Selected Voluntary Posters

Forest typology is a discipline which has as its final goal the definition and classification of forests and forest sites into distinctly restricted units (forest types) according to the existing potential productive capacities with an aim of easier and rational management and implementation of management guidelines, so as to achieve overall management goal. The term management types refers to a particular forest area and forest ground which has similar ecological and management properties responsible for regular manner of management. Ecological management forest type is determined on the basis of geological parent rock, forest community, soil type, climate silviculture properties and stand values. The most favourable stand form, rotation diameter of cutting maturity, normal production, and its value are determined for each ecological management type. Sustainablity of management can be achieved by appropriate management, surface stand distribution and dynamics.

This paper presents the development of structural stand elements of pendunculed oak in forest communitt Carpino betuli - Quercetum roboris and ecological-management type II G-10. Researches were carried out on about 170 plots, 20 to 140 years old. The aim of reseach was to determine development model for pendunculed oak and the elaboration of growth and yield tables.

The evolution of spatial, individual-based models of forest dynamics is reviewed and areas for future work are identified. The spatial considerations of several simulators of forest dynamics developed in the last few decades are compared. Although the individual-based approach is amenable to spatial simulation of sessile organisms, many individual-based models of forests lack spatial complexity. For example, ecological models developed prior to 1985 tend to be spatially simplistic, particularly in the horizontal dimensions. Nonetheless, they have been used to address a wide range of theoretical and applied problems. In certain cases, these parsimonious models do not appear to be substantially inferior to spatially complex models. Instances in which forest models with limited spatial structure are likely to be less effective are discussed in light of recent advances in vegetation science. Areas requiring further attention include the effects of propagule availability in space and the effects of diminished spatial resolution in the horizontal spatial domain ranging from individual trees to intrastand patches.

Ontario, as a party to the Canadian Council of Forest Ministers, has endorsed the Criteria and Indicators (C&I) approach to forest sustainability reporting. It is officially recognized in the Forest Resources Assessment Policy for Ontario, as specified by the 1994 Decision of the Class Environmental Assessment for Timber Management in Ontario and, the Crown Forests Sustainability Act. An issue with C&I, as currently specified by the Canadian Standards Association or the Montreal Process, is a lack of methods for roll-up of indicator reports to evaluations of overall sustainability. Where Ontario is obliged under legislation to evaluate and report on forest health the option to compile baseline indicators and then monitor and manage them individually is not available. Ontario is aware of at least four schools of thought (spatial or non-spatial simulation optimizations, ecosystem health, economics, multi-criteria analysis (MCA)) with some capacity to integrate the three pillars of forest sustainability: social, economic, environment. For Ontario's 2001 state of the forest report the MCA method proposed by The Center for International Forest Research has been adopted. The Ontario experience to date is presented, and clues to future directions are examined.

Tree growth depends on light absorption and should thus be related to leaf area. For advances in growth modelling it is urgent to develop the methods for leaf area estimation on standing trees. So far the leaf area/sapwood area relation has been used for this purpose, leaving a quite big variation unexplained and a systematic variation between stands. We examined the possibilities to use photos and image analysis to sharpen the leaf area estimation.

Eight trees in each of three seed tree stands in northern Sweden were selected. Photos from ground level at tree-length distance were taken with a high-resolution digital camera. Calculated values from the images were correlated to crown biomass from destructive sampling and measured tree variables. The measurement errors at destructive sampling were estimated at 5 % for leaf biomass and 8 % for leaf area. Residual standard deviation from the leaf area/sapwood area regression was 27 %. Supplementing with data from the image analysis reduced this error to about the half. The result is in line with earlier studies on seedlings and saplings. The conclusion is that image analysis is an efficient way to sharpen the estimation of leaf area and biomass on standing trees.

British Columbia (BC) possesses a variety of forest ecosystems and manages about 59 million ha of forestland. Complex stands (multi-species, multi-age) are an important component of these lands. The BC Ministry of Forests is adopting the Forest Vegetation simulator (FVS), developed in the United States, as a forest growth and yield model for use in such stands (called Prognosis^BC). The regeneration component of Prognosis^BC has been calibrated over last few years using regression approaches with data collected in the southern interior of BC. Previous studies showed difficulties with these approaches due to data shortages and lack of correspondence between the systems of site productivity used in the United States and BC. As result, Prognosis^BC is used at present without a regeneration component. Approaches that make efficient use of limited data and avoid the need for correspondence between site productivity systems should result in better regeneration prediction. Imputation techniques meet these requirements.

To test the use of imputation approaches in predicting regeneration in complex forests following partial cutting, stands from the Interior Cedar Hemlock moist warm zone variant (ICHmw2) in the vicinity of Nelson, BC were selected. These stands normally are comprised of several tree species, may contain a number of age cohorts, and are commonly harvested using partial cutting techniques. Two imputation approaches were examined: tabular and most similar neighbour (MSN). Comparison of the two approaches showed the MSN approach to provide superior predictions.

The objectives of this research are to identify potential long-term trade-offs of forest management options on aquatic habitat of small and intermediate streams in coastal British Columbia, Canada. Next to many biotic processes, the production of salmonids and other aquatic organisms is dependent on the structural complexity of aquatic habitat within stream systems. The approximation of timber harvest-induced changes in pattern of runoff generation, sediment input, and large woody debris (LWD) recruitment, and their likely impact on the channel network and hence aquatic habitat, is therefore essential for evaluating relative impacts of forest management options.

As no field experimentation on a large scale is possible, I have built a spatially-explicit, pixel-based, and stochastic management model using Microsoft Visual Basic and Access. I use a 3,200 ha catchment in the Tsitika valley on northeastern Vancouver Island as a management platform. I obtained process information and data from the literature. Semi-hypothetical landscapes are generated by running the model (without harvesting) for, say, 500 years. Forest management activities are then simulated for these landscapes. The rule-based model represents the major hydrology, coarse sedimentology, vegetation dynamics, and LWD recruitment processes. It approximates the cumulative, downstream-propagating impacts on the stream system and aquatic habitat. I evaluate the effects of management actions in a qualitative way. The focus is on trends in frequency distributions of processes and in relative answers, rather than on accurate and absolute predictions. The model is based on the assumption that major precipitation events and the ensuing storm runoff are 'driving' the system. These major events initiate debris torrents, mobilize sediments, and entrain LWD in stream channels, and result in changes in stream habitat characteristics.

I want to gain an understanding of the implications of road network configuration, and areal extent and disposition of cutblocks for cumulative impacts on the stream system, and hence aquatic habitat. I contrast the following management options: 1) rate of harvest (no logging versus logging at a long-term yield rate); 2) alternative spatial harvest patterns (small versus large cutblock sizes arranged in dispersed and concentrated patterns); 3) alternative temporal harvest patterns (continuous versus intermittent harvest); and 4) road density of alternative harvesting systems (grapple/highlead versus skyline yarding).

50 years of daily temperature and precipitation data from the closest weather station are input to a Thornthwaite water balance to track soil moisture over time based on precipitation, evapotranspiration and soil data. Using the meteorological data, the effective precipitation is calculated based on antecedent soil moisture and infiltrability. For threshold precipitation events, the model generates design storms from the historical data and calculates daily runoff and peakflows in stream channels using a modified Clark unit hydrograph where antecedent precipitation is accounted for. A high and a low density road network for the short- and long-yarding systems, respectively, were designed in conjunction with large (~ 80 ha) and small (~ 25 ha) cutblocks. An extension of the channel network by forest roads can be approximated by a change in the time of concentration of the unit hydrograph. After each critical discharge event, the model calculates erosional events on hillslopes and in stream channels. Probabilities for debris torrent initiation are based on hydraulic forcing due to major precipitation events, time since last disturbance, land conditions, hillslope gradient, and road construction (if a road is present). The deposition of a debris torrent is deduced from its relative LWD content and the hillslope gradient or stream power. Vegetation dynamics is simulated based on nine dominant stand types and alternative successional trajectories. Stand attribute data like height, density, stem diameter and volume are based on the stand level growth and yield model TIPSY. Vegetation dynamics and forest management (harvesting and road construction) are simulated yearly. Given the downslope path of a debris torrent, the quantity and quality of LWD reaching streams is tracked. The LWD channel input from riparian stands is approximated by an average annual rate. The formation of log jams is approximated based on the relative LWD load in the channel. Log jam dynamics with regard to size, height and dominant tree species are tracked and sediment storage capacity is computed. The occurrence of channel bedload movement is approximated by a stream power probability function. Bedload transport is computed by the Bagnold equation. Using the B.C. Channel Assessment Procedure classification of stream channel morphologies, the channel type and state is calculated based on hydraulic equations for bankfull width and depth. The state of stream channels can change from 'highly aggraded' via 'stable' to 'highly degraded' based on sediment and LWD input, and flow. Finally, the channel network accessible to coho salmon (Oncorhynchus kisutch) is rated for its spawning and incubation, summer rearing and over-wintering habitat capability based on fish expert rating of channel morphologies.

Results of sensitivity analysis for key parameters and Monte Carlo runs for management scenarios will be presented and implications for forest management will be discussed in my doctoral dissertation to be completed this year. In the future, this modelling approach may also be applied to other fish species, and other aquatic and riparian-dependent wildlife species.

More detailed information about the Fish-Forest Watershed Model is available on http://www.ForestWerks.ca.

Forest managers use empirical growth and yield type models to update forest inventories and evaluate the effects of proposed management strategies. Growth and yield models predict future growth based on current tree and stand characteristics, such as basal area, site index and individual tree species with little or no consideration of extreme or variable environmental conditions. Process-based models are used to increase understanding of ecological processes and predict variation in ecosystem production under different environmental conditions. Process-based models more accurately predict variability in forest productivity under a changing environment but often do not create outputs applicable to forest inventory updating or forest management. Previous attempts to account for variable environmental conditions have included hybrid models and internal adjustments to growth and yield models requiring extensive refitting of the models, while often ignoring interactions between various climatic variables. A growth index created using outputs from the PnET-II process-based model and applied to estimates from the STEMS growth and yield model provides for the inclusion of environmental sensitivity in the management-oriented outputs of a growth and yield model. This type of index is created and applied entirely external to both models providing a simple and practical means for improving the sensitivity of growth and yield estimates to the changing environment as a whole.

Forest planning models have exploited the vast increases in computing power throughout the 1990s to better represent both the state of the landscape as a detailed and evolving forest inventory, and prevailing forest policies as constraining policies and zones of application. While these improvements have clearly resulted in more accurate assessments of the impact of forest policy on long-term timber or fibre availability, neither the expression of regulation and control as constraints nor the presentation of timber/fibre supply as something to be maximized accurately represent the long-term realization of the types of "on the ground" forest management decisions needed to meet social and investment goals inherent in the use of forest landscapes.

Our newest forest planning model, SYMBOL (Sustainable Yield Model for Building Optimal Landscapes) uses a fundamentally different representation of forest policy. Policies are stated as goals, and penalty cost functions are developed that the model uses to drive toward a long-term forest plan that minimizes the penalty costs. Conversely, harvesting levels are represented as constraints, reflecting the necessity of achieving a minimum return-on-investment required for sustained operation of a viable forest products industry. SYMBOL is therefore a mechanism for finding a long-term harvest schedule that meets supply requirements while mitigating against deviations from the social and environmental goals stated for the forest landscape.

Our poster session will use a sample application of SYMBOL on an IFPA (Innovative Forest Practices Area) in British Columbia to demonstrate the formulation of these problems and the results achievable from this type of analysis.

The vast majority of growth and yield models can be classified into whole stand models, diameter class models or individual tree models. While the data used to develop the models may be the same, the results from different modelling approaches can be drastically different. Because there is no rigorous justification for determining the "best" approach, the incompatibilities among different types of models can sometimes cause serious problems for model users.

In this analysis, the historical development and the main characteristics of the Weibull distribution function are discussed. The procedures used for estimating the Weibull parameters are evaluated. It was found that the one-step parameter prediction approach, based on fitting the combined distribution data from all sample plots at the same time, gave the most reasonable estimate. The conventional two-step parameter recovery procedure of fitting separate Weibull distribution to the diameter-class frequency data from each individual plot, followed by regressing the estimated parameters to other plot-level variables, was found to be less efficient than the one-step parameter prediction approach.

The estimated Weibull distribution function was partitioned into an individual tree model based on the "inverse-distribution" approach. Diameter class and stand-level yields were obtained by accumulating predicted individual tree yields from the partitioned function. This allowed compatible projections on three levels of detail (whole stand, diameter class and individual tree) to be made without requiring the fitting of individual tree models. It also eliminated the need to develop sometimes fairly subjective procedures for reconciling the incompatible projections that typically arise when different types of model are used.

A growth model for oak in Denmark has been developed, based on data from 52 long-term experiments. The model combines a stand growth model for prediction of aggregated values for stands and models for individual tree growth, which can be constrained to ensure consistent estimates of stand values. The state variables chosen to describe even-aged oak stands are top height, stand basal area and stem number. A system of simultaneous difference equations is developed and parameters estimated for description of annual growth. Thinning activities are included as vectors of shifts, both in the estimation and in the application of the model. They represent the discrete changes in basal area and stem number due to thinning. The model uses one rate constant as indicator site quality. The rate constant is based on both height and basal area, thereby combining two main factors of volume production. The rate constant is estimated for each experiment and is correlated with traditionally used measures of site quality as well as soil properties. The rate constant can be estimated from historical records of growth, and can be updated continuously as more data are collected. Furthermore the state of a stand is directly used for prediction. The model can handle unthinned stands as well as managed stands. A non-spatial individual tree growth model for oak is developed as a system of simultaneous difference equations for annual growth of diameter and height. The model uses the stand rate constant as indicator site quality. The individual tree diameter growth model can be used for allocating stand basal area growth as predicted by the stand growth model. The models can be used for evaluating management alternatives and for projection of inventory observations for management planning purposes. The models can give valuable information on sustainable management, for stands as well as single trees.

As deforestation through logging in the tropics continues, conservation of remaining forests becomes more and more difficult. The overuse of the forests and a shortage in the recruitment rates as indicated in recent research warrant to draw serious attention to those processes. We enrich current discussion with a simulation study to assess various impacts of tree harvesting in tropical rain forests in Sabah, Malaysia of different fragmentation intensity. For this purpose, results of simulations of 40 different scenarios with the process-based forest growth model FORMIND2.0 were assessed. FORMIND2.0 is based on calculations of carbon balance of individual trees, which belong to 13 different plant functional types. Simulations suggest that natural recruitment acts as a strong buffer in response to disturbances. Thus, especially in fragmented forests, standing bole volume or number of saplings cannot indicate whether recruitment is sufficient to prevent forest degeneration and species loss. The recruitment rates (at small diameters 1 cm) need to become in focus of the field surveys. From the detailed description of different logging damages, simple regression functions emerge, which allows us to assess logging impacts. Disturbances in fragmented rain forests led to shifts in abundances of species and to species loss. Reduced-impact logging methods led up to three times higher yields than conventional methods. In the latter, yield increased with the length of the harvesting intervals, whereas reduced-impact methods had maximum yield for a cycle period of 40 years. Results were based on optimistic assumptions, but show already, that current logging practice in South-East Asia overuse forests to a large extent. The consequences of the results to other forest stands are widely discussed.

The stand simulator is a collection of several models of stand growth and treatment. All of them can be run under one interface including graphical, statistical and time-series tools to describe the development and current state of a stand.

The simulator can be used to estimate the growth of a stand over any given period of time with the precision of a sigle tree. It can also do various kinds of thinnings and estimate the costs related to them.

There are both empirical and process based models available in the simulator. While empirical models mainly are specified for a particular type of stands, process based models can be used more widely as long as the set of growth parameters required corresponds to the environmental conditions. The simulator compares the stand to the limitations of each model and informs the user which models are available for the case or gives a warning when the stand is leaving the region of validity of the model used.

The stand simulator is designed to be used for research purposes as well as a tool to help decision making on the field level.

Global change may already be affecting natural and managed forest ecosystems across Canada. Although the impacts will vary from site to site, likely effects of a warmer climate include increased losses from fires and insect outbreaks and shifts in species composition. Primary productivity is likely to decrease in water-limited areas, but in cooler and wetter regions (further north and at higher elevations), increases in productivity are possible.

Forest management will need to respond to the anticipated changes in forest distribution and growth rates, while also possibly accommodating the demand for carbon sequestration in forests to offset fossil fuel emissions. In this study, a forest gap model modified to include management options, FORSKA-M, was parameterised for western central Canada and used to investigate effects of implementing alternative management strategies on forest succession and productivity in the Foothills Model Forest (FMF, near Hinton, Alberta) under scenarios of climatic change. The model was first validated against permanent sample plot data from the managed forest area within the FMF. The model was then applied to a set of representative forest stand types in this region. Baseline simulations without forest management were run to investigate the impact of climatic change and increasing disturbance frequencies on average carbon storage. Management response strategies were then investigated, including adjusting harvest rotation periods and thinning schedules, improved fire control, and changing species composition when stand replacement occurred. These management options were tested with the simulation model to analyse their potential to increase both carbon storage and sustainable timber yield in the FMF.

LIGNUM is a functional-structural tree model that treats a tree as a collection of a large number of simple units that correspond to the organs of the tree. The model describes the three-dimensional structure of the tree crown and derives growth in terms of the metabolism taking place in these units. LIGNUM is capable of analyzing the behavior of individual trees in different environments with varying surrounding vegetation and environmental change. It is accurate at describing how trees use resources and how the size and tree structure responds to this. The ongoing work has shown that Lignum has a wide range of possible applications e.g. prediction of early stand development, simulation of an agroforestry system, study of 3D functional organization of trees and its importance plant herbivore interactions.

LIGNUM implements the tree (i.e. the structural units) as abstract data types that can be adapted to various tree species and modelling purposes. The growth of a tree in LIGNUM is described in terms of 1) Metabolism (photosynthesis, respiration, and nutrient flow), 2) Birth and senescence of tree parts and, 3) Architectural development. Different functions of LIGNUM have been realized as Workbench LIGNUM that allows several analyses through a graphical user interface. We demonstrate examples of how Workbench LIGNUM can be used for research purposes, including simulations of different tree species, examination of tree architecture and visualization inner structure of stem and branches.

The model LIGNUM belongs to a family of functional-structural tree models (FSMs) that offer a way of synthesizing both high resolution and aggregated information of trees. Presentation of a tree as repeating sequences of only a few basic structural units allows LIGNUM to be used for a large variety of plant analyses. It enables applying different methods to describe plant metabolism in the realistic structural framework. Since the FSMs contain an explicit description of the plants they are especially well suited for studying questions related to structural dynamics of plants and plant-environment interactions in heterogeneous environments. Detailed description of the plant architecture facilitates accurate estimation of actual resource uptake and, most of all; it facilitates the estimation of limits on the organ growth set by the resource uptake and tree-like growth habit.

Typically when modeling a tree species with LIGNUM the following processes are assessed: 1) Metabolism (photosynthesis, respiration, and nutrient flow), 2) Birth and senescence of tree parts and, 3) Architectural development. LIGNUM has been implemented using generic programming in terms of following components: 1) The Core Model implements the structural units and inherently the tree as abstract data types that can be adapted to various tree species and purposes, 2) The Function Layer implements the metabolic processes, 3) The Structure Layer describes the architectural development of a tree, 4) The Tree Species Layer implements various tree species by selecting appropriate components from the core model, function layer and the structure layer, and 5) The Visualization that provides new ways to study visually the results of simulations. We demonstrate examples of how LIGNUM can be used for different modelling tasks.

The objective of this study was to demonstrate the utility of a decision-support algorithm for stand density management within upland black spruce (Picea mariana (Mill.) B.S.P.) plantations. Specifically, a IBM-compatible PC-based algorithm was developed which (1) graphically illustrates expected 50-year size-density trajectories for each user-specified density management regime, (2) given (1), calculates and subsequently tabulates annual yield estimates including those obtained from thinning treatments (e.g., mean dominant height, density, mean volume, total volume, total merchantable volume, quadratic mean diameter, and basal area), and (3) given (2), graphically illustrates empirically-derived yield production curves for total and merchantable volume per hectare including cumulative values for each regime. The utility of the algorithm is demonstrated within an operational context: determining site-specific thinning regimes densities required to attain user-specified management targets. Procedures for acquiring the executable version via the Internet are also included.

The initial growth of a Norway spruce plantation is limited by several biotic and abiotic factors. The decreasing number of foresters in modern forestry has raised the need for decision support systems. A model for regeneration of Norway spruce stands is under construction where the growth and survival of individual planted Norway spruce seedlings will be the output. Effects of competition, nutrition, water availability, frost damage, row deer browsing and pine weevil damage etc. will be included. Influence of silvicultural methods, such as site preparation, shelters and plant protection, will also be considered in the model.

In the model, site properties, year to year variation and silvicultural treatments will affect both growth of the seedlings and occurrence and severity of damage. The occurrence of damage will result in mortality and growth reductions. Some preliminary results showing the influence of various silvicultural treatments and site properties will be shown.

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.

New Zealand as a signatory to the Montreal Process is required to report on the national forest estate on a regular basis. Amongst the indicators that must be reported upon are those related to soil nutrient supply. Reporting changes after the fact suggests that in the case of a decline in an indicator value some remedial action needs to be taken, and this may be costly or impractical. We are therefore taking a more pre-emptive approach to Montreal Process reporting by building predictive risk assessment models for sustainable forest management based upon an understanding of soil nutrient supply and nutrient uptake by radiata pine. This will allow development of best management practises to minimise or prevent any detrimental effects such as nutrient depletion, and give the opportunity to avoid negative trends in biophysical indicators. One component of our model predicts long term changes in soil and forest nutrient pools under different management regimes on major NZ soil types.

Data from our series of six Long-Term Site Productivity trials has been used to develop nutrient pool data for major soil types in New Zealand, and nutrient pools within the tree biomass are estimated using results from biomass studies. Productivity and nutrient contents at the end of a rotation are estimated using growth models and allometric relationships developed for radiata pine. Estimates of weathering, atmospheric deposition rates, slash decomposition and organic matter mineralisation are obtained from published studies and available datasets.

The model is spreadsheet based, and is linked to a GIS database to allow spatial representation of predicted changes in soil nutrient pools over multiple rotations. The main use of the model is in the development of scenarios for the future and for sensitivity analyses to determine the relative importance of various nutrient pools and fluxes within the systems.

This paper presents the results from two thinning experiments in sycamore (Acer pseudoplatanus L.) and discusses the implications for a growth model: Which allometric relationships should be considered. Thinnings were conducted from below at regular intervals, i.e. approximately 5-years. The results imply that thinnings, only to a very limited extent, increases the diameter growth of sycamore (Acer pseudoplatanus L.), and that severe thinnings causes a substantial decrease in volume growth. Additionally, the height growth is negatively correlated with thinning grade. So much so, that the traditional method of estimating site productivity from age-height relationships may lead to serious bias. The results are discussed in terms of self-tolerance and growth response, and it is proposed that sycamore (Acer pseudoplatanus L.) has a high degree of self-tolerance but a poor response in radial growth to increased growing space. The reduced height growth at lower densities is thought to occur as a result of weak epinastic control. These results contradict, to some extent, general theories of forest growth.

A sustainable forest management plan for the 58,000-acre Chesapeake Forest (on the Eastern Shore of Maryland) required a schedule of thinning and harvesting activity that would meet a variety of constraints. An important goal of the plan was to enhance habitat for the endangered Delmarva fox squirrel, while protecting water quality and providing a sustainable flow of forest products from the land. The preferred habitat for the Delmarva fox squirrel is mixed pine-hardwood stands with relatively large-diameter trees (16" DBH and larger). The Habplan harvest scheduling model was used with a GIS database to spatially allocate harvests and thinnings annually over a 50-year period. Management regimes with multiple thinnings were emphasized in areas focused on habitat enhancement to maximize diameter increment, while general management areas included shorter rotations to meet wood production goals. This poster will illustrate the application of the Habplan model and demonstrate the tradeoffs in habitat and harvest goals.

Modern growth and yield applications consist of various tree- and stand-level models as well as technical implementations of user interface and databases. As the systems become more complex, the cost of revising and recoding also tends to rise. On the other hand, there are viable technical solutions for extending the "life" of even old pieces of software by reusing them as subcomponentes of new applications.

MOTTI is a growth and yield simulation application running in MS-Windows -environment. The components of the application can be divided into stand- and tree -objects, user interface, database, and parameters. The core of MOTTI is a set of tree- and stand-level models implemented as FORTRAN routines. The user interface and database consists of freeware ActiveX -components and VCL-components available in Borland Delphi application development environment. The oldest FORTRAN subroutines were coded in the beginning of 1980's, but they were still usable without a need for modifications. Only new headers needed to be inserted before the subroutines were recompiled and inserted into a dynamic link library (DLL).

The current version of MOTTI is "a modellers workbench"; a tool to test and demonstrate growth and yield models. It includes tree-level models (e.g., growth and survival, taper curves and volumes, timber assortments and timber prices, costs and incomes), stand-level models (e.g., self-thinning, site classification, ditch network maintenance need, selection of trees to be thinned, calculation of mean, median and dominant values). Basically, by using the current components of MOTTI, new applications can be composed with only minor need for new coding.

Although it is technically easy to reuse procedural components in an object-oriented application, there are drawbacks in the "purity" of the design; the interconnections between original procedural routines might cause dependencies, which must be taken into account. It must also be underlined, that growth and yield models are seldom "plug-and-play" modules, which could be reorganized into new applications without careful study of their original purpose and role as a part of simulation flow.

An ecosystem management approach to Northern Arizona's ponderosa pine, Pinus ponderosa var. scopulorum, forests requires maintaining a portion of the landscape as old-growth habitat. The US Forest Service, North Kaibab Ranger District has some of the most extensive old-growth ponderosa pine forests remaining in the Southwest. Conservation areas covering 20% of each forest type on the Kaibab National Forest have been created to protect old-growth tracts (>400 hectares) since 1991. Management within protected areas may include activities that enhance old-growth characteristics. High tree densities, litter accumulations and tree disease are a forest health concern within some old-growth stands. However, activities are often postponed due to controversy or uncertainty regarding, past, present and future forest conditions at the stand and landscape scale. The Forest Vegetation Simulator (FVS) integrated with an ArcInfo, Geographic Information System (GIS) was applied to a 36,000-hectare ponderosa pine forest within a current land planning area. Five forest development stages and structural features were defined by university faculty, public land managers and interest groups based on historic and contemporary forest inventory data. FVS "compute functions" and "keywords" were used to calculate each stage. FVS output and forest health indicators linked to a spatial database were extracted using Arc Macro Language. Spatial information and FVS projections created a dynamic working environment to identify high disturbance risk areas within the planning landscape. Adapting locally defined old-growth conditions to FVS projection and spatial data systems demonstrates their capabilities as an easy-to-use decision support and modeling device. Management alternatives may be addressed interactively by linking these tools.

Recently developed tree list generation (TLG) models have been directly applied to forest operational planning to provide detailed information. Traditional operational harvest planning is based on inventory data and volume estimates from yield curves that are developed locally using sample plot data. During this process, the volume estimates are broad stratum based and detailed tree information is not available for individual polygons. To provide detailed and accurate tree information on the planning area, TLG models were developed recently for Sundance Industries based on the Alberta Vegetation Inventory (AVI) and temporary sample plot (TSP) data. Through mathematical modeling, tree lists were generated for each merchantable stand from TLG models within the Sundance Forest Management Agreement (FMA) Area. The predicted tree list information was then used, in conjunction with current forest inventory data, to produce estimates of the quantity of various sizes of logs. To support the process of harvest planning, the volume and log profile for planned cut blocks were predicted and analyzed. The predicted volumes and log profiles for cut blocks were compared with the actual scaling data harvested in year 2000. Comparison results indicated that TLG predictions could accurately represent the actual scaling records. Therefore, the predicted volume and log profile information was applied for the year 2001 cut block planning. The polygon based volume and log profile information from the TLG models provided detailed product information and solved the log allocation problem between the large log mill and small log mill. The detailed log profile information also makes it possible to optimize the log allocation in the future.

When constructing growth models for multilayered stands one have to include the recruitment of trees into the tree stratum. Norway spruce is the most important species when it comes to this issue in the boreal part of Sweden, as it is the only secondary species occurring. As a first step the occurrence and density of seedlings and saplings has to be evaluated. As step number two, the ingrowth from the sapling bank in terms of number of trees that are passing a certain size limit during a certain period has to be estimated. This study deals with step number one. It aims to estimate the probability of occurrence of small spruce trees (0-40 mm dbh) as a function of different variables, describing a variety of conditions all across the country. I used data from the Swedish National Forest Inventory (NFI), which include about 20000 plots with 10 m radius evenly distributed all over the forested part of the country. Trees are defined as individuals >40 mm dbh in NFI. During 1988-92, small trees were measured within a subplot of 20 m² along with the ordinary NFI measurements. Logistic regression showed that the occurrence of small spruce trees significantly varied with plot size, basal area, humidity, site fertility, spruce proportion in the tree layer, and temperature sum. The model explained 73% of the variation. The results are evaluated and further work is proposed.