Reprinted from . •. ForestEcologyand ; _Management Forest Ecology and Management 87 (1996) 27-39 Expanding the scale of forest management: allocating timber harvests in time and space Eric J. Gustafson 1 USDA Forest Service, North Central Forest Experiment Station, Forestry Sciences Laboratory, 5985 Highway K, Rhinelander, W! 54501, USA Accepted 26 April 1996 , . e ELSEVIER Forest Ecology and Management Aims and scope. Forest Ecology and Management publishesscientificarticlesconcernedwith forest management and conser- vation,and in particularthe applicationof biological,ecological and social knowledgeto the managementof man-made and nat- ural forests.The scope of the journal includesall forest ecosystemsof the world.A refereeingprocess ensuresthe qualityand internationalinterestof the manuscriptsacceptedfor publication.The journalaims to encouragecommunicationbetweenscien- tists in disparate fields who share a common interest in ecology and natural-resourcemanagement, and to bridge the gap between researchworkersand forest managersin the fieldto the benefit of both.The journal shouldbe of interestto research workers, managersand policymakersin forestry,naturalresources,ecologicalconservationand relatedfields. FOUNDING EDITOR LaurenceL. Roche, Murroe,Ireland EDITORS-IN-CHIEF -Forthe Americas,Australia,New Zealand and the Pacific: For the restof the world: R.F. Fisher G.M.J. Mohren Departmentof Forest Science ForestProductionEcologyGroup Departmentof VegetationEcology Texas A&M University DLO-Institutefor Forestryand NatureResearch College Station,TX 77843-2135, USA P.O. Box23 6700 AA Wageningen,The Netherlands BOOK REVIEW EDITOR D. Binkley,Forestand Wood Science,Colorado State University,Fort Collins,CO 80523, USA EDITORIAL ADVISORY BOARD G. Abrahamsen,AgriculturalUniversityof Norway,_s, J.A. Maghembe,SADCC/ICRAF AgroforestryProject, Norway Zomba, Malawi M.R. Ahuja, BFH, Instituteof ForestGenetics, F. Makeschin,InstitutfQrBodenkundeund Standortslehre, , Grosshansdorf,Germany Tharandt,Germany R. Alfaro, Canadian Forestry Service,Victoria,B.C., D.C. Malcolm, Universityof Edinburgh,Edinburgh,UK Canada E. Miilk6nen, FinnishForestResearch Institute,Vantaa, F. Andersson,SwedishUniversityof AgriculturalSciences, Finland Uppsala,Sweden M.A.R. Nahuz, Institutode PesquisasTecnologicas, P. Attiwill,Universityof Melbourne,Parkville,Vic., Australia S_ioPaulo,SP, Brazil J.C. Calvo, Instituteof Technology,Cartago, Costa Rica R. P_iivinen,EuropeanForestryInstitute,Joensuu,Finland J.D. Deans, Instituteof TerrestrialEcology,Penicuik, R.F. Powers,PacificSouthwestResearchStation,Redding, Midlothian,UK CA, USA S. Diamandis,ForestResearch Institute,Thessaloniki,Greece T. Pukkala,Universityof Joensuu,Joensuu,Finland D.P. Dykstra,CIFOR, Jakarta, Indonesia L. Rasmussen,DanishForestand LandscapeResearch E.P. Farrell,UniversityCollegeDublin,Dublin, Ireland Institute,Hoersholm,Denmark P.H. Freer-Smith,Forestry Commission,Farnham,UK D.D. Reed, MichiganTechnologicalUniversity,Houghton, O. Garcia, ENGREF, Nancy, France MI, USA R.A. Goyer, LouisianaState University,BatonRouge, LA, G. Ruark, USDA ForestService, Washington,DC, USA USA R. Sands, Universityof Canterbury,Christchurch,NZ J.B. Hall, UniversityCollegeof North Wales, Bangor,UK J.A. Stanturf,Stoneville,MS, USA F. Houllier,FrenchInstituteof Pondicherry,Pondicherry,India O. Sziklai, Universityof BritishColumbia,Vancouver, B.C., B.M. Kumar,KeralaAgriculturalUniversity,Kerala, India Canada J.P. Lassoie,CornellUniversity,Ithaca, NY, USA K. von Weissenberg,Universityof Helsinki,Helsinki,Finland J.N. Long,Utah State University,Logan, UT, USA A.E. Lugo, InternationalInstituteof TropicalForestry, Rio Piedras, PR, USA Publication Information: Forest Ecology and Management (ISSN 0378-1127). For 1996 volumes80-89 are scheduledfor pub- lication.Subscriptionpricesare available uponrequest from the Publisher.Subscriptionsare acceptedon a prepaid basis only and are entered on a calendaryear basis.Issuesare sentby surface mail exceptto the followingcountrieswhere air deliveryvia SA/ mail is ensured: Argentina,Australia,Brazil, Canada, Hong Kong, India, Israel, Japan, Malaysia, Mexico, New Zealand, Pakistan, PR China, Singapore, South Africa, South Korea, Taiwan, Thailand, USA. For all other countries airmail rates are available on request. Claims for missing issues should be made within six months of our publication(mailing) date. Please address all your requestsregarding orders and subscriptionqueries to: Elsevier Science B.V., Order Fulfilment Department, P.O. Box 211, 1000 AE Amsterdam, The Netherlands, tel. (+31-20) 4853642, fax (+31-20) 4853598. In the USA and Canada: For further informationon this and other Elsevierjournals please contact: Elsevier Science Inc., Journal Information Center, 655 Avenue of the Americas, New York, NY 10010, USA. Tel. (212) 6333750; fax (212) 6333764; telex 420-643 AEP UI. • Back volumes: Please contact the Publisher. Forest Ecology • and Management ELSEVIER Forest Ecology and Management 87 (1996) 27-39 Expanding the scale of forest management: allocating timber harvests in time and space Eric J. Gustafson 1 USDA Forest Service, North Central Forest Experiment Station, Forestry Sciences Laboratory, 5985 Highway K, Rhinelander, W154501, USA Accepted 26 April 1996 Abstract This study examined the effect of clustering timber harvest zones and of changing the land use categories of zones (dynamic zoning) over varying temporal and spatial scales. Focusing on the Hoosier National Forest (HNF) in Indiana, USA as a study area, I used a timber harvest allocation model to simulate four management alternatives. In the static zoning alternative, harvests were dispersed throughout the timber harvest land base (65% of HNF) for 15 decades. The three dynamic zoning alternatives varied in the degree to which harvests were clustered in time and space. Two levels of harvest intensity were simulated, and at each level of harvest intensity, the area harvested was held constant among all four zoning alternatives. The dynamic zoning strategies resulted in substantial increases in the amount of forest interior and reductions in the amount of forest edge across the landscape, as well as an increase in the average age of stands when harvested. The greatest reduction in fragmentation was produced by the alternative that most tightly clustered harvests in time and space (i.e. intensive harvesting of small blocks in a relatively short time). When harvest intensity was high, this alternative produced amounts of forest interior and edge comparable to those of the dispersed alternative with half the rate of harvest. The results suggest that the injection of dynamics in specifying disturbance regimes, and the clustering of disturbance in time and space_ can be used to sustain larger blocks of mature forest than can static zoning. Dynamic zoning encourages explicit specification of the disturbance regimes that will be imposed across the land base over long periods of time. Keywords- Forest managementplanning;Fragmentation;Disturbance;Forestinterior;Forest edge; Multiple use; Temporal scale; Clustering timberharvests;Simulation modeling 1. Introduction is mandatedby law to provide for multiple uses and values through the National Forest Management Act Forest management has become controversial, of 1976. Industrial forest owners have also made a stemming from fundamental differences in how for- commitment to provide multiple forest values in the est resources are viewed by different segments of management of their forest lands (Wallinger, 1995). society. The management of federally owned forests Forest management plans typically allocate the land base among several land use categories, and projec- • tions are made of the impacts of the plans on a suite t Tel.: 715-362-1152; fax: 715-362-1166; e-mail: of forest values, including biological diversity, recre- ericgus@newnorth.net, ational opportunities, and commodity production. 0378-1127/96)$15.00 Copyright © 1996 Elsevier Science B.V. All fights reserved. PH S0378-1 127(96)03838-8 28" E.I. Gustafson / Forest Ecology and Management 87 (1996) 27-39 . Because of this goal to provide for multiple uses, ditions of managed forests (Crow and Gustafson, planners often find themselves attempting to provide 1996). for mutually exclusive uses of land, such as timber A poorly understood consequence of static zoning production and old-growth forest. A typical model is is that forest age class distributions become skewed to designate several land use categories, and to allo- over long time periods (Gustafson and Crow, 1996). cate land to these categories, thus grouping suites of Stands in timber production areas are kept in rela- compatible land uses into spatially defined zones, tively early seral stages; other management areas Examples of objectives associated with various land experience little disturbance, and the forest in those use categories might be 'a physical setting to provide areas will eventually be dominated by late-seral types. opportunity for solitude and a feeling of closeness to Intermediate seral stages should become rare as a nature'; or 'provide for recreation facilities'; or consequence of the deterministic disturbance regime 'maintain habitat diversity, provide a sustained yield imposed by static management strategies over long of timber, and provide dispersed recreation opportu- periods of time, and community composition may nities.' Usually several blocks of land (management change markedly. Deterministic disturbance regimes areas) are allocated to each land use category, and may reduce the natural variability of landscapes, these blocks are dispersed throughout the forest, resulting in undesired ecological conditions ostensibly to provide the values associated with each (Mladenoff and Pastor, 1993; Swanson et al., 1994; category across the landscape. This approach pro- McCarthy and Burgman, 1995). vides for multiple uses at the landscape scale, but A recent trend in US National Forest management may not adequately integrate multiple uses within has been a reduction of more than 50% in timber each management area (Behan, 1990). For example, production since 1988, to the lowest levels since timber production has increasingly been viewed as about 1955 (Haynes et al., 1995). This trend has being incompatible with many non-commodity uses primarily been in response to pressure to provide for of the forest, and is often segregated from those uses. more non-commodity values from National Forests, Forest plans typically consider a 50-year planning such as wildlife habitat and a natural setting in which horizon. In many cases, it is not possible to provide to experience nature. As an example of this, the all potential uses within a management area over a Hoosier National Forest in southern Indiana amended .. 50-year period. However, interesting possibilities its 1985 Forest Plan, which emphasized clearcutting arise whenconsidering longer temporal scales, on 85% of the land base (USDA Forest Service, Should the designation of the land use(s) within a 1985), changing the management emphasis to un- management area be static for long periods of time even-age management.-This amendment reduced the (static zoning), or should it be dynamic, with several expected timber output by 60% and set aside 60% of land uses rotating among several management areas the land base for non-commodity purposes (USDA (dynamic zoning) at a scale of centuries? For exam- Forest Service, 1991). ple, timber production might be allowed periodically It remains to be seen if a policy of sharply in a non-timber production area to prevent native curtailed commodity production will be socially ac- oak-hickory forests from succeeding to beech-ma- ceptable in the long term. Virtually every member of ple. On the other hand, timber production areas society uses wood-based products, and the demand could be allowed to lay fallow to provide non-com- for wood is projected to rise more than 60% by the modity values for some period of time. The current year 2040 (Haynes et al., 1995). Forest products are management paradigm appears to allow for spatial renewable, unlike many substitutes. Reduced timber and temporal management dynamics within a man- production on federal lands increases demand for • agement area, but little thought has yet been given to private and foreign timber. Industrial forest owners .dynamics in the designation of management areas also experience pressure to provide non-commodity over long time periods (over 50 years). The interac- values from their forests. Forest planners have the tion of the spatial and temporal domains of manage- unenviable task of attempting to balance the conflict- "ment activity has been inadequately explored, but ing demands by society for commodity and non- has significant consequences for the ecological con- commodity values from forested lands. The chal- .,. x EJ. Gustafson / Forest Ecology and Management 87 (1996) 27-39 29 lenge will be to develop new management paradigms gated, the disturbance occurs in a relatively small that allow commodity production while maintaining area over a short time period, and a relatively long _.,_,_.,_,,_.... •, non-commodity values, period free from harvest disturbance follows. Thus, One of the potential ecological consequences of dynamic zoning produces a clustering of harvest timber harvest is a reduction in the amount of habitat disturbance in both space and time. Dynamic zoning for forest interior species, many of which are thought is a potential tool to produce dynamic landscape to be declining in abundance (Robbins et al., 1989; heterogeneity (Mladenoff and Pastor, 1993) by im- 1 Hill and Hagen, 1991). Most harvest methods create plementing harvesting cycles, and encouraging ex- Openings that perforate blocks of contiguous forest plicit specification of disturbance regimes over large I and introduce edge habitat within the forest. Many spatial and temporal scales. interior Species are thought to be sensitive to the size In this study, I used a timber harvest allocation • of forested blocks (Blake and Karr, 1987; Freemark model to compare four cutting strategies that differedt and Collins, 1992), and internal edges may provide in the spatial and temporal dispersion of harvest improved habitat for generalist predators and brood allocations. My objective was to quantify the changes parasites (Gates and Gysel, 1978; Brittingham and in forest interior habitat and forest edge produced by " ' Temple, 1983; Small and Hunter, 1988; Robinson et different harvest dispersion strategies, providing in- al., '1995). : sight into the utility of dynamic zoning strategies for The practice of dispersing cutting units has been forest management. Recent studies have demon- " " implicated as a major contributor to the reduction in strated the value of clustering harvests spatially forest interior habitat and the increase in linear edge through time (Li et al., 1993; Gustafson and Crow, (Franklin and Forman, 1987; Gustafson and Crow, 1994; Wallin et at., 1994), but here I also examine ' !994; Wallin et al., 1994). Progressive cutting across the effect of dynamically changing the locations of the landscape has been proposed as an alternative to timber harvest zones. the traditional approach of dispersing cutting units across the landscape (Li et al., 1993). Under this , strategy, timber harvesting would proceed somewhat 2. Methods systematicall _, across the landscape. Openings pro- duced by harvest would be clustered in both time 2.1. Study area and space, allowing more interior habitat to be sus- tained on the landscape as a whole. The practical The study was conducted on a rectangular study " application of this approach is complicated by dis- area (1058046 ha) that included the entire Hoosier continuous ownership of the landscape and the vari- National Forest (HNF) Purchase Area, located in ability in the suitability of stands for harvest at any southern Indiana, USA (Fig. 1). The HNF was used J given point in time. A variant of this approach might to provide realistic ownership and Management Area be tO progressively designate timber harvest manage- (MA) patterns for assessing alternative cutting strate- •ment areas across the landscape over successive gies. The HNF is typical of National Forests in the planning periods (dynamic zoning). This would also eastern USA in that the ownership pattern is highly . have the effect of clustering harvest openings within fragmented by privately-owned inholdings, and the , the larger landscape, but would allow more flexibil- HNF owns only about 43% of the land within the ity in the placement of individual harvest treatments Purchase Area. In published Forest Plans, MA , within the management area. Flexibility at the water- boundaries have been drawn that specify the man- shed scale is essential to mitigate the effects of agement direction for the federally owned land within cutting on stream flow and sediment production each MA. I defined the land base on which timber _ (Hombeck and Swank, 1992) and to protect special harvest was to be simulated using the MA bound- res0urce'features and habitats (Naiman et al., 1993). aries of the 1991 HNF Amended Plan (USDA Forest Spatial clustering of harvests by progressive cutting Service, 1991). Ownership boundaries were digitized also has, implications for disturbance (by harvest) from 1:24000 scale paper maps produced by the US return intervals. When harvests are highly aggre- Geological Survey (USGS) for the Forest Service • .o • , • . .. 30 EJ. Gustafson / Forest Ecology and Management 87 (1996) 27-39 UNITED STATES tized. A forest cover map of the entire Purchase Area t was generated from USGS-Land Use Data Acquisi- tion (LUDA) data, and all layers were gridded to a common cell size of 100 by 100 meters. It was not feasible to idigitize stand age maps of the entire HNF and stand age data were not available for private I assumed that the distribution of pastland, SO .... .. harvest activity (and therefore stand ages) is spatially-. •.' random on the HNF. I tested the assumption that stands reaching rotation age and past harvest alloca- I sTUDY AREAN tions are randomly distributed using nearest-neighbor • / analysis (Davis, 1986) on ten subsets of HNF stand maps (mean (+ SD) size of subsets 3366 + 1062 ha). ' The observed mean nearest-neighbor distance be-o tween stands of similar age was compared with the o Fig. 1. Location of study area. distance expected if stands were randomly dis- - ' tributed, and a z-statistic was computed. The null (Fig. 2(a)). MA bounclaries were manually trans- hypothesis that stands are randomly distributed could • ferred from _e Forest Plan maps (approximate scale, not be rejected at the 95% confidence level for eight 1"127.000) to 1"100000 scale USGS maps and digi- of the ten subsets (see Gustafson and Crow, 1996). __" 5 _"2"; .. .. r _'4 I '1"4 i _] " I J" " _.V_,. , 9"" '%r : '_'e" -_" ,._., 2: C "'*:" ' ' " ¢2';..-, ' % -_ ]l_ai,T;" 'I ,, SCALE '_.,,_-_ b'" 'l I KILOMETERS _ Fig. 2. Map Of (a) distribution of land owned by the HNF within the study area and (b) timber land base on which harvests were simulated. The solid-line rectangles represent the subsetsused for the 50-year and 100-year hiatus alternatives, and the dashedlines represent the subsetsused for the 120-year hiatus alternative. Upper case letters indicate the order in which timber harvest was allowed on the subsetsfor the 50- and 100-year hiatus alternatives, and the numbers indicate the order in which timber harvest was allowed on the subsets for the 120-year hiatus alternative. • • • EJ. Gustafson / Forest Ecology and Management 87 (1996) 27-39 31 2.2. Timber harvest allocation model existed in timber production areas during the initial . decades of simulation to meet target harvest levels. HARVEST is a timber harvest allocation model A consequence of this procedure was that the distri- that was constructed to allow the input of specific bution of stands less than 20 years of age (openings) rules to allocate forest stands for even-age harvest in the first two decades of simulation was not explic- (clearcuts and shelterwood) and group selection, us- itly modeled, so the initial forest condition appeared ing parameters commonly found in National Forest less fragmented than is probably the case. Plan standards and guidelines. The model produces landscape patterns that have spatial attributes result- 2.3. Experimental design ing from the initial landscape conditions and the propose d management activities. The model is sim- The land base harvested over a period of 15 plistic in that it does not attempt to optimize timber decades was determined by the Management Area ' production or quality, nor does it predict the specific boundaries specified in the 1991 HNF Amended locations of future harvest activity,, as it ignores Plan (USDA Forest Service, 1991). In the 1991 many considerations such as visual objectives and Amended Plan, timber harvest was allowed on 39 299 road access. Instead, the model stochastically mimics ha, but for the alternatives simulated in this study I , ° the ,allocation of stands, for harvest by forest plan- also allowed timber harvest on an additional 9585 " ners, using onlythe constraints of the standards and ha, to allow for higher timber outputs than projected . • guidelines and MA boundaries. Modeling this pro- under the 1991 Amended Plan. Timber production cess allowsexperimentation to link variation in man- was allowed on approx!mately 65% of the HNF land agement strategies with the resulting pattern of forest base, and only on land owned by the HNF (Fig. , openings. 2(b)). HARVEST was constructed to be used in con- The experimental treatments consisted of altema- junction with a grid-cell Geographic Information tive designations of timber harvesting areas on the System (GIS), with routines for direct input and HNF that varied as to where timber harvest was oUtput ofERDAS v. 7.5 GIS files, but supporting allowed during each decade and for how many ' files exported in text format from other raster GIS decades it was allowed there. The total land base that systems. Timber harvest allocations were made by was harvested (timber harvest land base) over a the model using a digital stand map, where grid-cell period of 15 decades was identical among all altema- values reflect the age (in years) of the forest in that tives. For the static zoning alternative, harvest was ceil. HARVEST takes a GIS stand age map as input, allowed throughout the timber harvest land base and produces a new stand age map incorporating during all 15 decades. Three dynamic zoning alterna- harvest allocations. HARVEST allows control of the tives were simulated. For the '50-year hiatus' alter- ," size distribution of harvests, the total area of forest to native, the timber land base was divided into three be harvested, and the rotation length (by specifying subsets; timber harvest was allowed on only two of the minimum age on the input stand map where these subsets at a time (beginning with subsets A and harvests may be allocated). HARVEST selects har- B, Fig. 2(b)), and the third was temporarily set aside vest locations randomly within currently active tim- from timber harvest for 50 years. The treatments ' her production MAs, checking first to ensure that the were rotated every 5 decades, so that each subset forest is old enough to meet rotation length require- was harvested for 10 successive decades and then set , ments. This is .consistent with the random distribu- aside from timber harvest for 5 decades. For the tion of past harvest activity, as discussed above. '100-year hiatus' alternative, the same three subsets Since the initial forest ages were unknown, but were used, but timber harvest was allowed on 0n!Y • assumed to be spatially random, I allowed the model one of these subsets at a time (beginningwith subset to choose harvest locations randomly from all cells A, Fig. 2(b)), and the other two were temporarily set within timber production areas by assigning all forest aside from timber harvest. Again, the treatments an initial age of 100 years. This assumed that suffi- were rotated every 5 decades, and each subset was cient area of forest old enough to be harvested harvested for 5 successive decades and then set aside • 32 EJ. Gustafson/ Forest Ecology and Management 87 (1996) 27-39 r .............. :7 --.-.-- F- for 10 decades. Finally, for the ' 120-year hiatus' pattern (static, 50-year hiatus, 100-year hiatus, and alternative, the timber harvest land base was divided 120-year hiatus), and two levels of harvest intensity into'five subsets, and each subset was harvested for 3 (1300 ha per decade and 2600 ha per decade). Three decades (beginning with subset 1, Fig. 2(b)) and then replicates of each factorial combination were pro- was set aside for 12 decades. Total area harvested, duced. : size of harvest openings, and rotation interval (minimum age for cells to be harvested) were held constant across all treatments and decades, so that 2.4. Analysis timber productiOn was the same for all four scenar- ios. At each decade, I used a GIS to determine the Timber harvest parameters were chosen to fall amount of forest interior habitat (forest over 300 m within the parameter space of the 1985 Plan and from an opening or edge). A simple FORTRAN 1991 Amended Plan alternatives simulated elsewhere routine was written to calculate linear forest edge. " (Gustafson and Crow, 1996) and are detailed in Clearcut stands in the HNF generally achieve canopy Table 1. Only harvest methods that produce forest closure in 12-20 years (T. Thake, personal commu- o openings (clearcut, shelterwood and seedtree) were nication, 1993), so cells harvested were assumed to simulated,, and harvest placement was not con- create openings in the forest for 20 years and were Strained by adjacency prohibitions. The intensity of then assumed to return to a closed canopy condition. • harvest (total area harvested) is an important deter- A different definition of canopy closure would change minant of forest interior and edge (Gustafson and the absolute amount of interior and edge, but the Crow, 1994), so I conducted simulations at two relative differences among alternatives would be levels of harvest intensity, one having twice as much similar to those reported here. The total amount of area harvested each decade as the other. This al- edge and interior was plotted over simulated time for lowed me to assess the relative contribution of both each of the alternative cutting patterns. For compari- dispersio n of harvests and intensity of harvest to the son, the results of simulating the original 1985 Forest amount of forest interior and edge. Higher levels of Plan and the 1991 Amended Plan (Gustafson and sustained harvest were not possible without increas- Crow, 1996) were also plotted. ing the timber land b_e. Thus, a complete factorial To evaluate the relative effects of harvest disper- design was implemented, with four levels of cutting sion and harvest intensity on forest interior and edge, Table 1 Harvest .intensities used in the simulation of timber harvest alternatives on the Hoosier National Forest. The High and Low Intensity parameters were used for the simulation of the static and dynamic zoning cutting alternatives. Analysis included three replicates of simulations conducted for 15 decades Model parameter Low Intensity High Intensity 1985 Plan 1991 Plan • " Mean clearcut opening size (ha) 5.0 5.0 4.0-7.0 2.8 Mean group 9Pening size (ha) NA NA 0.4 0.2 i • M_imum Opening Size (ha) 8.0 8.0 10.8 4.0 Total harvested per decade a (ha) 1300.0 2600.0 5709.6 1267.0 Harvest iate per decade t, (%) 2.6 5.3 10.5 3.2 Rotation length (years) 100 100 80-120 80 " Timber land base (ha) c 48884 48884 56279 39299 a Represents harvest activity across the entire forest. Total HNF ownership is approximately 84 774 ha. b Represents the percentage of forest within the timber harvest land base that is harvested each decade. c Total area where harvest is allowed during at least part of the 15-decade simulations. • EJ. Gustafson / Forest Ecology and Management 87 (1996) 27-39 33 an ANOVA. was used to test for treatment effects time periods were included in the analysis to account reflecting harvest dispersion (DISPERSION), harvest for the potential correlation of measures of forest " intensity (INTENSITY), and time (DECADE). The interior and edge in successive decades. [......"'"_ ........"- ' STATIC 50-YR a HIATUS b | •. , ,_i. __. _ _:_ _' _ Lz4\ • i_,'_,' . ,_' ,,.-. ,,,, .. . • _ NON-FOREST/HARVESTED SCALE FORESTEDGEHABITAT KILOMETERS FORESTINTERIOR 2o o4 .. 100-YR 120-YR.. •. HIATUS HIATUS C. d f ,'._-. :_:,. Fig. 3. Forest interior in the study area at the end of 15 decades of simulated harvest at the 'High intensity' harvest rate (2600 ha per decade) under the four zoning alternatives. The solid lines represent the approximate location of the HNF Purchase Boundary, and simulated harvests Occurred only on HNF land within those boundaries. • o 34 EJ. Gustafson / Forest Ecology and Management 87 (1996) 27-39 3. Results _ STATIC ZONING - -v- 120-YEAR HIATUS - - -a - 50-YEAR HIATUS - -O - 1985 PLAN J ;1_ •::_-._._+_,-__,+_+._..,..... r100-YEAR HIATUS + 1991 AMENDED PLAN The total amount of forest interior varied markedly 220o among the simulated altertlatives (Fig. 3), with the _ ^..} highest amount produced ,by the pattern that most 210o •• _ • Y • tightly clustered harvests in time (i.e. longest hiatus '_ 2000 - _period) and space (120-year hiatus, Fig. 3(d)). Under o I_ 1900 the static alternative, none of the timber land base • z 1800 was set aside at any time; large amounts of forest 7- • 00 t_ edge habitat can be seen scattered throughout the _ 17oo HNF Purchase Area, and few blocks of forest inte- o , , ," , , , , , , , , ,,,riorremain (Fig. 3(a)).+Under the dynamic zoning is°° 1 "_*"**'*"*'_°"'*_- i alternatives, increased amounts of forest interior can lsoo be seen in the.areas that had just completed their 0 3 6 9 12 15 fallow period; for+example, examine subset B (Fig. DECADE . 2(b), Fig. 3(b)) And subsets A and B (Fig. 2(b), Fig. Fig. 4. The amount of forestinteriorproducedby timberharvest 3(C)), representing the 50-ye_ and 100-year hiatus alternatives over simulated time on the Hoosier National Forest alternatives, respectively. Forest interior reaches its (HNF).The 'Static' alternative is the least clustered in space and • time, while the '120-year hiatus' alternative is the most clustered • highest levels on _e landscape as a whole under the in space and time. The results plotted are for the 'High intensity' 120-year hiatus alternative (Fig. 2(b), Fig. 3(d)). harvest rate (2600 ha per decade)- The 1985 Plan and 1991 Timber production is evident in subset 5 under this Amended Plan are simulations of published plans for the HNF +alternative, where the density of harvest openings is (Gustafson and Crow, 1996), and are shown for comparison. quite high, due to the high level of clustering. The replications of the simulations produced little variability in forest interior and edge. The variability openings existed on two zones, perforating contigu- was too low to show clearly with error bars on line ous forest habitat across a broader portion of the graphs, so error bars are not shown. The standard landscape. One might expect that forest edge would deviation from the mean of interior area produced by not show such a pattern, since edge is introduced three replicates never exceeded 0.5% in any combi- around an opening, regardless of the spatial disper- nation of treatments, and the standard deviation from sion of the openings. However, when harvest inten- the mean linear edge never exceeded 0.02%. sity was high, openings (cells less than 20 years old) The dynamic zoning strategies resulted in more within the timber production zones begin to coalesce, forest interior across the landscape than the static reducing edge. The oscillation in the amount of edge / zoning alternative (Fig. 3), with the highest amount seen in Fig. 5 reflects this periodic coalescence of produced by the pattern that most tightly clustered openings near the end of production in a zone, and harvests in time and space (120-year hiatus, Fig. generation of relatively higher amounts of edge when 3(d), Fig. 4). The dynamic zoning strategies also new cutting zones were opened. This oscillation is resulted inless forest edge across the landscape than not evident at low harvest intensity (not shown). .. the static zoning alternative, with the least amount Levels of fragmentation under a dynamic zoning againproduced by the pattern that most tightly clus- alternative with a high intensity of harvest were tered harvests in time and space (120-year hiatus, comparable with those produced by the static altema- " Fig. 5). tive with a low level of harvest (Fig. 6). Note in Fig. • The periodic rise and fall in the amount of forest 4 that even the 50-year hiatus alternative (high inten- interior and forest edge evident in the dynamic zon- sity is plotted) produced approximately the same ing alternatives' was caused by the initiation of cut- amount of forest interior as the 1991 Amended Plan, ting on a new cutting zone. Openings were produced even though the total area harvested under the 50-year in the new zone before all the openings closed on the hiatus alternative was twice that of the 1991 Amended previous zone, so that for a 2-decade period harvest Plan (Table 1). With a cutting intensity similar to [] --- / L ' " + • , " . E.I. Gustafson / Forest Ecology and Management 87 (1996) 27-39 35 ,.,' STATIC ZONING -.-v - 120-YEAR HIATUS 2200 . - -m- 50-YEAR HIATUS - 4- 1985 PLAN _ 120-YEAR HIATUS, HIGH INTENSITY •_ _ --D - STATIC, LOW INTENSITY100-YEAR HIATUS ---B-- 1991 AMENDED PLAN o_ 2150 -E 17500 _ _ e..e.e.t.e.o-_._ _ 2100 - - 0 _'. 170® " _- _, I_ 2050- - tu z_ 16500 I-- ° |_ 2000- -W w m 16000 0 tu u.. 19501 n,, . b-H...._ _ ..=-.H....m-0 a It. n," 15500 1900 ' ' i ' ' i , , i , , i , ' z 15000 0 3 6 9 12 15 ..I DECADE 1450o Fig. 6. Amount of forest interior produced over simulated time by 0 3 6 9 .12 15 the '120-year hiatus' alternative at high-intensity harvest (2600 ha DECADE per decade) and the 'Static' alternative at low-intensity harvest Fig. 5. Amount of forest edge produced by timber harvest altema- (1300 ha per decade). The 'Static' alternative is the least clustered . fives Over simulated time on the Hoosier National Forest (HNF). in space and time, while the '120-year hiatus' alternative is the The 'Static' altemat.ive is the least clustered in space and time, most clustered in space and time. while the '120-year hiatus' alternative is the most clustered in space and time. The results plotted are for the 'High intensity' forest. The 1991 Plan, with an intensity of harvest harvest rate (2600.ha per decade). The 1985 Plan and 1991 approximately half thafof the high-intensity dynamic Amended Plan are simulations of published Plans for the HNF zoning altemative_, produced higher amounts of edge (Gustafson and Crow, 1996), and are shown for comparison, due to the use of group selection. • Differences in the spatial dispersion of harvests that of the 1991 Plan (low intensity), all the altema- (zoning) appear to have a greater effect on the tires, including the static one, produced more forest amount of forest interior than do differences in har- . interior than the 1991 Plan (plot not shown). This vest intensity. All the main effects are highly signifi- wasdue to the use of smaller openings in the 1991 cant in the ANOVA models; however, examination Plan, including extensive use of group selection, of the sums of squares shows that the spatial disper- which resulted in more openings that perforated the sion of harvests (DISPERSION) explains 49.6% of L Table 2 Analysis of variance comparing the effects of harvest intensity (INTENSITY), the spatial dispersion of harvest activity (DISPERSION), and the time period simulated (DECADE) on the area of forest interior and linear forest edge maintained on the landscape. Analysis included three replicates of simulations conducted for 15 decades Source d.f. Forest interior (kin 2) Forest edge (kin) SS F Prob > F R 2 SS F Prob > F R2 INTENSITY 1 427298 402.6 0.0001 16178271 861.2 0.0001 DISPERSION = 3 1190387 373.8 0.0001 5792713 102.8 0.0001 , 'S v 50 _ 1 78501 74.0 0.0001 95386 5.1 0.0249 • " 50 v 100 b - 1 219118 206.4 0.0001 1000327 53.2 0.0001 , 100v 120 c 1 56061 52.8 0.0001 717132 38.2 0.0001 • ' S&50 v 100& 120 a 1 1055825 994.7 0.0001 4980195 " 265.1 0.0001 DECADE 14 421361 28.4 0.0001 8604716 32.7 0.0001 Error 341 361959 6406158 Total 359 2401006 0.85 36981858 0.83 a Orthogonal contrast of the Static (S) zoning alternative with the 50-year (50) hiatus alternative. b Orthog0nal contrast of the 50-year (50) hiatus alternative with the 100-year (100) hiatus alternative. c Orthogonal contrast of the 100-year (100) hiatus alternative with 120-year (120) hiatus alternative. a Orth0gona! contrast of the Static (S) and 50-year (50) hiatus alternatives with the 100-year (100) and 120-year (120) hiatus alternatives. • = _,_ii_i_!__¸_¸ 36 E.I. Gustafson/ ForestEcologyandManagement87 (1996)27-39 the total variance 0f forest interior, while harvest produced, but from the temporal and spatial configu- intensity (INTENSITY) explains 17.8% of the vari- ration of its extraction. Specifically, as harvests be ..... _,,_ ....... , ance and DECADE explains 17.5% (Table 2). Or- came more aggregated in time (longer hiatus inter- thogonal contrasts partitioning the variation caused val) and space, the level of fragmentation decreased, by DISPERSION (Table 2) show that the greatest the average age of forests in timber management variance is explained by differences between the zones inereased, and the disturbance interval neces- 50-year and the 100-year hiatus alternatives (9.1%), sary to achieve a given level of harvest was length- 1 and that the variance explained by differences be- ened. Relative to static zoning, dynamic zoning in- tween the two Ieast aggregated alternatives (static creases opportunities to reduce the amount of edge and 50-year hiatus) and the two most aggregated and increase both the amount of interior habitat and ' alternatives (100-year and 120-year hiatus) is 44.0%. timber production by clustering harvest activity and INTENSITY is more important in explaining the lengthening disturbance intervals. Although I simu- length of forest edge, explaining 43.7% of the total lated specific hiatus intervals, the important point is ' variance, while DISPERSION explains only 15.7% not the length of these intervals, but the temporal and of the variance a_ndDECADE explains 23.3% (Table spatial dynamics of the clustering that coincidentally 2). With a given harvest size, each opening produces produced these intervals. These results were obtained a fixed amount of edge, and the number of openings by simulating dynamic zoning on a National Forest, produced is proportional to harvest intensity. The but the principle of clustering harvests in both time - • • Spatial dispersion of openings has some impact on and space can be applied to the management of any edge, in that more aggregated harvests tend to pro- large land base. Industrial forests are managed to duce a coalescence of openings that reduces the maximize mean annual increment of timber volume • _ relative amount of edge produced. Orthogonal con- and to favor the regeneration of certain tree species. trasts partitioning, the variation in edge caused by Clustering disturbance by dynamic zoning with a DISPERSION show trends.similar to those of forest rotation interval < 100 years would produce less interior, but at lower levels of variance explained fragmentation than dispersing disturbance with a (Table 2). similar interval. Dynamic zoning can also serve to Clustered harvests with longer hiatus periods re- cluster operational activities such as road improve- suited in an increase in the average age of stands in ment, access control, and site preparation, lowering timber Production zones on subsequent re-entries, costs of production. The average age of forest cells at the end of 15 The simulations reported here did not include the decades under the static alternative was 110.2 years, effects of any disturbance other than timber harvest- The average age of cells in a zone after its hiatus ing. Such effects may be significant on some land- period was 118.0 years under the 50-year hiatus scapes, but are probably minimal on the HNF. On '1 alternative, 146.2 years under the 100-year hiatus the HNF, prescribed fire has been used to maintain alternative, and 163.3 years under the 120-year hia- barrens and oak-hickory communities, but wildfire tus alternative. The dynamic zoning alternatives had is rare and localized in this moist Central Hardwood the effect :of aggregating older forest stands by clus- region. Windthrow is more common, but its effects tefing disturbance. _ are also generally local. Natural disturbance in this • region would produce some fine-grained, local patch-. . iness, but its overall impact on landscape pattern 4. Discussion _ would likely be minimal on a landscape of this size, even over a period of 15 decades. •These results demonstrate the potential benefits of Consideration of-the temporal and spatial scale of enlarging the spatial and temporal scale of forest disturbance is critical for the understanding of eco- ' management planning and of incorporating long-term logical processes (Urban et al., 1987; Wiens, 1989; temporal dynamics (dynamic zoning) into manage- Reice, 1994). The designation of MAs on managed ment plans. I found differences in forest fragmenta- forests essentially specifies the disturbance regime tion that resulted not from the amount of timber for each area of the forest. Timber harvest imposes E.J. Gustafson / Forest Ecology and Management 87 (1996) 27-39 37 periodic disturbance that changes the community at a shorter rotation, helping to enhance biodiversity and spatial "scale of several hectares, and the intensity of to maintain soil productivity (Swanson and Franklin, harvest is a major determinant of the resulting corn- 1992). A greater diversity of seral stages would exist munity structure and composition. In other MAs, across the landscape, although the interspersion of disturbance may be suppressed, with very little man- the types would be less. In addition, stands would be agement that might directly change community corn- older when they are re-entered, introducing position Or landscape pattern. Thus, static zoning economies of scale in the harvesting and processing I causes specific disturbance regimes to exist in perpe- of larger trees. However, longer rotations may pro- tuity in Specific locations. It is not clear what the duce changes in community composition that in some impacts of static zoning on biotic diversity might be cases may be undesirable. A dynamic zoning strat- over long time periods. It has long been accepted egy allows for flexibility in rotation length (or distur- that disturbance produces the spatial heterogeneity bance interval), while still retaining the benefits of ! that is necessary to maintain diversity. However, the clustereddisturbance in time and space. long-term role of disturbance is sometimes mini- Clearly, many other factors besides forest frag- mized in the management of 'natural ecosystems' mentation impact forest management plans. For ex- (AttiwiU, -1994). Non-equilibrium theories of corn- ample, increasing the area of old-growth forest would . munity structure suggest that the diversity of species be difficult on a dynamically zoned land base and and the coexistence of similar species that is seen in would probably require integration of dynamic zon- " most communities are due to some level of distur- ing with old-growth islands (sensu the 'long-rotation bance and the resulting opportunityfor recruitment island' concept of Harris(1984)). Some flexibility in of new species to the community (Connell, 1978; rotation intervals may be required to meet vegetation Huston, 1979; Lewin, 1986; Reice, 1994). How management goals on other parts of the landscape. ecosystems will respond to novel disturbance regimes Public acceptance of periodic changes in the location is not often understood (Swanson et al., 1994). For of natural appearing recreation areas is difficult to example, in the Central Hardwood region, there ap- predict, and may be problematic for the implementa- pears to be a trend toward the conversion of native tion of a dynamic zoning strategy. These issues ' oak-hickory communities to beech-maple, thought certainly need to be investigated. Nevertheless, the t0be the result of fire suppression (Lorimer, 1985). results of these simulations suggest that it may be It is far from clear how the rest of the flora and technically possible to extract timber from a large fauna might respond to the development of a forest land base while maintainingmost of that land base in " type to which they are not adapted and that may not a relatively undisturbed state for long periods of have existed in many areas since Pleistocene glacia- time. tion. The dynamic zoning alternativessimulatedhere Most eastern forests still bear the legacy of "z" do not adequately mimic 'natural' disturbance widespread disturbance and abuse, and most are regimes, and were in fact deterministic with a long relatively young. It is prudent to protect parts of the temporal period. Furthermore, I simulated a very forest from timber harvest to develop a diversity of limited set of silvicultural and management options, forest conditions across the landscape. However, this . However, the dynamic management of disturbance may be wise only in the short term, and explicit . Overlong time periods allows managers greater flex- thought must be given to the nature of disturbance •ibility, and coupled with the clustering of disturbance regimes that will be imposed across large forested . in time and space can be used to sustain larger areas over the long term. blocks of mature forest than can a static alternative. As timber production zones are more tightly clus- ' teredin time and space, the effective rotation interval 5. Conclusion becomes 'longer, providing large blocks of mature forest habitat on zones nearing the end of rotation. It is perhaps inevitable that conflicting demands . These older forests would be expected to have greater on our forests will increase. The value of forested StructUralcomPlexity than forests managed on a ecosystems for recreation and as repositories of bio- , 38 E.I. Gustafson / Forest Ecology and Management 87 (1996) 27-39 logical diversity Will increasingly be recognized, References while the deriaand for wood products will also in- crease. Pressure to provide both commodity and Attiwill, P.M., 1994. The disturbance of forest ecosystems: the non-commodity valuesfrom our forests will require ecological basis for conservative management. For. Ecol. new and creative ways to manage these valuable Manage.,63: 247-300. resources. Behan, R.W., 1990. Multiresource forest management: a paradig- A criticism of timber harvest is that it reduces the maticchallengeto professionalforestry.J. For., 88(4):12-18. BBlake, J.G. and Karr, J.R., 1987. Breeding birds of isolated habitat values of certain species of concern and woodlots:area and habitatrelationships.Ecology,68: 1724- reduces the aesthetic enjoyment of the forest. A 1734. criticism of setting aside lands from timber harvest is Brittingham, M.C. and Temple, S.A., 1983. Have cowbirds caused ' that wood fiber is locked up and wasted, putting forestsongbirds to decline? BioScience, 33: 31-35. Connell, J.H., 1978. Diversity in tropical rainforests and reefs. more pressure on other woodlands and foreign coun- Science, 199:1302-1310. tries to produce the fiber to meet the demands of Crow, T.R. and Gustafson, E.J., 1996. Ecosystem management: society. A dynamic zoning paradigm begins to sat- managing natural resources in time and space. In: K.A. Kohm isfy both these criticisms through better and more and J.F. Franklin (Editors),Creatinga forestry for the twenty- judicious integration of multiple uses. When harvests first century. Island Press, Washington, DC, pp. 424-450. are clustered in both space and time, more forest Davis, J.C., 1986.Statisticsand Data Analysisin Geology.John * interior is preserved, and more areas are distant from Wiley, New York,646pp.Franklin, J.F. and Forman, R.T:T., 1987. Creating landscape any signs of recent harvest activity. The specific patternsby forestcutting:ecologicalconsequencesand princi- location of these areas would shift across the land- pies. Landscape Ecol., 1" 5-18. scape on a time scale of several decades, but the Freemark, K. and Collins, B., J992. landscape ecology of birds amount of land in these conditions would remain breeding in temperate forest fragments. In: J.M. Hagen and D.W. Johnston (Editors), Ecology and Conservation of constant, If timber harvest activity is moved progres- Neotropical Landbirds. Manomet Bird Observatory, Woods sively across the landscape, less of the forest is Hole,MA, pp. 443-454. permanently set aside, and harvested stands will be Gates, J.E. and Gysel, L.W., 1978. Avian nest dispersionand older on average, introducing economies of scale in fledgling success in field and forest ecotones. Ecology, 59: their harvest. Even within timber harvest areas, har- 871-883. Gustafson, E.J. and Crow, T.R., 1994. Modeling the effects of vestingwould be spread out over 30-50 years, so forest harvesting on landscape structure and the spatial distri- that when stands are cut at the end of the period, the butionof cowbirdbrood parasitism.LandscapeEcol., 9: 237- Stands cut first will have regenerated to a closed 248. canopy condition. Although a significant portion of a Gustafson, E.J. and Crow, T.R., 1996. Simulating the effects of managed forest might still be set aside from timber alternativeforest managementstrategieson landscapestruc- ture.J. Environ. Manage., 46: 77-94. _,- harvest; a dynamic zoning paradigm could support Harris, L.D., 1984. The Fragmented Forest. University of Chicago higher levels of timber extraction with less forest press,Chicago, IL, 211 pp. fragmentatio n than the static zoning alternative. Dy- Haynes, R.W., Adams, D.M. and Mills, J.R., 1995. The 1995 namic zoning also encourages explicit specification RPA timber assessment update. Gen. Tech. Rep. RM-259, of the disturbance regimes that will be imposed USDAForestService. across the land base over long time scales. Hill, N.P. and Hagen, J.M., III, 1991. Population trends of some. northeastern North American landbirds: a half-century of data.,. • Wilson Bull., 103: 165-182. Hombeck, J.W. and Swank, W.T., 1992. Watershed ecosystem • analysis as a basis for multiple-use management of eastern Acknowledgements forests. Ecol. Appl., 2: 238-247. Huston, M., 1979. A general hypothesis of species diversity. Am. Nat., 113: 81-101. I thank T. Crow, J. Johnson, S. Tang, C. Morgan, Lewin,R., 1986.Supply-sideecology.Science,234: 25-27. ' Li, H., Franklin, J.F., Swanson, FJ. and Spies, T.A., 1993. S. Shirley, D. Mladenoff and two anonymous re- Developing alternative forest cutting patterns: a simulation viewers for critical reviews of earlier drafts of the approach.LandscapeEcol., 8: 63-75. manuscript. I thank L. Burde for editorial assistance. Lorimer, C.G., 1985. The role of fire in the perpetuation of oak i[_l E.I. Gustafson / Forest Ecology and Management 87 (1996) 27-39 39 ' ]- _i'L;iiii¢....J forests. In: J.E.. Johnson (Editor), Proc. of Challenges in Oak Swanson, F.J. and Franklin, J.F., 1992. New forestry principles Management and Utilization. WEX Cooperative Extension from ecosystem analysis of Pacific Northwest forests. Ecol. - Service, University of Wisconsin-Extension, Madison, WI, pp. Appl., 2: 262-274. 8-25. Swanson, F.J., Jones, J.A., Wallin, D.O. and Cissel, J.H., 1994. McCarthy, M.A. and Burgman, M.A., 1995. Coping with uncer- Natural variabilitymimplications for ecosystem management. tainty in forest wildlife planning. For. Ecol. Manage., 74: lli: M.E. Jensen and P.S. Bourgeron (Editors), Vol. II: Ecosys- 23-36. tern Management: Principles and Applications. PNW-GTR- Mladenoff, D.J. and Pastor, J., 1993. Sustainable forest ecosys- 318, USDA Forest Service, pp. 80-94. / terns in the Northern Hardwood and Conifer Forest Region: Urban, D.L., O'Neill, R.V. and Shugart, H.H., 1987. Landscape / conceptsand management. In: G.H. Aplet, N. Johnson, J.T. ecology. BioScience, 37:119-127. Olson and V.A. Sample (Editors), Defining Sustainable USDA Forest Service, 1985. Land and Resource Management ' Forestry. Island Press, Washington DC, pp. 145-180. Plan, Hoosier National Forest. USDA Forest Service, Eastern , Naiman, RJ., Decamps, H. and Pollock, M., 1993. The role of Region, Hoosier National Forest, Bedford, IN, USA. • riparian corridors in maintaining regional biodiversity. Ecol. USDA Forest Service, 1991. Plan Amendment, Land and Re- Appl., 3: 209-212. source Management Plan, Hoosier National Forest. USDA _ Reice, S.R., 1994. Nonequilibrium determinants of biological Forest Service, Eastern Region, Hoosier National Forest, Bed-community structure. Am. Sci., 82: 424-435. ford, IN, USA. Robbins, C.S., Sauer, J.R., Greenberg, R.S. and Droege, S., 1989. Wallin, D.O., Swanson, F.J. and Marks, B., 1994. Landscape Population declines in North American birds that migrate to pattern response to changes in pattern generation rules: land-use . the neotropics. Proc. Natl. 'Acad. Sci., 86: 7658-7662. legacies in forestry. Ecol. Appl., 4: 569-580. " Robinson, S.K., Thompson, F.R., Donovan, T.M., Whitehead, Wallinger, S., 1995. A commitment to the future: AF&PA's . . D.R..and Faaborg, J., 1995. Regional forest fragmentation and sustainable forestry initiative. J. For., 93(1): 16-19. the nesting success of migratory birds. Science, 267: 1987- Wiens, J.A., 1989. Spatial scaling in ecology. Funct. Ecol., 3: 1990. • 385-397. Small, M.F. and Hunter, M.L., 1988. Forest fragmentation and - ' .. avian nest prex!ation in forested landscapes. Oecologia, 76: 62-64. , r Forest Ecology and Management Submission Of manuscripts. Manuscriptsshouldbe submittedin triplicate.Authors fromthe Americas,Australia,New Zealand and the Pacificare requested t°University,SendtheircollegemanuscriptSstation,t° F°reStTx77843-2135,Ec°l°gyandUsA;Management'allother authorsDr" RichardarerequestedF"Fisher, toDepartmentsendtheirOfmanuscriptsFOrestScience,totheTexasEditorialA&M B Office, Forest Ecology and Management, P.O. Box 181, 1000 AD Amsterdam, The Netherlands. Authors in Japan please note: Upon request, Elsevier Science Japan will provide authors with a list of people who can check and improve the Englishof their paper (before submission). Please contact our Tokyo office: Elsevier Science Japan, -- , 1-9-15 Higashi-Azabu,Minato-ku,Tokyo 106, Japan; tel. (03)-5561-5032; fax (03) 5561-5045. Electronic manuscripts: Electronicmanuscripts have the advantage that there is no need for the rekeying of text, thereby avoidingthe possibilityof introducingerrorsand resultingin reliableand fast deliveryof proofs. For the initial submissionof manuscriptsfor consideration,hardcopiesare sufficient.For the processingof accepted papers, electronicversionsare preferred.After final acceptance, yourdisk plustwo, finaland exactly matchingprintedversionsshould . be submittedtogether. Double density (DD) or highdensity (HD) diskettes(3.5 or 5.25 inch) are acceptable.It is importantthat the file saved is in the native format of the wordprocessorprogram used. Label the disk with the name of the computerand wordprocessingpackage used, your name, and the name of the file on the disk. Further informationmay be obtainedfromthe •Publisher. All questions,arising after acceptance of the manuscript,especiallythose relating to proofs, should be directed to Elsevier EditorialServices, MayfieldHouse, 256 Banbury Road, Oxford OX2 7DH, UK, tel. (018.65)314900, fax (01865) 314990. Advertising information: Advertising orders and enquiries may be sent to: International= Elsevier Science, Advertising Department, The Boulevard, Langford Lane, Kidlington, Oxford, OX5 1GB, UK; tel. (+44)(0) 1865 843565; fax (+44)(0) 1865 843976. USA and Canada: Weston Media Associates, Dan Lipner, P.O. Box 1110, Greens Farms, CT 06436-1110, USA; tel. (+1) (203) 261 2500; fax (+1) (203) 261 0101. Japan: Elsevier Science Japan, Marketing Services, 1-9-15 Higashi-Azabu, Minato-ku, , Tokyo 106, Japan; tel. (+81)-3-5561-5033; fax (+81)-3-5561-5047. US mailing info: Forest Ecology and Management (0378-1127) is published monthly by Elsevier Science B.V. (Molenwerf 1, Postbus 211, 1000 AE Amsterdam). Annual subscription price in the USA is US$ 2043.00 (valid in North, Central and South America), includingairspeed delivery. Second classpostagerate ispaid at Jamaica, NY 11431. _ USA POSTMASTERS: Send address changes to Forest Ecology and Management Publications Expediting, Inc., 200 Meacham Avenue, Elmont, NY 11003. AIRFREIGHT AND MAILING in the USA by Publications Expediting Inc., 200 Meacham Avenue, Elmont, NY 11003. Forest Ecology and Management has no page charges For a full and complete Guide for Authors, please refer to Forest Ecology and Management, Vol. 80, Nos. 1-3, pp. 287-290 y. Copyright © 1996, Elsevier Science B.V. All rights reserved. 0378-1127/96/$15.00 (_ The paper usedin this publicationmeets the requirementsof ANSI/NISO Z39.48-1992 (Permanence of Paper). Printedin The Netherlands "LANDSCAPE AND URBAN PLANNING = An International Journal of Landscape Ecology,• • Landscape Planning and Landscape design , o .'0 Editor-in-Chief: J.E. Rodiek, College of Architecture, Texas A & M University, College Station, "IX77843-3137, USA . . AIMS AND SCOPE the premise that research R.¢. Szaro, Washington,OC,USA, A journal concerned with linked to practice will J.W.Thomas,LaGrande,OR,USA..... " conceptual, scientific and ultimately improve the human P.J.Trowbridge,Ithaca,NY,USA,T.H.D.Turner,London,UK, • design approaches to land made landscape. M.J.Vroom,Wageningen, -use. By emphasizing TheNetherlands, _" ecological understanding and W.V.Wendler,CollegeStation,TX, a multi-disciplinary approach Editorial Advisory Board" USA, to analysis and planning and I.D.Bishop,Parkville,Vic.,Australia, B.-E.Yang,Seoul,Korea,• E.G.Bolen,Wilmington,NC,USA, E.H.Zube,Tucson,AZ,USA design, it attempts to draw I.D.Bruns,Schomdorf,Germany, attention to the interrelated J.B.Byrom,Edinburgh,UK, • nature of problems posed by T.C.Daniel,Tucson,AZ,USA, ABSTRACTED/INDEXED IN •nature and human use of R.M.DeGraaf,Amherst,MA,USA, Applied EcologyAbstracts, land. In addition, papers J.G.Fabos,Amherst,MA,USA, BiologicalAbstracts, " dealing with ecological S.GonzalezAlonso,Madrid,Spain, Current Contents B & S, processes and interactions M.Hough,Etobicoke,ON,Canada, Environmental Periodicals within urban areas and P.Jacobs,Montreal,PQ,Canada, Bibliography, Geobase, ' D.S.Jones,Melbourne,Vic.,Australia, GeographicalAbstracts, between these areas and the H.Lavery,Milton,QId.,Australia, LandSearch. surrounding natural systems W.M.Marsh,Flint,MI,USA, which supportthem, will be D.L.Mitchell,Dallas,TX,USA, considered. Papers in which D.G.Morrison,Athens,GA,USA, specific problems are J.l.Nassauer,St.PaulMN,USA, 1994SUBSCRIPTIONDATA examined are welcome. M.Helischer,Guelph,ON,Canada, D.D.Paterson,Vancouver,BC,Canada, Volumes28-30(in 9 issues) Topics might include but are notlimited to landscape A.Ramos,Madrid,Spain, Subscriptionprice: P.Shepard,C/aremont,CA,USA, Dfl.1080.00(US$ 584.00) ecology; landscape planning O.R.Skage,A/narp;Sweden, incl.Postage and landscape design. R.C.Smardon,Syracuse,NY,USA, • " Landscape ecology examines G.Sorte,Alnarp,Sweden, iSSN0169-2046 how heterogeneous F. Stearns,Rhinelander,WI,USA; p. ' combinations of ecosystems , are structu red, how they ElsevierSciencePublishers, function and how they - P.O.Box211,1000AEAmsterdam, • change. Landscape planning _"_". TheNetherlands examines the various ways ___ Fax:(020)5803-203 humans structure their land __L CustomersintheUSAandCanada:use changes Landscape ElsevierSciencePublishers "• P.O.Box945,MadisonSquareStation, design involves the physical Strategies and forms by which g LSEVI g R NewYork,NY10160-0757,USAFax:(212)633-3680 land use.change is actually _,_Out_,Guilderpricequotedappliesworldwide,exceptin_e _/Americas (Nor_, Cen_al and South America}. The US Dollar directed Landscape and SCIENCE PUBLISHERS pricequotedappliesin_eAmericasonly.NonVATregistered • customers in the European Community should add the Urban Planning is based on appropriate VAT rate applicable in their ¢ount_ tothewice. .