Distribution and Spread of Laurel Wilt Disease in Georgia: 2006-08 Survey and Field Observations By: R. Scott Cameron, Chip Bates, and James Johnson Georgia Forestry Commission September 2008 Funded partially by the USDA Forest Service Forest Health Protection Region 8 Table of Contents Introduction 1 Methods 2 Systematic Redbay Mortality Survey 4 Laurel Wilt in Sassafras, Pondberry and Pondspice 7 Results and Discussion 7 Laurel Wilt Infection by County 7 Comparison of 2006 & 2007 LW Infection Levels by County 9 Geographic Distribution of Laurel Wilt in Georgia 9 Spread of Laurel Wilt Disease in Georgia 13 Factors Influencing Natural Disease Spread 15 Anthropomorphic (Man-assisted) Spread 16 Laurel Wilt in Other Lauraceae Species 18 Other Insects and Diseases on Redbay 24 Conclusions and Implications for Management, Research, and Future Monitoring 26 References 29 Introduction Laurel wilt (LW), caused by the fungus Raffaelea lauricola, is a new disease of plants in the Lauraceae family in the United States, vectored by an introduced Asian ambrosia beetle, Xyleborus glabratus (Fraedrich et al. 2008, Harrington et al. 2008). Since the capture of the first X. glabratus in a monitoring trap near the Port of Savannah and the first reports of dying redbay trees (Persea borbonia) near Savannah in 2003 and 2004, this disease has spread rapidly through the abundant redbay in the maritime and coastal plain forests northward in South Carolina and south well into Florida, killing most of the large redbay trees in its path. Other plants in the laurel family known to be susceptible to varying degrees include: sassafras (Sassafras albidum), avocado (Persea americana), pondspice (Litsea aestivalis), and pondberry (Lindera melissifolia), the latter two being federally listed as threatened and endangered species, respectively (Fraedrich et al. 2008). This exotic disease episode is particularly noteworthy because it is caused by a previously unknown pathogen vectored by an ambrosia beetle that was not expected to be a serious threat to forests or wood products in the U.S. Yet in less than 10 years, it has essentially eliminated redbay from a large portion of the South Atlantic Coastal Plain and maritime forests. This disease may continue to expand throughout coastal plain forests from Virginia to Texas and is a threat to the avocado industry in south Florida and elsewhere. Ambrosia beetles generally attack dead and dying hosts and do not vector important diseases. The redbay ambrosia beetle is not known to be an important pest in its native range in Southeast Asia, but was known to be associated with members of the Lauraceae (laurel) family (Rabaglia, 2003). However, the unique association between insect, pathogen and host in which X. glabratus conveys the R. lauricola fungus into highly susceptible redbay trees has proven to be exceptionally fatal. The fungus spreads through the vascular system, blocking water transport, and causes redbay trees to wilt and die within months after initial infection. The wood of dying and recently killed redbay trees 1 serves as host material for X. glabratus, as well as other ambrosia beetles, which multiply rapidly, resulting in large numbers of beetles capable of spreading to new locations. Currently, X. glabratus is the only confirmed vector of R. lauricola, which causes vascular wilt disease in redbay and other Lauraceae species (Fraedrich 2008). This new disease may eventually become much more widely distributed if it continues to spread in sassafras, which is found in much of the eastern half of the U. S. There are many additional genera and species in the laurel family, concentrated mainly in the tropical and subtropical areas in Central and South America, which may also be susceptible to LW disease. Thus, it is important to document the advance of this disease both in geographic distribution and species affected. The spread of the laurel wilt disease has been tracked on the county-level as encountered by foresters and landowners since 2004 (Fig. 1). New counties were added to the map in GA when the presence of the pathogen in samples was confirmed by laboratory isolation. Based on this county-level data, it has been estimated that the spread of the disease has been about 20 miles per year over a four-year period in South Carolina (Boone 2008) and in all three states (SC, GA, FL) combined, the rate of spread has been estimated to be about 34 miles per year (Koch and Smith 2008). This pattern of expansion includes natural beetle dispersal as well as possible human-assisted spread. A single female beetle emerging from infested material can potentially start a new infection center. Female beetles carry the fungus in special pouches located at the base of each mandible and introduce the fungus when they chew through the bark into the wood in healthy trees. Human activities that move wood from beetle infested trees can certainly spread the disease over long distances to new locations. To track the geographic distribution and rate of spread of the disease in a more comprehensive and systematic fashion, surveys have been conducted in Georgia and South Carolina in 2006/2007 (2006 survey) and 2007/2008 (2007 survey) supported by the USDA Forest Service, Forest Health Protection and the corresponding state forestry agencies in Georgia and South Carolina. The objectives of the surveys conducted in Georgia summarized in this report include: 1) Systematically document the distribution of LW disease on a grid pattern over the area known to be infected and beyond the apparent advancing front, 2) Determine the severity of infection within the range of distribution of the disease, 3) Document the rate and direction of spread and elucidate possible causes of varying rates of spread in across Georgia, 4) Investigate the symptoms and incidence of LW disease in sassafras and other species in the laurel family in Georgia, and 5) Lay the foundation for future monitoring projects. Methods Some plant taxonomists recognize three closely related species of Persea, including: P. palustris (swampbay), P. borbonia (redbay), and P. humilis (silkbay). Silkbay only occurs in Florida, but the ranges of swampbay and redbay overlap in Georgia. Swampbay and redbay are difficult to tell apart and are not universally recognized as separate species. Thus, for purposes of this survey and report, swampbay and redbay are considered to be one species, referred to as redbay. 2 Figure 1. Counties in Florida, Georgia, and South Carolina confirmed to have laurel wilt disease by year of initial detection as of June 2008. 3 Systematic Redbay Mortality Survey A systematic sampling procedure was used to delineate the incidence and intensity of laurel wilt disease in Georgia based on visual symptoms of disease on redbay trees. Redbay with wilted and/or dead foliage was used as the primary means for determining the range of LW disease. A 5 x 5 mile grid of potential plot locations was generated on the Georgia Forestry Commission (GFC) GIS system covering a broad area encompassing the counties know to have laurel wilt disease from prior GFC confirmation and the 2006 Georgia survey (Beck 2007), plus a band of additional counties to the north and west of the known disease distribution in southeast Georgia. The coordinates of the 2006 GA survey plots also were layered into the GIS system. A map with the grid points and 2006 survey plot locations was generated with ArcMap and printed out to serve as a guide for selecting survey plots. Most of the 2006 survey plots were selected for reassessment and additional plots were added from the grid to provide a relatively uniform network of plots, approximately 10 miles apart. Additional plot locations were selected from the 5-mile grid as the survey progressed to better delineate the distribution of LW, especially in areas of sparse incidence along the advancing edge of the disease. A band about 15 to 20 miles wide, generally east of I-95, along the coast (including the barrier islands) was excluded from the present survey, since this area was considered to be generally infected. The Fort Stewart (Department of Defense) property in Bryan and Liberty counties and the Okefenokee National Wildlife Refuge in Charlton and Ware counties were not surveyed due to lack of access, with the notable exception of the Stephen Foster State Park Road in the Okefenokee Swamp. Coordinates of selected plots were transferred to an Excel spreadsheet, which was used to further refine plot locations and plan survey routes. The selected plot locations were plotted on Google Earth using latitude and longitude coordinates, and if necessary, the plots were moved to a site with a high likelihood of public road access and the presence of redbay within a 2 mile radius of the original grid plot position. This became increasingly essential as the survey moved away from the coast to where farmland is more prevalent and redbay grows primarily at the edges of streams and river bottoms. The 2006 plot locations and the revised new plot locations were plotted on Delorme Street Map USA, and the most efficient daily travel routes were then designed prior to departure to the field each day. A Delorme Earthmate GPS receiver attached to a computer mounted in the survey vehicle was used to follow the route and locate plots in the field. Field procedures. The actual 2007 plots were located where redbay would be included in the plot as near as possible to 2006 plots and the predetermined new plot positions. Sample plots consisted of one-tenth acre strips ca. 330 x 13.2 located ideally about 132 from the road edge and perpendicular to the road. A string line hip chain was used to measure the length of the plot and a plot center stick (Fig. 3) was used at arms length to estimate the plot width. In many areas, redbay distribution was not continuous or uniform and thus the plot direction was adjusted to incorporate as many representative redbay as possible. Some plots meandered to follow the redbay distribution and a few plots were doubled in width and cut to half the length if the plot ran into water, lack of redbay, or unsuitable land use (i.e. farmland). If redbay could not be located within about a 2 miles radius of the predetermined grid plot location, the plot was recorded with GPS coordinates as no redbay. 4 Each plot was initiated at a redbay tree, designated as the plot origin. A TDS Recon portable data recorder equipped with GPS was used to record all plot data, including: sample number and date, GPS coordinates at the plot origin, direction of travel compass bearing, number of redbay killed by LW, number of live (healthy) redbay, and number of dead redbay killed by other causes. Only redbay trees greater than 1 diameter at breast height (DBH) were tallied. Initially, the number of healthy and flagged (dead branchlets often infested with another exotic ambrosia beetle, Xyleborus campactus) redbay trees were recorded separately, but due to the subjective nature of these categories and uncertain utility of this information, the flagged designation was discontinued and merged with healthy in a live redbay category. Additional data recorded for each plot included: LW severity (0 = none, 1-33% = light, 34-66% = moderate, and >66% = severe), number of redbay sprouts and seedlings (<1 DBH) in a one-fiftieth acre circular plot at the plot origin, presence or absence of redbay stump sprouts in the plot, and numbers of live and dead sassafras or other laurel species present in the plots. All data were also recorded on field data sheets and in some cases, additional notes were recorded about redbay abundance and LW disease in the area, plot characteristics, other tree health problems and causes of mortality, and suitability of the plot for future monitoring. Redbay trees killed by laurel wilt are generally quickly and thoroughly colonized throughout the stem and branches by several species of ambrosia beetles, including Xyleborus glabratus, Xylosandrus crassiusculus, among others. The boles are also invaded by wood rotting fungi, which hasten the decomposition process. Thus, trees killed by laurel wilt tend to break up and fall apart within a couple years after being killed by the fungus. During the early phases of this decomposition process, however, wood moisture content remains high for an extended time, permitting the development of large numbers of ambrosia beetles. In plots evaluated in areas where the disease had already heavily impacted the redbay population, redbay stumps and broken stems were assumed killed by the disease and tallied as LW dead (Fig. 3). In plots where the cause of death of redbay was uncertain, patches of bark were removed and/or small stems cut to determine if the black staining of xylem tissue characteristic of LW was present (Fig. 2). If not, it was assumed that the mortality resulted from other causes. Sometimes the cause was apparent, for example in the case of fire damage. When dead redbay trees were encountered in counties not previously confirmed to have LW, sections of stems and/or wood chips were collected and sent to Dr. Stephen Fraedrich (USDA, Forest Service, Athens, GA) for laboratory culture to confirm the presence of the fungus before being documented as positive for LW. Figure 2. Redbay tree with bark removed to expose the black staining of the wood characteristic of LW infection. 5 Figure 3. Stump and sprouts of a redbay tree killed by laurel wilt (left of orange plot center stake). Although all the larger redbays in the area had been killed by laurel wilt, small redbay trees (orange flagging in the background), and most seedlings and sprouts are still alive. Survey timing. Redbay trees retain their dark-green leaves in the winter and trees killed by LW retain the red to chocolate brown dead leaves for many months after dying, both of which are easier to see during the winter when the woods are more open because deciduous vines, bushes, and trees are without leaves. It is much more difficult to see live or dead redbay trees once deciduous plants leaf out and redbay trees have new, light-green foliage in the spring. Thus, nearly all survey plots were installed and assessed from November 27, 2007 to March 31, 2008. A few special plots detected by private land owners and GFC foresters off the main survey grid were assessed for LW infections at other times of the year. Since this survey generally documented dead and dying trees sampled in the 2007/2008 winter, the survey was essentially assessing infections through the end of 2007. Thus, this survey will be referred to as the 2007 survey as compared to the grid survey conducted the previous year, which will be referred to in this report as the 2006 survey. Data entered on the Recon data recorder were transferred periodically during the sampling period into the GFC GIS/database system. These data were converted to Excel spreadsheet format for purposes of editing and subsequent analysis and presentation of survey results. Maps displaying the locations and relative LW infection levels in plots from the 2006 and 2007 surveys were developed using ArcMap on the GFC GIS system. 6 Laurel Wilt in Sassafras, Pondberry, and Pondspice One objective for this project was to characterize and investigate the incidence of laurel wilt in sassafras, pondspice, and pondberry. Since sassafras and redbay have different site preferences, they seldom overlap on a small plot of land or even on a broader geographic scale as illustrated on the distribution maps for redbay and sassafras presented in Koch and Smith (2008). In addition, sassafras is a deciduous tree and blends in with other deciduous trees in the winter when most of the systematic LW survey was conducted in 2007-2008. Furthermore, the health of sassafras trees is difficult to ascertain with leaves off in the winter. Thus, no sassafras was detected on any of the redbay grid survey plots. The authors became more adept at recognizing sassafras when they flowered and leafed out in the spring and a number of sassafras clumps, mostly along roadsides and fence rows, were identified and surveyed for disease incidence in the spring of 2008. These sassafras plots generally were located in association with travel for other purposes or special targeted sassafras survey trips and thus they are clustered in a few areas and do not represent the actual distribution of sassafras across the entire grid area sampled in 2007. Pondspice and pondberry are scarce and disperse, often occurring together on poorly drained, swampy depressions associated with small sand dunes. Special field visits to two sites with these two species were conducted during the 2007/2008 survey period. One was near Clyo in Effingham County and the other in Wheeler County near McRae, Georgia. Both site visits were coordinated by Tom Patrick with the Georgia Department of Natural Resources. Results and Discussion A total of 237 plots were recorded in the 2007 Georgia laurel wilt survey. Of this total, 188 had at least one redbay and 19 were in areas where no redbay could be found within a reasonable distance (ca. 2 miles) of the pre-established plot center. Another 30 plots were identified as sassafras plots with no redbay present. Only one redbay plot was assessed for each of Wheeler, Burke, and Chatham counties, which were disregarded and not included in the summary data due to the lack of multiple plots. Among the plots with redbay, 102 plots had no redbays with laurel wilt symptoms (none), 34 were lightly infected with laurel wilt disease (1-33% of redbay trees infected), 26 were moderate (34-66 % infected), and 26 were severe (66-100% infected). Laurel Wilt Infection by County in Georgia Data were summarized for each of 24 counties included in the 2007 grid survey by computing the mean among plots sampled in each county for percent laurel wilt dead, percent live, percent other dead, percent of plots with stump and/or basal stem sprouts, and the mean number of seedlings and root sprouts per acre (Table 1). Counties closest to Savannah that have had LW disease the longest, generally had the highest percent of the redbay trees killed by the disease. To illustrate, counties are listed in four percent infection classes as follows: 1) Greater than 45 percent dead redbay trees - Bryan, Bulloch, Effingham, and Liberty, 2) 30-40 percent dead - Camden, Evans, McIntosh, and Screven, 3) 10-30% dead redbay - Charlton, Glynn, Long, Tattnall, and Wayne, and 4) Less than 10 percent Appling, Brantley, Clinch, Jenkins, Pierce and Toombs. Bacon, Candler, Emanuel, and Jeff Davis counties had no laurel wilt disease detected. 7 Table 1. Numbers of plots and plot means for percent of redbay trees dead from laurel wilt, percent live, percent dead from other causes, percent of plots with stump/stem sprouts and mean number of redbay regeneration (root sprouts and seedlings) by county in the 2007 Georgia laurel wilt survey. Redbay Plots Plot Means by County # % Other % with # Regen County Plots % LW Dead % Live Dead Sprouts per Ac Appling 16 8.1 91.0 0.9 18.8 884 Bacon 4 0.0 100.0 0.0 0.0 38 Brantley 10 1.3 97.0 1.7 0.0 355 Bryan 4 55.0 44.4 0.6 50.0 313 Bulloch 13 48.6 48.4 3.1 69.2 554 Camden 10 35.1 64.9 0.0 30.0 675 Candler 3 0.0 100.0 0.0 0.0 33 Charlton 10 21.8 78.2 0.0 20.0 145 Clinch 5 2.9 97.1 0.0 20.0 400 Effingham 8 71.0 28.4 0.6 75.0 894 Emanuel 6 0.0 97.2 2.8 16.7 258 Evans 8 35.9 63.2 0.8 50.0 813 Glynn 3 12.8 87.2 0.0 33.3 150 Jeff Davis 5 0.0 100.0 0.0 0.0 140 Jenkins 7 1.3 90.6 8.1 0.0 264 Liberty 5 50.6 49.4 0.0 80.0 630 Long 8 14.0 84.5 1.6 62.5 463 McIntosh 5 41.2 58.8 0.0 40.0 550 Pierce 7 5.9 92.4 1.7 14.3 357 Screven 11 36.2 59.8 3.9 36.4 359 Tattnall 11 15.6 83.8 0.6 0.0 132 Toombs 6 0.6 99.4 0.0 0.0 108 Ware 7 2.6 95.1 2.3 28.6 179 Wayne 16 19.0 75.3 5.7 37.5 538 Total 188 21.2 77.0 1.8 29.8 440 Relatively few redbay trees observed in this survey had clearly died from other causes with an average over all the plots of 1.8 percent. Stump sprouts were often numerous around dying and dead redbay trees, although many eventually die (Fig. 3). The proportions of plots with sprouts and amount of regeneration (root sprouts and seedlings combined) appeared to be significantly correlated with LW mortality rate (Table 2). There were fewer plots with stump sprouts and less regeneration in plots without LW and increasing amounts of both with higher levels of damage (Table 2). However, this relationship could also be confounded by the distribution of redbay trees and LW damage levels. Many of the plots without damage were located near the limit of redbay distribution where redbay trees and regeneration are scarce. 8 Table 2. Percent of redbay plots with stump sprouts and mean amount of regeneration/Ac by percent redbay mortality class among all 2007 LW survey plots. Mean number % of plots with root sprouts & % LW mortality stump sprouts seedlings/Ac None 8 325 1-33% 20 474 33-66% 78 622 > 66% 87 733 It has been documented that large redbay trees become infected and die more rapidly than smaller diameter trees (Fraedrich et al. 2008). This was also observed throughout this survey, with very few large redbay trees remaining alive where the disease had been present for a number of years. However, there were many relatively small redbay trees, mostly in the 1 to 4 inch DHB range, still alive in plots, even in very heavily infected areas. Also, disease severity was not uniform within the general area of infection. There were a few plots near Savannah in Chatham, Bryan, and Effingham counties that had no disease or only light infection rates. At least for now, redbay has not been eliminated from the ecosystem as has been suggested might occur. The question remains unanswered as to what will happen to these remaining healthy redbay trees over time. Comparison of 2006 and 2007 Laurel Wilt Infection Levels by County Infection levels in plots sampled in the 2006 and 2007 were compared to demonstrate the increase in rate of infection over a one year time frame. Due to land use alterations, inaccessibility, and inconsistencies in data collection, only a portion of the plots could be used for this comparison. A total of 47 plots were compared in 15 counties. The mean percent infection was computed for the plots in each county for comparison (Figure 4). These means derived from selected plots differ somewhat from the county averages among all plots reported for the 2007 survey, but this comparison serves to illustrate that the infection rate increased rapidly in these counties over a period of one year, more than doubling from an average of 21 percent infection in 2006 to 46 percent in 2007. Geographical Distribution of Laurel Wilt in Georgia The coordinates and severity code for each plot in the survey were loaded on the GFC GIS system and a map was produced with color coded dots indicating the plot locations and laurel wilt infection severity (black = none, green = light, yellow = moderate, and red = severe) in the 24 counties sampled in the 2007 grid survey (Figure 5). The locations of sassafras clumps encountered during the survey and other travel opportunities were plotted on the map as stars using the same relative severity color codes as for redbay plots. 9