Modeling Regeneration Outcomes
Regeneration of a forest stand, in a silvicultural context, occurs when the harvest of trees, in one cutting or a sequence of cuttings, provides conditions for the establishment of a new cohort, or age class, and its subsequent development into canopy positions, at least for those species capable of attaining canopy status. Other arborescent species may regenerate as well, but may be species that characteristically achieve only midstory or understory status. Put another way, successful regeneration methods provide for timely and sufficient release (i.e. a level of release sufficient for even the most shade-intolerant species) of regeneration sources already present and suitable growing conditions for those arriving soon after the disturbance.
Forest managers require the ability to predict the outcome of their management activities, among them, regeneration cuts designed to secure natural regeneration. Predicting these regeneration outcomes in forests that have many species that form a variety of associations across environmental gradients is challenging.
The results of regeneration studies conducted by scientists at Bent Creek Experimental Forest are consistent with the Initial Floristic Composition concept. When extended to forest management activities, the concept suggests that after a substantial overstory removal (i.e., a level of release sufficient for even the most shade-intolerant species), species composition that develops is comprised of one or more of the following sources of regeneration: (1) new seedlings that become established during or soon after cutting; (2) advance reproduction, (i.e. seedlings and saplings that exist in mature stands at the time of harvest); and (3) stump and/or root sprouts from trees removed during the harvest.
Consequently, if we could enumerate regeneration sources present prior to cutting, and somehow predict the arrival of new regeneration sources during or after cutting, we would have some idea about the pool of regeneration sources that would compete for dominance in the new cohort. If, in addition, we had information regarding the competitive ability of the various individuals in the pool of regeneration sources, we could predict species composition at some point in time after cutting(s).
A model for forest regeneration – named REGEN – was formulated to apply the concepts of described above to predict regeneration outcomes following stand-replacing disturbance in upland hardwood forests in the southeastern United States (Boucugnani 2005). The model is driven by a pre-treatment inventory of all existing regeneration sources enumerated by species and size class. Based on probabilities, the model adds sprouts as well as seedlings and root suckers to the regeneration plots. REGEN picks the dominants/co-dominant trees on each regeneration plot at crown closure based on a ranking of the post-harvest performance of different regeneration sources which include new seedlings, various sizes of advance reproduction, and stump sprouts. Probabilities and rankings are provided in the form of “knowledge bases” which comprise lists of parameters that dictate the relative likelihood of a particular regeneration source being successful.
In addition to the stand-alone version of REGEN, we are working with the USDA Forest Service’s, Forest Management Service Center to link REGEN to the Southern variant of the Forest Vegetation Simulator (FVS). FVS is the USDA Forest Service’s nationally supported growth and yield modeling system that is used to forecast stand development with and without management or other disturbance events. Each FVS variant is a distance independent, individual tree growth model that has the capability of including silvicultural, fire, and insect and disease impacts on forest stands. Users are able to track outputs of individual tree characteristics and stand characteristics such as density, volume, wildlife habitat, fire and health related indices, and carbon stocks.
A known constraint of the Southern variant of FVS is the regeneration model. In FVS terminology, the regeneration model is a partial establishment model, meaning only sprouts are estimated when a tree is cut or killed by fire. All other regeneration must be entered by the user.
Forestry models provide land managers a means to assess potential effects of alternative treatments in forested stands and are especially useful when site-specific information regarding the potential effects of a treatment is lacking. Using FVS alone to predict regeneration outcomes is not acceptable, given the inability of the Southern variant of FVS to sufficiently predict regeneration success following overstory removal. Alternatively, using REGEN alone does not allow for the modeling of the effect of intermediate treatments (e.g. prescribed burn) on the regeneration pool or predict the long-term stand development patterns and tradeoffs under alternative management scenarios. By combining the two models, we were able to diminish weaknesses in both models, thus allowing for multi-scale comparisons of treatment alternatives.
Stage (1973) noted in his first publication on Prognosis, the predecessor of FVS, that our ecological and silvicultural knowledge is incomplete and as such our forestry models are incomplete. With this in mind it is important that forestry models allow land managers the ability to adjust model relationships as needed. FVS is flexible with respect to modifying regeneration inputs. It is also essential that forestry models are consistently maintained to facilitate the incorporation of new ecological findings; such is the case with FVS. The full integration of REGEN within FVS will provide resource practitioners with a valuable tool that can be used to assess the relative tradeoffs of varying management scenarios in terms of a variety of stand metrics, including stand density, volume, species composition, wildlife habitat, fire and health related indices, and carbon stocks.
Validation of REGEN rankings is currently underway for knowledge bases specific to the southern Appalachian Mountains. Preliminary and final results will be updated as they become available.
Research Principal Investigators
Dr. Tara L. Keyser, Research Forester, Southern Research Station – RWU-4157, Asheville, North Carolina
Dr. David L. Loftis, Research Forester (emeritus), Southern Research Station – RWU-4157, Asheville, North Carolina
Dr. Phil Radtke, Associate Professor, Virginia Tech University, Blacksburg, Virginia
David B. Boucugnai, Computer Programmer, Georgia
USDA Forest Service, Forest Management Service Center-Forest Vegetation Simulator, Fort Collins, Colorado. Contacts include: Chad Keyser and Mike Van Dyck.