Invited Speakers

Loblolly pine is an important commercial species that is highly responsive to intensive management practices and is grown for a wide variety of wood products. Intensively-managed plantations of loblolly are replacing natural stands as the primary source of pine raw material in the southeastern United States. Sustainable management of these planted stands requires that foresters consider not only wood production but biodiversity, wildlife habitat, water quality, scenic beauty, and other values as well. Growth and yield models that produce reliable estimates of stand production and stand structure for a wide range of management treatments are essential to evaluating the effectiveness of intensive plantation culture for achieving a range of resource goals. Data from permanent growth plots in operationally-established loblolly pine plantations and results from silvicultural research with the species have been used to develop growth and yield models that provide quantitative estimates for evaluating a host of sustainability measures.

A research programme aiming at the development of new systems for forest management analysis and planning was recently initiated at the Forest Faculty, SLU. Today, forest management in Sweden, as in many other places, is the management of multiple resources such as biodiversity, recreation, hunting, carbon sequestration and rein deer herding. Sustainability is no longer related only to timber production by embraces the composition, processes, and functions of the entire ecosystem. Moreover, the pressure from substitute products requires that the production of timber, wood fuel, etc, is made more competitive. The increased complexity of forest management has created a surge for improved instruments for analyses and decision support.

As interactions between processes, functions and different uses are obvious in forest ecosystems, integrated analyses and planning procedures are essential. Research efforts in different areas have also to be gathered in a common framework for the development of necessary models and methods. The Heureka research programme aims at such an integrated effort for the development of forest resource management systems. The new systems should be designed (i) with a modular structure to allow for the development of different applications, (ii) with the landscape as the basic planning unit, (iii) with the tree as the basic unit of projection of the tree layer, (iv) with models for processes interacting with the management of the tree layer, and (v) with models for risk and uncertainty of data and of model projections.

In the Pacific Northwest, the site curves commonly used for plantations of Douglas fir are based on natural stand data. It has been observed that early height growth on plantations often exceeds the best growth in natural stands, and that observed growth patterns often fail to follow the natural stand site curves. Furthermore, it is now documented that high planting densities promote faster early growth in Douglas fir plantations, both in diameter and in height. Accordingly, new site curves based on plantation data, and including a density effect, were sought. A good plantation data base from permanent plots was available, encompassing a wide range of density regimes and sites.

Height-age curves were developed which could be described as polymorphic, variable-asymptote, base-age invariant. The variables are top height and total age from seed. The height growth for any one-year period is modified with a multiplier term; that term is a function of age and current density (trees per hectare). At any given age, there is some density where the multiplier is at a maximum; lesser growth is predicted at higher and lower densities.

Growth models at different levels of resolution have been developing essentially independently. Their relative advantages and limitations are often misunderstood and controversial. I will discuss some aspects of their use and the potential for cross-fertilization, with emphasis on a stand-level approximation to the TASS tree-level model.

Concern about the impact on global climate from increasing atmospheric concentrations of greenhouse gases, caused largely by burning of fossil fuels, has resulted in steps toward an international agreement for legally binding commitments to reduce greenhouse gas emissions. Vegetation is an important component of the global carbon cycle, and there is a need to develop an improved understanding of the effects of management practices on carbon stocks in vegetation, particular forests. The effect of harvesting on forest carbon stocks is generally a function of:

• proportion of biomass removed at harvest
• site disturbance, damage to residual trees during harvesting and residue treatment
• regeneration stocking and growth rates
• post harvest mortality
• dynamics of woody debris and litter, and
• understorey response to overstorey changes and harvesting disturbance

In this paper, we use two computer simulation models to explore the impact of varying intensity of timber removal on carbon stocks in Australian native forests, an empirically-derived forest growth simulator and the ecosystem model, LINKAGES, which includes simulates forest growth carbon dynamics in forest floor and woody debris. The aim was to develop stand-level estimates of long-term average carbon stocks under three different management intensities in different forest types: conservation management (no current harvesting, but may be subject to major natural disturbances such as wildfire), extensive management (regular selection harvesting on a 20-40 year management cycle with removal of 20 - 40 percent of the timber volume) and intensive management (more frequent selection harvesting, with a greater amount of timber volume removed). These stand level results will be used in a landscape-level assessment framework to estimate the effect on carbon stocks of changing the area under different management regimes.

A current trend in the development of forest stand models is to use spatially-explicit, individual tree information in order to simulate forest dynamics with increased accuracy. By adding spatial information such as tree coordinates, crown shape and size, it is hypothesized that the computation of the model's driving function (e.g. diameter or height increment) is improved, especially when simulating multistoried stands. Since competition for resources by trees operates at the local level, another trend in these models is to compute sophisticated resource indices such as light availability and canopy structure. In some models, this is done using the fisheye view approach in addition to or in replacement of traditional competition indices. In this paper, we want to test whether these computationally-demanding resource indices such as relative light intensity and canopy structure outperform traditional competition indices in predicting individual tree growth. The study was undertaken in uneven-aged black spruce (Picea mariana) stands in northerneastern Québec, Canada. The predictability of individual tree growth rates was related to crown dimensions and other stand and tree variables measured in the field. Data were collected from 208 stems, using stem analysis, in 52 stands of varying site quality (range 9.5-18.1 m height @ 50 years). Mean stand age was 142 years (range 59-257 years). Various competition indices (distance-dependent and distance-independent) were computed and related to growth increment of individual trees. They were compared to canopy indices estimated by a fisheye view approach using standard algorithms that were integrated into the stand model. Results will be discussed in terms of the overall performance of each competition measure in predicting tree growth increment.

SORTIE was originally developed in the early 1990’s for mixed deciduous forests in eastern North America. One of the major distinctive feature of SORTIE, compared to the ZABOWA-FORET family of gap models, is that the model structure originated from field experiments and heavily relies on the input of data from field studies. During the past four years, the model has been extensively modified for the temperate mixed-species forests of northwestern British Columbia (SORTIE/BC) Significant changes were made to the original SORTIE model to take into account unique features of western forests and unique challenges posed by using of the model for management planning. Currently, work is underway to use SORTIE for southern hardwood forests in Quebec, boreal mixedwood and black spruce forests of Canada, temperate forests of New Zealand and tropical forests of Puerto Rico.

The SORTIE model consists of a mixture of mechanistically and empirically-derived relationships (i.e. hybrid model) and is comprised of four basic submodels: (1) seedling recruitment - a function of parent tree proximity and seedbed substrate; (2) resource availability - predicts understory light dynamics as a function of species specific light extinction coefficients; (3) subcanopy tree growth - function of light availability and previous growth history and; (4) tree mortality - function of recent growth rates. We are currently working to add an understory vegetation submodel and to improve our prediction of growth and mortality of mature canopy trees. The model is ideal for examining stand dynamic and succession after small- to intermediate-scale disturbance in structurally complex mixed or single-species stands. From a forest management perspective, the model has a very flexible user-interface to allow incorporation of a wide range of partial cutting strategies (e.g., understory protection, diameter limit, shelterwood, variable retention). The model easily simulates prescriptions that retain complex stand structures, especially the influence of retained canopy trees on recruitment, growth and mortality of sub-canopy trees. Current model predictions include: (1) spatial distribution and sizes of all individuals in a simulated stand; (2) DBH and height distributions by species; (3) changes in basal area and density, by species, over time; (4) tables of basal area and densities of both adult and juvenile trees and; (5) distribution of subcanopy light levels.

SORTIE can be used to address various ecological and silvicultural questions in the temperate and boreal forest of North America. We are currently re-engineering the model in order to facilitate and improve the continuously changing requirements of the SORTIE users and developers.

Over the last three decades, Research Branch (BC Ministry of Forests) has developed biologically oriented growth models that project the development and yield of even-aged stands. These models have contributed to silviculture planning, timber supply forecasts, and analyses of long-term sustainability. This paper focuses on the extension of these systems to include multi-cohort and mixed species stands, assessment of new harvesting strategies and other work germane to forest certification. Wood quality modelling is reported separately. The results of these investigations are being incorporated into the Tree and Stand Simulator (TASS), a spatially explicit individual-tree model. Other modules buck trees into logs, saw logs into lumber, grade products, perform financial analyses, and assess employment. Our approach is to understand the biology of stand development and build flexible, robust equations and numeric techniques that model height growth, branch extension, stem increment, bole form and wood properties. Simulated tree crowns grow and compete in a 3-dimensional matrix. The crowns, in turn, determine the quantity and quality of wood fibre added each year. TASS is calibrated to conform to the performance of 14 000 remeasured sample plots. Graphical displays and statistical comparisons confirm that crowns and stems develop very realistically and respond appropriately to planting density, genetic improvement, thinning, pruning, fertilization and other treatments.

Extending TASS to more complex stand types requires a better understanding of individual tree growth and competition dynamics under multi-layered canopies for a wider range of tree species. The flexibility needed to address an ever broadening range of scenarios demands even more emphasis on individual tree functions as opposed to stand level functions. The integration of a light model with TASS has been a key step in the development of more realistic simulations of complex stands. Field projects studying stem and crown growth, leaf area/sapwood relationships, and competition for moisture and light in multi-cohort stands of the Interior Douglas-fir zone and the mixed species Interior Cedar-Hemlock zone are planned or well underway. Where moisture and nutrients interact with light to limit growth, competition for below ground resources must also be considered. Adding a spatially explicit natural regeneration module will be a necessary though challenging endeavor. All of these studies incorporate both process modelling of physiological relationships and retrospective analysis of tree growth and stand development.

Increasing influence from the environmental community and other groups has led to rapid and dramatic changes in forest management strategies in the BC forest industry. Ecosystem-based management approaches, such as variable retention forestry, are being implemented to meet environmental concerns and to secure access to the global markets for certified forest products. Field experiments are being installed in many areas to evaluate these new silvicultural options, but they will not be useable for hypothesis testing or model development for many years. In the interim, spatial models such as TASS are well positioned to estimate the growth and yield effects of treatments with a strong spatial component. TASS simulations are presented that show how the spatial arrangement of residual trees affects the growth of regenerated trees in the understory.

LMS is a tool for evaluating forest management alternatives that uses standard inventory information to integrate many analyses and predict changes in stand and landscape conditions over many years. LMS is a Microsoft Windows application that coordinates the flow of information among existing growth models (for most regions of North America), computer visualizations software, and analysis tools. LMS allows the user to treat and grow stands and landscapes and view the outcomes using a "point-and-click" system. By providing graphical, tabular, and visual outputs, LMS facilitates the exploring of current and projected stand- and landscape-scale conditions under different management alternatives. The outputs can include stand structures (several different classifications), standing and harvested volume (by species, size class, and/or log sort), wind and fire hazard, financial analyses, habitat suitability indexes, and other sustainable forestry criteria. The information can readily be transferred to standard spreadsheets (e.g., Microsoft Excel) for further, user-specific analyses. LMS can be used in certification by the certifying agency specifying the objectives and measurable criteria-locally calibrated-which a landowner must meet to be certified. Each landowner could develop a management plan using his/her own copy of LMS that meets these targets, plus any other objectives that he/she may personally desire. Using LMS, sustainable forest values could be readily provided and monitored and the landowner could maintain confidentiality. The LMS software can be downloaded free of charge from: http://silvae.cfr.washington.ed.

To predict the potential impacts of future changes in global environment (such as climate, land-use, fire disturbance and forest harvesting) on the sustainability of forest ecosystems forest resource managers will require forest simulation models. There have been two basic approaches to modeling forest growth and dynamics: empirical and mechanistic forest simulation models. Generally, a weakness of the empirical growth and yield model is strength of the mechanistic process model, and vice versa. It is almost always possible to find an empirical model that provides a better fit for a given set of data due to the constraints imposed by the assumptions of process models. Nevertheless, empirical and process models can be married into hybrid models in which the shortcomings of both component approaches can be overcome to some extent. This is the rationale behind the hybrid simulation approach to forest growth and carbon dynamic modelling. In this paper, a hybrid, monthly-time-step model of forest growth and carbon dynamics (namely TRIPLEX) is presented. The TRIPLEX model integrates the forest production model of 3-PG (Landsberg and Waring, 1997), the forest growth and yield model of TREENYD3.0 (Bossel, 1996) and the soil-carbon-nitrogen model of CENTURY4.0 (Parton et al., 1993).

TRIPLEX aims at being comprehensive without becoming complex, and minimizing the number of input parameters required, but capturing key processes and important interactions between the carbon and nitrogen cycles of complex forest ecosystems. It is designed to hybrid both empirical and process-based models that can be used for 1) forest management (e.g., growth and yield prediction), 2) carbon budget, and 3) climate change impact assessments. The model calibration and testing for Canadian boreal forests will be reported in this study.

Accurate projections of forest growth and yield provide a foundation for the development of sustainable management practices. Historically, foresters have relied heavily upon empirical models of stand growth based upon statistically derived relationships between a simple set of predictor variables and measures of stand yield. While such models have proven effective for relatively simple applications (yield projection in single species, even-aged stands, for example), they are largely inadequate for projecting forest growth and development under the increasingly complex management systems required to meet the needs of today's forest resource managers. In an effort to cope with the increasing demands placed on growth and yield models, developers of such models have attempted to add flexibility to their models with the addition of selected ecosystem processes at varying levels of complexity. However, in a management context, the benefits of added complexity must be balanced against the higher costs in terms calibration data and uncertainty associated with that complexity.

Here we compare two stand-level models developed in British Columbia: TASS and FORECAST. TASS is an empirically based growth and yield model, but it possesses a representation of light competition. FORECAST is more complex and includes both light and nutrient competition, was well as understory representation. Both models were applied to simulate a set of management scenarios. The relative benefit of additional complexity in FORECAST was evaluated by incrementally including ecosystem processes in subsequent runs. Early results indicate that the two models begin to diverge as simulation scenarios increase in complexity. Implications are discussed in terms of core modelling objectives.

This paper attempts to define what constitutes a necessary and sufficient system for practical and reliable estimates of forest site productivity. Selected significant developments in characterising site productivity demonstrate the need to readdress the forestry profession's prevailing paradigm and suggest possible directions.

Traditionally the models used in forest planning have been restricted to characteristics for forest production and timber value. In this study the tree layer will form the basis for making relevant ecological connections to the characteristics of the trees, since the flora and fauna in the forest ecosystem are strongly connected with the structure and status of the tree layer. The study will put special emphasis on tree species that are of little (no) interest for forest industry, but are important in the ecological context.

Different models are developed with the objective to forecast tree characteristics that are better connected to ecological conditions. The basic features of the models are distance dependent individual tree models. Simultaneous regression techniques for a system of functions for basal area, height and crown height/length as dependent variables are used. The competition situation of the trees will be modelled explicitly in the models. Methods are used that consider spatial and temporal relations in the data set.

The functions are based on data from permanent sample plots in the Swedish National Forest Inventory from 1983 and onwards. The data set is representative of forest conditions and forest management throughout Sweden.