Evaluating the Effects of Varying Levels and Patterns of Green-Tree Retention: Experimental Design of the DEMO Study

Home , Gifford Pinchot

Keith B. Aubry, USDA Forest Service, Pacific Northwest Research Station, Olympia, Washington 98512 Michael P. Amaranthus, USDA Forest Service, Pacific Northwest Research Station, Grants Pass, Oregon 97526 Charles B. Haipern, College of Forest Resources, University of Washington, Seattle, Washington 98195 James D. White, USDA Forest Service, Pacific Northwest Region, Gifford Pinchot National Forest, Trout Lake, Washington 98650 Brenda L. Woodard, USDA Forest Service, Pacific Northwest Region, Umpqua National Forest, Roseburg, Oregon 97470 Charles E. Peterson, USDA Forest Service, Pacific Northwest Research Station, Portland, Oregon 97208 Cynthia A. Lagoudakisi, USDA Forest Service, Pacific Northwest Region. Umpqua National Forest, Roseburg, Oregon 97470 Alan J. Horton, USDA Forest Service, Pacific Northwest Region, Regional Off~ce,Portland, Oregon 97208

Evaluating the Effects of Varying Levels and Patterns of Green-tree Retention: Experimental Design of the DEMO Study

Abstract In western Oregon and Washington, recent changes in federal forest management policy contained in the Northwest Forest Plan have led to new harvest prescriptions on millions of acres of public lands. For example. on upland sites, standards and guidelines now require that live (green) trees are retained in at least 15% of the area within each harvest unit and recommend that at least 70% of this retention is in patches of moderate to larger size (0.2- 1.0 ha or more). These prescriptions [-'or green-tree retention were based on the professional judgement and collective knowledge of many of the biologists who have studied the organisms and ecological processes that characterize these forests, but they have not been rigorously tested nor implemented on a broad geographic scale. Several prescriptions for green-tree retention are being evaluated experimentally by the Demonstration of Ecosys- tem Management Options (DEMO) study. In this paper, we briefly review recent changes in forest management policy and existing information gaps that led to the establishment of the DEMO study. We then provide an overview of the criteria for site selection, the experimental design and harvest prescriptions, the scope of scientific inquiry, and the collaboration that has occurred between scientists and land managers. These discussio~isprovide the context and experimental framework for the individual research papers that comprise the remainder of this volume.

Introduction Bureau of Land Management (Tuchmann 1995). In April 1993. in response to this 'timber crisis', In the 1980s, public concern over the fate of the President Clinton commissioned the Forest Eco- northern spotted owl (Strix occidentalis cauri~za) system Management Assessment Team (FEMAT and of the old-growth Douglas-fir (Pseudotsuga 1993, see also articles in Vol. 92(4) of the Jour- i17erzziesii) forests on which it depends (Gutikrrez nal of Forestry) to formulate and assess the con- and Carey 1985, Thomas et al. 1990) led to un- sequences of an array of options for managing precedented local and regional interest in natural federal forest lands within the range of the north- resource management in the Pacific Northwest. ern spotted owl. This assessment was the first This interest, and the political activism it gener- attempt to develop a comprehensive plan for eco- ated, elevated these issues to national prominence system management on a broad geographic scale. (Ervin 1989, Norse 1990). Lawsuits initiated by The resulting management plan came to be known environmental groups resulted in injunctions from as the Northwest Forest Plan (Tuchmann et al. three separate federal district courts and brought 1996). The Record of Decision (ROD) and asso- timber sale programs to a standstill on federal lands ciated standards and guidelines (S & Gs) for the managed by the USDA Forest Service and USDI Plan were published in April 1994 (USDA and USDI 1994a) and lawsuits against the Plan were 'Current address: USDA Forest Service. Alaska ~kgion, resolved in favor of the government in December Tongass National Forest. Juneau, Alaska 99801 1994.

12 Northwest Science, Vol. 73, Special Issue. ! 999 Among the changes in forest management ten structurally and biologically simpler and the policy brought about by the Northwest Forest Plan development of late-successional characteristics is a greater emphasis on maintaining the diver- in these stands may require more intensive silvi- sity of species, structures, and physical and eco- cultural manipulations. Large-scale harvest studies logical processes that are important to the func- in mature, late-successional forests are lacking. tioning of mature, late-successional forest In the Northwest Forest Plan, these forests are ecosystems. These include microclimates asso- defined as single- and multi-storied stands domi- ciated with interior-forest conditions, spatial het- nated by conifers > 53 cm dbh (USDA and USDI erogeneity in forest composition and structure, 1994b; pp. 3&4-26). They comprise 682,300 ha fungi, invertebrates, saprophytic plants, nongame of Matrix lands and Adaptive Management Ar- wildlife, snags, down woody debris, and other eas (USDA and USDJ 1994b; Table 3&4-8) and, structural and compositional features generally thus, are available for timber harvest under the associated with old-growth forests (Franklin et Northwest Forest Plan. Furthermore, old-growth al. 1981). forests are dynamic and it will not be possible to This is not the first time that forest manage- provide cathedral groves of ancient forests for ment objectives have shifted in response to chang- future generations simply by preserving existing ing public perceptions or professional judgements. old-growth stands (DeBell and Curtis 1993). From the 1930s until the mid- to late- 1940s, se- Today's mature forests will become the old-growth lective cutting of the largest and most vigorous forests of the future; thus, if societal objectives trees was widely applied in old-growth Douglas- are to maintain wood production while increas- fir stands in the Pacific Northwest. Long-term ing the extent of forest stands with old-growth studies revealed, however, that subsequent characteristics, we must. begin to develop meth- windthrow mortality was significant and offset ods for managing mature forests for these objec- increases in residual stand volume for at least 25 tives. yr after harvest (Isaac 1956). Clearcu tting replaced In the early 1990s, several private research selective logging in most areas, and from the 1950s institutions and public interest groups in Wash- through the late 1980s, regeneration harvests con- ington and Oregon sought federal funds for re- sisted primarily of clearcuts of various sizes stag- search aimed at addressing these information needs. gered in time and space, followed by slash burn- In response to these lobbying efforts, the Pacific ing and planting. This strategy was designed to Northwest Region of the USDA Forest Service meet legal mandates for multiple-use management received Congressional direction in 1993 to es- on public lands while providing high volumes of tablish a large-scale silvicultural experiment (see timber at relatively low cost, dispersion of cu- Franklin et al. 1999). The Region, in coopera- mulative watershed effects, access for recre- tion with the Pacific Northwest Research Station, ation, and habitat for deer and other game spe- subsequently established the Demonstration of cies associated with early successional forests. Ecosystem Management Options (DEMO) study, Several studies are proposed or underway in a collaborative research effort with the Washing- the Pacific Northwest region to evaluate the ef- ton State Department of Natural Resources, Uni- fects of silvicultural prescriptions designed to versity of Washington, Oregon State University, accelerate the development of late-successional and University of Oregon to evaluate the ecological, forest characteristics in young, managed stands physical, and social effects of retaining live (green) that have regenerated after clearcutting (e.g., trees in harvest units in western Oregon and Wash- Tappeiner 1992, Hanington and Carey 1997, Carey ington. The DEMO study is designed to provide et al. in press). Such experiments are both rel- information to aid in the development of harvest evant and necessary. However, because of sub- strategies that will retain or accelerate the recov- stantial differences in initial conditions-especially ery of some of the species and key ecological fea- the vegetative and structural legacies typical of tures found in mature and old-growth forests. mature forests-responses to retention harvest in In this paper, we briefly review recent changes young, managed forests may be very difYerent from in forest management policy and existing infor- those in mature forests. Mature forests generally mation gaps that led to the establishment of the have a broader array of structures, species, and DEMO study. We then provide an overview of functions, whereas managed plantations are of- the criteria for site selection, the experimental

DEMO: A Study of Green-tree Retention 13 design and harvest prescriptions, the scope of logical processes in these forests. However, the scientific inquiry, and the collaboration that has ecological and silvicultural effects of retaining occurred between scientists and land managers. various amounts or configurations of green trees These discussions provide the context and experi- in harvest units have not been quantified (DeBell mental framework for the individual research and Curtis 1993). Aggregated and dispersed re- papers that comprise the remainder of this vol- tention strategies are each hypothesized to have ume. Facilitating the integration of research find- varying ecological con5equences and silvicultural ings among these disciplines was a guiding prin- applications. Dispersed retention is believed to ciple during development of many aspects of the be most appropriate where ecological objectives DEMO study. require that certain target structures or conditions be uniformly distributed, such as provision of down Management Needs woody debris and snags or- mitigating microcli- matological or hydrological impacts. Aggregated Adaptive management, a process by which in- retention, on the other hand, is expected to be formation gained from monitoring the conse- more effective in maintaining a broader array of quences of management actions is used to refine structural elements and ecological conditions. For and improve future management strategies example, aggregates enable the maintenance of (Bormann et al. 1994), is expected to play an all canopy layers including understory vegetation, important role in shaping future forest manage- as well as snags of various sizes and decay classes, ment in the Pacific Northwest. To ensure that in ways that are not possible with dispersed re- management prescriptions provide desired ben- tention. Intact patches of habitat also serve as efits, however, they should also be evaluated ex- refugia for various organisms, which may pro- perimentally using a rigorous and replicated study vide source populations for recolonization of design. Information gained through such experi- nearby regenerating areas (Franklin et al. 1997). mentation, which enables researchers to apply inferential statistics to data collected and to mini- Because empirical information on the ecological mize the confounding effects of extraneous sources effects of retaining green trees in harvest units in of variation, will provide reliable and broadly mature stands is extremely limited, the short- and applicable results. Several large-scale silvicul- long-term consequences for biodiversity of imple- tural experiments involving various levels of green- menting harvest prescriptions in the Northwest tree retention are underway in British Columbia Forest Plan are largely unknown. Green-tree re- in forests dominated by Pacific silver fir (Abies tention in mature forests will produce varying anzabilis) and western hemlock (Tsuga Izetero- responses depending upon the amount and con- pkjdla) (Arnott and Beese 1997, Coates et al. 1997, figuration of retained trees. For example, sub- see also Franklin et al. 1999). However, there is stantially reducing canopy coverage and dispers- little empirical information available to forest ing retained trees is expected to accelerate the managers in the Douglas-fir region on the eco- diameter growth of retained trees, but may have logical responses or social and economic tradeoffs adverse consequences for understory species that associated with various levels and patterns of green- have an affinity for late-sera1 or interior forest tree retention (Franklin et al. 1997). conditions (Halpern et al. 1999). Green-tree re- Standards and guidelines in the Northwest tention in mature stands will influence understory Forest Plan specify retaining green trees in at least composition and structure; shape future levels of 15% of the area within each cutting unit and rec- standing and down coarse woody debris; and af- ommend that 70% of this retention is in aggre- fect the diversity, abundance, and distribution of gates of moderate to larger size (0.2- 1 .0 ha or more), many sensitive plant and animal species. Thus, with the remainder dispersed either as single trees information on the ecological consequences of or in small clumps < 0.2 ha. In addition, to the green-tree retention in mature forests is needed extent possible, snags and large decadent trees in if we are to effectively manage both young plan- harvest units should be preserved in green-tree tations and mature forests. retention patches (USDA and USDI 1994a). These The DEMO study is expected to provide broadly prescriptions reflect the professional judgements applicable information on the ecological and sil- and collective knowledge of many of the biolo- vicultural effects of green-tree retention in ma- gists who have studied the organisms and eco- ture Douglas-fir forests. Results are expected to

14 Aubry et a]. be applicable over a large proportion of public sented in Table 1. Treatments were randomly and private lands in western Oregon and Wash- assigned among six units within each block; the ington. The study focuses on regeneration har- six treatments (Figure 1) are: vests of mature forests dominated by medium- (I) 100% reterztion-A control unit that pro- and large-sized conifers. In addition to the more vides a baseline for assessing the efYects of har- than 680,000 ha of forest in these size classes in vest treatments on ecosystem composition, struc- unreserved federal lands in the Pacific Northwest, ture, and function. DEMO results will be applicable to thousands of (2) 75% aggregated reteiztion-Three circu- hectares of younger forest as they mature over lar 1-ha patches are harvested in a triangular ar- the next several decades. ray, removing 25% of the stand area. The pattern Research Questions and distances between the gaps are consistent among blocks. The DEMO study is designed to address the fol- (3) 40% disper-sed retentioiz-Dominant and lowing general questions: co-dominant trees are retained in an even distri- (I) How does the proportion of green trees bution throughout the treatment unit. The total retained at a site influence various biological, basal area retained in this treatment varies among physical, and social values? blocks, and is determined by the combined basal (2) Given a particular level of green-tree re- area (-1 5%) in the five I -ha patches of the corre- tention, is the spatial pattern of that retention sponding 40% aggregated retention treatment. important? In particular, are the effects of re- (4) 40% aggregated reterztion-Five undis- taining trees in undisturbed patches different from turbed 1-ha circular patches are retained in the those that result from distributing them evenly treatment unit; distances among patches are con- throughout the harvested area? sistent among blocks (see Figure 2). (3) If organisms, processes, and values are af- (5) 15% dispersed retentioiz-Dominant and fected by the level andor pattern of green-tree co-dominant trees are retained in an even distri- retention in the short term, are the effects main- bution throughout the treatment unit. The total tained over time as stands develop? basal area retained in this treatment varies among (4) Do the effects of different levels or pat- blocks, and is determined by the combined basal terns of green-tree retention differ among geo- area (-1 5%) in the two l-ha patches of the corre- graphic areas or forest communities? sponding 15% aggregated retention treatment. (6) 15% aggregated retentiorz-Two undis- Experimental Design and Site Selection turbed 1-ha circular patches are retained in the The DEMO study consists of a randomized block treatment unit. Each patch is located in a diago- design involving six treatments representing vary- nal half of the treatment area; the distance being levels and patterns of green-tree retention tween patches is consistent among blocks. (Figure 1). Each treatment unit is a relatively Responses of forest organisms to these treat- homogeneous forest stand of about 13 ha, and is ments will vary depending on the life history char- square or slightly rectangular (Figure 2). Treat- acteristics and spatial requirements of the taxa ment units consist of upland forest habitats; large under consideration, as well as the time since treat- streams or wetlands were avoided. ments were implemented (see Halpern et al. 1999, The six treatments are replicated in eight geo- Lehmkuhl et al. 1999, Cazares et al. 1999, and graphic locations (blocks), four in southwestern Progar et al. 1999 for detailed discussions of pre- Washington and four in southwestern Oregon dicted responses). In general, however, we pre- (Figure 3). Although selected blocks may con- dict that the diversity and abundance of species tain a variety of forest communities, Douglas-fir associated with unmanaged mature and old-growth is the dominant tree species in all blocks (see Table forests (see Ruggiero et al. 1991) will decline with I in Halpern et al. 1999); forest ages range from harvest intensity and will be lower in dispersed about 70 to 200 yr or more (Table 1). Detailed than in aggregated treatments. We also predict descriptions of the physiographic environments that species closely associated with late-succes- and management histories for each block are pre- sional forests will be unable to persist in the 15%

DEMO: A St~~dvof Green-tree Retention 15 15% aggregated retention 15% dispersed retention

40% aggregated retention 40% dispersed retention

75% retention 100% retention

Figure 1. Schematic representation of the six treatments in the DEMO study (figure provided by Ken Bible). retention treatments, although species with very During the initial planning stages of the ex- snlall spatial requirements may survive in sefu- periment, we considered comparing additional gia created by the aggregates (see discussion of levels of retention or varying the sizes of aggre- 'lifeboating' in Franklin et al. 1997). gates, but practical constraints precluded these

16 Aubry et al. expected to be weak. Lehtnkuhl et al. (1991) evaluated the influence of surrounding stands and landscapes on bird, small mammal, and amphib- ian abundances in unmanaged Douglas-fir stands in southern Washington and found that landscape indices alone were not useful for predicting ver- tebrate species richness or abundance. For the organisms they studied, physiographic position (especially elevation) and stand structure were ~iiuchbetter predictors of abundance patterns than landscape indices. For other wildlife species that may be more strongly influenced by landscape context (e.g., flying squirrels, bats, woodpeck- ers) the influence of adjacent stands will be considered as covariates in post-hoc analyses of treatment responses. iz~gure2. Aerlal vlew of the 40% aggregated retention treat- rnent at the Parad~wHills block in Washington Methods shortly after harve~t(photo courtesy of Jon Nakae). options (see Franklin et al. 1999). Given the lim- Treatment Prescriptions ited availability of suitable areas large enough to The experimental prescrip.tions in DEMO were effectively study most species of wildlife (espe- extremely difficult to apply uniformly among all cially arboreal rodents) and the funds available blocks. This was due partly to differences in geo- for this research, evaluating other levels of reten- graphic location, physical environment, and for- tion or sizes and shapes of aggregates was not est composition and structure, but also to the prac- possible. Site suitability was constrained by many tical constraints of conducting harvest experiments factors, including variation in vegetative struc- of this magnitude on public lands (see Abbott et ture and composition, roads, management history al. 1999). We have addressed these issues in sev- and allocations, and the presence of streams or eral ways. All prescriptions were applied con- wetlands. In many of the blocks, implementa- sistently among the six treatment units within each tion of the treatments required variances from block. Harvest method, snag and log retention, current management requirements, such as those slash treatment, and planting were similar in all for riparian buffers in the Northwest Forest Plan treatment units within each block to ensure that (USDA and USDI 1994a). Justifications for these these activities introduced as little extraneous variances were reviewed by the interagency Re- within-block variation as possible. Maintaining gional Ecosysten~Office (REO) and evaluated on complete uniformity in treatments among the eight a case-by-case basis. The RE0 concluded "that blocks has proven impractical, but concerted ef- the review did not identify any unacceptable risks forts have been made to reduce variability in the to the objectives of the standards and guidelines implementation of treatments. that would require modification or cancellation of the project"'. The Systematic Sampling Grid It was not possible to select study areas in which In this study, we use either an 8 x 8 or 7 x 9 sam- the surrounding forests were of uniform stand age, pling grid (to provide flexibility in treatment unit structure, and management histoty. Thus, we were placement) with a 40-m spacing between sample not able to control the past or future management points. The edge of the treatment area extends at of surrounding stands; some treatment areas oc- least 40 m beyond the outer grid points (Figure cur within a uniform matrix, while others have a 4). The minimum size (13 ha) of the treatment more diverse landscape context. Consequently, units is based on the area necessary to generate landscape-scale ititluences of surrounding stands reliable abundance indices for northern flying on the treatment units may vary. For many re- squirrels (Gkaucom?ls sabrinus). Carey et al. (1 99 1) sponse variables, however, such influences are recommended 10 x 10 grids with 40-m spacing

DEMO: A Sttlrtv of Green-tree Retention 17 MOUNTAINS

PACIFIC OCEAN

A MOUNT JEFFERSOA

MOUNTAINS

, Figure 3. Location of'the eight experimental blocks of the DEMO study in Oregon and Washing- ton. Block codes from north to south are: CF = Capitol Forest, B = Butte, PH = Paradise Hills, WS = Little White S;~lnion,LC = Layng Creek. WF = Watson Falls, LR = Little River, and DP = Dog Prairie. ABLE I. Environmental features and stand characteristics of the eight experimental blocks in the DEMO study. Ranges arc based on mean values for each ofthe six treatment units in each block.

Slope Stand age ~~cation/Block Elevation (m) (degrees) Aspect (yr) Management history Riparian zones'

Oregon: Umpqua National Forest Watson Falls 945- 13 10 4-7 flat 1 10-130 salvage logged class IV stream (I unit) ( 1970- 1978)

1,ittle River 1220- 1400 1 4-40 varied 200-520' salvage logged class II strealns (2 units): (2 iinits) class Ill stt-earn (I unit): class IV streams (2 units)

Layng Creek 490-790 20-4 1 SE 60-80 second-growth; class Ill streams thinned ( 1984- 1989) (6 units)

Dog Prairie 1460- 1 7 1 0 34-62 SW 165 thinned ( 1986) class IV stream (I unit) IVashington: Gifford Pinchot National Forest 975- 1280 40-53 E-SE 70-80 none class Ill streams (2 units); class IV streams (4 units) Little White Salmon 825-975 40-66 NW-NE 140- 170 none class 111 stream . (I unit); class IV streams (5 units) Paradise Hills 850-1035 9-33 varied 110-140 none class 111 streams (3 units); class IV streams (4 units) ,Washington: Department of Natural Resources Capitol Forest 2 10-275 28-52 varied 65 second-growth class 111 streams (2 units); class IV streams (4 units)

'Class I1 stream = fish-bearing perennial stream; class 111 stream = non-fish bearing perennial stream; class IV stream = non-fish hearing intermittent stream. 'One treatment unlt with tree ages of 400-520 yr; remaining units with tree ages of 200-325 yr.

between sample points (16-ha sampling area) for to strongly influence flying squirrel populations estimating densities of flying squirrels, but indi- (see Lehmkuhl et al. 1999). Addressing the needs cated that smaller grids (e.g., 7 x 9; 10 ha) were of spotted owls was a primary objective during adequate, particularly if the goal was to calculate the development of the Northwest Forest Plan relative abundance values rather than actual den- (USDA and USDI 1994b; pp. 2-72), thus, the re- sities (see also Carey et al. 1996). sponse of flying squirrel populations to the DEMO Study grids were designed so that flying squirrel treatments will have important impIications for populations could be included as response vari- assessing the efficacy of green-tree retention strat- ables for several reasons: they are forest obli- egies for maintaining or enhancing spotted owl gates associated primarily with mature and old- foraging habitat in harvest units. growth conifer forests, use both canopy and The permanent grid system (or a subsample forest-floor habitats, have relatively large home thereof) will also be used to sample the responses ranges (about 0.8 ha), and are important prey for of various ecosystem components such as veg- the northern spotted owl in the DEMO study area etation, other wildlife, and ectomycorrhizal fungi. (Carey 199 1 ). Thus, the DEMO treatments, which The grid facilitates integration among these studies 13rimarily involve manipulating the density and and ensures that the same environments will be configuration of the forest canopy, are expected sampled within each stand for the duration of the

DEMO: A Study of Green-tree Retention 19 Minimum size of treatment area is 13 ha

, .: '..', .'i. .. . , .. . . x~:~izii~.~!zE!3~~z~rz:~:~:~. , spacing

40-m spacing to edge of unit

Figure 4. Schematic representation of the DEMO experitnental grid as configured for an 8 x 8 square grid in a 40% aggregated retention treatment unit. project. In the three aggregated retention treat- felling trees into retention patches. Four blocks ments (75,40, and 15%), in which two contrast- were helicopter logged, two shoveVloader logged, ing environments will be created (intact forest and one skyline cable logged, and one block (Capitol clearcuts), the grid will also permit analyses of Forest, Washington) was logged with a combina- within-treatment patterns of variability and the tion of systems due to unusual topographic con- contributions of each of these environments to ditions in one treatment unit. treatment-level responses. Snags Timber Harvest During harvest, existing snags that did not pose a Methods for harvesting timber in each block were safety hazard were retained. To provide snags determined by local forest managers. We recog- for future stands and to meet minimum snag re- nized that appropriate timber harvest methods quirements specified in the Northwest Forest Plan, would vary in different forest and topographic contractors left an additional 6.5 dominant or co- conditions; thus, it was not practical to use the dominant green trees per ha in all harvested ar- same harvest method in all blocks. Several con- eas within each treatment unit for snag creation. straints were imposed, however: the same ~uethod These 'snag trees' were retained in addition to was used in all treatment units within each block, the green trees that contribute to the 15, 40, or treatment units within a block were harvested in 75% retention levels in treatments 2 through 6. the same year and logging contractors avoided When possible, trees selected for snag creation

20 Aubry et a1 I were those that were already decayed or had bro- mum stocking densities of 3 12 trees per ha after I ken tops. These trees will be killed within one 5 years. Evaluation of adequate stocking will be growing season of harvest by removing the crown based on planted trees, advance regeneration, and I with a chain saw. subsequent natural regeneration. Although more than 3 12 trees per ha may be planted in some Down Woody Material blocks to ensure adequate survival, planting den- Down woody material on site was retained dur- sity will be set as low as possible to encourage ing harvest, but no prescriptions for creating large natural regeneration, minimize the need for site down woody material were implemented. Green preparation, and reduce potential confounding trees that inadvertently fell into unharvested ar- effects on understory vegetation. As with all other eas of the 75% retention treatment or unharvested prescriptions, planting densities and species com- patches of the 15% or 40% aggregated retention position will be consistent among treatment units treatments were removed. When necessary, such within each block and will be determined jointly trees were limbed prior to removal to minimize by forest managers and researchers. disturbance to reserve areas; resulting branches were also removed from the leave areas. Other Post-harvest Treatments The use of fertilizers, seeding for erosion con- SlashIFuels Treatment trol, or herbicides is prohibited on DEMO study After harvest, forest managers and researchers sites. Manual release of tree seedlings must be evaluated slash loadings to assess potential fire approved by both forest managers and DEMO hazards and effects on regeneration and accurate researchers. sampling of response variables. If further treat- Forest managers have agreed to protect the ment was necessary, it was accomplished with as DEMO study sites from activities that are not part little disturbance to the site as possible, using of the planned prescriptions for 10 years after methods determined jointly by forest managers harvest. At that time, managers and researchers and researchers. Although various methods for are expected to re-evaluate the status of experi- fuel-reduction (e.g., yarding tops or hand-piling mental sites and research progress, and assess the and burning) were used among blocks, slash treatments were applied consistently among treatment need for new study plans, contractual agreements, units within each block. Sites having low to mod- and additional silvicultural treatments. erate amounts of slash were left untreated to provide for soil stability and future nutrient inputs. Major Areas of Investigation Given limited time, funding, and human resources, Retention of Submerchantable Trees areas of investigation have been restricted to those To the extent possible, trees that were too small in of highest priority and with the greatest potential diameter to be merchantable, including advance for integrative analyses. These include vegeta- regeneration (seedlings established naturally under tion, wildlife, fungi, invertebrates, hydrology, social the canopy), were protected in all treatment units. perceptions, and harvest costs. Riparian Protection Vegetation The number of potential study areas that met the Vegetation studies will examine post-harvest size requirements of the study was extremely lim- changes in the composition, structure, and diver- ited; consequently, it was not possible to avoid sity of overstory and understory communities both small streams or seeps in all treatment units (Table among and within experimental treatments. Vari- I ). However, to avoid confounding the effects of ables of interest include the growth, damage, and the primary treatments, undisturbed riparian buffers mortality (including windthrow) of residual trees; were not maintained in any of the treatment units. recruitment, growth, and mortality of regeneration (natural and planted); changes in species Reforestation composition of ground-layer bryophytes and li- The Resource Management Plans of the Umpqua chens, herbs, and shrubs; and the dynamics of and Gifford Pinchot National Forests require mini- snags and down woody debris. Because understory

DEMO: A Study of Green-tree Retention 21 I and overstory measurements are spatially linked, tebrates are important food items for many ver- I understory patterns can be interpreted relative to tebrate species and are influential in forest eco- ' initial treatment effects as well as to changes in systems as both pests and biological control agents. overstory characteristics. Changes in many of In addition, they affect plant species composition, these variables will be used to interpret the re- leaf area, canopy-atmosphere interactions, and sponses of organisms and processes under inves- nutrient fluxes to the forest floor. Hence, responses , tigation in companion studies (see Halpern et al. of canopy invertebrate cornrnunities to forest , 1999 for additional discussion). management practices are particularly important to ecosystem function and forest health (see Progar 1 Wildlife et al. 1999 for additional discussion). Wildlife studies will evaluate patterns of habitat Hydrology use and provide estimates of relative abundance for diurnal forest birds, arboreal rodents, bats, Hydrologic ilivestigations will focus on snow forest-floor small mammals, and amphibians. hydrology, especially rain-on-snow events. Be- Wildlife research will focus primarily on wild- cause these winter studies are logistically diffi- life species with home ranges s~nallerthan the cult to implement simultaneously at multiple lo- 13-ha treatment areas but, whenever possible, cations, snow accumulation and melt will be information will be gathered on habitat use by measured at only one experimental block (Watson wider ranging mammals and birds. Because ani- Falls, Oregon). Supporting meteorological data mal populations vary annually, 2 yr of pre-treat- will be collected so that the physical processes ment data have been collected; additional years affecting snowmelt and accumulation can be un- of sampling would have been desirable, but were derstood for different harvest levels and patterns. precluded by limitations of both time and fund- Rain-on-snow events are among the most impor- ing. Pre-treatment data revealed that flying squirrel tant factors in cumulative watershed impacts, such populations were too low in several of the blocks as flooding, yet little is known about the influ- to enable meaningful comparisons with post-treat- ence of green-tree retention on this process (see ment data. Consequently, flying squirsel popu- Storck et al. 1999 for additional discussion). lations are only being studied in four of the experimental blocks (two in each state); other wildlife Social Perceptions sampling is being conducted in all eight blocks Social perceptions of varying approaches to green- (see Lehmkuhl et al. 1999 for additional discus- tree retention harvest will initially be measured ' sion). as judgments of acceptability and scenic beauty Fungi based on a mail survey of within-stand photo- graphs in several of the harvested blocks. This

Fungi research will investigate the abundance, research will include an evaluation of the influ- , community structure, dietary importance for wild- ence of demographic characteristics and attitudes life species, and dynamics of ectomyco~rhizalfungi toward (1) public forest management practices, ' in four of the eight experimental blocks (two in (2) the forest products industry, and (3) the po- each state). Fungi play a key role in wildlife food tential commodity and non-commodity benefits webs and influence forest recovery and produc- of the treatments on such perceptions. Once all tivity after disturbance. In addition, the economic treatments are implemented, a more intensive , importance of edible mushrooms has greatly in- survey involving focus groups will be conducted I creased in the Pacific Northwest in recent years; to evaluate public perceptions of the DEMO green- I I we know little, however, about the effects of green- tree retention treatments based on both within- tree retention harvest on these organisms (see stand and vista views (see Ribe 1999 for addi- , Cazares et al. 1999 for additional discussion). tional discussion). . I

I l nvertebrates Harvest Costs I Invertebrate studies will address treatment effects Initially, economic studies were proposed as a , on canopy invertebrate diversity, comnlunity struc- major component of the DEMO study (Anony- ture, and population abundances. Canopy inver- mous 1996) but attempts to track logging costs I

22 Aubry et al. among treatments were confounded by unantici- at 5-, lo-, or 20-year intervals, when changes in pated problems in cost accounting. Modifications canopy structure dominate stand-level responses. to the study design to overcome these proble~ns will be necessary to fully realize the potential were not possible given practical limitations and benefits of this study and to provide forest man- funding constraints; consequently, this component agers with comprehensive information on man- of the study was dropped. However, data col- agement options for maint,aining late-successional lected during helicopter logging on the Butte block conditions in mature forests managed for timber in Washington will be used to develop a helicop- production. tcr production and cost simulation model that will Data management is centralized at Oregon State he of immediate practical value to forest manag- University's Quanti tatjve Sciences Group, For- ers in the region. est Science Data Bank in Corvallis, Oregon. Comprehensive data entry, verification, and man- Additional Studies agement protocols have been developed, as well as programs to manipulate and reduce the data The DEMO harvest treatments and baseline data for subsequent statistical analyses. All data are f'rom each site will provide many short- and long- non-proprietary, but agreements will be developed rerm opportunities for additional studies that can among participants to govern publication activi- he accomplished at significantly lower costs than ties. An information management plan has been if they were conducted independently. For ex- developed that specifies data management poli- ample, the composition and diversity of arboreal cies, and discusses data input, access, sharing, and lichen communities have already been studied in security (Anonymous 1996). several of the Washington blocks as part of a master's thesis (Pipp 1998) and pre-treatment No previous studies have been conducted in vegetation data provided the ecological context Douglas-fir forests that involve the experimental for that study. The DEMO sites and infrastruc- treatments, response variables, and geographic ture are available to other researchers and we scope encompassed by the DEMO study. In ad- welcome additional research partners with extra- dition to standard statistical methods, some in- mural funding to establish new, non-destructive novative biometrical techniques will be applied. studies on the DEMO experimental sites to build Natural variation in forest ecosystems is high and upon the baseline research being conducted. Of occurs at all spatial scales; consequently, we will interest would be studies of treatments effects on partition the variation in ways that separate treat- microclimate and hydrology, nutrient cycling and ment responses from those associated with envi- decomposition, plant physiology, conifer seed ronmental variation and other sources of experi- production, and canopy epiphyte and soil inver- mental error. Initial analyses will involve the tebrate populations, among others. application of standard ANOVA (analysis of vari- ance) tests in accordance with the basic experimental design. Because of the scope and com- Data Collection, Management, and plexity of this study, however, data analyses will Analysis not be limited to standard statistical approaches. An interdisciplinary team of scientists from Or- Three of the treatments (15% and 40% aggre- egon State University, University of Oregon, gated retention and 75% retention) will generate University of Washington, and the Pacific North- two distinct environments within the treatment west Research Station are conducting pre- and area-undisturbed forest and clearcut. Response post-treatment sampling using standard sampling variables for which measurements in the undis- and data collection protocols (see other papers in turbed forest are not independent of those taken this volume). Collection of pre-treatment data in the clearcuts, such as bird or small mammal was completed in 1997 and harvest treatments abundance values, will be analyzed using treat- will be completed in 1999; post-harvest data col- ment-level means. However, for other organisms lection on several blocks began in spring 1998. with more discrete spatial requirements, these treat- Although many of the ecological effects will take ments will create two distinct responses. Thus, decades to express themselves, initial post-treat- important within-treatment variation may be ob- ment assessments will be completed within 2-3 scured by only analyzing treatment-level means; yr after harvest. Future assessments, conducted for these variables, analyses will also include finer

DEMO: A Study of Green-tree Retention 23 scale considerations of within-treatment responses lective appreciation of the respective roles and (e.g., Halpern et al. 1999). contributions of these two groups in the manage- In addition, there is substantial natural varia- ment of forested ecosystems. These interactions tion both within and among blocks that will in- have also presented some challenges, in large part fluence responses to the treatments; that is, the because of legal constraints and differing objec- natural variation present will contribute in spe- tives and approaches among scientists and 111an- cific ways to the effects of the treatments. Sources agers (see Abbott et al. 1999, Franklin et al. 1999). of variation may include differences in site envi- Managers seek to minimize or mitigate potential ronment, disturbance history, species coniposi- negative impacts, whereas researchers are con- tion, or stand structure (e.g., basal area, density, cerned with evaluating those impacts in a sys- and size structure). Furthermore, in treatments tematic, unbiased way. with aggregated retention, the patches contain an During the design and implementation of the intact canopy and have not been disturbed by log- DEMO study, numerous con~promiseswere nec- ging equipment; cut areas in both aggregated and essary to resolve conflicting objectives (see dispersed retention treatments, however, have had Franklin et al. 1999). Research objectives were their canopies altered and have also been impacted modified to reduce the number and area of har- in various ways by the process of removing the vest units, and potentially adverse environmen- trees. Thus initially, the effects of canopy reten- tal impacts (e.g., those that violate NEPA restric- tion will be difficult to separate from the varia- tions) were minimized by avoiding sites that were tion in harvest-related disturbance among treat- particularly sensitive to disturbance. Neverthe- ments. However, fine-scale sampling of the types less, harvest units are larger and silvicultural pre- and levels of disturbance (see Halpern et al. 1999) scriptions more rigid and restrictive than manag- will permit us to evaluate the relative effects of ers would otherwise implement. In addition, the disturbance and retention. Separating these po- research requirement for random assignment of tentially confounding influences may also require treatments contrasts with approaches that would the application of more complex analytical ap- optimize current land management objectives. In proaches, such as longitudinal data analysis most cases, however, managers have been will- (Anonymous 1996). ing to implement rigid treatments and involve the public in ways that differ from current practices Integration and Cooperation (see Abbott et al. 1999). The primary vehicles for achieving integration Scientists and managers will benefit equally of research activities are the common use of the from direct involvement in this study. Technical systematic sampling grid, consistent and uniform information will be transferred directly to man- implementation of treatments, a centralized and agers, resource specialists, and the public; research- coordinated data management system, and des- ers will gain a new understanding and apprecia- ignated scientific and management coordinators tion of the management process; and fundamental working in close collaboration. Many opportu- assumptions used to develop green-tree retention nities exist for integration across scientific disci- standards and guidelines in the Northwest Forest plines including relationships among vegetation, Plan will be tested. The constraints under which wildlife, fungi, and animal diets; vegetation, birds, resource managers must work are little known to bats, and invertebrates; down woody material, most researchers, and the experiences gained from snags, and wildlife; hydrology and forest struc- this study will be useful in designing future stud- ture; and visual quality and vegetation. ies that contribute to the information needs of those The DEMO study represents a large-scale at- who manage public forests. This sharing of knowl- tempt to conduct comprehensive studies that have edge and experience is central to the DEMO study, direct relevance to ecosystem management. Both and will become more critical as we implement planning and implementation have required ex- the concepts of adaptive ecosystem management plicit, ongoing interaction among researchers and and develop closer working relationships between managers and, for many, it represents the first scientists and land managers. The DEMO study opportunity for such collaboration (see Abbott et sites will serve as evolving demonstration areas al. 1999). This process has broadened our col- and outdoor laboratories for resource managers,

24 Aubry et al. scientists, educators, and the public for decades Bigley, Jeff DeBell, Jeff DeGraan, Jim Hotvedt, to come. Michael Perez-Gibson, John Shumway, and Lenny Young of the Washington State Department of Acknowledgements Natural Resources. A special thanks to Bob Wil- liams, Ted Stubblefield, and Don Ostby of the Many individuals contributed to the design and Pacific Northwest Region, USDA Forest Service; ilnplementation of the DEMO study including: Jennifer Belcher of the Washington State Depart- Larry Bednar, Gary Clendenen, Robert Curtis, ment of Natural Resources; and Tom Mills of the Dean DeBell, Richard Haynes, John Henshaw, Pacific Northwest Research Station, USDA Forest John Lehmkuhl, Steve MacDonald, Don Minore, Service for their significant contributions to estab- Martin Raphael, and Fred Swanson of the Pacific lishing this study and ensuring its completion. Northwest Research Station, USDA Forest Ser- vice; Rick Abbott, Debbie Anderson, Debbie This is a product of the Demonstration of Eco- Couche, ~arbaiaFontaine, Fred Henchell, Wayne system Management Options (DEMO) study, a Kleckner, Sue Macmeeken, Jon Nakae, and Ed joint effort of the USDA Forest Service Region 6 Tompkins of Region 6, USDA Forest Service; Dean and Pacific Northwest Research Station. Research Berg, E. David Ford, Jerry Franklin, and Joel partners include the University of Washington, Reynolds of the College of Forest Resources, Oregon State University, University of Oregon, University of Washington; Logan Norris, John Gifford Pinchot and Umpqua National Forests, Tappeiner, and David Hibbs of the College of and the Washington State Department of Natural Forestry, Oregon State University; and Richard Resources.

Literature Cited Cazares, E., D. L. Luoma, M. P. Amaranthus, C. L. Cham- bers, and J. F. Lehmkuhl. 1999. Interaction of fun- Abbott, R. S., J. D. White, and B. L. Woodard. 1999. The gal sporocarp production with small mammal abun- benefits and challenges of large-scale silvicultural dance and diet in Douglas-fir stands of the southern experiments: perspectives from forest managers on Cascade Range. Northw. Sci. 73 (Spec. Iss.):64-76. the Demonstration of Ecosystem Management Options (DEMO) study. Northw. Sci. 73 (Spec. Iss.):118-125. Coates, K. D., A. Banner, J. D. Steventon, P. LePage, and P. Anonymous. 1996. Demonstration of Ecosystem Manage- Bartemucci. 1997. The Date Creek silvicultural sys- ment Options: A study of green-tree retention levels tems study in the interior cedar-hemlock forests of and patterns in western Oregon and Washington. northwestern British Columbia: overview and treat- Unpubl. study plan, USDA For. Serv., Pacific Northw. ment summaries. Land Manage. Hndbk. 38, B.C. Min. Res. Sta., Olympia, Washington. 108 p. For., Victoria. Arnott, J. T. and W. J. Beese. 1997. Alternatives to clearcutting DeBell, D. S. and R. 0. Curtis. 1993. Silviculture and new in BC coastal montane forests. For. Chron. 73:670- forestry in the Pacific Northwest. J. For. 9 1 :25-30. 678. Ervin, K. 1989. Fragile Majesty: The Battle for North Bormann, B. T., P. G. Cunningham, M. H. Brookes, V. W. America's Last Great Forest. The Mountaineers, Manning, and M. W. Collopy. 1994. Adaptive eco- Seattle. system management in the Pacific Northwest. USDA For. Serv., Gen. Tech. Rep. PNW-GTR-34 1. FEMAT. 1993. Forest ecosystem management: an ecologi- Carey, A. B. 1991. The biology of arboreal rodents in Dou- cal, economic, and social assessment. Report of the glas-fir forests. USDA For. Serv., Gen. Tech. Rep. Forest Ecosystem Management Assessment Team. PNW-GTR-276. USDA For. Serv.; USDC Nat. Ocean. Atmos. Admin. Carey, A. B., B. L. Biswell, and J. W. Witt. 1991. Methods and Nat. Mar. Fish. Serv.: USDI Bur. Land Manage., for measuring populations of arboreal rodents. USDA Fish and Wildl. Sen., and Nat. Park Serv.; and US For. Serv., Gen. Tech. Rep. PNW-GTR-273. EPA, Washington, D.C. Carey, A. B.. D. R. Thysell, and A. W. Brodie. I12 prsss. The Franklin, J. F., D. R. Berg, D. A. Thornburgh, and J. C. Forest Ecosystem Study: Background, rationale, Tappeiner. 1997. Alternative silvicultural approaches implementation, baseline conditions, and silvicultural to timber harvesting: variable retention systems. 111 assessment. USDA For. Serv., Gen. Tech. Rep. PNW- K. A. Kohm and J. F. Franklin (eds.). Creating a For- GTR-XXX. estry for the 21st Century: The Science of Ecosys- Carey, A. B., D. R. Thysell, L. J. Villa, T. M. Wilson, S. M. tem Management. Island Press, Washington, D.C. Pp. Wilson, J. M. Trappe, W. Colgan 111, E. R. Ingram. and M. Holmes. 1996. Foundations of biodiversity in man- I11-139. aged Douglas-fir forests. In D. L. Peterson, and C. V. Franklin, J. F., K. Cromack, Jr., W. Denison, A. McKee. C. Klimas (eds.). The role of restoration in the ecosys- Maser, J. Sedell, F. Swanson, and G. Juday. 1981. tem. Proc. 2"* Symp. Ecol. Restor., Society for Eco- Ecological characteristics of old-growth Douglas-fir logical Restoration, Madison, Wisconsin. Pp. 68-82. forests. USDA For. Serv., Gen. Tech. Rep. PNW- I 18.

DEMO: A Study of Green-tree Retention 25 Franklin, J. F., L. A. Norris, D. R. Berg, and G. R. Smith. Progar, R. A,, T. D. Schowalter, and T. Work. 1999. Arbo- 1999. The history of an experiment in regeneration real invertebrate responses to varying levels and pat- harvest of northwestern forest ecosystems. Northw. terns of green-tree retention in northwestern forests. Sci. 73 (Spec. Iss.):3- I I. Northw. Sci. 73 (Spec. lss.):77-86. GutiCrrez. R. .I. and A. B. Carey (eds.). 1985. Ecology and Ribe, R. 1999. Regeneration harvests versus clearcuts: public management of the spotted owl in the Pacific North- views of the acceptability and aesthetics of Northwest west. USDA For. Serv., Gen. Tech. Rep. PNW-185. Forest Plan harvests. Northw. Sci. 73 (Spec. Iss.): 102- Halpern. C. B., S. A. Evans, C. R. Nelson, D. McKenzie, D. 117. Ruggiero, L. F., L. L. C. Jones, and K. B. Aubry. 1991. Plant A. Liguori, D. E. Hibbs, and M. G. Halaj. 1999. and animal habitat associations in Douglas-fir forests Response of forest vegetation to varying levels and of the Pacific Northwest: an overview. 112 L. F. patterns of green-tree retention: an overview ofa long- Ruggiero, K. B. Aubry, A. B. Carey, and M. H. Huff term experiment. Northw. Sci. 73 (Spec. lss.):27-44. (tech. coords.). Wildlife and vegetation of unliianaged Harrington, C. A. and A. B. Carey. 1997. The Olyrnpic Habitat Douglas-fir forests. USDA For. Set-v.,Gen. Tech. Rep. Developmen1 Study: Effects of thinning, manipula- PNW-GTR-285. tion of coarse woody debris and understory manage- Storck, P., T. Kern, and S. Bolton. 1999. Measurement of ment on stand structure and cornposition of under- differences in snow accumulation, melt, and microme- story plants and forest-floor small mammals. Unpubl. teorology due to forest harvesting. Northw. Sci. 73 study plan, USDA For. Serv., Pacific Northw. Res. (Spec. Iss.):87- 10 1. Sta., Olympia, Washington. 38 p. Tappeiner, J. C. 1992. Managing stands for northern spotted Isaac. L. A. 1956. Place of partial cutting in old-growth stands owl habitat. Appendix F i17 Final Recovery Plan for of the Douglas-fir region. USDA For. Serv., Res. Pap. the Northern Spotted Owl. USDI Fish and Wildlife 16. Serv., Portland, Oregon. Thomas, J. W., E. D. Forsman, J. B. Lint, E. C. Meslow, B. Lehmkuhl, J. F., L. F. Ruggiero, and P.A. Hall. 1991. Land- R. Noon, and J. Verner. 1990. A conservation strat- scape-scale patterns of forest fragmentation and wild- egy for the northern spotted owl: a report of the In- life richness and abundance in the southern Washing- teragency Scientific'Committee to Address the Con- ton Cascade Range. In L. F. Ruggiero, K. B. Aubry, servation of the Northern Spotted Owl. USDA For. A. B. Carey. and M. H. Huff (tech. coords.). Wildlife Serv. and USDI Bur. Land Manage., Fish Wildl. Serv., and vegetation of unmanaged Douglas-fir forests. and Nat. Park Serv., Portland, Oregon. USDA For. Serv., Gen. Tech. Rep. PNW-GTR-285. Tuchmann, E. T. 1995. Timber supply and the President's Pp. 425-442. forest plan: trials and triumphs. J. For. 93:6-10. Lehmkuhl. J. F., S. D. West, C. L. Chambers, W. C. McCornb, Tuchmann, E. T., K. P. Connaughton, L. E. Freedman, C. B. D. A. Manuwal, K. B. Aubry, J. L. Erickson, R. A. Moriwaki. 1996. The Northwest Forest Plan: A re- Gitzen, and M. Leu. 1999. An experiment for as- port to the President and Congress. USDA Off. For. sessing wildlife response to varying levels and pat- Econ. Asst., Portland, Oregon. terns of green-tree retention. Northw. Sci. 73 (Spec. USDA and USDI. 1994a. Record of decision for amend- ments to Forest Service and Bureau of Land Man- Iss.):45-63. agement planning documents within the range of the Norse, E. A. 1990. Ancient Forests of the Pacific North- northern spotted owl. USDA For. Serv., Portland, west: The Grandeur, Complexity, Diversity, and Im- Oregon. pending Destruction of a Fragile and Vital Ecosys- . 1994b. Final supplemental environmental impact tem. Island Press, Washington, D.C. statement on management of habitat for late-succes- Pipp, A. K. 1998. Effects of forest age versus forest struc- sional and old-growth related species within the range ture on epiphytic lichen biomass and diversity. Uni- of the northern spotted owl. USDA For. Serv., Port- versity of Montana, Missoula. M.S. Thesis. land, Oregon.

Endnote [Memo from Donald R. Knowles, Executive Director, Re- gional Ecosysten~Office to Charles W. Philpot, Station Di- rector, Pacific Northwest Research Station dated May 31, 1995.

26 Aubry et al.