Biomass Flow in Western Forests: Simulating the Effects of Fuel Reduction and Presettlement Restoration Treatments

and Presettlement Restoration Treatments Biomass Flow Western Forests in • Demonstrate the approach by applying the modeling framework to selected stands in the Sierra Nevada and central Rockies. The focus of this article is on results of the simulation with regard to fire resilience, stand structure development, and biomass flow under uneven-aged management regimes that will most likely be used to maintain desired ecosystem attributes in short-interval fire–adapted western forests. Defining Stands in Need of Treatment Two factors must be considered when determining whether a stand is at risk from an unnatural fire event and therefore suitable for thinning and biomass removal: • The stand’s potential to propagate stand-replacing fires that kill all trees and mineralize most of the organic soil material. This mainly depends on the available fuel load. • The frequency of successful ignition. This depends on the quality of the fuel load and also on environmental conditions (e.g., precipitation and wind). Neither the fuel load nor the ignition frequency alone is sufficient to define unnatural fire regimes. For example, boreal forests often have high fuel loads and chaparrals experience frequent ignitions, but these fire regimes are not unnatural (Keeley et al. 1999). However, many forests, including midelevation forests of the Rockies and Sierra Nevada, were historically subject to fires with a return interval in the order of decades. These low-intensity fires regularly eliminated ground, surface, and ladder fuels; partially reduced crown fuels; and, by killing seedlings, kept stand densities low. In the Sierra Nevada and Southwest, these return intervals have increased by an order of magnitude (SNEP 1996; Moore et al. 1999). Current fuel loads and tree densities create fire intensities comparable to boreal forests (Alexander 1982), where survival of trees is unlikely. Total mortality over large areas was historically a rare fire pattern in ponderosa pine and mixed conifer forests (Moore et al. 1999). Defining Fire Resilience Fire resilience can be defined as a stand’s ability to survive fires without permanent loss of functional or structural elements. The upper canopy with the oldest and largest trees represents such a structural element. A stand can be considered fire resilient if the probability of a complete loss of the upper canopy is reasonably low. This is the case if (1) propagation of a fire within the upper canopy and (2) continuous lethal scorching of canopy trees from crowning surface fires is avoided. Both of these spread paths can be largely eliminated by reducing the fuel density below a critical threshold (for a detailed discussion, see Agee 1996). Experiences from real fires and simulations (e.g., van Wagtendonk 1996) suggest that this threshold can be approximated using a canopy closure value of 40 percent. Therefore, in our modeling we use 40 percent canopy closure as our target for fire resilience (fig. 1). 13 October 2001 • Journal of Forestry Figure 1. Stand structures and fire spread. Any tree whose crown intersects with a fire-spread path is assumed killed. Solid thick arrows show spread paths affecting upper canopy trees. Only the structures in case V and case VIII reliably prevent continuous loss of upper canopy by keeping the fuel density subcritical in the canopy and the ladder layer. Critical fire spread path Noncritical fire spread path Irrelevant fire spread path Silvicultural Concept The need to address the vertical dimension of the stand when restoring and maintaining short-interval fire– adapted western forests makes evenaged management approaches inappropriate. The historic stand structure of the forests of interest was not evenaged. Frequent fire-return intervals under historic fire regimes created forests of many age classes with a diverse canopy structure and spatial distribution of trees. We therefore chose an individual tree selection model (Alexander and Edminster 1977) to define desired stand structure because it is easy to simulate and well suited to handle a continuum of tree sizes. The model controls stocking across defined upperand lower-diameter classes (hereafter called dbhQmin and dbhQmax) using a negative exponential Q ratio defined as the number of trees in one diameter class divided by the number of trees in the next larger class. The Q ratio and desired basal area stocking define the numbers of trees to be left in each diameter class. We discovered that maintaining a 40 percent canopy cover target and a diameter range of ≥ 40 inches (realistic in the Sierra Nevada) resulted in extremely flat diameter class distributions; i.e., the Q ratio was close to 1, or required extremely open (low basal area) stands. To avoid this, we created a hybrid management scheme for Sierra Nevada forests that could accommodate the larger trees needed to meet desired presettlement conditions. In our hybrid scheme, trees with less than a target dbhQmax (usually 30 inches) are managed under the uneven-aged individual tree selection model. Of the trees that exceed dbhQmax in the Q stocking curve, a fixed percentage termed “large tree removal intensity” (LTRI) is harvested in every selection cutting cycle such that (1 – LTRI) percent of the trees with dbh > dbhQmax remain on site. This hybrid approach facilitates the desired retention of large trees (Covington et al. 1997) in presettlement restoration and allows evaluation of the long-term effects of retaining large trees on stand structure and yield. Recent work by Graham et al. (1999) shows that type of thinning has a crucial influence on fire resilience. The uneven-aged regime modeled here results in the desirable conditions of low crown-fuel bulk densities but high crown bases. Simulation Tools We chose the USDA Forest Service Forest Vegetation Simulator (FVS) (Teck et al. 1996, 1997) as the primary tool for simulating the effect of our hybrid management model on typical mid-elevation stands in the Sierra Nevada, the central Rockies, and the Southwest. We used the WS variant of FVS for Sierra Nevada data and the CR variant for central Rockies and Southwest data. Using FVS, however, poses the challenge that the WS and CR variants do not simulate natural regeneration. Because small trees represent a source of surface and ladder fuels, regeneration must be included. Our solution to this was to assume that enough trees regenerated following each cutting cycle to fill the smallest diameter class of the Q distribution. Because the uneven-aged selection automatically eliminated surplus trees in the next cutting cycle, a greater number of trees regenerated at the time of treatment has no effect on long-term stand structure. Regeneration density is irrelevant to fire-resilience considerations provided the small trees have not grown rapidly enough to provide fuel ladders into the canopy (unlikely with 14 Journal of Forestry • October 2001 Table 1. Data set information for Sierra Nevada, Coconino, and Manitou stands at the beginning of the simulation period. Trees Basal area Volume Accretion Mortality Data set, (stems (square feet Dmax (cubic feet (cubic feet per (cubic feet per stand type per acre) per acre) (inches) per acre) acre per year) acre per year)