Designing and Analyzing a Habitat Model of American Ginseng in the Southern United
Seth D. Webinger
Department of Resource Analysis, Saint Mary's University of Minnesota, Winona, MN,

Keywords: American Ginseng, Habitat Model, Raster Analysis, Statistical Analysis,
Geographic Information Systems (GIS)

American ginseng (Panax quinquefolius L.) is a threatened plant harvested for its root, which
when dried, can sell for $125-$500 per pound domestically and $1500-$2000 per pound
internationally. Starting during the 2012 field season, resource management staff and law
enforcement officials at the research study area (omitted for data privacy) began proactive
efforts to help protect the plant and catch poachers within the study area's boundary. To aid
in the effort of locating potential ginseng growth sites, a habitat model was created consisting
of different habitat variables most favorable for ginseng growth and analyzed using point
data of known ginseng locations. Statistical analysis was used to examine the legitimacy and
usefulness of the model in being an effective tool.
and all National Park Service lands prohibit harvesting of wild ginseng while American ginseng (Panax quinquefolius some US Forest Service lands will allow L.) is a plant that grows in North harvesting with a permit (USFWS, 2015). American eastern deciduous forests, Ginseng is intensively harvested stretching from portions of southern due to the high price at which its roots can Canada to the Midwest, Southeast, and the be sold. The price for ginseng roots varies Northeast portions of the United States from year to year but Anderson, Anderson, (USDA, 2014). The American Ginseng and Houseman (2002) acknowledge that growth range in North America is dry ginseng roots can sell from $125-$500 provided in Appendix A. Despite the vast per pound domestically. More recently, the coverage in which ginseng grows, the State of Kentucky listed the price of dried United States Department of Agriculture ginseng roots in the range of $300-$350 lists the plant as threatened or endangered per pound domestically and $1500-$2000 in Connecticut, Maine, Massachusetts, per pound internationally (University of Michigan, New Hampshire, New York, Kentucky College of Agriculture, 2012). North Carolina, Pennsylvania, Rhode Nearly 95% of all ginseng roots harvested Island, and Tennessee (USDA, 2014). within the United States end up being sold Nineteen States allow ginseng to China (Snow and Snow, 2009). harvest on private lands. Each state has its Cultivated ginseng can be harvested at any own regulations for sustainable age, but Hu (1976) states prices of management of the plant, but generally a cultivated roots are generally much lower plant five years old and with at least three due reduced chemical potency. leaves can be harvested. Most state lands Webinger, Seth. 2015. Designing and Analyzing a Habitat Model of American Ginseng in the Southern United States. Volume 17, Papers in Resource Analysis. 13 pp. Saint Mary's University of Minnesota Central Services Press. Winona, MN. Retrieved (date) from Because of the staggering price at ginseng is being harvested is greater than which ginseng can be sold, law the number of wild plants being grown, enforcement efforts have increased within causing a decrease in the total population some locations to combat poaching. At of ginseng (McGraw et al., 2010). Great Smoky Mountains National Park When considering the ecological (GRSM), 13,000 plants were seized by questions within a spatial context, "The Park Rangers over an 18-year period environmental education and research leading up to 2011 (National Parks potential for GIS applications is Conservation Association, 2011). In 2012, substantial" (Snow and Snow, 2009). This resource management staff and law research explores the question of whether enforcement officials at this project's or not historical ginseng data at the study study area began taking proactive area can successfully validate the creation measures to try and reduce ginseng of a custom-built ginseng habitat model. poaching. The study area is combed for ginseng plants and when found roots of Ginseng Habitat
plants are exposed and marked with a special dye and metal chip which are both As with any flora, certain habitat specific identifiers to the study area. This characteristics must be present for ginseng does not simply prevent plants from being to grow. Soil type, elevation, slope, aspect, extracted, it allows law enforcement and canopy cover can all play a role in officials to prove if plants came from the achieving suitable growth. Each habitat study area when they make contact with variable can have a range of values someone with ginseng roots in their favorable for ginseng growth. possession. Without dye and metal chips, it can be very difficult for law enforcement officials to make a case against an individual suspected of poaching roots. Nutrient-rich soils with a pH level greater A report furnished by the US Fish than 5.5 are most suitable for ginseng and Wildlife Service in 2000 lists both growth (Rock, Tietjen, and Choberka, ginseng dealers and hunters confirming 1999). The University of Kentucky that American Ginseng is increasingly College of Agriculture (2012) adds that more difficult to find (Anderson et al., ginseng tends to grow in soils that are 2002). Over-harvesting of ginseng, or any "moist, well-drained, and high in organic plant, can "drive populations to commercial and biological extinction" (Van der Voort, Bailey, Samuel, and McGraw, 2003). A study by McGraw, Souther, and Lubbers (2010) discusses the In a study (Rock et al., 1999) conducted at difficult recovery ginseng faces after an GRSM, ginseng was found to grow at intense harvest. They focused on two lower elevations within the park, ranging study areas intensely harvested, one in from 2160-3620 feet. GRSM and the study Missouri and one in West Virginia. After area are both within the Appalachian five years in Missouri and 11 years in West Virginia, ginseng growth had yet to reach the pre-harvest number of plants. Findings suggest the number in which According to Anderson et al. (2002), This study analyzed different ginseng is usually found growing on habitat variables to create a predictive slopes that are 10-40% and as far up to model to target areas containing habitat 60%. In the GRSM study, field most suitable for ginseng growth. Desired verification was performed at locations habitat was the result of a weighted sum of known to have ginseng and observed slope five reclassified habitat layers. The ranged from 8-36% with an average of weighed sum's output was reclassified into 24.2% (Rock et al., 1999). three categories: least suitable for ginseng growth, suitable for ginseng growth, and most suitable for ginseng growth. Known locations of ginseng plants were obtained Ginseng can grow on any aspect but Snow to validate the model. and Snow (2009) determined north or east orientation is best. Rock et al. (1999) found ginseng was typically found growing on aspects to the north, ranging To test the effectiveness of the model, between 354° and 10°. In addition, if west statistics were used to determine if the facing slopes or slopes of any other sample population of ginseng points came orientation are covered by shade for part from one of the suitability levels created of the day, ginseng may grow there as well by the weighted sum of habitat variables. (Anderson et al., 2002). It would be best if sample ginseng points came from the most suitable level of the weighted sum. Therefore, the null hypothesis (HO) and alternative hypothesis Excessive amounts of sunlight, or even too (HA) were stated as: much shade, can be detrimental to the lifespan of a ginseng plant. At GRSM, HO: The sample of ginseng points ginseng was found to grow underneath a collected in the study area came from canopy ranging from 20-70% with an locations considered to be most average canopy cover at 61.3% (Rock et suitable for ginseng growth according al., 1999). The University of Kentucky College of Agriculture (2012) suggests HA: The sample of ginseng points ideal canopy cover for ginseng growth is collected in the study area came from a combination of habitat suitability levels created by the model. Because ginseng is protected within the study area and because of high prices in Data Collection
which the plant's roots can be sold, the protection of ginseng is an ongoing A Digital Elevation Model (DEM) of the process. The combined effort of searching study area was created from the National for ginseng, carefully exposing and Elevation Dataset furnished by the United marking plants roots and collecting GPS States Geological Survey. Using the data points is a tedious process requiring Spatial Analyst within ArcGIS, Slope and increased time and personnel. Aspect raster datasets were derived from

the elevation data. Soils vector data was cell. The higher the value assigned, the collected using the Web Soil Survey tool more likely ginseng could grow at that from the United States Department of particular location for each variable. Agriculture, National Resource The z-unit (altitude) for the Conservation Service. Finally, canopy elevation data was originally in meters and cover vector data was collected from the then converted to feet by multiplying the LANDFIRE data site, a program created dataset by 3.28083 using Spatial Analyst. by the wildland fire management programs Because ginseng was observed more of the US Forest Service and Department frequently at lower elevations at GRSM (Rock et al., 1999), cells between 1111- GPS waypoints of ginseng 1911 feet were assigned a value of 3, cells locations were collected during ginseng between 1911-2710 feet were assigned a growing seasons from 2012-2014. One value of 2, and cells between 2710-3509 waypoint did not specifically represent one feet were assigned a value of 1 (Figure 1). plant; in some instances a waypoint may The greater the value assigned to a cell, represent a cluster of plants within close the more desirable the habitat proximity of each other. In total, 940 characteristic is at that location. waypoints were collected from 2012-2014 while 4597 ginseng roots were marked
during that same period. The ginseng
vector point data and study area boundary
vector data were provided by resource
management staff at the study area.

Data Analysis
It should be noted that spatial and textual
references to data and the study location
was purposely omitted from the paper and
Figures 1-5 due to the need to protect
sensitive locational data. For this same Figure 1. Elevation: 1111-1911 feet (green), 1911-2710 feet (yellow), 2710-3509 feet (red). reason, a map depicting actual ginseng plant locations was also omitted from this Aspect was consistent with north- facing slopes best for ginseng growth, while east and west slopes were suitable as Raster Processing well. South-facing slopes were not recommended, but the possibility for Each dataset was clipped to the study area growth exists depending on slope position and projected to the coordinate system and shelter from sunlight (Anderson et al., UTM NAD 1983 Zone 17N. Habitat 2002). Therefore, north-facing slopes were variables in vector format were converted given a value of 3, east- and west-facing to raster data. Cells for each raster dataset slopes were given a value of 2, and south- were set to 30 x 30 meters. Once all the facing slopes were given a value of 1 data were in raster format, cells in each habitat variable were reclassified Average slope for each 30 x 30 according to attributes assigned to each meter cell within the slope layer ranged

anywhere from 0% to 80%. Relying College of Agriculture (2012), heavily on the study conducted at GRSM approximately 70-80% canopy cover is and other literature (Anderson et al., needed for ginseng growth; the average 2002), slopes between 10-40% were shade found at GRSM was 61.3%. For the canopy layer, cells with cover between 50-80% were reclassified as 3, cells between 20-50% were reclassified as 2, and cells between 10-20% and between 80-100% were reclassified as 1 (Figure 4). Figure 2. Aspect: North (green), East and West (yellow), South (red). Areas in green represent orientations most suitable for ginseng growth while areas in red represent orientations least suitable for ginseng growth. assigned values of 3, slopes between 40- Figure 4. Canopy Cover: 50-80% (green), 20-50% (yellow), 10-20% and 80-100% (red). A vast 60% were assigned values of 2, and slopes majority of the study area contains canopy cover between 0-10%, and 60% and greater were least suitable for ginseng growth (red) while some assigned values of 1 (Figure 3). pockets of the study area contain suitable canopy Ginseng requires a balance of cover (yellow) and most suitable canopy cover shade and sunlight to generate growth. According to the University of Kentucky The drainage class for each soil type was used to determine ideal conditions for ginseng growth. Well drained soils were given a value of 3, somewhat well drained soils were given a value of 2, and excessively drained soils, somewhat excessively drained soils, and poorly drained soils were given a value of 1 (Figure 5). An example of the map unit soil description report is provided in Appendix B. Figure 3. Slope: 10-40% (green), 40-60% (yellow), 0-10% and 60% and greater (red). Suitable With each habitat layer reclassified to (yellow) and most suitable (green) slope for represent the likelihood for ginseng ginseng growth dominate the study area with some growth, the weighted sum function in least suitable slope (red) scattered throughout. ArcGIS was used to combine the raster

datasets. Aspect, slope, canopy cover, and Three separate shapefiles of ginseng points soils layers were all given equal weight were provided for each year from 2012- and multiplied by a coefficient of .225. 2014. To start, each of the three point shapefiles were merged together to create Table 1. Breakdown of acreage between ginseng habitat suitability levels for the study area. Suitability Level
Figure 5. Soils: Well Drained (green), Somewhat one master shapefile of all ginseng Well Drained (yellow), Excessively Drained, Somewhat Excessively Drained, and Poorly waypoints collected. ArcGIS was used to Drained (red). The study area is practically split in assign a suitability score from the cells of half between most suitable soils for ginseng growth the ginseng habitat suitability raster to (green) and least suitable soils for ginseng growth each ginseng waypoint by its location. This value was appended to the ginseng point attribute table. Because literature only suggested ginseng Because each ginseng waypoint tends to grow at lower altitude, but failed had a suitability score assigned to it, to give specifics (Rock et al., 1999), the further analysis was performed to see how elevation layer was multiplied by a many points fell into each suitability coefficient of .1. category. Eleven points were assigned a A resulting dataset was created value of -9999 meaning that the point fell with each cell value between 0-3—the outside of the area composed of the habitat higher the cell value, the greater chance ginseng suitability raster. This could be for ginseng growth. This dataset was due to the accuracy of the GPS signal reclassified into three classes using the during data collection which can include, natural breaks (Jenks) method. The Jenks but is not limited to, the number of GPS natural breaks optimization method satellites available, GPS receiver quality, clusters data based on natural groupings. or user error. Cell size used in the analysis Similar values are clustered into a group could have also been a factor, potentially while groups are divided based on the leaving gaps along the study area's largest gaps between values in the data boundary (Figure 6). For this study, these distribution. Due to the sensitive nature of 11 points were excluded from further the results, a map of the output is restricted from published results; however, a breakdown of acreage between classes is Statistical Analysis
provided in Table 1. To determine the effectiveness of the spatial model, parametric statistics were

used to compare the mean of ginseng most suitable for ginseng growth) (Table points to determine if it equaled the mean of the cells from the most suitable habitat It is assumed the variances of the level. The mean of the ginseng points was two populations are unequal because they also compared to the mean of the least come from two unrelated datasets. This suitable habitat level and compared to the was confirmed in Microsoft Excel by mean of the suitable level for further performing a two-sample F-test for variances for each of the following Table 2. Description of ginseng points across ginseng habitat suitability levels for the study area. Suitability Level
# of Points
% of Points
scenarios. Welch's t-test, a variation of the student's t-test, was used because of the Figure 6. Example of gaps along study area unequal variances in this study. For the boundary due to raster cell size. first two tests the null hypothesis was adjusted to be specific to each test, aiming analysis. Ruxton (2006) suggests when to prove or disprove the ginseng points comparing "2 populations based on came from the suitability level being samples of unrelated data, then the unequal variance t-test should always be Table 3 shows descriptive statistics used in preference to the Student's t-test or of the first t-test in which ginseng points Mann-Whitney U test." Thus using were compared to cells from the least Microsoft Excel, a two-sample t-test suitable level. In this test, the t statistic assuming unequal variances was used. If was highly significant (p < .001) and the null hypothesis was rejected, the therefore the null hypothesis was rejected. hypothesis was modified to state ginseng The mean of the ginseng points is not plant locations came from a combination equal to the mean of cells from the least of the habitat levels created by the model. Table 3. Two-sample t-test assuming unequal variances descriptive statistics, ginseng points (GP) against least suitable cells (LS). Out of the 929 ginseng waypoints collected within the study area, 434 of them, or 46.7%, were within cells labeled as most suitable for ginseng growth. Along with this an additional 462 waypoints were within suitable cells. Together these points comprised 97.4% of the ginseng points Table 4 shows descriptive statistics collected from 2012-2014 (suitable or of the two-sample t-test comparing the mean of ginseng points with cells from the suitable habitat level. Once again, the t ‘Suitable, were performed to either statistic was highly significant (p < .001) eliminate or show that the ginseng points and the null hypothesis was rejected. came from one of the less desirable habitat Although the end result was the same as levels created by the model. Because the the least suitable level, there is a shift in null hypothesis was rejected in both the mean of the suitable cells towards the scenarios, it can be inferred that ginseng mean of the ginseng points. The same points are not located in areas containing results of the previous two tests can be habitat specific to least suitable or suitable seen in results of comparing the mean of cells. If the null hypothesis was accepted, ginseng points to the mean of cells from then the mean of the ginseng points would the most suitable level (Table 5). be equal to the mean of the cells from the least suitable or suitable level and model Table 4. Two-sample t-test assuming unequal variables may need to be re-examined. variances descriptive statistics, ginseng points (GP) Evaluating results of the second against suitable cells (S). and third t-tests, ginseng points vs. suitable and ginseng points vs. most suitable, show the mean of the suitable level was smaller than the mean of the ginseng points and the mean of the most suitable level was larger than the mean of The t statistics was highly significant (p < the ginseng points. One could suggest that .001), rejecting the null hypothesis. The an equal mean between ginseng points and mean for the most suitable cells showed a cells from the habitat model could be large increase compared to the means of found by combining the two suitability the least suitable and suitable levels and levels. A fourth t-test was performed to ended up being larger than the mean of the determine if the mean of the ginseng points was equal to the mean of cells from the suitable and most suitable levels. Table 5. Two-sample t-test assuming unequal The difference between means variances descriptive statistics, ginseng points (GP) showed a large decrease for the fourth t- against most suitable cells (MS). test (Table 6). This test differed from the previous three tests as the t statistic was not significant (p > .05) and therefore the null hypothesis was not rejected here. Table 6. Two-sample t-test assuming unequal Discussion
variances descriptive statistics, ginseng points (GP) against suitable and most suitable cells (S/MS). Dividing cells into three sections would allow a solid starting point for resource management staff, provided that the model proves to be accurate. Many reclassifications could take place in the future to target areas with the next highest The fourth t-test helps show the suitability score. mean of the ginseng points is equal to the The first two t-tests, ginseng points mean made up of cells from the suitable vs. least suitable and ginseng points vs. and most suitable levels. The original null

hypothesis is still rejected for the first the 11 points were not included in this three tests, but it can be determined that the ginseng points do not come from cells Accessibility is also a factor that of the least suitable level. could have led to potential error. Because of the large size of the study area and the Sources of Potential Error
short timeframe in which ginseng can be found, it is much more advantageous for Data were collected on many different resource management staff to target areas occasions, by different individuals over a that are fairly easy to get to save time and three-year period, during different times of also to target areas that are easy for day, and under different atmospheric poachers to access as well. Because of conditions. A combination of these factors this, ginseng points collected may not be a could affect the horizontal accuracy of the normal representation of the entirety of GPS receiver during the data collection ginseng within the study area. Of the process. In addition, data were collected ginseng points collected, habitat on different GPS receivers of different characteristics may be much more similar quality in which precision of location than a random collection of ginseng points could have varied. that covers the whole study area. GPS horizontal accuracy error can It should also be noted once again reach upwards to 15 meters depending on that each ginseng point is not the conditions described above (Unger, representative of a single ginseng plant. Hung, Zhang, and Kulhavy, 2014). Some Instead of taking a GPS waypoint for the or all of the 11 ginseng points that nearly 4,600 plants whose roots were received a valued of -9999 could be within marked, it was much more efficient to take 15 meters of a cell in the habitat suitability one point for a cluster of plants, which was raster (Figure 7). A useable value assigned often the case. There was no attribute field in the ginseng point data that described how many plants associated with each point. With this being known, the results of the parametric statistics could potentially be affected. Had a waypoint been collected for every marked plant, the number of plants and percentage of plants within a certain suitability level could have changed the results. Error could have also occurred during the spatial analysis portion of this study. Thirty by thirty meter cells were used to create a general snapshot of habitat Figure 7. Ginseng points outside of suitability characteristics within a given area within raster with 15-meter buffer. the study area. Smaller cells could lead to lengthy processing times and crashes to any number of these points could within the ArcMap program while larger change the outcome of the statistical cells may generalize too much the habitat analysis. Because a definitive solution characteristics over a large area. Within an could not be found to resolve this problem, arbitrary area, there can be many changes in elevation, aspect, slope, canopy, or even in say, New York, Missouri, Georgia, etc. soil. The aim of the model in this study and that the model should be based on was to locate general areas where ginseng which habitat characteristics are present at is most likely to be found. this specific study area. A new model The results of the analysis will also based off of this information could result be affected by the manner in which the in more favorable statistical results and final habitat suitability raster was possibly help crews locate areas better reclassified and the map scale at which suited for ginseng growth. raster and vector data was derived. Depending on the reclassification method Conclusions
chosen and the number of classes chosen could affect how many points were within Although the results of the statistical each suitability level. If raster and vector analysis proved that many of the ginseng data created under National Map Accuracy points did not come from the most suitable Standards was derived at a map scale of habitat level, it does not necessarily mean 1:24,000, data has a positional accuracy of that the model created for this project is +/- 40 feet. This means that any cell or not a useful tool; it just means that it could point used in this study could potentially be reworked to get a more favorable result. shift any direction up to 40 feet (USGS, A balance can be found between relying 1999). As map scale decreases positional on statistical analysis and using reason. accuracy decreases. When looking back at Table 2, 97.4% of the waypoints collected were Further Research
within the suitable or most suitable level, with nearly half of those points in the Another way to test the effectiveness of latter. Given the fact that habitat layers the habitat model would be to perform were reclassified based on what scholars field verification. A set number of cells suggested through literature, a conclusion could be randomly selected from each of can be made that the model has potential the three suitability levels. A small team to be a useful tool, but further research could mark out a 30 x 30 meter area at the should be performed to make coordinates of the center point for a given cell. The number of ginseng plants could There is potential to rework the then be recorded as well as the type of model based on characteristics of the associated plants at that location. The ginseng waypoints. Information could be habitat variables at each cell could also be extracted from each habitat layer and recorded and compared to those that were assigned to each point. Using that used for that cell in the model. information, a new habitat suitability layer Further analysis could also be could be formed to garnish a more conducted on the ginseng data points. favorable end result. Instead of relying solely on literature to But, that may not tell the whole determine values of habitat characteristics story. Field work must be performed be to best for ginseng growth, best habitat assist in creating a more accurate picture characteristics could be extracted from the of the usefulness of the model. Actually known locations of ginseng points. It may finding ginseng in the wild by using not be the case that habitat characteristics suggested locations created by the model are the same in the study area as they are is the only real way to show that a habitat suitability model can work for such a quinquefolius L.) in Great Smoky Mountains National Park. 2-29. Retrieved November 13, 2013 from Special thanks go resource management Ruxton, G. D. 2006. The unequal variance staff at the study area for providing data t-test is an underused alternative to critical to this analysis. In addition, thanks Student's t-test and the Mann-Whitney U goes to the rest of the staff at the study test. Behavioral Ecology, 17, (4). area who contributed to the many Retrieved January 27, 2015 from Google strenuous hours in the field marking ginseng roots and collecting data points. Snow, M., and Snow, R. 2009. The Furthermore, thanks go to John Reestablishment of American Ginseng Ebert, Greta Poser, and Dr. Dave (Panax quinquefolius). WSEAS McConville of Saint Mary's University of Transactions on Biology and Minnesota for their guidance throughout Biomedicine, 6, (2), 38-47. Retrieved the entirety of this process. January 29, 2014 from Google Scholar. USDA. 2014. U.S. Department of References
Agriculture, NRCS. The PLANTS Database. Retrieved February 13, 2014 Anderson, R. C., Anderson, R., and from Houseman, G. 2002. Wild American USFWS. 2015. U.S. Fish and Wildlife Ginseng. Native Plants Journal, 3, (2), Service. American Ginseng. Retrieved 94-105. Retrieved January 29, 2014 from January 6, 2015 from Hu, S. Y. 1976. The Genus Panax american-ginseng.html. (Ginseng) in Chinese Medicine. USGS. 1999. U.S. Geological Survey. Economic Botany, 30, (1), 11-28. Map Accuracy Standards: U.S. Retrieved February 19, 2015 from Geological Survey Fact Sheet 171-99. 1- 2. Retrieved April 8, 2015 from McGraw, J. B., Souther, S., and Lubbers, A. E. 2010. Rates of Harvest and Compliance with Regulations in Natural Unger, D. R., Hung, I-K., Zhang, Y., and Populations of American Ginseng Kulhavy, D. L. 2014. Evaluating GPS (Panax quinquefolius L.). Natural Areas Effectiveness for Natural Resource Journal, 30, (2), 202-210. Retrieved Professional: Integrating Undergraduate October 28, 2014 from Google Scholar. Students in the Decision-Making National Parks Conservation Association. Process. Journal of Studies in Education, 2011. Case Study: Great Smoky 4, (4), 30-44. Retrieved April 6, 2015 Mountains National Park. Made In from Google Scholar. America, 40. Retrieved January 29, 2014 University of Kentucky College of from Agriculture. 2012. Ginseng. UK Cooperative Extension Service, 1-6. Rock, J. J., Tietjen, J., and Choberka, E. Retrieved February 4, 2014 from 1999. Habitat Modeling and Protection of American Ginseng (Panax Van der Voort, M. E., Bailey, B., Samuel, D. E., and McGraw, J. B. 2003. Recovery of Populations of Goldenseal (Hydrastis canadensis L.) and American Ginseng (Panax quinquefolius L.) Following Harvest. The American Midland Naturalist, 149, (2), 282-292. Retrieved October 28, 2014 from Google

Appendix A. American ginseng range in North America (USDA, 2014).
Appendix B. Example of Map Unit Description report for soils (USDA, 2014).



EnvironmentalImpacts of Polyvinyl ChlorideBuilding Materials by Joe Thornton, Ph.D. A Healthy Building Network Report EnvironmentalImpacts of Polyvinyl ChlorideBuilding Materials by Joe Thornton, Ph.D. A Healthy Building Network Report This report was prepared by the author and does not represent the opinions of The University of Oregon or any of its affiliates.


April 2013 – March 2014 IAP-Network StUDyS Developing crucial Statistical methods for Understanding major complex Dynamic Systems in natural, biomedical and social sciences ordinator: Ir ene Gijbels, KUL-1 Phase VII, Contract P7/06 1 General information and list of abbreviations Main achievements per work package . . . . . . . . . . . Most important network activities . . . . . . . . . . . .