in a lagged fashion, forestvolumes have increased on a timberland basethat is 4 percent smaller than the one in 1952.Demand for wood products is expected to keepgrowing, driven by the same population increasesand economic development that affect demandsfor other major land uses. However, intensifiedtimber management on the supply side repre-sents some of the largest projected changes in-volving private timberland (Alig et al. 1999a,2002), and could act to moderate or temper up-ward pressure on future wood prices. Private tim-berlands have the biological potential to provide

72

larger quantities of timber in an environmentallysound manner than they do today. Many of theopportunities for intensified forest managementcan be undertaken with positive economic returns,but changing social and institutional aspects mayaffect actualization. Most of the timber manage-ment intensification opportunities in the UnitedStates are in the South and the Pacific Northwestwest side. Further, more forests around the worldare managed today compared to earlier decades,and this includes an expanding area of relativelyproductive plantations. Even if projected intensifi-cation were implemented, most of the NIPFtimberland would still be concentrated in low man-agement-intensity classes that involve naturallyregenerated stands. Sustainability analyses canbe enhanced if both land use and land investmentoptions are examined. Analyses should be explicitwith respect to timing of tradeoffs in addition tothe growing attention given to spatial details. Thiswould increase the usefulness of such tools inpolicy analyses, such as in the 2000 RPA assess-ment and forest carbon analyses.

Metric Equivalents

1 acre = 0.405 hectares

Literature Cited

Adams, D.; Alig, R.; McCarl, B. [et al.]. 1996a.The forest and agricultural sector optimizationmodel (FASOM): model structure and policyapplications. Res. Pap. PNW-RP-495. Port-land, OR: U.S. Department of Agriculture, For-est Service, Pacific Northwest ResearchStation. 60 p.

Adams, D.; Alig, R.; McCarl, B.A. [et al.].1996b. An analysis of the impacts of publictimber harvest policies on private forest man-agement in the U.S. Forest Science. 42(3):343-358.

Alig, R.; Adams, D.; Chmelik, J.; Bettinger, P.1999a. Private forest investment and long-runsustainable harvest volumes. New Forests. 17:307-327.

Alig, R.; Adams, D.; McCarl, B. 1998. Impacts ofland exchanges between the forest and agri-cultural sectors. Journal of Agriculture andApplied Economics. 90(5): 31-37.

Alig, R.; Benford, F.; Moulton, R.; Lee, L.1999b. Long-term projections of urban anddeveloped land area in the United States. In:Otte, K., ed. Balancing working lands and de-velopment [CD-ROM]. Philadelphia, PA: Ameri-can Farmland Trust and Land Trust Alliance:[Pages unknown]

Alig, R.; Butler, B. 2002. Land use and forestcover projections for the United States: 1997-2050. In: Proceedings of the 2001 nationalSociety of American Foresters (SAF) conven-tion. Bethesda, MD: Society of American For-esters: 93-115.

Alig, R.; Healy, R. 1987. Urban and built-up landarea changes in the United States: an empiri-cal investigation of determinants. Land Eco-nomics. 63: 215-226.

Alig, R.; Mills, J.; Butler, B. 2002. Private timber-lands: growing demands, shrinking land base.Journal of Forestry. 100 (2): 32-37.

Alig, R.; Zheng, D.; Spies, T.; Butler, B. 2000.Forest cover dynamics in the Pacific Northwestwest side: regional trends and projections.Res. Pap. PNW-RP-522. Portland, OR. U.S.Department of Agriculture, Forest Service,Pacific Northwest Research Station. 22 p.

Alig, R.J.; Plantinga, A.J.; Kline, J.D. 2003.Land use changes involving forestry in theUnited States: 1952 to 1997, with projectionsto 2050. Gen. Tech. Rep. PNW-GTR-587.Portland, OR: U.S. Department of Agriculture,Forest Service, Pacific Northwest ResearchStation. 92 p.

Clawson, M. 1979. Forests in the long sweep ofAmerican history. Science. 204: 1168-1174.

Cohen, W.; Spies, T.; Alig, R. [et al.]. 2002.Characterizing 23 years (1972-1995) of standreplacement disturbance in western Oregonforests with Landsat imagery. Ecosystems. 5:122-137.

Easterling, D.; Evans, J.; Groisman, P. [et al.].2000. Observed variability and trends in ex-treme climate events: a brief review. Bulletinof the American Meteorological Society. 81:417-425.

73

Franklin, J.F. 1989. The “new forestry.” Journal ofSoil and Water Conservation. 44(6): 549.

Kingsbury, L. 1999. Oregon’s conservation re-serve enhancement program: likely participa-tion and recommendations for implementation.Corvallis, OR: Oregon State University. M.S.thesis.

Kline, J.; Alig, R.; Johnson, R. 2000. Fosteringthe production of nontimber services amongforest owners with heterogenous objectives.Ecological Economics. 33: 29-43.

Sampson, N. 2000. People, forests, and forestry:new dimensions in the 21st century. In:Sampson, N.; DeCoster, L., eds. Proceedings,forest fragmentation 2000. Washington, DC:American Forests: 53-59.

U.S. Department of Agriculture, Forest Serv-ice. 2001. 2000 RPA assessment of forest andrange lands. Report FS-687. Washington, DC.78 p.

U.S. Department of Agriculture, Natural Re-sources Conservation Service. [NRCS].2001. The national resources inventory, 1997.Revised. Washington, DC.www.nhq.nrcs.usda.gov/NRI/1997. [Date lastaccessed unknown].

U.S. Department of Commerce, Bureau of theCensus. 2002. Gateway to census 2000.Washington, DC. www.census.gov.(10 November 2003).

Zheng, D.; Alig, R. 1999. Changes in the non-federal land base involving forestry in westernOregon, 1961-1994. Res. Pap. PNW-RP-518.Portland, OR: U.S. Department of Agriculture,Forest Service, Pacific Northwest ResearchStation. 22 p.

74

Worldviews Panel

75

Presentation Objectives

Mason, Bruce & Girard has been asked to giveyou our impressions of the private landowner per-spective on landscape management. To do thissubject justice, we need to discuss private land-owner goals and objectives, state and federalregulations, and forest land management issuesas they relate to landscape management. As Iprepared these remarks, I have drawn on mynearly 30 years of experience within the forestproducts industry while with Georgia-Pacific Cor-poration and through the contacts I have madewhile with Mason, Bruce & Girard. I have tried tocapture questions we hear being asked and tohopefully give you an understanding of why thesequestions are important to the private forest land-owners within our region. This region needs diver-sity, flexibility, room for dissension, and morenonbureaucratic processes. Any perceived ben-efits of landscape management, at this point intime, are being viewed as far outweighing thecosts, which will be borne, in a large part, by theindividual forest-land owner.

Private Landowners’ Goals andObjectives

The first question we need to ask is, Why dopeople own forest land? What is their driving moti-vation? Is it profit, is it enjoyment of being able tomanage a forest based on their own interests, is itfor recreational purposes, or is it ownership forthe beauty of it?

Although there are some who own timberland toenjoy it, most private landowners manage theirlands with the objective of profiting from it insome manner, whether on an annual basis orover some other timeframe. There are many dif-ferent ideas on what forest land managementshould look like, from very intensive to a hands-offapproach.

These reasons for ownership will shape the land-owners’ goals and objectives and thus how theywill proceed with operations on their property.Ultimately, these reasons also will shape how theywill form any perspectives on landscape manage-ment.

State and Federal Regulations

The three Western states (Washington, Oregon,and California) have very comprehensive lawsgoverning timberland operations. These lawsprimarily regulate forest management operationsthat have the potential to impact public resourcessuch as fish and wildlife habitat and clean water.They may be modified somewhat, depending onthe size of a given ownership, although generallyspeaking most will apply to all landowners if man-agement activities occur.

Landowners, regardless of size of holdings, mustdetermine how best to fit their goals and objec-tives with how these laws impact their individualownership. For example, a landowner with smallacreage with fish-bearing streams flowing throughthe property may have few management optionsas compared to a landowner whose property con-tains only non-fish-bearing streams and, whotherefore, may have several management op-tions. Corporate ownerships generally are largeenough to work around the regulations and stillmeet corporate mandates.

As an example of how regulations can impact themanagement of our private forest lands, Mason,Bruce & Girard completed a study for the OregonForest Industries Council in the early 1990s to testthe requirements of Oregon Senate Bill 1125. Thisbill, which was backed by forest landowners, setnew requirements designed to spread timber har-vest over the landscape by restricting clearcutareas to 120 acres in size and requiring regenera-tion to be 4 feet high, 4 years old, and “free togrow” before harvesting an adjacent timber stand.“Free to grow” is defined as “a tree or stand ofwell-distributed trees, of acceptable species andgood form, with a high probability of remaining or

Private Landowner Perspective on Landscape Management

Carl F. Ehlen1

1 Forestry consultant, Mason, Bruce & Girard, Inc., 707 SWWashington Street, Suite 1300, Portland, OR 97205; Tel: 503-224-3445; e-mail: cehlen@masonbruce.com

76

becoming vigorous, healthy, and dominant overundesired competing vegetation.” For the purposeof this definition, trees are considered well distrib-uted if 80 percent or more of the area contains atleast 200 trees per acre and not more than 10percent contains fewer than 100 trees per acre.The study determined that the “free to grow” re-quirement, when combined with leave areas be-tween harvest units and maximum regenerationharvest unit size, created the following:

1. Significant impacts on the size, number, distri-bution, and timing of regeneration harvests.

2. A slowing of the rate of harvest on individualtracts.

3. The potential for significantly increasing themiles of roads, which not only adds to a land-owner’s cost but also adds to the probabilityfor increased environmental damage.

4. Land being taken out of production.

5. Increased stand fragmentation as the maxi-mum regeneration harvest size is further re-duced, thus potentially jeopardizing the abilityof a landowner to fully harvest the annualgrowth on the property.

6. A significant reduction in the size of regenera-tion harvest units with the average alwaysbeing less than the maximum allowed.

7. More broken ownership, which makes it moredifficult to efficiently schedule harvesting ac-tivities.

8. More dispersion of regeneration units over thelandscape.

Private landowners have long supported forestregulations backed by scientific evidence. Now,however, as regulatory proposals pass from sci-entific to political, private landowners are becom-ing more and more leery of what might be ex-pected of them as they continue to managetheir property and risk the potential loss of theirinvestments.

Will more regulations, particularly if seen as politi-cal, drive more disinvestment of forest land?

Landscape Management Perspectivesand Questions

Just as management goals and objectives differamong private landowners, so do their perspec-tives, and therefore their definitions, of landscapemanagement. To our knowledge, there is no ob-jective, all-encompassing definition of landscapemanagement or what it means to the individuallandowner in terms of impacting their timber re-source.

Is “landscape management” providing for all pub-lic resources on every acre, or is it providing for allpublic resources through a mosaic of age andsize classes of forests across the landscape?

Is it important that all landowners provide thismosaic, regardless of size of ownership?

Does landscape management include all land-owners—private, state, and federal alike—in agiven watershed or viewshed or larger area?

Are all landowners to be treated equally?

Will a landowner be able to factor in individual orcorporate goals and objectives?

Given the above, we believe that the following, ata minimum, must be studied and analyzed beforeprivate landowners would even consider buyinginto any landscape management scheme.

Economic Analysis

“Land” or “soil” expectation is an indicator of thevalue of a piece of bare land as a forestry invest-ment. As additional regulations are imposed, soilexpectation values on low-site lands begin to be-come marginal, particularly east of the Cascadecrest, but the same is true on the west side aswell. This concern begins to beg the issue of dis-investment of lower site lands.

Current state and federal regulations are costly tothe private landowner. To date we know of noanalysis to determine the additional costs andbenefits to the private landowner operating underany landscape management framework. An eco-nomic analysis is imperative, as any regulations

77

in addition to what we have today would be ex-tremely costly. As environmental regulation in-creases, we arrive at the point of getting smallmarginal benefits at an enormous cost, most ofwhich will be borne by the landowner.

Has anyone taken the time to do an analysis ofthis sort?

Rotation Length Issues

Landowners must be able to determine the requi-site rotation length for their property based on theproductive capacity of the land and their goalsand objectives, while at the same time being ableto harvest their timber in response to marketchanges in accordance with current regulations.

How will landscape management schemes allowfor this within the required timing?

How will final rotations be impacted?

Will landowners be allowed to set their own rota-tions and harvest ages depending on silviculturalobjectives and market fluctuations?

Silvicultural Issues

With the serious decline in federal timber salevolume, combined with the continuing demand forwood products, many private landowners are en-deavoring to harvest timber equal to the growthrates of their property. Many private landownersare investing heavily in various silvicultural activi-ties to assure that their forest land is producingtimber at its productive capacity. These activitiesinclude, among others, planting genetically im-proved seedlings, brush control, early and mid-rotation fertilization, and midrotation thinning. Weare seeing a quality of forest management activi-ties that is unprecedented. And yet, landownerscontinue to improve on these efforts.

Will these efforts and investments go for naught?In other words, will the financial investments inplanting, vegetation control, precommercial thin-ning, etc. made early in the life of a rotation belost if a landowner is not allowed to harvest hiscrop?

How does even-age management fit with land-scape management goals and objectives?

Are we as a society only interested in how ourprivate forest lands look and their resource pro-

tection attributes as opposed to keeping themproductive and profitable while continuing to pro-tect our public resources?

Is there a balance that can be reached betweenprivate timber production and habitat require-ments?

Administration of LandscapeManagement Requirements

A great deal of thought must go into how land-scape management issues are to be adminis-tered. These issues must be examined by eachlandowner rather than on a watershed or land-scape basis. Problems will erupt as soon as onelandowner is restricted or impacted to a greaterdegree than other landowners.

Who will be the regulator?

Who determines if landowners are complying withthe requirements?

How will anyone be able to make a determinationin an objective and legal manner?

Summary Comments

Our principal issue at this point is that we do notknow what landscape management looks like nordo we know what, if anything, a private landownerwill have to give up. Is this another attempt tofurther regulate the private landowner? Will land-scape management become a form of “central-ized planning?”

We do not know of any private landowner in theNorthwest who does not embrace the need for theprotection and the sustainability of public re-sources. The current regulatory framework isinherently designed to provide for both.

How one landowner’s goals fit with an adjacentlandowner’s goals and how a given landowner’soperations fit within the definition of managing ona landscape scale, we assume, are still to be de-termined. In the end, however, private landownersmust be included in any discussions of landscapemanagement, particularly if regulatory proposalsare being considered. There is too much value atstake for discussions on this issue to proceedotherwise.

78

Describing past conditions is part of a coarse-filtermanagement strategy for sustaining biologicaldiversity (Cissel et al. 1994, Haufler 1994, Haufleret al. 1996, Hunter et al. 1989, Swanson et al.1994). A description of past ecosystem structureand its variability is useful in exploring the causesand consequences of change in ecosystem char-acteristics over time, and thus for understandingthe set of conditions and processes that sustainedecosystems prior to their recent alterations byhumans. It provides a context for interpretingnatural processes, especially disturbance, and itallows variability in patterns and processes to beunderstood in terms of a dynamic system. Naturalresource managers can use such information toevaluate the magnitude, direction, and causes oflandscape change, especially change induced bypeople. An assessment of current conditions rela-tive to the historical range of variability (HRV),also called natural variability, also can be used toidentify management goals, particularly wherethose goals encompass sustainability, ecologicalrestoration, conservation of biological diversity,and maintenance of natural processes (Hann andBunnell 2001, Landres et al. 1999, Morgan et al.1994, Swanson et al. 1994, White and Walker1997). The concept is less useful when manage-ment objectives are narrowly focused on maximiz-ing commodity production or sustaining thepopulation of an endangered plant or animal(Landres et al. 1999).

In this paper, I give a brief overview of the HRVconcept, which is also called natural variabilityand reference variability, and its use in under-standing and managing dynamic landscapes. Ibriefly discuss the value and limitations of history,and then summarize what we have learned fromhistorical ecology that is of use to managers ofdynamic landscapes.

Defining Historical Range of Variability

Historical range of variability characterizes fluc-tuations in ecosystem conditions or processesover time (Landres et al. 1999, Manley et al. 1995,Morgan et al. 1994, Swanson et al. 1994), andthus provides researchers and managers with areference against which to evaluate recent andpotential ecosystem change. Landres et al. (1999)defined natural variability as “the ecological condi-tions and their variability over space and timerelatively unaffected by people.” Thus, HRV de-fines bounds of system behavior that remain rela-tively consistent over time. It is described for atimeframe relevant to understanding the behaviorof contemporary ecosystems and the implicationsfor management, usually prior to the intensiveinfluence of European-Americans and with rela-tively consistent climate. Many people use atimeframe of 100 to 700 years before present tointerpret successional dynamics and to character-ize past variation in forest structure and composi-tion, but that depends on the available historicalinformation and the ecosystem of interest. Histori-cal data on disturbance occurrence and effectsare increasingly limited the further back in time welook. Using recent centuries includes burning andother actions of American Indians, and encom-passes climatic variability, such as the Little IceAge. Hann et al. (1997) used the last 2,000 yearsas the timeframe for HRV for the interior Colum-bia River basin.

Utility of This Concept

The utility of this concept is based on two con-cepts: (1) past conditions and processes providecontext and guidance for managing ecologicalsystems today, and (2) spatial and temporal vari-ability caused by disturbance is vital to ecosystemintegrity (Landres et al. 1999). However, the utilitywill also depend on the management issues andthe social and ecological context for decisions(Landres et al. 1999). The HRV often is used in

Back to the Future: The Value of History in Understandingand Managing Dynamic Landscapes

Penelope Morgan1

1 Professor, College of Natural Resources, University ofIdaho, Moscow, ID 83844-1133; Tel: 208-885-7507; e-mail:pmorgan@uidaho.edu.

79

combination with societal values and other infor-mation to identify desired future conditions(Haufler et al. 1996, Landres et al. 1999). Histori-cal information has been used to guide manage-ment of waterflow on the Colorado River (Poff etal. 1997) and in the Everglades (Harwell 1997),and in management of fire (Brown et al. 1994,Skinner and Chang 1996) and forest structure(Baker 1992, Camp et al. 1997, Cissel et al. 1999,Gauthier et al. 1996, Hann and Bunnell 2001,Keddy and Drummond 1996, Ripple 1994). It alsohas been used in ecological restoration (Allen etal. 2002, Covington and Moore 1997, White andWalker 1997), and to deepen our understandingof the processes driving forest change (Lertzmanet al. 1997; Lesica 1996; Mladenoff and Pastor1993; Swetnam and Betancourt 1990, 1998;Wimberley et al. 2000). Ecosystem managementoften relies heavily on a description of past vari-ability in defining desired future conditions(Christensen et al. 1996, Kaufmann et al. 1994,Manley et al. 1995).

The greatest values for characterizing HRV areunderstanding (1) how the processes and factorsdriving ecosystem change differ from one placeand time to another, (2) the influence of drivingvariables in the past, and (3) how those drivingvariables might influence ecosystems today andin the future (Landres et al. 1999). Managers de-veloped the HRV concept in a search for legallydefensible strategies for sustaining biological di-versity and ecological integrity.

Data for describing HRV differ in temporal andspatial scale (Morgan et al. 2001, Swetnam et al.1999). Both natural (fire-scarred trees, pollen,charcoal, etc.) and human archives (old photo-graphs, early survey data, etc.) are sources ofhistorical data (Swetnam et al. 1999). Althoughwe know more about fire history than about otherdisturbances, none of our fire history data arecomprehensive across both time and space, andmost are better at describing frequency than se-verity, size, or spatial pattern of disturbances. Fewadequately inform us about spatial pattern anddynamics at the landscape scale (Morgan et al.2001, Swetnam et al. 1999). Also, we have moredata for points and small areas than we have forlandscapes and watersheds (Landres et al. 1999,Morgan et al. 2001, Swetnam et al. 1999). Thus,extrapolation is required, but extrapolation from

points to large areas, and from short time inter-vals (years and decades) to long time intervals(centuries) is problematic and fraught with error(Morgan et al. 2001). We can more readily recon-struct past vegetation structure than past pro-cesses, and when we reconstruct past foreststructure, we are more confident about the num-ber of large trees than we are about how manysmall trees there were (Allen et al. 2002). Hierar-chy theory can be helpful in interpreting data fromdifferent spatial and temporal scales.

Historical Range of Variability andDesired Future Conditions

Comparing current and desired future conditions(DFCs) to HRV is useful for identifying manage-ment strategies and priorities (fig. 1). I sometimesdemonstrate this with a ball (existing conditions)rolling around on a large plate (HRV). When con-ditions are within the HRV, the ball rolls around onthe plate. The edges of the plate keep the ball onthe plate (the system is bounded), until somethingpushes it over the edge (outside of HRV). Thenthe ball drops (change is rapid, and though it maybe somewhat predictable, the ball might not returnto the plate). Further, I can hold the ball steady,but doing so takes energy on my part (an externalsubsidy to the system in the form of human ac-tion). Now imagine a small plate representingDFCs. This plate is smaller because DFCs aretypically more limited than HRV. It is possible tohave all three stacked up: the ball on the smallplate and both on the large plate. We do haveforest ecosystems where current conditions arewithin the DFCs and those are within HRV. Thisincludes, for example, some subalpine forests. Insuch cases (row 1 of fig. 1), our managementdecision might be to maintain this desirable condi-tion, which we view as sustainable. We can, how-ever, imagine other combinations. Many urbanand agricultural areas, for instance, would becharacterized with the ball on a small plate (cur-rent conditions within DFCs) but both separatefrom the large plate (both outside of HRV).Whether or not such conditions are sustainabledepends on whether we as a society are willing toassume the external subsidies (e.g., financialsubsidies to farmers, and fertilizer and herbicidemade from fossil fuels), effects (e.g., quality of lifeand environment), and risks (e.g., climatechanges).

80

Thus, current conditions, DFCs and HRV mayor may not overlap. All three or any two might besimilar, or all might be different. Depending onsocial, economic, and political issues, restorationis suggested where desired conditions are withinHRV, but current conditions are different (row 2of fig. 1). When DFCs differ from HRV or currentconditions, desired future conditions may need tobe reevaluated (rows 3, 4 and 5 of fig. 1). Externalsubsidies will be required to maintain such a con-dition, but people can choose to bear both thesubsidies and the risk of change to the system.Ideally, scientists and managers can quantify theimplications of different choices as the basis forinformed decisions by the stakeholders. Haufler etal. (1996) suggest combining knowledge of HRVwith knowledge of species needs to identify “ad-equate representation” of elements of ecosystemdiversity.

Examples of Applications of HistoricalRange of Variability in LandscapeManagement

Hann and Bunnell (2001) describe the develop-ment of landscape management prescriptionsbased upon assessment of changes from HRV.They give examples at several spatial scales,including the mapping of historical fire regimes,current fire regimes, and condition classes reflect-ing the degree of change for the conterminousUnited States done by Hardy et al. (2001) andavailable on the Web (http://www.fs.fed.us/fire/fuelman). Hann and Bunnell (2001) also describe

the use of HRV description, modeling succession,and identifying areas of departures from HRV asa means for prioritizing for restoration in Colorado.Their approach is similar to one applied in theSierras by Caprio and Graber (2000). Cisselet al. (1999) contrast two approaches to land-scape management in the Blue River watershedin Oregon. An approach based on using logging toapproximate past disturbance regimes while alsomaintaining stream buffers created less forestfragmentation than an alternative based on matrixand corridors under the Northwest Forest Plan.

What Have We Learned?

Historical ecology provides many lessons. Forinstance, we now understand that disturbancewas and is ubiquitous. At any given time, mostecosystems are in some state of recovery fromdisturbance. Directly and indirectly, disturbancesstrongly influence the structure, composition,diversity, and function of the ecological systemswe manage. Further, disturbances interact, andthe relative importance of disturbances varies withscale. For instance, it is likely that at some broadspatial scales, regional and global climatesupercedes local fuel and topography in determin-ing fire patterns (Swetnam and Betancourt 1998).It is likely, for instance, that El Niño drives firepatterns across the Southwestern United States(Swetnam and Betancourt 1990) and in thePatagonian region of South America (Kitzbergerand Veblen 1997). The results of historical analy-sis can sometimes be surprising. For instance,

Status

Graphical representation

Management action

HRV ≥ DFC ≥ CC

|…xxx...|

|__________________|

Maintain

HRV ≥ DFC ≠ CC |…..…...| xxx |__________________|

Restore

HRV ≠ DFC ≥ CC |__________________| |…xxx...|

HRV ≥ CC ≠ DFC |________xxx_______| |…..…...| HRV ≠ DFC ≠ CC

|_________________| |……...| xxx

Evaluate carefully to assess risks, sustainability, and the external subsidies of DFC; reevaluate DFC

Figure 1—Comparing both current ecological conditions (CC, xxx) and desired future conditions (DFC |———|), to historical

range of variability (HRV, |_____|) is useful for identifying management actions. (Redrawn from Landres et al. 1999).

81

Swetnam and Lynch (1993) found that sprucebudworm (Choristoneura occidentalis Freeman),a defoliating insect in conifers, is more prevalentduring wet than during dry years, counter to thecommonly held notion that defoliation occurs indrought-stressed trees.

Scale matters. Wimberley et al. (2000) modeledthe historical range of variability of old forests inthe Oregon Coast Range. The historical range ofvariability was progressively narrower as theyincreased the analysis area by orders of magni-tude from a small watershed to a large watershedand then to a region.

Changes in forests at the landscape scale aregreat and variable (Hann et al. 1997). Forestsat high elevation have changed in different waysthan forests at low elevations over the last 60years, and both biophysical and social factorsare correlated with different types and degreesof change (Black 1998). Although changes infire regimes are extensive, the changes are notthe same everywhere (Hardy et al. 2001, http://www.fs.fed.us/fire/fuelman).

Rigorous testing of hypotheses by using historicaldata has even greater potential for rapidly advanc-ing ecological understanding. Time series analy-sis, comparative case studies, and analyzingsynchronous events across broad spatial scaleshave taught us a great deal about the relativeinfluences of climate, land use, topography andvegetation on fire regimes (Rollins et al. 2000a,2000b, 2001; Swetnam 1993; Swetnam et al.1999; Swetnam and Betancourt 1998), and aboutthe interactions between insects, forests, andclimate (Swetnam and Lynch 1993).

Through modeling (e.g., Miller and Urban 1999,Wimberley et al. 2000), comparison of managedand unmanaged landscapes, and examination ofhistorical data, we are coming to understand howand why landscapes change. That understandingis very important to managers of dynamic land-scapes, even when those landscapes are subjectto introduced species, structures, and processes.

Implications and Challenges forManagers

We need new approaches to landscape planningthat effectively integrate the three very differentapproaches currently applied.2 Traditional ap-proaches to evening out the flow of commoditieshave taught us a lot about modeling complex andinteracting systems, but it is often difficult to incor-porate ecological uncertainty and change into thisapproach. Conservation planning based uponcorridors and matrices accommodates the needsof many species but doesn’t recognize that land-scapes are dynamic. Basing landscape manage-ment prescriptions on disturbance regimes buildsupon our knowledge of how and why landscapeschange (Cissel et al. 1994, 1999) but may notprovide for the needs of species and people.Thus, we need to develop approaches to land-scape planning that accommodate multiple spe-cies and processes in ecosystems subject tointroduced species, processes, and humanpopulation pressure. For whitebark pine (Pinusalbicaulis Engelm.) ecosystems, for instance,landscape planning must incorporate the needsand dynamics of grizzly bears, Clark’s nutcrack-ers, squirrels, fire, the introduced blister rust, andclimate change, along with scenic values andwatershed protection. Historical information canhelp us to understand this system and to buildmodels, which are then used to compare alterna-tive futures for this ecosystem (Keane and Arno2001).

Human-induced changes to our global, regional,and local environments pose great challenges tonatural resource managers. Our climate is chang-ing (IPCC 2001). Introduced species have alteredour environment as well (Vitousek et al. 1996).Some will find HRV less useful in addressingmanagement of ecosystems that are clearlyundergoing changes that are to some degreeunprecedented when coupled with human popula-tion pressures. However, effective management

2 Swanson, F. 1998. Personal communication. Researchgeologist, U.S. Department of Agriculture, Forest Service,Pacific Northwest Research Station, Forestry SciencesLaboratory, 3200 SW Jefferson Way, Corvallis, OR 97331;Tel: 541-750-7355; e-mail: fswanson@fs.fed.us.

82

will continue to depend on our understanding ofecosystems. The ecological implications of cli-mate change and the dynamics of ecosystemsreinforce the need for a broader focus on restor-ing ecological integrity, resilience, and sustain-ability, rather than on restoring some “vignette”or past condition (Landres et al. 1999, White andWalker 1997).

The potential consequences to biological diversityare great if current forest conditions are outside ofand remain outside of the historical range of vari-ability. Consequences include increased vulner-ability to stand-replacing fires where few everoccurred historically, which can put soils, streams,and fish at risk (Hann et al. 1997). This does notmean, however, that we must return our forestscompletely to historical conditions to sustain bio-logical diversity. Thoughtful evaluation of changefrom historical conditions; evaluation of socialneeds; presence of exotic species, structures,and processes; and future climate all should playa role in identifying DFCs and the managementalternatives to achieve them (Landres et al. 1999).

Acknowledgments

I am deeply indebted to the many creative andthoughtful managers and scientists who helpedme develop my ideas. Wendel Hann was one ofthe managers who originated the HRV concept.Discussions with him and with Tom Swetnam,Peter Landres, Dave Parsons, Bob Keane, ArtZack, Steve Bunting, and many others havehelped me shape my ideas. Lee Harry, a silvicul-turist on the Beaverhead National Forest, origi-nated the bowl, ball, and plate demonstration,and thus his ideas were the genesis for figure 1.

Literature Cited

Allen, C.D.; Savage, M.; Falk, D.A. [et al.]. 2002.Ecological restoration of Southwestern ponde-rosa pine ecosystems: a broad perspective.Ecological Applications. 12(5): 1418-1433.

Baker, W.L. 1992. Effects of settlement and firesuppression on landscape structure. Ecology.73: 1879-1887.

Black, A.E. 1998. Forest patterning in the interiorColumbia River basin: 60 years of social andbiophysical interaction. Moscow, ID: Universityof Idaho. 137 p. Ph.D. disseration.

Brown, J.K.; Arno, S.F.; Barrett, S.W.; Menakis,J.P. 1994. Comparing the prescribed naturalfire program with presettlement fires in theSelway-Bitterroot Wilderness. InternationalJournal of Wildland Fire. 4: 157-168.

Camp, A.; Oliver, C.; Hessburg, P.; Everett, R.1997. Predicting late-successional fire refugiapre-dating European settlement in theWenatchee Mountains. Forest Ecology andManagement. 95: 63-77.

Capiro, A.C.; Graber, D.M. 2000. Returning fireto the mountain: Can we successfully restorethe ecological role of pre-Euroamerican fireregimes to the Sierra Nevada? In: Cole, D.M.;McCool, S.F.; Borie, W., O’Laughlin, J., comps.Wilderness science in a time of change confer-ence. Vol. 5: Wilderness ecosystems, threats,and management Proc. RMRS-P-15-VOL-5.Wilderness ecosystems, threats, and manage-ment. Ogden, UT: U.S. Department of Agricul-ture, Forest Service, Rocky MountainResearch Station: 233-241.

Christensen, N.L.; Baruska, A.M.; Brown, J.H.[et al.]. 1996. The report of the Ecological So-ciety of America committee on the scientificbasis for ecosystem management. EcologicalApplications. 6(3): 665-691.

Cissel, J.H.; Swanson, F.J.; McKee, W.A.;Burditt, A.L. 1994. Using the past to plan thefuture in the Pacific Northwest. Journal of For-estry. 92: 30-31, 46.

Cissel, J.H.; Swanson, F.J.; Weisberg, P.J.1999. Landscape management using historicalfire regimes: Blue River, Oregon. EcologicalApplications. 9(4): 1217-1231.

Covington, W.W.; Moore, M.M. 1997. Restoringecosystem health in ponderosa pine forests ofthe Southwest. Journal of Forestry. 95: 23-29.

Gauthier, S.; Leduc, A.; Bergeron, Y. 1996. For-est dynamics modelling under natural firecycles: a tool to define natural mosaic diversityfor forest management. Environmental Moni-toring and Assessment. 39: 417-434.

83

Hann, W.J.; Bunnell, D.L. 2001. Fire and landmanagement planning and implementationacross multiple scales. International Journal ofWildland Fire. 10: 389-403.

Hann, W.J.; Jones, J.L.; Karl, M.G. [et al.].1997. Landscape dynamics of the basin. In:Quigley, T.M.; Arbelbide, S.J., tech. eds. Anassessment of ecosystem components in theinterior Columbia basin and portions of theKlamath and Great Basins, Gen. Tech. Rep.PNW-GTR-405. Portland, OR: U.S. Depart-ment of Agriculture, Forest Service, PacificNorthwest Research Station: 363-1055. Vol. 2

Hardy, C.C.; Schmidt, K.M.; Menakis, J.P.;Sampson, R.N. 2001. Spatial data for nationalfire planning and fuel management. Interna-tional Journal of Wildland Fire. 10: 353-372.

Harwell, M.A. 1997. Ecosystem managementof south Florida. BioScience. 47: 499-512.

Haufler, J.B. 1994. An ecological framework forplanning for forest health. Journal of Sustain-able Forestry. 2(3/4): 307-316.

Haufler, J.B.; Mehl, C.A.; Roloff, G.J. 1996.Using a coarse-filter approach with a speciesassessment for ecosystem management.Wildlife Society Bulletin. 24(2): 200-208.

Hunter, M.L.; Jacobson, G.L.; Webb, T. 1989.Paleoecology and the coarse-filter approach tomaintaining biological diversity. ConservationBiology. 2: 375-385.

Intergovernmental Panel on Climate Change[IPCC]. 2001. Climate change 2001: impacts,adaptation and vulnerability. http://www.ipcc.ch/(1 May).

Kaufmann, M.R.; Graham, R.T.; Boyce, D.A.[et al.]. 1994. An ecological basis for ecosys-tem management. Gen. Tech. Rep. RM-GTR-246. Fort Collins, CO: U.S. Department ofAgriculture, Forest Service, Rocky MountainForest and Range Experiment Station. 22 p.

Keane, R.E.; Arno, S.F. 2001. Restoration con-cepts and techniques. In: Tomback, D.F.; Arno,S.F.; Keane, R.E., eds. Whitebark pine com-munities: ecology and restoration. Washington,DC: Island Press: 367-401.

Keddy, P.A.; Drummond, C.G. 1996. Ecologicalproperties for the evaluation, management,and restoration of temperate deciduous for-est ecosystems. Ecological Applications.6: 748-762.

Kitzberger, T.; Veblen, T.T. 1997. Influencesof humans and ENSO on fire history ofAustrocedrus chilensis woodlands in northernPatagonia, Argentina. Ecoscience. 4: 508-520.

Landres, P.B.; Morgan, P.; Swanson, F.J. 1999.Overview of the use of natural variability con-cepts in managing ecological systems. Eco-logical Applications. 9(4): 1179-1188.

Lertzman, K.; Spies, T.; Swanson, F. 1997.From ecosystem dynamics to ecosystem man-agement. In: Schoonmaker, P.K.; von Hagen,B.; Wolf, E.C., eds. The rain forests of home:profile of a North American bioregion. Wash-ington, DC: Island Press: 361-382.

Lesica, P. 1996. Using fire history models toestimate proportions of old growth forest innorthwestern Montana, USA. Biological Con-servation. 77: 33-39.

Manley, P.N.; Brogan, G.E.; Cook, C. [et al.].1995. Sustaining ecosystems: a conceptualframework. R5-EM-TP-001. Albany, CA: U.S.Department of Agriculture, Forest Service,Pacific Southwest Region and Pacific South-west Research Station.

Millar, C.I.; Woolfenden, W.B. 1999. The role ofclimate change in interpreting historical vari-ability. Ecological Applications. 9: 1207-1216.

Miller, C.; Urban, D.L. 1999. Forest pattern, fireand climate change in the Sierra Nevada. Eco-systems. 2: 76-87.

Mladenoff, D.J.; Pastor, J. 1993. Sustainableforest ecosystems in the northern hardwoodand conifer region: concepts and manage-ment. In: Aplet, G.H.; Olsen, J.T.; Johnson, N.;Sample, V.A., eds. Defining sustainable for-estry. Washington, DC: Island Press: 145-180.

Morgan, P.; Aplet, G.H.; Haufler, J.B. [et al.].1994. Historical range of variability: a usefultool for evaluating ecosystem change. Journalof Sustainable Forestry. 2(1/2): 87-111.

84

Morgan, P.; Hardy, C.; Swetnam, T.W. [et al.].2001. Mapping fire regimes across time andspace: understanding coarse- and fine-scalepatterns. International Journal of Wildland Fire.10: 1-14.

Poff, N.L.; Allan, J.D.; Bain, M.B. [et al.]. 1997.The natural flow regime. BioScience. 47: 769-784.

Ripple, W.J. 1994. Historic spatial patterns of oldforests in western Oregon. Journal of Forestry.92: 45-49.

Rollins, M.; Swetnam, T.W.; Morgan, P. 2000a.Twentieth century fire patterns in the Gila/AldoLeopold Wilderness Complex in New Mexicoand the Selway-Bitterroot Wilderness AreaIdaho/Montana. In: Crossing the millennium:integrating spatial technologies and ecologicalprinciples for a new age in fire management:Proceedings from the Joint Fire Science con-ference and workshop. Moscow, ID: Universityof Idaho: 161-169.

Rollins, M.; Swetnam, T.W.; Morgan, P. 2000b.Twentieth-century fire patterns in the Selway-Bitterroot Wilderness Area, Idaho/Montanaand the Gila/Aldo Leopold Wilderness Com-plex. In: Cole, D.N.; McCool, S.F.; Borrie, W.T.;O’Loughlin, J., comps. Wilderness science in atime of change conference: Vol. 5. Wildernessecosystems threats and management. Proc.RMRS-P-15. Fort Collins, CO: U.S. Depart-ment of Agriculture, Forest Service, RockyMountain Research Station: 283-287.

Rollins, M.G.; Swetnam, T.W.; Morgan, P. 2001.Evaluating a century of fire patterns in twoRocky Mountain wilderness areas using digitalfire atlases. Canadian Journal of Forest Re-search. 31(12): 2107-2123.

Skinner, C.N.; Chang, C. 1996. Fire regimes,past and present. In: Status of the Sierra Ne-vada, Volume II: assessments and scientificbasis for management options. Wildland Re-sources Center Report 37. Davis, CA: Univer-sity of California, Centers for Water and Wild-land Resources: 1041-1070. Sierra NevadaEcosystem Project final report to Congress.

Swanson, F.J.; Jones, J.A.; Wallin, D.O.;Cissel, J.H. 1994. Natural variability—implica-tions for ecosystem management. In: Jensen,M.E.; Bourgeron, P.S., tech. coords. Ecosys-tem management: principles and applications,Volume II. Eastside forest ecosystem healthassessment. Gen. Tech. Rep. PNW-GTR-318.Portland, OR: U.S. Department of Agriculture,Forest Service, Pacific Northwest ResearchStation: 80-94.

Swetnam, T.W.1993. Fire history and climatechange in giant sequoia groves. Science.262: 885-889.

Swetnam, T.W.; Allen, C.D.; Betancourt, J.L.1999. Applied historical ecology: using the pastto manage for the future. Ecological Applica-tions. 9(4): 1189-1206.

Swetnam, T.W.; Betancourt, J.L. 1990. Fire-southern oscillation relations in the Southwest-ern United States. Science. 249: 1017-1020.

Swetnam, T.W.; Betancourt, J.L. 1998. Meso-scale disturbance and ecological response todecadal climatic variability in the AmericanSouthwest. Journal of Climate. 11: 3128-3147.

Swetnam, T.W.; Lynch, A.M. 1993. Multicentury,regional-scale patterns of western spruce bud-worm outbreaks. Ecological Monographs.63(4): 339-424.

Vitousek, P.M.; D’Antonio, C.M.; Loope, L.L.;Westbrooks, R. 1996. Biological invasions asglobal environmental change. American Scien-tist. 84(5): 468-478.

White, P.S.; Walker, J.L. 1997. Approximatingnature’s variation: selecting and using refer-ence information in restoration ecology. Con-servation Biology. 5(4): 338-349.

Wimberley, M.D.; Spies, T.A.; Long, C.J.;Whitlock, C. 2000. Simulating historical vari-ability in the amount of old forests in the Or-egon Coast Range. Conservation Biology.14(1): 167-180.

85

86

Case Studies

87

In the early 1990s, the forest industry in the Pa-cific Northwest was in a condition that has beendescribed as “gridlock” (FEMAT 1993). Becauseof mounting evidence that old-growth forest eco-systems were declining in health and integrity,and because of the listing or potential listing ofspecies (such as the northern spotted owl (Strixoccidentalis caurina) and various salmon stocks)by the U.S. Fish and Wildlife Service as endan-gered or threatened, forest management prac-tices were under intense criticism. Researchsuggested that a large proportion of the historicalacreage of old growth had been lost to timberharvest, and that the remaining stands wereheavily fragmented, causing risks to old-growth-dependent species (FEMAT 1993). In addition,habitat loss from roads, culverts, and degradationof riparian areas was contributing to problemswith fish populations. On federal forest lands(those lands managed by the USDA Forest Ser-vice and USDI Bureau of Land Management[BLM]) within the range of the northern spottedowl (western Washington and Oregon, and north-ern California), a series of court injunctionsagainst logging and delays from appeals of timbersales had resulted in a dramatic decline of timberharvest operations, with major economic, social,and political consequences.

In 1993, President Clinton convened a largegroup (ultimately over 100 contributors) of fed-eral agency and university scientists, the ForestEcosystem Management Assessment Team(FEMAT), and gave them the task of producing aplan that would both provide for viability of old-growth ecosystems and their associated speciesand provide for a sustainable level of economicbenefits from forest products (Johnson et al.1999). Led by Jack Ward Thomas (later Chief ofthe USDA Forest Service), the FEMAT group de-veloped several options for consideration by

agency officials; the selected alternative wasadopted in 1994 (USDA USDI 1994) and becamewhat is now known as the Northwest Forest Plan(NFP) (fig.1). The NFP guides management of 24million acres of federal forest lands (includingnational forest and BLM lands, as well as nationalparks) within the range of the northern spottedowl.

The Northwest Forest Plan–AnEcosystem Management Approach

The Northwest Forest Plan was one of the earliestand largest attempts to implement what was atthat time a relatively new concept in public landuse planning–that of managing whole systems,including broad-scale vegetation patterns, bio-physical processes, and whole suites of species.This approach has become widely known as “eco-system management.”

The NFP treats the land base as a system of in-terconnected parts, providing a single conserva-tion biology strategy for wildlife, plants, fish, andfungi associated with late-successional and old-growth forests. The old-growth orientation of theNFP distinguishes it from other biodiversity con-servation efforts. It weaves together efforts tar-geted at individual species potentially at risk (the“survey and manage” component, described inthe next section), with broad-scale strategies thatsweep in groups of species and ecological pro-cesses (reserves, special guidelines for activelymanaged areas). It specifically provides for activemanagement for both commodity production andrestoration. And finally, it is a long-term plan (100-year timeframe in which target conditions becomefully implemented) with checkpoints along the wayto determine if adjustments are needed in order tomeet the goals.

A primary feature of ecosystem management isthat it is science based. To achieve sustainability,plans must be built with an understanding of the

Northwest Forest Plan

Nancy M. Diaz1

1Chief, Branch of Physical Sciences, OR/WA State Office,U.S. Department of the Interior (USDI) Bureau of LandManagement, P.O. Box 2965, Portland, OR 97208-2965; Tel:503-808-6036; e-mail: ndiaz@or.blm.gov

88

interactions and possibilities among the compo-nents of forest ecosystems and their associatedhuman communities. The NFP not only incorpo-rated science, it was developed largely by scien-tists. This mode of plan development helped gainthe scientific credibility needed to reverse thecourt injunction against timber harvest, and asthe plan proceeded to be implemented, resultedin significant growth in scientist-manager collabo-rations.

The NFP includes three major facets:

• A strategy (composed of land allocations andstandards and guidelines) for managing 24million acres of federal forest land.

• The Northwest Economic Adjustment Initia-tive, a $1.2 billion program (spread over5 years) that provides support to timber-de-pendent communities, tribes, and businesses.

• An interagency model of policymaking,wherein the participating federal agencies act

as partners in formulating policies and majorimplementation steps. The participating agen-cies are:

o USDA Forest Service—Pacific Northwestand Pacific Southwest Regions, and Pa-cific Northwest and Pacific SouthwestResearch Stations

o USDI Bureau of Land Management

o USDI National Park Service

o USDI Fish and Wildlife Service

o USDI Geological Survey

o U.S. Environmental Protection Agency

o USDC National Marine Fisheries Service

o U.S. Army Corps of Engineers

o USDI Bureau of Indian Affairs

o USDA Natural Resources ConservationService

Northwest Forest Plan Land Alloca-tions, Standards, and Guidelines

The heart of the NFP is in the system of reserves,actively managed lands (matrix), and operationalguidelines it lays out. The key features are:

• Matrix lands–Making up about 16 percent ofthe total NFP acreage, matrix lands are wheremost active timber harvest is expected tooccur. Within managed areas, special stan-dards and guidelines protect elements of late-successional and old-growth forests to provideconnectivity for old-forest-dependent organ-isms across the landscape, and provide refu-gia for them during the time that managedareas are in earlier successional stages.These standards include leaving 15 percentof mature green trees in managed areas andspecific requirements for protection of snagsand large down wood.

• Late successional reserves (LSRs)–About 7.5million acres (31 percent of the total) of federalforest lands within the NFP area are in largereserves within which late-successional andold-growth forest structures and processesare maintained and protected. The purpose ofLSRs is to provide large blocks of habitat forspecies associated with old forests. The LSRs

Figure 1–Federal land allocations within the range of thenorthern spotted owl.

89

are scattered throughout the NFP area suchthat organisms can disperse among themacross the entire area (standards for interven-ing matrix areas help provide connectivityacross managed lands). Significant acreage inLSRs is currently in younger stands (formerplantations); up to age 80, silvicultural treat-ments may be used to foster old-growth standstructures and habitats. Once a stand reachesage 80, no further management typically oc-curs.2 All activities (for example, recreation) inLSRs must be neutral or beneficial with regardto late-successional and old-growth qualities.When LSRs are combined with other types ofreserves (congressionally and administrativelydesignated areas, such as wilderness areasand national parks), the resulting acreage isapproximately 78 percent of the total NFParea.

• Aquatic conservation strategy (ACS)–TheNFP provides for the viability of native fishstocks and overall riparian values through theACS. There are four components to this ele-ment of the NFP:

o Riparian reserves–Similar in policy andintent to LSRs, riparian reserves consistof buffers of varying widths (generally 100to 300 feet) adjacent to all streams andwetlands in the matrix. About 11 percentof the total NFP acreage is in riparianreserves. Besides providing benefits toboth riparian and aquatic habitats andprocesses, riparian reserves also providelate-successional habitat connectivityacross the matrix between LSRs.

o Key watersheds–Selected watershedswithin the NFP area are designated as“key watersheds,” with special require-ments for the protection of at-risk fishstocks and water quality.

o Watershed analysis–A “systematic proce-dure to characterize the aquatic, riparian,and terrestrial features within a water-shed,” used to “refine riparian reserveboundaries, prescribe land managementactivities, including watershed restoration,and develop monitoring programs” (USDAUSDI 1994).

o Restoration–Special emphasis has beenplaced on proactively identifying and re-storing degraded aquatic and ripariansites (e.g., road crossings with inad-equate culverts or stream reaches withinsufficient large instream wood.

• Monitoring and adaptive management–Aprime goal of the NFP is to learn from experi-ence, and adjust the plan when appropriate.This cycle of doing, learning, and adjusting istermed “adaptive management.” Much ofadaptive management application occurs in-formally and opportunistically. Two “institution-alized” facets of adaptive management in theNFP are:

o Monitoring–The NFP calls for monitoringat multiple scales for an array of factors.The most formalized monitoring occursat the NFP area-wide scale, in ongoing,systematic assessments of northern spot-ted owl populations, marbled murrelets(Brachyramphus marmoratus), aquaticand riparian conditions, late-successionaland old-growth habitats, and adherenceto NFP guidelines in implementation ofland management activities. Monitoringfor biological diversity, socioeconomicfactors, and tribal relations is beingplanned as well.

o Adaptive management areas (AMAs)–Ten areas were designated AMAs in theNFP, constituting about 6 percent of thetotal acreage. The AMAs were geographi-cally located to capitalize on opportunitiesto develop partnerships (“collaborativestewardship”) with local communities, andwere intended to be places where experi-mentation outside the matrix standardsand guidelines would be allowed (to testthe adequacy of the standards and guide-lines) and where new management ap-proaches could be developed and tested.

2 There may be exceptions to this rule on a case-by-casebasis, if needed, to reduce risk to old-growth characteristics.For example, without thinning or other mechanical treatmentsfor fuels reduction and enhancement of stand structure, somestands on the eastern flank of the Cascades with a high-frequency, low-intensity natural fire regime may be at risk ofstagnation, insect and disease outbreaks, or destruction bywildfire. Conceivably, treatments after age 80 could beallowed in these stands; however, such exceptions have notbeen widely sought.

90

• Survey and manage (S&M) mitigation–Whenthe consequences of implementing all thecomponents listed thus far were evaluated, itwas determined that there were still approxi-mately 400 species of old-growth-associatedplants, animals, and fungi whose persistencewas still uncertain. For many of the species,lack of information on distribution and abun-dance was a major factor in their being in-cluded. This is not surprising, as the groupcontains many rather obscure nonvascularplants, fungi, invertebrates, and arthropods, inaddition to the better known flora and fauna.The NFP sets out a special set of rules man-dating surveys for S&M species prior to imple-menting management activities, and variousdegrees of protection, depending on theneeds. Provision was made for S&M speciesto be evaluated periodically as new informa-tion is obtained, with the goal of eventuallyeither determining that species should nothave been included in the first place, or pro-viding an adequate plan of protection.

Has the NFP Succeeded?

For the most part, it is too early to tell whether theNFP has succeeded in meeting its goals. Cer-tainly, the harvest levels envisioned have not beenachieved, largely because of recent lawsuits, lackof political support for harvest of old-growth for-ests on matrix lands, and other factors. Results ofmonitoring efforts for species and aquatic/ripariansystems are still several years away. Agencies arestill struggling with adaptive management, both asa concept, and in the context of the AMAs. On theother hand, the injunction against timber harvestwas lifted, the agencies involved are working to-gether proactively and collaboratively, and com-munities are increasingly involved in managementdecisions.

In 1998, the Interagency Steering Committee(ISC) commissioned a review of the NFP (Pipkin1998). The report identified several elements ofthe NFP as “useful and proven models” worthy ofinclusion in future assessments:

• A common regional “vision” for forests and thetimber economy that all participants can worktoward.

• “Institutionalized” collaboration (standing com-mittees and chartered groups with specificmandates) among agencies and the public.

• Provision for direct participation by nonfederalgovernments including tribes.

• Consistent data collection and mappingacross ownerships.

• Specific inclusion of research and monitoringfor decisionmaking and adaptive manage-ment.

Challenges

The NFP was the first plan of its kind to attemptsuch a comprehensive, ecologically based effortwith diverse agencies on a large land base. Asplan implementation has proceeded, questionsand issues have emerged that present significantchallenges to the interagency partners.

At the head of the list of compelling questions isthat of meshing ecosystem-based and single-species approaches to conserving late-succes-sional and old-growth biological diversity. Exceptfor the survey and manage component, the NFPassumes that providing for adequate habitat, con-nectivity, and ecological processes across thelandscape will protect biological diversity. TheFEMAT group viewed this assumption as a hy-pothesis that would be tested through monitoring,with subsequent plan adjustments if necessary.3

However, an experimental approach to land man-agement presumes that some degree of risk isacceptable, a point around which there clearly islack of consensus. On the other hand, attemptingto manage a large set of cryptic species at a verylow level of risk is far beyond the resources of anyof the agencies; in fact, it is probably an unattain-able goal to begin with. Strategies that minimizerisk to individual species by focusing on providinghabitat may be the only practical solution tobiodiversity conservation concerns. How to createsuch strategies that are scientifically sound andlegally defensible remains a productive avenue forfurther exploration.

3 To date, a framework for testing this hypothesis has notbeen developed.

91

The need to work at multiple scales was explic-itly stated as a goal for the NFP (FEMAT 1993).However, technological tools for sliding up anddown the continuum of scales of space and timehave been slow in coming. Such tools would haveto provide the ability to both scale up (for ex-ample, assemble the cumulative effects of finerscale phenomena) and scale down (disaggregatelarger scale phenomena into smaller temporally orspatially explicit fragments). Currently, the NFP isbeing implemented at several scales (stand, wa-tershed, landscape, province, region), but it isoften impossible to translate data, findings, oreffects to larger or smaller scales because ofinconsistencies in methodologies, assumptions,and so on.

There is an apparent tension between what I willterm static versus dynamic approaches to de-fining target conditions for a variety of ecosystemelements within the NFP. The static approach ismanifested in the creation of rules, standards, andguidelines that set a threshold or narrow range ofconditions that must be achieved and maintained.This approach often has the terms “regulatory” or“statutory” applied to it. In contrast, the dynamicapproach is evidenced in the notion of adaptivemanagement (experimental management with aspecific goal of learning and potentially adjustingpractices), and in the idea that because ecosys-tems are themselves dynamic, a wider range ofconditions will naturally (and should be allowed to)exist through time. Both approaches have a placein the NFP, but the extent to which each applies tothe various components of the NFP has yet to beresolved. The main difference between the ap-proaches is the perception of risk associated withvariation.

One of the most problematic examples of thetension between the two approaches has been inadaptive management areas. Although AMAswere delineated with the express purpose ofevaluating the assumptions of the NFP throughtesting a wide range of practices (some of whichmay fall outside standards and guidelines), landmanagers have been understandably reluctant toventure outside the prescriptions of the NFP be-cause of the very real threat of appeals, lawsuits,and other forms of resistance from outside enti-ties that may lack confidence in managers’ abili-ties to meet broad NFP goals. There are also

differing points of view among the agencies re-garding how much and what type of experimenta-tion is appropriate within AMAs.

Another example is fire-prone late-successionalreserves on the east flank of the Cascade crest,where managers have hesitated, again becauseof the risk of appeals and lawsuits, to seek ex-emptions from LSR guidelines in order to imple-ment prescriptions that may reduce the risk ofstand damage from fire and insects. Both of theseexamples illustrate that application of a static setof rules to an inherently dynamic situation mayactually increase risk in the long run (through lostopportunities for learning and improved manage-ment in the case of AMAs, and through increasedthreat to old-growth values in LSRs).

Finally, because the NFP applies exclusively tofederal lands, the role of nonfederal lands, es-pecially private lands, in the overall ecological,social, economic, and political context of the re-gion cannot be fully assessed. It is especially diffi-cult to understand how private lands contribute tothe overall whole when large elements are un-regulated and landowner behavior is difficult topredict.

The federal agencies involved in the NFP haveset in motion numerous efforts to resolve theseand other issues. The NFP has not proved to bethe ultimate solution to conflicts over the use andmanagement of federal forest lands in the PacificNorthwest. Although the agencies have achievedmajor successes, small and large, the challengeslisted above, combined with the continuing needto build public trust, have constituted real barriers.The NFP’s strength lies in the strong science un-derpinnings on which it is founded, in the scope ofthe ecological, social, and political issues it at-tempted to address, and in its built-in flexibility toadapt and improve. It remains to be seen whetherthe agencies and public together can capitalize onthese strengths to overcome the difficulties citedhere and realize the NFP’s promise of protectionfor old-growth ecosystems while contributing toregional economic health.

Metric Equivalents

1 foot = 0.304 meters

1 acre = 0.405 hectares

92

Literature Cited

Forest Ecosystem Management AssessmentTeam [FEMAT]. 1993. Forest ecosystem man-agement: an ecological, economic, and socialassessment. Portland, OR: U.S. Departmentof Agriculture; U.S. Department of the Interior[et al.]. [Irregular pagination].

Johnson, K.N.; Swanson, F.J.; Herring, M.;Greene, S. 1999. FEMAT assessment–casestudy. In: Johnson, K.N.; Swanson, F.J.; Her-ring, M.; Greene, S., eds. Bioregional assess-ments. Washington, DC: Island Press: 87-116.

Pipkin, J. 1998. The Northwest Forest Plan revis-ited. Washington, DC: U.S. Department of theInterior, Office of Policy Analysis. 117 p.

U.S. Department of Agriculture, Forest Serv-ice; U.S. Department of the Interior, Bureauof Land Management. 1994. Record of deci-sion for amendments to Forest Service andBureau of Land Management planning docu-ments within the range of the northern spottedowl. [Place of publication unknown.] 74 p. [Plusattachment A: standards and guidelines].

93

Abstract

We characterized recent historical and current vegetation composition and structure of a representativesample of subwatersheds on all ownerships within the interior Columbia River basin and portions of theKlamath and Great Basins. For each selected subwatershed, we constructed historical and current veg-etation maps from 1932 to 1966 and 1981 to 1993 aerial photos, respectively. Using the raw vegetationattributes, we classified and attributed cover types, structural classes, and potential vegetation types toindividual patches within subwatersheds. We characterized change in vegetation spatial patterns using asuite of class and landscape metrics, and a spatial pattern analysis program. We then translated changein vegetation patterns to change in patterns of vulnerability to wildfires, smoke production, and 21 majorforest pathogen and insect disturbances. Results of change analyses were reported for province-scaleecological reporting units (ERUs). Here, we highlight significant findings and discuss management impli-cations.

Twentieth-century management activities significantly altered spatial patterns of physiognomies, covertypes and structural conditions, and vulnerabilities to fire, insect, and pathogen disturbances. Forest landcover expanded in several ERUs, and woodland area expanded in most. Of all physiognomic conditions,shrubland area declined most due to cropland expansion, conversion to semi- and non-native herblands,and expansion of forests and woodlands. Shifts from early to late seral conifer species were evident inforests of most ERUs; patch sizes of forest cover types are now smaller, and current land cover is morefragmented. Landscape area in old multistory, old single story, and stand initiation forest structures de-clined with compensating increases in area and connectivity of dense, multilayered, intermediate foreststructures. Patches with medium and large trees, regardless of their structural affiliation are currentlyless abundant on the landscape. Finally, basin forests are now dominated by shade-tolerant conifers,and exhibit elevated fuel loads and severe fire behavior attributes indicating expanded future roles ofcertain defoliators, bark beetles, root diseases, and stand replacement fires. Although well intentioned,20th-century management practices did not account for landscape-scale patterns of living and dead veg-etation that enable forest ecosystems to maintain their structure and organization through time, or for thedisturbances that create and maintain them. Improved understanding of change in vegetation spatialpatterns, causative factors, and links with disturbance processes will assist managers and policymakersin making informed decisions about how to address important ecosystem health issues.

Recent Changes (1930s–1990s) in Spatial Patterns of InteriorNorthwest Forests, USA1

P.F. Hessburg,2 B.G. Smith,3 R.B. Salter,2 R.D. Ottmar,4 and E. Alvarado5

1 Hessburg, P.F.; Smith, B.G.; Salter, R.B.; Ottmar, R.D.; Alvarado, E. 2000. Recent changes (1930s–1990s) in spatialpatterns of interior Northwest forests, USA. Forest Ecology and Management. 136: 53-83. Reprinted with permission fromElsevier.

2 Research ecologist, (Hessburg) and geographer (Salter), U.S. Department of Agriculture, Forest Service, Pacific NorthwestResearch Station, Forestry Sciences Laboratory, 1133 N Western Ave., Wenatchee, WA 98801; Tel: 509-662-4315; e-mail:phessburg@fs.fed.us, bsalter@fs.fed.us.

3 Quantitative ecologist, U.S. Department of Agriculture, Forest Service, 1230 NE 3rd St., Suite 262, Bend, OR 97701;Tel: 541-383-4023; e-mail: bgsmith01@fs.fed.us.

4 Research forester, U.S. Department of Agriculture, Forest Service, Pacific Northwest Research Station, Forestry SciencesLaboratory, 400 N 34th, Suite 201, Seattle, WA 98103; Tel: 206-732-7828; e-mail: rottmar@fs.fed.us.

5Research forester, U.S. Department of Agriculture, Forest Service, Pacific Northwest Research Station, Forestry SciencesLaboratory, 400 N 34th, Suite 201, Seattle, WA 98103; Tel: 206-732-7842; e-mail: ealvarado@fs.fed.us.

94

Landscape Methodology

95

Introduction

Several broad-scale ecoregional assessmentshave been conducted in the United States in re-cent years (Everett et al. 1994, FEMAT 1993,SNEP 1997, USDA FS 1996), and several moreare in progress. Each assessment has used anow well-standardized approach to define theanalytical problem. A scoping process was usedto identify and evaluate critical issues deservingconsideration. A needs assessment was per-formed to identify data requirements and analyti-cal methods needed to respond to the issues. Awide array of statistical, simulation, and optimiza-tion procedures has been used to address variouscomponents of the overall assessment problem.To assert that the suite of analyses used in theseassessments was conducted ad hoc would be adisservice. However, it is fair to say that, althoughassessment teams may have carefully coordi-nated the conduct of analyses with integration inmind, there is little evidence in the reports thateffectively integrated analysis was achieved.

Landscape Analysis With EcosystemManagement Decision Support

The USDA Forest Service Pacific Northwest Re-search Station released the first production ver-sion of the ecosystem management decisionsupport (EMDS) system in February 1997. TheEMDS integrates a knowledge-base engine intothe ArcView3 (Environmental Systems Research

Institute, Redlands, CA) geographic informationsystem (GIS) to provide knowledge-based rea-soning for landscape-level ecological analyses(fig. 1). Major components of the EMDS systeminclude the NetWeaver knowledge-base system,the EMDS ArcView application extension, and theassessment system (Reynolds 1999a, 1999b;Reynolds et al. 1996, 1997a, 1997b).

A knowledge base is a formal logical specificationfor interpreting information and is therefore a formof meta database in the strict sense (Jackson1990, Waterman 1986). Interpretation of data bya knowledge-base engine (a logic processor)provides an assessment of system states andprocesses represented in the knowledge base astopics. Use of logical representation for assessingthe state of systems frequently is desirable ornecessary. Often, the current state of knowledgeabout a problem domain is too imprecise for sta-tistical or simulation models or optimization, eachof which presumes precise knowledge about rel-evant mathematical relations. In contrast, knowl-edge-based reasoning provides solutions forevaluating this more imprecise information, andsome knowledge-based systems can provideuseful analyses even in circumstances in whichdata are incomplete.

Knowledge-based solutions are particularly rel-evant to ecosystem management because thetopic is conceptually broad and complex, involvingnumerous, often abstract, concepts (e.g., health,sustainability, ecosystem resilience, ecosystemstability, etc.) for which assessment depends onnumerous interdependent states and processes.Logical constructs are useful in this context be-cause the problem can be evaluated as long asthe entities and their logical relations are under-stood in a general way and can be expressed bysubject matter authorities.

Analysis and Communication of Monitoring Data WithKnowledge-Based Systems

Keith M. Reynolds1 and Gordon H. Reeves2

1 Research forester, Pacific Northwest Research Station,Forestry Sciences Laboratory, 3200 SW Jefferson Way,Corvallis, OR 97331; Tel: 541-750-7434; e-mail:kreynolds@fs.fed.us.

2 Research fisheries biologist, Pacific Northwest ResearchStation, Forestry Sciences Laboratory, 3200 SW JeffersonWay, Corvallis, OR 97331; Tel: 541-750-7314; e-mail:greeves@fs.fed.us

3 The use of trade of firm names in this publication is forreader information and does not imply endorsement by theU.S. Department of Agriculture of any product or service.

96

Nestucca Basin Example: Evaluationof Salmon Habitat Suitability

The Nestucca basin is a 5th-code hydrologic unitlocated in the northern portion of the OregonCoast Range province (fig. 2) and contains 45true and composite 6th-code watersheds. In 1998,a watershed assessment team from the SiuslawNational Forest rated each true 6th-code water-shed in the basin with respect to its suitability forproviding salmon habitat, following the rating sys-tem of the decision matrix of the National MarineFisheries Service (NMFS). In subsequent discus-sions with the authors, assessment team mem-bers expressed a number of reservations aboutthe approach that had been used for rating water-shed condition.

To improve on the basic approach to watershedevaluation for salmon habitat suitability as re-

quired to implement the Aquatic ConservationStrategy of the Northwest Forest Plan, the authorsdesigned a prototype knowledge base suitable forapplication to true 6th-code watersheds in the Or-egon Coast Range province (fig. 2). Starting withbasic requirements identified in the NMFS matrix,Reeves summarized logical requirements for theknowledge base. The authors then worked to-gether to implement the knowledge-base designin NetWeaver, and then reviewed the initial designwith the assessment team to make refinements,corrections, and additions to the initial version.

The most basic outputs from an EMDS analysisare maps of evaluated indices for topics includedin an analysis. In our example, the primary map ofinterest (fig. 3) displays the evaluated index forsalmon habitat suitability for each 6th-code water-shed in the basin. Composite watersheds, forwhich the knowledge base was not designed, are

Figure 1—Architecture of EMDS system. The acronym, GUI, indicates a graphic user interface through which a userinteracts with components of the EMDS application. The assessment system uses a custom GUI to manage com-munication between the NetWeaver logic engine and code internal to the EMDS ArcView extension. ESRI = Environ-mental Systems Research Institute.

97

Figure 2—The Nestucca basin is in the centralOregon Coast Range.

Figure 3— Salmon habitat suitability in 6th-code watersheds in the Nestucca basin.Patterned features are composite watersheds for which the knowledge base is notappropriate. The scale expresses strength of support for the proposition that thebiophysical condition of a watershed provides suitable salmon habitat. The extremes ofthe scale, -1 and 1, indicate no support and full support, respectively. Intermediate valuesexpress some degree of support.

98

displayed with a lined pattern. All information thatis logically antecedent to topics selected for analy-sis also must be evaluated by the logic engineand so is displayable either in maps, graphs, or intabular output generated by the logic engine.

Maps characterizing landscape condition are pow-erful communication tools, but it is at least asimportant to be able to explain the basis for ob-served results in a clear, intuitive manner. One ofthe most significant features of knowledge-basesystems is their ability to display a logic trace thatexplains the derivation of conclusions. In this ex-ample, the top portion of the browser windowdisplays an expandable outline of knowledge-base structure, and the bottom portion displaysthe logic structure of the topic currently selected inthe outline (fig. 4).

The NetWeaver logic engine reasons with incom-plete information, when needed, by using fuzzymath (Zadeh 1965, 1968), and the engine evalu-ates the influence of any missing information withrespect to its contribution to completeness of ananalysis. Input data for the Netsucca basin analy-sis contained a few missing observations on datafields (fig. 5). However, the engine implementsevidence-based reasoning, producing partialevaluations of topics with incomplete information(figs. 3 and 4). The knowledge base is, in effect, amental map of the problem. Its structure containsinformation that can be used to compute the influ-ence of missing information (fig. 6). The computa-tion of influence is based on the level at whichinformation enters the logic structure, how manyreferences there are to it, and how frequently theassociated database field contains missing obser-vations. The assessment system in EMDS alsoincludes tools to optionally synthesize informationabout data influence and the logistics of acquiringdata to prioritize data needs (Reynolds 1999a).

Finally, the logic engine generates tabular outputfrom which an analyst can summarize additionalinformation such as the frequency distribution ofindex values (fig. 7) or the frequency with whichdata substantially contribute to a particular conclu-sion (figs. 8 and 9). Frequency distributions ofindices provide a simple synoptic view of condi-tions over an assessment area at a point in time(fig. 7), and useful statistical inferences about theefficacy of restoration programs, for example,could be drawn from comparisons of such fre-quency distributions over consecutive assess-

ment times. Similarly, more detailed summaries ofconditions in an assessment area (figs. 8 and 9)provide useful background information for designof restoration programs as well as a basis formore detailed statistical inferences about the effi-cacy of these programs.

Identifying Restoration Objectives andCriteria

Problem specification for ecological assessmentmay well deserve to be classified as a wickedproblem (Allen and Gould 1986). Evaluation moni-toring must consider potentially numerous statesand processes of biophysical, social, and eco-nomic components of an ecosystem. Many enti-ties may have both deep and broad networks oflogical dependencies as well as complex intercon-nections. Although constructing such complexrepresentations in NetWeaver is not trivial, it is atleast rendered feasible by the precision and com-pactness of fuzzy logic relations, and by thegraphic, object-based representation of logic net-works in the system interface. No subject-matterauthority, nor for that matter any group of authori-ties, is capable of holding a comprehensive cogni-tive map of such a complex problem domain intheir consciousness. On the other hand, thegraphic, object-based form of knowledge repre-sentation in NetWeaver is highly conducive to theincremental evolution of knowledge-base designfrom simple to complex forms.

Two essential aspects of the knowledge-basedapproach to evaluation monitoring that we havedescribed and illustrated are its goal orientationand use of formal logic in problem specification.The implications for realizing truly integratedanalysis are significant. Knowledge-base designbegins with identification of the questions (formu-lated as propositions) that the analysis ultimatelyis designed to address. Specification proceeds byidentifying lines of reasoning that decompose theoriginal propositions into progressively more con-crete ones until links to data are identified. Theresult, effectively, is a mental map of the problem,including not only identification of all topics perti-nent to the problem, but their logical interdepen-dencies. The final specification thus not onlyprovides logical links between data requirementsand the original motivating questions, but providesa specification for how the data are to be inter-preted, taking into account interrelations amongpieces of information.

99

Figure 4—The knowledge base with the EMDS hotlinkbrowser tool displays the state of the knowledge basefor features (e.g., watersheds) selected in the EMDSAssessment View.

Figure 5—Frequency of missing data fields in the input data table for the Nestucca basin analysis.

100

Figure 6—Influence of missing information in the Nestucca basin analysis calculated by the NetWeaver logic engine.

Figure 7—Frequency distribution of the salmon habitat suitability index for the Nestucca basin analysis. Truth value interval isa measure of the degree of support that data provide for a proposition. This index also can be interpreted as a measure ofsuitability.

101

Figure 8—Frequency with which data in high-gradient drainages contributed to the conclusion of a compromisedor degraded biophysical condition. Drainages with an average reach gradient greater than 4 percent wereclassified as high gradient; in-channel conditions were not available.

Figure 9—Frequency with which data in low-gradient drainages contributed to the conclusion of a compro-mised or degraded biophysical condition. Drainages with an average reach gradient less than or equal to4 percent were classified as low gradient; data on in-channel conditions were available.

102

Performing Evaluations WithIncomplete Information

It is probably fair to say that almost all significantmonitoring and assessment programs start outwith significant data gaps. The NetWeaver logicengine is particularly well suited for use in applica-tions such as evaluation monitoring in particularand ecosystem analysis in general because it isable to reason with incomplete information andstill provide meaningful results, and because itcan evaluate the influence of missing information.

We noted earlier that the number of topics in anyreasonably realistic problem and the number ofinterdependencies among them can easily over-whelm the capacity of the human brain to reasoneffectively about such questions as, What is themost useful information that I could acquire at thispoint in time to improve the completeness of myanalysis? It is important to appreciate that theinfluence of information is highly dynamic: theinfluence of what is missing depends both on theobservations that are currently available, and onthe values in those observations. Consequently,use of iterative analysis of data influence in EMDSfollowing successive, incremental additions ofnew data could conceivably reduce resourcesdevoted to data collection on the order of 50 per-cent in typical applications.

In our very small example, the distributions ofmissing data (fig. 5) and data influence (fig. 6)mirror each other. More generally, and particularlyfor larger and more complex logic models, the twodistributions may be very different owing to inter-dependencies among data elements. Even in thissimple case, however, it is clear that missing dataon water temperature are the largest source ofuncertainty in the current analysis (fig. 6). It also isquite possible that if some of the missing watertemperature observations were acquired, poolquality, fines, and farm use might no longer havenonzero influence.

Explanation of Monitoring Results

The ability to explain monitoring results to col-leagues, line officers, interested publics, etc. is atleast as important as the ability to perform ananalysis in the first place. If the logic of underlyingmodels cannot be presented in clear, unambigu-ous terms to such audiences, results will not be

trusted. Powerful and sophisticated linear pro-gramming systems such as FORPLAN largelyfailed on this account.

Our experience to date with EMDS, and with itsknowledge-base browser interface in particular,has been that the presentation of results fromlogic models is easy to grasp at an intuitive levelso the results are accessible to broad audienceswith only limited explanation. An analyst can easilyrun the browser interactively in front of an audi-ence, navigating the structure of the evaluatedknowledge base, and explaining the derivation ofthe logic as he or she goes. The earlier logic dia-gram (fig. 4) provides a good example. Compositewatersheds (WS type = 1) are screened out of theanalysis with a logic switch and remain in an un-determined logic state. For true watersheds, theright-hand path under the WS type logic switch isfollowed. Biophysical condition of a true water-shed evaluates as suitable to the degree thatthere is suitable riparian vegetation and a suit-able upland environment. A second logic switchchecks whether the data for the watershed in-clude an evaluation of in-channel conditions.If a low-gradient stream survey was performed(lowGradSurvey > 0), then the evaluation includesthe upland, in-channel, and riparianVeg topics;otherwise the evaluation of biophysical conditiononly depends on the upland and riparianVeg top-ics. Each of the three topics, upland, in-channel,and riparianVeg, has its own logic specification(not shown).

Evaluation Within Larger Contexts

Almost all landscape analyses are performedwithin the context of some broader scale. Ourknowledge base for evaluation of salmon habitatsuitability is a prototype constructed for the spe-cific context of the Oregon Coast Range. Conse-quently, it is unlikely that the current version issufficiently general for application throughout theplanning region of the Northwest Forest Plan,which encompasses the range of the northernspotted owl (Strix occidentalis caurina).

Fortunately, it is not difficult to generalize proto-types such as ours for broader applicability. Insome cases, for example, it may be sufficient toreplace hard-wired arguments with a calculatedone that contains a more general fuzzy member-ship function whose parameters vary with addi-tional context information. For more dramatic

103

variations in logic structure, it may be necessaryto add new logic switches that similarly selectalternative logic pathways based on additionalcontext information. In a technical sense, knowl-edge-base generalization is a relatively minorissue. On the other hand, it leads naturally to dis-cussion of the more interesting question ofmultiscale implementations.

Extending Ecosystem ManagementDecision-Support Applications toMultiple Spatial Scales

It is relatively easy in principle to extend integratedanalysis via knowledge-based reasoning overmultiple spatial scales (fig. 10). Data from fine-scale landscape features such as 6th-code water-sheds are first processed by a knowledge basedesigned for that scale. Knowledge-base output,shown as evaluated states in the middle of thefigure, then go through an intermediate filter (typi-

cally implemented in a spreadsheet or databaseapplication) to synthesize information for input tothe next coarser scale. A second knowledge baseprocesses the synthesized information to providean assessment of landscape-level attributes atthe top of the figure. Finally, knowledge-base out-puts at the broader landscape scale may feedback to the fine scale as context information thatinfluences evaluations at the fine scale. Thissimple conceptual model provides the basis for aformal logical specification of analyses that areconsistent across scales. Hierarchies, or evennetworks, of knowledge-based analyses as sug-gested would be highly consonant with ecosystemtheories concerning the hierarchical organizationecosystems.

English Equivalents

1 kilometer = 0.62 miles

Figure 10—An example of knowledge-basedintegration across spatial scales.

104

Literature Cited

Allen, G.M.; Gould, E.M., Jr. 1986. Complexity,wickedness, and public forests. Journal ofForestry. 84: 20-23.

Everett, R.; Hessburg, P.; Jensen, M.;Bormann, B. 1994. Eastside forest ecosystemhealth assessment. Volume I: executive sum-mary. Gen. Tech. Rep. PNW-GTR-317. Port-land, OR: U.S. Department of Agriculture,Forest Service, Pacific Northwest ResearchStation. 61 p.

Forest Ecosystem Management AssessmentTeam [FEMAT]. 1993. Forest ecosystem man-agement: an ecological, economic, and socialassessment. Portland, OR: U.S. Departmentof Agriculture, Forest Service; U.S. Departmentof the Interior, Bureau of Land Management;[et al.]. [Irregular pagination].

Jackson, P. 1990. Introduction to expert systems.Reading, MA: Addison-Wesley Publishers.526 p.

Reynolds, K.; Cunningham, P.; Bednar, L.[et al.]. 1996. A design framework for a knowl-edge-based information management systemfor watershed analysis in the Pacific NorthwestU.S. AI Applications. 10: 9-22.

Reynolds, K.M. 1999a. EMDS users guide (ver-sion 2.0): knowledge-based decision supportfor ecological assessment. Gen. Tech. Rep.PNW-GTR-470. Portland, OR: U.S. Depart-ment of Agriculture, Forest Service, PacificNorthwest Research Station. 63 p.

Reynolds, K.M. 1999b. NetWeaver for EMDSuser guide (version 1.1): a knowledge basedevelopment system. Gen. Tech. Rep. PNW-GTR-471. Portland, OR: U.S. Department ofAgriculture, Forest Service, Pacific NorthwestResearch Station. 75 p.

Reynolds, K.M.; Saunders, M.; Miller, B. [etal.]. 1997a. An application framework for deci-sion support in environmental assessment. In:GIS 97 Proceedings of the eleventh annualsymposium on geographic information sys-tems. Washington, DC: GIS World: 333-337.

Reynolds, K.M.; Saunders, M.; Miller, B. [etal.]. 1997b. Knowledge-based decision sup-port in environmental assessment. In: Pro-ceedings 1997 ACSM/ASPRS annual con-vention and exposition. Bethesda, MD: Ameri-can Society for Photogrammetry and RemoteSensing: 344-352. Vol. 4

Sierra Nevada Ecosystem Project [SNEP].1997. Final report to Congress, addendum.Davis, CA: University of California, Centers forWater and Wildland Resources. 328 p.

U.S. Department of Agriculture, Forest Ser-vice. 1996. The Southern Appalachian assess-ment summary report. Tech. Pap. R8-TP-25.Atlanta, GA: Southern Region [Region 8].118 p.

Waterman, D.A. 1986. A guide to expert systems.Reading, MA: Addison-Wesley Publishers.419 p.

Zadeh, L.A. 1965. Fuzzy sets. Information andControl. 8: 338-353.

Zadeh, L.A. 1968. Probability measures of fuzzyevents. Journal of Mathematical Analysis andApplications. 23: 421-427.

105

Introduction

The appearance of forested landscapes and indi-vidual stands after forest management activitiesis critical to public acceptance of these activities.Even with thorough planning, detailed site-specificanalysis, and careful monitoring, many manage-ment activities will not be truly successful if thepublic views the resulting landscape as an eye-sore. Unfortunately, many people judge the suc-cess or failure of management activities based onthe visual impact of the activity to an otherwise“natural” landscape.

Forestry professionals have long used visualiza-tion techniques to address a variety of forestmanagement problems. Prior to the advent ofcomputerized methods, they used “artists’ rendi-tions” to communicate the effects of land man-agement activities. Perspective sketches andscale models continue to help communicate thespatial arrangement and extent of managementactivities to the lay public. However, current man-agement practices use more detailed silviculturalprescriptions involving small treatment areas scat-tered over larger landscapes and the removal ormodification of specific stand components. Inaddition, treatment regimes are designed toachieve a desired condition over long timespansby using a series of treatments. Given the broadrange of treatments considered and changes tostand conditions over time, it is difficult to producetraditional artists’ renditions that reflect the ex-pected condition accurately and with a minimumof artist-contributed bias. In addition, the timerequired to produce artists’ renditions limits theiruse when depicting several alternative treatmentstrategies.

Foresters charged with selecting stands for treat-ment and designing silvicultural prescriptions of-ten find it difficult to comprehend the complex

spatial and temporal interactions that occuracross landscapes. Traditional work methodsinvolving fieldwork, maps, and aerial photographsprovide enough information to assess individualtreatments but may not provide adequate informa-tion to fully evaluate the cumulative impact oftreatments or the impact of treatments imple-mented over time. Computer-based landscapesimulations are a recognized tool for assessingthe potential visual impact of land use decisionsand management activities. Visualization toolsthat include a stand projection component or pro-vide linkages to such models can simulate anddepict stand and landscape changes over time.Such presentations help to communicate standand landscape conditions and how these condi-tions change as a result of management activities,natural disturbances, and growth over time. Dur-ing the treatment design process, visualizationsdepicting treatments within their landscape con-text provide important feedback. Such feedbackcan help resource specialists develop and imple-ment better landscape management plans. Fur-thermore, visualizations help communicatemanagement activities to other resource special-ists and public stakeholders.

Overview of Visualization Techniques

Computerized visualization methods range fromsimple diagrams to complete virtual realities. Fourmethods, described in detail and compared inMcGaughey (1998), are commonly used to pro-duce visual representations of forest operations:

• Geometric modeling

• Video imaging

• Geometric video imaging

• Image draping

Geometric modeling methods build geometricmodels of individual components (ground surface,trees, other plants, and structures) and then as-semble the component models to create a scenethat represents a forest stand or landscape.

Seeing the Forest and the Trees: Visualizing Stand andLandscape Conditions

Robert J. McGaughey1

1 Research forester, U.S. Department of Agriculture, ForestService, Pacific Northwest Research Station, University ofWashington, P.O. Box 352100, Seattle, WA 98195-2100; Tel:(206) 543-4713; e-mail: mcgoy@u.washington.edu,rmcgaughey@fs.fed.us

106

Video imaging uses computer programs to modifyscanned full-color video or photographic imagesto represent changes to stand and landscapeconditions. To portray a wide range of landscapeconditions, video-imaging techniques use a libraryof images that represent different forest condi-tions to replace portions of an original image.

A hybrid approach, called geometric video imag-ing by this author, combines geometric modelingand video-imaging techniques to produce realisticimages that accurately represent data describingthe effects of forest management activities. Op-erators use geometric modeling to produce im-ages that specify the location, arrangement, andscale of stand or landscape features. These im-ages then guide video-imaging manipulations thatmodify a photographic image to reflect proposedstand or landscape changes. By combining theprecise spatial location capabilities of geometricmodeling and the photo-manipulation capabilitiesof video imaging, hybrid methods result in spa-tially accurate, photo-quality images. However,hybrid techniques, like video imaging, requireextensive libraries of tree and stand images torepresent an appropriate range of species, treesizes, growth forms, landscape positions, andtreatment options.

Image draping mathematically “drapes” an imageover a digital terrain model to create a texturedsurface. The draped image is typically a satellitescene, aerial photograph, orthophoto, or scannedmap sheet.

Of the four methods, geometric modeling pro-vides the most consistent link between data de-scribing stand and landscape conditions andfeatures in a generated image. In general, there isa one-to-one relationship between data elementsand objects in the final image. Geometric model-ing produces images that are generally less real-istic than images produced from photographs.However, if photographic icons are used to repre-sent individual trees and other objects, geometricmodeling can produce images that contain manyof the details normally associated with photo-graphs.

Visualization of Stand/Plot-ScaleAreas

Foresters use stand tables, showing the relativeabundance of plant species and size classes, andsimple graphs, showing single or multiple stand

attributes as a function of stand age, to presentinformation describing forest stands and standconditions. These communication tools, suitablefor conveying simple data relationships to scien-tific and professional audiences, do not alwaysprovide an easily understood representation of aforest stand or the changes in stand structure thatcan occur after disturbances or natural processes.The Stand Visualization System (SVS) was de-signed to help foresters understand and commu-nicate stand conditions and changes in standconditions resulting from management activitiesor natural processes (McGaughey 1997).

The SVS uses geometric modeling techniques tocreate images depicting stand conditions. As partof the stand construction process, SVS convertstables describing the population of standing trees,down material, and understory shrubs into a list ofindividual stand components, e.g., trees, downlogs, and shrubs. Standing trees and down mate-rial are defined by using the number of trees orlogs per hectare. Understory shrubs can be de-fined by using the number of plants per hectarebut are more commonly defined by using a per-centage of cover. The SVS generates enoughplants to achieve the desired cover. Plant loca-tions are generated by using a variety of spatialpatterns including patterns that mimic natural andplanted stands. The SVS also reads output fromthe Forest Vegetation Simulator (Wykoff et al.1982) and ORGANON (Hester et al. 1989). Theimages produced by SVS, although abstract, pro-vide a readily understood representation of standconditions and help communicate silviculturaltreatments and forest management alternatives toa variety of audiences. The SVS provides thefollowing specific capabilities:

• Display overhead, profile, and perspectiveviews of a stand.

• Differentiate between stand components byusing different plant forms, colors, or othertypes of marking as specified by users.

• Provide tabular and graphical summaries ofinformation represented in a stand image.

• Facilitate the design of silvicultural treatmentsby allowing users to select individual standcomponents and specify treatments.

• Display information describing individual standcomponents as they are selected by the user.

107

Once an SVS tree list has been created, userscan simulate silvicultural treatments by selectingindividual trees or groups of trees and specifyinga treatment action such as removal or pruning.As an alternative, treatment rules such as “re-move 75 percent of the stand basal area frombelow” can be applied to the entire stand. Figure 1shows a 1-hectare stand before and after a simplethinning treatment that removes all of the alder(Alnus Hill.) trees and a portion of the Douglas-fir (Pseudotsuga menziesii (Mirb.) Franco) trees.Figure 2 shows the same stand after a treat-ment that removed all of the alder and six smallpatches containing 30 percent of the Douglas-firbasal area.

Visualization of Landscape-ScaleAreas

A wide array of techniques is available for pre-senting landscape-scale information ranging incomplexity from simple tables and graphs tocolor-coded maps to photo-realistic computervisualizations. Computer visualization techniquesare especially useful given the complexity of standtreatments, the spatial distribution of treatmentunits, and the desire to view landscape conditionsover time. Landscape visualization provides theviewer with an easily identified image of a land-scape complete with topographic features andstand conditions. A new visualization system be-ing developed by the author, EnVision, is de-signed to portray landscape conditions by usingdata describing individual stands. EnVision pro-vides several capabilities not previously availableor not available in the same application. For ex-ample, EnVision can fill individual polygons with atexture derived from an aerial photograph whilerendering individual trees in other polygons.

An EnVision scene is based on a gridded digitalterrain model2 that defines the ground surface.EnVision provides a variety of methods to repre-sent the ground including gridlines and contourlines, with and without hidden area removal, andas a shaded, lighted surface. Surface features

such as roads, streams, and points of interest canbe represented as points, lines, or polygons onthe ground surface or as solid walls or point mark-ers sitting on the ground surface. To representground surface texture, EnVision can fill polygonfeatures with textures derived from aerial photo-graphs. This “synthetic aerial photo” capability canbe used to provide ground texture representingvegetation under a canopy of larger trees that aredrawn as geometric models.

EnVision can use a variety of data to describeconditions within individual stands. The simplestform of stand data consists of the distribution oftree sizes for each species present in a stand.More complex data include individual tree recordsfrom inventory plots that include species, diameterat breast height, total tree height, live crown ratio,crown width, and the number of trees per unitarea represented by the record. Stand data aremerged with data describing the growth form andgeneral crown shape for each tree species toconstruct a geometric model for each tree in astand.

Multiscale Visualization

The ability to visualize individual stands by usingSVS has significantly enhanced a forester’s abilityto understand and communicate stand conditionsand the effects of silvicultural treatments on astand. Through linkages with the Forest Vegeta-tion Simulator (Teck et al. 1996, Wykoff et al.1982), SVS provides visualizations of standgrowth and development. However, understandingthe dynamics within an individual stand does notnecessarily lead to a better understanding of thedynamics of an entire landscape. The complexityof landscape management plans, variety of standtreatments, and time period for which most plansare developed make it difficult to understand andcommunicate the plan’s impact on the landscape.A system that links the design and visualization ofindividual stand treatments with the ability to dis-play the stands within their landscape context isneeded. EnVision provides this linkage by allow-ing the use of an SVS tree list to represent eachstand on a landscape. In operation, EnVision rep-licates the SVS tree list to fill the stand polygon by

2 Gridded digital terrain models that correspond to the U.S.Geologic Survey’s 7.5-minute quadrangle series are availablefor most areas in the United States. Other terrain modeltypes, e.g., TIN models, can be used to generate a griddedmodel by using geograpic information system (GIS) software.

108

using a simple tiling algorithm. This linkage be-tween SVS and EnVision makes it possible todesign spatially explicit treatments within SVS andvisualize that treatment within the landscape con-text. Figure 3 shows a landscape where the 1-hectare stand in figure 2 has been used to fill a21-hectare stand polygon. In this example, SVStile reflection has been used to minimize any pat-tern effects that might result when filling the standpolygon.

Conclusions

This paper has discussed a variety of techniquesthat can help land managers visualize, under-stand, and communicate stand- and landscape-scale management actions. In particular, thispaper has highlighted the SVS and EnVision ap-plications developed by the author.

Figure 1—SVS visualization of a 1-hectare stand (a) before and (b) after a uniform thinning treatment. The thinning removedall of the alder and 25 percent of the Douglas-fir basal area from below.

Figure 2—SVS visualization showing an over-head view of the stand shown in figure 1 butafter a treatment that removed all of the alderand six small patches containing 30 percent ofthe Douglas-fir basal area.

Figure 3—Landscape visualization produced by EnVisionshowing a stand polygon filled by using the SVS stand shownin figure 2. Individual trees have been rendered only in thepolygon of interest. Other stand polygons have been filledwith textures that represent the stand structure.

109

Experience with SVS, UTOOLS/UVIEW land-scape analysis and visualization system (Agerand McGaughey 1997), and Vantage Point proto-type (Bergen et al. 1998) indicates that such visu-alization tools help resource specialists and landmanagers make better decisions. However, suchtools also can produce photo-realistic images thatcan mislead viewers. Practitioners must be care-ful when using such tools that they do not inten-tionally or unintentionally misrepresent existing orexpected conditions (Wilson and McGaughey2000). Failure to do so will seriously limit the use-fulness of these visualization techniques for futureprojects.

English Equivalents

1 hectare (ha) = 2.47 acres

Literature Cited

Ager, A.A.; McGaughey, R.J. 1997. UTOOLS:microcomputer software for spatial analysisand landscape visualization. Gen. Tech. Rep.PNW-GTR-397. Portland, OR: U.S. Depart-ment of Agriculture, Forest Service, PacificNorthwest Research Station. 15 p.

Bergen, S.D.; McGaughey, R.J.; Fridley, J.L.1998. Data-driven simulation, dimensionalaccuracy and realism in a landscape visualiza-tion tool. Landscape and Urban Planning. 40:283-293.

Hester, A.S.; Hann, D.W.; Larsen, D.R. 1989.ORGANON: southwest Oregon growth andyield model user manual, Version 2.0.Corvallis, OR: Oregon State University, ForestResearch Laboratory. 59 p.

McGaughey, R.J. 1997. Visualizing forest standdynamics using the Stand Vsualization Sys-tem. In: Proceedings of the 1997 ACSM/ASPRS annual convention and exposition.Bethesda, MD: American Society of Photo-grammetry and Remote Sensing: 248-257.Vol. 4.

McGaughey, R.J. 1998. Techniques for visualiz-ing the appearance of forestry operations.Journal of Forestry. 96(6): 9-14.

Teck, R.; Moeur, M.; Eav, B. 1996. Forecastingecosystems with the Forest Vegetation Simula-tor. Journal of Forestry. 94(12): 7-10.

Wilson, J.S.; McGaughey, R.J. 2000. Presentinglandscape-scale forest information: What issufficient and what is appropriate? Journal ofForestry. 98(12): 21-28.

Wykoff, W.R.; Crookston, N.L.; Stage, A.R.1982. User’s guide to the stand prognosismodel. Gen. Tech. Rep. INT-133. Ogden, UT:U.S. Department of Agriculture, Forest Service,Intermountain Forest and Range ExperimentStation. 112 p.

110

Introduction

Interest in the management of forests at the land-scape level has increased dramatically over thepast several decades. An area where this hasrecently received much attention is on lands man-aged by the USDA Forest Service in the PacificNorthwest. Prior approaches to forest manage-ment on these lands focused landscape-level

planning on development of a road network to aidin wildfire suppression and dispersion of harvestactivities. Stand treatments were primarily clear-cuts designed to meet several objectives includingrapid stand regeneration and the creation of edgeand early seral wildlife habitat. Fueled by changein public perception of clearcuts and a growingconcern over the effect of forest fragmentationon the loss of old-growth forest habitat on wild-life, especially the northern spotted owl (Strixoccidentalis caurina), and water quality during thelate 1980s, all Forest Service timber sales weretemporarily suspended. To end the suspension,the Northwest Forest Plan (USDA and USDI1994) was developed and is the guiding plan for9.7 million ha of federally managed forest land.

The Northwest Forest Plan (NWFP) is based on asystem of static reserves, riparian corridors, andgeneral forest land called matrix lands. A criticismof the NWFP is the use of static reserves on a

A Method to Simulate the Volume and Quality of WoodProduced Under an Ecologically Sustainable LandscapeManagement Plan

Glenn Christensen,1 R. James Barbour,2 and Stuart Johnston3

Abstract

Adopting the Northwest Forest Plan into federal land management brought with it many new challengesto forest planning. One challenge is to determine what effect using new and untested silvicultural treat-ments to meet ecologically based objectives will have on the production of wood or other products thatsociety values. Determining the quantity and quality of potential outputs as a consequence of ecologicalrestoration activities is important to helping understand the cost to society of implementing such a plan.The difficulty comes from the scale of the problem. No longer are short-term stand-level analyses ad-equate; evaluation of forest outputs has to be at meaningful spatial and temporal scales. To be success-ful, new methods will have to be developed.

The objective of this paper is to present a method used to evaluate wood removals from a landscapemanagement plan that was developed to meet ecological objectives by mimicking past disturbancecycles. This required extending stand-level techniques for simulating quality and quantity of wood to 980stands in the Blue River watershed of western Oregon. Silvicultural prescriptions include a range of thin-ning intensities and frequencies combined with extended rotation ages (100 to 260 years).

Study results include the range of products that might be manufactured under different silvicultural alter-natives and evaluate the volume as well as the quality of primary and secondary products. Primary prod-uct quality estimates include log attributes such as diameter, branch size, juvenile wood, and growth ringcount. Secondary product estimates include lumber or veneer volume recovered by grade and likely enduse.

1 Forester, U.S. Department of Agriculture, Forest Service,Pacific Northwest Research Station, Forestry SciencesLaboratory, P.O. Box 3890, Portland, OR 97208; Tel: 503-808-2064; e-mail: gchristensen@fs.fed.us

2 Research scientist, U.S. Department of Agriculture, ForestService, Pacific Northwest Research Station, ForestrySciences Laboratory, P.O. Box 3890, Portland, OR 97208;Tel: 503-808-2542; e-mail: jbarbour01@fs.fed.us

3 Forester, U.S. Department of Agriculture, Forest Service,Siuslaw National Forest, Mapleton Ranger District, 4480Highway 101, Building G, Florence, OR 97439; Tel: 541-902-6958; e-mail: sjohnston@fs.fed.us

111

dynamic landscape. Recently, however, consider-able interest has emerged as a different approachto landscape-level planning, and two studies havebeen initiated in the central Willamette NationalForest of Oregon. Both the Augusta Creek study(Cissel et al. 1998) and the Blue River landscapeplan (Cissel et al. 1999) have focused upon inte-grating historical disturbance regimes into land-scape- and watershed-level management plans.These approaches use information on historicaland current landscape conditions, disturbancehistory, and social goals to set objectives for fu-ture landscape structures that provide desiredhabitat, watershed, timber supply, and other func-tions (Cissel et al. 1999). The intent is not tomimic historical conditions but rather to use themas a reference in developing and evaluating man-agement alternatives to meet these goals.

The Blue River landscape plan is an integratedlandscape management strategy that was devel-oped to achieve both ecological and social objec-tives. As stated in the plan, the “primary goal is tosustain native habitats, species, and ecologicalprocesses while providing sustained flow of woodfiber for conversion to wood products” (Cissel etal. 1999). The key underlying concept is that bysimulating certain aspects of the historical fireregime through forest management activities, wecan sustain the historical range of variability nec-essary to preserve these native habitats, species,and ecological processes while still providing apredictable flow of timber.

Regardless of which approach is used, method-ologies and protocols are needed that describethe quantity, quality, and value of the wood pro-duced under landscape-level plans. Managersalso need to know when wood will be removedand from what locations on the landscape. In ad-dition to difficulties from planning at large spatialand temporal scales, most modern forest man-agement requires the use of complex silviculturalprescriptions in the form of thinnings, group selec-tions, and individual tree selections to meet spe-cific stand structure objectives. To address thesecomplexities, the Landscape Management Sys-tem (LMS) was developed at the University ofWashington (McCarter 1997).

Study Area

The 23 900-ha Blue River watershed is almostentirely managed by the USDA Forest Serviceand includes the H.J. Andrews Experimental For-est, an area with an extensive history of ecosys-tem research. Of the 298 million ha of forest landwithin the United States, 19 percent (56 millionha) is managed by the USDA Forest Service(Smith et al. 1994). The Blue River watershed islocated in the Willamette National Forest withinthe McKenzie River watershed, a tributary of theWillamette River in western Oregon (fig. 1).

The landscape is steep, highly dissected volcanicterrain of the Cascade Range. Annual precipita-tion exceeds 2500 mm falling mostly in Octoberthrough April as rain at lower elevations and snowin higher areas. The landscape ranges from 317to 1639 m in elevation and is covered largely byconiferous forests dominated by Douglas-fir(Pseudotsuga menziesii (Mirb.) Franco), westernhemlock (Tsuga heterophylla (Raf.) Sarg.), andPacific silver fir (Abies amabilis Dougl. ex Forbes)(Cissel et al. 1999).

Approach

The approach for this analysis is to use the BlueRiver landscape plan as the basis to determinescheduling of stand treatments and silviculturalprescriptions. The analysis focuses on wood pro-duction as trees are harvested to meet broaderecological objectives. The main tool used forsimulation is the LMS, a computerized systemthat integrates landscape-level spatial information,stand-level inventory data, and distance-indepen-dent individual tree growth models to projectchanges though time across forested landscapes(McCarter 1997). Once current conditions aredetermined for each stand as accurately as pos-sible, silvicultural treatments are scheduled andgrowth projections made. As stands are grownand treatments made, information is collected onindividual tree volume and quality characteristics.

112

This analysis uses methods similar to those de-veloped by Christensen (1997) and Barbour et al.(1996) to evaluate the volume and quality of woodfrom a range of silvicultural regimes. Tree volumeand quality information is analyzed with theTREEVAL (Briggs 1989, Sachet et al. 1989)model to estimate potential lumber volume andgrade recovery. The TREEVAL model was devel-oped by the USDA Forest Service, Pacific North-west Research Station to provide financial in-formation and analysis of product recovery tosupport silvicultural decisions in coast Douglas-fir(P. menziesii var. menziesii). Analyses conductedby Christensen (1997) and Barbour et al. (1996)are temporally explicit, evaluating wood producedfrom individual stands through time. They arespatially implicit in terms of where the wood camefrom. The current analysis adds the spatial dimen-sion to this process, providing not just wood pro-duction information as a stand develops overtime, but also providing information on the sourceof wood from across many stands within a water-shed. This gives planners the ability to understandhow stand treatments will interact to provide apredictable flow of wood in terms of volume andgrade recovery from a defined landscape.

Modeling Wood Production by Usingthe Landscape Management System

The LMS model was chosen for its landscape-level analysis capabilities, as well as its integra-tion of various component programs to providedetailed tree volume and wood quality information.Of particular interest is the ability to chooseamong several different growth and yield modelswithin the program to allow maximum flexibility inthe number of species, kinds of silvicultural treat-ments, and output produced. Included in LMS arestand- and landscape-level visualization programsthat permit, in addition to traditional quantitativeanalysis, generalized visual representations ofdifferent management options.

A detailed description of the methodology for eachpart of the project can be found in Christensen etal.4 For this paper, a brief overview is providedgiving the basic approach used during eachphase of the study.

Figure 1—Location of Blue River watershed.

4 Christensen, G.A.; Stuart, J.; Malinick, T.E. [In preparation].Simulating the volume and quality of wood produced underan ecologically sustainable landscape management plan inthe Oregon Cascade Range: results from the Blue RiverLandscape Plan. Res. Pap. Portland, OR: U.S. Department ofAgriculture, Forest Service, Pacific Northwest ResearchStation.

113

Current Stand Conditions

As part of the analysis for the Blue River Land-scape Plan, much basic information had alreadybeen assembled. The most important of this infor-mation was geographic information system (GIS)data files. From the GIS data, we knew the standtype, vegetation association, and how the stand isto be managed. Now we needed an inventory foreach stand with specific tree measurements. TheLMS software requires specific stand inventorydata including stand or polygon number, species,diameter at breast height, height, crown ratio, andexpansion factor, on a per-acre basis. The spatialdata requirements vary depending upon thegrowth-and-yield model selected but can include,by stand or polygon number, number of plots,regional location, site index, habitat type, age,slope, aspect, and elevation. Among the availabledata sets, it was determined that the inventoryplots maintained by the current vegetation survey(CVS) in the Willamette National Forest had themost applicable stand inventories in addition toproviding the most widely distributed sample plots.

By using CVS data for stand inventories we wereable to determine current vegetation condition formany stands. However, a significant number ofstands remained that had no available tree-leveldata. To overcome this, we used all available in-formation to assign inventory data to thesestands. By using a combination of vegetation as-sociation, stand type (tree species and structure),age, elevation, slope, aspect, and site index wewere able to select the closest match from standsthat had inventory data. Once this process wascomplete, we had an intact inventory for 980stands giving us the current vegetation conditionof the entire watershed.

Silvicultural Prescriptions, StandTreatment Scheduling, and GrowthProjections

As part of the Blue River Landscape Plan, thewatershed was divided into two basic manage-ment categories, reserves and landscape areas.Reserves were established to protect special in-terest areas such as late-successional spottedowl habitat and riparian areas along fish-bearingstreams. Landscape areas are the areas de-signed to meet a variety of ecological and socialobjectives where some level of timber harvest willoccur (fig. 2).

For this study, reserve areas were assumed toreceive no timber harvest and are not included inthe analysis. Landscape areas were divided intothree types based on interpretation of historicalfire-return intervals and intensity. The three land-scape areas are (Cissel et al. 1999):

• Landscape area 1–Frequent fire-return inter-val and moderate severity (40 to 60 percentmortality).

• Landscape area 2–Moderate fire-return inter-val and moderate-to-high severity (60 to 80percent mortality).

• Landscape area 3–Infrequent fire-return inter-val and high severity (>80 percent mortality).

Cissel et al. (1999) developed silvicultural pre-scriptions for each landscape area to closelymimic natural stand development following a fire(table 1). Treatments were designed to producestand structures and wildlife habitat closely re-sembling historical conditions within the water-shed. These prescriptions were used in the LMSmodel for this analysis. It is assumed that for eachprescription, reforestation (through replanting) willsucceed under some level of overstory. It is alsoassumed that, as after a natural fire, overstoryretention will be nonuniform and distributed aspatches and scattered individual trees throughoutthe stand. Patchy overstory retention also willstimulate understory growth in the gaps aidingreplanting success and development of verticalstructural diversity. Further structural diversity,both horizontal and vertical, will be developedthrough several thinning entries prescribed foreach landscape area.

Simulation of stand growth within LMS used theORGANON growth and yield model (Hann et al.1994). ORGANON has the advantage of allow-ing detailed silvicultural treatments and providesquantitative estimates of important stand attri-butes such as percentage of crown cover follow-ing a thinning. Another advantage of ORGANONis that it provides estimates of tree quality charac-teristics as an optional output. Tree characteristicsestimated by ORGANON include height to eachbranch, largest branch diameter, diameter of thejuvenile wood core at each branch whorl, anddiameter inside bark of the stem at each branchwhorl. A similar output from LMS has been devel-oped but has not been fully tested. A limitation of

114

Figure 2—Blue River Landscape Plantreatment areas (Cissel et al. 1999).

115

Table 1—Blue River silvicultural prescriptions

Prescription elements Landscape area 1 Landscape area 2 Landscape area 3

Rotation age (years)/%regeneration harvestedannually 100/1.0 180/0.56 260/0.38

Landscape block sizes<40 ha (% of area) 60 20 2040-80 ha (% of area) 20 40 4080-160 ha (% of area) 20 20 40

Retention level (% of existingoverstory crown closure) 50 30 15

Retention mixture Shade intolerant – Shade intolerant – Shade intolerant –(% of species dependent 65 80 95on plant association) Shade tolerant– Shade tolerant– Shade tolerant–

35 20 5

Reforestation density(trees per hectare) 500 750 1000

Reforestation mixture Shade intolerant– Shade intolerant– Shade intolerant–(% of species dependent 40 60 75on plant association) Shade tolerant– Shade tolerant– Shade intolerant–

60 40 25

First thinning (tph)a 500 at year 35 500 at year 20 to 25 500 at year 12 to 15

Second thinning (tph) 250 at year 35 275 at year 40 275 at year 40

Third thinning (tph) 150 at year 65 200 at year 70 200 at year 60

Fourth thinning (tph) Not planned 125 at year 100 125 at year 100

Low-severity fire (in Not planned Once between years 100 Twice between yearsaddition to fuel and 180 100 and 260treatments)

aTph = trees per hectare.

Source: Cissel et al. 1999.

growth models is the inability to simulate patchesand gaps within a stand. To accurately estimatethe effect of these treatments on tree growth andquality, models will need to have this failing cor-rected. Currently the best we can do is to usespatially uniform stand treatments and assumethe effect on overall stand growth and quality isnegligible.

A key part of the Blue River Landscape Pan iswhen and where stand treatments occur. To moveaway from further forest fragmentation, Cissel etal. (1999) applied treatments to areas designated

as landscape blocks. An individual block rangesin size from about 40 to 160 ha and may containseveral stands. Harvest entries were scheduledwith an area-control approach by using the land-scape block as the basic unit. Blocks were desig-nated for entry in 20-year periods over 200 years.Scheduling of harvest entries in LMS is accom-plished through a scenario file that contains theyear of treatment, stand number, and treatmentdescription. By using this file, silvicultural treat-ments can be programmed for multiple standsacross the entire landscape through time. De-

116

spite being able to simulate the growth of manystands for a large area, the current model cannotsimulate interactions among stands.

Analysis of Wood Volume and Quality

Analysis of wood production is completed afterstands have been grown and silvicultural treat-ments applied by using the LMS model. LMS canbe configured to simulate bucking of trees intouser-defined log lengths. Currently, logs can bedefined by only diameter and length, but work isbeing completed that incorporates additional qual-ity characteristics such as number of branches,branch diameter, growth ring count, and diameterof juvenile wood core. For this study we focusedon lumber production from each regime and dif-ferences in tree characteristics. Tree-quality char-acteristics are summarized for each regime frominformation in the ORGANON wood quality outputfile. To simulate lumber production, the TREEVALmodel is used. TREEVAL is designed to acceptORGANON wood-quality output files as input.TREEVAL calculates financial information allow-ing comparisons of different silvicultural regimes.For this study we were interested primarily in thevolume and quality of lumber recovered for eachregime. Lumber recovery equations from Fahey etal. (1991) have been programmed into the modelestimating yields of veneer and lumber. Lumberrecovery can be combined for visually graded andmachine-stress rated (MSR) lumber. TREEVALsimulates the bucking of trees to a nominal loglength of 4.9 m and a minimum small-end diam-eter of 13 cm. Volumetric lumber grade recoverieswere estimated for each log by using theTREEVAL model.

Conclusion

For this paper, we wanted to demonstrate that bylinking existing simulation models, a detailedanalysis can be made from a landscape-scaleperspective of wood volume and quality frommany individual stands. This information can beused to evaluate wood removals based on differ-ent approaches to planning on a large scale. Tim-

ber production comes from tree removals as partof stand treatments used to develop specificstructural attributes through time. All approacheshave an eventual rotation age that is longer thanthat used in traditional management for timberproduction. The combination of using uniquestand treatments with long rotations will have aneffect on the volume and quality of wood availablefor commodity production. It is unknown what themagnitude of this effect will be, and simulationmodels give us an opportunity to look at potentialtrends.

English Equivalents

1 hectare (ha) = 2.47 acres

1 millimeter (mm) = 0.039 inches

1 centimeter (cm) = 0.39 inches

1 meter (m) = 3.28 feet

Literature Cited

Barbour, R.J.; Johnston, S.; Hayes, J.P.;Tucker, G.F. 1996. Simulated stand character-istics and wood product yields from Douglas-firplantations managed for ecosystem objectives.Forest Ecology and Management. 91(1997):205-219.

Briggs, D.G. 1989. Tree value system: descrip-tion and assumptions. Gen. Tech. Rep. PNW-GTR-239. Portland, OR: U.S. Department ofAgriculture, Forest Service, Pacific NorthwestResearch Station. 24 p.

Christensen, G.A. 1997. Development of Doug-las-fir log quality, product yield, and standvalue after repeated thinnings in westernOregon. Corvallis, OR: Oregon StateUniversity. 176 p. M.S. thesis.

Cissel, J.H.; Swanson, F.J.; Grant, G.E. [et al.].1998. A landscape plan based on historical fireregimes for a managed forest ecosystem: theAugusta Creek study. Gen. Tech. Rep. PNW-GTR-422. Portland, OR: U.S. Department ofAgriculture, Forest Service, Pacific NorthwestResearch Station. 82 p.

117

Cissel, J.H.; Swanson, F.J.; Weisberg, P.J.1999. Landscape management using historicalfire regimes: Blue River, Oregon. EcologicalApplications. 9(4): 1217-1231.

Fahey, T.D.; Cahill, J.M.; Snellgrove, T.A.;Heath, L.S. 1991. Lumber and veneer recov-ery from intensively managed young-growthDouglas-fir. Res. Pap. PNW-RP-437. Portland,OR: U.S. Department of Agriculture, ForestService, Pacific Northwest Research Station.25 p.

Hann, D.W.; Olsen, C.L.; Hester, A.S. 1994.ORGANON users manual. Corvallis, OR: De-partment of Forest Resources, Oregon StateUniversity. 113 p.

McCarter, J.B. 1997. Integrating forest inventory,growth and yield, and computer visualizationinto a landscape management system. In:Teck, R.M.; Moeur, M.; Adams, J., eds. Pro-ceedings of the Forest Vegetation Simulatorconference. Gen. Tech. Rep. INT-GTR-373.Ogden, UT: U.S. Department of Agriculture,Forest Service, Intermountain ResearchStation: 159-167.

Sachet, J.K.; Briggs, D.G.; Fight, R.D. 1989.Tree value system: users guide. Gen. Tech.Rep. PNW-GTR-234. Portland, OR: U.S. De-partment of Agriculture, Forest Service, PacificNorthwest Research Station. 45 p.

Smith, W.B.; Faulkner, J.L.; Powell, D.S. 1994.Forest statistics of the United States, 1992,metric units. Gen. Tech. Rep. NC-GTR-168.St. Paul, MN: U.S. Department of Agriculture,Forest Service, North Central Forest Experi-ment Station. 147 p.

U.S. Department of Agriculture, Forest Serv-ice; U.S. Department of the Interior, Bureauof Land Management. 1994. Recordof decision for amendments to Forest Serviceand Bureau of Land Management planningdocuments within the range of the northernspotted owl. [Place of publication unknown].74 p. [Plus attachment A: standards andguidelines].

118

Transforming Information From Many Perspectives toAction on the Ground

119

Scale and the Human Frame ofReference

As Allen and Hoekstra (1992), among others,have noted, scale is a social construction. Thismeans that scale is not something that is outthere on the ground that exists apart from humansbut rather it is something we create by assigningdefining attributes to it. A particular scale does notexist in reality and has no meaning until we give itmeaning (Tuan 1974). Thus, those who select aparticular scale have a responsibility to clarify itsmeaning to others.

In addition, scale only has meaning relative to thehuman frame of reference. Recall the relationGulliver had to the lands and people he visited.The differences in scale that he experienced wereall relative to his own size and personal past ex-perience. Because each of us brings a differentpast experience with us, we will see the samethings differently. This means that we need to lookat places and projects together so we can sharewith each other what each of us “sees” ratherthan assuming we are all seeing and experiencinga place in the same way.

Most of us (as humans) experience places at thescale of a site–a place we go fishing, camping, orhiking, for example. Some of us may experiencea sense of place or special attachment to certain

places, usually at small scales. However, whenasked to comment on a forest plan at the scale ofa national forest, many of us (as citizens) may beat a loss as to how to express how we feel aboutthose special places. Thus, both agency planningprocesses and research questions need to recog-nize the interconnectedness across multiplescales and scale down to smaller scales so thatcitizens can better understand the implications ofa decision or finding on “their place.” Often theboundaries identified for purposes such as plan-ning and research are not socially significant ormeaningful to citizens.

Nassauer (1997) suggests that landscape scaledoes not have a universal definition. She identifiestwo common uses of the concept:

(1) a heterogeneous combination of eco-systems, which affect each other acrossspace and time, and (2) a “middle” scale,within a hierarchy, of ecological processesthat affects smaller-scale processes andis affected by larger-scale processes. Boththe concepts of heterogeneity and hierar-chy emphasize the connectedness of thelandscape across space and time. Con-nectedness, then, is an essential propertyof the landscape scale. (Nassauer1997: 73)

Nassauer (1997: 73) goes on to suggest that “atthe scale of human beings, connectedness refersto landscape structure that allows flows of water,nutrients, energy or species that people havenoticed and believe to have ecological value.”

A Social Science Perspective on the Importance of Scale

Linda E. Kruger1

All ecological processes and types of ecological structure are multiscaled. Each particular struc-ture relates to a particular scale used to observe it such that, at the scale of perception, the entityappears most cohesive, explicable, and predictable. The scale of a process becomes fixed onlyonce the associated scaled structures are prescribed and set in their scaled context. Scaling isdone by the observer; it is not a matter of nature independent of observation. (Allen andHoekstra 1992: 11, emphasis added.)

This paper offers a very brief introduction to some scale-related considerations from the perspective of asocial scientist. The paper does not provide an indepth discussion, but my hope is that enough informa-tion is provided to provoke thought and additional exploration in these areas.

1 Research social scientist, U.S. Department of Agriculture,Forest Service, Pacific Northwest Research Station, ForestrySciences Laboratory, 400 N 34th Street, Suite 201, Seattle,WA 98103; Tel: 206-732-7800; e-mail: lkruger@fs.fed.us

120

From this perspective, landscape scale is a cul-tural concept based on what we determine tohave value. Because we as humans managelandscapes, we often arrive at a definition of land-scape scale by identifying what we perceive asmanagement units (Nassauer 1997). Based onthis perspective, landscape scale could meansomeone’s backyard, a watershed, ranger district,or national forest. “In each case, the ecologicalfunctions of individual patches of lakes, streams,turf, fields, forests, or even pavement are con-nected to one another” (Nassauer 1997: 73).Thus, landscape as a scale of analysis is similarto the social concept of community in that land-scapes, like communities, are nested and embed-ded in each other at various scales, rather thanrepresenting one specific scale.

It is this connectedness of ecological functionsacross space that makes working across land-scapes essential. However, working across land-scapes is often easier said than done. Patchworkownership patterns put landscape-level decisionsin conflict with our culture’s belief in private prop-erty rights. More work is needed to develop toolsand incentives for working across boundaries inways that minimize the threat to people’s sense ofprivate property rights. (See Kalinowski’s [1996]discussion of Leopold’s land ethic, specifically hisdifferentiation between land as property and landas territory.)

The ability to take a cross-boundary perspective—looking across management jurisdictions, owner-ships, and administrative and legal boundaries—expands the possibility of considering questions atthe appropriate scale (Knight and Landres 1998).Like scale, boundaries are created by society fora variety of social purposes. Boundaries organizespace. They often cut across other social relationsthat define space. For example, the city I live inlies within a school district that includes two othercities. It is bisected by a county line. Boundariesdefine to whom we pay taxes, for whom we canvote, and who will represent us in government,where our kids go to school, where our watercomes from, and in some parts of the world evenwhere people can and cannot travel. Creatingboundaries is a process of place-making or con-structing an identity that differentiates what andwho are inside the boundary from what and whoare outside the boundary.

Determining appropriate planning or study bound-aries is important. Just as important is keeping inmind that boundaries are always porous and openacross time and space (Massey 1995). The identi-ties of places are products of links to other timesand places. Although each place is unique, itsinterconnections to other times and places areimportant in defining its uniqueness.

Appropriate Scale

Scale is an important consideration for both man-agers and scientists. Every resource issue andquestion has a spatial and temporal componentor context. In choosing the appropriate scale, wemust consider at what scale our knowledge isreliable and at what scale we are most comfort-able making predictions (Haskell et al. 1992).What the appropriate scale is depends on thequestion(s) being asked and can range from asmall personal space (How will this plant look onmy desk?) to the entire earth (How might thisactivity influence global change?). For social in-quiry the most common scales are community,county, state, and region. Counties are frequentlyused for analysis because they are a major unitused by the USDC Bureau of the Census. Machliset al. (1995) participated in the social assessmentprocess portion of the Interior Columbia BasinEcosystem Management Project (ICBEMP). Theysuggest the county as the best scale for socialanalysis, noting that data are readily available andthat counties are increasingly involved in resourcepolicy through planning and zoning activities. Theyalso suggest that the county scale fits best withlandscape-level analysis (Machlis et al. 1995:13-16).

Although there are advantages to the county scaleand major studies such as ICBEMP and ForestEcosystem Management Assessment Team(FEMAT) do not support higher resolution com-munity-level studies, there also are limitations toworking at the county scale. The effects of re-source decisions are often felt at the local level,and communities within a county can differ dra-matically, particularly in a county that has a mix ofurban and rural areas. Working at the larger,county-level scale also tends to disenfranchisepeople whose interests are more local. Theseare often the people who will feel a study “hasbeen done to them.” It also can be hard to makecounty-level data meaningful to a community. As

121

a society, we need to explore methodologies thatenable people to learn about themselves and theirplaces at a variety of scales that are meaningful tothem.

The ability to consider a question at the appropri-ate scale may be constrained by a variety of fac-tors including the scale at which data are avail-able, time or funding, political considerations,logistical problems, and project location (Commit-tee on the Applications of Ecological Theory toEnvironmental Problems 1986). In many cases,ownership, management jurisdiction, or anotheradministrative or legal constraint instead of thequestion or issue itself determines the spatial areathat is considered.

Working across disciplines or multiple types ofinformation may require synthesizing finer scaleinformation to a coarser scale common with oth-ers. For example, socioeconomic data may beavailable only at the county scale, whereas bio-physical data may be available at a combinationof scales including the watershed, stream reach,stand, or province. The identification of a cross-cutting question may be a helpful first step in iden-tifying an appropriate scale at which to integrateacross disciplines.

The Committee on the Applications of EcologicalTheory to Environmental Problems (1986: 97)determined that a “fundamental problem in pre-dicting and controlling cumulative effects is thefrequently large mismatch between the scales orjurisdictional boundaries of management authorityand the scales of the ecological phenomena in-volved or their effects.” The committee notes thatan affected environment often crosses severaljurisdictions. In addition the impacts of an actionmay be felt in a different jurisdiction far fromwhere the offending action took place.

Thus, the scale that is chosen must be largeenough to encompass interactions, larger scaleprocesses, and cumulative effects (over time andspace) without losing sight of important details,smaller scale processes, and interactions. Atbroader scales, important variability can bemasked; at narrower scales important effects,levels, and nature of impacts can be missed; andat scales that are mismatched, it can be impos-sible to develop linkages between human commu-nities and forests.

Considerations in Choosing aParticular Scale

In choosing a particular scale to use when design-ing a study or planning area, asking the followingquestions may be helpful.

• What assumptions underlie the choice of thisscale? Are these reasonable assumptions?

• What is the justification for choosing this scaleover others?

• What is being overlooking or ruled out bychoosing this scale over others? Are we miss-ing any critical questions?

• Are we choosing a particular scale becausewe have data available or because measure-ment is easier than at other scales?

• What are the implications of working at thisscale?

• Does our choice of scale match ourobjectives?

How does the choice of a particular scale affectecosystem management efforts? In an attempt tobegin to address this issue, Norton (1992) identi-fies four ways in which scale affects ecosystemmanagement.

1. Spatial boundary issues. Jurisdictional bound-aries usually do not correspond to naturalfeatures or biogeography.

2. Temporal perspective. Biological (and social)systems are dynamic. Thus managers mayhave to decide whether to manage for currentsocial values, production of “essential ser-vices,” or features that are critical to maintain-ing natural processes.

3. Organizational structure and scale. This as-pect of scale has to do with substructure.Once a whole system is defined, subunits,each with their own boundaries, must be iden-tified. Taking an example from social science,a nation may be made of states, which mightbe subdivided into counties, which have withinthem cities and towns and neighborhoodswhere families and individuals live.

122

4. Scale and degree of impact. The degree ofimpact an action may have on a system de-pends both on the scale of the activity and thescale of the system. Activities that might de-grade a small site might have negligible effectat a larger scale. As the scale of the activityincreases, it might have higher degrees ofimpact at ever increasing system scales. Of-ten the effects of our actions have impacts ata variety of scales.

So what happens if our choice of scale is lessthan optimal? There are several possible out-comes of analysis or action at the wrong scale.The following are drawn from the Committee onthe Applications of Ecological Theory to Environ-mental Problems report Ecological Knowledgeand Environmental Problem-Solving (1986).

1. Well-intended efforts can have minimal oreven adverse effects if planned for too shorta timeframe or too small an area.

2. Key processes might be overlooked at largertemporal or spatial scales.

3. Averaging over larger areas can mask theimportance of processes at smaller scalesbecause it is more difficult to detect subtledifferences.

In addition to these concerns, a focus on singleactions at whatever scale can result in obscuringcumulative effects over time and space. Cumula-tive effects can take several forms ranging fromactivities that add materials to the environment,such as discharging effluents, to those that re-move materials from the environment, such asharvesting timber. Cumulative effects can take theform of changes over large areas for long periodsof time that result from management actions,such as when a forest is managed based onsingle stands. Cumulative effects also occur whenseveral actions compound each other; for ex-ample, when logging roads are built, logging, rec-reation, hunting, poaching, and other activitiesincrease and affect a variety of forest species.The importance of understanding and consideringthe implications of cumulative effects makes astrong case for the consideration of multiple spa-tial and temporal scales.

The Inclusion of Citizens in Planningand Decisionmaking

Transforming information into implementation onthe ground requires both recognition that humansare part of ecosystems (rather than an outsideforce) and the inclusion of citizens in planningand implementation. This requires us to shift ourthinking from “unfortunately we have to work withpeople” to “citizens are critical to an informed,effective process.” Collaborative planning, whenused in Forest Service planning processes, re-quires participation of researchers, managers,and citizens. It recognizes that citizens haveknowledge and real world experiences to contrib-ute to the planning process. Frequently citizensalso bring a passion for action, and citizen partici-pation can result in better decisions and an im-proved process. Land managers can share re-sponsibility for environmental quality and healthwith those outside the agency only if they provideopportunities that include citizens as meaningfulparticipants in planning and decisionmaking pro-cesses.

It also requires an understanding of the variousscales at which people identify a landscape as aplace that they relate to. Geographer Gillian Rose(1995) suggests that people develop a sense ofbelonging or connection to places at multiplescales including a local scale, a regional scale, anational scale, a supranational scale, like Europe,and even at a global scale, as our social and eco-nomic systems become more globalized. Thismeans that as managers consider who should beinvolved in local planning processes, they mustthink beyond the local geographic community.

“What kinds of evidence, scientific or otherwise,are necessary to justify a given choice regardingthe scale on which environmental problems areaddressed?” Norton (1992: 35) answers his ownquestion by suggesting an interdisciplinary ap-proach that acknowledges that questions of scaleinvolve value judgments. “The correct scale onwhich to address a management problem is de-termined by what society wants to accomplishwith that system” (Norton 1992: 36). Norton joinsothers in suggesting that citizen involvement indialogue and debate is necessary in order to fa-cilitate consideration of alternative actions anddesired conditions.

123

Scale may influence citizen involvement in man-agement and planning. It can be much more diffi-cult for citizens to comment on large-scaleproposals that are not linked to specific small-scale places they can personally identify with.Forest plans are at a scale many citizens finddifficult to respond to. We might speculate thatwatersheds, especially at the local communitylevel, are a more meaningful scale as evidencedby the number of grassroots watershed groups.

Frequently, federal land management agenciesconduct planning processes at the scale of a na-tional forest. Ecosystem management is based onplanning at a large scale, often including areaslarger than a national forest. The FEMAT and theICBEMP each covered several states. At thesebroad scales it is easy for people and their inter-ests and concerns to get lost.

How meaningful public participation can be isoften a function of the scale at which a study isinitiated. The scale or scope of a place beingstudied is important in the sense that it has to bea scale that people can relate to. What is impor-tant is that the scale is appropriate to the questionor issue being addressed and that it does notobscure what is meaningful and what people canrelate to.

We must also take care that the tools and tech-niques that we use do not obscure what it is thatis meaningful and important to people. Technol-ogy has provided us with tools that enable us tocreate pictures of what a stand or landscape willlook like immediately after harvest and at intervalsthereafter. In addition to its use in presenting infor-mation, this technology is often used to solicitpublic response to harvest practices (Shindler etal. 1995). Although the technology can serve auseful purpose in helping citizens visualize whatthe land will look like over time as the forestgreens up, some researchers have concerns withits use in measuring public judgments of accept-ability.

Studies have found that judgments of acceptabilityof management practices are based on more thanjust looking at a scene and responding to what isthere (Brunson 1993, Shindler and Cramer 1999).Research done by some of our collaborators hasidentified a number of factors that play a role inthe acceptability of resource management deci-sions and actions. For example, the respondents’

personal experiences both with the specific placeand the agency proposing the activity and theirlevel of trust in the agency often play a role. Asense that “too much has already been cut” orgeneral concerns over wildlife, recreation, or sce-nic values also may play important roles. Finally,how the decisionmaking process is carried out isimportant to some people. If people feel theydidn’t have opportunities for meaningful participa-tion or if they didn’t feel heard, they may object toan activity regardless of how the project looks onthe ground. I think a quote from the child’s story, ALittle Prince (Saint-Exupery 1943: 87) aptly de-scribes this situation: “What is essential is invis-ible to the eye.” What is essential to understand-ing how acceptable a forest practice is may haveless to do with how the forest appears and moreto do with aspects of how and why it looks theway it does and where it is located relative toplaces people care about.

Literature Cited

Allen, T.F.H.; Hoekstra, T.W. 1992. Toward aunified ecology. New York: Columbia UniversityPress. 384 p.

Brunson, M. 1993. “Socially acceptable” forestry:What does it imply for ecosystem manage-ment? Western Journal of Applied Forestry.8(4): 1-4.

Committee on the Applications of EcologicalTheory to Environmental Problems. 1986.Ecological knowledge and environmental prob-lem-solving. Washington, DC: National Acad-emy Press. 388 p.

Haskell, B.D.; Norton, B.G.; Costanza, R. 1992.What is ecosystem health and why should weworry about it? In: Costanza, R.; Norton, B.G.;Haskell, B.D., eds. Ecosystem health: newgoals for environmental management. Wash-ington, DC: Island Press: 3-20.

Kalinowski, F.A. 1996. Aldo Leopold as hunterand communitarian. In: Vitek, W.; Jackson, J.,eds. Rooted in the land: essays on communityand place. New Haven, CT: Yale UniversityPress: 140-149.

Knight, R.L.; Landres, P.B., eds. 1998. Steward-ship across boundaries. Washington, DC: Is-land Press. 371 p.

124

Machlis, G.E.; Force, J.E.; Dalton, S.E. 1995.Monitoring social indicators for ecosystemmanagement. Walla Walla, WA: Interior Co-lumbia Basin Ecosystem Management Project.Unpublished report. On file with: Pacific North-west Research Station, Human and NaturalResources Interactions Program, ForestrySciences Laboratory, 400 N 34th Street, Suite201, Seattle, WA 98103.

Massey, D. 1995. The conceptualization of place.In: Massey, D.; Jess, P., eds. A place in theworld? Places, cultures and globalization. Ox-ford, United Kingdom: Oxford University Press:45-85.

Nassauer, J.I. 1997. Cultural sustainability: align-ing aesthetics and ecology. In: Nassauer, J.I.,ed. Placing nature: culture and landscape ecol-ogy. Washington, DC: Island Press: 67-83.

Norton, B.G. 1992. A new paradigm for environ-mental management. In: Costanza, R.; Norton,B.G.; Haskell, B.D., eds. Ecosystem health:new goals for environmental management.Washington, DC: Island Press: 23-41.

Rose, G. 1995. Place and identity: a sense ofplace. In: Massey, D.; Jess, P., eds. A place inthe world? Places, cultures and globalization.Oxford, United Kingdom: Oxford UniversityPress: 88-132.

Saint-Exupery, A. 1943. The little prince. NewYork: Harcourt, Brace and World, Inc. 113 p.

Shindler, B.; Cramer, L.A. 1999. Shifting publicvalues for forest management: making senseof wicked problems. Western Journal of Ap-plied Forestry. 14(1): 1-7.

Shindler, B.; Peters, J.; Kruger, L. 1995. Socialvalues and acceptability of alternative harvestpractices on the Tongass National Forest.101 p. Unpublished report. On file with: PacificNorthwest Research Station, Human and Natu-ral Resources Interactions Program, ForestrySciences Laboratory, 400 N 34th Street, Suite201, Seattle, WA 98103.

Tuan, Yi-Fu. 1974. Space and place: humanisticperspective. Progress in Human Geography.6: 213-252.

125

126

Conclusion

127

I was asked not to prepare remarks ahead of theworkshop, but rather to respond to what I sawand heard. Let me say at the outset, I have beenmuch impressed with the general thrust of thisworkshop. Many of the papers were technical andrepresented sound research, but the real strengthof the presentations as a whole was in the genu-ine eagerness in tackling big problems. The bigproblem, as I see it, is that we have many large-scale problems with regard to renewable re-sources, and that there is a huge divide betweenhow science prefers to work and what needs tobe done. We should be impressed with the highlevel of appreciation for the difficulties, and thevigor of the response in so many of the presenta-tions. I saw courage in the presenters I heard.

As an example, we are really quite good at quanti-fying arcane models about cutting regimes. Butwhat is the use of this to the end consumer ofthese results? The models mean nothing until wevisualize the landscape in images that anyonecan see and understand. I saw exactly that at thisworkshop. The new gigabytes and gigahertz ofcomputer capability are powerful, because theylet us translate abstractions, such as a 10-percentcut, into an image that the public can see andevaluate. New technology really does change thegame, as we fly down streams and summarizetheir condition with thermal imagery. The prob-lems are large, but the new technology moveshuman achievements upscale to meet many ofthem. Even the papers that used no technological“fancyware,” were generally very cognizant of theneed to serve the interface between humans onthe landscape and the biogeophysical landscapeitself. When we manage an ecosystem, we do notmanage the material ecological system itself;rather we manage the presence of the peoplewho impact the system. And that does not meanwe manipulate the public covertly or dishonestly.

No. Ecosystem management will not work unlessit involves explicitly giving people options that theyare prepared to use, and a sound basis to selectamong them.

What I have described above is what SilvioFuntowicz and Jerry Ravetz call postnormal sci-ence (Ravetz 1999). Although normal, merelymodern science is meticulous as it tries to ap-proach reality with its models; the postnormalscientist cannot afford to be so idealistic. Post-normal situations are characterized by (1) highstakes, (2) insufficient data, and (3) a short timefuse. The data from normal science will not becoming in time, for the issue will have passed, winor lose, before meticulous experiments and confi-dence limits can be achieved. Even so, scientistsmust offer their best advice anyway, and the man-ager must act on that best advice. Clearly findingout what is the reality of the situation is irrelevant,but, if it is any comfort, reality was never achiev-able anyway.

Beyond postnormal science is postmodern sci-ence. In the classical world and modern world thenovice became dexterous, and the apprenticebecame a craftsman, while the master went be-yond all to create something in a new framework.The novice and apprentice achieve quality bybeing meticulous, delivering exactly what isneeded every time. That is called structural qual-ity. The master introduces a different quality, adynamical quality of good change, which itselfundermines the premises that underlie structuralquality. Quality classical work was approved bycommon standards, where the external consen-sus was that everyone liked the picture of thePatron. The difference between classical andmodern is that in the modern world, external real-ity becomes the external reference. Elites, be theycubist painters or 20th-century scientists, assertthat by looking at the world their special way, ulti-mate reality may be approached. But in apostmodern world there is no external reference,be it consensus or reality. In postmodernity, it isnot that it is all in our heads and anything goes, itis rather that we only deal with data, not with ob-

Summary Remarks on Views From the Ridge

T.F.H. Allen1

1 Professor, Botany Department, University of Wisconsin,Madison, WI 53706-1381; Tel: 608-262-2692; e-mail:tfallen@facstaff.wisc.edu

128

jective reality. Without an external reference, thepostmodern scientist must make the process ofscience justify itself, but how? The answer is thatwe need to keep science a high-quality activityin both structural and dynamic features. Eventhough we cannot now pretend that science isobjective, it is quality that is endogenous to doingscience that continues to give science its de-served privileged position (Funtowicz and Ravetz1992). Science is meticulous so it can repeatanything it does, and that amounts to structuralquality. Science also challenges its products atevery level from the null hypothesis all the way tofighting over new paradigms. That is dynamicalquality.

Structural and dynamical quality, that is, honestcarefulness and creativity, make science the bestgame in town. And we saw lots of both sorts ofquality at this workshop. Some papers were tech-nical and were splendid displays of structuralquality. Other papers were full of dynamical qual-ity, as managers, politicians, scientists, and thelay public made visionary statements of what todo next, as we reach out to a world full of peoplewith values living in systems that are the purviewof the Forest Service.

It is good to see such courage and commitmentto dealing with the values of people in the bio-sphere, because the problems are huge. JosephA. Tainter (1988), of the Rocky Mountain Re-search Station, points out that societies are prob-lem-solving units that elaborate processes orapproaches to deal with resource issues. It ap-pears inevitable that problem solving proceeds togreater cost and less benefit with vicious dimin-ishing returns. Eventually societies lose either thepolitical will or the ability to deal with the last prob-lem, whatever form it may take, and the peoplejust walk away from civilization.

Some societies on the edge of collapse find a newlease on life by switching to some fundamentallynew resource, such as coal when Britain ran outof wood. Looking at attoid ants, we see that theprimitive species farms fungi on grasshopperdroppings, which amounts to jet fuel for growingfungi. The problem is that insect droppings are ahighly processed, scarce material, albeit of high

quality. Leaf-cutting ants have evolved further touse leaves instead of guano, and so overcomethe scarcity issue. But leaves are a low-qualityresource, and so demand very organized pro-cessing. Diminishing returns on a high-qualityresource, such as guano, end in collapse whenthe supply runs out. Diminishing returns on a low-quality resource, such as leaves, end in collapsewhen the huge system grows into excessive de-mand. Counterintuitively, more capital is built ondiffuse, low-quality resources than on high-qualityresources.

It appears that human society often switches fromhigh-quality resources (HQR) to low-quality re-sources (LQR) in cycles of elaboration. The ener-getically easy entry to a general type of resourceis on HQR, and the elaboration upscale is onLQR. Thus (1) hunting (HQR) leads to (2) agricul-ture (LQR). In an agricultural setting, (3) lootingneighbors (HQR) gives way to (4) imperial taxa-tion of peasants (LQR). In the First World we liveon (5) industrialization (HQR). Without meaning todescribe any sort of grand historical narrative, wenote that some political arrangements collapsedat various stages, e.g., Mongols in the West atstep 3, or the Romans at step 4.

We look as if we are about to come off the backside of step 5, fossil carbon, and need a new LQRphase. Because renewable energy sources are alllow quality, they will have to be extensively cap-tured, rather like sun on crops. Much more eco-logical damage has been done by LQR agriculturecompared to HQR fossil fuels. The change to theLQR hydrogen economy using wave, wind, bio-mass, and solar energy may be wrenching, ashuman settlement decentralizes. The diffuse na-ture of high-technology information systems al-lows for energy self-sufficient production systemsscattered across the landscape. There are alter-native scenarios of the move to renewable energywhere decentralization is delayed or circum-vented, but these have their downside. They in-volve consumption of the mountains of dirty coal,as a society builds massive LQR infrastructurewhere forests of windmills sit on sculptured coast-lines (Allen et al. 2001). Avoiding renewable en-ergy appears to invite the four Horsemen of theApocalypse, so we cannot afford to fail. And fail

129

we will unless we can encourage the populace atlarge that the transition is worth it. The humanfocus and the emphasis on advice from lay opin-ion at this workshop are most encouraging at atime when deep concern for public opinion is notmisplaced. The coming LQR society is set to dev-astate coastlines, deserts, forests, and wildlife. Itis for this reason that the visionary tone of thisworkshop is so important. Even under the happi-est outcome, ecological stress will be great. Pre-cisely because of impending ecological disaster inthe move to extensive use of renewable energy,only the best stewardship will do. The problem islarge, so the view of the stewards of natural re-sources such as forests, wildlife, and fisheriesmust be at least as visionary as the most expan-sive view from the ridge.

Literature Cited

Allen, T.F.H.; Tainter, J.A.; Pires, J.C.;Hoekstra, T.W. 2001. Dragnet ecology, “Justthe facts ma’am”: the privilege of science in apost-modern world. BioScience. 51: 475-485.

Funtowicz, S.O.; Ravetz, J.R. 1992. The good,the true and the post-modern. Futures. 24(10):963-974.

Ravetz, J.R. 1999. What is post-normal science?Futures. 31: 647-653.

Tainter, J.A. 1988. The collapse of complex soci-eties. Cambridge, MA: Cambridge UniversityPress.

130

Key Drivers of Landscape-LevelResearch

Station scientists at the Pacific Northwest Re-search Station (PNW) work to advance the basicunderstanding of how systems function, especiallyterrestrial, aquatic, and socioeconomic systems.This provides the foundation of scientific under-standing to better understand the consequencesand risks of choices made in managing the land.It also provides the foundation on which to helpdesign new land management options.

In the 1990s, in response to key issues, Stationscientists increasingly focused their research onthe landscape scale. This workshop reflects abody of work designed by PNW to consciouslyexamine this scale. In many ways, we have onlystarted to understand systems at the landscapescale in the Pacific Northwest, but it is an impor-tant part of our research program that is alreadydemonstrating that it holds considerable promise.We appreciate being able to share informationwith those of you who make land managementdecisions or affect those decisions and the scien-tists here who have validated and built on ourresearch results.

Much of our landscape-scale research has fo-cused on national forest land issues because theyoffer landscape-scale laboratories. Our research-ers often work side by side with land managers aswe conduct studies and assessments on the verylarge scales of the national forests. We also con-duct landscape-scale research on other landssuch as the state lands of Washington. Much ofwhat we are learning is applicable to all owner-ships. As we study forests at the landscape scale,I believe we can find some additional options thatmay create compatibility among interests andacross ownerships.

What put us on the landscape trek? In part, it wasthe result of our fine-scale and single-disciplineresearch. Although often initiated from a singledisciplinary perspective, this research has pro-duced important scientific findings. At the sametime, it has shown that some of those fine-scalephenomena also were components of large sys-tems. We have found that some relations existand manifest themselves only at a large scale.

As we examined outbreaks of insects and diseaseduring the past several decades, for example, webegan to understand the need to look beyond thestand. Also our research on fire disturbance pro-cesses and smoke management led to a broaderperspective. The ecological effects of forest frag-mentation and the interactions between fragmen-tation and habitat quality for some wildlife speciesbegged a landscape-scale view. Increasingly weunderstand that pattern and patches across thelandscape affect relations such as those amongthe nesting, feeding, and migrating of some wild-life species.

In addition to our fine-scale research, some defin-ing land management issues helped us to look atbroader scaled relations. Issues such as manage-ment of old-growth forests, recovery of threatenedand endangered species, and forest health areexamples. These types of issues have been driv-ers of our research program, just as our scientificunderstanding has helped shape the character ofthese issues.

Land management issues such as these, rein-forced by our past research, have led to the fol-lowing overarching land management questionsthat demand broader scale, functionally integratedresearch: How do the effects of land managementactions accumulate over space and time? Whatare the effects of interactions among differentland units, and how do those interactions affectoverall outcomes? Another related question, giventhe inability to treat all parcels, is, How can weprioritize among the numerous parcels to maxi-mize our objectives?

The Future of Landscape-Level Research at the PacificNorthwest Research Station

Thomas J. Mills1

1 Deputy Chief, Business Operations, U.S. Department ofAgriculture, Forest Service, 201 14th Street SW, Washington,DC 20250; Tel: 202-205-1707; e-mail: tjmills@fs.fed.us

131

These questions require a landscape-scale per-spective and integration among different scientificdisciplines. Moving to broader scale is usuallycoupled with the need to study the interactionsamong different components of the land, as wellas the interaction among the different land com-ponents within the landscape. Integration amongphysical and biological disciplines is needed, butintegration between the biophysical and socialsciences also is essential if we are to understandthe full array of consequences of land manage-ment actions and to create new land managementoptions.

The Station has a deep history of long-term re-search studies and an emerging body of land-scape-scale research, but research integratedacross disciplines at the landscape scale is onlynow emerging. Although working across disci-plines can be complicated and time consuming, itis in the resulting comprehensive picture thatpolicymakers likely will find solutions to complexproblems.

Land Management OpportunitiesAmong the Challenges

Not only does the landscape-scale perspectiveadd significant complexity beyond what is appar-ent through a finer scale look at the land, it alsoholds the promise of land management opportuni-ties not apparent at the finer scale. For example,if the focus is solely on the fine scale, a manageror the public might conclude that only the areabeing studied can provide some desired value.They might also conclude that all such sites mustprovide the same set of values. Our research hasshown, however, considerable variability acrossthe landscape. The land is not now uniform norwas it ever, even when nature was managing it.For example, not all sites on the west side of Or-egon and Washington were prime salmonid habi-tat at any one time even in the past. And some ofthe most constraining laws, such as the Endan-gered Species Act, require only that values beprovided on enough sites to meet an overall goal,not that those values need to be provided fromevery site. A landscape view helps us understandhow a mosaic of uses might be scattered acrossthe landscape. Each parcel might be devoted todelivery of some set of values with the designateduses moving across the landscape over time, asnature moved them.

Policymakers have tried to solve this challenge byimplementing land allocations akin to land usezoning. Land allocations in national forest plansare a simple version of this. Our research chal-lenge is to sufficiently understand the broad-scalephenomena so that creative solutions may bedeveloped to provide an array of values that blenduses across the landscape, and even shift usesacross time in a way that increases benefits andreduces negative effects and risks.

Building on the Foundation of PastResearch

Many PNW scientists and their colleagues havepresented examples of landscape-scale researchat this workshop. The PNW Station is making aconcerted effort to address the landscape scale.Broad-scale studies like that of land cover dynam-ics and land use changes described by Ralph Aligand the study of the implications of scale for thestudy of terrestrial wildlife populations by MartinRaphael are only two examples of the work pre-sented here. The poster session showed a varietyof projects related to landscapes that PNW hasunderway. And yet during this workshop, althoughwe presented much, it was not all work conductedat the landscape scale by the Station. Some land-scape research spans decades for issues suchas insects and disease. Some of our researchhas just begun, and information is yet to bedeveloped.

As with the rest of our research, much of thiswork on the landscape scale is conducted withcolleagues at universities and through partner-ships with land managers. John Cissel’s presenta-tion of the Augusta Creek study was an example.Critical in our success is the ability to work withthe National Forest System and other land man-agers on large tracts of land. The contribution wehave made so far, our continued cooperation withuniversities, and our partnerships with land man-agers create a strong foundation from which tocontinue studies at a broad scale.

Challenges to Landscape-LevelResearch

In accomplishing landscape-level research, espe-cially with research integrated across many disci-plines, we face significant challenges. To makemeaningful progress, we need to face and over-come these challenges.

132

We need to fully develop concepts, models, andespecially testable hypotheses. For the most part,these do not yet exist. Questions abound, butrigorous scientific methods to answer the ques-tions need to be developed further. Without thebenefit of stronger methods, there is always therisk that the personal values of scientists will be-come embedded in the framing of questions orthe interpretation of results.

Landscape-scale research often but not always,requires integration of several scientific disci-plines. Success depends on our ability to under-stand interactions among biophysical componentsand between biophysical and social systems. Butthe science perspective of the different disciplinesoften focuses at widely divergent scales, and typi-cally with very different boundaries. For example,most social science is conducted at broad scalesand creates behavioral models developed at thebroader spatial scales but still useful at smallerscales. The biophysical sciences, on the otherhand, are challenged to “scale up” with both dataand models. For scientists to address landscapeissues, synthetic and interdisciplinary sciencemust overcome these technical and conceptualdifficulties. To date, we have mostly accomplishedintegrated research by synthesis of knowledgealready gained. Instead, we need to plan for inte-grated approaches as we develop hypotheses,concepts, models, and methods. We must gainthe ability to determine early in our research howwe want to address the pieces to understand thewhole.

Through its ability to bring together staff scientistsand those from universities, the Station has con-tributed to and will continue to work toward suchintegrated research. The Station has an advan-tage in being able to bring scientists together, andwe need to exercise that advantage.

Landscape science requires at least some large-scale experiments. Large-scale experiments arechallenges from several perspectives. One is thedifficulty of convincing land managers and thepublic that the search for knowledge should re-quire testing ideas, and we are testing them be-cause we do not know what the outcome will be,even as we are guided by a hypothesis. Even ifland managers are willing to accept and partici-pate in these experiments, management withuncertain outcomes is not consistent with the

emphasis to define objectives and hold managersaccountable for performance results. We haveovercome this challenge in some places, such asthe large-scale studies in the Willamette NationalForest, but we need to build on those few suc-cesses, including working across ownershipboundaries.

Even when we are successful with large-scale,designed experiments, we invariably have fewreplications. It is difficult to commit extensiveacreages, and it is hard to find enough compa-rable sites. We need to somehow develop meansof testing hypotheses that are not so dependenton large sample sizes. We will never have thesample sizes typical of fine-scale studies, but thatcannot stop us from drawing defensible infer-ences.

We need to expand our capability to draw infer-ences from other than designed experiments.Modeling has long been a tool to learn about rela-tions. Models will play an even more importantrole in large-scale research, but we need to de-velop better means to measure the confidence wecan place in their results. We need to developnew methods to reveal the confidence, or in turn,the uncertainty, of our scientific information thatdescribes large-scale relations.

Landscape research requires vast amounts ofdata. Other opportunities for collection of impor-tant landscape data may emerge as monitoringdata are generated, such as is proposed for thefederal lands in the range of the northern spottedowl. We will have to develop more rigorous dataaccumulation and data management systemsthan customarily are needed for finer scale re-search. This technological challenge emergedwith the study of landscape relations in the interiorColumbia basin as in the work described by PaulHessburg. Just as important as the technology,however, is the discipline to use it.

Successfully addressing landscape-scale issuesrequires complex decisionmaking, often involvingmultiple agencies and landowners. A willingnessto research transboundary effects is necessary,and success in both science and implementationwill require public-private cooperation. Thesetransboundary challenges highlight the impor-tance of socioeconomic landscape research.Biophysical research alone simply will not be suc-cessful in providing the scientific information todeal with larger scale issues.

133

Landscape science may be burdened with thea priori expectation that it will produce more thanscientific information. Scientists, managers, andthe public may develop exaggerated expectationsabout the nature of results of studies about land-scapes, perhaps even more than about finerscaled research studies. Sound scientific informa-tion is an essential ingredient to sound decisions,but information alone does not make decisions.Any decision invariably requires the integrationof multiple components that can only be accom-plished by value-based weighing of tradeoffsamong those components. That is the stuff ofdecisionmaking, not science. While insisting thatthe available science must be fully and faithfullyconsidered in the decision process, those respon-sible for research results must be careful to avoidthe misperception that science “makes” decisions.

Conclusion

We have a strong foundation of research onwhich to build future landscape-scale research,and the PNW Station and others in the sciencecommunity have already begun to do landscape-scale research. There are compelling scientificand policy reasons for conducting landscape-scale research, especially research integratedacross scientific disciplines. Working with ourresearch collaborators and land managementpartners, we hope to build on this foundation andmake meaningful advances—advances that maynot only help us better estimate the conse-quences and risks of current management op-tions but also help us all create new options thatmight embody more compatibility among themany values society seeks from the land.

This page has been left blank intentionally.Document continues on next page.

This page has been left blank intentionally.Document continues on next page.

This page has been left blank intentionally.Document continues on next page.

Pacific Northwest Research Station

Web site http://www.fs.fed.us/pnwTelephone (503) 808-2592Publication requests (503) 808-2138FAX (503) 808-2130E-mail pnw_pnwpubs@fs.fed.usMailing address Publications Distribution

Pacific Northwest Research StationP.O. Box 3890Portland, OR 97208-3890

The Forest Service of the U.S. Department of Agriculture is dedicated to the principle of multiple usemanagement of the Nation’s forest resources for sustained yields of wood, water, forage, wildlife, andrecreation. Through forestry research, cooperation with the States and private forest owners, andmanagement of the National Forests and National Grasslands, it strives—as directed by Congress—toprovide increasingly greater service to a growing Nation.

The U.S. Department of Agriculture (USDA) prohibits discrimination in all its programs and activities on thebasis of race, color, national origin, gender, religion, age, disability, political beliefs, sexual orientation, ormarital or family status. (Not all prohibited bases apply to all programs.) Persons with disabilities whorequire alternative means for communication of program information (Braille, large print, audiotape, etc.)should contact USDA’s TARGET Center at (202) 720-2600 (voice and TDD).

To file a complaint of discrimination, write USDA, Director, Office of Civil Rights, Room 326-W, WhittenBuilding, 14th and Independence Avenue, SW, Washington, DC 20250-9410 or call (202) 720-5964 (voiceand TDD). USDA is an equal opportunity provider and employer.

Pacific Northwest Research Station, 333 S.W. First Avenue, P.O. Box 3890, Portland, Oregon 97208-3890.

U.S. Department of AgriculturePacific Northwest Research Station333 S.W. First AvenueP.O. Box 3890Portland, OR 97208-3890

Official BusinessPenalty for Private Use, $300