Explore the vast range of titles available.

Forest inventory attributes are an important source of information for a variety of strategic and tactical forest management purposes. However, it is not possible or feasible for field inventories to be conducted contiguously across large areas, especially at a resolution fine enough to be useful for operational management. Therefore, a large number of quantitative modeling and prediction methods have been and are being developed and applied to predict and map forest attributes, with the goal of providing an accurate, spatially continuous, and detailed information base for practitioners of forestry and ecosystem management. This article reviews the most commonly used prediction techniques in the context of a comprehensive modeling framework that includes a discussion of methods, data sources, variable selection, and model validation. The methods discussed include regression, nearest neighbor, artificial neural networks, decision trees, and ensembles such as random forest. No single technique is revealed as universally superior for predicting forest inventory attributes; the ideal approach depends on goals, available training and ancillary data, and the modeler's interest in tradeoffs between realism and statistical considerations. Useful ancillary data included in the models tend to include climate and topographic variables as well as vegetation indices derived from optical remote sensing systems such as Landsat. However, the use of airborne LiDAR in modeling of forest inventory attributes is increasing rapidly and shows promise for operational forest management applications. Different considerations are encapsulated within a generalized model development framework that provides a structure against which tradeoffs can be evaluated.

Although forest edges have been studied extensively as an important consequence of fragmentation, a unifying theory of edge influence has yet to be developed. Our objective was to take steps toward the development of such a theory by (1) synthesizing the current knowledge of patterns of forest structure and composition at anthropogenically created forest edges, (2) developing hypotheses about the magnitude and distance of edge influence that consider the ecological processes influencing these patterns, and (3) identifying needs for future research. We compiled data from 44 published studies on edge influence on forest structure and composition in boreal, temperate, and tropical forests. Abiotic and biotic gradients near created forest edges generate a set of primary responses to edge creation. Indirect effects from these primary responses and the original edge gradient perpetuate edge influence, leading to secondary responses. Further changes in vegetation affect the edge environment, resulting in ongoing edge dynamics. We suggest that the magnitude and distance of edge influence are a direct function of the contrast in structure and composition between adjacent communities on either side of the edge. Local factors such as climate, edge characteristics, stand attributes, and biotic factors affect patch contrast. Regional factors define the context within which to assess the ecological significance of edge influence (the degree to which the edge habitat differs from interior forest habitat). Our hypotheses will help predict edge influence on structure and composition in forested ecosystems, an important consideration for conservation. For future research on forest edges in fragmented landscapes, we encourage the testing of our hypotheses, the use of standardized methodology, complete descriptions of study sites, studies on other types of edges, synthesis of edge influence on different components of the ecosystem, and investigations of edges in a landscape context.

The effect of forest roads on species distribution and dispersal is an important conservation and management issue. We examined distributions of understory plants and their relationships to unpaved forest roads in a northern hardwood landscape in the Chequamegon National Forest, Wisconsin (U.S.A.). At six different sites, we recorded species cover, canopy cover, litter depth and cover, and bare ground at 11 distances (0, 5, 10, 15, 20, 30, 45, 60, 90, 120, and 150 m) from the road edge. At each of the 11 distances, we established a 60-m transect parallel to the road edge, within which we sampled 10 randomly placed 1 x 1 m plots (660 plots). We examined changes in species abundance (percent species cover per plot), richness, and Shannon-Wiener diversity (H′) with distance from the roads in an effort to determine the degree and magnitude of road effects on plant distribution. The species richness and H′ of native plants and the abundance of exotic species were clearly related to distance from the roads. Exotic species were most prevalent within 15 m of roads, occurring infrequently in the interior forest. The richness and H′ of native species were lower on the roadsides but reached interior-forest levels within a short distance (5 m) from the roads. The roads appeared to be associated with a disturbance corridor that affected site variables up to 15 m into the hardwood stands. At our six sites we detected 117 species, 25% of which occurred more frequently near the road, with only 12% having a 90% or greater preference for the forest interior. Our results suggest that roads have associated effects that alter interior-forest conditions and thus plant species composition and abundance; however, these effects are limited in depth of penetration into managed forests.

To be able to estimate the cumulative carbon budget at broader scales, it is essential to understand net ecosystem exchanges (NEE) of carbon and water in various ages and types of ecosystems. Using eddy-covariance (EC) in Douglas-fir dominated forests in the Wind River Valley, Washington, USA, we measured NEE of carbon, water, and energy from July through September in a 40-year-old stand (40YR) in 1998, a 20-year-old stand (20YR) in 1999, and a 450-year-old stand (450YR) during both years. All three stands were net carbon sinks during the dry, warm summers, with mean net daily accumulation of$-0.30\\ {\\rm g}\\ {\\rm C}\\ {\\rm m}^{-2}\\ {\\rm d}^{-1},-2.76\\ {\\rm g}\\ {\\rm C}\\ {\\rm m}^{-2}\\ {\\rm d}^{-1}$, and$-0.38\\ {\\rm g}\\ {\\rm C}\\ {\\rm m}^{-2}\\ {\\rm d}^{-1}$, respectively, in the 20YR, 40YR, and 450YR (average of 1998, 1999) stands; but for individual years, the 450YR stand was a carbon source in 1998$(0.51\\ {\\rm g}\\ {\\rm C}\\ {\\rm m}^{-2}\\ {\\rm d}^{-1})$and a sink in 1999$(-1.26\\ {\\rm g}\\ {\\rm C}\\ {\\rm m}^{-2}\\ {\\rm d}^{-1})$. The interannual differences for the summer months were apparent for cumulative carbon exchange at the 450YR stand, which had$46.9\\ {\\rm g}\\ {\\rm C}\\ {\\rm m}^{-2}$loss in 1998 and$115.9\\ {\\rm g}\\ {\\rm C}\\ {\\rm m}^{-2}$gain in 1999. As predicted, the 40YR stand assimilated the most carbon and lost the least amount of water to the atmosphere through evapo-transpiration.

Riparian zones are vital components of the landscape. Much attention has been focused on the question of how wide a buffer is needed to protect the original riparian environment. We sampled five streams 2-4 m wide and associated riparian ecosystems before and after clearcutting in western Washington. Buffers ranging from 17 to 72 m wide were left intact at all sites when harvesting. Our objectives were: (1) to characterize pre-harvest microclimatic gradients across riparian ecosystems, from the stream to the upland; (2) to identify effects of harvesting on these gradients; and (3) to describe effects of buffer width and near-stream microclimate on stream microclimate. Six weather stations measuring air temperature, soil temperature, surface air temperature, relative humidity, short-wave solar radiation, and wind speed were installed along transects running across the stream and into the upland, and two reference stations were established, one in an upland clearcut and one in an upland interior forest. Pairwise comparison tests were used to evaluate statistical differences between stations along transects for determination of gradient extent. Pre-harvest riparian gradients existed for all variables except solar radiation and wind speed, and values generally approached forest interior values within 31-62 m from the stream. After harvesting, microclimate values at the buffer edge and each subsequent location toward the upland began to approximate clearcut values instead of forest interior values, indicating an interruption or elimination of the stream-upland gradient. In addition, regression analyses showed that stream microclimate was affected to some degree by buffer width and microclimate in the surrounding area. We conclude that a buffer at least 45 m on each side of the stream is necessary to maintain a natural riparian microclimatic environment along the streams in our study, which were characterized by moderate to steep slopes, 70-80% overstory coverage (predominantly Douglas-fir and western hemlock), and a regional climate typified by hot, dry summers and mild, wet winters. This buffer width estimate is probably low, however, since it assumes that gradients stabilize within 30 m from the stream and that upslope edge effects extend no more than 15 m into the buffer (a low estimate based on other studies). Depending on the variable, required widths may extend up to 300 m, which is significantly greater than standard widths currently in use in the region (i.e., ∼10-90 m). Our results indicate that even some of the more conservative standard buffer widths may not be adequate for preserving an unaltered microclimate near some streams. Additional site-specific data are needed for different site conditions in order to determine whether generalizations can be made regarding near-stream microclimate.

We studied the relationships of landscape ecosystems to historical and contemporary fire regimes across 4.3 million hectares in northern lower Michigan (USA). Changes in fire regimes were documented by comparing historical fire rotations in different landscape ecosystems to those occurring between 1985 and 2000. Previously published data and a synthesis of the literature were used to identify six forest-replacement fire regime categories with fire rotations ranging from very short (<100 years) to very long (>1,000 years). We derived spatially-explicit estimates of the susceptibility of landscape ecosystems to fire disturbance using Landtype Association maps as initial units of investigation. Each Landtype Association polygon was assigned to a fire regime category based on associations of ecological factors known to influence fire regimes. Spatial statistics were used to interpolate fire points recorded by the General Land Office. Historical fire rotations were determined by calculating the area burned for each category of fire regime and dividing this area by fifteen (years) to estimate area burned per annum. Modern fire rotations were estimated using data on fire location and size obtained from federal and state agencies. Landtype Associations networked into fire regime categories exhibited differences in both historical and modern fire rotations. Historical rotations varied by 23-fold across all fire rotation categories, and modern forest fire rotations by 13-fold. Modern fire rotations were an order of magnitude longer than historical rotations. The magnitude of these changes has important implications for forest health and understanding of ecological processes in most of the fire rotation categories that we identified.

Land uses, especially harvesting and road building, are considered to be the primary cause of forest fragmentation in many parts of the world. To test this perception, we (1) quantified changes and rates of change in vegetative composition and structure within the Washburn Ranger District in northern Wisconsin using Landsat images, (2) examined changes in landscape structure, (3) assessed changes within the area of road influence (ARI), and (4) investigated changes in landscape composition and structure within the context of forest management activities. Our landscape classifications included six dominant cover types: mixed hardwood (MH), jack pine (JP), red pine (RP), mixed hardwood/conifer (MHC), non-forested bare ground (NFBG), and regenerating forest or shrub (RFS). Increases in NFBG and RFS, by 196% and 28% respectively, reflect expansion of the pine-barrens. Windthrow in the mature hardwoods during the late 1970s and jack pine budworm outbreaks during the mid-1990s correlated with decreases in those classes over the corresponding intervals. A 69% decrease in mean patch size and a 60% increase in edge density reflect increased fragmentation. An inverse relationship existed between the compositional trends of forested (excluding JP) cover types and RFS and NFBG cover types. ARI covered 8% of the landscape affecting species composition within the MH, RFS, and NFBG. Results from this study are key in assessing the links between management activities and ecological consequences and thereby facilitate adaptive management.

Understanding relative influences of ecological and anthropogenic factors on wildfire occurrence can assist decisionmakers in allocating fire management resources. We examined the influences of ecological and anthropogenic variables on probability of modern fire occurrence in the Mark Twain National Forest (MTNF), Missouri, using classification and regression tree (CART) and logistic regression analyses. Models were developed for five classes of fire size. Although CART distinguished some effects of fire size on results, logistic regression indicated a single model developed for all fires was sufficient for predictions. Ecological subsection was a dominating influence on fire occurrence for final CART and logistic models, highlighting the potential usefulness of ecosystem classification as a framework for considering factors influencing modern wildfires. Other influential predictors included ecosystem fire resistance; distance to roads, cities, and railroads; road density; mean October precipitation; elevation; median house value; and population density. Wildfires in the MTNF are caused overwhelmingly by arson, which, when combined with our results, suggests that arsonists may seek out flammable fuel types in remote areas with easy access. Within this general anthropogenic fire regime, we found a more subordinate effect of specific human variables (e.g., population density) on modern fire occurrence than did similar studies in the Upper Midwest, perhaps because our study area encompassed primarily federal forestlands with low population density.

We propose a working framework for future studies of net carbon exchange (NCE) in disturbed landscapes at broad spatial scales based on the central idea that landscape-level NCE is determined by the land mosaic, including its age structure. Within this framework, we argue that the area-of-edge-influence (AEI), which is prevalent in many disturbed, fragmented landscapes, should constitute a distinct ecosystem type since numerous studies have indicated unique ecological properties within these areas. We present and justify four working hypotheses currently being tested in northern Wisconsin, based on this framework: (1) the area of an ecosystem that is influenced by structural edges (e.g., AEI) has NCE that is significantly different from the ecosystem interior; (2) age structure and composition of an ecosystem play critical roles in determining the ecosystem's contribution to the cumulative net ecosystem production (NEP) of the landscape mosaic; (3) the relative importance of different structural and biophysical controls of carbon exchange is ecosystem dependent; and (4) the frequency and intensity of disturbances in time and space control the cumulative NCE of the land mosaic through alteration of ecosystems that vary in age, structure, physical environment, and interactions. In addition, we describe five different research approaches to quantify NCE at broad scales, including biometric estimations, ecophysiological methods, micrometeorological methods, applications of remote sensing and GIS, and ecosystem models.