A forest is an exceedingly complex biological unit. It comprises not only a more or less diversified aggregation of trees, but numerous species of shrubby and herbaceous plants, fungi, insects, herbivorous animals and a complex soilfauna andflora. In other words, it consists of a very large number of mutually interacting organisms which are affected by, and themselves affect, a complex of environmental factors. Those words could have been written today, but they come from a 1929 report on forestry research coauthored by I.W. Bailey of Harvard University and H.A. Spoehr of the Carnegie Institution. This insightful report goes on to describe the fact that, while much of forest management depends on the modification of the forest through treatment of the forest vegetation itself, the manager must be skillful because: ... such gross treatments have highly diversified and far-reaching effects upon the biology of the forest, not only upon the soil and the trees but also upon the minor vegetation, insects, fungi, and other elements of the complex. The latter effects cannot safely be ignored since they in turn may later profoundly influence the future growth of the forest. Today, three professional generations after Bailey and Spoehr's advice, the attention of forest managers is focused on how best to do the things they do in a forest so that when they "profoundly influence" its future, that future is healthy and sustainable. What today's managers enjoy (that Bailey's generation lacked) is a better basis for understanding forest ecosystems and how they function, as well as vastly more sophisticated tools to model and predict what kinds of forest responses may result from a given action. That understanding includes better methods to evaluate conditions over large areas so that individual actions can be placed in the context of landscape or regional effects, and better methods to portray what will happen over longer time periods, so that people can see the long-term effects of a particular management action. These are enormously complex systems, so saying that people today "know more" is not meant to imply that they know it all, or that they know enough. Most experts still advise addressing forest management questions as an exercise in experimental design, designing each action on the basis of the best hypothesis available, but doing enough monitoring so that one learns from the outcome. (left) Wetlands are an improtant component of forests, and retaining wetland quality is one of the management objectives on this northeastern forest. These new tools, however, applied in conjunction with basic ecosystem management concepts, can help people better realize how short-term changes to organisms or populations (cutting some trees or affecting vegetation with a prescribed fire program) may affect forests, landscapes, and even larger areas over longer time periods. People are often repelled by the sight of a cougar killing a deer, or a logger cutting down a tree. Those actions, lethal in the immediate, local sense, may be contributing to the long-term integrity of larger landscapes. On the other hand, if they are done wrong or taken to extremes, those same actions may be destroying the long-term integrity of the forest. Understanding the differences involved is a critical factor in achieving forest health treatment that contributes to a sustainable forest. One example of modem capability is the availability of computer models based on expert systems. These models allow a person to simulate the likely effects of a management action or disturbance into the future. Graphics drawn from the Landscape Management System, developed by the University of Washington in cooperation with the Forest Service. This program, which is available free of charge on the World Wide Web at https://silvae.cfr.washington.edu, can produce images that illustrate not just how a forest system may change, but how it will appear, in the future. It is available for only a few forest types however, and a great deal of data about forest growth and successional processes is required to adapt it to new forest conditions. Fortunately, many people are at work developing such tools and they are rapidly becoming available. With tools like these, forest managers can more accurately predict the outcome of actions. For a skeptical public, who want to see productive forests but lack the expertise to envision long-term effects, the programs provide a way to gain information and confidence about different management approaches. The other approach that is gaining increasing usage is the concept of risk assessment. None of us would buy a stock offering, drive a car, or mortgage a home without thinking of the risks involved. While we aren't hoping for a market crash, car wreck, or financial reversal, we realize that those can happen. So we evaluate the level of risk that we are willing to accept, and pay for some kind of insurance or other safeguard to cover the unlikely, but catastrophic, events we wish to avoid. Forest managers today have the same opportunity. The growing capability of aerial imagery, computer modeling, and ecosystem understanding allows managers to identify places where risks are highest and explore courses of action that are likely to reduce risk. One such example was recently demonstrated in a wildfire hazard/risk modeling exercise conducted in the State of Colorado (Sampson et al. forthcoming). This exercise used a 10-year fire ignition history---combined with satellite imagery of vegetation and data on soil types, climate, elevation, and hydrology---to identify the wildfire hazard by watershed area. The probability of a future ignition that could grow into an uncontrollable wildfire is reflected by past records that show where ignition history is highest and vegetation most likely to support an intense wildfire. Knowing the areas most at risk from large wildfires does nothing to predict events in the near future, but it allows managers to take a closer look for areas where treatment might be the most effective at reducing hazards. A modern geographic information system (GIS) allows the user to locate past events on the landscape, and combine them with the current conditions that might affect fire behavior. In addition, the GIS allows the study of "distributed risk," or the risk one area faces because of the conditions in an adjoining area. Thus, if one place in a watershed faces a very high ignition risk, and the entire watershed is covered with flammable vegetation, the chances of the entire watershed being affected are very high---even if the ignition risk is isolated in one small area. Using soil, slope, vegetation, and fire effects information, the model also illustrated areas where soils might be most susceptible to development of hydrophobic (water repel-lent) conditions if they were subjected to intense wildfire. Hydrophobic layers can be created when the heat of a fire volatilizes organic compounds from the vegetation and drives some of them down into the soil. As the heated compounds move down through soil pores, the soil cools them until they condense, leaving a waxy organic residue. In soils with limited pore space (mainly coarse-textured soils), these compounds can seal the soil, creating a water barrier that lasts for a year or more until the compound breaks down or new roots penetrate and open up the layer. If rainfall or snowmelt occurs before the hydrophobic layer is broken down, the top layers of soil will become saturated, then start to flow downhill under the pressure of the excess water. The result can be damaging soil erosion, coupled with sediment and debris flows that affect stream channels and reservoirs far downstream. While tools such as this offer increasing insight into the risks inherent in current or predicted forest conditions, it is important to note that major gaps in our knowledge still exist, as well as major limitations in the feasibility of treating certain forest conditions or restoring forests after some kinds of damage. Any risk assessment, no matter how well conceived, can be thrown into immediate disarray by an unforseen natural disturbance. (The Colorado wildfire modeling exercise, for example, estimated that the forests in the nortlicentral watersheds faced only a low hazard. The 1997 blowdown of 20,000 acres of spruce in an unusual windstorm may have significantly changed that situation, at least in one place. Events like this are known to occur historically, but the chances of accurately assessing the risk-or managing it if one could assess it-are minimal.) Another important factor is the inherent limits of different ecosystems. A pine forest in the southeast may, after harvest and replanting, be a vigorous young forest again within two-three years. On a dry, sunbaked south slope in the Inland West, a similar clearcut pine forest may be virtually impossible to restore, as the altered microclimate becomes too harsh for seedlings to endure. Similarly, a wildfire that depletes soil nutrients and organic matter may, on a marginal site, deplete the soil to the extent that reforestation may not occur for generations, if ever. Where the soil starts into a downward spiral of erosion and degradation, the site is more likely to become desert-like in the future than it is to return to its original condition. Once that process begins, the chances of reversing it and restoring the site are pretty low in most places, even with heavy investments of expertise and money. The limits we face may be lack of knowledge, or simply the fact that the forces of nature in some places are overwhelming. Whatever the case, it is well to remember that limits exist when the growing confidence of scientists and managers begins to sound as if all such obstacles had been conquered. Much forest management in the past was focused on producing useable supplies of wood, forage, or wildlife from the forest. Early exploitation took advantage of established forests with little or no regard for the future. Those actions reflected largely local, short-term considerations. As conservation pressures increased throughout the 20th Century, people initiated major efforts to restore damaged forests through tree planting and fire protection. In the management of existing forests, conservation concerns led first to the concept of sustained yields, then to the more recent focus on sustainable forest systems. People's thinking about forests and their future has become increasingly longer-term. It is the longest-term framework---sustainable forests---that shapes today's forest health treatment efforts. The problems addressed are complex, and usually tied to a particular place and the conditions that exist there. Standardized management approaches are seldom the most useful, and any approach can become a problem when it is applied in the wrong place or in the wrong way. There are, however, some general factors that seem to be common in addressing forest health questions. Often, these are similar to the factors cited in recent attempts to define ecosystem management. They include:
- Maintenance of a plant community (in terms of composition, size, arrangement and density) that is suited to the environmental conditions (soil, microclimate) on the site.
- Protection of basic soil quality and productivity.
- Protection of genetic and biological diversity within the forest.
- Management to achieve an array of structural conditions across time and space.
- Maintenance of essential ecosystem processes such as fire, nutrient cycling, carbon cycling, and water partitioning within the system.
- Consideration of cumulative effects over time as well as landscape and regional impacts.
- The need to create disturbances where needed to mimic missing events or features in the system (such as a lack of fire, or missing predators).
- The need for a good monitoring system to provide the information needed to effectively adapt management and treatment to changing conditions within the forest.