Modeling changes in wildlife habitat and timber revenues in response to forest management

Modeling changes in wildlife habitat and timber revenues in response to forest management

Marzluff, John M

ABSTRACT. Few models evaluate the effects of forest management options on wildlife habitat and incorporate temporal and spatial trends in forest growth. Moreover, existing habitat models do not explicitly consider economic trade-offs or allow for landscape level projections. To address these concerns, we linked standard wildlife habitat suitability models with habitat projections from the Landscape Management System (LMS). LMS integrates spatially explicit forest inventories with forest growth, decay, and silviculture treatment (e.g., planting, thinning, harvesting) models to compare some economic and biological impacts of forest management on wildlife habitats at spatial scales ranging from the individual forest stand to the landscape. We used LMS to quantify pileated woodpecker (Dryocopus pileatus), Cooper’s hawk (Accipiter cooperl), and southern red-backed vole (Clethrionomys gapperl) habitat qualities across landscapes and to project habitat changes through time. We modeled five forest management scenarios (including no action, intensive management for timber production, moderate management and intensive management to enhance mature forest structure, and mixed management for wildlife and timber) proposed for the 566 ha Satsop Forest in western Washington. The selected wildlife species’ habitats (mixture of mature conifer and deciduous forests) were maximized by simply allowing the forest to grow with no management. However, this cost an estimated $929,539/yr in lost timber revenue. Intensive thinning, replanting, and retention of large trees increased habitat for all focal species and provided $384,558/yr return from timber. Because of potential short-term reductions in habitat that occur during long-term enhancement strategies (e.g., intensive thinning) and limited ability to model understory growth dynamics, monitoring and validation of predictions would be necessary. FOR. Sci. 48(2):191-202.

Key Words: Economics, forest modeling, habitat suitability, landscape ecology, forest management, wildlife habitat.

QUANTIFYING THE EFFECTS of landcover modification on wildlife habitat has necessitated the use of habitat models. Consequently, a number of models have been developed to aid in the evaluation of forest management activities on the quality of wildlife habitat (Rushton et al. 1997, Bevers and Hof 1999, Ji and Jeske 2000, Moore et al. 2000). A habitat “accounting” system, termed Habitat Evaluation Procedures (HEP), developed by the U.S. Fish and Wildlife Service has become a standard tool for impact assessment on wildlife species (USDI 1981). Habitat quality is expressed in terms of a habitat suitability index (HSI), derived through a modeling process based on measurements of specific habitat features and environmental variables deemed most important to a species’ presence, distribution, and abundance. Suitability scores for each habitat feature or environmental variable range from 0 (lowest suitability) to 1 (highest suitability). The combination of all individual suitabilities results in the computation of a final HSI value, ranging from 0 to 1, that is a measure of habitat quality for that species for that particular habitat. HSI-based models have been used extensively to quantify the effect of timber management activities on a variety of wildlife (e.g., Rempel et al. 1997, Kliskey et al. 1999). To assess the cumulative positive or negative effects of a management action, the HSI for each habitat type is multiplied by the number of ha of that habitat type to obtain Habitat Units that can be summed for all habitat types in the study area. In this way, Habitat Units are the “currency” used to assess how much wildlife habitat has been lost or gained as a result of management activities.

Despite the widespread use and acceptance of HSI models, there are few mechanisms to evaluate habitat management options on wildlife habitat while incorporating temporal and spatial trends in forest growth. Silvicultural prescriptions are among the most popular techniques to manage wildlife habitat, and much attention has focused on formulating guidelines for optimal ratios and juxtaposition of habitat types. For example, among the most cited and implemented timber guidelines designed to improve elk (Cervus elaphus) habitat is the 60:40 ratio of forage to cover habitat per land unit as recommended by Thomas et al. (1979). Despite such consideration of silvicultural treatments on wildlife habitat, there are few mechanisms to evaluate the effects of such activities over the long term. Instead, HSI scores are computed for one forest condition without explicitly incorporating how forest structures may change through time. Moreover, economic trade-offs associated with management alternatives are rarely integrated with wildlife habitat models. Sustainable ecosystem management requires the integration of both ecological and economic considerations (Boyce and Haney 1997). The evaluation of alternative management activities is a major challenge (McCarter et al. 1998).

We explored the use of the Landscape Management System (LMS) with existing stand inventory information to evaluate five forest management prescriptions on the quality of wildlife habitat for three species inhabiting the Satsop Forest in southwestern Washington. The goal of this project was not to determine an optimum management plan for this forest or consider all economic values of habitat. Instead, our objective was to illustrate the usefulness of decision-support tools such as LMS in planning for wildlife habitat and economic return from timber harvest; such tools are rarely incorporated into wildlife habitat planning. We met this objective by interfacing LMS with commonly used and accepted habitat models that estimate habitat suitability for three species. Other habitat models and economic valuations could be combined in a similar way. In particular, we encourage future models to include explicit valuation of wildlife. Wildlife has economic, ecologic, social, and aesthetic values that we have not included in our assessments.

Study Area

The Satsop Forest encompasses roughly 566 ha south of the Chehalis River in southwest Washington State owned by the Grays Harbor Public Development Authority (PDA). About 340 ha are forested, and 89 ha are in grassy meadows, shrubs, or riparian areas and not included in timber management analysis. The remaining 137 ha are part of the developed infrastructure of the Satsop Development Park. Elevation is relatively constant at 75 in, and forest productivity is moderate to high [56% of the forest is Site Class II, Site Index 35 m-41 in (50 yr index age); Figure 1A]. Stands range from pure conifer to pure hardwood, but most are complex mixtures of Douglas-fir (Pseudotsuga menziesii), big leaf maple (Acermacrophyllum), western hemlock (Tsuga hetrophylla), red alder (Alnus rubra), and western redcedar (Thuja plicata; Figure lB). Most of the area is covered by forest 5-10 yr old; forests >120 yr were present in small proportion on the site (Figure 1B).

Currently Satsop Forest management is dictated by a wildlife mitigation agreement. This agreement was put in place to mitigate for wildlife habitat losses caused by the partial construction of Washington Public Power Supply System’s Nuclear Plants No. 3 and No. 5. Even though this agreement was created prior to the PDA’s acquisition of the land, it is still in place. Management under the agreement is equivalent to the “Moderate Enhancement” scenario defined and modeled below.

The PDA’s management objectives for Satsop Forest extend beyond strictly the creation and enhancement of wildlife habitat. There is a desire to create a demonstration forest where alternative silvicultural practices that create both wildlife habitat and income can be showcased to other landowners with the primary focus being nonindustrial private forest (NIPF) owners.


The Landscape Management System

The effect of forest management activities on habitat quality was projected using LMS (McCarter et al. 1998). LMS is a flexible, computer-based, landscape-level analysis tool that integrates forest inventory information, growth models, and graphical and visualization software to project changes in forest structure and composition over time (McCarter 1997, McCarter et al. 1998). Component programs include growth models [all versions of the Forest Vegetation Simulator (FVS; Edminster et al. 1991) and other models], interactive stand treatment programs, table and graph outputs, and stand and landscape visualization programs. Data sources necessary for LMS include stand inventory information (tree-based measurements), landscape data (slope, aspect, elevation, site quality), and Geographic Information System (GIS) data (stand boundaries, streams, roads, etc.). Growth modeling for this study used the Pacific Northwest Variant of FVS and 5 yr growth projection periods. Using LMS it is possible to project stands and landscapes forward in time to predict potential future stand and landscape forest conditions, while treating stands through harvesting, planting, and other activities to simulate potential management practices. Analysis of the projected future stand conditions and product outputs can be done with a variety of tables and graphs produced through LMS as well as visualizations of projected future stand and landscape conditions using Stand Visualization System (SVS) and Envision programs from the USDA Forest Service embedded in LMS (McGaughey 1997).

Forest Management Scenarios

We simulated forest growth and resulting wildlife habitat and timber values using five different management scenarios over an 80 yr planning horizon (1998-2078). The initial year of the simulations (1998) was the year that the forest inventory was performed on Satsop Forest.

Scenario 1: No Action. We allowed forests to grow without silvicultural treatment for 80 yr.

Scenario 2: Intensive Management for Timber Production. We modeled an industry standard (Michaelis 2000), 45 yr rotation by precommercially thinning dense stands to 741 trees/ha at age 15 yr old, clearcutting leaving 12 trees/ha to meet WA Forest Practices Rules at age 45, and planting 1,110 trees/ha of Douglas-fir. We chose this scenario to maximize revenue generated by timber harvest.

Scenario 3: Moderate Management to Enhance Mature Forest Structures. We accelerated the development of complex forest structure by thinning any conifer stands in the 25– 40 yr range from below to 370 trees/ha between 2018 and 2038, leaving the biggest trees, and then underplanted with 124 trees/ha of western hemlock and western redcedar. We chose this scenario as a method of accelerating multilayered canopy development on younger planted stands with minimal silvicultural activities and resulting potential disturbances to wildlife.

Scenario 4: Intensive Management to Enhance Mature Forest Structures. We designed stand-specific pathways to manipulate each stand through a series of thinnings to promote the development of late-successional structural characteristics (multiple canopy layers and large diameter trees). We designed each thinning to open the stand enough to allow understory development and canopy regeneration, while maintaining residual trees from each canopy layer present before treatment. We planted multiple species including Douglas-fir, western hemlock, and western redcedar to promote species diversity and structural development. We divided stands into six groups according to age and species composition.

We precommercially thinned stands in Group 1 (conifer stands

We commercially thinned stands in Group 2 (conifer stands >40 yr old; “old”) between 1998 to 2018 to maintain the existing canopy layers, promote the release of advanced regeneration, and establish regeneration of shade tolerant species. Because two canopy layers already existed in these stands, we designed the thinning prescription to retain trees from each layer and underplanted to create stands with three or more layers. Between 1998 and 2013, we commercially thinned stands to 124 trees/ha by leaving the 62 tallest trees in the >50 cm dbh size class and the 62 trees/ha with the largest dbh

Group 3 (young deciduous stands) contained many dense hardwood stands that established through natural seeding. We designed a pathway to convert these stands into conifer stands that would be thinned and underplanted several times to promote the development of multiple canopy layers. We clearcut stands between 1998 and 2038 and replanted with 1,111 trees/ha of Douglas-fir. At age 20, we precommercially thinned these stands from below to 618 trees/ha. At age 40 we thinned stands to 62 trees/ha and underplanted 741 trees/ha of Douglas-fir, western hemlock, and western redcedar.

Group 4 (old deciduous stands) contained hardwood stands with mature forest structures. We designed a pathway to maintain existing mature forest structures while increasing the conifer component in the stands. Between 2018 and 2033, we thinned stands by removing all trees 50 cm dbh to 62 trees/ha, reduced trees 15-50 cm dbh to 87 trees/ha, and retained all trees

Group 5 (young mixed conifer/deciduous stands) contained dense mixed conifer/ hardwood stands that resulted from planting conifers after earlier clearcutting combined with natural seeding by conifers and hardwoods. We designed this pathway to change stands to pure conifer and encourage a multilayered canopy. Between 1998 and 2008, we thinned stands by removing all hardwoods. Between 2038 and 2053, we thinned stands to 62 trees/ha from below and underplanted with 741 trees/ha of Douglas-fir, western hemlock, and western redcedar.

Group 6 (old mixed conifer/deciduous stands) contained older mixed stands that had some mature forest characteristics. We designed a pathway to maintain older forest structure while still allowing silvicultural activities. Between 2018 and 2033, we harvested stands leaving the largest 62 trees/ha and underplanted 741 trees/ha of Douglas-fir, western hemlock, and western redcedar. We did a second, diameter-limiting thin between 2058 and 2073 that retained the largest 62 trees/ ha in the 15-53 cm dbh size class and the largest 62 trees/ha in the >53 cm dbh class. Following thinning, we underplanted 741 trees/ha of Douglas-fir, western hemlock, and western redcedar.

Scenario 5: Mixed Management for Wildlife and Timber Values. We managed young stands on the most productive soils (50 yr site index of >43 m; Figure IA) for intensive timber production as in scenario 2, but left the remaining forest as an untreated reserve for wildlife habitat. This resulted in 34 stands totaling 118 ha managed for timber with the remaining 178 ha managed for wildlife habitat. In our simulations, mature forest was the preferred wildlife habitat (see wildlife models below), so we did not silviculturally treat stands designated as wildlife habitat. We selected this scenario to simulate coupling intensive timber production with wildlife reserves on a small landscape.

Habitat Models

To assess the effects of the five management alternatives, we implemented the HER We used published habitat suitability models for the pileated woodpecker (Schroeder 1983), Cooper’s hawk (USDI 1980), and southern red-backed vole (Allen 1983), three important nongame species in the Satsop Forest area. These species were selected because models were readily available, and each species represents an ecologically important component of the Satsop Forest.

Pileated woodpecker was included because it is a large cavity nester that prefers late-successional forests (Schroeder 1983). Their optimal habitat was defined as forests with >75% canopy closure, >74 trees/ha with >76 cm dbh, >25 stumps/ha >0.3 m tall and 18 cm diameter or logs >18 cm diameter, >0.43 snags >51 cm dbh per ha, and snag average dbh of >76 cm (Schroeder 1983).

Cooper’s hawk was included because it prefers mature deciduous stands. According to the HSI model, their optimal habitat is forests with >60% canopy cover, >51 cm average dbh, and 10-30% conifer canopy closure (USDI 1980).

Southern red-backed voles are an important prey source for locally threatened species that prefer mature and late– successional conifer forests. Their optimal habitat was defined as sites containing >31 cm average dbh, >20% ground cover of downfall >7.8 cm, >80% grass cover, and >50% evergreen canopy closure (Allen 1983).

For each species, Habitat Unit values are reported. Habitat Unit values are relative measures of both quantity and quality of habitat. For each species, these are calculated by multiplying the habitat suitability index (HSI) by the acreage of habitat for the species. HSI values range from 0.0, no habitat value, to 1.0, optimal habitat value.

Timber Volume and Timber Economics

LMS reports both standing timber volume and harvested timber volume through time. Standing volume is calculated as the total standing volume at the end of each growth period. Cut volume is the amount of timber harvest that occurred during each 5 yr growth period. Volumes are calculated by FVS based on the Scribner 32 ft (9.75 in) log rule with a minimum top diameter outside bark of 4.5 in. (11.5 cm). We report these values at 5 yr intervals separately for three size classes. “Poles” are trees 61cm dbh that provide high quality wood for lumber including specialty, clear, tight-grained woods used in boat planking, siding, molding, and ladders.

We approximated the relative economic value of wood products using current timber prices from the November 2000 issue of Log Lines. Reported prices are average October prices per thousand board feet (mbf) of Douglas-fir for the local (Twin Harbors) region of Washington State. Pole timber produces chip and saw logs for the domestic market that have a price of $481/mbf. Small sawlogs produce chip and saw logs for the domestic market and Japan (J12 and J8), China (C 12H), and Korea (K8) log export markets. The price is $880/mbf, $755/ mbf, $718/ mbf, and $668/ mbf for J12, J8, C12H, and K8 logs, respectively. Large sawlogs provide all the above log grades plus J14 for export. The price for Douglas-fir J14 is $913/ mbf. The determining factor in export quality is log surface quality. We used an average of the above prices for calculations resulting in values of $700/ mbf for small sawlogs and $736/ mbf for large sawlogs. These values were then used to calculate discounted total revenue for each management scenario. This was done for the full 80 yr simulation with a discount rate of 4%.

Assessing Forest Stand Attributes

During the original HEP, Satsop Forest was divided into 116 forest stands with similar management histories, site potentials, and standing vegetation that ranged in size from 0.17 ha to 17.1 ha. Cover typing was done according to rules set forth in the original HEP conducted in 1994. Cover types were delineated on aerial photos based on vegetation type and stand timber characteristics. Forested stands were classified into 13 cover types based on canopy closure, percent conifer/ deciduous, stocking, average dbh, average height, number of canopy layers, and number of big trees (>53 cm) per ha (Table 1). At the end of our simulated treatment and growth period for each scenario, we reclassified the cover type for each stand in the Satsop Forest according to Table 1.

During the summer of 1998, a timber inventory was conducted to collect tree, snag, and downed wood data for each of the 116 stands on the 340 forested ha. Stands sampled were those delineated in the 1994 HEP process with some splitting of stands where definite cover type breaks were identified on aerial photographs taken in August 1997. We measured 248 plots with an average of 2.2 plots per stand or about 1 plot per 2.9 ha. The number of plots ranged from 1 for the smallest stands to 12 plots for the largest stand.

The Satsop Forest inventory followed USFS inventory protocol for plot layout (USDA 1989). Initially, a 100 m grid covered Satsop Forest. Because Satsop Forest contains numerous small, narrow stands it was quickly discovered that some stands were missed by the grid. To overcome this, we chose to perform a “representative” inventory that allowed us to sample at least one plot per stand in the small stands that were missed by the grid. Samples consisted of two nested plots: a variable radius plot and a fixed radius plot. In the variable radius plot, a basal area factor (baf) of 20 or 40 ftZ/ac (approximately 4.6 or 9.2 m2/ha) was used, depending on tree size. The goal was to maintain approximately eight trees per plot. The fixed radius plot was 0.00 13 ha where all trees with a dbh of 12 cm or less were measured. For all trees, we recorded species, dbh, and height; age, crown ratio and crown class were taken for the tallest dominant tree, or site tree, in each plot. Site tree information was used to measure site index and age of the stand. Tree heights were calculated by the FVS growth model based on age, dbh, and site index.

Height and dbh were measured on snags in the variable radius plot if they were counted as “in” using the appropriate baf. Downed wood diameter, length, and depth were measured if it was all or partially in the fixed radius plot.

Landscape Attributes.-We calculated landscape attributes (mean elevation, mean aspect, mean slope, and polygon acreage) for each polygon on Satsop Forest using ArcInfo GIS processing coverages created from cover type maps generated during the original HEP, USGS topographic quadrangle maps, and USGS digital elevation models (DEM).

HIS Input Variables.-HSI models require input data for various habitat attributes for each species in the HEP evaluation. We calculated tree-based variables, percent canopy closure, percent conifer canopy closure, percent deciduous canopy closure, dominant tree height, dominant tree dbh, average dbh, number of large trees, number of canopy layers, using programs embedded in LMS based on current and future stand inventories projected with the FVS growth model. Conifer/deciduous percentages were determined based on percentage of total stand basal area in either conifer or deciduous species. Canopy closure for all species, conifer species only, and deciduous species only were calculated using an equation from Crookston and Stage (1999) that converts total crown basal area to canopy closure assuming crown overlap. Calculation of the number of trees/ha >53 cm dbh was performed by summing the expansion factors for all tree records with dbh >53 cm. An estimation of the number of canopy layers was done using an algorithm developed by Baker and Wilson (2000). Understory and standing dead and downed wood variables, which are not included in LMS, were taken from the 1994 HEP analysis and related to stands after assigning a cover type to each stand. Once a stand is assigned a cover type, it is assigned values for understory and standing dead and downed wood from the HEP data table. Habitat Units for each stand were calculated by multiplying HIS per stand by the stand’s land area.


Stands developed differently through time depending on the scenario we used to manage them (Figure 2). Forests simply matured in the absence of silvicultural treatments or with moderate thinning. With intensive timber harvest, the forest cycled through dense regenerating forests, dense preharvest forests, and clearcuts. Thinning and replanting to accelerate the development of mature forest structure transformed regenerating forests into multistoried forests in 80 yr. Depending on the management scenario, stand structures after 80 yr were drastically different (Figure 2).

No Action

Wildlife habitat value increased and remained well balanced among the three focal species in the absence of any silvicultural management (Figure 3). Vole and woodpecker habitat increased as forests matured and became increasingly dominated by larger trees (Figure 4). Cooper’s hawk habitat decreased slightly as large conifers increasingly dominated mixed stands. No timber was harvested by definition with this scenario, so no timber revenues were produced.

Intensive Management for Timber

All species’ habitats remained constant or declined with intensive timber management (Figure 3). Vole and hawk habitat declined slightly during the first decade, but after 40 and 80 yr had returned to original conditions. In contrast, pileated woodpecker habitat declined severely with intensive management for timber production (Figure 3C). The decline resulted from reductions in standing volume of trees by harvest, especially the volume of large trees (Figure 5). Few snags were left on the landscape, further reducing woodpecker habitat. After 80 yr of intensive management, woodpecker habitat was reduced to 21 % of the starting amount.

Considerable timber was harvested each yr, but the short rotation produced a shift over time in harvest from primarily large sawlogs to small sawlogs (Figure 5). On average, 133.3 mbf of poles, 852.6 mbf of small sawlogs, and 365.0 mbf of large sawlogs were harvested each yr. This produced an annual average income from timber of $929,539 and discounted total revenue of $19,779,691.

Moderate Management to Enhance Mature Forest Structures

Habitat values resulting from this scenario did not differ appreciably from the no action scenario. Large trees dominated the forest after 80 yr (Figure 4) and habitat for all three focal wildlife species occurred in stable or increasing amounts (Figure 3). Pileated woodpeckers benefited most under this management regime because our thinning and underplanting increased development of multilayered forests that were assumed to have many large snags. Pileated woodpecker habitat increased by 106% over 80 yr.

Timber harvest was minimal, concentrated during a 20 yr period and produced only poles and small sawlogs for the market (Figure 5). Annually, an average of 19.5 mbf of poles and 25.8 mbf of small sawlogs worth $27,415 were harvested. For the 80 yr simulation, the discounted revenue from timber was $641,462.

Intensive Management to Enhance Mature Forest Structures

Intensive management to increase structural complexity using silvicultural techniques increased all species’ habitats (Figure 3). Vole and hawk habitat increased and then remained relatively constant throughout the simulation. Over the 80 yr, vole habitat increased by 35% and hawk habitat increased by 15%. Woodpecker habitat increased steadily to improve by 117% after 80 yr.

Repeated thinning, retention of residuals, and replanting resulted in substantial harvest (Figure 5) and minimal standing volume (Figure 4). On average, each yr 108.5 mbf of poles, 317.3 mbf of small sawlogs, and 124.2 mbf of large sawlogs worth $365,668 were harvested (Figure 5). Total discounted revenue from this harvest was $6,593,858.

Mixed Management for Wildlife and Timber

Concentrating timber harvest on the most productive forest stands and managing the rest of the ownership for wildlife maintained habitat for all three species. Pileated woodpecker habitat increased by 40% over 80 yr. Cooper’s hawk habitat remained relatively unchanged, and vole habitat cycled at a 40 yr interval (Figure 3). This cyclic behavior likely resulted from concentrating harvest at 45 yr intervals. The forest grew to a considerable standing volume dominated by large trees on nonharvested areas (Figure 4).

Harvest was increased over the moderate enhancement scenario (Figure 5), but remained composed primarily of small sawlogs (275.9 mbf /yr). Timber production was concentrated around year 45 because of the age class distribution of the highest site stands. Economic return from timber was moderate, averaging $244,736/yr with a discounted total revenue of $3,545,627 for the 80 yr simulation.


Satsop Forest Management

Our three focal species clearly benefited from allowing the forest to mature naturally. The benefit of mature forest was most striking in the response of woodpeckers to management predicated on a 45 yr cutting rotation (Figure 3). However, the economic consequences of forgoing such intensive management are equally striking; 1,350 mbf /yr of poles, small sawlogs, and large sawlogs worth $929,539/yr were not harvested (Figure 5). Total discounted revenue from timber forgone by choosing this management scenario is also striking at $19,779,691.

Our models suggest some compromises are possible. Moderate management to enhance mature forest structural components maintains wildlife habitat (Figure 3) and allowed timber valued at $27,415 to be harvested each yr with total discounted revenue value of $641,462. A slightly more aggressive strategy of managing the most productive sites for timber and allowing the lower quality sites to age naturally allowed an annual return of $244,736 (26% of the return obtained with a pure 45 yr rotation) and total discounted revenue value of $3,545,626 (18% of the total discounted revenue value of a pure 45 yr rotation) without appreciably reducing Cooper’s hawk and pileated woodpecker habitat. However, vole habitat declined at 45 yr intervals. Such reductions may be tolerable if vole populations remain viable at resulting lowered sizes, or perhaps if larger areas are maintained in this less ideal habitat. Adaptive monitoring would be needed to evaluate wildlife responses.

Extensive thinning to accelerate forest growth produced substantial economic returns and increased wildlife habitat for our long-term projections (Figure 3). During the 80 yr simulation, thinning produced $365,668/yr, 39% of what was produced with an aggressive 45 yr rotation, with total discounted revenue value of $6,593,858, 33% of the 45 yr rotations. Woodpeckers fared substantially better with thinning than with the 45 yr rotation because a mixture of deciduous and coniferous trees was maintained, and large trees developed quickly and were retained. However, species that require the decadent aspects of old forests (e.g., large snags) may have difficulty maintaining viable populations in the face of intensive thinning. Specific monitoring of such species will be needed if aggressive thinning scenarios are implemented. Targeted management for habitat elements required by such species (e.g., creating snags or decay) may also be necessary during the middle years of thinning programs.

Although some timber revenue appeared compatible with maintaining wildlife habitat, we identified a potentially important problem with the timing of some returns. Because harvest scheduling was based on stand age, and older stands were set aside for wildlife and excluded from harvesting, a boom-bust cycle of timber harvest often resulted with moderate harvest scenarios (Figure 5). This would translate into a cyclic economy that could disrupt cash flow, employment, and community stability. This cycle was purely a result of focusing wildlife management on the oldest stands and focusing timber management on the youngest stands. It could be alleviated by selecting a range of stand ages to be managed for timber production at the onset. Then, some harvest could be scheduled each year and a more stable economy could develop.

Modeling Wildlife Habitat and Economic Returns

The ability to rapidly model forested landscape responses to silvicultural treatments through time, while projecting wildlife habitat quality and economic return from timber harvest, is a real strength of our approach. With few exceptions, current approaches to wildlife habitat modeling do not incorporate landscape projection tools, or allow the user to consider a variety of forest management options (i.e., a variety of thinning levels, plantings). Instead, wildlife managers consider the effects of forest management activities at the time of activity to evaluate effects on wildlife habitat. This may provide a much less complete and perhaps less accurate appraisal of the effects of silvicultural treatments compared to the dynamic evaluation allowed in LMS. Exceptions to this generalization include recent uses of DYNamically Analytic Silviculture Technique (DYNAST) (Boyce 1980) and FORPLAN (Kirkman et al. 1984) to calculate habitat capability for wildlife species (Boyce and McNab 1994). However, Morrison et al. (1998) report that these models typically lack sensitivity to spatial patterns of forest stages, do not allow landscape level projections, often lead to erroneous management recommendations, and lack appropriate output format and information to effectively compute habitat suitability for a variety of wildlife species. Further-more, these stand growth-and-yield models of forests used by wildlife managers depict stand growth without allowing additional management actions such as artificial planting or prescribed levels of forest thinning (Bettinger et al. 1996, 1997, Morrison et al. 1998). In contrast, as this case study illustrates, LMS was sensitive to spatial patterns, produced useful output because HSI models were contained explicitly within the program, and allowed for an assortment of realistic forest management options.

Of the concerns raised by Morrison et al. (1998), the only one not specifically addressed in our case study is the chance that forest growth models may lead to erroneous management recommendations. However, we believe this problem is due primarily to user applications of the resultant information rather than the decision support tool per se. Managers must realize that LMS and other decision support tools (i.e., models) illustrate our best guess of how complex processes work. Therefore, managers should not implement models blindly and instead use the modeling process in a hypothesis-testing framework to evaluate ecological and economic trade-offs.

We used LMS to simulate only one component of economic return, revenue generated by timber harvest. Certainly, wildlife and its habitat have other economic benefits that could be quantified using the techniques of ecological economics (Loomis 2000). These techniques could allow managers to estimate potential economic losses from reducing or degrading wildlife habitat by cutting timber. Such losses could be modeled along with economic gains from the harvest to allow a more complete assessment of economic trade-offs resulting from a variety of forest management scenarios. We encourage managers to incorporate complete economic assessments in their evaluations of forest management scenarios and then link such evaluations to changes in wildlife habitat as we have done here.

Rather than modeling habitat suitability for individual wildlife species, as was done here, we also could easily project predefined habitat types through time to quantify changes in the structure of the forested landscape. This would allow us to project changes in biodiversity and evaluate ecosystem responses to landscape management options. However, we favor an approach that models habitat suitability of individual wildlife species so that the ecosystem responses can be understood mechanistically by knowing how representative populations respond to management. Provided model availability, many more species could be evaluated simultaneously just as rapidly, and trade-offs among them weighed. For example, the three species we modeled all responded positively to increasing forest maturation, but the preponderance of small conifers in intensively managed forests has less impact on vole and hawk habitat than it did on woodpecker habitat (Figure 3). The economic gains of short rotations (Figure 5) could then be evaluated in terms of biodiversity costs to woodpeckers, given stability in hawk and vole habitat. To conduct such an analysis requires better wildlife habitat models, preferably ones accurately linking population viability to habitat. Incorporation of habitat patch size, shape, and configuration on the landscape could also improve models. Applications of optimization models by Hof et al. (1993, 1994) examined spatial and temporal habitat and harvesting arrangements on an area. These were limited in the scope of the species used and applicable harvest methods. With refinement these could be used in concert with LMS to more accurately evaluate the trade-offs between wildlife habitat and economics. Any model that converts forest attributes to wildlife metrics can be used by LMS to determine how management of individual forested stands affects the quality of wildlife habitat over entire landscapes.

Future Needs

Modeling the habitat availability of a host of wildlife species across a landscape and through time very rapidly, such as is possible with LMS, holds great promise for science-based ecosystem management. However, we need to improve understory modeling greatly to take full advantage of this promise. Available growth models predict forest overstory growth and response to silvicultural activities with good accuracy. However, models to predict how shrub cover, ground cover, or specific understory species respond to changes in forest structure are rudimentary or completely lacking. We suspect many of the best wildlife-habitat models will depend on understory attributes. For example, 53% of 15 HSI models for forest species currently require at least one understory attribute and 33% require more than one (USDI 1981). Modeling understory attributes requires detailed collection of current understory habitat attributes and accurate models that project understory changes through time. Recent improvements in predicting understory distribution and abundance have been made (McKenzie and Halpern 1999, McKenzie et al. 2000), but are scale-dependent. Additional advancements such as adding a “triggered treatments” module that could be set to harvest a stand when it meets certain criteria also would be useful. Similarly, one could specify other habitat features of interest, such as a desired level of canopy closure or dead and down woody material for wildlife benefits, and modeling could optimize these performance measures by providing cutting specifications.

Ideally, in the future, we will have validated models for many wildlife species that relate overstory and understory forest attributes to population viability. Even when such models are available, managers must be careful to monitor the response of species to landcover changes. Care must be given to confirm that wildlife actually uses the habitat that develops over time. Just confirming that the expected habitats develop as predicted is not enough. We need to confirm that wildlife is surviving and reproducing in the habitats as predicted.

Literature Cited

ALLEN, A.W. 1983. Habitat suitability index models: Southern red-backed vole (Western U.S.). USDI Fish Wildl. Serv. FWS/OBS-82/10.42. 14 p.

BAKER, PT, AND J.S. WIsoN. 2000. A quantitative technique for the identification of canopy stratification in tropical and temperate forests. For. Ecol. Manage. 127:77-86.

BET-riNGER, P., K. N. JOHNSON, AND J. SESSIONS. 1996. Forest planning in an Oregon case study: Defining the problem and attempting to meet goals with a spatial-analysis technique. Environ. Manage. 20:565-577.

BErrINGER, P., J. SESSIONS, AND K. BOSTON. 1997. Using Tabu search to schedule timber harvests subject to spatial wildlife goals for big game. Ecol. Model. 94:111-123.

BEVERS, M., AND J. HoF. 1999. Spatially optimizing wildlife habitat edge effects in forest management linear and mixed-integer programs. For. Sci. 45:249-258.

BoycE, S.G. 1980. Management of forests for optimal benefits (DYNASTOB). USDA Forest Serv. Res. Pap. SE-204. 92 p.

BOYCE, M.S., AND A. HANEY. 1997. Ecosystem management: applications for sustainable forest and wildlife resources. Yale University Press, New Haven, CT. 361 p.

BOYCE, S.G., AND W.H. McNAB. 1994. Management of forested landscapes: Simulations of three alternatives. J. For. 92:27-32.

CROOKSTON, N.L., AND A.R. STAGE. 1999. Percent canopy cover and stand structure statistics from the forest vegetation simulator. USDA For. Serv. Gen. Tech. Rep. RMRS-GTR-24.18 p.

EDmiNSTER, B., H.T. MowRER, T.M. MATHiAsEN, W.K. SCHULER, AND F.G. HAWKSWORTH. 1991. GENGYM: A variable density stand table projection system calibrated for mixed conifer and ponderosa pine stands in the Southwest. USDA For. Serv. Res. Pap. RM-297.

HOF, J., AND L. JOYCE. 1993. A mixed integer linear programming approach for spatially optimizing wildlife and timber in managed forest ecosystems. For. Sci. 39:816-834.

HoF, J., M. BEVERS, L. JOYCE, AND B. KENT. 1994. An integer programming approach for spatially and temporally optimizing wildlife populations. For. Sci. 40:177-191.

Ji, W., AND C. JEsKE. 2000. Spatial modeling of the geographic distribution of wildlife populations: A case study in the lower Mississippi River region. Ecol. Model. 132:95-104.

KING, J.E. 1966. Site index curves for Douglas-fir in the Pacific Northwest. Weyerhaeuser For. Pap. 8. Weyerhaeuser Company Res. Cent., Centralia, WA. 49 p.

KIRKMAN, R.L., J.A. EBERLY, W.R. PoRAT, AmD R.R. Trus. 1984. A process for integrating wildlife needs into forest management planning. P. 347350 in Wildlife 2000: Modeling habitat relationships of terrestrial vertebrates, Verner, J., M.L. Morrison, and C.J. Ralph (eds.). Univ. of Wisconsin Press, Madison, WI.

KEY, A.D., E.C. LoFRoTH, W.A. THOMPSON, S. BROWN, AND H. SCHREIER. 1999. Simulating and evaluating alternative resource-use strategies using GIS-based habitat suitability indices. Land. Urb. Plan. 45:163-175.

LOG LnPs. 2000. Log Lines log price reporting service, Mount Vernon, WA. Vol. 12, n. 11.

LOOMIS, J.B. 2000. Can environmental economic valuation techniques aid ecological economics and wildlife conservation? Wildl. Soc. Bull. 28:52-60.

McCARTER, J.B. 1997. Integrating forest inventory, growth and yield, and computer visualization into a landscape management system. Proc.: For. Veg. Simul. Conf. 373:159-167.

McCARTER, J.B., J.S. WION, P.J. BAKER, J.L. MOFFETT, AND C.D. OLiVER. 1998. Landscape management through integration of existing tools and emerging technologies. J. For. 96:17-23.

McGAuGHEY, R.J. 1997. Visualizing forest stand dynamics using the stand visualization system. Proc. Am. Soc. Photogram. Rem. Sens. 4:248-257.

McKENZIE, D., AND C.B. HALPERN. 1999. Modeling the distributions of shrub species in Pacific northwest forests. For. Ecol. Manage. 114:293-307.

McKENZIE, D., C.B. HALPERN, AND C.R. NELSON. 2000. Overstory influences on herb and shrub communities in mature forests of western Washington, USA. Can. J. For. Res.30:1655-1666.

MICHAELIS, L.O. 2000. Assuring a future for private timber management. P. 65-70 in Summit 2000 Washington Private Forests Forum. J. K. Agee [ed.]. Institute of For. Resour., Seattle, WA.

MOORE, C.T., M.J. CONROY, AND K. BOSTON. 2000. Forest management decisions for wildlife objectives: system resolution and optimality. Comput. Electron. Agric. 27:25-39.

MoRmoN, M.L., B.G. MARcoT, AND R.W. MANNAN. 1998. Wildlife-habitat relationships: Concepts and applications. The University of Wisconsin Press, Madison, Wl. 435 p.

RE,EL, R.S., P.C. ELKIE, A.R. RODGERS, AND M.J. GLUCK. 1997. Timber– management and natural-disturbance effects on moose habitat: landscape evaluation. J. Wild]. Manage. 61:517-524.

RUSHTON, S.P., P.W.W. LURz, R. FULLER, AND P.J. GARSON. 1997. Modelling the distribution of the red and grey squirrel at the landscape scale: a combined GIS and population dynamics approach. J. Appl. Ecol. 34:1137-1154.

SCHROEDER, R.L. 1983. Habitat suitability index models: pileated woodpecker. FWS/OBS-82/10,39. 26 p.

THOMAS, J.W., H. BLACK, JR., R.J. SCHERZINGER, AND R.J. PEDERSEN. 1979. Deer and elk. P. 104-127 in Wildlife habitats in managed forests-the Blue Mountains of Oregon and Washington. Thomas, JW. (ed.). USDA Handb. No. 533.

UNITED STATES DEPARTMENT OF AGRICULTURE. 1989. Stand examination program. Forest Service. Field Procedures Guide. Pacific Northwest Region. Olympia, WA.

UNITED STATES DEPARTMENT OF THE INTERIOR. 1980. Habitat suitability index model: Cooper’s hawk. Fish and Wildl. Serv. Review Copy. 8 p.

UNITED STATES DEPARTMENT OF THE INTERIOR. 1981. Standards for the development of habitat suitability index models for use in the habitat evaluation procedure. Fish and Wildl. Serv. Div. of Ecol. Serv. ESM 103, Washington, DC. 167 p.

John M. Marzluff-Fax: (206) 685-0790; E-mail:; Chadwick D. Oliver, Kevin R. Ceder, John Withey, James B. McCarter, C. L. Mason, and Jeffrey Comnick, College of Forest Resources, University of Washington, Box 352100, Seattle, WA 98195. Joshua J. Millspaugh, Department of Fisheries and Wildlife Sciences, University of Missouri, 302 A-BNR Building, Columbia, MO 65211.

Acknowledgments: Funding for our modeling efforts was provided bythe State and Private Forestry Initiative of the USDA Forest Service, Grays Harbor County, Grays Harbor County Public Development Authority, the U.W./W.S.U. Rural Technology Initiative, the U.W. Landscape Management Project, and Columbia-Pacific Resource Conservation and Development. Liana Aker and Curt Leigh of Washington Department of Fish and Wildlife assisted us in implementing HSI models at Satsop.

Manuscript received November 30, 2000. Accepted November 13, 2001.

Copyright Society of American Foresters May 2002

Provided by ProQuest Information and Learning Company. All rights Reserved