Controls and reference conditions in forestry: The role of old-growth and retrospective studies

mental evidence obtained by a 5-year experiment with harvesting in a cornfield takes centuries to obtain from a forest. Most foresters have careers that last only 30–40 years. Changes may be occurring in productivity too slowly to be detected by any one generation of foresters. Periodic comparison of long-term controls and harvested forests can show unequivocally whether forest management activities are maintaining productivity and other important attributes over time. For example, old-growth forest remnants often are used and paired to secondgrowth stands that originated under similar conditions. At some time in the past the stands were similar, but the second-growth stands were harvested and the old-growth remnants were not. Thus, one may jump several decades into an experiment that was inadvertently started at the time of settlement by Europeans. Short-term controls are used to test hypotheses about the effects of specific silvicultural treatments. In this setting, secondgrowth stands may be divided into control stands that will not be harvested and experimental stands that will receive a harvest treatment hypothesized to move stands more rapidly toward a desired reference condition. The effectiveness of silvicultural treatments patterned after natural disturbance processes can be evaluated in this manner. Note that the same old-growth stands used as long-term controls discussed in the previous paragraph also may serve here as reference stands, and the objective of the experiment would be to see which silvicultural treatment creates the reference condition faster or more efficiently. Natural disturbance processes are entirely compatible with the use of long-term controls, because the goal is to assess whether harvesting does as good a job as natural disturbance at maintaining productivity and species richness over time. However, natural disturbances pose a major interference in the use of short-term controls, necessitating the use of multiple replicates to ensure that some persist without the confounding factors of additional disturbances. Concerns may be raised about the value of control and reference stands with regard to global-scale environmental changes. If the two sets of stands are in the same region, however, they will generally experience similar magnitude and rate of climate change, nitrogen deposition, acid rain, and other regional human influences. For example, much of the boreal forest of central North America has experienced similar changes in fire frequency caused by climate change over the last century, regardless of nearness to human activities (Johnson 1992). Harvesting therefore remains the major difference, leading to a valid control. The point of the control is to determine whether harvested stands differ from stands with natural disturbance over time, not whether stands remain the same as some presettlement condition. Although global-scale factors generally influence control, reference, and managed stands in the same way within one region, other confounding factors may vary locally. In the United States, overabundance of deer and exotic species invasions are having major impacts in some forests, and their impacts are likely to be more pronounced in managed stands than reference stands, which are often in more remote locations (Vitousek 1990, Côté et al. 2004). Thus, although in general harvesting remains the major difference between managed stands and controls or references, other factors sometimes must be considered. Reference conditions also are used in two different senses. In this case the main difference is spatial scale, namely, reference stands and reference landscapes. For reference stands, attributes such as tree size-class distribution, composition, gaps, and coarse woody debris are measured and compared with second-growth stands. Reference landscape conditions refer to the proportion of stands in various vegetation growth stages, which in turn is a function of the disturbance regime. The types of stand-leveling disturbances determine which species regenerate in young stands, on what successional trajectory the stands will progress, and what proportions of stands will progress along various trajectories. The frequency of standleveling disturbances determines the distribution of stand ages and the relative abundance of young and old stands. Often, disturbance regimes are complex, and there are many successional and developmental stages of forest present on a landscape with intricate relationships among them. Retrospective studies are necessary to examine these “successional webs.” At the landscape scale, if the size, frequency, and intensity of each natural disturbance regime can be reconstructed, then range of natural variation (RNV) estimates can be established for proportions of stands in various stages of succession and development or growth stages (Frelich 2002). Successional stages differ in composition over time, whereas stand development connotes change in structure over time. Growth stages are defined as a combination of succession and development or composition and structure. Retrospective studies of disturbance regimes can be based on a variety of data sources, including unlogged forest remnants, historical records, or paleoecological evidence. These tools can be used individually or in combination to determine the proportion of stands in each growth stage within the landscape. User-friendly simulation models (e.g., the Vegetation Dynamics Development Tool; Beukema et al. 2003) are available for determining a target proportion of forest in each successional stage.