Non Technical Summary
The goal of this project is to improve sustainable forest management through enhanced harvest planning, data driven decision-making, and reduced timber harvest site impacts enabled by advanced modeling of the interaction between heavy machine traffic on soil function and tree growth. Project objectives include: 1) characterize the travel patterns of machinery harvesting southern pine (Pinus spp.) plantation stands in different soil types and moisture conditions in both first thinnings and regeneration harvests; 2) link changes in soil physical properties with machine traffic patterns, disturbance categories, and soil conditions; 3) quantify the impact of machine traffic on tree growth following first thinnings in a range of soil types and moisture levels; and 4) develop a harvest planning decision support system that incorporates soil type and moisture, machine induced soil impacts on tree growth, harvest patterns and prescriptions, and depth to water.This project is a collaborative effort between researchers at the University of Georgia (UGA), the National Council for Air and Stream Improvement (NCASI), and the Swedish Forest Research Institute (Skogforsk). We will leverage the resources and previous experiences of Skogforsk, the UGA Plantation Mangement Research Cooperative, the Harley Langdale Jr. Center for Forest Business, the two largest timber Real Estate Investment Trusts (REITs) in the United States (Weyerhaeuser and Rayonier) and Resource Management Services. This partnership will allow us to improve the sustainability of managed forest plantations in both the US South and Sweden.Our research will significantly advance the understanding of soil health and tree growth functions as related to current management systems. Findings from this project will be used in decision-making related to the long-term sustainability of planted forests and the critically important fiber that they supply. Effects of changes in management could be substantial given that our forest sector partners consist of the largest private forest landowners in the United States who manage over 23.5 million acres of planted forest stands.

Classification

Knowledge Area (KA)	Subject of Investigation (SOI)	Field of Science (FOS)	Percent
123	0611	3100	100%

Knowledge Area
123 - Management and Sustainability of Forest Resources;

Subject Of Investigation
0611 - Conifer forests of the South;

Field Of Science
3100 - Management;

Goals / Objectives
The goal of this project is to improve sustainable forest management through enhanced harvest planning, data driven decision-making, and reduced negative effects of timber harvest on soil enabled by advanced modeling of the interaction between heavy machine traffic on soil function and tree growth. Project objectives include: 1) characterize the travel patterns of machinery harvesting southern pine (Pinus spp.) stands in different soil types and moisture conditions in both first thinnings and regeneration harvests; 2) link changes in soil physical properties with machine traffic patterns, disturbance categories, and soil conditions; 3) quantify the effect of machine traffic on tree growth following first thinnings in a range of soil types and moisture levels; and 4) develop a harvest planning decision support system that incorporates soil type and moisture, machine-induced soil effects on tree growth, harvest patterns and prescriptions, and depth to water table (DTW).

Project Methods
Task 1.1: We will install GPS tracking devices on both feller-bunchers and grapple skidders to monitor and record all machine traffic on active harvest sites. Sites will be planted loblolly pine (Pinus taeda L.) stands located in both the Piedmont and Coastal Plain physiographic regions of the southeastern U.S. We will sample active harvest sites spread spatially across the predominant planted loblolly pine region of the southeastern U.S. and temporally across seasons during the project period.Task 2.1: We will characterize machine-induced soil disturbance to correlate visually assessed effects with changes in soil physical properties and relative machine traffic. We will measure and map machine disturbance in categories ranging from no disturbance, light disturbance, moderate disturbance, heavy disturbance, and rutting depth if applicable. Disturbance mapping will then be georeferenced with machine tracking data from Task 1.1 to correlate machine traffic with disturbance severity and location (i.e., distance from landings). ur data will provide information on where machine traffic overlapped between the machines and where the machines operated separately from each other indicating the effect of felling vs. skidding.Task 2.2: Changes in soil physical properties due to harvesting will be assessed by evaluating soil strength using both georeferenced recording soil penetrometers and slide hammers to obtain bulk density samples. At each site, the length of the main skid trail will be determined. Soil penetrometer sampling locations will be established at approximately 0%, 25%, 50%, and 75% of the trail length from the landing. At each of the four locations, the main skid trail will be sampled in a zig-zag pattern with a soil strength profile recorded every 3 m, yielding a minimum of 10 profiles per sampling location (4 locations x 10 profiles = 40 profiles). Off trail sampling will be approximately 40 m from the main skid trail in areas without obvious machine traffic. Transects arranged parallel to the main skid trail will be established and a soil strength profile will be collected every 3 m yielding a minimum of 10 profiles per sampling location (4 locations x 10 profiles = 40 profiles). The entire study will yield 7,680 soil strength profiles (80 profiles x 96 sites).To determine changes in bulk density, 2.5 × 5 cm soil cores will be systematically collected from the soil surface along the main skid trail using a hammer driven double core soil sampler and sealed for later analysis of bulk density (Blake and Hartge 1986).At each sampling location, five cores will be collected in a perpendicular transect across the trail with three samples in the trail and 2 samples off the trail, yielding 20 cores per site (five cores x four locations) and 1,920 for the entire study (20 cores x 96 sites). Soil penetrometer measurements and bulk density samples will be georeferenced with GPS data collected from machines during harvesting (Task 1.1) to determine the effect of repeated machine travel. We will sample across a gradient of machine disturbance severity to develop a curve of physical property changes including penetration resistance (kPa) and bulk density (g/cm3) changes.Task 3.1: We will use depth to water table (m), excess water, water deficit, precipitation (mm), and minimum, maximum, and mean temperatures (oC) to develop annual environmental clusters for site selection. We will work with forest sector partners to identify 40 sites that underwent first thinning (row thinning with selection) to a residual density in the range of 14-20 m2 ha-1 across a range of moisture conditions within the previous 3-5 years based on these environmental clusters. A site will be defined as a single landing area ranging from approximately 8-20 ha, with at least 20 sites containing significant deep rutting (>30 cm). Pairing sites with these annual clusters will provide information on site conditions at the time of thinning and ensure that selected sites fall into a range of moisture conditions.Task 3.2: We will map the skid trails on each site and assign each section of the skid trail to the following categories: deep ruts (>30 cm), moderate ruts (7.5-29.9 cm), and no/minor rutting (<7.5 cm). We will measure soil strength using an electronic and GPS-enabled penetrometer, soil moisture using a soil moisture meter, and soil horizon depths using a soil auger. We will establish rectangular plots extending two planted rows on either side of the skid trails and measure a minimum of eight trees per plot (four trees on each side of the skid trail). For each loblolly pine tree within each plot, we will measure diameter at breast height (dbh; cm), total height (m), height to live crown (m), crown width (m), crown class (dominant, codominant, intermediate, or suppressed), number of sides released by thinning, size and number of bole injuries due to thinning, and distance to skid trail (m). Finally, we will collect an increment core from each tree within each plot. We will obtain data from industry partners on historical management that may influence growth (e.g., site preparation, fertilizer application, competition control, genetics, and stand density). Task 3.3: We will model the annualized growth response to mechanized thinning based on soil conditions and disturbance level. These relations will indicate which factors are the most significant on post-thinning growth response. Generalized Additive Models (GAMs) will be employed to allow for flexible modeling of the non-linear relationships that might exist between the predictors (soil conditions and disturbance levels) and the response variable (annualized tree growth). Task 4.1:We will identify environmental conditions that have a notable effect on the level of site disturbance from harvesting. Key environmental layers will be selected (e.g., depth to the water table, excess water, water deficit, precipitation, soil moisture, soil type, and temperature) using contemporary machine learning algorithms that incorporate variable selection techniques. These selected layers will enhance the system's accuracy and relevance in assessing disturbances.Task 4.2:Seasonal southeastern U.S. maps will be generated based on the key environmental layers identified in Task 4.1. These parameters will be used to visualize a coarse granular map indicating potentially sensitive areas (i.e., geographic areas that may experience significant disturbance that may affect soil characteristics and tree growth) and normal areas (i.e., geographic areas not as prone to significant disturbance). Task 4.3:An RShiny application will be designed to offer a more detailed and user-friendly resource than the coarse maps produced in Task 4.2. Both intra-annual and interannual variations can lead to differing conditions, resulting in varying levels of disturbance due to harvesting. This application will allow users to adjust the key environmental layers identified in Task 4.1, based on a specific geographic area and timeframe of interest to assess potential effects from mechanized thinning or clearcuts. By bridging the gap between intricate data and its accessible representation, this system will assist end users in making informed and sustainable decisions in forestry management and provide documentation for certification bodies.Task 4.4:We will provide protype decision support systems to logging contractors and collaborating companies to obtain feedback on potential modifications that may assist with harvest planning and what additional outputs, if any, would be helpful when supplying certification auditors documentation on their sustainable forest management efforts. This will enable us to create a decision that truly supports decision-making by end users in the field, increasing efficiency, reducing errors, and contributing to the overall success of this project.