<p>BELONGER, PAUL JAMES. Variation of selected juvenile wood properties in four southern provenances of loblolly pine. (Under the direction of Steven E. McKeand.)Gravimetric wood density, x-ray densitometry, and latewood tracheids were used to investigate the relative importance of genetic and environmental effects on various assessments of wood density, tracheid characteristics, and growth traits in loblolly pine (Pinus taeda L.). Breast-height wood samples were taken from four 12-year-old plantings of a genetics trial that included approximately 50 open-pollinated families of loblolly pine from diverse sources. The densitometry and tracheid analysis included wood samples from only two of the test sites and assessed variation in 51 families, and 38 families, respectively. Moderate provenance differences, but strong family and environmental differences were found for wood density and volume, and the pooled genetic correlation between volume and density was -0.30. The Atlantic Coastal and Lower Gulf sources had higher average wood density than the Marion County and Gulf Hammock sources, and the Lower Gulf source had the lowest stem volume. Provenance variation in wood density was not consistent with geographic trends indicating a need for field testing in the area of intended deployment. Location effects were very important and sites which promoted high volume production also appeared to cause low wood density. Strategies are available to combat the unfavorable negative environmental correlation (-0.91, P < 0.01) between stem volume and wood density. Provenance variation was important for disk densities of early ring segments, but diminished with age and was not significant (P > 0.10) beyond the segment consisting of rings 3-6. Pooled narrow-sense heritability estimates for the consecutive-ring group disk densities ranged from 0.142 to 0.225, and all groups were highly correlated with average (tree) disk density (rA > 0.90). Mean latewood density and mean latewood percent both showed a strong positive genetic correlation with average disk density and the disk density of the ring 3-5 segment. Early selection for disk density can be effective.The transition to "mature" wood occurred at ring number 5.9 at the flatwoods location with higher wood density and ring number 8.9 at the upland test site with lower density. The point of transition also showed a strong provenance component (P < 0.05); the higher density Atlantic Coastal and Lower Gulf Coastal Plain sources transitioned sooner than the lower density Gulf Hammock and Marion County sources. Trachied length, total diameter, lumen diameter, and cell wall thickness were measured using outer-ring latewood tracheids. Of the trachied traits, only cell wall thickness showed a marginal location effect (P < 0.10). Measured in the middle-third of the cells, tracheids sampled at the Florida location were about 11% larger in total diameter, lumen diameter, and cell wall thickness, but no difference was detected for tracheid length. Provenance differences were strongest for cell wall thickness (P < 0.01) and marginally important for tracheid length (P < 0.16) and total cell diameter (P < 0.18).Variation among families within provenances was large (P < 0.01 for all tracheid traits) and resulted in relatively high narrow-sense heritability estimates of 0.58 for length, 0.34 for total diameter, 0.22 for lumen diameter, and 0.37 for cell wall thickness. The genetic correlations among these traits were all greater than 0.69 except the correlation between length and lumen diameter which was 0.42. Measurement of tracheid length can be used to assess family differences for cell diameter and cell wall thickness.None of the tracheid traits were strongly correlated with average disk density or the disk density of rings 3-5. Therefore, genetic selection for high average density in 10 to 12 year-old loblolly pines from the provenances studied here will not produce a predictable pattern in the character of outer-ring latewood tracheids.<P>

<p>PARKER, DOUGLAS SEABROOK.Using Botanical Analysis to Shape aLongleaf Restoration Project. (Under the direction of Gary Blank.)<p> This thesis focuses on the rare and endangered Piedmont TransitionalLongleaf Pine Community (PTLC) found on the Harris Research Tract (HRT)located in southern Wake County, North Carolina. The goal of this thesis is to laya foundation of knowledge to guide the restoration and preservation of the PTLC. This guidance is via three different papers with three different perspectives(current, historical, and social).<p> The current condition and range of the PTLC is assessed and verifiedthrough the use of the North Carolina Vegetation Survey (NCVS). During thesummer of 1997, 56 plots 20m X 50m were intensively surveyed. The data wereanalyzed using cluster analysis, detrended correspondence analysis, canonicalcorrespondence analysis, and non-metric multidimensional scaling. The 1222 acreHRT was subsequently divided into eight different forest communities of whichtwo were identified as variants of the PTLC.<p> The historical land use practices, identified in a previous study by ScottBode, are combined with the vegetation survey data to further refine the PTLC intoinput levels for restoration. One portion of the PTLC is seen to contain nolongleaf pine due to the practices of marine stores production, hog grazing, andselective timbering and would require high input for restoration. Another portionis seen to have only suffered fire suppression as evidenced by old photos, landrecords, and species composition and would require low input for restoration. Inaddition, the vegetative survey data is compared to other PTLC in North Carolinaand the historical record to generate a target list of species for longleaf communityrestoration. <p> As the study progressed, a social aspect that could become an obstacle tothe PTLC restoration effort presented itself. The negative attitude of theneighboring communities of New Hill, Holly Springs, Fuquay-Varina, and Apex toward the use of prescribed fire as a management tool became apparent after anincident in the spring of 1998. This attitude based on old information couldbecome an obstacle if unchallenged. Thus, a white paper was written to theneighboring communities to confront the mythos of the evilness of fire. Arationale discussing the need for the white paper was also included.<p><P>

<p>The objective of this project was to develop an optimization model for wetland mitigation site selection using a geographic information system for rating wetlands and traditional operations research methodologies. This project was conducted using a GIS based wetland assessment procedure entitled North Carolina Coastal Region Evaluation of Wetland Significance (NC-CREWS) developed by the North Carolina Department of Environmental Resources' Division of Coastal Management (DCM). NC-CREWS rates possible mitigation sites by the wetland functions they could perform if fully restored. Using this component of NC-CREWS, a conceptual model to optimize the selection of restoration sites based on their functions was developed and tested on a small, fabricated example to test workability. This model was adapted to include possible restoration site ratings from an actual watershed in Craven County, NC, provided by DCM. The 0-1 integer programming model that was developed was tested using trials to address issues of problem size, functional unit level required, and order of sites used in the model. Of the 180 tests, all but 37 reached an optimal solution by 200 million iterations of a branch-and-bound algorithm. The problem size and number of functional units required had little impact on the solution time. The ordering of sites as supplied to the model resulted in nonfeasible solutions if sites were chosen based on a physical characteristic such as size or perimeter length. Given the assumptions made in the model it is possible to derive a list of possible mitigation sites to use for improving field recognizance. <P>

<p> The purpose of this research is to generate forest stands for sampling simulations on anoperational level. In this research, a simulated forest stand consists of x and y coordinates ofstems and their attributes. One-acre stem maps are mapped stem locations and attributes from thefield. The simulation procedure is a two step process. The first step is to create discrete samplesfrom a one-acre stem map. Secondly, the discrete samples are selected randomly withreplacement and placed adjacently until a simulated forest stand of desired size and shape is built. The simulated forest stands are then compared with the one-acre stem map from which they werecreated by their respective average clump size, spatial point pattern, and spatial variability of stemdiameter. Average clump size is estimated using stem counts from grids of contiguous quadrats(Greig-Smith, 1952). Spatial point pattern is classified based on the mean and variance of first,second, third, and fourth nearest neighbor distances (Smith, 1977). The spatial variability of stemdiameter is assessed using the robust semi-variogram estimator (Cressie et al., 1980) The abilityof the simulation procedure to reproduce the above mentioned spatial characteristics is related tothe second order stationarity of stem diameter in the one-acre stem map. In this case, secondorder stationarity is the premise that the mean stem diameter is constant in the one-acre stem mapand the variance is only dependant on stem separation distance. Stem locations, the speciespercentages, and the coefficient of variation of stem diameter should be assessed to determine theapplicability of this simulation procedure for a particular one-acre stem map. <P>

<p>Seedlings of 60 loblolly pine (Pinus taeda L.) families representing four populations, Atlantic Coastal, Piedmont, Coastal x Piedmont, and Piedmont x Coastal were compared for early seedling growth and cold hardiness. The objectives of this study were to examine seedling growth of the four populations in a greenhouse environment and cold hardiness of the four populations in controlled environments (preconditioning) designed to mimic environmental conditions in the Piedmont areas planned for deployment. In addition family differences within each population for seedling growth characteristics (height, root collar diameter, and volume index) and cold hardiness were evaluated. The Atlantic Coastal population was the tallest in average, followed by the Coastal x Piedmont hybrids, the Piedmont x Coastal hybrids, and the Piedmont population. All improved populations were superior in seedling growth to the unimproved Piedmont checklot. The height superiority appears to be related to length of the growing season in the provenance of origin. There was not much distinction between populations for root collar diameter although the improved populations were all superior to the checklot. Differences in volume index were related to family rather than population effects, and differences in height contributed more to volume growth than differences in diameter. Cold injury was not significantly different based on preconditioning treatments, but differences in cold hardiness were related to population and families. Hybrids behaved in a manner similar to the maternal parent. Heritability estimates were moderate for injury assessment, indicating that cold injury is under strong genetic control. Coastal families had the highest narrow sense heritability for injury (h2=0.45) followed by the Piedmont families and Coastal x Piedmont hybrids. Coastal x Piedmont hybrids had the highest heritability for survival followed by the Piedmont families. There was a strong genetic correlation between height and cold hardiness at the population level. The taller trees from Atlantic Coastal population tended to suffer more damage than the shorter trees from Piedmont population. Seven of the top ten families ranked for height were in the bottom 1/3 of families ranked by injury. These tended to be Coastal or Coastal x Piedmont families. Seven of the top twenty families ranked for height fell into the middle 1/3 of families ranked by injury. These tended to be hybrid families, although a few Coastal families were present. Based on these early results, it appears that intraspecific hybridization between the two sources of loblolly pine may provide taller trees similar to Coastal parent and cold hardiness similar to the Piedmont parent. It would appear that there is an advantage to deploy Piedmont x Coastal hybrids on the more adverse (e.g. colder Piedmont, cold area) sites. On the miler sites, deployment of Coastal x Piedmont or hardy Coastal families might be appropriate. There are risks associated with deploying families that are not hardy on these sites. Excessive mortality would cancel out the growth advantage if non-hardy tree genotypes were deployed.<P>

<p>ABSTRACTRANDOLPH, HENRY ANTHONY. Evaluation of Shoreline Plantings for WetlandMitigation at Coddle Creek Reservoir in Piedmont North Carolina. (Under the directionof E. Carlyle Franklin and Douglas J. Frederick.)Wetland mitigation provides a method of compensating for loss of vital wetlandfunctions while allowing unavoidable wetland degradation. Coddle Creek Reservoir,located in northwest Cabarrus County, North Carolina, has been developed as a drinking water reservoir for nearby municipalities experiencing dramatic populationincreases. Approximately 356 acres of shoreline plantings were installed to mitigatewetland losses associated with development of the reservoir. Recently, considerable controversy has developed over mitigation projects and their effectiveness in replacingwetland functions. The objective of this study was to evaluate the mitigation plantings atCoddle Creek Reservoir by comparing them to naturally developed shoreline vegetation at a nearby reservoir. Results will be used for long-range management of vegetation atCoddle Creek as well as future reservoir construction projects.Vegetation, soil, and hydrologic data were collected at Mountain Island Lake to determine natural community development along reservoir shorelines. Natural successionhas created distinct community types progressing from the Mountain Island Lake shorelineupslope along an elevational (moisture) gradient. Project design at Coddle CreekReservoir delineated three specific planting zones to mimic natural community developmentalong a moisture gradient. Results of this study suggest that design and initialestablishment of vegetation zones at Coddle Creek Reservoir closely resemble the naturalzonation of vegetation adjacent to Mountain Island Lake. A narrow fringe of wetlandvegetation occurs along the reservoir shoreline that progresses upslope into a zone of transitional vegetation that blends into the upland zone that dominated prior to reservoirconstruction. This wetland zone is most limited by the amount of wetland soils presentadjacent to the reservoir. The extent of wetland soil conversion is strongly correlated to the slope of the site.<P>

<p>Recent studies of agricultural watersheds have indicated that vegetated filter zones (VFZ) maynot function optimally because of channelized surface runoff through the zone. One proposedsolution has been to disperse channelized surface runoff in the receiving portion of the filter zonewith level spreaders. An initial feasibility study demonstrated that level spreaders substantiallyimproved filter strip performance and reduced non-point source pollution (NPSP) outputs tosurface water. However, widespread adoption of level spreaders as a best management practicerequires designs that are both functional and cost effective over a range of site characteristicsincluding filter zone (FZ) condition and with varying sources of NPSP and under differing ratesof loading.The main objective of this research was to evaluate different level spreader designs andconfigurations for dispersing channelized agricultural runoff and to evaluate them for enhancingFZ effectiveness on several sites with greatly differing watershed characteristics. Otherobjectives included estimating construction and maintenance costs and developingrecommendations for level spreaders for specific watershed and FZ conditions. Level spreaders with associated instrumentation were constructed on eight watersheds from 1989to 1997 representing a wide variety of watershed and FZ conditions. Spreaders withoutassociated instrumentation were constructed on three watersheds. Source areas included cropsunder both conventional and conservation tillage, a pasture, a dry-lot for dairy cattle, and a pavedand partly-roofed cattle containment area. All spreaders tested were designed to be permanentinstallations. Designs tested included commercial galvanized gutters, treated wood, fabric-linedditches with gravel just above and below the ditch, and vegetated berm and trench. Reductions in NPSP through-puts were a function of filter zone size, input concentration, runoffvolume, and season. Filter zones where FZ area was greater than two percent of source-area sizegenerally removed a least a third of each analyte. Of the six sites with dispersed flow which didnot have large unaccounted contributions to the FZ, only one failed to reduce N by 30 percent.All spreader designs improved FZ performance. Level spreaders with larger cross-sectional areaswere more effective for high peak-flow events. However, spreaders with limited cross sectionalsuch as above-ground gutters have potential where excavation of ditches or shaping of spreaderswith large equipment is a problem such as in forests or on steep slopes. The most easilymaintained design is a vegetated berm and trench spreader shaped from soil. However, its use ispractical only where tree roots are minimum and where farm equipment can maneuver duringinstallation. This design also allows limited vehicle traffic over the spreader.<P>

<p>AbstractBIN XIANG Genetic Analysis of Diallel Tests of Loblolly Pine (Pinus taeda L.) (Under the direction of Dr. Bailian Li)A new approach was developed for analyzing diallel tests with SAS PROC MIXED and PROC IML. The new method can estimate variance components, obtain BLUE (best linear unbiased estimators) of fixed effects and BLUP (best linear unbiased predictors) of random genetic effects simultaneously. A new formula based on BLUP was further developed to predict individual tree breeding values. This new analytical method was validated using computer simulation and was compared with other existing programs. To analyze disconnected diallel mating designs with more than one diallel, simulated data generated with known parameters were analyzed using BLUP to compare three alternative models, which include diallel as fixed effect (Model 1), random effect (Model 2) or no diallel effect (Model 3). Both Model 1 and Model 3 produced unbiased GCA (general combining ability) variance estimates, while Model 2 resulted in downward biased GCA variance estimate. The accuracy of BLUP prediction for three models was very close, with Model 3 slightly better than the other two. Statistical approaches were also evaluated for combining multiple disconnected diallel test series in a given region. The best GCA sample variance prediction in the class of linear combination of local variance estimates was derived. Simulation study showed that a checklot adjustment was very critical to improve the prediction of genetic values obtained using BLUP analysis. Additional adjustment with improved GCA sample variance prediction could improve the correlation slightly beyond checklot adjustment.Analysis of annual measurement through age 8 from a total of 275 parents, 690 full-sib families from 23 diallel tests of loblolly pine in Northern, Coastal and Piedmont test regions showed: 1) dominance variance was small (20-40% of total genetic variance) relative to additive variance; 2) heritability increased over time, and the magnitude of heritabilities for diameter at breast height (DBH) and volume was comparable with the corresponding heritabilities for height; 3) DBH and volume had higher genetic correlation with 8-year volume than height. Genetic gain prediction in year-8 volume for selection on height and volume indicated that: 1) selection on volume yielded more gain than selection on height; 2) Coastal population had the greatest correlated response, followed by Piedmont and Northern population; 3) family plus within family selection based on total genetic component can capture the most genetic gain; 4) for all selection methods, additional gain (10-40%) can be achieved by capturing non-additive genetic component.Selection efficiency study of height and volume for three test regions indicated that earlier selection appeared to be more efficient than direct selection on year-8 volume in most selection methods. Family selection can be performed at age 2 or 3 for height and at age 4 for DBH and volume. Combined selection (family plus within family) was highly efficient at age 3 or 4. <P>

<p>Short-term events such as festivals are important components of heritage tourism. Heritage tourism provides economic benefits and cultural sustainability to local communities. This dissertation analyzes perceived levels of authenticity and spending by visitors to heritage festivals. The economic impact of visitor expenditures upon the host regions in North Carolina (USA) is also analyzed. Existing and potential visitor target markets are identified. The festivals studied are Scottish Highland Games, and analyses are based on survey data collected from Grandfather Mountain Highland Games and Flora Macdonald Highland Games held in North Carolina. Perceived authenticity is measured as the average authenticity rating of events offered at the Games and statistically related to visitor characteristics. Second, visitor expenditures are compared across age and income groups and regressed on other visitor characteristics, including the visitors' experience at the event. Finally, economic impact of total visitor expenditures is estimated with an input-output model that calculates income, output and value added impacts. The results provide useful insights for Highland Games and local government officials with respect to future promotion. Highland Games in the United States are staged authenticity, but this does not make them inauthentic. There are important differences in perceptions of authenticity among visitors with regard to gender and historical background. Tourist expenditures are positively related to annual household income and distance traveled. When potentially endogenous attributes acquired at the Games (such as enjoyment) are added to the model, they increase the statistical significance of the model. Finally, the Highland Games visitor expenditures have the biggest economic impact on the lodging industry. This is followed by impacts on eating and drinking, admission and registration, gasoline and retail industries. Since the multipliers are small, the games have a rather small percentage impact on their host regions. However, they generate both large direct expenditures as well as indirect and induced effects. The results of this study show that the Highland Games boost heritage tourism in North Carolina by providing events perceived as authentic and economic benefits to the local community.<P>

<p>ABSTRACTCasey, James Francis. Examining the Adoption of Agroforestry in Southeastern Mexico: Three Essays from a Survey with Farmers in Calakmul, Campeche. (Under the direction of Robert C. Abt.)In three essays, farmer preferences are analyzed using Contingent Valuation and Choice Modeling. Particular attention is paid to the role of uncertainty, investment in human capital and the importance of forest conservation. The data analyzed in each of the three essays was collected by the author from July 1997 through March 1998. The first essay empirically tests the hypothesis that investment in human capital increases the likelihood of participation in an agroforestry development program. Results support the hypothesis that human capital investments that reduce uncertainty, improve the likelihood of participation in an agroforestry development program. In the second essay a choice modeling experiment (CM) is used to assess the value farmers have for different agroforestry systems and the individual attributes that make up each system. Results suggest that farmers in Calakmul place a value on 1) conserving and increasing local forest cover, 2) working in a nursery in order to obtain seedlings, and 3) receiving enough technical assistance to get started with a program. The third essay uses an application of the discrete-choice contingent valuation method to estimate the willingness of farmers, in southeastern Mexico, to participate in forest conservation efforts. The estimated values of the marginal effects from a probit model (Cameron 1987) are used to develop welfare estimates. Results suggest farmers are willing to work slightly less than three days per month in order to preserve an area of community forest. <P>