Recipient Organization
MISSISSIPPI STATE UNIV
(N/A)
MISSISSIPPI STATE,MS 39762
Performing Department
Dept of Forestry
Non Technical Summary
Forest growth and yield models provide estimates of forest stand characteristics at a future point in time and hence information needed to make management and policy decisions about a given forest resource. Growth and yield models are dynamic systems. As more data and knowledge from research become available, models are enhanced to improve their prediction accuracy, provide alternative modeling frameworks that address different user needs, and enable them to make predictions under different management and biotic and abiotic environmental conditions. Models for loblolly, longleaf, and slash pinedeveloped for Mississippi and the mid Gulf region were put together in the 1980s and early 1990s. Since then, significant developments in modeling techniques have occurred and more data has been collected for some of the species. In addition, almost all slash and loblolly pine stands are currentlyestablished using advanced generation genetically improved planting stock.With the significant amount of new knowledge, there is need for model revision. There is also a need for individual tree models of these species to provide a higher resolution modeling alternative. Research effort on modeling of hardwood stands in Mississippi and the mid Gulf region has been small and hence models for hardwoods are still not well developed or are lacking.Objectives: Theobjectives of the researchare to: 1)develop new stand level diameter distribution models for plantation loblolly and slash pine and even-aged natural longleaf pine, 2) develop individual tree models for plantation loblolly and slash pine and even-aged natural longleaf pine, 3)establish new growth and yield plots in managed natural stands of red oak-sweetgum bottomland hardwoods and inLower Mississippi Alluvial Valleyhardwood plantations, 4) propose new ways of accounting for genetic improvement effects in loblolly pine growth and yield models and incorporate some of the approaches in the models developed in Objectives 1 and 2,5)propose the best statistical modeling approaches and model forms for the forest types in Mississippi and the mid Gulf region, and 6) develop new Windows® computer graphic and web-based user interface for the models developed in Objectives 1 and 2.Approach: Existing data and data that will be collected from ongoing and yet to be established studies will be used in modeling activities for developing whole stand diameter distribution models, individual tree models, genetic improvement response functions, and thinning response functions for loblolly, slash, and longleaf pine and hardwood stands in Mississippi and the mid Gulf region. Various regression techniques will be used and techniques of accounting for hierarchical and repeated measures nature of the data will be applied where necessary. The effect of soil and physiographic factors will be investigated. Alternative model formulations from the literature and those that might be developed during this research will be tested and those with the lowest prediction bias and variance and exhibit logical behavior will be adopted. Performance of the model system will be evaluated using cross validation techniques and sensitivity and uncertainty analysis techniques.The developed models will be assembled into computer instalable simulators using thesoftware development tools inMicrosoft Visual Studio® software.
Animal Health Component
50%
Research Effort Categories
Basic
50%
Applied
50%
Developmental
(N/A)
Goals / Objectives
Major Goal of the Project:The overal goal of the project is to provide forest landowners and scientists ofMississippi and mid-Gulf region of southern United Stateswith upto date forest growth and yieldmodel simulatorsthat areunbiased and precise, address the different forest types in the region, and providealternative modeling resolutions.The specific objectives are:Objective 1: Develop new stand level diameter distribution models for plantation loblolly and slash pine and even-aged natural longleaf pineObjective 2: Develop individual tree models for plantation loblolly and slash pine and even-aged natural longleaf pineObjective 3: Establish new growth and yield plots in managed natural stands of red oak-sweetgum bottomland hardwoods and in Lower Mississippi Alluvial Valleyhardwood plantationsObjective 4: Propose new ways of accounting for genetic improvement effects in loblolly pine growth and yield models and incorporate some of the approaches in the models developed in Objectives 1 and 2Objective 5: Propose the best statistical modeling approaches and model forms for the forest types in Mississippi and the mid Gulf region.Objective 6: Develop new Windows® computer graphic and web-based user interface for the models developed in Objectives 1 and 2
Project Methods
Objective 1: Stand level diameter distribution models for plantation loblolly and slash pine and even-aged natural longleaf pineDataData that were used to develop the CLob and NLongleaf models are currently available. These will, respectively, be used to develop the new plantation loblolly and the natural longleaf pine models. For the longleaf pine model, data from an additional 134 plots are available through a data use agreement with the US Forest Service and these will be included in the development of this model. Data that were used to develop the CSlash model will be used for the new slash pine model. These data are currently being sought from the US Forest Service.AnalysisIndividual model components will be fitted using various regression techniques. Techniques of accounting for hierarchical and repeated measures nature of the data will be applied where necessary. The effect of soil and physiographic factors will be investigated. Alternative model formulations from the literature and those that might be developed during this research will be tested and those with the lowest prediction bias and variance and exhibit logical behavior will be adopted. Performance of the model system will be evaluated using 'leave-one-cluster-out' cross validation techniques and using sensitivity and uncertainty analysis techniques.Objective 2: Individual tree models for plantation loblolly and slash pine and even-aged natural longleaf pineDataThe same data used for Objective 1 will be used in this objective.AnalysisThe analysis will follow the same approach as described for Objective 1 except that the focus here will be on models that operate on individual trees.Objective 3: Establish in managed natural stands of red-oak sweetgum bottomland hardwoods and in LMAV hardwood plantationsThis objective will provide a set of long term study plots that will be measured at the time of plot establishment and re-measured at intervals of three to five years in the future. Data from these plots will eventually be used to develop growth and yield models for managed and unmanaged hardwood plantations in the LMAV and to develop thinning and stand improvement treatment response functions for the recently developed unmanaged natural red oak-sweetgum bottomland hardwoods model (Howard 2011; Jeffreys 2014).A total of at least 450 plots will be established, which will be distributed as: at least 150 in the managed natural red-oak sweetgum minor stream bottom stands, at least 150 in the unmanaged LMAV plantations, and a minimum of 150 in managed LMAV plantations. In the natural red-oak sweetgum minor stream bottom stands, the plots will be required to have hardwoods making up at least 70% of the total basal area with that of red oaks being at least 30% of the total. For the LMAV plantations, the species to be considered will depend on interest of the stakeholders and may include two-species mix plantations e.g. cottonwood-Nuttall oak (Quercus nuttallii Palmer) mixed species stands or single species stands. In the managed stands, each research plot location will have 4 plots - an unthinned control and 3 thinning, or related treatment, plots of light, moderate, and heavy thinning or related treatment intensities. The plots for each stand type will be selected to represent the widest possible range of age, stand density, and site quality classes. Circular plots will be used and plot size will vary between one-tenth of an acre and a size large enough (to a maximum of 1 acre) to capture at least 50 trees. Regeneration/Ingrowth data will be collected from one-hundredth acre plots within each research plot.On each main research plot, all trees with a diameter breast height (dbh) of at least 3.5 inches will be numbered, tagged, and the following data recorded for each tree:1. species,2. dbh,3. crown class,4. damage codes (if applicable, e.g., cankered, split, lightning damaged, stem rot), and5. distance and magnetic azimuth from plot center.On a representative sub sample of trees, additional measurements will be taken including:1. total tree height,2. height to base of the live crown, and3. crown width.On all trees containing 1, 2, or 3 factory sawlog(s), the following additional information will be collected:1. Length of each sawlog, and2. USDA Forest Service specifications sawlog grade.On the one-hundredth acre regeneration plot, seedlings will be numbered, tagged, and the following data recorded for each one of them:1. species,2. height, and3. seedling condition (if applicable, e.g., diseased, damaged).Objective 4: Propose new ways of accounting for genetic improvement effects in loblolly pine growth and yield models and incorporate some of the approaches in the models developed in Objectives 1 and 2DataStand growth and development data from genetic variety block plots experiments will be used to achieve this objective. Data will be obtained from ongoing studies, and from yet to be established studies, through formal and informal agreements or collaborations with other scientists at MSU and other universities in the region and with forest industry. Currently, informal agreements with other scientists at MSU are in place to use re-measurement data from a 29-year old half-sib genetic family by planting density study at MSU's John Starr Memorial Forest and also from a 7-year old clonal genetic variety by initial planting density by management intensities in Newton County, Mississippi. Other suitable data will be sought and incorporated in the research as they become available.AnalysisThe first stage of the analysis will involve an evaluation of the prediction behavior of the models in the loblolly pine growth and yield simulators currently available for use in the southern United States. Simulators to be considered include FASTLOB, PTAEDA, CLob., and the North Carolina State University Managed Pine Plantation Growth and Yield Simulator (Hafley and Smith 1989). The outputs of the whole model and the various sub-models of a simulator will be compared, across time, for different levels of genetic improvement effects that get incorporated in the simulator as site index change. In the second stage of the analysis, the degree of agreement between the changes predicted by the simulators and those observed from experimental plots data will be evaluated. Non-parametric statistical techniques will be used in this evaluation. Information on the nature and frequency of any inconsistencies between model predicted and observed changes will be used to formulate robust approaches of accounting for genetic improvement effects in loblolly pine growth and yield models and to suggest aspects of genetic improvement effects on growth and yield principles that might need to be investigated through experimentation. The formulated approaches will be incorporated into the loblolly pine models developed in Objectives 1 and 2.Objective 5: Propose the best statistical modeling approaches and model forms for the forest types in Mississippi and the mid Gulf region.This objective will be achieved alongside Objectives 1, 2, and 4. The modeling activities in these projects will involve evaluation of both existing and new modeling approaches to identify those that perform best. The best performing modeling approaches and model forms will be identified and recommended for use in the mid Gulf region.Objective 6: Develop new Windows® computer graphic and web-based user interface for the models developed in objectives 1 and 2The Microsoft Visual Studio® C++ programing platform will be used to integrate the models from Objectives 1 and 2 into Windows® application growth and yield simulators that will be availed as installable programs and also as web-based user interfaces.