Landscape Silviculture for Late- Successional Reserve Management

The effects of different combinations of multiple, variable-intensity silvicultural treatments on fire and habitat management objectives were evaluated for a ±6,000 ha forest reserve using simulation models and optimization techniques. Our methods help identify areas within the reserve where opportunities exist to minimize conflict between the dual landscape objectives. Results suggest that most of the trees removed by silvicultural treatments designed to support fire and habitat objectives, while generating enough revenue to break-even, would be medium-sized (17-40 cm), shade-tolerant conifers. The study produced information that was used by a planning team on the Gifford Pinchot National Forest to develop stand-level treatments based on mid-scale landscape patterns. New contracting authorities give the Forest Service ways to offer sales that support landscape management objectives in the reserve, but the contracts are time-consuming to prepare and award. Implementation of a stewardship contract associated with the study reserve is scheduled to begin in summer 2006. Introduction The Northwest Forest Plan (Plan) designated late-successional forest reserves on some federal lands in Oregon, Washington, and California. One goal of the reserve network is to sustain habitat for the northern spotted owl (Strix occidentalis caurina) and other species associated with older, late seral forests (USDA and USDI 1994). Plan guidelines require land managers to protect these reserves, or LSR, from largescale natural and human disturbances. An ongoing challenge for managers of LSR in drier Plan provinces is to conserve and develop older forests in ways that support their resilience in fire-adapted ecosystems. In such places, LSR managers are concerned about effects to late seral forest habitat structures associated both with severe wildfires and with silvicultural treatments done to reduce fire severity. The problem of potentially conflicting effects from forest management is not confined to LSR, however. Throughout fire-adapted forest ecosystems of the western US, federal land managers seek ways to promote forest structures and processes that are consistent with pre-fire exclusion conditions, while preserving some of the attributes of existing conditions that people have come to value. Our interest lies in developing methods to quantify tradeoffs among various forest management objectives. A need for analytical methods of this type occurs everywhere that multiple resource objectives exist but is acute in the West, because of extensive areas of public land combined with an increasing human population. 1 A version of this paper was presented at the National Silviculture Workshop, June 6-10, 2005, Tahoe City, California. 2 Research Forester and Program Manager, respectively, Pacific Northwest Research Station, USDA Forest Service, P.O. Box 3890, Portland, OR 97208. Silvicultural Options—Landscape Silviculture—Hummel and Barbour USDA Forest Service Gen. Tech. Rep. PSW-GTR-203. 2007. 158 Many people are interested in land management issues. At times, their preferences and the best ways to provide them appear to be in conflict. In this paper, our objective is to describe a method for identifying silvicultural solutions to potentially conflicting landscape management objectives. We summarize results from a study in which we investigated how treatments to moderate fire behavior could impact late seral forest structure in one LSR, if treatment expenses might be offset by revenue generated from harvest activities, and the dimensions of the trees removed. We use the term “landscape silviculture” for treatments applied to a stand but evaluated collectively according to objectives for an entire reserve. Site Description The Gotchen LSR lies on the eastern flank of the Cascade Range in Washington State, covering about 6,070 ha of the Mount Adams Ranger District on the Gifford Pinchot National Forest. Like many other reserves in the drier provinces of the Plan area, the Gotchen LSR includes a mix of older, mixed-conifer forests and plantations. Tree species include Douglas-fir (Pseudotsuga menziesii), grand fir (Abies grandis), subalpine fir (Abies lasiocarpa), ponderosa pine (Pinus ponderosa), western larch (Larix occidentalis), and lodgepole pine (Pinus contorta). Six documented spotted owl nest sites exist (Mendez-Treneman 2002). Defoliation of true firs (Abies) associated with an outbreak of western spruce budworm is contributing to increasing fuel loads and to declining crown cover, which affects owl habitat quality (Hummel and Agee 2003). Managers responsible for the Gotchen LSR seek ways to moderate ongoing risks to owl habitat associated with potential stand-replacement severity fire, while retaining older forest structures within the reserve landscape (Hummel and Holmson 2003). Methods and Analysis Characterize Landscape Conditions At the outset of the Gotchen LSR study, we considered it important to use a simulation model that could recognize the contribution of individual trees to forest structure at both within-stand and among-stand (landscape) scales. We needed a model that could track residual stand structure and forest dynamics following a silvicultural treatment, and account for any trees cut during the treatment both by size and species. In addition, because wildfire can affect multiple stands, we wanted the model to have spatial database capabilities, so that the influence of conditions in neighboring stands on fire behavior and effects within a stand (and vice-versa) could be simulated. We ultimately selected the Forest Vegetation Simulator East Cascades variant (FVS) (Stage 1973, Johnson 1990, Crookston and Havis 2002). We began by using aerial resource photos to identify vegetation patches in the Gotchen LSR, and then spatially described them in a geographic information system (GIS) database (fig. 1). The patches were stratified into a summary matrix of stand types based on structure class and potential vegetation type (details in Hummel et al. 2001). By selecting patches within stand types using probability proportional to size, field samples made in 2000 and 2001 covered the range of existing conditions. The data were used to create “tree lists” for sampled patches following FVS procedures (Dixon 2003). We randomly assigned a FVS tree list to any unsampled patches Silvicultural Options—Landscape Silviculture—Hummel and Barbour USDA Forest Service Gen. Tech. Rep. PSW-GTR-203. 2007. 159 within the same stand type, our assumption being that within-stratum variation in forest structure is lower than among-strata variation. Figure 1−Landscape patches in the Gotchen LSR identified by using aerial photographs. Each colored patch is a different combination of forest structural stage and vegetation cover type (Source: Hummel et al. 2001). We considered it vital that vegetation patterns be able to change with time in our analysis and not be constrained by existing landscape geometry. Some of the large patches in the southern part of the Gotchen LSR, for example, result from previous logging and fire suppression activities, and tend to be bigger than regional studies of disturbance ecology would suggest for areas like this one with mixed-severity fire regimes. We therefore introduced the ability for new patterns to emerge by substratifying the original patches into smaller “projection units.” These units represent the smallest area to which a treatment could be applied. Each unit received the FVS tree list associated with its original patch, but individual unit growth trajectories could differ based on stochastic variation within the model, and on the treatment schedules ultimately selected for each one (details in Hummel et al. 2002, Calkin et al. 2005). Use Forest Structure to Describe Fire and Habitat Objectives Once we had a database representing existing forest vegetation, we turned our attention to how its structural dynamics related to fire and to owl habitat. We focus on structure, or the arrangement and variety of living and dead forest vegetation, because it can be measured and it is physically and biologically relevant to fire behavior and to owl habitat. We selected a 30-year analysis period by considering both fire return intervals for mixed-conifer forests in the region and model capabilities. For fire threat (FT), we used three variables: flame length, crown fire Silvicultural Options—Landscape Silviculture—Hummel and Barbour USDA Forest Service Gen. Tech. Rep. PSW-GTR-203. 2007. 160 initiation, and crown fire spread, which we estimated for each unit using the Fire and Fuels Extension to FVS (FVS-FFE) (Reinhardt and Crookston 2003). A unit’s FT index (low, moderate, or high) was a weighted combination of these variables within a unit and its adjacent units (details in Calkin et al. 2005): Threat10i=FLi+w1*(Torchi+Crowni)+(w2*∑j[Edgej*{Torchj+Crownj}] / ∑jEdgej)