<%BANNER%>

Ecology of flowering dogwood (Cornus florida L.) in response to anthracnose and fire in Great Smoky Mountains National P...


PAGE 1

ECOLOGY OF FLOWERING DOGWOOD ( Cornus florida L.) IN RESPONSE TO ANTHRACNOSE AND FIRE IN GREAT SMOKY MOUNTAINS NATIONAL PARK, USA By ERIC HOLZMUELLER A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLOR IDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2006

PAGE 2

Copyright 2006 by Eric Holzmueller

PAGE 3

iii ACKNOWLEDGMENTS I would like to express my thanks to my graduate committee chair, Dr. Shibu Jose, for his guidance, suggestions, and support in co mpleting this project. Thanks also go to my committee members, Drs. Alan Long, De bbie Miller, and Wendell Cropper, for their contributions to this study. Special thanks go to committee member Dr. Mike Jenkins for his input, help, and support with all phases of this project. I am also deeply grateful to the Great Smoky Mountains National Park Inve ntory and Monitoring Program for providing me with data collection a ssistance, and an office and housing during multiple field seasons. I would also like to thank the faculty at the IFAS-Statistics Division for their help in data analyses and Ken Ford for hi s assistance in the potted plant experiment. I would like to thank the National Pa rk Service Southeast Region Natural Resources Preservation Program, Great Smoky Mountains Association, and the University of Florida College of Agriculture and Life Sciences for providing funding for this project. Without their fi nancial support, this project would not have been possible. Finally I would like to thank my family and friends for their suppor t and encouragement.

PAGE 4

iv TABLE OF CONTENTS page ACKNOWLEDGMENTS.................................................................................................iii LIST OF TABLES.............................................................................................................vi LIST OF FIGURES.........................................................................................................viii ABSTRACT....................................................................................................................... ..x CHAPTER 1 INTRODUCTION........................................................................................................1 Statement of the Problem..............................................................................................1 Review of Literature.....................................................................................................2 Cornus florida .......................................................................................................2 Biology of Dogwood Anthracnose........................................................................3 Impacts of Dogwood Anthracnose........................................................................4 Factors Affecting Dogwood Anthracnose.............................................................5 Ecological Significance of Cornus florida ............................................................6 Specific Objectives.......................................................................................................7 2 INFLUENCE OF FIRE ON THE DENSITY AND HEALTH OF Cornus florida L. (FLOWERING DOGWOOD) POPULATIONS IN GREAT SMOKY MOUNTAINS NATIONAL PARK...........................................................................11 Introduction.................................................................................................................11 Materials and Methods...............................................................................................15 Study Site.............................................................................................................15 Field Sampling.....................................................................................................16 Data Analysis.......................................................................................................16 Results........................................................................................................................ .19 Cornus florida Stem Density...............................................................................19 Foliage Health and Crown Dieback....................................................................20 Stand Structure....................................................................................................21 Overstory Community Composition...................................................................21 Understory Community Composition.................................................................22 Importance Values for Understory Species.........................................................22 Tsuga canadensis Stem Density..........................................................................23

PAGE 5

v Discussion...................................................................................................................23 Management Implications..........................................................................................28 Conclusion..................................................................................................................29 3 INFLUENCE OF CALCIUM, POTASSIUM, AND MAGNESIUM ON Cornus florida L. DENSITY AND RESISTANC E TO DOGWOOD ANTHRACNOSE.....39 Introduction.................................................................................................................39 Materials and Methods...............................................................................................41 Study Site.............................................................................................................41 Forest Soil Sampling...........................................................................................42 Lab Analysis........................................................................................................42 Potted Plant Experiment......................................................................................43 Statistical Analysis..............................................................................................45 Forest Soil....................................................................................................45 Potted Plant Experiment...............................................................................45 Results........................................................................................................................ .45 Forest Soil Cation Saturation...............................................................................45 Potted Plant Experiment......................................................................................46 Calcium Treatments.....................................................................................46 Potassium Treatments..................................................................................46 Magnesium Treatments................................................................................47 Foliar Cation Concentrations.......................................................................47 Discussion...................................................................................................................48 Conclusion..................................................................................................................52 4 INFLUENCE OF Cornus florida L. ON CALCIUM MINERALIZATION IN TWO SOUTHERN APPALACH IAN FOREST TYPES...........................................64 Introduction.................................................................................................................64 Materials and Methods...............................................................................................66 Study Area...........................................................................................................66 Field Sampling.....................................................................................................67 Laboratory Analysis............................................................................................68 Statistical Analysis..............................................................................................68 Results........................................................................................................................ .69 Forest Floor Mass and Depth..............................................................................69 Soil pH.................................................................................................................69 Initial Exchangeable Ca.......................................................................................70 Ca mineralization.................................................................................................70 Discussion...................................................................................................................71 Conclusion..................................................................................................................76 5 SUMMARY AND CONCLUSION...........................................................................83 LIST OF REFERENCES...................................................................................................86 BIOGRAPHICAL SKETCH.............................................................................................97

PAGE 6

vi LIST OF TABLES Table page 2-1 Scales for foliage health (% of foliage with signs of anthracnose) and crown dieback (% of crown dieback) used to a ssess the level of di sease severity of dogwood anthracnose on Cornus florida trees ........................................................30 2-2 Mean foliage and crown health ( 1 SE) for Cornus florida for five diameter classes in the four different sampli ng categories using the Mielke-Langdon Index (Mielke and Langdon 1986)...........................................................................31 2-3 Mean understory and overstor y basal area and stem density ( 1 SE) in the four sampling categories..................................................................................................32 2-4 Mean species richness an d Shannons diversity index ( 1 SE) for the understory and overstory in the four sampling categories.........................................................33 2-5 Overstory and understory indicator values (IndVal) (percent of perfect indication) and associated sampling category..........................................................34 2-6 Mean importance values ( 1 SE) of selected underst ory species in the four sampling categories..................................................................................................35 3-1 Content of the base fertilizer mix.............................................................................53 3-2 Inputs (%) added to the base fertilize r mix (separately) for each treatment............54 3-3 Scale used to assess the severity of dogwood anthracnos e infection on the foliage of Cornus florida seedlings. Scale was ba sed on the Mielke-Langdon Index (Mielke and Langdon 1986)...........................................................................55 3-4 Biweekly infection ratings for Cornus florida seedlings for length of the experiment................................................................................................................56 3-5 Foliar calcium (Ca), potassium (K), a nd magnesium (Mg) concentrations (%) for selected species in a southern Appalach ian forest. Data presented is from Day and Monk (1977)......................................................................................................57 4-1 Mineral soil and fore st floor mean pH ( 1 SE) for summer and winter incubations in the cove hardwood and oak hardwood forest types..........................77

PAGE 7

vii 4-2 Mean yearly Ca mineralization ( 1 SE) in forest floor, mineral soil, and combined total (forest floor plus mineral soil) for the three Cornus florida sampling densities in the cove hard wood and oak hardwood forest types...............78 4-3 Foliar calcium concentrations (%) from dominant species in the two forest types (oak hardwood and cove hardwood) sampled in this study.....................................79

PAGE 8

viii LIST OF FIGURES Figure page 1-1 Native range of Cornus florida and the documented range of dogwood anthracnose in the eastern United States....................................................................9 1-2 The effects of a 1976 wildfire on Cornus florida stem density in Great Smoky Mountains National Park..........................................................................................10 2-1 Mean Cornus florida stem density ( 1 SE) in the four sampling categories for three diameter classes and total stems ha-1 for all diameter classes.........................36 2-2 Nonmetric multidimensional scaling sample ordination of overstory and understory communities, showing the relative differences in community composition separated by sampling categories........................................................37 2-3 Mean Tsuga canadensis stem density ( 1 SE) in five diameter size classes and total (all diameter classes combined ) in the four sampling categories.....................38 3-1 Linear regression between so il calcium (Ca) saturation and Cornus florida stem density and basal area...............................................................................................58 3-2 Linear regression between so il potassium (K) saturation and Cornus florida stem density and basal area...............................................................................................59 3-3 Linear regression between so il magnesium (Mg) saturation and Cornus florida stem density and basal area......................................................................................60 3-4 Biweekly mortality (%) of Cornus florida seedlings for the four treatment levels of each cation...........................................................................................................61 3-5 Foliar calcium (Ca), potassium (K), and magnesium (Mg) concentrations of Cornus florida seedling foliage for the four tr eatment levels of each cation...........62 3-6 Precipitation data for 2004 and prev ious 5 year aver age (1999-2003) during April-September at the Twin Creeks Natural Resources Center, Great Smoky Mountains National Park..........................................................................................63 4-1 Mean initial exchangeable Ca levels ( 1 SE) in the forest floor and mineral soil in the cove hardwood and oak hardwood forest types during summer and winter collection times .......................................................................................................80

PAGE 9

ix 4-2 Mean Ca mineralization ( 1 SE) for the forest floor and mineral soil in the cove hardwood and oak hardwood forest type s during summer and winter incubation periods......................................................................................................................81 4-3 Conceptual model of calcium (Ca) cyc ling in an eastern United States hardwood forest. Arrow thickness indicates amount of Ca movement and box size indicates size of available Ca pool..........................................................................................82

PAGE 10

x Abstract of Dissertation Pres ented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy ECOLOGY OF FLOWERING DOGWOOD ( Cornus florida L.) IN RESPONSE TO ANTHRACNOSE AND FIRE IN GREAT SMOKY MOUNTAINS NATIONAL PARK, USA By Eric Holzmueller May 2006 Chair: Shibu Jose Major Department: Forest Resources and Conservation Cornus florida L. (flowering dogwood), a common understory tree species in eastern forests, is currently threat ened throughout its range by a fungus ( Discula destructiva Redlin) that causes dogwood anthracnos e. This disease rapidly kills C. florida trees and mortality has exceeded 90% in some forest types. The health and ecological integrity of forest ecosystems throughout the eastern United States are threatened by the decline of C. florida populations, but management techni ques to control an thracnose have received little attention. Hence, the objectives of this project were to determine (1) the influence of past burning on C. florida density and health, (2 ) how nutrient levels influence the density and health of C. florida and (3) the role of C. florida in calcium (Ca) cycling. We examined C. florida populations in burned and unbur ned oak-hickory stands in Great Smoky Mountains National Park to de termine if burning pr ior to anthracnose

PAGE 11

xi infection has reduced the imp acts of the disease. Burned stands contained greater C. florida densities and lower disease se verity than unburned stands. Nutrient availability has been hypothesized as a factor that influences dogwood anthracnose severity on C. florida We studied the influence of Ca, potassium (K), and magnesium (Mg) on C. florida density and resistance to dogwood anthracnose. We found positive correlations between soil Ca, K, and Mg saturation and C. florida density in oakhickory forests. We also found that seedlings grown in soil with lower inputs of Ca and K cations exhibited higher disease severity ear lier in the growing season than seedlings grown with greater inputs of Ca and K. Cornus florida is thought to play an important role in the Ca cycle because of the high Ca concentration found in the folia ge. We quantified the relationship of C. florida density on Ca mineralization in the mineral so il and forest floor. Calcium mineralization occurred primarily in the forest floor and was generally greatest in the high density C. florida plots. Our research showed a positive correlation between C. florida density and soil Ca, K, and Mg saturation. Higher levels of soil Ca and K may alleviate disease severity in C. florida Further, our results indicate that pr escribed fire may provide an important management tool to reduce dise ase incidence and severity in oak-hickory forests. We also found that the loss of C. florida from eastern forests has reduced the rate of soil and forest floor Ca mineralization, which may have nega tive effects on many associated flora and fauna.

PAGE 12

1 CHAPTER 1 INTRODUCTION Statement of the Problem Over the last quarter century, a f ungal disease has severely impacted Cornus (dogwood) species along the Pacific seaboard and throughout the ea stern United States. Redlin (1991) identified the fungus Discula destructiva Redlin as the causal agent for the disease, dogwood anthracnose. Two species, Cornus nuttallii Aud. (Pacific dogwood) and Cornus florida L. (flowering dogwood), are th e most susceptible to dogwood anthracnose and have experienced heavy mort ality because of the disesae (Daughtery and Hibben 1994). Although not proven, it is believed that dogwood anthracn ose is an exotic disease brought to North America from Asia (Britton 1994). Because Cornus kousa L. (Oriental dogwood) is quite resi stant to anthracnose, it is suspected that the disease was brought from overseas on trees of this spec ies that did not show symptoms (Britton 1994). In a study of Discula genetic diversity, Trigia no et al. (1995) found that D. destructiva was highly homogenous acro ss its broad continental ra nge compared to other Discula species, suggesting that D. destructiva is a recently introduced fungus still undergoing intense selection pressure. The sudden appearance of the disease on both coasts of the United States and its rapid spread in the eastern half of the country support this assumption. Historically, C. florida was one of the most common understory trees in the eastern United States (Muller 1982, Elliott et al 1997, Jenkins and Parker 1998). However, dogwood anthracnose has severely impacted this species and heavy mortality of C.

PAGE 13

2 florida (attributed to dogwood anth racnose) has occurred throughout the eastern United States (Anagnostakis and Ward 1996, Sh erald et al. 1996, Hiers and Evans 1997, Schwegman et al. 1998, Carr and Banas 1999, W illiams and Moriarity 1999, Jenkins and White 2002). For example, Anagnostakis and Wa rd (1996) reported a mortality rate of 86% between 1977 and 1987 on long-term plots in Connecticut and a 77% reduction of stems was observed in Catoctin Mounta in Park in Maryland between 1976 and 1992 (Sherald et al. 1996). Currently, there ar e no management options for controlling dogwood anthracnose in large forested areas. Review of Literature Cornus florida Cornus florida is primarily an understory spec ies found in the eastern United States, from the southern tip of Maine to the northern half of Florid a and as far west as Oklahoma (McLemore 1990). Most common as a small understory tree, average height for C. florida ranges from 5-12 m and average st em diameter ranges from 3-8 cm. Although C. florida grows best on fertile well-drained soils, it can be found on Ultisols, Inceptisols, Alfisols, Spodosols, and Entisols (McLemore 1990) Cornus florida is a shade tolerant species and maximum photosynthesis occurs at about one-third of full sunlight (McLemore 1990) Cornus florida is a thin barked species that is easily damaged by logging and fire, however, it will prolifically stump sprout when damaged (Buell 1940, Kuddes-Fischer and Arthur 2002, Bla nkenship and Arthur 2006). Its large geographical range (Figure 1-1), ability to grow on a variety of sites, tolerance of shade, and sprouting ability enables C. florida to grow in association with a wide-range of tree species and in multiple stand conditions Cornus florida is commonly associated with Quercus species and mesic hardwoods such as Liriodendron tulipifera L., and also occurs

PAGE 14

3 under the shade of Pinus species (McLemore 1990) Cornus florida is common in many second growth forests, and is also common in old growth and undist urbed areas as well (Harrod et al. 1998, Jenkins and Parker 1998). Biology of Dogwood Anthracnose Dogwood anthracnose progressively attacks all aboveground parts of infected trees. Although the disease can affect trees of any size, smaller trees are more susceptible (Mielke and Langdon 1986, Hiers and Evans 1997). The edges of the l eaves on the lower branches show the first symptoms of the di sease, developing black spots that extend down the leaf margins under suitable conditions (Britton 1994). Anthracnose then spreads from the leaves into the tw igs of the infected trees where the fungus overwinters. In the spring, reproduc tive structures of D. destructiva form underneath leaf spots and on the surface of twig cankers. Numerous asexual spores ooze out in slimy beige droplets from buds and twigs. Local dispersa l of the spores occurs by splashing rain, while longer distance dispersal may be facili tated by insects and birds (Sherald et al. 1996). The fungus eventually reaches the bole of the tree where cankers develop, girdling and killing the tree (Mielke and Langdon 1986). Trees also die from repeated defoliation, with smaller trees dying first (Britton 1994) Trees may die within 1-5 years of first infection; saplings may die in th e same year they are infected. There are many signs of dogwood anthrac nose infection that can be readily identified. Leaves of infected trees typically have one or both of two types of leaf spots. Irregular light brown spots with reddish brown borders are formed when environmental conditions are less conducive to the fungus. These may be concentrated around the lower edges of the leaf, but may also appear s cattered across the lamina. Under favorable disease conditions, the other type of leaf spot forms, a leaf blight with black, water-

PAGE 15

4 soaked lesions typically initiating at the leaf tip and expanding upward along the midvein into the twig. Infected trees may disp lay an umbrella-like ca nopy due to the loss of lower branches (Mielke and Langdon 1986, Da ughtrey et al. 1996). In place of lower branches, epicormic shoots often develop, but are quickly infected with anthracnose. Since C. florida has the ability to produce stump sprouts, a dead stem surrounded by many sprouts is a common occurrence followi ng anthracnose-caused mortality. These new shoots, however, are typically infected a nd will not likely mature to replace the dead tree. Impacts of Dogwood Anthracnose Since its first appearance in New York in 1978 (Pirone 1980), the spread of anthracnose throughout the eastern half of the United States has been rapid (Figure 1-1). The disease was widespread in nine northeas tern States by 1987 (Connecticut, Delaware, Maryland, Massachusetts, New Jersey, New Yo rk, Pennsylvania, Virginia, and West Virginia) (Anderson 1991). By 1992 the disease had reached the Carolinas, Georgia, Alabama, Tennessee, and Kentucky (Knight en and Anderson 1993). Daughtrey et al. (1996) reported that the disease had also move d west to the states of Missouri, Illinois, Indiana, Ohio, and Michigan. A recent study by Wyckoff and Clark (2002) at the Coweeta Hydrologic Laboratory in the southern Appalachian M ountains documented the rapid decline of C. florida with a mortality rate of 15% within a 5 year (1993-1998) period. In Great Smoky Mountains National Park (GSMNP), analysis of long-term vegetation monitoring data revealed several alarming trends (Jenkins and White 2002). Between the two sampling intervals (1977-1979 and 1995-2000) dramatic decreases in C. florida stem density were observed in typic cove, acid cove, alluvial oak-hickory, and oak-pine forest types.

PAGE 16

5 Overall decline was greatest in ac id cove forests, where mean C. florida stem density decreased by 94% (101 stems ha-1 to 6 stems ha-1). Typic cove and alluvial forests exhibited the next greatest decreases in C. florida stem density (92% for both forest types). Mean density decreased from 180 stems ha-1 to 14 stems ha-1 in typic cove forests and from 364 stems ha-1 to 28 stems ha-1 in alluvial forests. The two driest forest types, oak-hickory and oak-pine, exhibited declin es of 80% and 69%, respectively. Prior to anthracnose, C. florida stem density in oak-hickory fore sts was the second highest of any forest type (298 stems ha-1), however, mean density decreased to 61 stems ha-1 in 19952000. Mean density decreased from 132 stems ha-1 to 41 stems ha-1 in oak-pine forests. New C. florida seedlings are not replacing dead C. florida trees in infected stands. Trees infected with anthracnose produce fe wer seeds (Rosell et al. 2001), and reduced seed production combined with the susceptibi lity of smaller trees to the disease has drastically decreased regeneration. Seedlings and saplings were reported to be absent in several studies (Sherald et al. 1996, Hiers and Evans 1997), indi cating the severity of the problem. Factors Affecting Dogwood Anthracnose Many environmental variables influence the spread and vi rulence of dogwood anthracnose, but moisture is probably the most critical (Britt on 1993). The disease is more severe on shaded and moist northeast-f acing slopes than on southwest-facing slopes with open canopies (Chellemi et al. 1992, Clinton et al. 2003). In a long-term vegetation study in GSMNP, Jenkins and White (2002) report ed higher levels of mortality attributed to dogwood anthracnose in more mesic forest s (typic and acid coves and alluvial communities) compared to more xeric forest types (oak-hickory and oak-pine communities). Jenkins and White (2002) al so reported an increased number of C. florida

PAGE 17

6 stems on three plots after a 1976 wildfire (Fi gure 1-2), indicating stand conditions can be manipulated to increase C. florida survival from dogwood anthracnose. However, the effects of past burning on C. florida populations have not be en fully investigated. Although moisture may be the most important environmental variable affecting the impacts of dogwood anthracnose on C. florida another key variable that has yet to be fully explored is nutrient availability. In a greenhouse study by Britton et al. (1996), simulated acid rain increased the susceptibility of C. florida to anthracnose. The authors hypothesized that the in crease in susceptibility was largely because of soil-mediated impacts that reduced the availability of so il cations. Calcium (Ca) potassium (K), and magnesium (Mg) are nutrients that have been linked to disease resistance in other plant species and diseases (Sij et al. 1985, Anglberger and Halmschl ager 2003, Sugimoto et al. 2005). Sugimoto et al. ( 2005) found a reduction of Phytophthora sojae Kaufmann and Gerdemann (stem rot) with the app lication of Ca on two cultivars of Gycline max L. Merr. cv. Chusei-Hikarikuro (black soybean) and cv. Sachiyutaka (white soybean) in Japan. Sij et al. (1985) reported that increased rates of K fer tilizer significantly decreased Colletotrichum dematium (Pers.ex Fr.) Grove var. truncata (Schuv.) Arx. (anthracnose) in field grown G. max plants in Texas. In Austria, severity of Sirococcus conigenus (shoot blight) was increased on Picea abies (Norway spruce) trees that had needles with low levels of Mg (Anglberger and Halmschlag er 2003). It is possible that bioavailability of soil Ca, K and Mg plays a role in resistance to dogwood anthracnose in C. florida as well. Ecological Significance of Cornus florida The loss of C. florida from stands throughout the eastern United States will likely have serious ecological effects. An individua l tree may produce up to 10 kg of fruit each

PAGE 18

7 fall (Rossell et al. 2001). There are more than 50 species of birds, including neotropical birds during fall migration, as well as a numbe r of small game species that are known to eat the fruit (Martin et al. 1951, Stiles 1980) Cornus florida twigs are also an important source of browse for white -tailed deer and other he rbivores (Blair 1982). Cornus florida is also important in nutrient cycling within forest communities. In eastern mixed hardwood forests, Ca releas ed through mineral weathering is generally insignificant (Huntington et al. 2000, Dijkstra and Smits 2002). As a re sult, the re lease of Ca through organic matter decomposition (minera lization) is considered the major source of Ca for immediate plant uptake for all tree species (Dijkstra and Smits 2002). Dijkstra (2003) reported that for most species, Ca mineralization beneath the canopy of a given species primarily occurs in th e forest floor (from leaf litte r) as opposed to the mineral soil. Decomposition of C. florida foliage is very rapid compar ed to other species (Blair 1988, Knoepp et al. 2005), and C. florida litter contains high concentrations (2.0-3.5%) of Ca (Thomas 1969, Blair 1988). Because of th e high Ca concentration in its foliage, quick decomposition, and a bundance in the understory, C. florida has long been believed to influence Ca availability in the mineral soil and forest floor by acting as a Ca pump in forests (Thomas 1969, Jenkins et al 2006). High Ca concentration in C. florida foliage could mean potentially high rates of Ca minera lization in the forest floor and mineral soil If high mineralization rates occur, there is a potential for high Ca availability in the soil. Apart from anecdotal evidence, the impacts of C. florida on Ca cycling, however, have not been quantified. Specific Objectives As discussed previously, the ro les of past burning and bioa vailability of nutrients in determining C. florida population dynamics follo wing infection with dogwood

PAGE 19

8 anthracnose have not been fully investig ated. In addition, whether higher foliar Ca concentration in C. florida foliage translates into higher ra tes of Ca mineralization in the forest floor and mineral soil remains undeterm ined. Therefore, the objectives of this research project were to: 1. Determine the influence of past burning on (1) C. florida density and health and (2) overall stand structure and species composition in oak-hickory forests where C. florida is historically found. 2. Examine the effects of soil cation availa bility (Ca, K, and Mg) on the health and survival of C. florida. 3. Determine the relationship between C. florida density and Ca mineralization rates in two forest types (cove hardwood and oak hardwood) where C. florida is a common understory species. The following three chapters describe the results of field surveys and experiments conducted in Great Smoky Mountains National Park to address these three objectives.

PAGE 20

9 Figure 1-1. Native range of Cornus florida and the documented range of dogwood anthracnose in the eastern United States (Based on data from the U.S. Forest Service Southern and Northeastern Forest Research Stations). Native range of Cornus florida Counties reporting dogwood anthracnose

PAGE 21

10 0 200 400 6001-4.9 cm5.0-9.9 cm 1-4.9 cm5.0-9.9 cm Diameter class (cm) by burn treatment Cornus florida density (stems ha-1) 1979 2000 UnburnedBurned Figure 1-2. The effects of a 1976 wildfire on Cornus florida stem density in Great Smoky Mountains National Park. While stem density decreased drastically in unburned areas during the 21 year study period (P < 0.001), density increased (245%, P = 0.159) in the 1.0-4.9 cm class and remained stable in the 5.0-9.9 cm class in burned areas (P = 0.334) (Based on data from Jenkins and White 2002).

PAGE 22

11 CHAPTER 2 INFLUENCE OF FIRE ON THE DENSITY AND HEALTH OF Cornus florida L. (FLOWERING DOGWOOD) POPULATION S IN GREAT SMOKY MOUNTAINS NATIONAL PARK Introduction Historically, Cornus florida L. (flowering dogwood) wa s one of the most common understory species in eastern United States hardwood forests (Muller 1982, Elliott et al. 1997, Jenkins and Parker 1998). According to McLemore (1990), C. florida occurs on a variety of soils (Ultisols, Inceptisols, Alfiso ls, Spodosols, and Entisols), is shade tolerant (maximum photosynthesis occurs at about one-third of full sunlight), and has a large geographical range (southern Ma ine to northern Florida and as far west as Oklahoma). These factors enable the species to grow in association with a wide-range of tree species including Quercus species, Pinus species, and Liriodendron tulipifera L. Cornus florida is most often found in post-logged secondary fore sts, and also occurs within tree-fall gaps of old growth forests (Muller 1982, Jenkins a nd Parker 1998). Because of its ability to grow in shaded conditions and thin bark, C. florida is not considered a fire dependent species. However, it sprouts pr olifically when top-killed by fire (Kuddes-Fishcher and Arthur 2002, Blankenship and Arthur 2006). Cornus florida is an ecologically important spec ies in forests throughout the eastern United States. An individual C. florida tree may produce up to 10 kg of protein-rich fruit each fall, and more than 50 species of birds as well a number of small mammal species are known to eat the fruit (Martin et al. 1951, Rossell et al. 2001). However, C. florida s most important role may be in the annual cycling of calcium in forest communities.

PAGE 23

12 Cornus florida foliage decomposes very rapidly compared to other species, and C. florida litter contains large amount s of calcium (2.0-3.5%) (Thomas 1969, Blair 1988, Knoepp et al. 2005). As a result, C. florida serves an important ecol ogical function as a calcium pump for associated plant sp ecies and forest floor biota. Dogwood anthracnose, a dis ease caused by the fungus Discula destructiva Redlin (believed to be an exotic disease from Asia) has become a serious disease of C. florida over the past 20 years. Dogwood anthracnose was first identified in the late 1970s in New York (Pirone 1980), and has since spread ra pidly throughout the eastern United States, infecting C. florida populations throughout mu ch of its range (Hol zmueller et al. 2006). Following infection by anthrac nose, mortality rates of C. florida have been as high as 90% (Anagnostakis and Ward 1996, Sherald et al. 1996, Hiers and Evans 1997, Jenkins and White 2002). Although the disease can infect trees of any size, smaller trees are more susceptible than large trees and often die fr om repeated defoliation within 1-5 years of infection (Mielke and Langdon 1986, Hibben and Daughtery 1988). The fungus also causes twig dieback and stem cankers, which can eventually girdle the tree (Hibben and Daughtery 1988). The rapid spread of dogwood anth racnose and high mortality of C. florida make it somewhat similar to the effects of Cryphonectria parasitica (Murill) Barr (Asian chestnut blight fungus) on Castenea dentata Marsh. (American chestnut). Cryphonectria parasitica was introduced in New York City in th e early 1900s and spread rapidly down the Appalachian Mountains, infecting C. dentata throughout most of its range by 1926 and effectively extirpating the species by the 1950s (Anagonostakis 2001). Whereas the effects of C. parasitica on C. dentata were fairly uniform in all forest types and stand

PAGE 24

13 conditions, disease severity of dogwood anthr acnose is influenced by many variables, particularly shading and moisture (Britton 1993 ). The disease is more severe on shaded and moist northeast-facing slopes than on rela tively drier southwest facing slopes with open canopies (Chellemi and Britton 1992, Che llemi et al. 1992). Britton (1993) reported that given adequate amounts of rainfall, the disease could develop throughout the growing season. In addition, Br itton et al. (1996) found that acid rain increases the susceptibility of C. florida to anthracnose. Since dogwood anthracnose is known to be sensitive to environmental characteristics, there is a possibility that management techniques could be used to manipulate stand structure to reduce the impacts of dogwood anthracnose. However, research examining the effects of stand ma nipulations on dogwood anthracnose in forest stands has been limited. Britton et al. (1994) examined the effect of past silvicultural practices on C. florida density and disease severity and reported C. florida densities were highest and disease severity lo west in stands that had been clearcut in 1939 and again in 1962. This effect was attributed to increased light levels in these plots. Prescribed burning may also offer a technique to manipulate stand conditions to favor C. florida survival in infected stands. Jenkins a nd White (2002) reporte d a 200% increase in C. florida stem density on three long-term vegeta tion plots that burne d in 1976 in Great Smoky Mountains National Park (GSMNP), however, the effects of burning on stand structure and C. florida health and survival we re not fully explored. Dogwood anthracnose was first reported in GSMNP in March of 1988 when a park-wide survey revealed that 23 out of 58 plots that contained C. florida were infected with anthracnose (Windham and Montgom ery 1990). Seven years later, in a study

PAGE 25

14 conducted in the western half of GSMNP, W ilds (1997) observed signs of anthracnose on 98% of the plots on which C. florida occurred. By 2000, C. florida mortality attributed to dogwood anthracnose ranged from 69% in oak-pi ne stands to 94% in acid coves on longterm vegetation plots located within GS MNP (Jenkins and White 2002). Prior to anthracnose, C. florida was the dominant understory woody sp ecies in oak-hickory forest of GSMNP. Following anthracnose infection, C. florida density decreased by 80% in this forest type. Within eastern North America, oak-hic kory forests comprise over 34% of total forest cover (Smith et al. 2001). Within this forest type, fire was historically common and influenced species composition and stand stru cture (Brose et al. 2001). Because fire has played an important ecological role in this widely distribu ted forest type, we undertook a study with the following objectives: to dete rmine the influence of past burning on (1) C. florida density and health and (2) overall sta nd structure and species composition in oakhickory forests in GSMNP. We hypothesize that past burning has altered stand conditions (structure and species com position) in ways that redu ced the impacts of dogwood anthracnose compared to unburned stands. Burning typically reduces stand density, increases light penetratio n through the canopy, and decrea ses understory moisture content, which, we hypothesize, favors C. florida survival. These conditions will not last indefinitely, however, and we further hypothesize that repe ated burning is needed to maintain stand conditions that reduce the impacts of the disease. Because C. florida is a thin barked species that is frequently t op-killed by fire, we further hypothesize that C. florida will display reduced surviv al once a threshold of burni ng frequency is reached.

PAGE 26

15 Materials and Methods Study Site This study was conducted in GSMNP, USA, which encompasses slightly over 200 000 ha, and straddles the Tennessee and No rth Carolina state line. The Park is internationally renowned as a center of biological diversity within North America and was designated as an International Biosphere Reserve in 1976. Due to its biotic diversity, large size, and protected status, changes that occur within the biological communities of GSMNP often serve as baselines for comparis on to other state and federal lands. Mean annual temperature in Gatlinburg, Tennessee (440 m a.s.l. and adjacent to GSMNP) is 12.9 C and mean annual precipitation is 142 cm. Our study sites ranged in elevation from 287 to 975 m. Although C. florida occurs in a variety of forest types, we focused our study in oak-hickory forests. These fore sts were typically found on moderately steep to steep slopes with southeastsouth-northwest facing aspects. Cornus florida is a common understory species in the oak-hickory forest type, and this type has frequently burned in some areas of GSMN P over the last 30 years. During June-August of 2001-2004, we sampled seventy-nine 0.04 ha (20 m x 20 m) plots in burned and unburned stan ds. Fifty-five plots were established in fourteen stands that had burned up to three times over a 20 ye ar period (late 1960s to the late 1980s). In addition, twenty-four plots were establishe d in six unburned stands. These areas were divided into four sampling categ ories: (1) single burn (seven stands, thirty plots), (2) double burn (four stands, sixteen plots), (3) triple burn (three stands, nine plots), and (4) unburned (six stands, twenty-four plots). We used historic park maps and fire hi story records to sel ect burned (single, double, and triple burns) and unburned areas. Within each burned area, we selected

PAGE 27

16 stands from vegetation associ ations within the Montane Oa k-Hickory Forest Ecological Group (White et al. 2003). Associat ions within this group contained C. florida as a major understory component prior to the onset of dogwood anthracnose (White et al. 2003). Individual polygons (stands) of each associa tion were derived from vegetation maps of GSMNP based upon 1:12000 color-IR aerial photos (Welch et al. 2002). Unburned (reference) plots were esta blished in nearby areas with similar slopes, aspects, topography, and vegetation associations as the burn plots. All burns selected were at least 10 hectares in size. Plots were located with in the burn areas by placing a 50 m buffer inside of each area, and randomly selecting pl ot coordinates within appropriate vegetation associations. A minimum of three 20 x 20 m pl ots were placed within each burn area, with up to five plots placed in larger burns. Field Sampling We recorded the diameter at br east height (dbh) of all living overstory stems (> 10.0 cm dbh) by species to th e nearest 0.1 cm. Living stems 10.0 cm dbh (understory) were tallied by species into f our diameter classes: 0-1.0 cm, 1.1-2.5 cm, 2.65.0 cm, and 5.1-10.0 cm. We measured the dbh of all C. florida stems, regardless of overstory or understory classi fication, to the nearest 0.1 cm Foliage and crown health were assessed for each living C. florida stem using the Mielke-Langdon Index (Mielke and Langdon 1986; Table 2-1). Data Analysis Data were analyzed for differences in the four sampling categories (single burn, double burn, triple burn, and unburned) fo r the following response variables: C. florida stem density, C. florida foliar and crown health, overstory basal area and stem density, understory basal area and stem density, unde rstory species importance values [IV =

PAGE 28

17 ((relative density + relati ve basal area)/2)*100], Tsuga canadensis (L.) Carr. stem density, plot species richness, and speci es diversity (Shannons diversity index) Cornus florida stem density differences among the four sampling categories were analyzed in three diameter classes: 0-5.0 cm, 5.1-10.0 cm, and > 10.0 cm. Total C. florida stem density of the four sampling categories was al so analyzed. Because smaller trees are more susceptible to dogwood anthracnose (Hiers and Evans 1997, Jenkins and White 2002), C. florida foliage and crown health were analyzed in five diameter size classes that better represented smaller diameters: 0-1.0 cm 1.1-2.5 cm, 2.6-5.0 cm, 5.1-10.0 cm, and > 10.0 cm. Tsuga canadensis stem density was analyzed in five diameter size classes (0-2.5 cm, 2.6-5.0 cm, 5.1-10.0 cm, 10.1-20.0 cm, and > 20.0 cm), plus total stem density. Before statistical comparison, all data (ex cept categorical) were tested for normality using the Kolmogorov-Smirnov goodness-of-fit te st for normal distribution. Only the C. florida stem density data violated the test for normality (P < 0.05). These data were natural log transformed to improve normality and equalize variances. For ease of interpretation, non-tr ansformed values are presente d. All response variables were analyzed with analysis of variance (ANOVA) using the Mixed procedure in SAS (SAS 2002). The model was made up of two factor s. The first factor was fixed, sampling category, and the other was random, burn area nested within sampling category. When ANOVA revealed a clear differe nce between the sampling categories, we used the PDIFF option for post-hoc pairwise comparisons (S AS 2002). All means presented in the paper are least square means calculated by SAS using the mixed procedure. To test for differences in overstory and understory community composition in the four sampling categories we used MRPP (Multi-Response Permutation Procedures;

PAGE 29

18 Biondini et al. 1985, Lesica et al. 1991, Peterson and McCune 2001). MRPP is a nonparametric procedure to test for multivaria te differences in pre-existing groups (i.e. single burn, double burn, triple burn, and unburned stands) (Mie lke 1984). It provides a test statistic ( A ) and P-value to determine whether the sampling categories occupied the same regions of species space ( A measures within-group agreement, if A = 1 then items within groups are identical and 1 is the highest possible value for A A = 0 when heterogeneity within groups equa ls expectation by chance, and A < 0 with less heterogeneity within groups than expected by chance). Interpretation of this test statistic was done using nonmetric multidimensional scali ng (NMS) and indicato r species analysis (IndVal) (Peterson and McCune 2001, McCune and Grace 2002). NMS is an ordination technique that is ideal fo r data that are nonnormal or nonlinear and contains large numbers of zero values. It uses ranked distances to linearize the relation of degrees of difference between community samples and di stances on an environmental gradient, and is the most effective ordination technique available for community data (Clarke 1993, McCune and Grace 2002). IndVal is used to describe the relationship of species to categorical variables by combining species a bundance in a specific category plus the faithfulness of occurrence of that species in that specific category (Dufrne and Legendre 1997, Qian et al. 1999, Peterson and McCune 2001). The analysis pr oduces a value of abundance for each species in each group (IndVal ) and a test statistic produced from Monte Carlo tests (1000 itera tions) to determine if o ccurrence in the maximum (indicator) group is greater than w ould be expected from chance. All multivariate analyses were performed using PC-ORD (McCune and Medford 1999). To reduce the effects of rare species, we deleted those species occurring in less

PAGE 30

19 than three plots prior to multivariate analyses ; eight species were deleted from overstory analyses and 16 species were deleted from understory analyses (McCune and Medford 1999). MRPP and NMS were performed using the quantitative version of Srensons distance measure and NMS ordination was di splayed on two axes. Additional axes did not significantly improve the expl anatory power of the ordination. Results Cornus florida Stem Density Cornus florida stem density differed significantly among sampling categories in the smallest (0-5.0 cm) size class (double burn = 691 stems ha-1, triple burn = 175 stems ha-1, single burn = 154 stems ha-1, and unburned stands = 35 stems ha-1, P = 0.0015; Figure 21). Double burn stands contained four times more C. florida stems than single burn stands (P = 0.05) and twenty times more C. florida stems than unburned stands (P = 0.0002). The stem density of double burn stands was not significantly different from that of triple burn stands (P = 0.39). The single burn stand and triple burn st and stem densities in this size class were not significantly different (P = 0.39), but stem densities in both of these categories were significantly greater than the unburned stand stem density (P < 0.007). There was no statistical differen ce between sampling categories in C. florida stem density in the 5.1-10 cm size class (P = 0.229; Fi gure 2-1). Although differences were not significant due to high inter-p lot variability (P = 0.167), doubl e and triple burn stands contained greater densities of st ems >10 cm dbh (22 and 17 stems ha-1, respectively) than unburned and single burned stands (5 and 6 stems ha-1, respectively). Total stem density of C. florida differed significantly among the four sampling categories (double burn = 770 stems ha-1, triple burn = 233 stems ha-1, single burn = 225 stems ha-1, and unburned stands = 70 stems ha-1, P = 0.0003; Figure 2-1). This difference

PAGE 31

20 can be largely attributed to the greater percen tage of smaller trees in burned stands. Trees in the 0-5.0 cm size class contri buted 90% of the total stems ha-1 in double burn stands, 75% in triple burn stands, 68% in single bur n stands, and just 50% in unburned stands. We observed a significantly greater total dens ity of stems in double burn stands than in single burn stands (P = 0.036) and unburned stands (P < 0.0001), but triple burn stands did not differ significantly from double burn st ands (P = 0.53). Single burn and triple burn stands were not significantly diffe rent from each other (P = 0.26), but both categories were significantly greater than the unburned stands (P < 0.001). Foliage Health and Crown Dieback Overall, mean foliage health ranged from 3.1 to 3.9 for all size classes and sampling categories, and we did not observe any differences in the foliar health among the sampling categories in each of the five si ze classes (P > 0.38; Table 2-2). Mean crown dieback ratings of all size classes ranged from 2.4 to 3.7, and ther e were no significant differences in crown health in three of the size classes (0-1.0 cm, P = 0.64; 5.1-10.0 cm, P = 0.45; and > 10.0 cm, P = 0.93; Table 2-2). Howeve r, in two of the smaller size classes (1.1-2.5 cm and 2.6-5.0 cm) there was a signif icant difference in mean crown dieback ratings among the four sampling categories (P = 0.04 and P = 0.01, respectively; Table 22). Further analyses in the 1.1-2.5 cm size cla ss showed that the rating of unburned stands (2.4) was significantly lower (less healthy) th an those of the burned stands (3.3 3.4, P < 0.02). In the 2.6-5.0 cm size class, mean crown dieback ratings for burned stands (3.6 3.2) were significantly higher (healthier) than th at of unburned stands (2.7, P < 0.04). Differences among the sampling categories were not significant for the 0 0.1 cm class, however, plots in unburned stands did not contai n any living trees in this size class.

PAGE 32

21 Stand Structure Overstory (> 10.0 cm dbh) basal area was similar among the four sampling categories (single burn = 21.7 m2 ha-1, double burn = 22.1 m2 ha-1, triple burn = 20.6 m2 ha-1, and unburned stands = 23.2 m2 ha-1, P = 0.52; Table 2-3). Ov erstory stem density differed significantly among the four samp ling categories (single burn 436 stems ha-1, double burn 323 stems ha-1, triple burn 317 stems ha-1, and unburned stands 564 stems ha-1, P < 0.0001; Table 2-3). Comparisons of ove rstory stem density among the four sampling categories revealed that double and triple burn stands had similar stem densities (P = 0.92). Densities in these tw o categories were significantly lower than that of single burn stands (P < 0.05) and unburned stands (P < 0.001). Finally, single burn stands had significantly fewer stems than unburned stands (P = 0.016). In the understory ( 10.0 cm dbh), basal area was similar between sampling categories, ranging from 6.5 7.4 m2 ha-1 (P = 0.93; Table 2-3), while stem density was significantly different (s ingle burn 2851 stems ha-1, double burn 4594 stems ha-1, triple burn 5072 stems ha-1, and unburned stands 2292 stems ha-1, P=0.024; Table 2-3). Comparisons of understory stem density am ong the four sampling categories revealed that double and triple burn stands had similar densities (P = 0.66), as did single burn stands and unburned stands (P = 0.46). Doubl e and triple burn stands had significantly greater understory stem densities than the single burn and unburne d stands (P < 0.05). Overstory Community Composition Shannons diversity index did not differ significantly in the overstory in the four sampling categories (P = 0.48; Ta ble 2-4). However, species ri chness of plots in the four sampling categories did differ slightly (P = 0.06), and was greatest in unburned stands compared to burned stands (6.9 versus < 5.7, respectively; Table 2-4). MRPP indicated

PAGE 33

22 that species composition differed slightly between plots in th e sampling categories (MRPP: P < 0.0001, A = 0.09). Three species were asso ciated with unburned stands ( Acer rubrum L., Oxydendrum arboreum (L.) DC., and T. canadensis ) and one each for double burn stands ( Quercus alba L.) and triple burn stands ( Quercus velutina Lam.) (IndVal: P < 0.08 each; Table 2-5). These differences, how ever, were not strong enough to clearly separate the sampling categories in the ordination (Figure 2-2). Understory Community Composition Shannons diversity index and species ri chness did not differ in the understory among the four sampling categories (P = 0.81 and P = 0.48, respectively; Table 2-4). Species composition did differ slightly be tween the sampling categories (MRPP: P < 0.0001, A = 0.06). Numerous species were indicativ e of sampling categories, primarily in the triple burn stands (unburned: T. canadensis ; double burn: C. florida and Tilia americana L.; triple burn: Carya alba L., Carya glabra Mill., Pinus virginiana Mill., Q. alba Quercus prinus L., Quercus rubra L., Q. velutina and Robinia pseudoacacia L.; Table 2-5). These differences, however, were not strong enough to cl early separate the sampling categories in the ordination (Figure 2-2). Importance Values for Understory Species In addition to greater stem densities, mean importance value (IV) of C. florida was four times greater in double burn stands than in unburned stands (IV = 21.6 versus 5.1, P = 0.001; Table 2-6) Cornus florida importance values were also significantly greater in single burn (IV = 12.1) and triple burn (IV = 14.8) stands compared to unburned stands (IV = 5.1, P = 0.05) Cornus florida had the greatest importance value of any species in double burn stands and second highest in the triple and single bur n stands. Six other species had higher importance values than C. florida in unburned stnads, including T.

PAGE 34

23 canadensis which was ten times greater in importa nce in unburned stands compared to burned stands (P < 0.05; Table 2-6). In triple burn stands, the importance values for A. rubrum (IV = 7.8) was significantly lower (P = 0.03) th an that of single burn sta nds (IV = 24.2). Also within triple burn stands, C. glabra and C. alba L. importance values were three times greater than in any other category (P < 0.004 and P < 0.01, respectivel y). In addition, the R. pseudoacacia importance value was three times greater in triple burn stands compared to the other sampling categories (P < 0.04) and the importance value of P. virginiana was greatest in the triple burn stands. Tsuga canadensis Stem Density Overall, total T. canadensis stem density was significantly greater in unburned stands (216 stems ha-1) compared to single burn (42 stems ha-1), double burn (23 stems ha-1), and triple burn stands (11 stems ha-1, P < 0.001; Figure 2-3). Most of the T. canadensis stems were in the smallest (0-2.5 cm) size class, T. canadensis stem density was over four times greater in unburned stands (83 stems ha-1) compared to single (18 stems ha-1), double (14 stems ha-1), and triple burn stands (11 stems ha-1, P < 0.001; Figure 2-3). Unburned stands had significantly more T. canadensis stems ha-1 in the 2.55.0 cm, 5.1-10.0 cm, and 10.1-20.0 cm size classes as well (P < 0.005). In the largest diameter classes (> 20 cm), single burn stands and unburned stands were similar (4 and 6 stems ha-1, respectively, P = 0.49). This size class was absent on plots in double and triple burn stands. Discussion The results of our study demonstrate the potential role of fi re in regulating population dynamics of C. florida in post-anthracnos e stands. Overall, burned stands in

PAGE 35

24 our study had greater C. florida stem densities, C. florida trees with heal thier crowns, and higher C. florida importance values than unburned stands. The greater C. florida stem densities and healthier C. florida trees in burned stands are likely the result of reduced shading following the burns that created a re latively drier microclimate that was less favorable to D. destructiva Studies have shown that shaded conditions increase the severity of dogwood anthracnose (Gould and Peterson 1994, Erbaugh et al. 1995). For example, Chellemi and Britton (1992) re ported an inverse relationship between evaporative potential an d disease severity on C. florida in the southern Appalachians. Discula destructiva growth has been found to be greater under moist conditions. In a study involving artific ial inoculation of C. florida with dogwood anthracnose, Ament et al. (1998) reported that D. destructiva lesions on C. florida leaves were five times larger when placed inside moistened bags for seven days compared to lesions that developed on leaves that spent four, two, and zer o days inside moistened bags. In our study, the greatest densities of C. florida stems were found in double burn stands. Although single bur n stands had greater C. florida stem densities than unburned stands, it appears that a singl e burn within a 20 year period is not sufficient to maintain C. florida Increases in overstory stem densities fo llowing a single burn appear to provide sufficient shading for anthracnose to increase in virulence. Studies have shown repeated burns better maintain lower overstory stem densities in oak forests (Huddle and Pallardy 1996, Peterson and Reich 2001, Hutchinson et al. 2005), which favors C. florida survival from dogwood anthracnose. Our results indicate however, that benefits of multiple burns are reduced when burning is increased to thr ee burns in a 20 year period. It appears that although larger (> 5 cm) C. florida trees survived the third burn, smaller trees displayed

PAGE 36

25 less resprouting after the third burn. This re duction in resprouting ma y be attributed to fire induced mortality of smaller stems, as opposed to lack of root carbohydrates, resulting in fewer stems per hectare capabl e of producing sprouts (Arthur et al. 1998), and suggests that this was too short an in terval between burns. Increased importance values of R. pseudoacacia and P. virginiana on triple burn stands, (indicator species of triple burn stands as well), suggest that the third burn sh ifted stands towards an earlier successional composition (Harrod et al. 1998). This shift to an earlier successional composition may perhaps be a nother reason for decreased C. florida stem densities in triple burn stands Although there were some differences in overstory community composition among the sampling categories, these differences were not strong enough to classify the sampling categories as unique communiti es. However, the identification of A. rubrum O. arboreum and T. canadensis as indicator species in unburne d stands suggest that this sampling category is shifting to a later succession al stage. This is not surprising, however, considering the lack of disturbance in these stands for the past 80-100 years. In the understory, we found that differ ences in overall community composition were present, but limited to a few species, such as C. florida and T. canadensis We observed higher understory stem densities in multiple burn stands, which is a result of reprouting by deciduous trees a nd is consistent with othe r studies (McGee et al. 1995, Elliott et al. 1999, Kuddes-Fischer and Arthur 2002). However, while burned stands had greater total understory stem densities, fire decreased the density and importance of T. canadensis in the understory of burned stands. In fire suppressed stands, this species often dominates the understor y and produces a dense sub-ca nopy that results in moist,

PAGE 37

26 heavily shaded conditions (Godman and Lancaster 1990, Woods 2000, Jenkins and White 2002, Galbraith and Martin 2005) that favor dogwood anthracnose development (Chellemi and Britton 1992). Th erefore, the reduction in T. canadensis density following fire has likely contributed to the greater densities of C. florida in burned areas. In our study, the positive effects on C. florida density observed in burned stands were likely a result of the indirect effects of fire, produced by cha nges in stand structure and composition. The direct effect s of fire (smoke and heat) on D. destructiva are unknown. However, studies suggest that fire may reduce the amount of inoculum of fungal diseases (Parmeter and Uhrenholdt 1975, Schwartz et al 1991). In laboratory experiments, Parmeter and Uhrenholdt ( 1975) found that exposure to smoke from burning pine needles reduced the amount of rust and gall fungi on cellophane sheets. Schwartz et al. (1991) suggest ed that smoke from upland fi res may have settled into unburned mesic ravines and helped re duce mortality of the endangered Torreya taxifolia Arnott (Florida torreya) by reducing fungal disease. In addition, burning typically produces a flush of nutrients in the soil afte r a burn (Kutiel and Shaviv 1992, Boerner et al. 2004, Tuininga and Dighton 2004). This flush in nutrients, especially cations such as calcium, magnesium, and potassium, may benefit C. florida survival. Studies have documented the importance of nutrients in plan t survival from diseases (Sij et al. 1985, Yamazaki et al. 1999, Sugimoto et al. 2005). Holzmueller et al. (2006b) reported that cation availability played a role in C. florida survival and resistance to dogwood anthracnose. We observed prolific sprouting by C. florida in stands that burned prior to anthracnose infection in our study. However, th e use of fire as a management tool for

PAGE 38

27 anthracnose is dependent upon th e ability of diseased and weakened trees to resprout after fire in anthranconse-in fected stands. Encouragingly, Blankenship and Arthur (2006) reported high levels of C. florida sprouts in recently burned areas that had been previously infected with dogw ood anthracnose. However, the long-term survival rate of these sprouts in anthracnose-infected stands is unknown. In addition, in the burned stands we sampled, most living C. florida stems were relatively small in diameter. The amount of fruit produced by these populations of smalle r individuals relative to pre-anthracnose production is unknown. The amount of fruit a nd seeds produced is critical to the successful reproduction of C. florida and its role as a source of soft mast for wildlife. Prescribed fire may offer the best means of control of dogwood anthracnose in oakhickory forest stands. Although Britton et al. (1994) documented the highest stem densities of C. florida in clearcut areas in a study of timber harvest practices on C. florida populations, it is unlikely that clearcutting large areas for the benefit of a single understory species would fit into many ecosyst em management plans. Furthermore, in clearcut stands, the overstory will likely re develop during the stem exclusion stage within 20 years after a harvest (Oliver and Larson 1990), and shading will again increase. The return interval for clearcutti ng (60-100 years) will likely be too long to serve as an effective long-term control. Prescribed burning in oak forests may offer a more applicable management technique across larg e forested areas and multiple ownerships, particularly those where mechanical harv ests are not an optio n. Although eastern oak forests have been subjected to fire suppr ession for about 100 years, resource managers have increased efforts to manage oak forests with prescribed fire (Brose et al. 2001), and these efforts offer a framework for managing C. florida populations as well. Burning on a

PAGE 39

28 10-15 year return interval w ould probably be best for C. florida survival, and would fit in with the historic fire regime of eastern oak-hickory forests (Harmon 1982). In single, double, and triple burned stands in our st udy, we observed less crown dieback compared to unburned stands, however, the lack of differe nce in foliar infection may indicate that the interval since the last bur n (nearly 20 years) has allowed anthracnose to return to a level of infection similar to unburned stands. This suggests that these stands will require additional burns to slow the loss of C. florida Management Implications Our results suggest that prescribed fire may offer an effective and practical management technique to alleviate the symp toms of anthracnose in oak-hickory and, perhaps, oak-pine forests. Other forest types, such as cove and alluvial forests, where C. florida was once a frequent component (Jenkins and White 2002) burne d infrequently, if at all, in the past. Therefore these forest types are unlikely to sustain fire frequencies of 10-15 years; the frequency that appears to best reduce the effects of anthracnose. Although this study was conducted in GSMNP, we believe that its methods and results are applicable across the eastern United States in forest types that contain C. florida and have regimes of relatively frequent fire. It is unlikely, however, that burning could be returned to all oak-hickory and oakpine forests due to many external fact ors such as time and budget constraints and management objectives. Consequently, certain ar eas may be deemed higher in priority for burning. These areas include currently uninfected stands that are in close proximity to infected areas, stands only recently showing signs of infection, or infected stands with large C. florida populations. Land managers would be more successful using prescribed

PAGE 40

29 burning in maintaining C. florida populations than attempting to reintroduce C. florida to areas where it once occurred. Conclusion The results of our study suggest that burning has reduced the impacts of anthracnose on C. florida populations. Burned stands, espe cially double burn stands, had significantly greater C. florida stem densities than unburned stands. In addition, the density of T. canadensis a species that creates sta nd conditions favorable for dogwood anthracnose, was greatly reduced in burned st ands. Past burning did not drastically affect overall overstory or understory species composition, but the importance values of selected overstory and unders tory species were highly in dicative of specific burn frequencies. The results indi cate that prescribed burning may offer an effective and practical technique to control the impacts of dogwood anthracnose and prevent the loss of C. florida from oak-hickory forests.

PAGE 41

30 Table 2-1. Scales for foliage health (% of fo liage with signs of anthracnose) and crown dieback (% of crown dieback) used to a ssess the level of di sease severity of dogwood anthracnose on Cornus florida trees (based on the Mielke-Langdon Index, Mielke and Langdon 1986). Rating Foliage health Crown dieback 1 >76 >76 2 51-75 51-75 3 26-50 26-50 4 1-25 1-25 5 none none

PAGE 42

31 Table 2-2. Mean foliage and crown health ( 1 SE) for Cornus florida for five diameter classes in the four different sampli ng categories using the Mielke-Langdon Index (Mielke and Langdon 1986). 1 P-value from ANOVA 2 Means followed by the same letter in the same row are not statisti cally different (P < 0.05) using post-hoc pairwise comparis ons among sampling categories when ANOVA P-value < 0.05 3A = Absent, no trees from this sampling category were found in this size class Diameter class (cm) Single burn Double burn Triple burn Unburned P-value1 Foliage Health 0 1.0 3.6 (0.2) 3.5 (0.2) 3.1 (0.3) A3 0.38 1.1 2.5 3.6 (0.2) 3.6 (0.2) 3.6 (0.2) 3.7 (0.3) 0.94 2.6 5.0 3.6 (0.2) 3.8 (0.2) 3.6 (0.3) 3.7 (0.3) 0.88 5.1 10.0 3.8 (0.1) 3.7 (0.1) 3.9 (0.2) 3.8 (0.3) 0.90 >10.1 3.6 (0.3) 3.6 (0.2) 3.7 (0.2) 3.6 (0.3) 0.99 Crown Dieback 0 1.0 3.2 (0.3) 3.5 (0.3) 3.0 (0.5) A 0.63 1.1 2.5 3.3 (0.2) a2 3.4 (0.2) a 3.3 (0.3) a 2.4 (0.3) b 0.04 2.6 5.0 3.3 (0.1) a 3.6 (0.2) a 3.2 (0.2) a 2.7 (0.3) b 0.01 5.1 10.0 3.3 (0.1) 3.4 (0.2) 3.7 (0.2) 3.3 (0.3) 0.45 >10.1 3.5 (0.4) 3.7 (0.4) 3.4 (0.5) 3.2 (0.3) 0.93

PAGE 43

32 Table 2-3. Mean understory and over story basal area and stem density ( 1 SE) in the four sampling categories. 1 P-value from ANOVA 2 Means followed by the same letter in the sa me column for understo ry and overstory are not statistically different (P < 0.05) using post-hoc pairwise comparisons among sampling categories when ANOVA P-value < 0.05 Understory Basal Area (m2 ha-1) Stem density (stems ha-1) Single burn 7.1 (0.5) 2851 (360) b2 Double burn 6.9 (0.7) 4594 (726) a Triple burn 6.5 (1.0) 5072 (1090) a Unburned 7.4 (0.4) 2292 (274) b P-value 0.931 0.024 Overstory Single burn 21.7 (0.8) 436 (31) b Double burn 22.1 (1.1) 323 (46) c Triple burn 20.6 (1.8) 317 (43) c Unburned 23.2 (0.8) 564 (25) a P-value 0.52 <0.0001

PAGE 44

33 Table 2-4. Mean species richne ss and Shannons diversity index ( 1 SE) for the understory and overstory in th e four sampling categories. Understory Species richness Shannons diversity index Single burn 11.4 (0.9) 1.9 (0.08) Double burn 11.0 (1.3) 1.9 (0.11) Triple burn 13.5 (1.5) 2.0 (0.12) Unburned 10.7 (1.0) 1.9 (0.09) P-value 0.481 0.81 Overstory Single burn 5.7 (0.4) 1.3 (0.09) Double burn 5.4 (0.5) 1.2 (0.12) Triple burn 5.2 (0.7) 1.2 (0.14) Unburned 6.9 (0.4) 1.4 (0.09) P-value 0.06 0.48 1 P-value from ANOVA

PAGE 45

34 Table 2-5. Overstory and understory indica tor values (IndVal) (percent of perfect indication) and associated sampli ng category. P-value represents the proportion of randomized tria ls that the indicator value was equal to or exceeded the observed indicator value. Species IndVal P-value Indicator group Overstory Acer rubrum 32.2 0.080 Unburned Oxydendrum arboreum 29.8 0.073 Unburned Quercus alba 47.4 0.001 Double burn Quercus velutina 29.7 0.026 Triple burn Tsuga canadensis 36.0 0.006 Unburned Understory Carya alba 46.0 0.001 Triple burn Carya glabra 42.6 0.012 Triple burn Cornus florida 40.2 0.015 Double burn Pinus virginiana 31.5 0.018 Triple burn Quercus alba 18.9 0.065 Triple burn Quercus prinus 37.8 0.033 Triple burn Quercus rubra 42.5 0.002 Triple burn Quercus velutina 22.3 0.076 Triple burn Robinia pseudoacacia 42.9 0.005 Triple burn Tilia americana 18.7 0.023 Double burn Tsuga canadensis 46.9 0.007 Unburned

PAGE 46

35 Table 2-6. Mean importance values ( 1 SE) of selected underst ory species in the four sampling categories. 1 Means followed by the same letter in the same row are not statisti cally different (P < 0.1) using post-hoc pairwise comparisons among sampling categories when ANOVA Pvalue < 0.05 2A = Absent, no trees from this sampling category were found in this size class Species Single burn Double burn Triple burn Unburned Acer pensylvanicum 9.6 (5.5) 3.5 (7.3) A 9.8 (6.0) Acer rubrum 24.2 (3.3) a 17.8 (4.4) ab 7.8 (1.8) b 19.2 (3.6) ab Carya alba 0.9 (0.9) a 2.0 (1.8) a 7.0 (1.5) b 2.1 (1.0) a Carya glabra 1.6 (1.0) a 1.8 (2.5) a 10.3 (2.0) b 2.5 (1.2) a Cornus florida 12.1 (2.3) a 21.6 (3.2) b 14.8 (4.2) ab 5.1 (2.6) c Halesia tetraptera 3.2 (1.7) 3.4 (2.4) A 1.8 (1.9) Kalmia latifolia 5.3 (3.6) 2.8 (4.8) 5.4 (6.0) 6.6 (4.0) Liriodendron tulipifera 4.9 (1.6) a 2.4 (2.2) ab 0.3 (2.6) ab 0.4 (1.8) b Nyssa sylvatica 6.4 (1.8) ab 9.6 (2.3) 0.9 (3.1) a 6.1 (1.9) ab Oxydendrum arboreum 6.3 (1.7) 7.3 (2.3) 6.4 (2.9) 5.2 (1.9) Pinus strobus 3.1 (2.0) 1.6 (2.6) A 4.9 (2.2) Pinus virginiana 0.2 (0.7) a 0.8 (0.9) ab 2.5 (1.1) b 1.1 (0.8) ab Quercus prinus 3.8 (1.0) 1.3 (1.4) 3.0 (2.2) 2.4 (1.2) Quercus rubra 0.7 (0.3) a 0.8 (0.4) ab 2.0 (0.6) b 0.1 (0.4) a Quercus velutina 0.5 (0.7) a 0.6 (0.9) a 3.1 (1.0) b 0.4 (0.8) a Rhododendron maximum 4.3 (2.3) 0.3 (3.1) 2.0 (3.8) 4.1 (2.5) Robinia pseudoacacia 1.3 (0.9) a 2.0 (1.2) a 6.4 (1.6) b 0.2 (1.0) a Sassafras albidum 1.5 (0.7) 1.1 (0.9) 2.9 (1.2) 1.3 (0.8) Tsuga canadensis 0.9 (2.6) a 0.5 (3.5) a 0.3 (4.2) a 9.9 (2.8) b

PAGE 47

36 Figure 2-1. Mean Cornus florida stem density ( 1 SE) in the four sampling categories for three diameter classes and total stems ha-1 for all diameter classes. Within each graph, P-value is from ANOVA, a nd bars with same letters are not significantly different from each othe r (P < 0.05) using post-hoc pairwise comparisons among sampling categories when ANOVA P-value < 0.05; note the scale of the y-axis for each graph. Sampling category Cornus florida density (stems ha-1) 0 300 600 900 Single burnDouble burnTriple burnUnburned0-5 cm class P-value = 0.0015 a b ab c 0 20 40 60 80 Single burnDouble burnTriple burnUnburned5.1-10 cm class P-value = 0.229 0 10 20 30Single burnDouble burnTriple burnUnburned>10.1 cm class P-value = 0.167 0 300 600 900 Single burnDouble burnTriple burnUnburned Total stems P-value = 0.0003 b a ab c

PAGE 48

37 Figure 2-2. Nonmetric multidimensional scal ing sample ordination of overstory and understory communities, showing the relative differences in community composition separated by sampling categories. Graphs indicate a lack of distinct compositional changes in overstory or understory with respect to burn frequency.

PAGE 49

38 Figure 2-3. Mean Tsuga canadensis stem density ( 1 SE) in five diameter size classes and total (all diameter classes combin ed) in the four sampling categories. Same letters in each diameter class represent no significant differences in mean stems ha-1 (P < 0.05) using post-hoc pairwise comparisons among sampling categories when ANOVA P-value < 0.05. 0 50 100 150 200 250 300 Total0-2.52.6-5.05.1-10.010.1-20.0>20.0 Unburned Single burn Double burn Triple burna a a a aa b b b b b bb b a bb bTsuga canadensis density (stems ha-1) Diameter class (cm)

PAGE 50

39 CHAPTER 3 INFLUENCE OF CALCIUM, POTASSIUM, AND MAGNESIUM ON Cornus florida L. DENSITY AND RESISTANCE TO DOGWOOD ANTHRACNOSE Introduction Dogwood anthracnose, caused by the fungus Discula destructiva Redlin, is a major disease of Cornus florida L. in forests of the eastern United States. Where dogwood anthracnose is present, mortality rates of C. florida have been up to 90% (Sherald et al. 1996, Hiers and Evans 1997). The fungus causes leaf necrotic blotches, leaf blight, twig dieback, and stem cankers, which eventually lead to tree death (Hibben and Daughtery 1988). Disease severity and rate of infection, however, vary with several environmental factors. The disease is most se vere in cool, wet areas of high elevation with shaded slopes (Chellemi et al. 1992). Within individual st ands, disease severity increases with decreased light availability, increased moisture, and decrea sed evaporative potential of the leaf surface (Chellemi and Britton 1992). While the link between mineral nutrition and resistance to dogwood anthracnose has not been examined in detail, soil availabi lity of cations may re duce the impacts of the disease and increase the survival of C. florida trees. Britton et al (1996) found that disease severity did not increas e with application of pH 2.5 simulated rain to the foliage, but did increase with applic ation of pH 2.5 simulated ra in to the growing medium. Consequently, the increase in the severity of infection was attributed to nutrient bioavailability, not foliar da mage. However, since soil nut ritional changes were not quantified, this hypothesis still remains unproven. In a review of calcium (Ca) physiology

PAGE 51

40 and terrestrial ecosystem processes, McLa ughlin and Wimmer (1999) hypothesized that low Ca saturation decreases the natural resistance of C. florida to dogwood anthracnose. Anderson et al. (1991) reported that liming reduced disease severity of anthracnose on ornamental C. florida trees. While the link between soil cation availa bility and anthracnose has not been examined, calcium applications have been s hown to reduce the impacts of fungal disease on soybeans ( Glycine max (L.) Merr.) and bacterial wilt on tomato ( Lycopersicon esculentum L.) seedlings (Muchovej et al. 1980, S ugimoto et al. 2005, Yamazaki et al. 1999). Other cations, particularly potassium (K) and magnesium (Mg), have also been linked to disease resistance in plants. Sij et al. (1985) reported that increased rates of K fertilizer significantly decreased the impact s of a fungal disease on field grown soybean plants. Jeffers et al. (1982) reported lower numbers of seeds infected with seed mold on tomato plants fertilized with K. Likewise, th e severity of fungal shoot blight was greater on Picea abies (Norway spruce) trees whose needle s contained low levels of Mg (Anglberger and Halmschlager 2003). Calcium, K, and Mg play essential roles in plant growth and development (Epstein 1972, Mengel et al. 2001). Calcium strengthen s plant cell walls, which may help in disease resistance (Muchovej et al. 1 980, Conway et al. 1992, Sugimoto 2005). Potassium and Mg are essential for many plant metabolic functions (Epstein 1972, Mengel et al. 2001). Disease resistance w ith optimal K and Mg nutrition may be attributed to increased energy used to offset the impact of plant diseases (Mengel et al. 2001). In addition, K may also increase diseas e resistance by increasing the thickness of outer walls in epidermal cells (Mengel et al. 2001)

PAGE 52

41 The objective of this chapter was to dete rmine whether the availability of soil cations (Ca, K, and Mg), influe nce the health and survival of C. florida. Because high levels of Ca, K, and Mg have been associated with fungal disease resistance in other plant species, we hypothesize (hypothesi s 1) that forested stands with high densities of C. florida trees would have higher conc entrations of these cations in the soil. We further hypothesize (hypothesis 2) that a dditional input of soil cations decreases disease severity of dogwood anthracnose on C. florida Materials and Methods Study Site This study was conducted in Great Smoky Mountains National Park (GSMNP), USA. Great Smoky Mountains National Park straddles the Tennessee and North Carolina state line, encompassing slightly over 200 000 ha in the southern Appalachian Mountains. The varied geology and topographic structure of GSMNP results in a wide-range of soil conditions and associated vegetation comm unities. Mean annual temperature in Gatlinburg, Tennessee (440 m a.s.l. and adjace nt to GSMNP) is 12.9 C and mean annual precipitation is 142 cm. Our study sites rang ed in elevation from 287 to 975 m. Although C. florida occurs in a variety of forest types, from mesic coves to xeric oak-pine woodlands, we focused our study on oak-hick ory forests on moderate ly steep to steep slopes with southeast-south-north west facing aspects. Oak-hickory is a major forest type in GSMNP, covering 43,337 ha, (21% of the Pa rks total forest cover; Madden et al. 2004) and prior to anthracnose C. florida was the dominant woody species in the understory (Jenkins and White 2002).

PAGE 53

42 Forest Soil Sampling During June-August of 2001-2004 soil samples were collected from seventy-nine long-term vegetation plots (20 m x 20 m; 0.04 ha) established in twenty oak-hickory stands in GSMNP. Cornus florida stem density in the stands ranged from 0 to 1150 stems ha-1 and basal area ranged from 0 to 0.7 m2 ha-1. Mineral soil was co llected from four random locations within each plot and then pooled together to create one composite sample per plot. Samples were collected by sc raping away the forest floor and collecting the top 10 cm of mineral soil (A and upper B horizons). Lab Analysis Samples were analyzed for soil Ca, K, and Mg by A&L Analytical Laboratories, Inc., Memphis, Tennessee and respective soil cation saturation (%) was calculated. Samples were dried at 36 C for 6 hours, and then sieved through a 2 mm sieve. A 3 g sample of dried soil was shaken with 30 mL of Mehlich III extrac ting solution (Mehlich 1984) for 5 minutes and then centrifuged. The solution was analyzed for Ca, K, Mg, and Na by using an inductively coupled plas ma emission spectrometer. Exchangeable hydrogen (H) was calculated by leaching 5 g of dried soil w ith, 20 mL of 0.2 M triethanolamine and 0.25 M barium chloride buffer solution (pH 8.1), then by 20 mL of 0.25 M barium chloride solution. The amount of standard acid needed to back titrate the leachate to the methyl red and bromcresol green endpoint was used to calculate the concentration of exchangeable acidity. Cati on exchange capacity (CEC) was determined by the summation of exchangeable base cati ons (Ca, K, Mg, and Na) and exchangeable H. Soil cation saturation was calculated as the percentage of the respective cation that contributed to total CEC.

PAGE 54

43 Potted Plant Experiment We tested the influence of th ree cations (Ca, K, and Mg) on C. florida seedling survival and resistance to dogwood anthrac nose by using a potted plant experiment. Six hundred 1-0 bareroot C. florida seedlings were purchased fr om a commercial nursery in Bartow, Florida, located approximately 65 km east of Tampa, Florida. Dogwood anthracnose has not been reported in Florida; therefore these seedlings were presumed to be disease free and showed no evidence of an thracnose infection before the experiment began. The 1-0 seedlings were transplanted in to three-gallon pots in a 40% Florida peat, 40% pine bark, and 20% sand mi xture and grown in a nursery for one year. The seedlings were watered every other day during the grow ing season. During the transplant into the three-gallon pots, the seedli ngs were randomly selected to receive one of twelve fertilization treatments. The treatments incl uded three cations (Ca, K, and Mg) at four levels (0, 50, 100, and 200%) of a standard nur sery fertilization rate Fertilization mixes for the twelve treatments were mixed separate ly at the nursery usi ng a base mix (Table 31) with the addition of the appropriate amount of Ca, K, and Mg to create the required treatments (Table 3-2). The seedlings were fe rtilized three times; the first fertilization treatment was at the time of transplant (M ay 2003), the second was in October 2003, and the final fertilization treatment was in Marc h 2004, before leaf ons et occurred. Depending on the treatment, 59 to 68 g (base rate + the tr eatment addition) of fertilizer was placed within each pot during e ach fertilization period. In April 2004, the seedlings were trans ported to Twin Creeks Natural Resources Center in GSMNP (35 42 51N, 83 30 37W). Dogwood anthra cnose has been widely reported in GSMNP, and heavy C. florida dieback has been reporte d in every forest type

PAGE 55

44 where C. florida is found (Jenkins and White 2002). Seedlings were placed under infected C. florida trees in a 70 year-old Liriodendron tulipifera L. and Acer rubrum L. stand. This allowed C. florida trees infected with dogwood anthracnose to serve as a source of natural inoculum (Britton et al. 1996). All seedlings were placed directly underneath the C. florida canopy in a randomized complete block design with four blocks. Foliage samples were collected from the C. florida seedlings one day prior to inoculation to test for foliar nutrient concentration. Afte r collection, samples were dried in an oven at 65 C for 72 hours and then ground using a tissue grinder. Analysis of foliage samples was performed using an inductively coupled plasma emission spectrometer (ICPES) at the University of Fl orida Analytical Research Lab (Gainesville, Florida). A 750 mg sample of dried plant material was weighed into a 20 mL high form silica crucible and dry ashed at 485 C for 12 hours. The ash was equilibrated with 5 mL of 20% HCl at room temper ature for 30 minutes. Then 5 mL of deionized water was added, gently swirled and the sample was allo wed to settle for 3 hours. The solution was decanted into a 15 ml plastic vial for direct determination by ICPES. Results of tissue concentrations are presented in mg g-1. Seedlings were measured for height and r oot collar diameter in April 2004 before inoculation and no significant differences were found among treatments (P > 0.36). After inoculation, seedling foliage was assessed ever y 2 weeks for presence of anthracnose. Any nonanthracnose lesions were disregarded. We used a scale based on the MielkeLangdon index (Mielke and Langdon 1986) to asse ss disease severity on the seedlings (Table 3-3).

PAGE 56

45 Statistical Analysis Forest Soil In order to reduce plot variability found in the forest soil data plot values were averaged together for each stand. We then used linear regression mode ls to describe the relationship between soil cation sa turation (Ca, K, and Mg) and C. florida stem density and basal area. All statistical anal yses were done using SAS (SAS 2002). Potted Plant Experiment In the potted plant experiment, ANOVA was conducted to test for differences in infection ratings among the treatments. When ANOVA revealed significant main effects, we separated the means with post-hoc pairwise comparisons. A logist ical model was used to test for differences in seedling mortality at the end of the experiment. We used curvilinear regression to desc ribe the relationship between C. florida tissue concentration of Ca, K, and Mg to fertiliza tion input. All statistical anal yses were done using SAS (SAS 2002). Results Forest Soil Cation Saturation We found significant positive correlations between the three cations (Ca, K, and Mg) and C. florida stem density and basal area. So il Ca saturation ranged from 5.7 26.3% and exhibited the strongest rela tionship of the three cations with C. florida stem density (R2 = 0.70, P < 0.0001; Figure 3-1). Soil Mg saturation ranged from 2.3 8.2% and soil K saturation ranged from 2.0 5.0%. We also found significan t relationships of soil K saturation (R2 = 0.54, P < 0.001; Figure 3-2) and soil Mg saturation (R2 = 0.62, P < 0.0001; Figure 3-3) with C. florida stem density.

PAGE 57

46 When C. florida basal area was used as th e independent variable, R2 and P-values decreased for every cation. The strongest relationship was observed between soil Ca saturation and C. florida basal area (R2 = 0.59, P = 0.001; Figure 3-1). A weak relationship was observed be tween soil K saturation and C. florida basal area (R2 = 0.23, P = 0.08; Figure 3-2), but a significant re lationship was observed between soil Mg saturation and C. florida basal area (R2 = 0.45, P = 0.008; Figure 3-3) Potted Plant Experiment Calcium Treatments Among the four levels in the Ca treatments, the seedlings did not show a significant difference among treatments until 6 weeks afte r inoculation. After 6 weeks, the 0% treatment had the lowest (less healthy) inf ection rating (Table 34). Eight weeks after inoculation, all treatments ex cept for the 100% treatment, wh ich had a higher (healthier) infection rating, were statistica lly the same (P > 0.19; Tabl e 3-4). This trend continued until the end of the experiment. At the end of the experiment, mortality for the Ca treatments ranged from 62 100% and was significantly different among the four treatments (P = 0.0013; Figure 3-4). The 0% and 200% Ca treatments had 100 and 87% mortality, respectively, which was significantly greater than the 72% mortality observed in the 50% Ca treatment and 62% mortality observed in the 100% Ca treatment (P < 0.05). Potassium Treatments Seedlings in the K treatments began to show significant differences in infection ratings 4 weeks after inoculation (P = 0.025; Table 3-4). Throughout the experiment, the 0% treatment had the lowest (less healthy) infection rating and the highest mortality (Table 3-4 and Figure 3-4). The 100% treatmen t had a high (healthier) infection rating in

PAGE 58

47 the beginning of the experiment and after 10 weeks was significantly higher (healthier) than all other treatments (P < 0.02; Table 34). After 14 weeks, there were no significant differences in infection ratings among the four K treatments (P = 0.44) and by the end of the experiment (24 weeks) all treatments ha d suffered heavy mortality (> 85%, P = 0.26; Figure 3-4). Magnesium Treatments There appeared to be very little differe nce in the anthracnose infection ratings among the Mg treatments. Inf ection ratings were simila r throughout the experiment; however, the 200% treatment did have slightly lower (less healthy) infection ratings in weeks 4 and 6 (Table 3-4). After 10 weeks there were no significant differences among treatments (P = 0.17; Table 3-4), and at the end of the experiment there were no significant differences in mo rtality among the four Mg treatments (> 88%, P = 0.26; Figure 3-4). Foliar Cation Concentrations Mean Ca concentration in the foliage of the Ca treatments ranged from 3.59 mg g-1 (0% treatment) to 3.66 mg g-1 (100% treatment). There was not a significant relationship between foliar Ca concentration and Ca input perhaps due to the high sample variability (R2 = 0.17, P = 0.43; Figure 3-5). In the K treatm ents, mean K concentration was highest in the 100% treatment (2.05 mg g-1) and lowest in the 0% and 200% treatments (1.76 and 1.77 mg g-1, respectively). The relationship betw een K concentration and K input was significant (R2 = 0.61, P = 0.02; Figure 3-5). In the Mg treatments, Mg concentration was greatest in the 50% treatment (0.83 mg g-1) and lowest in the 200% treatment (0.72 mg g-1). In general, after peaking in the 50% trea tment, Mg concentration decreased as input

PAGE 59

48 increased in the Mg treatments although the relationship was weak (R2 = 0.40, P = 0.098; Figure 3-5). Discussion We found significant positive correlations between C. florida (stem density and basal area) and soil Ca, K, and Mg satu ration. Although correlation does not imply causation, there are several possible explan ations for these strong correlations. One explanation is that highe r soil cation concentrations help maintain healthy C. florida populations even in the presence of dogwood anthracnose. Because these nutrients are readily available in the forest soil, ther e is a decrease in in ter and intraspecific competition for nutrient resources. This enables C. florida to obtain more nutrients to be used for plant defense and allocate more res ources to developing stronger cells walls to resist infection of D. destructiva The higher nutritional status could also help in replacing lost foliage and or repairing stem cankers arising from anthracnose infections. It is also reasonable to argue that nutrient levels are greater on high C. florida stem density sites because of the presence of C. florida Multiple studies have shown the influence of individual species on forest fl oor and mineral soil nutri ent levels (Dijkstra and Smits 2002, Washburn and Arthur 2003, Fu jinuma et al. 2005), which occurs through several different mechanisms. Certain species ar e able to uptake highe r levels of nutrients than others and secure nutrients in bioma ss before they are lost through soil leaching (Dijkstra and Smits 2002). Anot her mechanism is the influence of plant species on chemical weathering of the soil by modifying so il acidity (Augusto et al. 2000). It has been hypothesized that because of the high Ca concentration and rapid decomposition of C. florida foliage (Blair 1988, Knoepp et al. 2005) compared to other associated species, C. florida acts as a Ca pump in forest so ils (Thomas 1969). In addition to high

PAGE 60

49 concentration of Ca in C. florida foliage, studies performed in western North Carolina found that C. florida foliage contained high concentra tions of K and Mg compared to other species in oak hardwood forests (D ay and Monk 1977, Elliot et al. 2002; Table 35). High levels of these nutrients, particularly Mg, compared to other species were also found in the wood, twigs, and bark of C. florida (Day and Monk 1977). High soil K and Mg saturation in stands with high densities of C. florida indicates that this species may also be acting as K and Mg pumps as well. It is possible that these two hypotheses are not independent of each other. The results of our potted plant experime nt showed greater inputs of Ca and K cations slowed the rate of anthracnose infec tion. We observed differenc es in the rates of infection over the 6 month period with resp ect to the treatments. Calcium nutrition has been linked to the resistance of many plant species to fungal and bacterial diseases (Muchovej et al. 1980, Yamazaki et al. 1999, Sug imoto et al. 2005). Calcium plays a key role in development of plant cell walls (E pstein 1972, Mengel et al. 2001) and studies suggest that disease severity is reduced due to increased Ca concentrations in cell walls (Muchovey et al. 1980, Conway et al. 1992, Sugimoto 2005). In our experiment, seedlings in the 100% Ca treatment exhibi ted fewer signs of anthracnose and lower mortality compared to 0% and 200% Ca tr eatments throughout the experiment. At the end of the experiment, mortality for the Ca treatments were significantly greater in the 0% and 200% treatment compared to the 50% and 100% treatments. This indicates that Ca availability is an important factor in C. florida resistance to dogwood anthracnose. However, we did not see a relationship betw een foliar Ca concentr ation and Ca inputs.

PAGE 61

50 Perhaps the high variability in foliar Ca c oncentration within each treatment precluded a significant relations hip (Figure 3-3). Potassium is a macronutrient essential to the performance of multiple plant enzyme functions (Epstein 1972, Mengel et al. 2001). Studies have indicat ed higher disease resistance with increased levels of K (Je ffers et al. 1982, Sij et al. 1985), which, although the mechanisms are not completely understood, may be attributed to increased energy and epidermal wall thickness (Mengel et al. 2001). In our experiment, the 100% K treatment had a healthier infection rating in the early weeks of the experiment. After 10 weeks, the infection rating was still significan tly healthier than all other treatments. The 100% K treatment also had the highest foliar concentration of K compared to other K treatments (Figure 3-3). Therefore, increased disease resistance may be attributed to increased K foliar concentration. Despite the additional K inputs in the 200% K treatment, this treatment had lower K foliar concentration levels and unhealthier infection ratings compared to the 100% K treatment. Decreased foliar concen tration in the 200% K treatment compared to the 100% K treatment can be attributed to nutrient imbalances created by excess soil K; similar results ha ve been reported by Wilmot et al. (1996). Although we found a significant relationship between C. florida density and soil Mg saturation, it did not app ear that Mg input had any e ffect on disease severity or mortality. Other studies of plant disease have found si milar results (Nwoboshi 1980, Wisniewski et al. 1995). Nwoboshi (1980) found that Mg fertilization rates had no effect on the resistance of Manihot esculenta L. (cassava) to anthracnose. Wisniewski et al. (1995) reported that Mg did not inhibit the germination or gr owth of two fungal diseases, Botrytis cinerea Pers. or Penicillium expansum Link, whereas increased Ca

PAGE 62

51 concentrations decreased spore germina tion and growth of both pathogens. Although Anglberger and Halmschlager ( 2003) reported that severity of Sirococcus shoot blight decreased on P. abies trees that had needles with high leve ls of Mg, in our experiment we found that the treatment that produced the highest concentration of foliar Mg (50% treatment) fared just as poorly as the other treatments. Overall, all treatments suffered heavy mort ality by the end of the growing season after they were exposed to D. destructiva (62-100%; Figure 3-2). Th is is not particularly surprising considering the fact that multiple st udies have shown smaller trees to be very susceptible to the disease and often die within the first year of to exposure the disease (Mielke and Langdon 1986, Hibben and Daught ery 1988, Hiers and Evans 1997). In addition, it should be noted th at the 2004 growing season (year of the potted plant experiment) had above average precipitation in GSMNP (National Climatic Data Center 1999-2004; Figure 3-6) and dogw ood anthracnose throughout the Park was virulent (personal observation). The overwhelming pres ence of the disease might have decreased some of the differences among the treatments. Dogwood anthracnose is a disease that is most severe in cool, moist, and heavily shaded conditions (Chellemi and Britton 1992, Chellemi et al. 1992, Britton 1993), and higher le vels of mortality have been reported in mesic forests compared to xeric forests (Jenkins and White 2002). Britton (1993) reported that given adequate amounts of rainfall, dogwood anthracnose could develop throughout the gr owing season. Factors affecting relative humidity and evaporative potential of leaf surfaces in a stand, such as stand density, slope, and elevation, probably influe nce impacts of dogwood anthracnose on C. florida more than nutrient conditions. We still concl ude, however, that low av ailability of Ca and

PAGE 63

52 K in forested stands containing C. florida increases the susceptibility of C. florida to dogwood anthracnose. Decrease d nutrient availab ility in eastern forests from acid deposition (Likens et al. 1996, McLaughlin a nd Wimmer 1999) has likely had a negative impact on remaining C. florida populations, which may further inhibit the annual calcium cycling of cations by C. florida (Holzmueller et al. 2006c). Conclusion The results of this project indicate that there is a correlation between soil cation saturation (Ca, K, and Mg) and C. florida stem density and basa l area in oak-hickory forests. High concentrations of these cations in C. florida foliage suggest that this species may play an important role in nutrient cycl ing by acting as a pump that draws cations from deeper in a soil profile and cycles th em through the forest floor and surface soil. The results of this project also suggest that increased levels of Ca and K in the soil may lead to increased resistance to dogwood anthracnose. We conc lude that soil fertility in forest stands should not be overlooked when applying management techniques to reduce the impacts of dogw ood anthracnose.

PAGE 64

53 Table 3-1. Content of the base fertilizer mix. Element Rate (%) NH4 6.8 NO3 5.2 P2O5 6.0 AgSul 90 6.2 FeSO4 0.8 MnSO4 0.3 ZnSO4 0.1 CuSO4 0.05

PAGE 65

54 Table 3-2. Inputs (%) added to the base fert ilizer mix (separately) for each treatment. Treatment Calcium (CaSO4) rate (%) Potassium (K20) rate (%) Magnesium (MgSO4) rate (%) Ca 0% 0.0 8.0 1.8 Ca 50% 0.75 8.0 1.8 Ca 100% 1.5 8.0 1.8 Ca 200% 3.0 8.0 1.8 K 0% 1.5 0.0 1.8 K 50% 1.5 4.0 1.8 K 100% 1.5 8.0 1.8 K 200% 1.5 16.0 1.8 Mg 0% 1.5 8.0 0.0 Mg 50% 1.5 8.0 0.9 Mg 100% 1.5 8.0 1.8 Mg 200% 1.5 8.0 3.6

PAGE 66

55 Table 3-3. Scale used to assess the seve rity of dogwood anthrac nose infection on the foliage of Cornus florida seedlings. Scale was ba sed on the Mielke-Langdon Index (Mielke and Langdon 1986). Rating % of foliage with signs of anthracnose 0 Dead 1 76-100 2 51-75 3 26-50 4 1-25 5 0

PAGE 67

56Table 3-4. Biweekly infection ratings for Cornus florida seedlings for length of the experiment. Treatment Week 2 Week 4 Week 6 Week 8 Week 10 Week 12 Week 14 Week 16 Week 18 Week 20 Week 22 Week 24 Ca 0% 3.9 a1 3.5 a 2.8 a 2.5 a 1.5 a 1.1 a 0.8 a 0.6 a 0.4 a 0.1 a 0.1 a 0.0 a 50% 3.8 a 1 3.8 a 3.4 b 2.9 ab 1.9 a 1.6 a 1.4 ab 1.1 ab 1.1 ab 0.7 ab 0.6 ab 0.4 ab 100% 3.8 a 1 3.8 a 3.4 b 3.2 b 2.8 b 2.3 b 1.7 b 1.3 b 1.2 b 0.8 b 0.7 b 0.4 b 200% 3.8 a 1 3.6 a 3.0 a 2.7 a 1.8 a 1.3 a 0.9 a 0.8 a 0.7 a 0.3 a 0.3 a 0.2 a SE 0.1 a 0.1 a 0.2 a 0.2 0.2 a 0.2 a 0.2 a. 0.2 a 0.2 a 0.1 a 0.1 a 0.1 a K 0% 4.0 a 1 3.3 a 2.5 a 1.3 a 0.8 a 0.6 a 0.5 a 0.3 a 0.1 a 0.1 a 0.1 a 0.1 a 50% 3.8 a 1 3.6 b 2.9 ab 1.8 b 1.1 a 0.9 ab 0.6 a 0.4 a 0.4 a 0.3 a 0.3 a 0.2 a 100% 3.8 a 1 3.7 b 3.2 b 2.0 b 1.7 b 1.2 b 0.7 a 0.6 a 0.5 a 0.3 a 0.2 a 0.2 a 200% 4.0 a 1 3.7 b 3.2 b 1.7 a 1.2 a 0.8 ab 0.6 a 0.4 a 0.4 a 0.2 a 0.2 a 0.2 a SE 0.1 a 1 0.1 0.2 0.2 0.2 0.2 a 0.2 a. 0.1 a 0.1 a 0.1 a 0.1 a 0.1 a Mg 0% 3.9 a 1 3.9 a 3.4 a 2.1 a 1.6 a 0.9 a 0.6 a 0.5 a 0.5 a 0.3 a 0.2 a 0.2 a 50% 3.8 a 1 3.6 ab 3.0 ab 1.9 ab 1.2 a 0.9 a 0.6 a 0.4 a 0.3 a 0.1 a 0.1 a 0.1 a 100% 3.7 a 1 3.8 a 3.4 a 2.4 a 1.6 a 1.2 a 1.0 a 0.8 a 0.8 a 0.3 a 0.3 a 0.2 a 200% 3.9 a 1 3.4 b 2.6 b 1.5 b 1.2 a 0.7 a 0.7 a 0.7 a 0.6 a 0.4 a 0.3 a 0.3 a SE 0.1 1 0.1 0.2 0.2 0.2 0.2 a 0.2 0.1 a 0.1 a 0.1 a 0.1 a 0.1 a 1 Means with different letters in same co lumn for each cation for each week are stat istically different (P < 0.05) using post-hoc pairwise comparisons among sampling categories when ANOVA P-value < 0.05

PAGE 68

57 Table 3-5. Foliar calcium (Ca) potassium (K), and magnesium (Mg) concentrations (%) for selected species in a southern Appa lachian forest. Data presented is from Day and Monk (1977). Species Foliar concentration (%) Ca K Mg Cornus florida 1.60 1.18 0.90 Quercus alba 0.50 0.75 0.14 Quercus coccinea 0.45 0.62 0.14 Quercus prinus 0.59 1.09 0.19 Quercus rubra 0.75 0.89 0.33 Quercus velutina 0.71 0.95 0.17 Acer rubrum 0.62 0.53 0.20 Carya glabra 0.95 0.58 0.82 Liriodendron tulipifera 1.39 1.04 0.61 Oxydendrum arboreum 0.96 0.78 0.27 Nyssa sylvatica 0.96 1.04 0.51 Magnolia fraseri 1.07 1.25 0.38 Betula lutea 1.11 1.10 0.37 Sassafras albidum 0.52 1.08 0.27

PAGE 69

58 Figure 3-1. Linear regr ession between soil calcium (Ca) saturation and Cornus florida stem density and basal area. y = 0.0145x + 8.4861 R2 = 0.7042 P < 0.00010 5 10 15 20 25 30 050010001500Soil Ca saturation (%) y = 19.408x + 5.7707 R2 = 0.5933 P = 0.0010 5 10 15 20 25 30 0.00.20.40.60.8Soil Ca saturation (%)Cornus florida density (stems ha-1) Cornus florida basal area (m2 ha-1)

PAGE 70

59 Figure 3-2. Linear regr ession between soil potassium (K) saturation and Cornus florida stem density and basal area. y = 2.2548x + 2.5135 R2 = 0.2319 P = 0.080 2 4 6 0.00.20.40.60.8Soil K saturation (%) y = 0.0022x + 2.7002 R2 = 0.539 P < 0.0010 2 4 6 050010001500Soil K saturation (%)Cornus florida density (stems ha-1) Cornus florida basal area (m2 ha-1)

PAGE 71

60 Figure 3-3. Linear regr ession between soil magnesi um (Mg) saturation and Cornus florida stem density and basal area. y = 4.4378x + 3.5256 R2 = 0.4502 P = 0.0080 3 6 9 0.00.20.40.60.8Soil Mg saturation (%)Cornus florida density (stems ha-1) Cornus florida basal area (m2 ha-1) y = 0.0043x + 3.7915 R2 = 0.623 P < 0.00010 3 6 9 050010001500Soil Mg saturation (%)

PAGE 72

61 Figure 3-4. Biweekly mortality (%) of Cornus florida seedlings for the four treatment levels of each cation. Calcium Treatments0 20 40 60 80 100Mortality (%) 0% 50% 100% 200% Potassium Treatments0 20 40 60 80 100Mortality (%) 0% 50% 100% 200% Magnesium Treatments 0 20 40 60 80 1005/17/20046/16/20047/16/20048/ 15/20049/14/200410/14/2004Mortality (%) 0% 50% 100% 200%

PAGE 73

62 Figure 3-5. Foliar calcium (Ca) potassium (K), and magnesium (Mg) concentrations of Cornus florida seedling foliage for the four treatment levels of each cation. Treatment y = -1E-04x2 + 0.015x + 35.8 R2 = 0.17 P = 0.4273.3 3.6 3.9 0%50%100%150%200%250%Ca concentration (mg g-1) y = -0.0002x2 + 0.047x + 17.2 R2 = 0.6132 P = 0.021.6 1.8 2.0 2.2 0%50%100%150%200%250%K concentration (mg g-1) y = -4E-05x2 + 0.0053x + 7.8 R2 = 0.4021 P = 0.0980.6 0.7 0.8 0.9 0%50%100%150%200%250%Mg concetration (mg g-1)

PAGE 74

63 Figure 3-6. Precipitation data for 2004 and previous 5 year av erage (1999-2003) during April-September at the Twin Creeks Natural Resources Center, Great Smoky Mountains National Park (Data from National Climatic Data Center 19992004). 0 5 10 15 20 25AprilMayJuneJulyAugust SeptemberMonthMonthly precipitation (cm) 2004 precipitation Average precipitation

PAGE 75

64 CHAPTER 4 INFLUENCE OF Cornus florida L. ON CALCIUM MINERALIZATION IN TWO SOUTHERN APPALACHIAN FOREST TYPES Introduction Over the past two decades, numerous studies have raised concerns about calcium (Ca) depletion in forest soils of the eastern United States (Likens et al. 1998, Huntington et al. 2000, Johnson et al. 2000, Ya nai et al. 2005). In many fore sts, this depletion has been well documented. For example, between 1965-1992, Likens et al. (1998) estimated a loss of 9.9-11.5 kmol ha-1 of total Ca from the complete soil profile at Hubbard Brook Experimental Forest in New Hampshire. In a 60-80 year old southern Piedmont forest in Georgia, Huntington et al. ( 2000) estimated that the soil Ca depletion rate was 12.7 kg ha-1 y-1. Calcium depletion has been attribut ed to leaching caused by acid deposition (Lawrence et al. 1995, Likens et al. 1996) a nd uptake and sequestering of nutrients in woody biomass (Johnson and Todd 1990, Hun tington et al. 2000). The ecological consequences of soil Ca depletion could be devastating since long-term forest ecosystem health and sustainability have been closel y linked to pools of available Ca in the soil (Graveland et al. 1994, NAPAP 1998, Driscoll et al. 2001, Hamburg et al. 2003). In mixed hardwood forests of eastern Nort h America, Ca released through mineral weathering is generally an insignificant cont ributor to total calcium cycling (Huntington et al. 2000, Dijkstra and Smits 2002). As a re sult, the release of Ca through organic matter decomposition (mineralization) is considered the major source of Ca for immediate uptake by forest plants (Likens et al. 1998, Dijkstra and Smits 2002). In a

PAGE 76

65 study in northwestern Connecticut, Dijkstra (2003) reported that Ca mineralization occurred primarily in the forest floor (fro m leaf litter) as opposed to the mineral soil. Foliar Ca concentrations vary greatly among tree species (Metz 1952, Day and Monk 1977, Elliot et al. 2002), which influences the amount of Ca mineralized in the forest floor beneath the canopy of a gi ven tree species (Dijkstra 2003). Cornus florida L. foliage, on average, has a high er Ca concentration (2.0-3.5%) and more rapid decomposition than that other woody species (Thomas 1969, Blair 1988, Knoepp et al. 2005). Cornus florida was, historically, one of the most common understory species in eastern United States hardwood forests (Muller 1982, Elliott et al. 1997, Jenkins and Parker 1998). Because of the hi gh Ca concentration of its foliage, rapid decomposition of its litter and it s abundance in the understory, C. florida has long been believed to influence Ca availability in the soil and forest floor by acting as a Ca pump that draws calcium from deep in the soil pr ofile and deposits it on the biotically-rich forest floor and surface soil (Thomas 1969, Jenkin s et al. 2006). However, during the past 20 years, C. florida has suffered heavy mortality (ove r 90% in some forest types) throughout most of its range due to the rapid spread of the fungus Discula destructiva Redlin, which causes the di sease dogwood anthracnose (A nagnostakis and Ward 1996, Sherald et al. 1996, Hiers and Evans, 1997, Je nkins and White 2002, Holzmueller et al. 2006). The loss of C. florida foliar biomass from the unders tories of eastern hardwood forests has the potential to further redu ce Ca availability in these forests. Although many have suggested that C. florida plays an important role in Ca cycling and availability in eastern forests (Thomas 1969, Hiers and Evans 1997, Jenkins and White 2002), no studies have been c onducted to quantify the impact of C. florida on Ca

PAGE 77

66 mineralization in forests where the species occurs. Understanding the influence of C. florida on the mineralization rate of forest stands is critical to unde rstanding the impacts of dogwood anthracnose on calcium cycling. Th e objective of this study was to quantify Ca mineralization in the forest floor and mineral soil along na tural gradients of increasing C. florida stem density. Because C. florida litter decomposes very rapidly compared to other species (Thomas 1969, Blair 1988, Knoepp et al. 2005), and its foliage contains high concentration of Ca (Thomas 1969, Blai r 1988), we hypothesized that stands with high densities of C. florida have higher rates of Ca minera lization than stands with lower C. florida densities. Materials and Methods Study Area We conducted this study in the north-cen tral portion of Great Smoky Mountains National Park (GSMNP), near Gatlinburg, Tennessee. Mean annual temperature in Gatlinburg, Tennessee (440 m a.s.l.) is 12.9 C and mean annual precipitation is 142 cm. Study site elevation ranged from 487 to 762 m. All sampling was performed in two forest types; oak hardwood and cove hardwood. The most common species in the oak hardwood forest type were Quercus alba L., Quercus prinus L., Quercus coccinea Muenchh., Carya alba (L.) Nutt., Pinus strobus L., Oxydendrum arboreum (L.) DC., and Nyssa sylvatica Marsh. In the cove hardwood forest type, the most common species were Liriodendron tulipifera L., Acer rubrum L., Tsuga canadensis (L.) Carr., Fagus grandifolia Ehrh., Betula lenta L., and Magnolia fraseri Walt. All study sites were located in secondary forests that were logged prior to park esta blishment in 1934 (Pyle 1988). Soils in this area were predominantly Junaluska-Tsali complex, Soco-Stecoah complex, and SpiveySanteetlah-Nowhere complex. These complexes, found in both forest types, are typically

PAGE 78

67 well drained, form on moderate slopes ( 15-45%), are sometimes stony, and are derived from soft metasandstone (Ant hony Khiel, soil scientist, N RCS, personal communication). Field Sampling We determined Ca mineralization in the forest floor and mineral soil using the buried bag in situ incubation method described by Eno (1960). This technique has recently been utilized to quantify Ca mine ralization in forested ecosystems (Dijkstra 2003) and is commonly used to estimate N mi neralization as well (Prescott et al. 2003, Allen et al. 2005). We collected data for 2 years; bags were buried in early June 2003/2004 (summer incubation) and again in early December 2003/2004 under freshly fallen leaf litter (winter incubation). Overall, sixty-eight 10 m x 10 m plots were sampled every year, thirty in the cove hardwood forest type and thirty-eight in the oak hardwood forest type. For each forest type, plots were divided into three sampling categories based on C. florida stem density: (0 stems ha-1, 200-300 stems ha-1, and > 600 stems ha-1), hereafter referred to as zero, low, and high density, respectively. Each plot was surrounded by a 20 m buffer from the outside edge of the plot that was void of other C. florida stems. There was a minimum of 100 m separating plots from each other. In each plot, two forest floor samples (20 cm x 20 cm) were taken underneath the canopy of the C. florida trees, but were at least 1 m aw ay from the nearest tree base. Forest floor mass and depth were determined from these samples. On plots where no C. florida trees were present, two fo rest floor samples were randomly collected within the 10 m x 10 m plot, and were at le ast 1 m away from the nearest tree base. Where the forest floor was removed, a soil core (4 cm x 15 cm) was extracted. Each forest floor and underlying soil sample was divided into two equa l parts, transferred to polyethylene bags, and closed with a knot. Two litte r and two soil bags (initial sample) from each plot were

PAGE 79

68 returned to the lab to determine dry mass, pH, and exchangeable Ca. The remaining sample bags (final sample) on each plot were then returned to the spot from which they were collected, and the soil bags were buried in the core holes and the forest floor bags were placed in the litter laye r. The bags containing the fo rest floor were covered with fresh forest litter. Six months after incuba tion, the bags were retr ieved and brought back to the laboratory for further analysis. Laboratory Analysis Once in the lab, the contents of the bags were dried in an oven at 70 C for 72 hours. After drying, the mineral soil was sieved through a 2 mm sieve and the forest floor was ground using a tissue grinder. Subsamples of the forest floor and mineral soil were dried at 105 C for 48 hours to measure grav imetric moisture content. Samples of the forest floor and mineral soil were then measur ed for pH in de-ionized water slurry (10:1 ratio for the forest floor and 2:1 ratio for th e soil). Samples were stirred initially and again after 15 minutes. After settling for 30 minutes followi ng the final stirring, pH was measured. We extracted both the mineral soil and forest floor (separately) samples using 10 g of mineral soil and 5 g of forest floor mixed with 100 ml of 0.1M BaCl2 in a 120 ml vial. Samples were shaken for 2 hours on a soil shaker and filtered after settling for 24 hours using a coffee filter. Exchangeable Ca was measured using an inductively coupled plasma emission spectrometer at the Univ ersity of Florida Analytical Research Laboratory (Gainesville, Flor ida). Calcium mineralization was determined as the difference between final and initial exchangeable Ca in the bags. Statistical Analysis Summer and winter initial Ca concentrati ons and pH for the 2 years were compared using paired t-tests. Because there was no st atistical difference (P > 0.05) between the 2

PAGE 80

69 years for any variable the data were combin ed for further analysis. Differences between initial and final pH, exchangeable Ca from initial summer and initial winter periods, and Ca mineralization from summer and winter incu bations were tested using paired t-tests for each forest type. ANOVA was used to test for differences of forest floor mass and depth, initial exchangeable Ca concentra tions, and Ca mineraliz ation for the three C. florida densities for each forest type. We al so tested the relationship between Ca mineralization and stem density, basal area, and foliar biomass (Martin et al. 1998) using step-wise multiple regression. Step-wise multiple regression did not yi eld any significant relationships among any variab les and as a result data ar e not shown. All statistical analyses were done using SAS (SAS 2002). Results Forest Floor Mass and Depth There were no significant differences in forest floor mass (2.5 2.9 kg m-2) in cove hardwood plots for both summer and winter periods (P > 0.61). There also were no significant differences in fo rest floor mass (2.7 3.1 kg m-2) in oak hardwood plots during summer and winter peri ods (P > 0.37). There were no significant differences in forest floor depth (3.4 4.0 cm) in cove hardwood plots during summer and winter periods (P > 0.39). There were al so no significant differences in forest floor depth (3.5 3.8 cm) in the oak hardwood plots during summer and winter periods (P > 0.47). Soil pH In cove hardwood plots, mean forest floor pH ranged from 4.63 to 5.77 in the forest floor and from 4.01 to 4.95 in the mineral soil (Table 4-1). In oak hardwood plots, mean forest floor pH ranged from 4.01 to 5.40 and mean mineral soil pH ranged from 3.65 to 4.49 (Table 4-1). In cove hardwood plots, th e mean pH of the forest floor increased

PAGE 81

70 slightly during summer incubation on zero and low density plots (P < 0.01), but did not significantly change on high de nsity plots (P = 0.50). Mean mineral soil pH did not change significantly for any density class during either incubation period (P > 0.43). In oak hardwood plots, mean pH increased sligh tly in the forest floor and mineral soil of zero density plots (P = 0.1 and P = 0.01) and in the forest floor of low density plots (P = 0.005). During the winter incubation, there wa s a significant increase in pH between the initial and final samples in both forest types for all C. florida densities (P < 0.001; Table 4-1). Initial Exchangeable Ca Initial mean values for exchangeable Ca varied with C. florida density in both the forest floor and mineral soil for both forest t ypes. Mean forest floor values in the cove hardwood plots ranged from 5.5 to 8.4 g kg-1, which was about ten times greater than mean values found in the cove hardwood forest mineral soil 0.45 to 0.85 g kg-1 (P < 0.001; Figure 4-1). In oak hardwood plots, fore st floor mean values ranged from 3.6 to 7.4 g kg-1, and mineral soil mean values ranged from 0.19 to 0.68 g kg-1 (Figure 4-1). Initial exchangeable Ca mean va lues generally increased with C. florida density for both forest types for summer and wi nter incubation periods and we re significantly greater in the high C. florida density plots compared to the zero de nsity plots in both the forest floor and mineral soil in both forest types (P < 0.01). Comparisons of initial exchangeable Ca mean values made between the winter a nd summer incubations showed no significant differences between the tw o periods (Figure 4-1). Ca mineralization Ca mineralization was greater in the forest floor than in the mineral soil for both forest types (P < 0.0001). Mean values ranged from 2.09 to 9.16 mg kg-1 day-1 for the

PAGE 82

71 cove hardwood forest floor and from -1.10 to 0.20 mg kg-1 day-1 in the cove hardwood mineral soil (Figure 4-2). For the oak hardw ood forests, mean values ranged from 1.31 to 7.16 mg kg-1 day-1 in the forest floor and from -0.40 to 0.23 mg kg-1 day-1 in the mineral soil (Figure 4-2). Mean values for the winter incubation period were significantly lower than summer incubation period for most of the mineral soil comparisons, but were not significantly different from the summer incubation periods for the forest floor comparisons for both forest types (P > 0.21) Cornus florida density had an effect on the forest floor in both forest types (P < 0.1), with increasing C. florida density leading to increased Ca mineralization (Figure 4-2). Yearly Ca mineralization was greater in the high C. florida density plots compared to the zero density plots for both forest types; cove hardwood, high density (3.3 g kg-1 yr-1) versus zero density (0.6 g kg-1 yr-1, P = 0.04) and oak hardwood, high density (2.4 g kg-1 yr-1) versus zero density (1.1 g kg-1 yr-1, P = 0.09; Table 4-2). In most cases, yearly Ca mineralization in the mi neral soil was negative, indicating Ca immobilization. Discussion Most of the yearly Ca mineralization in our study can be attri buted to the forest floor, which is similar to Dijkstras (2003) findings in northwestern Connecticut that, under most tree species, forest floor Ca mineralization fa r exceeded mineral soil Ca mineralization. The two species that did have mineral soil exchang eable Ca inputs that were comparable to the forest floor in Dijkstras (2003) study were Acer saccharum Marsh. and Fraxinus americana L., and this increase in mine ral soil inputs was attributed to high earthworm activity. Neither of these tw o species was in high abundance in either forest type in our study. In addition, in glaciated areas, su ch as Connecticut, exotic earthworm species tend to dominate over native species. The exotic species tend to break

PAGE 83

72 down litter and duff at a much more rapid rate than natives. In non-glaciated areas, such as the southern Appalachians, native earthwo rms dominate and the break down of coarse organic matter is much slower (Hendrix and Bohlen 2002). Ca mineralization differed significantly in the forest floor among the three C. florida densities in both forest types and incuba tion periods. Within the forest floor, Ca mineralization was significantly higher in the high density C. florida plots, except for the winter incubation period in the oak hard wood forest type which did not show a significant difference due to high plot variability. Increased mineralization in high density C. florida plots could be attribut ed to the high Ca concen tration (Thomas 1969, Blair 1988) and rapid decomposition of C. florida foliage (Thomas 1969, Blair 1988, Knoepp et al. 2005). Mineralization in the forest fl oor did not differ betw een winter and summer incubation periods for any density level or forest type. Dijkstra (2003) reported Ca mineralized was greater in the summer in cubation period compared to the winter incubation period and attributed this to warmer temperatures during the summer incubation period. The warmer winters of Tennessee compared to Connecticut may have offset this difference and resulted in comparab le winter and summer values. It should be noted though, that mineral soil Ca minerali zation values were si gnificantly lower during the winter incubation period of our study. Initial exchangeable Ca values in the zero density C. florida plots were slightly higher in the cove hardwood forest type than in the oak hardwood fore st type in both the mineral soil and forest floor (P < 0.001). This re sult could be attributed to the different Ca concentrations found in the foliage of the dom inant species in each forest type. Numerous studies have shown how soil pr operties can be influenced by tree species (Boettcher and

PAGE 84

73 Kalisz 1990, Finzi et al.1998, Dijkstra and Smits 2002, Fujinuma et al.2005). The cove hardwood forest type was primarily dominated by L. tulipifera which has an average of 1.74% Ca concentration in its foliage, comp ared to the dominant species in the oak hardwood forest type, Q. alba which has much lower average Ca foliar concentration (0.73%) (Jenkins et al. 2006) (See Table 4-3 for a listing of specie s and corresponding Ca concentration). Despite the differences in initial values of exchangeable Ca between the two forest types, values for initial exchangeable Ca were significantly greater in high density C. florida plots compared to zero density plots in both the forest floor and mineral soil for both forest types for both incubation periods. While previous research has focused on the relationship between overstory trees and so il chemistry, our study shows that a single understory woody species can have considerable influence on soil chemical properties. One would assume that given the larger size of overstory trees, most of the forest floor biomass comes from overstory trees a nd not from understory trees, therefore overwhelming any effect understory trees mi ght have on soil chemical properties. However, in a study by Jenkins et al. (2006) in GSMNP, the authors reported that understory foliar biomass contributed up to 49% of total stand foliar biomass, depending on forest type and stand developmental stage. Because of dogwood anthracnose, C. florida density has greatly declined across the eastern United States (Anagnostakis and Wa rd 1996, Sherald et al. 1996, Hiers and Evans 1997, Jenkins and White 2002), dramatically reducing C. florida foliar biomass added to the forest floor. In GSMNP, there has been a significant reduction in the amount of Ca cycled to the forest floor in forest types containing C. florida Typic cove forests

PAGE 85

74 experienced a 86% decline in C. florida leaf litter over a 20 year period since 1977, resulting in a corresponding d ecline (85%) in annual Ca inpu ts and oak-hickory forests experienced a 78% reduction in C. florida leaf litter during the sa me period, resulting in a 78% reduction in annual Ca inputs (Jenkins et al. 2006). Throughout much of the Park, and likely across the southern Appalachians as well, C. florida trees have largely been replaced by T. canadensis a species with more acidic litter that contai ns little Ca (Jenkins and White 2002). Increased T. canadensis densities in the forest understory could further disrupt Ca cycling in eastern forests. In a study that compar ed base cation levels beneath three tree species ( Acer saccharum Marsh., Tilia Americana L., and Tsuga canadensis ) in Ottawa National Forest in western Upper Mich igan, the authors repo rted high levels of base cation leaching underneath T. canadensis canopies which was attributed to the low uptake of these cations by T. canadensis (Fujinuma et al. 2005). The hemlock woolly adelgid ( Adelges tsugae Annand), however, is spreadi ng rapidly within GSMNP, and forests across the southern Appalachians may experience heavy T. canadensis mortality, similar to that observed in the northeastern Un ited States (Johnson et al. 1999). If this occurs, the importance of shade tolerant hardwood species, such as A. rubrum may increase in the forest types we sampled. Wh ile this and other hardwood species typically contribute more calcium to annual cycling than T. canadensis their contributions are still much lower than that of C. florida In a study of regional forest plant species diversity in central Europe, Cornwell and Grubb (2003) reported the highest levels of plant species richness were found on nutrient rich soils. Although it is not cl ear how loss of Ca inputs from C. florida has affected the vigor and growth of other species and the overall stand dynamics in eastern hardwood

PAGE 86

75 forests, several other common species co-occurring with C. florida ( Q. coccinea Q. rubra L., and Robinia pseudoacacia L.) have been reported to show increased levels of mortality in the southern Appalachian M ountains while dogwood anthracnose has been dramatically reducing C. florida in eastern forests (W yckoff and Clark 2002). Cornus florida litter is a major source of Ca and a decline in foliar biomass could negatively affect Ca mineralization in the soil, the primary source of uptake in eastern forests (Dijkstra and Smits 2002), disr upting the Ca cycle (Figure 4-3). This negative impact, combined with acid deposition, may eventua lly result in further Ca depletion (Hamburg et al. 2003), which has been associated with canopy dieback in some eastern hardwood forests (Wilmot et al. 1996). Lack of Ca can affect other components of forest ecosystems, including soil fauna. Decreased land snail abundance has been corr elated with decreased exchangeable Ca levels in Sweden (Wreborn 1992) and th e central Appalachian Mountains (Hotopp 2002). In the Netherlands, poor reproductive success in Parus major L. (great tit, a passerine bird) because of a lack of Ca in eggshells was attribut ed to reduced snail abundance on soils depleted of calcium by acid deposition (Graveland et al. 1994, Graveland 1996). Because of the important role C. florida plays in the Ca cycle, preventing its loss may be of critical importance in eastern forests. This is a difficult task due to the presence of dogwood anthracnose. However, there has been some indication that prescribed burning with proper frequency can help retain C. florida as a component of stands infected with this disease (Holzmueller et al. 2006a).

PAGE 87

76 Conclusion These results suggest that C. florida density significantly affects Ca mineralization in both cove hardwood and oak hardwood forest types, primarily in the forest floor. The influence of C. florida on Ca mineralization may be attributed to the high Ca concentration and rapid decomposition of its foliage. Because mineralized Ca in the forest floor is the primary source of availa ble Ca in eastern hardwood forests, loss of C. florida may further alter Ca cycling in these fo rests with subsequent negative impacts on associated flora and fauna.

PAGE 88

77 Table 4-1. Mineral soil and forest floor mean pH ( 1 SE) for summer and winter incubations in the cove hardwood and oak hardwood forest types. 1 Statistical comparisons were made between th e initial and final samples for each density in each forest type using paired t-tests, ns = P > 0.1, = P < 0.1, ** = P < 0.01, *** = P < 0.001 Summer Winter Initial Final Initial Final Cove hardwood Forest Floor None 4.63 (0.20) 5.31 (0.25)*** 4.71 (0.15) 5.38 (0.21)*** 1 Low 5.30 (0.19) 5.77 (0.15)** 5.30 (0.17) 5.72 (0.19)*** High 5.14 (0.17) 5.20 (0.25) ns 5.31 (0.12) 5.71 (0.14)*** Mineral Soil None 4.26 (0.16) 4.19 (0.16) ns 4.01 (0.11) 4.95 (0.09)*** Low 4.29 (0.12) 4.27 (0.15) ns 4.15 (0.09) 4.60 (0.09)*** High 4.26 (0.11) 4.11 (0.07) ns 4.04 (0.07) 4.63 (0.07)*** Oak hardwood Forest Floor None 4.13 (0.11) 4.35 (0.16)* 4.01 (0.08) 4.41 (0.07)*** Low 4.47 (0.19) 4.78 (0.20)** 4.41(0.12) 4.77 (0.15)** High 5.23 (0.19) 5.40 (0.22) ns 5.11 (0.10) 5.48 (0.16)** Mineral Soil None 3.65 (0.05) 3.49 ( 0.08) ** 3.65 (0.27) 4.14 (0.04)*** Low 3.74 (0.14) 3.76 (0.19) ns 3.86 (0.09) 4.45 (0.16)*** High 4.02 (0.12) 3.95 (0.18) ns 4.01 (0.09) 4.49 (0.11)***

PAGE 89

78 Table 4-2. Mean yearly Ca mineralization ( 1 SE) in forest floor, mineral soil, and combined total (forest floor plus mineral soil) for the three Cornus florida sampling densities in the cove hardw ood and oak hardwood forest types. 1 Means with different letters in same colu mn for each forest type are statistically different (P < 0.1) using post-hoc pairwise comparisons among categories when ANOVA P-value < 0.1 Forest Floor (g kg-1 yr-1) Mineral Soil (g kg-1 yr-1) Total (g kg-1 yr-1) Cove hardwood Zero 0.7 (0.8) a1 -0.1 (0.05) b 0.6 (0.8) a Low 1.1 (0.9) a -0.2 (0.07) a 0.9 (0.9) a High 3.3 (0.8) b 0.0 (0.04) c 3.3 (0.8) b Oak hardwood Zero 1.1 (0.8) a -0.04 (0.01) a 1.1 (0.8) a Low 2.6 (0.8) b -0.04 (0.04) a 2.5 (0.8) b High 2.4 (0.9) b -0.03 (0.07) a 2.4 (0.9) b

PAGE 90

79 Table 4-3. Foliar calcium concentrations (%) from dominant species in the two forest types (oak hardwood and cove hardwood) sampled in this study. Data from trees sampled within Great Smoky Mount ains National Park on long-term vegetation plots (NPS unpublished data). Cove hardwood species Calcium concentration (%) Oak hardwood species Calcium concentration (%) Liriodendron tulipifera 1.74 Quercus spp. 0.73 Acer rubrum 0.82 Carya spp. 0.98 Tsuga canadensis 0.46 Pinus strobus 0.29 Betula lenta 0.95 Oxydendrum arboreum 0.91 Magnolia fraseri 1 1.07 Nyssa sylvatica 0.78 Cornus florida 1.73 Cornus florida 1.73 1 Data from study by Day and Monk (1977) in Co weeta Hydrologic Labo ratory located in southwestern North Carolina

PAGE 91

80 Figure 4-1. Mean initial ex changeable Ca levels ( 1 SE) in the forest floor and mineral soil in the cove hardwood and oak hardwood forest types during summer and winter collection times. Comparisons between summ er and winter values in each panel were made for each density; all were nonsignificant (ns; P > 0 .05). Bars for the same collecti on period in each panel with different letters are significantly diff erent (P < 0.05) using post -hoc pairwise comparisons among categories when ANOVA P-value < 0.05. Forest floor 0 2 4 6 8 10NoneLowHigha b b a c a ns ns ns Forest floor 0 2 4 6 8 10NoneLowHigha b b b b a ns ns ns Cove hardwood forest t yp e Cornus florida stem density Exchangeable Ca (g kg-1) Initial summer Initial winter Oak hardwood forest t yp e Mineral Soil 0.0 0.3 0.5 0.8 1.0 1.3NoneLowHigha b b b b a ns ns ns Mineral Soil0.0 0.3 0.5 0.8 1.0 1.3NoneLowHigha b a b a ns ns ns a

PAGE 92

81 Figure 4-2. Mean Ca mineralization ( 1 SE) for the forest floor and mineral soil in the cove hardwood and oa k hardwood forest types during summer and winter inc ubation periods. Comparisons between summer and winter values in each panel were made for each density (ns = P > 0.05, = P < 0.05, ** = P < 0.001, *** = P < 0.0001). Bars from the same incubation period in each panel with different lette rs are significantly different (P < 0.1), using post-hoc pairwise comparisons among categories when ANOVA P-value < 0.1. Forest floor 0 2 4 6 8 10 12ZeroLowHigha a b b a a ns ns nsCa mineralization (mg kg-1 day-1) Cornus florida stem density Summer Incubation Winter Incubation Cove hardwood forest t yp e Oak hardwood forest t y p e Mineral Soil-1.4 -1.0 -0.6 -0.2 0.2 0.6 1.0a a a a a *** *** *** a Zero Low High Forest floor 0 2 4 6 8 10 12ZeroLowHigha a b a b a ns ns ns Mineral Soil -1.4 -1.0 -0.6 -0.2 0.2 0.6 1.0a b a a c a ** ns *** Zero Low High

PAGE 93

82 Figure 4-3. Conceptual model of calcium (C a) cycling in an eastern United States hardwood forest. Arrow thickness indicat es amount of Ca movement and box size indicates size of ava ilable Ca pool based on da ta from Johnson et al. (1985) and Yanai et al. (2005). Loss of Cornus florida may decrease the size of the forest floor Ca pool and therefore overall Ca availability may be less in oak hardwood and cove hardwood forest types. Wood increment Forest Floor MineralSoil Parent Material Streamwater Litterfall Fauna Ste mfl ow Atmospheric deposition and precipitation

PAGE 94

83 CHAPTER 5 SUMMARY AND CONCLUSION This research project examined the influence of dogwood anthracnose and the ecological role of Cornus florida L. in Great Smoky Mountains National Park (GSMNP). Specifically, the effects of past burning on C. florida survival and health (Chapter 2), the effects of calcium (Ca), potassium (K), and magnesium (Mg) on dogwood density and health (Chapter 3), and role of C. florida in Ca mineralization (Chapter 4) were examined over a three year period. Findi ngs from these three interre lated studies are briefly summarized below. In Chapter 2, we examined C. florida populations in burned and unburned oakhickory stands to determine if burning prio r to anthracnose inf ection has reduced the impacts of anthracnose. We hypothesized that fire has altered stand structure and created open conditions less conducive to dogwood anthr acnose, which is most virulent in moist heavily shaded stands. We compared C. florida density, C. florida foliar infection and crown dieback, stand structure, species composition, Tsuga canadensis (L.) Carr. density, plot species richness, and plot diversity among four sa mpling categories: unburned stands, and stands that had burned once, twice, and three times (single, double, and triple burn stands, respectively) over a 20 year pe riod (late 1960s to la te 1980s). We also analyzed community composition using mu ltivariate analyses. Double burn stands contained the greatest density of C. florida stems (770 stems ha-1) followed by triple burn stands (233 stems ha-1), single burn stands (225 stems ha-1) and unburned stands (70 stems ha-1). While foliar infection ratings did no t differ between categories, we observed

PAGE 95

84 less crown dieback in small trees (< 5 cm dbh) in burned stands than in unburned stands (P < 0.05). Total overstory density was greater in unburned stands (564 stems ha-1) than in double and triple burned stand (317-436 stems ha-1, P < 0.0001), but understory stem density was greater in burned stands (2851-5072 stems ha-1) than unburned stands (2292 stems ha-1, P = 0.024). However, the understory density and importance value of T. canadensis a coniferous species that creates hea vy shading in forest understories, were considerably greater in unburne d stands than in burned stan ds. The results of our study suggest that prescribed fire may offer a management tool to reduce the impacts of dogwood anthracnose in easte rn hardwood forests. In Chapter 3 we found positive correlations between soil Ca, Mg, and K saturation and C. florida stem density and basal area. We tested the effect of these cations at four levels (0, 50, 100, and 200%) of a sta ndard nursery fertilization rate on C. florida seedling survival and resistance to dogwood an thracnose. Although most of the seedlings died after one season of exposure to dogwood an thracnose, we found that seedlings that had lower inputs of Ca and K cations showed higher levels of disease severity sooner than seedlings in other treatments, sugge sting these nutrients play a role in C. florida survival from anthracnose. Magnesium treatment levels did not appear to have an effect on C. florida disease severity or mortality. In Chapter 4 we sampled sixty-eight 10 m x 10 m plots in two fo rest types, cove hardwood and oak hardwood, to quantify the influence of C. florida density on initial exchangeable Ca and Ca mineralizati on in the mineral soil and forest floor Cornus florida density was classified into three levels in both forest type s (zero = 0 stems ha-1, low = 200-300 stems ha-1 and high = > 600 stems ha-1). We found significantly greater

PAGE 96

85 levels of initial exchangeable Ca on high dens ity plots, compared to low density plots in both forest types in the fore st floor and mineral soil (P < 0.01). Calcium mineralization occurred primarily in the forest floor and not in the mineral soil in both forest types. Yearly Ca mineralization was greatest in the high density C. florida plots (cove hardwood, high density 3.3 g kg-1 yr-1 versus zero density 0.6 g kg-1 yr-1, P = 0.04 and oak hardwood, high density 2.4 g kg-1 yr-1 versus zero density 1.1 g kg-1 yr-1, P = 0.09). These results indicate that the loss of C. florida from eastern United States forests will further alter the Ca cycle and may negatively affect the health of eastern hardwood forests. Overall, this project indicates that nu trient availability plays a role in C. florida survival from dogwood anthracnose. Our result s also indicate that prescribed burning offers a management technique to maintain C. florida as a component in eastern hardwood forests. Additionally, our project showed the importance of C. florida in the Ca cycle in eastern hardwood forests. Al though this project took place in Great Smoky Mountains National Park, because of the la rge study area and wide distribution of the forest types we sampled, we believe that our findings are ap plicable in forests across the eastern United States where C. florida occurs.

PAGE 97

86 LIST OF REFERENCES Allen, S. C., S. Jose, P. K. R. Nair, B. J. Brecke, V. D. Nair, D. A. Graetz, and C. L. Ramsey. 2005. Nitrogen mineralization in a pecan ( Carya illinoensis K. Koch)cotton ( Gossypium hirsutum L.) alley cropping system in the southern United States. Biology and Fertility of Soils 41:28-37. Ament, M. M., R. M. Auge, L. F. Gra nd, and M. T. Windham. 1998. An inoculation technique for dogwood anthracnose. Journa l of Environmental Horticulture 16:3741. Anagnostakis, S. L. 2001. The effect of multiple importations of pests and pathogens on a native tree. Biological Invasions 3:245-254. Anagnostakis, S. L., and J. S. Ward. 1996. Th e status of flowering dogwood in five longterm forest plots in Connect icut. Plant Disease 80:1403-1405. Anderson, R. L. 1991. Background, pages 5-9. In R. L. Anderson (ed.), Results of the 1990 dogwood anthracnose impact assessment a nd pilot test in the southeastern United States. U.S. Forest Service S outhern Region. Protection report R8-PR 20. Anderson, R. L., J. L. Knighten, S. E. Do wsett, and C. Henson. 1991. Effectiveness of cultural techniques to control dogwood anthra cnose, pages 39-46. In R. L. Anderson (ed.), Results of the 1990 dogwood anthracnose impact assessment and pilot test in the southeastern United Stat es. U.S. Forest Service Southern Region. Protection report R8-PR 20. Anglberger, H., and E. Halmschlager. 2003. The severity of Sirococcus shoot blight in mature Norway spruce stands with rega rd to tree nutrition, topography and stand age. Forest Ecology and Management 177:221-230. Arthur, M. A., R. D. Paratl ey, and B. A. Blankenship. 1998 Single and repeated fires affect survival and regeneration of woody and herbaceous species in an oak-pine forest. Journal of the Torrey Botanical Society 125:225-236. Augusto, L., M. P. Turpault, and J. Ranger. 2000. Impact of forest tree species on feldspar weathering ra tes. Geoderma 96:215-237. Biondini, M. E., C. D. Bonham, and E. F. Redente. 1985. Secondary successional patterns in a sagebrush ( Artemisia tridentata ) community as they relate to soil disturbance and soil biological activity. Vegetatio 60:25-36.

PAGE 98

87 Blair, J. M. 1988. Nutrient release from deco mposing foliar litter of three tree species with special reference to calcium, magne sium, and potassium dynamics. Plant and Soil 110:49-55. Blair, R. M. 1982. Growth and nonstructural carbohydrate content of southern browse species as influenced by light intensit y. Journal of Range Management 35:756-760. Blankenship, B. A., and M. A. Arthur. 2006. Stand structure over 9 years in burned and fire-excluded oak stands on the Cumber land Plateau, Kentucky. Forest Ecology and Management 225:134-145. Boettcher, S. E., and P. J. Kalisz. 1990. Si ngle-tree influence on soil properties in the mountains of eastern Kentucky. Ecology 71:1365-1372. Boerner, R. E. J., J. A. Brinkman, and E. K. Sutherland. 2004. Effects of fire at two frequencies on nitrogen transformations a nd soil chemistry in a nitrogen-enriched forest landscape. Canadian Journa l of Forest Research 34:609-618. Britton, K. O. 1993. Anthracnose infection of dogwood seedlings exposed to natural inoculum in western North Ca rolina. Plant Disease 77:34-37. Britton, K. O. 1994. Dogwood anthracnose, pages 17-20. In C. Ferguson and P. Bowman (eds.), Threats to forest health in the s outhern Appalachians. Southern Appalachian Man and the Biosphere Cooperative. Gatlinburg, TN. Britton, K. O., W. D. Pepper, D. L. Loftis and D. O. Chellemi. 1994. Effect of timber harvest practices on populations of Cornus florida and severity of dogwood anthracnose in western North Carolina. Plant Disease 78:398-402. Britton, K. O., P. Berrang, and E. Mavity. 1996. Effects of pretreatment with simulated acid rain on the severity of dogwood anthracnose. Plant Disease 80:646-649. Brose, P., T. Schuler, D. Van Lear, and J. Berst. 2001. Bringing fire back: the changing regimes of the Appalachian mixed-oak fo rests. Journal of Forestry 99:30-35. Buell, J. H. 1940. Effect of season of cutting on sprouting of dogwood. Journal of Forestry 38:649-650. Carr, D. E., and L. E. Banas. 1999. Dogwood anthracnose ( Discula destructiva ): effects of and consequences for host ( Cornus florida ) demography. American Midland Naturalist 143:169-177. Chellemi, D. O., and K. O. Britton. 1992. In fluence of canopy microclimate on incidence and severity of dogwood an thracnose. Canadian J ournal of Botany 70:1093-1096. Chellemi, D. O., K. O. Britton, and W. T. Swank. 1992. Influence of site factors on dogwood anthracnose in the Natahala Mount ain Range of western North Carolina. Plant Disease 76:915-918.

PAGE 99

88 Clarke, K. R., and M. Ainsworth. 1993. A method of linking multivariate community structure to environmental variables. Marine Ecology Progress Series 46:213-226. Clinton, B. D., J. A. Yeakley, and D. K. Apsley. 2003. Tree growth and mortality in a southern Appalachian deciduous forest following extended wet and dry periods. Castanea 68:189-200. Conway, W. S., C. E. Sams, R. G. McGuir e, and A. Kelman. 1992. Calcium treatment of apples and potatoes to reduce posth arvest decay. Plant Disease 76:329-334. Cornwell, W. K., and P. J. Grubb. 2003. Regi onal and local patterns in plant species richness with respect to res ource availability. Oikos 100:417-428. Daughtery, M. L., and C. R. Hibben. 1994. Dogwood anthracnose: a new disease threatens two native Cornus species. Annual Review of Phytopathology 32:61-73. Daughtrey, M. L., C. R. Hibben, K. O. Britt on, M. T. Windham, and S. C. Redlin. 1996. Dogwood anthracnose: unders tanding a disease new to North America. Plant Disease 80:349-357. Day, F. P., and C. D. Monk. 1977. Seasonal nutrient dynamics in the vegetation on a southern Appalachian watershe d. Journal of Botany 64:1126-1139. Dijkstra, F. A. and M. M. Smits. 2002. Tree sp ecies effects on calcium cycling: the role of calcium uptake in deep soils. Ecosystems 5:385-398. Dijkstra, F. A. 2003. Calcium mineralization in the forest floor a nd surface soil beneath different tree species in the northeaste rn US. Forest Ecology and Management 175:185-194. Driscoll, C. T., G. B. Lawrence, A. J. Bulger, T. J. Butler, C. S. Cronan, C. Eager, K. F. Lambert, G. E. Likens, J. L. Stoddard, and K. C. Weathers. 2001. Acid deposition in northeastern United States: sources and inputs, ecosystem effects, and management strategies. BioScience 51:180-198. Dufrne M., and P. Legendre. 1997. Species asse mblages and indicator species: the need for a flexible asymmetrical appr oach. Ecological M onographs 67:345-366. Elliott, K. J., L. R. Boring, W. T. Swank, and B. R. Haines. 1997. Successional changes in plant species diversity and compos ition after clearcutting in a southern Appalachian watershed. Forest Ecology and Management 92:67-85. Elliott, K. J., R. L. Hendrick, A. E. Major, J. M. Vose, and W. T. Swank. 1999. Vegetation dynamics after a prescribed fire in the southern Appalachians. Forest Ecology and Management 114:199-213.

PAGE 100

89 Elliot, K. J., L. R. Boring, and W. T. Swank. 2002. Above ground biomass and nutrient accumulation 20 years after clear-cutting a southern Appalachian watershed. Canadian Journal of Forest Research 32:667-683. Eno, C. F. 1960. Nitrate producti on in the field by incubating the soil in polyethylene bags. Soil Science Society of America Proceedings 24:277-279. Epstein, E. 1972. Mineral nutrition of plants: principles and perspe ctives. John Wiley and Sons Inc., New York. 412 p. Erbaugh, D. K., M. T. Windham, A. J. W. St odola, and R. M. Auge. 1995. Light intensity and drought stress as predisposition fact ors for dogwood anthracnose. Journal of Environmental Horticulture 13:186-189. Finzi, A. C., C. D. Canham, and N. va n Breemen. 1998. Canopy tree-soil interactions within temperate forests: species effects on pH and cations. Ecological Applications 8:440-446. Fujinuma, R., J. Bockheim, and N. Balster. 2005. Base cation cyc ling by individual tree species in old-growth forests of Uppe r Michigan, USA. Biogeochemistry 74:357376. Galbraith, S. L., and W. H. Martin. 2005. Thr ee decades of overstory and species change in a mixed mesophytic forest in eastern Kentucky. Castanea 70:115-128. Godman, R. M., and K. Lancaster. 1990. Tsuga canadensis (L.) Carr.Eastern hemlock, pages 604. In R. M. Burns and B. H. Honkala (eds.), Silvics of North America. Volume 1. Conifers. Agricultural Handbook 654. U.S. Forest Service, Washington, D.C. Gould, A. B., and J. L. Peterson. 1994. The effect of moisture and sunlight on the severity of dogwood anthracnose in street trees Journal of Arboriculture 20:75-78. Graveland, J. R., 1996. Avain eggshell formation in calcium-rich and calcium-poor habits: importance of snail shells and anthropogenic calcium sources. Canadian Journal of Zoology 74:1035-1044. Graveland, J. R., r. van der Wal, J. H. van Balen, and A.J. van Noordwijk. 1994. Poor reproduction in passerines from decline of small abundance on acidified soils. Nature 368:446-448. Hamburg, S. P., R. D. Yanai, M. A. Arthur J. D. Blum, and T. G. Siccama. 2003. Biotic control of calcium cycling in northern hardwood forest s: acid rain and aging forests. Ecosystems 6:399-406. Harmon, M. 1982. Fire history of the wester nmost portion of Great Smoky Mountains National Park. Bulletin of the Torrey Botany Club 109:74-79.

PAGE 101

90 Harrod, J., P. S. White, and M. E. Harmon. 1998. Changes in xeric forests in western Great Smoky Mountains National Park, 1936-1995. Castanea 63:346-360. Hendrix, P. F., and P. H. Bohlen. 2002. Exotic earthworm invasions in North America: ecological and policy implic ations. Bioscience 52:801-811. Hibben, C. R., and M. L. Daughtery. 1988. Dogw ood anthracnose in northeastern United States. Plant Disease 72:199-203. Hiers, J. K., and J. P. Evans. 1997. Eff ects of anthracnose on dogwood mortality and forest composition of the Cumberland Plateau (U.S.A.). Conservation Biology 11:1430-1435. Holzmueller, E. J., S. Jose, and M. A. Jenki ns. 2006a. Influence of fire on the density and health of Cornus florida L. (flowering dogwood) populations in Great Smoky Mountains National Park. Ar ticle in preparation. Holzmueller, E. J., S. Jose, and M. A. Jenkins. 2006b. Influence of calcium, potassium, and magnesium on Cornus florida L. density and resistance to dogwood anthracnose. Article in preparation. Holzmueller, E. J., S. Jose, and M. A. Jenkins. 2006c. Influence of Cornus florida L. on calcium mineralization in two southern Appalachian forest types. Article in preparation. Holzmueller, E., S. Jose, M. Jenkins, A. Camp, and A. Long. 2006. Dogwood anthracnose in eastern ha rdwood forests: what is known and what can be done? Journal of Forestry 104:21-26. Hotopp, K. P. 2002. Land snails and soil calciu m in central Appalachian Mountain forest. Southeastern Naturalist 1:27-44. Huddle, J. A., and S. G. Pallardy. 1996. Effects of long-term annual and periodic burning on tree survival and growth in a Missour i Ozark oakhickory forest. Forest Ecology and Management 82:1-9. Huntington, T. G., R. P. Hooper, C. E. J ohnson, B. T. Aulenbach, R. Cappellato, and A. E. Blum. 2000. Calcium depletion in a south eastern United States forest ecosystem. Soil Science Society of Am erica Journal 64:1845-1858. Hutchinson, T. F., E. K. Sutherland, and D. A. Yaussy. 2005. Effects of repeated prescribed fires on the structure, co mposition, and regenera tion of mixed-oak forests in Ohio. Forest Ec ology and Management 218:210-228. Jeffers, D. L., A. F. Schmitthenner, and M. E. Kroetz. 1982. Potassium fertilization effects on phomopsis seed infection, seed qual ity, and yield of soybeans. Agronomy Journal 74:886-890.

PAGE 102

91 Jenkins, M. A., and G. R. Pa rker. 1998. Composition and dive rsity of woody vegetation in silvicultural openings of southern Indiana forests. Forest Ecol ogy and Management 109:57-74. Jenkins, M. A., and P. S. White. 2002. Cornus florida L. mortality and understory composition changes in western Great Sm oky Mountains National Park. Journal of the Torrey Botanical Society 129:194-206. Jenkins, M. A., S. Jose, and P. S. White. 2006. Impacts of an exotic fungal disease and associated changes in community com position and structure on foliar calcium cycling in a mixed-species temperat e forest. Article in preparation Johnson, C. E., C. T. Driscoll, T. G. Siccama, and G. E. Likens. 2000. Element fluxes and landscape position in a northern hardwood fo rest watershed ecosystem. Ecosystems 3:159-184. Johnson, D. W., D. D. Richter, G. M. Love tt, and S. E. Londberg. 1985. The effects of atmospheric deposition on potassium, cal cium, and magnesium cycling in two deciduous forests. Canadian Journa l of Forest Research 15:773-782. Johnson, D. W. and D. E. Todd. 1990. Nutrient cycling in forest of Walker Branch Watershed, Tennessee: roles of uptake a nd leaching in causing soil changes. Journal of Environmental Quality 19:97-104. Johnson, K. D., F. P. Hain, K. S. Johnson, and E. Hastings. 1999. Hemlock resources at risk in the Great Smoky Mountai ns National Park, pages 111-112. In K. A. McManus, K. S. Shields, and D. R. Souto (eds.), Proceedings: Symposium on sustainable management of hemlock ecosy stems in eastern North America. U.S. Forest Service Northeast Forest Experime nt Station. General Technical Report NE267. Knighten, J. L. and R. L. Anderson. 1993. D ogwood anthracnose impact assessment in the southeast, pages 24-28. In : J. L. Knighten and R. L. Anderson (eds.) Results of the 1992 dogwood anthracnose impact assessmen t and pilot test in the southeastern United States. U.S. Forest Service S outhern Region. Protection report R8-PR 24. Knoepp, J. D., B. C. Reynolds, D. C. Cr ossley, and W. T. Swank. 2005. Long-term changes in forest floor processes in sout hern Appalachian forests. Forest Ecology and Management 220:300-312. Kutiel, P., and A. Shaviv. 1992. Effects of soil type, plant composition and leaching on soil nutrients following a simulated forest fire. Forest Ecology and Management 53:329-343. Kuddes-Fischer, L. M., and M. A. Arthur 2002. Response of unders tory vegetation and tree regeneration to a single prescribed fire in oak-pi ne forests. Natural Areas Journal 22:43-52.

PAGE 103

92 Lawrence, G. B., M. B. David, and W. S. Shortle. 1995. A new mechanism for calcium loss in forest floor soils. Nature 378:162-165. Lesica, P., B. McCune, S. Cooper, and W. S. Hong. 1991. Differences in lichen and bryophyte communities between old-growth and managed second-growth forests. Canadian Journal of Botany 69:1745-1755. Likens, G. E., C. T. Driscoll, and D. C. Buso. 1996. Long-term effects of acid rain: responses and recovery of a fore st ecosystem. Science 272:244-246. Likens, G. E., C. T. Driscoll, D. C. Buso, T. G. Siccama, C. E. Johnson, G. M. Lovett, T. J. Fahey, W. A. Reiners, D. F. Ryan, C. W. Martin, and S. W. Bailey. 1998. The biogeochemistry of calcium at Hubba rd Brook. Biogeochemistry 41:89-173. Madden, M., R. Welch, T. Jordan, and P. Jackson. 2004. Digital vegetation maps for Great Smoky Mountains National Park: fina l report. Center for Remote Sensing and Mapping Science, Department of Geogr aphy, University of Georgia, Athens. 120 p. Martin, A. C., H. S. Zim, and A. L. Nels on. 1951. American wildlife and plants: a guide to wildlife food habits. Dover Publishing, Inc., New York. 500 p. Martin, J. G., B. D. Kloeppel, T. L. Schaefer, D. L. Kimbler, and S. G. McNulty. 1998. Aboveground biomass and nitrogen allo cation of ten deciduous southern Appalachian tree species. Canadian J ournal of Forest Research 28:1648-1659. McCune, B., and J. B. Grace. 2002. Analysis of ecological commun ities. MjM software, Gleneden Beach. 300 p. McCune, B., and M. J. Mefford. 1999. PC-O RD. Multivariate Anal ysis of Ecological Data. Version 4.0. MjM software, Gleneden Beach, OR. McGee, G. G., D. J. Leopold, and R. D. Nyland. 1995. Understory response to springtime prescribed fire in two New York tr ansition oak forests. Forest Ecology and Management 76:149. McLaughlin, S. B., and R. Wimmer. 1999. Tansley Review No. 104. Calcium physiology and terrestrial ecosystem pr ocesses. New Phytology 142:373-417. McLemore, B. F. 1990. Cornus florida L.-Flowering dogwood, pages 278-283. In R. M. Burns and B. H. Honkala (eds.), Silvics of North America, Volume 2. Hardwoods. Agriculture Handbook 654. U.S. Forest Service, Washington, D.C. Mehlich, A. 1984. Mehlich 3 soil test extractan t: a modification of mehlich 2 extractant. Communication in Soil Scien ce Plant Analysis 15:1409-1416. Mengel, K. E. A. Kirkby, H. Kosegarten, and T. Appel. 2001. Principles of plant nutrition, 5th edition. Kluwer Academic Publishers, Dordrecht. 864 p.

PAGE 104

93 Metz, L. J. 1952. Calcium content of hardwood litter four times that from pine; nitrogen double. U.S. Forest Service Southeastern Forest Experiment Station. Research Notes Number 14. Mielke, M., and K. Langdon. 1986. Dogwood anthr acnose fungus threatens Catoctin Mt. Park. USDI National Park Service Park Science 6:6-8. Mielke, P. W., Jr. 1984. Permutation methods : a distance function approach. Springer Publishing, New York. 352 p. Muchovej, J. J., R. M. C. Muchovej, O. D. Dhingra, and L. A. Maffia. 1980. Suppression of anthracnose of soybeans by calcium. Plant Disease 64:1088-1089. Muller, R. N. 1982. Vegetation patterns in the mixed mesophytic forest of eastern Kentucky. Ecology 63:1901-1917. NAPAP (National Acid Preci pitation Assessment Program). 1998. Biennial report to Congress: an integrated assessm ent. Washington, D.C. 148 p. National Climatic Data Center. 199-2004. A nnual climatological study, Gatlinburg, TN. National Oceanic and Atmospheric Administration. Asheville, NC. Nwoboshi, L. C. 1980. The influence of pot assium, nitrogen, phosphorus and magnesium fertilization on anthracnos e disease and tuber yield of cassava, pages 193-198. In Proceedings: Potassium Workshop, Internati onal Potash Institute, Ibadan, Nigeria. Oliver, C. D., and B. C. Larson. 1990. Forest stand dynamics. McGraw-Hill Publishing Company, New York. 467 p. Parmeter, J. R., and B. Uhrenholdt. 1975. Some effects of pine-needle or grass smoke on fungi. Phytopathology 65:28-31. Peterson, E. B., and B. McCune. 2001. Diversit y and succession of epiphytic macrolichen communities in low-elevation managed conifer forests in western Oregon. Journal of Vegetation Science 12:511-524. Peterson, D. W., and P .B. Reich. 2001. Prescrib ed fire in oak savanna: fire frequency effects on stand structure and dynamics Ecological Applications 11:914-927. Pirone, P. P. 1980. Parasitic fungus affects regions dogwood. New York Times 24 Feb., pages 34,37. Prescott, C. E., G. D. Hope, and L. L. Blevins. 2003. Effect of gap size on litter decomposition and soil nitrate concentrations in a high-elevation spruce-fir forest. Canadian Journal of Forest Research 33:2210-2220. Pyle, C. 1988. The type and extent of anthr opogenic vegetation distur bance in the Great Smoky Mountains before National Park Service acquisition. Castanea 53:183-196.

PAGE 105

94 Qian, H., K. Klinka, and X. Song. 1999. Cr yptogams on decaying wood in old-growth forests of southern coastal British Colu mbia. Journal of Vegetation Science 10:883894. Redlin, S. C. 1991. Discula destructiva sp. nov., cause of dogwood anthracnose. Mycologia 83:633-642. Rossell, I. M., C. R. Rossell, K. J. Hining, and R. L. Anderson. 2001. Impacts of dogwood anthracnose ( Discula destructiva Redlin) on the fruits of flowering dogwood ( Cornus florida L.): implications for wildlife. American Midland Naturalist 146:379-387. SAS (Statistical Analysis System) Institute Inc. 2002. Version 9.0. SAS Institute Inc., Cary, NC. Schwartz, M. W., S. M. Hermann, and C. S. Vogel. 1995. The catastrophic loss of Torreya taxifolia : assessing environmental induc tion of disease hypotheses. Ecological Applications 5:501-516. Schwegman, J. E., W. E. McClain, T. L. Esker, and J. E. Ebinger, 1998. Anthracnosecaused mortality of flowering dogwood ( Cornus florida ) at the Dean Hill Nature Preserve, Fayette, County, Illinois, US A. Natural Areas Journal 18:204-207. Sherald, J. L., T. M. Stidham, J. M. Hadi dian, and J. E. Hoel dtke. 1996. Progression of the dogwood anthracnose epidemic and the status of flowering dogwood in Catoctin Mountain Park. Plant Disease 80:310-312. Sij, J. W., F. T. Turner, and N. G. Wh itney. 1985. Suppression of anthracnose and phomopsis seed rot on soybean with pota ssium fertilizer and benomyl. Journal of Agronomy 77:639-642. Smith, W. B., J. S. Vissage, D. R. Darr, a nd R. M. Sheffield. 2001. Forest resources of the United States, 1997. U.S. Forest Serv ice North Central Research Station. General Technical Report NC-219. Stiles, E. W. 1980. Patterns of fruit presentati on and seed dispersal in bird-disseminated woody plants in the eastern deciduous fo rest. American Naturalist 116:670-688. Sugimoto, T., M. Aino, M. Sugimoto, and K. Watanabe. 2005. Reduction of Phytophthora stem rot disease on soybeans by the application of CaCl2 and Ca(NO3)2. Journal of Phytopathology 153:536-543. Thomas, W. A. 1969. Accumulation and cycling of calcium by dogwood trees. Ecological Monographs 39:101-120. Tuininga, A. R., and J. Dighton. 2004. Change s in ectomycorrhizal communities and nutrient availability following prescribed burns in two upland pine-oak forests in the New Jersey pine barrens. Canadian Journal of Forest Research 34:1755-1765.

PAGE 106

95 Trigiano, R. N., G. Caetano-Anolles, B. J. Bassam, and M. T. Windham. 1995. DNA amplification fingerprinti ng provides evidence that Discula destructiva the cause of dogwood anthracnose in No rth America, is an introduced pathogen. Mycologia 87:490-500. Wreborn, I. 1992. Changes in land mollusk fauna and soil chemistry in an inland district in southern Sweden. Ecography 15:62-69. Washburn, C. S. M., and M. A. Arthur. 2003. Spatial variability in soil nutrient availability in an oak-pine forest: potentia l effects of tree species. Canadian Journal of Forest Research 33:2321-2330. Welch, R., M. Madden, and T. Jordan. 2002. Photogrammetric and GIS techniques for the development of vegetation databa ses of mountainous areas: Great Smoky Mountains National Park. Journal of P hotogrammetry and Remote Sensing 57:5368. White, R. D., K. D. Patterson, A., Weakley, C. J., Ulrey, and J. Drake. 2003. Vegetation classification of Great Smoky Mountains National Park. Report submitted to BRDNPS Vegetation Mapping Program. Nature Serve, Durham, North Carolina. 376 p. Wilds, S. P. 1997. Gradient analysis of the distribution of a fungal disease of Cornus florida in the southern Appalachian Mountai ns, Tennessee. Journal of Vegetation Science 8:811-818. Williams, C. E., and W. J. Moriar ity. 1999. Occurrence of flowing dogwood ( Cornus florida L.), and mortality by dogwood anthracnose ( Discula destructiva Redlin), on the northern Allegheny Plateau. Journal of the Torrey Botanical Society 126:313319. Wilmot, T. R., D. S. Ellsworth, and M. T. Tyree. 1996. Base cati on fertilization and liming effects on nutrition and growth of Vermont sugar maple stands. Forest Ecology and Management 84:123-134. Windham, M. T., and M. E. Montgomery. 1990. A survey to determine the incidence and severity of dogwood anthracnose in Gr eat Smoky Mountains National Park. Summary report to the National Park Se rvice, Great Smoky Mountains National Park, Gatlinburg, TN. 18 p. Wisniewski, M., S. Droby, E. Chalutz, and Y. Eilam. 1995. Effects of Ca2+ and Mg2+ on Botrytis cinerea and Penicillium expansum in vitro and on the biocontrol activity of Candida oleophila Plant Pathology 44:1016-1024. Woods, K. 2000. Dynamics in late-successi onal hemlock-hardwood forests over three decades. Ecology 81:110.

PAGE 107

96 Wyckoff, P. H., and J. S. Clark. 2002. The re lationship between grow th and mortality for seven co-occurring tree species in the southern Appalach ian Mountains. Journal of Ecology 90:604-615. Yamazaki, H., S. Kikuchi, T. Hoshina, and T. Kimura. 1999. Effect of calcium concentration in nutrient solution be fore and after inoculation with Ralstonia solanacearum on resistance of tomato seedlings to bacterial wilt. Soil Science and Plant Nutrition 45:1009-1014. Yanai, R. D., J. D. Blum, S. P. Hamburg, M. A. Arthur, C. A. Nezat, and T. G. Siccama. 2005. New insights into calcium depletion in northeastern forests. Journal of Forestry 103:14-20.

PAGE 108

97 BIOGRAPHICAL SKETCH Eric Holzmueller was born in Des Moin es, Iowa, in 1978. He received a Bachelor of Science and masters degree in forestry in 1999 and 2002, respectively, from Iowa State University, Ames, Iowa. In May of 2002 he began field work for his doctoral degree in Great Smoky Mountains National Park and began ta king classes in September 2002 at the University of Florida, Gainesville, FL. After graduation, Eric will continue to work at the University of Florida School of Forest Resources and Conservation as a Postdoctoral Research Associate wi th his advisor, Dr. Shibu Jose.


Permanent Link: http://ufdc.ufl.edu/UFE0013140/00001

Material Information

Title: Ecology of flowering dogwood (Cornus florida L.) in response to anthracnose and fire in Great Smoky Mountains National Park, USA
Physical Description: Mixed Material
Copyright Date: 2008

Record Information

Source Institution: University of Florida
Holding Location: University of Florida
Rights Management: All rights reserved by the source institution and holding location.
System ID: UFE0013140:00001

Permanent Link: http://ufdc.ufl.edu/UFE0013140/00001

Material Information

Title: Ecology of flowering dogwood (Cornus florida L.) in response to anthracnose and fire in Great Smoky Mountains National Park, USA
Physical Description: Mixed Material
Copyright Date: 2008

Record Information

Source Institution: University of Florida
Holding Location: University of Florida
Rights Management: All rights reserved by the source institution and holding location.
System ID: UFE0013140:00001


This item has the following downloads:


Full Text












ECOLOGY OF FLOWERING DOGWOOD (Cornusflorida L.) IN RESPONSE TO
ANTHRACNOSE AND FIRE IN GREAT SMOKY MOUNTAINS NATIONAL PARK,
USA















By

ERIC HOLZMUELLER


A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY

UNIVERSITY OF FLORIDA


2006





























Copyright 2006

by

Eric Holzmueller















ACKNOWLEDGMENTS

I would like to express my thanks to my graduate committee chair, Dr. Shibu Jose,

for his guidance, suggestions, and support in completing this project. Thanks also go to

my committee members, Drs. Alan Long, Debbie Miller, and Wendell Cropper, for their

contributions to this study. Special thanks go to committee member Dr. Mike Jenkins for

his input, help, and support with all phases of this project. I am also deeply grateful to the

Great Smoky Mountains National Park Inventory and Monitoring Program for providing

me with data collection assistance, and an office and housing during multiple field

seasons. I would also like to thank the faculty at the IFAS-Statistics Division for their

help in data analyses and Ken Ford for his assistance in the potted plant experiment.

I would like to thank the National Park Service Southeast Region Natural

Resources Preservation Program, Great Smoky Mountains Association, and the

University of Florida College of Agriculture and Life Sciences for providing funding for

this project. Without their financial support, this project would not have been possible.

Finally I would like to thank my family and friends for their support and encouragement.
















TABLE OF CONTENTS



A C K N O W L E D G M E N T S ......... .................................................................................... iii

LIST OF TABLES ............. ..... ......................... .......... ............ vi

LIST OF FIGURES ............. .. ..... ...... ........ ....... .......................... viii

A B ST R A C T .......... ..... ...................................................................................... x

CHAPTER

1 IN TR OD U CTION ............................................... .. ......................... ..

State ent of the P problem .................................................................................. 1
R review of Literature .................. ..................................... .. ........ .. ..
Cornusfl orida .............................................................. 2
Biology of Dogwood Anthracnose............................................... .................. 3
Im pacts of D ogw ood Anthracnose .................................. ............ .................. 4
Factors Affecting Dogwood Anthracnose..........................................................5
Ecological Significance of Cornus florida........................... ................. ...6
Specific Objectives ........ ............ ................. ............ ...................

2 INFLUENCE OF FIRE ON THE DENSITY AND HEALTH OF Cornusflorida
L. (FLOWERING DOGWOOD) POPULATIONS IN GREAT SMOKY
M OUN TA IN S N A TION AL PARK ..........................................................................11

Introdu action ...................................... ................................ ......... ...... 11
M materials and M methods ....................................................................... .................. 15
Study Site............................................. 15
F ie ld S a m p lin g ............................................................................................... 1 6
D ata A n aly sis ................................................................................ 16
R esu lts .....................................19.............................
Cornusflorida Stem Density ............................................................19
Foliage Health and Crown Dieback ..................................... ......... ......20
Stand Structure ............................................................ 21
Overstory Community Composition .................................. ...............21
Understory Community Composition ........................................ ......22
Importance Values for Understory Species .................................................. 22
Tsuga canadensis Stem Density ............................................. ............... 23









D isc u ssio n ............................................................................................................. 2 3
M anagem ent Im plications ................................................ .............................. 28
C conclusion ...................................................................................................... ....... 29

3 INFLUENCE OF CALCIUM, POTASSIUM, AND MAGNESIUM ON Cornus
florida L. DENSITY AND RESISTANCE TO DOGWOOD ANTHRACNOSE..... 39

In tro d u ctio n ...................................... ................................................ 3 9
M materials and M methods ....................................................................... ..................4 1
S tu d y S ite .......................................................4 1
F orest Soil Sam pling ................................................... .. ........ ...... ............42
L ab A n aly sis .....................................................42
Potted Plant Experim ent ................................................................ ............43
Statistical A naly sis ........................ ................ .. .. .... ........... 45
F o re st S o il ..............................................................4 5
Potted Plant E xperim ent...................................... ........................... ........ 45
R e su lts ................... .................................... .................... ................4 5
Forest Soil C ation Saturation......................................... .......................... 45
Potted Plant Experim ent .......................................................... ............... 46
Calcium Treatm ents .................................. .....................................46
Potassium Treatm ents ........................................................................... 46
M agnesium Treatm ents ...................................................... ..... .......... 47
Foliar Cation Concentrations ............................................ ............... 47
D isc u ssio n ............................................................................................................. 4 8
C conclusion ...................................................................................................... ....... 52

4 INFLUENCE OF Cornusflorida L. ON CALCIUM MINERALIZATION IN
TWO SOUTHERN APPALACHIAN FOREST TYPES ..........................................64

In tro d u ctio n .......................................................................................6 4
M materials and M methods ....................................................................... ..................66
S tu d y A re a ..................................................................................................... 6 6
F ie ld S a m p lin g ............................................................................................... 6 7
L laboratory A analysis ............................................ .. ........ .... ...........68
Statistical A naly sis ........................ ................ .. .. .... ........... 68
R esu lts ....................... .............. .........................................................69
Forest Floor M ass and D epth ........................................ ......................... 69
Soil pH .......................................... .......... 69
Initial E x changeable C a............................................................ .....................70
Ca mineralization..................... ..... ... .. .. ..................... 70
D isc u ssio n ............................................................................................................. 7 1
C conclusion ...................................................................................................... ....... 76

5 SUMMARY AND CONCLUSION ................................... ......................................83

L IST O F R E F E R E N C E S ........................................................................ .....................86

B IO G R A PH IC A L SK E TCH ..................................................................... ..................97


v















LIST OF TABLES


Table page

2-1 Scales for foliage health (% of foliage with signs of anthracnose) and crown
dieback (% of crown dieback) used to assess the level of disease severity of
dogwood anthracnose on Cornusflorida trees ....................................... ........... 30

2-2 Mean foliage and crown health ( 1 SE) for Cornusflorida for five diameter
classes in the four different sampling categories using the Mielke-Langdon
Index (Mielke and Langdon 1986)................. ........... ...............31

2-3 Mean understory and overstory basal area and stem density ( 1 SE) in the four
sam pling categories. ............................. ......... .. ...... ..... ...... ...... 32

2-4 Mean species richness and Shannon's diversity index ( 1 SE) for the understory
and overstory in the four sampling categories ............................................ ............ 33

2-5 Overstory and understory indicator values (IndVal) (percent of perfect
indication) and associated sampling category ................................... ..... .......... 34

2-6 Mean importance values ( 1 SE) of selected understory species in the four
sampling categories ............. ...... ...... ........... .. .......... .... 35

3-1 Content of the base fertilizer mix.... ....................................53

3-2 Inputs (%) added to the base fertilizer mix (separately) for each treatment ............54

3-3 Scale used to assess the severity of dogwood anthracnose infection on the
foliage of Cornusflorida seedlings. Scale was based on the Mielke-Langdon
Index (Mielke and Langdon 1986)........... .................... ..................55

3-4 Biweekly infection ratings for Cornusflorida seedlings for length of the
ex p erim en t ...................... .. .. ......... .. .. .............................................. 5 6

3-5 Foliar calcium (Ca), potassium (K), and magnesium (Mg) concentrations (%) for
selected species in a southern Appalachian forest. Data presented is from Day
an d M onk (1977) ................................................................................. 57

4-1 Mineral soil and forest floor mean pH ( 1 SE) for summer and winter
incubations in the cove hardwood and oak hardwood forest types........................77









4-2 Mean yearly Ca mineralization ( 1 SE) in forest floor, mineral soil, and
combined total (forest floor plus mineral soil) for the three Cornusflorida
sampling densities in the cove hardwood and oak hardwood forest types...............78

4-3 Foliar calcium concentrations (%) from dominant species in the two forest types
(oak hardwood and cove hardwood) sampled in this study .................................79















LIST OF FIGURES


Figure page

1-1 Native range of Cornusflorida and the documented range of dogwood
anthracnose in the eastern United States .............. ......... ................... .............. 9

1-2 The effects of a 1976 wildfire on Cornusflorida stem density in Great Smoky
M mountains N national Park................................................. ............................. 10

2-1 Mean Cornusflorida stem density (+ 1 SE) in the four sampling categories for
three diameter classes and total stems ha-1 for all diameter classes......................36

2-2 Nonmetric multidimensional scaling sample ordination of overstory and
understory communities, showing the relative differences in community
composition separated by sampling categories ............................. ...................37

2-3 Mean Tsuga canadensis stem density ( 1 SE) in five diameter size classes and
total (all diameter classes combined) in the four sampling categories...................38

3-1 Linear regression between soil calcium (Ca) saturation and Cornusflorida stem
den sity an d b asal area ............. ......................................................... .... .. .. .... 58

3-2 Linear regression between soil potassium (K) saturation and Cornusflorida stem
density and basal area............... .... .. ................. ..... ....... 59

3-3 Linear regression between soil magnesium (Mg) saturation and Cornusflorida
stem density and basal area ............................................. ............................. 60

3-4 Biweekly mortality (%) of Cornusflorida seedlings for the four treatment levels
o f ea ch catio n ...................................................................... 6 1

3-5 Foliar calcium (Ca), potassium (K), and magnesium (Mg) concentrations of
Cornusflorida seedling foliage for the four treatment levels of each cation...........62

3-6 Precipitation data for 2004 and previous 5 year average (1999-2003) during
April-September at the Twin Creeks Natural Resources Center, Great Smoky
M mountain s N national P ark ............................................................... .....................63

4-1 Mean initial exchangeable Ca levels ( 1 SE) in the forest floor and mineral soil
in the cove hardwood and oak hardwood forest types during summer and winter
co llectio n tim es .................................................... ................ 8 0









4-2 Mean Ca mineralization ( 1 SE) for the forest floor and mineral soil in the cove
hardwood and oak hardwood forest types during summer and winter incubation
p erio d s ......... ..... ........... .............. ...... ... .................. ............... 8 1

4-3 Conceptual model of calcium (Ca) cycling in an eastern United States hardwood
forest. Arrow thickness indicates amount of Ca movement and box size indicates
size of available C a pool .......................... .................. ............ .... ...... ...... 82















Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor of Philosophy

ECOLOGY OF FLOWERING DOGWOOD (Cornusflorida L.) IN RESPONSE TO
ANTHRACNOSE AND FIRE IN GREAT SMOKY MOUNTAINS NATIONAL PARK,
USA

By

Eric Holzmueller

May 2006

Chair: Shibu Jose
Major Department: Forest Resources and Conservation

Cornusflorida L. (flowering dogwood), a common understory tree species in

eastern forests, is currently threatened throughout its range by a fungus (Discula

destructive Redlin) that causes dogwood anthracnose. This disease rapidly kills C. florida

trees and mortality has exceeded 90% in some forest types. The health and ecological

integrity of forest ecosystems throughout the eastern United States are threatened by the

decline of C. florida populations, but management techniques to control anthracnose have

received little attention. Hence, the objectives of this project were to determine (1) the

influence of past burning on C. florida density and health, (2) how nutrient levels

influence the density and health of C. florida, and (3) the role of C. florida in calcium

(Ca) cycling.

We examined C. florida populations in burned and unburned oak-hickory stands in

Great Smoky Mountains National Park to determine if burning prior to anthracnose









infection has reduced the impacts of the disease. Burned stands contained greater C.

florida densities and lower disease severity than unburned stands.

Nutrient availability has been hypothesized as a factor that influences dogwood

anthracnose severity on C. florida. We studied the influence of Ca, potassium (K), and

magnesium (Mg) on C. florida density and resistance to dogwood anthracnose. We found

positive correlations between soil Ca, K, and Mg saturation and C. florida density in oak-

hickory forests. We also found that seedlings grown in soil with lower inputs of Ca and K

cations exhibited higher disease severity earlier in the growing season than seedlings

grown with greater inputs of Ca and K.

Cornusflorida is thought to play an important role in the Ca cycle because of the

high Ca concentration found in the foliage. We quantified the relationship of C. florida

density on Ca mineralization in the mineral soil and forest floor. Calcium mineralization

occurred primarily in the forest floor and was generally greatest in the high density C.

florida plots.

Our research showed a positive correlation between C. florida density and soil Ca,

K, and Mg saturation. Higher levels of soil Ca and K may alleviate disease severity in C.

florida. Further, our results indicate that prescribed fire may provide an important

management tool to reduce disease incidence and severity in oak-hickory forests. We also

found that the loss of C. florida from eastern forests has reduced the rate of soil and forest

floor Ca mineralization, which may have negative effects on many associated flora and

fauna.














CHAPTER 1
INTRODUCTION

Statement of the Problem

Over the last quarter century, a fungal disease has severely impacted Cornus

(dogwood) species along the Pacific seaboard and throughout the eastern United States.

Redlin (1991) identified the fungus Discula destructive Redlin as the causal agent for the

disease, dogwood anthracnose. Two species, Cornus nuttallii Aud. (Pacific dogwood)

and Cornusflorida L. (flowering dogwood), are the most susceptible to dogwood

anthracnose and have experienced heavy mortality because of the disesae (Daughtery and

Hibben 1994). Although not proven, it is believed that dogwood anthracnose is an exotic

disease brought to North America from Asia (Britton 1994). Because Cornus kousa L.

(Oriental dogwood) is quite resistant to anthracnose, it is suspected that the disease was

brought from overseas on trees of this species that did not show symptoms (Britton

1994). In a study of Discula genetic diversity, Trigiano et al. (1995) found that D.

destructive was highly homogenous across its broad continental range compared to other

Discula species, suggesting that D. destructive is a recently introduced fungus still

undergoing intense selection pressure. The sudden appearance of the disease on both

coasts of the United States and its rapid spread in the eastern half of the country support

this assumption.

Historically, C. florida was one of the most common understory trees in the eastern

United States (Muller 1982, Elliott et al. 1997, Jenkins and Parker 1998). However,

dogwood anthracnose has severely impacted this species and heavy mortality of C.









florida (attributed to dogwood anthracnose) has occurred throughout the eastern United

States (Anagnostakis and Ward 1996, Sherald et al. 1996, Hiers and Evans 1997,

Schwegman et al. 1998, Carr and Banas 1999, Williams and Moriarity 1999, Jenkins and

White 2002). For example, Anagnostakis and Ward (1996) reported a mortality rate of

86% between 1977 and 1987 on long-term plots in Connecticut and a 77% reduction of

stems was observed in Catoctin Mountain Park in Maryland between 1976 and 1992

(Sherald et al. 1996). Currently, there are no management options for controlling

dogwood anthracnose in large forested areas.

Review of Literature

Cornusflorida

Cornusflorida is primarily an understory species found in the eastern United

States, from the southern tip of Maine to the northern half of Florida and as far west as

Oklahoma (McLemore 1990). Most common as a small understory tree, average height

for C. florida ranges from 5-12 m and average stem diameter ranges from 3-8 cm.

Although C. florida grows best on fertile well-drained soils, it can be found on Ultisols,

Inceptisols, Alfisols, Spodosols, and Entisols (McLemore 1990). Cornusflorida is a

shade tolerant species and maximum photosynthesis occurs at about one-third of full

sunlight (McLemore 1990). Cornusflorida is a thin barked species that is easily damaged

by logging and fire, however, it will prolifically stump sprout when damaged (Buell

1940, Kuddes-Fischer and Arthur 2002, Blankenship and Arthur 2006). Its large

geographical range (Figure 1-1), ability to grow on a variety of sites, tolerance of shade,

and sprouting ability enables C. florida to grow in association with a wide-range of tree

species and in multiple stand conditions. Cornusflorida is commonly associated with

Quercus species and mesic hardwoods such as Liriodendron tulipifera L., and also occurs









under the shade of Pinus species (McLemore 1990). Cornusflorida is common in many

second growth forests, and is also common in old growth and undisturbed areas as well

(Harrod et al. 1998, Jenkins and Parker 1998).

Biology of Dogwood Anthracnose

Dogwood anthracnose progressively attacks all aboveground parts of infected trees.

Although the disease can affect trees of any size, smaller trees are more susceptible

(Mielke and Langdon 1986, Hiers and Evans 1997). The edges of the leaves on the lower

branches show the first symptoms of the disease, developing black spots that extend

down the leaf margins under suitable conditions (Britton 1994). Anthracnose then spreads

from the leaves into the twigs of the infected trees where the fungus overwinters.

In the spring, reproductive structures ofD. destructive form underneath leaf spots

and on the surface of twig cankers. Numerous asexual spores ooze out in slimy beige

droplets from buds and twigs. Local dispersal of the spores occurs by splashing rain,

while longer distance dispersal may be facilitated by insects and birds (Sherald et al.

1996). The fungus eventually reaches the bole of the tree where cankers develop, girdling

and killing the tree (Mielke and Langdon 1986). Trees also die from repeated defoliation,

with smaller trees dying first (Britton 1994). Trees may die within 1-5 years of first

infection; saplings may die in the same year they are infected.

There are many signs of dogwood anthracnose infection that can be readily

identified. Leaves of infected trees typically have one or both of two types of leaf spots.

Irregular light brown spots with reddish brown borders are formed when environmental

conditions are less conducive to the fungus. These may be concentrated around the lower

edges of the leaf, but may also appear scattered across the lamina. Under favorable

disease conditions, the other type of leaf spot forms, a leaf blight with black, water-









soaked lesions typically initiating at the leaf tip and expanding upward along the mid-

vein into the twig. Infected trees may display an umbrella-like canopy due to the loss of

lower branches (Mielke and Langdon 1986, Daughtrey et al. 1996). In place of lower

branches, epicormic shoots often develop, but are quickly infected with anthracnose.

Since C. florida has the ability to produce stump sprouts, a dead stem surrounded by

many sprouts is a common occurrence following anthracnose-caused mortality. These

new shoots, however, are typically infected and will not likely mature to replace the dead

tree.

Impacts of Dogwood Anthracnose

Since its first appearance in New York in 1978 (Pirone 1980), the spread of

anthracnose throughout the eastern half of the United States has been rapid (Figure 1-1).

The disease was widespread in nine northeastern States by 1987 (Connecticut, Delaware,

Maryland, Massachusetts, New Jersey, New York, Pennsylvania, Virginia, and West

Virginia) (Anderson 1991). By 1992 the disease had reached the Carolinas, Georgia,

Alabama, Tennessee, and Kentucky (Knighten and Anderson 1993). Daughtrey et al.

(1996) reported that the disease had also moved west to the states of Missouri, Illinois,

Indiana, Ohio, and Michigan.

A recent study by Wyckoff and Clark (2002) at the Coweeta Hydrologic

Laboratory in the southern Appalachian Mountains documented the rapid decline of C.

florida, with a mortality rate of 15% within a 5 year (1993-1998) period. In Great Smoky

Mountains National Park (GSMNP), analysis of long-term vegetation monitoring data

revealed several alarming trends (Jenkins and White 2002). Between the two sampling

intervals (1977-1979 and 1995-2000), dramatic decreases in C. florida stem density were

observed in typic cove, acid cove, alluvial, oak-hickory, and oak-pine forest types.









Overall decline was greatest in acid cove forests, where mean C. florida stem density

decreased by 94% (101 stems ha-1 to 6 stems ha-1). Typic cove and alluvial forests

exhibited the next greatest decreases in C. florida stem density (92% for both forest

types). Mean density decreased from 180 stems ha-1 to 14 stems ha-1 in typic cove forests

and from 364 stems ha-1 to 28 stems ha-1 in alluvial forests. The two driest forest types,

oak-hickory and oak-pine, exhibited declines of 80% and 69%, respectively. Prior to

anthracnose, C. florida stem density in oak-hickory forests was the second highest of any

forest type (298 stems ha-1), however, mean density decreased to 61 stems ha-1 in 1995-

2000. Mean density decreased from 132 stems ha-1 to 41 stems ha-1 in oak-pine forests.

New C. florida seedlings are not replacing dead C. florida trees in infected stands.

Trees infected with anthracnose produce fewer seeds (Rosell et al. 2001), and reduced

seed production combined with the susceptibility of smaller trees to the disease has

drastically decreased regeneration. Seedlings and saplings were reported to be absent in

several studies (Sherald et al. 1996, Hiers and Evans 1997), indicating the severity of the

problem.

Factors Affecting Dogwood Anthracnose

Many environmental variables influence the spread and virulence of dogwood

anthracnose, but moisture is probably the most critical (Britton 1993). The disease is

more severe on shaded and moist northeast-facing slopes than on southwest-facing slopes

with open canopies (Chellemi et al. 1992, Clinton et al. 2003). In a long-term vegetation

study in GSMNP, Jenkins and White (2002) reported higher levels of mortality attributed

to dogwood anthracnose in more mesic forests typicc and acid coves and alluvial

communities) compared to more xeric forest types (oak-hickory and oak-pine

communities). Jenkins and White (2002) also reported an increased number of C. florida









stems on three plots after a 1976 wildfire (Figure 1-2), indicating stand conditions can be

manipulated to increase C. florida survival from dogwood anthracnose. However, the

effects of past burning on C. florida populations have not been fully investigated.

Although moisture may be the most important environmental variable affecting the

impacts of dogwood anthracnose on C. florida, another key variable that has yet to be

fully explored is nutrient availability. In a greenhouse study by Britton et al. (1996),

simulated acid rain increased the susceptibility of C. florida to anthracnose. The authors

hypothesized that the increase in susceptibility was largely because of soil-mediated

impacts that reduced the availability of soil cations. Calcium (Ca), potassium (K), and

magnesium (Mg) are nutrients that have been linked to disease resistance in other plant

species and diseases (Sij et al. 1985, Anglberger and Halmschlager 2003, Sugimoto et al.

2005). Sugimoto et al. (2005) found a reduction of Phytophthora sojae Kaufmann and

Gerdemann (stem rot) with the application of Ca on two cultivars of Gycline max L.

Merr. cv. Chusei-Hikarikuro (black soybean) and cv. Sachiyutaka (white soybean) in

Japan. Sij et al. (1985) reported that increased rates ofK fertilizer significantly decreased

Colletotrichum dematium (Pers.ex Fr.) Grove var. truncata (Schuv.) Arx. anthracnosee)

in field grown G. max plants in Texas. In Austria, severity of Sirococcus conigenus

(shoot blight) was increased on Picea abies (Norway spruce) trees that had needles with

low levels of Mg (Anglberger and Halmschlager 2003). It is possible that bioavailability

of soil Ca, K and Mg plays a role in resistance to dogwood anthracnose in C. florida as

well.

Ecological Significance of Cornusflorida

The loss of C. florida from stands throughout the eastern United States will likely

have serious ecological effects. An individual tree may produce up to 10 kg of fruit each









fall (Rossell et al. 2001). There are more than 50 species of birds, including neotropical

birds during fall migration, as well as a number of small game species that are known to

eat the fruit (Martin et al. 1951, Stiles 1980). Cornusflorida twigs are also an important

source of browse for white-tailed deer and other herbivores (Blair 1982).

Cornusflorida is also important in nutrient cycling within forest communities. In

eastern mixed hardwood forests, Ca released through mineral weathering is generally

insignificant (Huntington et al. 2000, Dijkstra and Smits 2002). As a result, the release of

Ca through organic matter decomposition mineralizationn) is considered the major source

of Ca for immediate plant uptake for all tree species (Dijkstra and Smits 2002). Dijkstra

(2003) reported that for most species, Ca mineralization beneath the canopy of a given

species primarily occurs in the forest floor (from leaf litter) as opposed to the mineral

soil. Decomposition of C. florida foliage is very rapid compared to other species (Blair

1988, Knoepp et al. 2005), and C. florida litter contains high concentrations (2.0-3.5%)

of Ca (Thomas 1969, Blair 1988). Because of the high Ca concentration in its foliage,

quick decomposition, and abundance in the understory, C. florida has long been believed

to influence Ca availability in the mineral soil and forest floor by acting as a "Ca pump"

in forests (Thomas 1969, Jenkins et al. 2006). High Ca concentration in C. florida foliage

could mean potentially high rates of Ca mineralization in the forest floor and mineral soil.

If high mineralization rates occur, there is a potential for high Ca availability in the soil.

Apart from anecdotal evidence, the impacts of C. florida on Ca cycling, however, have

not been quantified.

Specific Objectives

As discussed previously, the roles of past burning and bioavailability of nutrients in

determining C. florida population dynamics following infection with dogwood









anthracnose have not been fully investigated. In addition, whether higher foliar Ca

concentration in C. florida foliage translates into higher rates of Ca mineralization in the

forest floor and mineral soil remains undetermined. Therefore, the objectives of this

research project were to:

1. Determine the influence of past burning on (1) C. florida density and health and
(2) overall stand structure and species composition in oak-hickory forests where
C. florida is historically found.

2. Examine the effects of soil cation availability (Ca, K, and Mg) on the health and
survival of C. florida.

3. Determine the relationship between C. florida density and Ca mineralization
rates in two forest types (cove hardwood and oak hardwood) where C. florida is
a common understory species.

The following three chapters describe the results of field surveys and experiments

conducted in Great Smoky Mountains National Park to address these three objectives.






























Native range of Counties reporting
Cornus florida dogwood anthracnose


Figure 1-1. Native range of Cornusflorida and the documented range of dogwood
anthracnose in the eastern United States (Based on data from the U.S. Forest
Service Southern and Northeastern Forest Research Stations).












600


01979
S-.400 02000
itu

E

S2000



0
1-4.9 cm 5.0-9.9 cm 1-4.9 cm 5.0-9.9 cm
Unburned Burned
Diameter class (cm) by burn treatment


Figure 1-2. The effects of a 1976 wildfire on Cornusflorida stem density in Great
Smoky Mountains National Park. While stem density decreased drastically in
unburned areas during the 21 year study period (P < 0.001), density increased
(245%, P = 0.159) in the 1.0-4.9 cm class and remained stable in the 5.0-9.9
cm class in burned areas (P = 0.334) (Based on data from Jenkins and White
2002).














CHAPTER 2
INFLUENCE OF FIRE ON THE DENSITY AND HEALTH OF Cornusflorida L.
(FLOWERING DOGWOOD) POPULATIONS IN GREAT SMOKY MOUNTAINS
NATIONAL PARK

Introduction

Historically, Cornusflorida L. (flowering dogwood) was one of the most common

understory species in eastern United States hardwood forests (Muller 1982, Elliott et al.

1997, Jenkins and Parker 1998). According to McLemore (1990), C. florida occurs on a

variety of soils (Ultisols, Inceptisols, Alfisols, Spodosols, and Entisols), is shade tolerant

(maximum photosynthesis occurs at about one-third of full sunlight), and has a large

geographical range (southern Maine to northern Florida and as far west as Oklahoma).

These factors enable the species to grow in association with a wide-range of tree species

including Quercus species, Pinus species, and Liriodendron tulipifera L. Cornusflorida

is most often found in post-logged secondary forests, and also occurs within tree-fall gaps

of old growth forests (Muller 1982, Jenkins and Parker 1998). Because of its ability to

grow in shaded conditions and thin bark, C. florida is not considered a fire dependent

species. However, it sprouts prolifically when top-killed by fire (Kuddes-Fishcher and

Arthur 2002, Blankenship and Arthur 2006).

Cornusflorida is an ecologically important species in forests throughout the eastern

United States. An individual C. florida tree may produce up to 10 kg of protein-rich fruit

each fall, and more than 50 species of birds as well a number of small mammal species

are known to eat the fruit (Martin et al. 1951, Rossell et al. 2001). However, C. florida's

most important role may be in the annual cycling of calcium in forest communities.









Cornusflorida foliage decomposes very rapidly compared to other species, and C. florida

litter contains large amounts of calcium (2.0-3.5%) (Thomas 1969, Blair 1988, Knoepp et

al. 2005). As a result, C. florida serves an important ecological function as a calcium

"pump" for associated plant species and forest floor biota.

Dogwood anthracnose, a disease caused by the fungus Discula destructive Redlin

(believed to be an exotic disease from Asia) has become a serious disease of C. florida

over the past 20 years. Dogwood anthracnose was first identified in the late 1970s in New

York (Pirone 1980), and has since spread rapidly throughout the eastern United States,

infecting C. florida populations throughout much of its range (Holzmueller et al. 2006).

Following infection by anthracnose, mortality rates of C. florida have been as high as

90% (Anagnostakis and Ward 1996, Sherald et al. 1996, Hiers and Evans 1997, Jenkins

and White 2002). Although the disease can infect trees of any size, smaller trees are more

susceptible than large trees and often die from repeated defoliation within 1-5 years of

infection (Mielke and Langdon 1986, Hibben and Daughtery 1988). The fungus also

causes twig dieback and stem cankers, which can eventually girdle the tree (Hibben and

Daughter 1988).

The rapid spread of dogwood anthracnose and high mortality of C. florida make it

somewhat similar to the effects of Cryphonectriaparasitica (Murill) Barr (Asian chestnut

blight fungus) on Castenea dentata Marsh. (American chestnut). Cryphonectria

parasitica was introduced in New York City in the early 1900s and spread rapidly down

the Appalachian Mountains, infecting C. dentata throughout most of its range by 1926

and effectively extirpating the species by the 1950s (Anagonostakis 2001). Whereas the

effects of C. parasitica on C. dentata were fairly uniform in all forest types and stand









conditions, disease severity of dogwood anthracnose is influenced by many variables,

particularly shading and moisture (Britton 1993). The disease is more severe on shaded

and moist northeast-facing slopes than on relatively drier southwest facing slopes with

open canopies (Chellemi and Britton 1992, Chellemi et al. 1992). Britton (1993) reported

that given adequate amounts of rainfall, the disease could develop throughout the

growing season. In addition, Britton et al. (1996) found that acid rain increases the

susceptibility of C. florida to anthracnose.

Since dogwood anthracnose is known to be sensitive to environmental

characteristics, there is a possibility that management techniques could be used to

manipulate stand structure to reduce the impacts of dogwood anthracnose. However,

research examining the effects of stand manipulations on dogwood anthracnose in forest

stands has been limited. Britton et al. (1994) examined the effect of past silvicultural

practices on C. florida density and disease severity and reported C. florida densities were

highest and disease severity lowest in stands that had been clearcut in 1939 and again in

1962. This effect was attributed to increased light levels in these plots. Prescribed

burning may also offer a technique to manipulate stand conditions to favor C. florida

survival in infected stands. Jenkins and White (2002) reported a 200% increase in C.

florida stem density on three long-term vegetation plots that burned in 1976 in Great

Smoky Mountains National Park (GSMNP), however, the effects of burning on stand

structure and C. florida health and survival were not fully explored.

Dogwood anthracnose was first reported in GSMNP in March of 1988 when a

park-wide survey revealed that 23 out of 58 plots that contained C. florida were infected

with anthracnose (Windham and Montgomery 1990). Seven years later, in a study









conducted in the western half of GSMNP, Wilds (1997) observed signs of anthracnose on

98% of the plots on which C. florida occurred. By 2000, C. florida mortality attributed to

dogwood anthracnose ranged from 69% in oak-pine stands to 94% in acid coves on long-

term vegetation plots located within GSMNP (Jenkins and White 2002). Prior to

anthracnose, C. florida was the dominant understory woody species in oak-hickory forest

of GSMNP. Following anthracnose infection, C. florida density decreased by 80% in this

forest type.

Within eastern North America, oak-hickory forests comprise over 34% of total

forest cover (Smith et al. 2001). Within this forest type, fire was historically common and

influenced species composition and stand structure (Brose et al. 2001). Because fire has

played an important ecological role in this widely distributed forest type, we undertook a

study with the following objectives: to determine the influence of past burning on (1) C.

florida density and health and (2) overall stand structure and species composition in oak-

hickory forests in GSMNP. We hypothesize that past burning has altered stand conditions

(structure and species composition) in ways that reduced the impacts of dogwood

anthracnose compared to unburned stands. Burning typically reduces stand density,

increases light penetration through the canopy, and decreases understory moisture

content, which, we hypothesize, favors C. florida survival. These conditions will not last

indefinitely, however, and we further hypothesize that repeated burning is needed to

maintain stand conditions that reduce the impacts of the disease. Because C. florida is a

thin barked species that is frequently top-killed by fire, we further hypothesize that C.

florida will display reduced survival once a threshold of burning frequency is reached.









Materials and Methods

Study Site

This study was conducted in GSMNP, USA, which encompasses slightly over 200

000 ha, and straddles the Tennessee and North Carolina state line. The Park is

internationally renowned as a center of biological diversity within North America and

was designated as an International Biosphere Reserve in 1976. Due to its biotic diversity,

large size, and protected status, changes that occur within the biological communities of

GSMNP often serve as baselines for comparison to other state and federal lands. Mean

annual temperature in Gatlinburg, Tennessee (440 m a.s.l. and adjacent to GSMNP) is

12.90 C and mean annual precipitation is 142 cm. Our study sites ranged in elevation

from 287 to 975 m. Although C. florida occurs in a variety of forest types, we focused

our study in oak-hickory forests. These forests were typically found on moderately steep

to steep slopes with southeast-south-northwest facing aspects. Cornusflorida is a

common understory species in the oak-hickory forest type, and this type has frequently

burned in some areas of GSMNP over the last 30 years.

During June-August of 2001-2004, we sampled seventy-nine 0.04 ha (20 m x 20 m)

plots in burned and unbumed stands. Fifty-five plots were established in fourteen stands

that had burned up to three times over a 20 year period (late 1960s to the late 1980s). In

addition, twenty-four plots were established in six unbumed stands. These areas were

divided into four sampling categories: (1) single burn (seven stands, thirty plots), (2)

double bum (four stands, sixteen plots), (3) triple burn (three stands, nine plots), and (4)

unburned (six stands, twenty-four plots).

We used historic park maps and fire history records to select burned (single,

double, and triple bums) and unbumed areas. Within each burned area, we selected









stands from vegetation associations within the Montane Oak-Hickory Forest Ecological

Group (White et al. 2003). Associations within this group contained C. florida as a major

understory component prior to the onset of dogwood anthracnose (White et al. 2003).

Individual polygons (stands) of each association were derived from vegetation maps of

GSMNP based upon 1:12000 color-IR aerial photos (Welch et al. 2002). Unburned

(reference) plots were established in nearby areas with similar slopes, aspects,

topography, and vegetation associations as the bum plots. All burns selected were at least

10 hectares in size. Plots were located within the burn areas by placing a 50 m buffer

inside of each area, and randomly selecting plot coordinates within appropriate vegetation

associations. A minimum of three 20 x 20 m plots were placed within each burn area,

with up to five plots placed in larger bums.

Field Sampling

We recorded the diameter at breast height (dbh) of all living overstory stems

(> 10.0 cm dbh) by species to the nearest 0.1 cm. Living stems < 10.0 cm dbh

(understory) were tallied by species into four diameter classes: 0-1.0 cm, 1.1-2.5 cm, 2.6-

5.0 cm, and 5.1-10.0 cm. We measured the dbh of all C. florida stems, regardless of

overstory or understory classification, to the nearest 0.1 cm. Foliage and crown health

were assessed for each living C. florida stem using the Mielke-Langdon Index (Mielke

and Langdon 1986; Table 2-1).

Data Analysis

Data were analyzed for differences in the four sampling categories (single burn,

double bum, triple burn, and unbumed) for the following response variables: C. florida

stem density, C. florida foliar and crown health, overstory basal area and stem density,

understory basal area and stem density, understory species importance values [IV =









((relative density + relative basal area)/2)*100], Tsuga canadensis (L.) Carr. stem

density, plot species richness, and species diversity (Shannon's diversity index). Cornus

florida stem density differences among the four sampling categories were analyzed in

three diameter classes: 0-5.0 cm, 5.1-10.0 cm, and > 10.0 cm. Total C. florida stem

density of the four sampling categories was also analyzed. Because smaller trees are more

susceptible to dogwood anthracnose (Hiers and Evans 1997, Jenkins and White 2002), C.

florida foliage and crown health were analyzed in five diameter size classes that better

represented smaller diameters: 0-1.0 cm, 1.1-2.5 cm, 2.6-5.0 cm, 5.1-10.0 cm, and > 10.0

cm. Tsuga canadensis stem density was analyzed in five diameter size classes (0-2.5 cm,

2.6-5.0 cm, 5.1-10.0 cm, 10.1-20.0 cm, and > 20.0 cm), plus total stem density.

Before statistical comparison, all data (except categorical) were tested for normality

using the Kolmogorov-Smirnov goodness-of-fit test for normal distribution. Only the C.

florida stem density data violated the test for normality (P < 0.05). These data were

natural log transformed to improve normality and equalize variances. For ease of

interpretation, non-transformed values are presented. All response variables were

analyzed with analysis of variance (ANOVA) using the Mixed procedure in SAS (SAS

2002). The model was made up of two factors. The first factor was fixed, sampling

category, and the other was random, bum area nested within sampling category. When

ANOVA revealed a clear difference between the sampling categories, we used the PDIFF

option for post-hoc pairwise comparisons (SAS 2002). All means presented in the paper

are least square means calculated by SAS using the mixed procedure.

To test for differences in overstory and understory community composition in the

four sampling categories we used MRPP (Multi-Response Permutation Procedures;









Biondini et al. 1985, Lesica et al. 1991, Peterson and McCune 2001). MRPP is a

nonparametric procedure to test for multivariate differences in pre-existing groups (i.e.

single burn, double burn, triple burn, and unburned stands) (Mielke 1984). It provides a

test statistic (A) and P-value to determine whether the sampling categories occupied the

same regions of species space (A measures within-group agreement, if A = 1 then items

within groups are identical and 1 is the highest possible value for A, A = 0 when

heterogeneity within groups equals expectation by chance, and A < 0 with less

heterogeneity within groups than expected by chance). Interpretation of this test statistic

was done using nonmetric multidimensional scaling (NMS) and indicator species analysis

(IndVal) (Peterson and McCune 2001, McCune and Grace 2002). NMS is an ordination

technique that is ideal for data that are nonnormal or nonlinear and contains large

numbers of zero values. It uses ranked distances to linearize the relation of degrees of

difference between community samples and distances on an environmental gradient, and

is the most effective ordination technique available for community data (Clarke 1993,

McCune and Grace 2002). IndVal is used to describe the relationship of species to

categorical variables by combining species abundance in a specific category plus the

faithfulness of occurrence of that species in that specific category (Dufrene and Legendre

1997, Qian et al. 1999, Peterson and McCune 2001). The analysis produces a value of

abundance for each species in each group (IndVal) and a test statistic produced from

Monte Carlo tests (1000 iterations) to determine if occurrence in the maximum

(indicator) group is greater than would be expected from chance.

All multivariate analyses were performed using PC-ORD (McCune and Medford

1999). To reduce the effects of rare species, we deleted those species occurring in less









than three plots prior to multivariate analyses; eight species were deleted from overstory

analyses and 16 species were deleted from understory analyses (McCune and Medford

1999). MRPP and NMS were performed using the quantitative version of Sorenson's

distance measure and NMS ordination was displayed on two axes. Additional axes did

not significantly improve the explanatory power of the ordination.

Results

Cornusflorida Stem Density

Cornusflorida stem density differed significantly among sampling categories in the

smallest (0-5.0 cm) size class (double burn = 691 stems ha-1, triple burn = 175 stems ha-1,

single burn = 154 stems ha-1, and unburned stands = 35 stems ha-1, P = 0.0015; Figure 2-

1). Double burn stands contained four times more C. florida stems than single burn stands

(P = 0.05) and twenty times more C. florida stems than unburned stands (P = 0.0002).

The stem density of double bum stands was not significantly different from that of triple

burn stands (P = 0.39). The single burn stand and triple bum stand stem densities in this

size class were not significantly different (P = 0.39), but stem densities in both of these

categories were significantly greater than the unburned stand stem density (P < 0.007).

There was no statistical difference between sampling categories in C. florida stem density

in the 5.1-10 cm size class (P = 0.229; Figure 2-1). Although differences were not

significant due to high inter-plot variability (P = 0.167), double and triple burn stands

contained greater densities of stems >10 cm dbh (22 and 17 stems ha-1, respectively) than

unburned and single burned stands (5 and 6 stems ha-1, respectively).

Total stem density of C. florida differed significantly among the four sampling

categories (double bum = 770 stems ha-l, triple burn = 233 stems ha-l, single bum = 225

stems ha-1, and unburned stands = 70 stems ha-l, P = 0.0003; Figure 2-1). This difference









can be largely attributed to the greater percentage of smaller trees in burned stands. Trees

in the 0-5.0 cm size class contributed 90% of the total stems ha'1 in double burn stands,

75% in triple bum stands, 68% in single burn stands, and just 50% in unbumed stands.

We observed a significantly greater total density of stems in double burn stands than in

single burn stands (P = 0.036) and unbumed stands (P < 0.0001), but triple bum stands

did not differ significantly from double burn stands (P = 0.53). Single burn and triple

burn stands were not significantly different from each other (P = 0.26), but both

categories were significantly greater than the unbumed stands (P < 0.001).

Foliage Health and Crown Dieback

Overall, mean foliage health ranged from 3.1 to 3.9 for all size classes and

sampling categories, and we did not observe any differences in the foliar health among

the sampling categories in each of the five size classes (P > 0.38; Table 2-2). Mean crown

dieback ratings of all size classes ranged from 2.4 to 3.7, and there were no significant

differences in crown health in three of the size classes (0-1.0 cm, P = 0.64; 5.1-10.0 cm, P

= 0.45; and > 10.0 cm, P = 0.93; Table 2-2). However, in two of the smaller size classes

(1.1-2.5 cm and 2.6-5.0 cm) there was a significant difference in mean crown dieback

ratings among the four sampling categories (P = 0.04 and P = 0.01, respectively; Table 2-

2). Further analyses in the 1.1-2.5 cm size class showed that the rating of unburned stands

(2.4) was significantly lower (less healthy) than those of the burned stands (3.3 3.4, P <

0.02). In the 2.6-5.0 cm size class, mean crown dieback ratings for burned stands (3.6 -

3.2) were significantly higher (healthier) than that of unburned stands (2.7, P < 0.04).

Differences among the sampling categories were not significant for the 0 0.1 cm class,

however, plots in unburned stands did not contain any living trees in this size class.









Stand Structure

Overstory (> 10.0 cm dbh) basal area was similar among the four sampling

categories (single bum = 21.7 m2 ha-1, double burn = 22.1 m2 ha-1, triple bum = 20.6 m2

ha-1, and unbumed stands = 23.2 m2 ha-1, P = 0.52; Table 2-3). Overstory stem density

differed significantly among the four sampling categories (single burn 436 stems ha-1,

double burn 323 stems ha-1, triple burn 317 stems ha-1, and unbumed stands 564 stems

ha-1, P < 0.0001; Table 2-3). Comparisons of overstory stem density among the four

sampling categories revealed that double and triple burn stands had similar stem densities

(P = 0.92). Densities in these two categories were significantly lower than that of single

burn stands (P < 0.05) and unbumed stands (P < 0.001). Finally, single burn stands had

significantly fewer stems than unburned stands (P = 0.016).

In the understory (< 10.0 cm dbh), basal area was similar between sampling

categories, ranging from 6.5 7.4 m2 ha-1 (P = 0.93; Table 2-3), while stem density was

significantly different (single bum 2851 stems ha-1, double burn 4594 stems ha-1, triple

burn 5072 stems ha-1, and unbumed stands 2292 stems ha-l, P=0.024; Table 2-3).

Comparisons of understory stem density among the four sampling categories revealed

that double and triple bum stands had similar densities (P = 0.66), as did single burn

stands and unburned stands (P = 0.46). Double and triple burn stands had significantly

greater understory stem densities than the single burn and unbumed stands (P < 0.05).

Overstory Community Composition

Shannon's diversity index did not differ significantly in the overstory in the four

sampling categories (P = 0.48; Table 2-4). However, species richness of plots in the four

sampling categories did differ slightly (P = 0.06), and was greatest in unburned stands

compared to burned stands (6.9 versus < 5.7, respectively; Table 2-4). MRPP indicated









that species composition differed slightly between plots in the sampling categories

(MRPP: P < 0.0001, A = 0.09). Three species were associated with unbumed stands (Acer

rubrum L., Oxydendrum arboreum (L.) DC., and T canadensis) and one each for double

burn stands (Quercus alba L.) and triple bum stands (Quercus velutina Lam.) (IndVal: P

< 0.08 each; Table 2-5). These differences, however, were not strong enough to clearly

separate the sampling categories in the ordination (Figure 2-2).

Understory Community Composition

Shannon's diversity index and species richness did not differ in the understory

among the four sampling categories (P = 0.81 and P = 0.48, respectively; Table 2-4).

Species composition did differ slightly between the sampling categories (MRPP: P <

0.0001, A = 0.06). Numerous species were indicative of sampling categories, primarily in

the triple bum stands (unbumed: T. canadensis; double burn: C. florida and Tilia

americana L.; triple bum: Carya alba L., Carya glabra Mill., Pinus virginiana Mill., Q.

alba, Quercus prinus L., Quercus rubra L., Q. velutina, and Robiniapseudoacacia L.;

Table 2-5). These differences, however, were not strong enough to clearly separate the

sampling categories in the ordination (Figure 2-2).

Importance Values for Understory Species

In addition to greater stem densities, mean importance value (IV) of C. florida was

four times greater in double burn stands than in unbumed stands (IV = 21.6 versus 5.1, P

= 0.001; Table 2-6). Cornusflorida importance values were also significantly greater in

single burn (IV = 12.1) and triple burn (IV = 14.8) stands compared to unburned stands

(IV = 5.1, P = 0.05). Cornusflorida had the greatest importance value of any species in

double bum stands and second highest in the triple and single burn stands. Six other

species had higher importance values than C. florida in unbumed stnads, including T.









canadensis, which was ten times greater in importance in unburned stands compared to

burned stands (P < 0.05; Table 2-6).

In triple burn stands, the importance values for A. rubrum (IV = 7.8) was

significantly lower (P = 0.03) than that of single burn stands (IV = 24.2). Also within

triple burn stands, C. glabra and C. alba L. importance values were three times greater

than in any other category (P < 0.004 and P < 0.01, respectively). In addition, the R.

pseudoacacia importance value was three times greater in triple burn stands compared to

the other sampling categories (P < 0.04) and the importance value of P. virginiana was

greatest in the triple burn stands.

Tsuga canadensis Stem Density

Overall, total T canadensis stem density was significantly greater in unburned

stands (216 stems ha-1) compared to single burn (42 stems ha-1), double burn (23 stems

ha-1), and triple bum stands (11 stems ha-1, P < 0.001; Figure 2-3). Most of the T

canadensis stems were in the smallest (0-2.5 cm) size class, T canadensis stem density

was over four times greater in unburned stands (83 stems ha-1) compared to single (18

stems ha-1), double (14 stems ha-1), and triple bum stands (11 stems ha-1, P < 0.001;

Figure 2-3). Unburned stands had significantly more T canadensis stems ha-1 in the 2.5-

5.0 cm, 5.1-10.0 cm, and 10.1-20.0 cm size classes as well (P < 0.005). In the largest

diameter classes (> 20 cm), single burn stands and unburned stands were similar (4 and 6

stems ha-1, respectively, P = 0.49). This size class was absent on plots in double and triple

burn stands.

Discussion

The results of our study demonstrate the potential role of fire in regulating

population dynamics of C. florida in post-anthracnose stands. Overall, burned stands in









our study had greater C. florida stem densities, C. florida trees with healthier crowns, and

higher C. florida importance values than unbumed stands. The greater C. florida stem

densities and healthier C. florida trees in burned stands are likely the result of reduced

shading following the bums that created a relatively drier microclimate that was less

favorable to D. destructive. Studies have shown that shaded conditions increase the

severity of dogwood anthracnose (Gould and Peterson 1994, Erbaugh et al. 1995). For

example, Chellemi and Britton (1992) reported an inverse relationship between

evaporative potential and disease severity on C. florida in the southern Appalachians.

Discula destructive growth has been found to be greater under moist conditions. In a

study involving artificial inoculation of C. florida with dogwood anthracnose, Ament et

al. (1998) reported that D. destructive lesions on C. florida leaves were five times larger

when placed inside moistened bags for seven days compared to lesions that developed on

leaves that spent four, two, and zero days inside moistened bags.

In our study, the greatest densities of C. florida stems were found in double burn

stands. Although single bum stands had greater C. florida stem densities than unburned

stands, it appears that a single burn within a 20 year period is not sufficient to maintain C.

florida. Increases in overstory stem densities following a single burn appear to provide

sufficient shading for anthracnose to increase in virulence. Studies have shown repeated

burns better maintain lower overstory stem densities in oak forests (Huddle and Pallardy

1996, Peterson and Reich 2001, Hutchinson et al. 2005), which favors C. florida survival

from dogwood anthracnose. Our results indicate, however, that benefits of multiple burns

are reduced when burning is increased to three burns in a 20 year period. It appears that

although larger (> 5 cm) C. florida trees survived the third bum, smaller trees displayed









less resprouting after the third burn. This reduction in resprouting may be attributed to

fire induced mortality of smaller stems, as opposed to lack of root carbohydrates,

resulting in fewer stems per hectare capable of producing sprouts (Arthur et al. 1998),

and suggests that this was too short an interval between burns. Increased importance

values of R. pseudoacacia and P. virginiana on triple burn stands, (indicator species of

triple burn stands as well), suggest that the third burn shifted stands towards an earlier

successional composition (Harrod et al. 1998). This shift to an earlier successional

composition may perhaps be another reason for decreased C. florida stem densities in

triple burn stands.

Although there were some differences in overstory community composition among

the sampling categories, these differences were not strong enough to classify the

sampling categories as unique communities. However, the identification ofA. rubrum, 0.

arboreum, and T. canadensis as indicator species in unburned stands suggest that this

sampling category is shifting to a later successional stage. This is not surprising,

however, considering the lack of disturbance in these stands for the past 80-100 years.

In the understory, we found that differences in overall community composition

were present, but limited to a few species, such as C. florida and T canadensis. We

observed higher understory stem densities in multiple burn stands, which is a result of

reprouting by deciduous trees and is consistent with other studies (McGee et al. 1995,

Elliott et al. 1999, Kuddes-Fischer and Arthur 2002). However, while burned stands had

greater total understory stem densities, fire decreased the density and importance of T.

canadensis in the understory of burned stands. In fire suppressed stands, this species

often dominates the understory and produces a dense sub-canopy that results in moist,









heavily shaded conditions (Godman and Lancaster 1990, Woods 2000, Jenkins and White

2002, Galbraith and Martin 2005) that favor dogwood anthracnose development

(Chellemi and Britton 1992). Therefore, the reduction in T canadensis density following

fire has likely contributed to the greater densities of C. florida in burned areas.

In our study, the positive effects on C. florida density observed in burned stands

were likely a result of the indirect effects of fire, produced by changes in stand structure

and composition. The direct effects of fire (smoke and heat) on D. destructive are

unknown. However, studies suggest that fire may reduce the amount of inoculum of

fungal diseases (Parmeter and Uhrenholdt 1975, Schwartz et al. 1991). In laboratory

experiments, Parmeter and Uhrenholdt (1975) found that exposure to smoke from

burning pine needles reduced the amount of rust and gall fungi on cellophane sheets.

Schwartz et al. (1991) suggested that smoke from upland fires may have settled into

unburned mesic ravines and helped reduce mortality of the endangered Torreya taxifolia

Arnott (Florida torreya) by reducing fungal disease. In addition, burning typically

produces a flush of nutrients in the soil after a burn (Kutiel and Shaviv 1992, Boerner et

al. 2004, Tuininga and Dighton 2004). This flush in nutrients, especially cations such as

calcium, magnesium, and potassium, may benefit C. florida survival. Studies have

documented the importance of nutrients in plant survival from diseases (Sij et al. 1985,

Yamazaki et al. 1999, Sugimoto et al. 2005). Holzmueller et al. (2006b) reported that

cation availability played a role in C. florida survival and resistance to dogwood

anthracnose.

We observed prolific sprouting by C. florida in stands that burned prior to

anthracnose infection in our study. However, the use of fire as a management tool for









anthracnose is dependent upon the ability of diseased and weakened trees to resprout

after fire in anthranconse-infected stands. Encouragingly, Blankenship and Arthur (2006)

reported high levels of C. florida sprouts in recently burned areas that had been

previously infected with dogwood anthracnose. However, the long-term survival rate of

these sprouts in anthracnose-infected stands is unknown. In addition, in the burned stands

we sampled, most living C. florida stems were relatively small in diameter. The amount

of fruit produced by these populations of smaller individuals relative to pre-anthracnose

production is unknown. The amount of fruit and seeds produced is critical to the

successful reproduction of C. florida and its role as a source of soft mast for wildlife.

Prescribed fire may offer the best means of control of dogwood anthracnose in oak-

hickory forest stands. Although Britton et al. (1994) documented the highest stem

densities of C. florida in clearcut areas in a study of timber harvest practices on C. florida

populations, it is unlikely that clearcutting large areas for the benefit of a single

understory species would fit into many ecosystem management plans. Furthermore, in

clearcut stands, the overstory will likely redevelop during the stem exclusion stage within

20 years after a harvest (Oliver and Larson 1990), and shading will again increase. The

return interval for clearcutting (60-100 years) will likely be too long to serve as an

effective long-term control. Prescribed burning in oak forests may offer a more

applicable management technique across large forested areas and multiple ownerships,

particularly those where mechanical harvests are not an option. Although eastern oak

forests have been subjected to fire suppression for about 100 years, resource managers

have increased efforts to manage oak forests with prescribed fire (Brose et al. 2001), and

these efforts offer a framework for managing C. florida populations as well. Burning on a









10-15 year return interval would probably be best for C. florida survival, and would fit in

with the historic fire regime of eastern oak-hickory forests (Harmon 1982). In single,

double, and triple burned stands in our study, we observed less crown dieback compared

to unburned stands, however, the lack of difference in foliar infection may indicate that

the interval since the last bum (nearly 20 years) has allowed anthracnose to return to a

level of infection similar to unburned stands. This suggests that these stands will require

additional bums to slow the loss of C. florida.

Management Implications

Our results suggest that prescribed fire may offer an effective and practical

management technique to alleviate the symptoms of anthracnose in oak-hickory and,

perhaps, oak-pine forests. Other forest types, such as cove and alluvial forests, where C.

florida was once a frequent component (Jenkins and White 2002) burned infrequently, if

at all, in the past. Therefore these forest types are unlikely to sustain fire frequencies of

10-15 years; the frequency that appears to best reduce the effects of anthracnose.

Although this study was conducted in GSMNP, we believe that its methods and results

are applicable across the eastern United States in forest types that contain C. florida and

have regimes of relatively frequent fire.

It is unlikely, however, that burning could be returned to all oak-hickory and oak-

pine forests due to many external factors such as time and budget constraints and

management objectives. Consequently, certain areas may be deemed higher in priority for

burning. These areas include currently uninfected stands that are in close proximity to

infected areas, stands only recently showing signs of infection, or infected stands with

large C. florida populations. Land managers would be more successful using prescribed









burning in maintaining C. florida populations than attempting to reintroduce C. florida to

areas where it once occurred.

Conclusion

The results of our study suggest that burning has reduced the impacts of

anthracnose on C. florida populations. Burned stands, especially double bum stands, had

significantly greater C. florida stem densities than unbumed stands. In addition, the

density of T. canadensis, a species that creates stand conditions favorable for dogwood

anthracnose, was greatly reduced in burned stands. Past burning did not drastically affect

overall overstory or understory species composition, but the importance values of

selected overstory and understory species were highly indicative of specific bum

frequencies. The results indicate that prescribed burning may offer an effective and

practical technique to control the impacts of dogwood anthracnose and prevent the loss of

C. florida from oak-hickory forests.






30


Table 2-1. Scales for foliage health (% of foliage with signs of anthracnose) and crown
dieback (% of crown dieback) used to assess the level of disease severity of
dogwood anthracnose on Cornusflorida trees (based on the Mielke-Langdon
Index, Mielke and Langdon 1986).


Rating Foliage health Crown dieback
1 >76 >76
2 51-75 51-75
3 26-50 26-50
4 1-25 1-25
5 none none









Table 2-2. Mean foliage and crown health (+ 1 SE) for Cornusflorida for five diameter
classes in the four different sampling categories using the Mielke-Langdon
Index (Mielke and Langdon 1986).

Diameter
las (m) Single bum Double burn Triple bum Unburned P-value1
class (cm)
Foliage
Health
0-1.0 3.6(0.2) 3.5 (0.2) 3.1(0.3) A3 0.38
1.1-2.5 3.6(0.2) 3.6(0.2) 3.6(0.2) 3.7(0.3) 0.94
2.6-5.0 3.6 (0.2) 3.8 (0.2) 3.6 (0.3) 3.7 (0.3) 0.88
5.1-10.0 3.8(0.1) 3.7(0.1) 3.9(0.2) 3.8(0.3) 0.90
>10.1 3.6 (0.3) 3.6 (0.2) 3.7 (0.2) 3.6 (0.3) 0.99
Crown
Dieback
0- 1.0 3.2 (0.3) 3.5 (0.3) 3.0 (0.5) A 0.63
1.1-2.5 3.3 (0.2)a2 3.4 (0.2)a 3.3 (0.3)a 2.4 (0.3)b 0.04
2.6-5.0 3.3 (0.1)a 3.6 (0.2)a 3.2 (0.2)a 2.7 (0.3)b 0.01
5.1- 10.0 3.3 (0.1) 3.4 (0.2) 3.7 (0.2) 3.3 (0.3) 0.45
>10.1 3.5 (0.4) 3.7 (0.4) 3.4 (0.5) 3.2 (0.3) 0.93
1 P-value from ANOVA
2 Means followed by the same letter in the same row are not statistically different (P <
0.05) using post-hoc pairwise comparisons among sampling categories when ANOVA
P-value < 0.05
3A = Absent, no trees from this sampling category were found in this size class









Table 2-3. Mean understory and overstory basal area and stem density (+ 1 SE) in the
four sampling categories.


Understory
Single burn
Double bum
Triple bum
Unburned
P-value

Overstory
Single burn
Double bum
Triple bum
Unburned
P-value


Basal Area
(m2 ha-1)
7.1 (0.5)
6.9 (0.7)
6.5 (1.0)
7.4 (0.4)
0.931


21.7(0.8)
22.1 (1.1)
20.6(1.8)
23.2(0.8)
0.52


Stem density
(stems ha'1)
2851 (360)
4594 (726)a
5072 (1090)a
2292 (274)b
0.024


436 (31)
323 (46)
317 (43)c
564 (25) a
<0.0001


1 P-value from ANOVA
2 Means followed by the same letter in the same column for understory and overstory are
not statistically different (P < 0.05) using post-hoc pairwise comparisons among
sampling categories when ANOVA P-value < 0.05











Table 2-4. Mean species richness and Shannon's diversity index (+ 1 SE) for the
understory and overstory in the four sampling categories.


Understory
Single burn
Double bum
Triple bum
Unburned
P-value


Species
richness
11.4(0.9)
11.0 (1.3)
13.5 (1.5)
10.7(1.0)
0.481


Overstory
Single burn 5.7 (0.4)
Double bum 5.4(0.5)
Triple bum 5.2(0.7)
Unburned 6.9 (0.4)
P-value 0.06
P-value from ANOVA


Shannon's
diversity index
1.9 (0.08)
1.9(0.11)
2.0 (0.12)
1.9 (0.09)
0.81


1.3 (0.09)
1.2 (0.12)
1.2(0.14)
1.4(0.09)
0.48









Table 2-5. Overstory and understory indicator values (IndVal) (percent of perfect
indication) and associated sampling category. P-value represents the
proportion of randomized trials that the indicator value was equal to or
exceeded the observed indicator value.

Species IndVal P-value Indicator group
Overstory
Acer rubrum 32.2 0.080 Unburned
Oxydendrum arboreum 29.8 0.073 Unburned
Quercus alba 47.4 0.001 Double bum
Quercus velutina 29.7 0.026 Triple burn
Tsuga canadensis 36.0 0.006 Unburned

Understory
Carya alba 46.0 0.001 Triple burn
Caryaglabra 42.6 0.012 Triple burn
Cornusflorida 40.2 0.015 Double bum
Pinus virginiana 31.5 0.018 Triple burn
Quercus alba 18.9 0.065 Triple burn
Quercus prinus 37.8 0.033 Triple burn
Quercus rubra 42.5 0.002 Triple burn
Quercus velutina 22.3 0.076 Triple burn
Robiniapseudoacacia 42.9 0.005 Triple burn
Tilia americana 18.7 0.023 Double bum
Tsuga canadensis 46.9 0.007 Unburned









Table 2-6. Mean importance values (+ 1 SE) of selected understory species in the four
sampling categories.


Species
Acer pensylvanicum
Acer rubrum
Carya alba
Carya glabra
Cornusflorida
Halesia tetraptera
Kalmia latifolia
Liriodendron tulipifera
Nyssa sylvatica
Oxydendrum arboreum
Pinus strobus
Pinus virginiana
Quercus prinus
Quercus rubra
Quercus velutina
Rhododendron maximum
Robinia pseudoacacia
Sassafras albidum
Tsuga canadensis


Single bum
9.6(5.5)
24.2 (3.3)a
0.9 (0.9)a
1.6 (1.0)a
12.1 (2.3)a
3.2(1.7)
5.3 (3.6)
4.9 (1.6)a
6.4 (1.8)ab
6.3 (1.7)
3.1 (2.0)
0.2 (0.7)a
3.8 (1.0)
0.7 (0.3) a
0.5 (0.7)a
4.3 (2.3)
1.3 (0.9)a
1.5 (0.7)
0.9 (2.6)a


Double bum
3.5 (7.3)
17.8 (4.4)ab
2.0 (1.8)a
1.8 (2.5)a
21.6 (3.2)b
3.4(2.4)
2.8 (4.8)
2.4 (2.2)ab
9.6 (2.3)
7.3 (2.3)
1.6(2.6)
0.8 (0.9)ab
1.3 (1.4)
0.8 (0.4)ab
0.6 (0.9)a
0.3 (3.1)
2.0 (1.2)a
1.1 (0.9)
0.5 (3.5)a


Triple bum
A
7.8 (1.8)b
7.0 (1.5)b
10.3 (2.0)b
14.8 (4.2)ab
A
5.4 (6.0)
0.3 (2.6)ab
0.9 (3.1)a
6.4 (2.9)
A
2.5 (1.1)b
3.0(2.2)
2.0 (0.6)b
3.1 (1.0)b
2.0(3.8)
6.4 (1.6)b
2.9(1.2)
0.3 (4.2)a


Unbumed
9.8 (6.0)
19.2 (3.6)ab
2.1 (1.0)a
2.5 (1.2)a
5.1 (2.6)c
1.8 (1.9)
6.6 (4.0)
0.4 (1.8)b
6.1 (1.9)ab
5.2(1.9)
4.9 (2.2)
1.1 (0.8)ab
2.4(1.2)
0.1 (0.4)a
0.4 (0.8)a
4.1 (2.5)
0.2 (1.0)a
1.3 (0.8)
9.9 (2.8)b


1Means followed by the same letter in the same row are not statistically different (P <
0.1) using post-hoc pairwise comparisons among sampling categories when ANOVA P-
value < 0.05
2A = Absent, no trees from this sampling category were found in this size class











0-5 cm class
Sa P-value = 0.0015






Single burn Double burn Triple burn Unburned


80

60

40

20

0
Single burn


5.1-10 cm class
P-value = 0.229


Double burn Triple burn


30 900
>10.1 cm class
P-value = 0.167
20 -600




0 0
Single burn Double burn Triple burn Unburned Single burn


I .


Double burn Triple burn


Total stems
P-value = 0.0003


Unburned


Sampling category

Figure 2-1. Mean Cornusflorida stem density (+ 1 SE) in the four sampling categories
for three diameter classes and total stems ha-1 for all diameter classes. Within
each graph, P-value is from ANOVA, and bars with same letters are not
significantly different from each other (P < 0.05) using post-hoc pairwise
comparisons among sampling categories when ANOVA P-value < 0.05; note
the scale of the y-axis for each graph.


Unburned









37






Overstory

Burn
Unburned
A Single
15 Double
E Triple l











SA xis






















Burn
A T
















AA E
A E





O V
-15 6
-I







-20 -1 0 0 0 0 20

Axis 1






















Undersunderstory communities, showing the relative differences in community
Burn



















SUnburdistinct compositional changes in overstory or understory with respect to burn



















frequency.
A AA SingleA




VA
20 05 10 00 10'





X Unburned

15 A Single A A
STriple T






T @
L&





a
A







D T
1i 05 A AA ba A








~ yA Y




















frequency.















* Unburned
OSingle burn
E Double burn
O Triple burn


300


250


200


150


100


50


a a


a


Lbb b


0-2.5 2.6-5.0 5.1-10.0
Diameter class (cm)


10.1-20.0 >20.0


Figure 2-3. Mean Tsuga canadensis stem density (+ 1 SE) in five diameter size classes
and total (all diameter classes combined) in the four sampling categories.
Same letters in each diameter class represent no significant differences in
mean stems ha1 (P < 0.05) using post-hoc pairwise comparisons among
sampling categories when ANOVA P-value < 0.05.


bb a

b
b t bb b b a


U ca

E
c '
c u,
^3 V


cI,~
2, S


Total














CHAPTER 3
INFLUENCE OF CALCIUM, POTASSIUM, AND MAGNESIUM ON Cornusflorida
L. DENSITY AND RESISTANCE TO DOGWOOD ANTHRACNOSE

Introduction

Dogwood anthracnose, caused by the fungus Discula destructive Redlin, is a major

disease of Cornusflorida L. in forests of the eastern United States. Where dogwood

anthracnose is present, mortality rates of C. florida have been up to 90% (Sherald et al.

1996, Hiers and Evans 1997). The fungus causes leaf necrotic blotches, leaf blight, twig

dieback, and stem cankers, which eventually lead to tree death (Hibben and Daughtery

1988). Disease severity and rate of infection, however, vary with several environmental

factors. The disease is most severe in cool, wet areas of high elevation with shaded slopes

(Chellemi et al. 1992). Within individual stands, disease severity increases with

decreased light availability, increased moisture, and decreased evaporative potential of

the leaf surface (Chellemi and Britton 1992).

While the link between mineral nutrition and resistance to dogwood anthracnose

has not been examined in detail, soil availability of cations may reduce the impacts of the

disease and increase the survival of C. florida trees. Britton et al. (1996) found that

disease severity did not increase with application of pH 2.5 simulated rain to the foliage,

but did increase with application of pH 2.5 simulated rain to the growing medium.

Consequently, the increase in the severity of infection was attributed to nutrient

bioavailability, not foliar damage. However, since soil nutritional changes were not

quantified, this hypothesis still remains unproven. In a review of calcium (Ca) physiology









and terrestrial ecosystem processes, McLaughlin and Wimmer (1999) hypothesized that

low Ca saturation decreases the natural resistance of C. florida to dogwood anthracnose.

Anderson et al. (1991) reported that liming reduced disease severity of anthracnose on

ornamental C. florida trees.

While the link between soil cation availability and anthracnose has not been

examined, calcium applications have been shown to reduce the impacts of fungal disease

on soybeans (Glycine max (L.) Merr.) and bacterial wilt on tomato (Lycopersicon

esculentum L.) seedlings (Muchovej et al. 1980, Sugimoto et al. 2005, Yamazaki et al.

1999). Other cations, particularly potassium (K) and magnesium (Mg), have also been

linked to disease resistance in plants. Sij et al. (1985) reported that increased rates of K

fertilizer significantly decreased the impacts of a fungal disease on field grown soybean

plants. Jeffers et al. (1982) reported lower numbers of seeds infected with seed mold on

tomato plants fertilized with K. Likewise, the severity of fungal shoot blight was greater

on Picea abies (Norway spruce) trees whose needles contained low levels of Mg

(Anglberger and Halmschlager 2003).

Calcium, K, and Mg play essential roles in plant growth and development (Epstein

1972, Mengel et al. 2001). Calcium strengthens plant cell walls, which may help in

disease resistance (Muchovej et al. 1980, Conway et al. 1992, Sugimoto 2005).

Potassium and Mg are essential for many plant metabolic functions (Epstein 1972,

Mengel et al. 2001). Disease resistance with optimal K and Mg nutrition may be

attributed to increased energy used to offset the impact of plant diseases (Mengel et al.

2001). In addition, K may also increase disease resistance by increasing the thickness of

outer walls in epidermal cells (Mengel et al. 2001)









The objective of this chapter was to determine whether the availability of soil

cations (Ca, K, and Mg), influence the health and survival of C. florida. Because high

levels of Ca, K, and Mg have been associated with fungal disease resistance in other plant

species, we hypothesize (hypothesis 1) that forested stands with high densities of C.

florida trees would have higher concentrations of these cations in the soil. We further

hypothesize (hypothesis 2) that additional input of soil cations decreases disease severity

of dogwood anthracnose on C. florida.

Materials and Methods

Study Site

This study was conducted in Great Smoky Mountains National Park (GSMNP),

USA. Great Smoky Mountains National Park straddles the Tennessee and North Carolina

state line, encompassing slightly over 200 000 ha in the southern Appalachian Mountains.

The varied geology and topographic structure of GSMNP results in a wide-range of soil

conditions and associated vegetation communities. Mean annual temperature in

Gatlinburg, Tennessee (440 m a.s.l. and adjacent to GSMNP) is 12.90 C and mean annual

precipitation is 142 cm. Our study sites ranged in elevation from 287 to 975 m. Although

C. florida occurs in a variety of forest types, from mesic coves to xeric oak-pine

woodlands, we focused our study on oak-hickory forests on moderately steep to steep

slopes with southeast-south-northwest facing aspects. Oak-hickory is a major forest type

in GSMNP, covering 43,337 ha, (21% of the Park's total forest cover; Madden et al.

2004) and prior to anthracnose C. florida was the dominant woody species in the

understory (Jenkins and White 2002).









Forest Soil Sampling

During June-August of 2001-2004 soil samples were collected from seventy-nine

long-term vegetation plots (20 m x 20 m; 0.04 ha) established in twenty oak-hickory

stands in GSMNP. Cornusflorida stem density in the stands ranged from 0 to 1150 stems

ha-1 and basal area ranged from 0 to 0.7 m2 ha-1. Mineral soil was collected from four

random locations within each plot and then pooled together to create one composite

sample per plot. Samples were collected by scraping away the forest floor and collecting

the top 10 cm of mineral soil (A and upper B horizons).

Lab Analysis

Samples were analyzed for soil Ca, K, and Mg by A&L Analytical Laboratories,

Inc., Memphis, Tennessee and respective soil cation saturation (%) was calculated.

Samples were dried at 360 C for 6 hours, and then sieved through a 2 mm sieve. A 3 g

sample of dried soil was shaken with 30 mL of Mehlich III extracting solution (Mehlich

1984) for 5 minutes and then centrifuged. The solution was analyzed for Ca, K, Mg, and

Na by using an inductively coupled plasma emission spectrometer. Exchangeable

hydrogen (H) was calculated by leaching 5 g of dried soil with, 20 mL of 0.2 M

triethanolamine and 0.25 M barium chloride buffer solution (pH 8.1), then by 20 mL of

0.25 M barium chloride solution. The amount of standard acid needed to back titrate the

leachate to the methyl red and bromcresol green endpoint was used to calculate the

concentration of exchangeable acidity. Cation exchange capacity (CEC) was determined

by the summation of exchangeable base cations (Ca, K, Mg, and Na) and exchangeable

H. Soil cation saturation was calculated as the percentage of the respective cation that

contributed to total CEC.









Potted Plant Experiment

We tested the influence of three cations (Ca, K, and Mg) on C. florida seedling

survival and resistance to dogwood anthracnose by using a potted plant experiment. Six

hundred 1-0 bareroot C. florida seedlings were purchased from a commercial nursery in

Bartow, Florida, located approximately 65 km east of Tampa, Florida. Dogwood

anthracnose has not been reported in Florida; therefore these seedlings were presumed to

be disease free and showed no evidence of anthracnose infection before the experiment

began. The 1-0 seedlings were transplanted into three-gallon pots in a 40% Florida peat,

40% pine bark, and 20% sand mixture and grown in a nursery for one year. The seedlings

were watered every other day during the growing season. During the transplant into the

three-gallon pots, the seedlings were randomly selected to receive one of twelve

fertilization treatments. The treatments included three cations (Ca, K, and Mg) at four

levels (0, 50, 100, and 200%) of a standard nursery fertilization rate. Fertilization mixes

for the twelve treatments were mixed separately at the nursery using a base mix (Table 3-

1) with the addition of the appropriate amount of Ca, K, and Mg to create the required

treatments (Table 3-2). The seedlings were fertilized three times; the first fertilization

treatment was at the time of transplant (May 2003), the second was in October 2003, and

the final fertilization treatment was in March 2004, before leaf onset occurred. Depending

on the treatment, 59 to 68 g (base rate + the treatment addition) of fertilizer was placed

within each pot during each fertilization period.

In April 2004, the seedlings were transported to Twin Creeks Natural Resources

Center in GSMNP (350 42' 51"N, 830 30' 37"W). Dogwood anthracnose has been widely

reported in GSMNP, and heavy C. florida dieback has been reported in every forest type









where C. florida is found (Jenkins and White 2002). Seedlings were placed under

infected C. florida trees in a 70 year-old Liriodendron tulipifera L. and Acer rubrum L.

stand. This allowed C. florida trees infected with dogwood anthracnose to serve as a

source of natural inoculum (Britton et al. 1996). All seedlings were placed directly

underneath the C. florida canopy in a randomized complete block design with four

blocks.

Foliage samples were collected from the C. florida seedlings one day prior to

inoculation to test for foliar nutrient concentration. After collection, samples were dried

in an oven at 650 C for 72 hours and then ground using a tissue grinder. Analysis of

foliage samples was performed using an inductively coupled plasma emission

spectrometer (ICPES) at the University of Florida Analytical Research Lab (Gainesville,

Florida). A 750 mg sample of dried plant material was weighed into a 20 mL high form

silica crucible and dry ashed at 4850 C for 12 hours. The ash was equilibrated with 5 mL

of 20% HC1 at room temperature for 30 minutes. Then 5 mL of deionized water was

added, gently swirled and the sample was allowed to settle for 3 hours. The solution was

decanted into a 15 ml plastic vial for direct determination by ICPES. Results of tissue

concentrations are presented in mg g1.

Seedlings were measured for height and root collar diameter in April 2004 before

inoculation and no significant differences were found among treatments (P > 0.36). After

inoculation, seedling foliage was assessed every 2 weeks for presence of anthracnose.

Any nonanthracnose lesions were disregarded. We used a scale based on the Mielke-

Langdon index (Mielke and Langdon 1986) to assess disease severity on the seedlings

(Table 3-3).









Statistical Analysis

Forest Soil

In order to reduce plot variability found in the forest soil data, plot values were

averaged together for each stand. We then used linear regression models to describe the

relationship between soil cation saturation (Ca, K, and Mg) and C. florida stem density

and basal area. All statistical analyses were done using SAS (SAS 2002).

Potted Plant Experiment

In the potted plant experiment, ANOVA was conducted to test for differences in

infection ratings among the treatments. When ANOVA revealed significant main effects,

we separated the means with post-hoc pairwise comparisons. A logistical model was used

to test for differences in seedling mortality at the end of the experiment. We used

curvilinear regression to describe the relationship between C. florida tissue concentration

of Ca, K, and Mg to fertilization input. All statistical analyses were done using SAS (SAS

2002).

Results

Forest Soil Cation Saturation

We found significant positive correlations between the three cations (Ca, K, and

Mg) and C. florida stem density and basal area. Soil Ca saturation ranged from 5.7 -

26.3% and exhibited the strongest relationship of the three cations with C. florida stem

density (R2 = 0.70, P < 0.0001; Figure 3-1). Soil Mg saturation ranged from 2.3 8.2%

and soil K saturation ranged from 2.0 5.0%. We also found significant relationships of

soil K saturation (R2 = 0.54, P < 0.001; Figure 3-2) and soil Mg saturation (R2 = 0.62, P <

0.0001; Figure 3-3) with C. florida stem density.









When C. florida basal area was used as the independent variable, R2 and P-values

decreased for every cation. The strongest relationship was observed between soil Ca

saturation and C. florida basal area (R2 = 0.59, P = 0.001; Figure 3-1). A weak

relationship was observed between soil K saturation and C. florida basal area (R2 = 0.23,

P = 0.08; Figure 3-2), but a significant relationship was observed between soil Mg

saturation and C. florida basal area (R2 = 0.45, P = 0.008; Figure 3-3)

Potted Plant Experiment

Calcium Treatments

Among the four levels in the Ca treatments, the seedlings did not show a significant

difference among treatments until 6 weeks after inoculation. After 6 weeks, the 0%

treatment had the lowest (less healthy) infection rating (Table 3-4). Eight weeks after

inoculation, all treatments except for the 100% treatment, which had a higher (healthier)

infection rating, were statistically the same (P > 0.19; Table 3-4). This trend continued

until the end of the experiment. At the end of the experiment, mortality for the Ca

treatments ranged from 62 100% and was significantly different among the four

treatments (P = 0.0013; Figure 3-4). The 0% and 200% Ca treatments had 100 and 87%

mortality, respectively, which was significantly greater than the 72% mortality observed

in the 50% Ca treatment and 62% mortality observed in the 100% Ca treatment (P <

0.05).

Potassium Treatments

Seedlings in the K treatments began to show significant differences in infection

ratings 4 weeks after inoculation (P = 0.025; Table 3-4). Throughout the experiment, the

0% treatment had the lowest (less healthy) infection rating and the highest mortality

(Table 3-4 and Figure 3-4). The 100% treatment had a high (healthier) infection rating in









the beginning of the experiment and after 10 weeks was significantly higher (healthier)

than all other treatments (P < 0.02; Table 3-4). After 14 weeks, there were no significant

differences in infection ratings among the four K treatments (P = 0.44) and by the end of

the experiment (24 weeks) all treatments had suffered heavy mortality (> 85%, P = 0.26;

Figure 3-4).

Magnesium Treatments

There appeared to be very little difference in the anthracnose infection ratings

among the Mg treatments. Infection ratings were similar throughout the experiment;

however, the 200% treatment did have slightly lower (less healthy) infection ratings in

weeks 4 and 6 (Table 3-4). After 10 weeks there were no significant differences among

treatments (P = 0.17; Table 3-4), and at the end of the experiment there were no

significant differences in mortality among the four Mg treatments (> 88%, P = 0.26;

Figure 3-4).

Foliar Cation Concentrations

Mean Ca concentration in the foliage of the Ca treatments ranged from 3.59 mg g-1

(0% treatment) to 3.66 mg g-1 (100% treatment). There was not a significant relationship

between foliar Ca concentration and Ca input, perhaps due to the high sample variability

(R2 = 0.17, P = 0.43; Figure 3-5). In the K treatments, mean K concentration was highest

in the 100% treatment (2.05 mg g-) and lowest in the 0% and 200% treatments (1.76 and

1.77 mg g- respectively). The relationship between K concentration and K input was

significant (R2 = 0.61, P = 0.02; Figure 3-5). In the Mg treatments, Mg concentration was

greatest in the 50% treatment (0.83 mg g-) and lowest in the 200% treatment (0.72 mg

g-1). In general, after peaking in the 50% treatment, Mg concentration decreased as input









increased in the Mg treatments although the relationship was weak (R2 = 0.40, P = 0.098;

Figure 3-5).

Discussion

We found significant positive correlations between C. florida (stem density and

basal area) and soil Ca, K, and Mg saturation. Although correlation does not imply

causation, there are several possible explanations for these strong correlations. One

explanation is that higher soil cation concentrations help maintain healthy C. florida

populations even in the presence of dogwood anthracnose. Because these nutrients are

readily available in the forest soil, there is a decrease in inter and intraspecific

competition for nutrient resources. This enables C. florida to obtain more nutrients to be

used for plant defense and allocate more resources to developing stronger cells walls to

resist infection ofD. destructive. The higher nutritional status could also help in

replacing lost foliage and or repairing stem cankers arising from anthracnose infections.

It is also reasonable to argue that nutrient levels are greater on high C. florida stem

density sites because of the presence of C. florida. Multiple studies have shown the

influence of individual species on forest floor and mineral soil nutrient levels (Dijkstra

and Smits 2002, Washburn and Arthur 2003, Fujinuma et al. 2005), which occurs through

several different mechanisms. Certain species are able to uptake higher levels of nutrients

than others and secure nutrients in biomass before they are lost through soil leaching

(Dijkstra and Smits 2002). Another mechanism is the influence of plant species on

chemical weathering of the soil by modifying soil acidity (Augusto et al. 2000). It has

been hypothesized that because of the high Ca concentration and rapid decomposition of

C. florida foliage (Blair 1988, Knoepp et al. 2005) compared to other associated species,

C. florida acts as a "Ca pump" in forest soils (Thomas 1969). In addition to high









concentration of Ca in C. florida foliage, studies performed in western North Carolina

found that C. florida foliage contained high concentrations of K and Mg compared to

other species in oak hardwood forests (Day and Monk 1977, Elliot et al. 2002; Table 3-

5). High levels of these nutrients, particularly Mg, compared to other species were also

found in the wood, twigs, and bark of C. florida (Day and Monk 1977). High soil K and

Mg saturation in stands with high densities of C. florida indicates that this species may

also be acting as K and Mg "pumps" as well. It is possible that these two hypotheses are

not independent of each other.

The results of our potted plant experiment showed greater inputs of Ca and K

cations slowed the rate of anthracnose infection. We observed differences in the rates of

infection over the 6 month period with respect to the treatments. Calcium nutrition has

been linked to the resistance of many plant species to fungal and bacterial diseases

(Muchovej et al. 1980, Yamazaki et al. 1999, Sugimoto et al. 2005). Calcium plays a key

role in development of plant cell walls (Epstein 1972, Mengel et al. 2001) and studies

suggest that disease severity is reduced due to increased Ca concentrations in cell walls

(Muchovey et al. 1980, Conway et al. 1992, Sugimoto 2005). In our experiment,

seedlings in the 100% Ca treatment exhibited fewer signs of anthracnose and lower

mortality compared to 0% and 200% Ca treatments throughout the experiment. At the

end of the experiment, mortality for the Ca treatments were significantly greater in the

0% and 200% treatment compared to the 50% and 100% treatments. This indicates that

Ca availability is an important factor in C. florida resistance to dogwood anthracnose.

However, we did not see a relationship between foliar Ca concentration and Ca inputs.









Perhaps the high variability in foliar Ca concentration within each treatment precluded a

significant relationship (Figure 3-3).

Potassium is a macronutrient essential to the performance of multiple plant enzyme

functions (Epstein 1972, Mengel et al. 2001). Studies have indicated higher disease

resistance with increased levels ofK (Jeffers et al. 1982, Sij et al. 1985), which, although

the mechanisms are not completely understood, may be attributed to increased energy

and epidermal wall thickness (Mengel et al. 2001). In our experiment, the 100% K

treatment had a healthier infection rating in the early weeks of the experiment. After 10

weeks, the infection rating was still significantly healthier than all other treatments. The

100% K treatment also had the highest foliar concentration of K compared to other K

treatments (Figure 3-3). Therefore, increased disease resistance may be attributed to

increased K foliar concentration. Despite the additional K inputs in the 200% K

treatment, this treatment had lower K foliar concentration levels and unhealthier infection

ratings compared to the 100% K treatment. Decreased foliar concentration in the 200% K

treatment compared to the 100% K treatment can be attributed to nutrient imbalances

created by excess soil K; similar results have been reported by Wilmot et al. (1996).

Although we found a significant relationship between C. florida density and soil

Mg saturation, it did not appear that Mg input had any effect on disease severity or

mortality. Other studies of plant disease have found similar results (Nwoboshi 1980,

Wisniewski et al. 1995). Nwoboshi (1980) found that Mg fertilization rates had no effect

on the resistance ofManihot esculenta L. (cassava) to anthracnose. Wisniewski et al.

(1995) reported that Mg did not inhibit the germination or growth of two fungal diseases,

Botrytis cinerea Pers. or Penicillium expansum Link, whereas increased Ca









concentrations decreased spore germination and growth of both pathogens. Although

Anglberger and Halmschlager (2003) reported that severity of Sirococcus shoot blight

decreased on P. abies trees that had needles with high levels of Mg, in our experiment we

found that the treatment that produced the highest concentration of foliar Mg (50%

treatment) fared just as poorly as the other treatments.

Overall, all treatments suffered heavy mortality by the end of the growing season

after they were exposed to D. destructive (62-100%; Figure 3-2). This is not particularly

surprising considering the fact that multiple studies have shown smaller trees to be very

susceptible to the disease and often die within the first year of to exposure the disease

(Mielke and Langdon 1986, Hibben and Daughtery 1988, Hiers and Evans 1997). In

addition, it should be noted that the 2004 growing season (year of the potted plant

experiment) had above average precipitation in GSMNP (National Climatic Data Center

1999-2004; Figure 3-6) and dogwood anthracnose throughout the Park was virulent

(personal observation). The overwhelming presence of the disease might have decreased

some of the differences among the treatments. Dogwood anthracnose is a disease that is

most severe in cool, moist, and heavily shaded conditions (Chellemi and Britton 1992,

Chellemi et al. 1992, Britton 1993), and higher levels of mortality have been reported in

mesic forests compared to xeric forests (Jenkins and White 2002).

Britton (1993) reported that given adequate amounts of rainfall, dogwood

anthracnose could develop throughout the growing season. Factors affecting relative

humidity and evaporative potential of leaf surfaces in a stand, such as stand density,

slope, and elevation, probably influence impacts of dogwood anthracnose on C. florida

more than nutrient conditions. We still conclude, however, that low availability of Ca and









K in forested stands containing C. florida increases the susceptibility of C. florida to

dogwood anthracnose. Decreased nutrient availability in eastern forests from acid

deposition (Likens et al. 1996, McLaughlin and Wimmer 1999) has likely had a negative

impact on remaining C. florida populations, which may further inhibit the annual calcium

cycling of cations by C. florida (Holzmueller et al. 2006c).

Conclusion

The results of this project indicate that there is a correlation between soil cation

saturation (Ca, K, and Mg) and C. florida stem density and basal area in oak-hickory

forests. High concentrations of these cations in C. florida foliage suggest that this species

may play an important role in nutrient cycling by acting as a "pump" that draws cations

from deeper in a soil profile and cycles them through the forest floor and surface soil.

The results of this project also suggest that increased levels of Ca and K in the soil may

lead to increased resistance to dogwood anthracnose. We conclude that soil fertility in

forest stands should not be overlooked when applying management techniques to reduce

the impacts of dogwood anthracnose.






53


Table 3-1. Content of the base fertilizer mix.


Element Rate (%)
NH4 6.8
NO3 5.2
P205 6.0
AgSul 90 6.2
FeSO4 0.8
MnSO4 0.3
ZnSO4 0.1
CuSO4 0.05










Table 3-2. Inputs (%) added to the base fertilizer mix (separately) for each treatment.


Calcium (CaSO4)
rate (%)
0.0
0.75
1.5
3.0
1.5
1.5
1.5
1.5
1.5
1.5
1.5
1.5


Potassium (K20)
rate (%)
8.0
8.0
8.0
8.0
0.0
4.0
8.0
16.0
8.0
8.0
8.0
8.0


Magnesium (MgSO4)
rate (%)
1.8
1.8
1.8
1.8
1.8
1.8
1.8
1.8
0.0
0.9
1.8
3.6


Treatment
Ca 0%
Ca 50%
Ca 100%
Ca 200%
K 0%
K 50%
K 100%
K 200%
Mg 0%
Mg 50%
Mg 100%
Me 200%


Y






55


Table 3-3. Scale used to assess the severity of dogwood anthracnose infection on the
foliage of Cornusflorida seedlings. Scale was based on the Mielke-Langdon
Index (Mielke and Langdon 1986).

Rating % of foliage with
signs of anthracnose
0 Dead
1 76-100
2 51-75
3 26-50
4 1-25
5 0













Table 3-4. Biweekly infection ratings for Cornusflorida seedlings for length of the experiment.


Week Week Week Week Week Week Week Week Week Week Week Week
Treatment 2 4 6 8 10 12 14 16 18 20 22 24
Ca
0% 3.9 al 3.5 a 2.8a 2.5 a 1.5 a 1.1 a 0.8a 0.6a 0.4 a 0.1 a 0.1 a 0.0 a
50% 3.8a 3.8a 3.4b 2.9ab 1.9a 1.6a 1.4ab 1.1ab 1.1ab 0.7ab 0.6ab 0.4ab
100% 3.8 a 3.8 a 3.4 b 3.2 b 2.8 b 2.3 b 1.7 b 1.3 b 1.2 b 0.8 b 0.7 b 0.4 b
200% 3.8a 3.6a 3.0 a 2.7a 1.8 a 1.3 a 0.9a 0.8a 0.7a 0.3 a 0.3a 0.2a
SE 0.1 0.1 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.1 0.1 0.1
K
0% 4.0 a 3.3 a 2.5 a 1.3 a 0.8a 0.6a 0.5 a 0.3 a 0.1 a 0.1 a 0.1 a 0.1 a
50% 3.8 a 3.6 b 2.9 ab 1.8 b a.1 0.9 ab 0.6 a 0.4 a 0.4 a 0.3 a 0.3 a 0.2 a
100% 3.8 a 3.7 b 3.2 b 2.0 b 1.7 b 1.2 b 0.7 a 0.6 a 0.5 a 0.3 a 0.2 a 0.2 a
200% 4.0 a 3.7 b 3.2 b 1.7 a 1.2 a 0.8 ab 0.6 a 0.4 a 0.4 a 0.2 a 0.2 a 0.2 a
SE 0.1 0.1 0.2 0.2 0.2 0.2 0.2 0.1 0.1 0.1 0.1 0.1
Mg
0% 3.9a 3.9 a 3.4a 2.1 a 1.6 a 0.9a 0.6 a 0.5 a 0.5 a 0.3 a 0.2a 0.2a
50% 3.8 a 3.6 ab 3.0 ab 1.9 a 1.2 a 0.9a 0.6a 0.4a 0.3 a 0.1 a 0.1 a 0.1 a
100% 3.7 a 3.8 a 3.4a 2.4a 1.6 a 1.2a 1.0 a 0.8a 0.8a 0.3 a 0.3a 0.2a
200% 3.9 a 3.4 b 2.6 b 1.5 b 1.2 a 0.7 a 0.7 a 0.7 a 0.6 a 0.4 a 0.3 a 0.3 a
SE 0.1 0.1 0.2 0.2 0.2 0.2 0.2 0.1 0.1 0.1 0.1 0.1


using post-hoc


Means with different letters in same column for each cation for each week are statistically different (P < 0.05)
pairwise comparisons among sampling categories when ANOVA P-value < 0.05












Table 3-5. Foliar calcium (Ca), potassium (K), and magnesium (Mg) concentrations (%)
for selected species in a southern Appalachian forest. Data presented is from
Day and Monk (1977).


Species


Cornusflorida
Quercus alba
Quercus coccinea
Quercus prinus
Quercus rubra
Quercus velutina
Acer rubrum
Carya glabra
Liriodendron tulipifera
Oxydendrum arboreum
Nyssa sylvatica
Magnolia fraseri
Betula lutea
Sassafras albidum


Foliar concentration (%)
Ca K Mg
1.60 1.18 0.90
0.50 0.75 0.14
0.45 0.62 0.14
0.59 1.09 0.19
0.75 0.89 0.33
0.71 0.95 0.17
0.62 0.53 0.20
0.95 0.58 0.82
1.39 1.04 0.61
0.96 0.78 0.27
0.96 1.04 0.51
1.07 1.25 0.38
1.11 1.10 0.37
0.52 1.08 0.27











30

S25

CO
20-
2 15
15

0)
o 5
C,


500


1000


1500


Corus florida density (stems ha-1)


30

"25

O 20
20
3 15

0
5 10
L)

0 5
C,
0-


Corus florida basal area (m2 ha-1)

Figure 3-1. Linear regression between soil calcium (Ca) saturation and Cornusflorida
stem density and basal area.


Sy = 0.0145x + 8.4861
R2 = 0.7042
*P < 0.0001


y = 19.408x + 5.7707
SR2 = 0.5933
P = 0.001











6



04 .*4-



S2 y = 0.0022x + 2.7002
R2 = 0.539
c, P < 0.001

0
0 500 1000 1500

Cornus florida density (stems ha1)
6







2 y = 2.2548x + 2.5135
2
SR2 = 0.2319
SP = 0.08
0 -
0.0 0.2 0.4 0.6 0.8

Cornus florida basal area (m2 ha-1)

Figure 3-2. Linear regression between soil potassium (K) saturation and Cornusflorida
stem density and basal area.






60




9



6-


,* y = 0.0043x + 3.7915
S R2 = 0.623
0 P < 0.0001
c,
0 -
0 500 1000 1500

Corus florida density (stems ha-1)

9




6-


p 3 y = 4.4378x + 3.5256
R2 = 0.4502
P = 0.008
0 -
0.0 0.2 0.4 0.6 0.8

Corus florida basal area (m2 ha-1)

Figure 3-3. Linear regression between soil magnesium (Mg) saturation and Cornus
florida stem density and basal area.






61


100
Calcium Treatments -

80 -
o#0 .0 ,- "



t 40 '"
o -0%
0 O -50%
20 V -o 100%
S. -200%



0
100
Potassium Treatments .. ^_ / < .-- "



60 / -0%


S40 ff
oo=
/,-/' 0%
-50%
20 -/ -- .100%
-- -200%
100
100
Magnesium Treatments .,,
80 "


S60 -X


t 40 -
0 -0%
S-50%
20 -- o100%
-200%
0 5, 07---------------








5/17/2004 6/16/2004 7/16/2004 8/15/2004 9/14/2004 10/14/2004

Figure 3-4. Biweekly mortality (%) of Cornusflorida seedlings for the four treatment
levels of each cation.



















y = -1E-04x2 + 0.015x + 35.8
R2 = 0.17
P = 0.427

0% 50% 100% 200%


y =-0.0002x2 + 0.047x + 17.2
R2 = 0.6132
P = 0.02









0% 50% 100% 200%


y = -4E-05x2 + 0.0053x + 7.8
R2 = 0.4021
TP = 0.098


50% 100%


200%


Treatment

Figure 3-5. Foliar calcium (Ca), potassium (K), and magnesium (Mg) concentrations of
Cornusflorida seedling foliage for the four treatment levels of each cation.


E
c2.0
0


0 1.8
C
0
O

S1.6


0.9


E
S0.8
0


| 0.7
0
0

0.6


0%







63


25


S20







i 5' 2004 precipitation
= 5

0 Average precipitation
0
April May June July August September
Month

Figure 3-6. Precipitation data for 2004 and previous 5 year average (1999-2003) during
April-September at the Twin Creeks Natural Resources Center, Great Smoky
Mountains National Park (Data from National Climatic Data Center 1999-
2004).














CHAPTER 4
INFLUENCE OF Cornusflorida L. ON CALCIUM MINERALIZATION IN TWO
SOUTHERN APPALACHIAN FOREST TYPES

Introduction

Over the past two decades, numerous studies have raised concerns about calcium

(Ca) depletion in forest soils of the eastern United States (Likens et al. 1998, Huntington

et al. 2000, Johnson et al. 2000, Yanai et al. 2005). In many forests, this depletion has

been well documented. For example, between 1965-1992, Likens et al. (1998) estimated

a loss of 9.9-11.5 kmol ha-1 of total Ca from the complete soil profile at Hubbard Brook

Experimental Forest in New Hampshire. In a 60-80 year old southern Piedmont forest in

Georgia, Huntington et al. (2000) estimated that the soil Ca depletion rate was 12.7 kg

ha-1 y1. Calcium depletion has been attributed to leaching caused by acid deposition

(Lawrence et al. 1995, Likens et al. 1996) and uptake and sequestering of nutrients in

woody biomass (Johnson and Todd 1990, Huntington et al. 2000). The ecological

consequences of soil Ca depletion could be devastating since long-term forest ecosystem

health and sustainability have been closely linked to pools of available Ca in the soil

(Graveland et al. 1994, NAPAP 1998, Driscoll et al. 2001, Hamburg et al. 2003).

In mixed hardwood forests of eastern North America, Ca released through mineral

weathering is generally an insignificant contributor to total calcium cycling (Huntington

et al. 2000, Dijkstra and Smits 2002). As a result, the release of Ca through organic

matter decomposition mineralizationn) is considered the major source of Ca for

immediate uptake by forest plants (Likens et al. 1998, Dijkstra and Smits 2002). In a









study in northwestern Connecticut, Dijkstra (2003) reported that Ca mineralization

occurred primarily in the forest floor (from leaf litter) as opposed to the mineral soil.

Foliar Ca concentrations vary greatly among tree species (Metz 1952, Day and Monk

1977, Elliot et al. 2002), which influences the amount of Ca mineralized in the forest

floor beneath the canopy of a given tree species (Dijkstra 2003).

Cornusflorida L. foliage, on average, has a higher Ca concentration (2.0-3.5%) and

more rapid decomposition than that other woody species (Thomas 1969, Blair 1988,

Knoepp et al. 2005). Cornusflorida was, historically, one of the most common

understory species in eastern United States hardwood forests (Muller 1982, Elliott et al.

1997, Jenkins and Parker 1998). Because of the high Ca concentration of its foliage, rapid

decomposition of its litter and its abundance in the understory, C. florida has long been

believed to influence Ca availability in the soil and forest floor by acting as a "Ca pump"

that draws calcium from deep in the soil profile and deposits it on the biotically-rich

forest floor and surface soil (Thomas 1969, Jenkins et al. 2006). However, during the past

20 years, C. florida has suffered heavy mortality (over 90% in some forest types)

throughout most of its range due to the rapid spread of the fungus Discula destructive

Redlin, which causes the disease dogwood anthracnose (Anagnostakis and Ward 1996,

Sherald et al. 1996, Hiers and Evans, 1997, Jenkins and White 2002, Holzmueller et al.

2006). The loss of C. florida foliar biomass from the understories of eastern hardwood

forests has the potential to further reduce Ca availability in these forests.

Although many have suggested that C. florida plays an important role in Ca cycling

and availability in eastern forests (Thomas 1969, Hiers and Evans 1997, Jenkins and

White 2002), no studies have been conducted to quantify the impact of C. florida on Ca









mineralization in forests where the species occurs. Understanding the influence of C.

florida on the mineralization rate of forest stands is critical to understanding the impacts

of dogwood anthracnose on calcium cycling. The objective of this study was to quantify

Ca mineralization in the forest floor and mineral soil along natural gradients of increasing

C. florida stem density. Because C. florida litter decomposes very rapidly compared to

other species (Thomas 1969, Blair 1988, Knoepp et al. 2005), and its foliage contains

high concentration of Ca (Thomas 1969, Blair 1988), we hypothesized that stands with

high densities of C. florida have higher rates of Ca mineralization than stands with lower

C. florida densities.

Materials and Methods

Study Area

We conducted this study in the north-central portion of Great Smoky Mountains

National Park (GSMNP), near Gatlinburg, Tennessee. Mean annual temperature in

Gatlinburg, Tennessee (440 m a.s.l.) is 12.90 C and mean annual precipitation is 142 cm.

Study site elevation ranged from 487 to 762 m. All sampling was performed in two forest

types; oak hardwood and cove hardwood. The most common species in the oak hardwood

forest type were Quercus alba L., Quercusprinus L., Quercus coccinea Muenchh., Carya

alba (L.) Nutt., Pinus strobus L., Oxydendrum arboreum (L.) DC., and Nyssa sylvatica

Marsh. In the cove hardwood forest type, the most common species were Liriodendron

tulipifera L., Acer rubrum L., Tsuga canadensis (L.) Carr., Fagus grandifolia Ehrh.,

Betula lenta L., and Magnolia fraseri Walt. All study sites were located in secondary

forests that were logged prior to park establishment in 1934 (Pyle 1988). Soils in this area

were predominantly Junaluska-Tsali complex, Soco-Stecoah complex, and Spivey-

Santeetlah-Nowhere complex. These complexes, found in both forest types, are typically









well drained, form on moderate slopes (15-45%), are sometimes stony, and are derived

from soft metasandstone (Anthony Khiel, soil scientist, NRCS, personal communication).

Field Sampling

We determined Ca mineralization in the forest floor and mineral soil using the

buried bag in situ incubation method described by Eno (1960). This technique has

recently been utilized to quantify Ca mineralization in forested ecosystems (Dijkstra

2003) and is commonly used to estimate N mineralization as well (Prescott et al. 2003,

Allen et al. 2005). We collected data for 2 years; bags were buried in early June

2003/2004 (summer incubation) and again in early December 2003/2004 under freshly

fallen leaf litter (winter incubation). Overall, sixty-eight 10 m x 10 m plots were sampled

every year, thirty in the cove hardwood forest type and thirty-eight in the oak hardwood

forest type. For each forest type, plots were divided into three sampling categories based

on C.florida stem density: (0 stems ha-1, 200-300 stems ha-1, and > 600 stems ha-l),

hereafter referred to as zero, low, and high density, respectively. Each plot was

surrounded by a 20 m buffer from the outside edge of the plot that was void of other C.

florida stems. There was a minimum of 100 m separating plots from each other.

In each plot, two forest floor samples (20 cm x 20 cm) were taken underneath the

canopy of the C. florida trees, but were at least 1 m away from the nearest tree base.

Forest floor mass and depth were determined from these samples. On plots where no C.

florida trees were present, two forest floor samples were randomly collected within the

10 m x 10 m plot, and were at least 1 m away from the nearest tree base. Where the forest

floor was removed, a soil core (4 cm x 15 cm) was extracted. Each forest floor and

underlying soil sample was divided into two equal parts, transferred to polyethylene bags,

and closed with a knot. Two litter and two soil bags (initial sample) from each plot were









returned to the lab to determine dry mass, pH, and exchangeable Ca. The remaining

sample bags (final sample) on each plot were then returned to the spot from which they

were collected, and the soil bags were buried in the core holes and the forest floor bags

were placed in the litter layer. The bags containing the forest floor were covered with

fresh forest litter. Six months after incubation, the bags were retrieved and brought back

to the laboratory for further analysis.

Laboratory Analysis

Once in the lab, the contents of the bags were dried in an oven at 700 C for 72

hours. After drying, the mineral soil was sieved through a 2 mm sieve and the forest floor

was ground using a tissue grinder. Subsamples of the forest floor and mineral soil were

dried at 1050 C for 48 hours to measure gravimetric moisture content. Samples of the

forest floor and mineral soil were then measured for pH in de-ionized water slurry (10:1

ratio for the forest floor and 2:1 ratio for the soil). Samples were stirred initially and

again after 15 minutes. After settling for 30 minutes following the final stirring, pH was

measured. We extracted both the mineral soil and forest floor (separately) samples using

10 g of mineral soil and 5 g of forest floor mixed with 100 ml of 0.1M BaC12 in a 120 ml

vial. Samples were shaken for 2 hours on a soil shaker and filtered after settling for 24

hours using a coffee filter. Exchangeable Ca was measured using an inductively coupled

plasma emission spectrometer at the University of Florida Analytical Research

Laboratory (Gainesville, Florida). Calcium mineralization was determined as the

difference between final and initial exchangeable Ca in the bags.

Statistical Analysis

Summer and winter initial Ca concentrations and pH for the 2 years were compared

using paired t-tests. Because there was no statistical difference (P > 0.05) between the 2









years for any variable the data were combined for further analysis. Differences between

initial and final pH, exchangeable Ca from initial summer and initial winter periods, and

Ca mineralization from summer and winter incubations were tested using paired t-tests

for each forest type. ANOVA was used to test for differences of forest floor mass and

depth, initial exchangeable Ca concentrations, and Ca mineralization for the three C.

florida densities for each forest type. We also tested the relationship between Ca

mineralization and stem density, basal area, and foliar biomass (Martin et al. 1998) using

step-wise multiple regression. Step-wise multiple regression did not yield any significant

relationships among any variables and as a result data are not shown. All statistical

analyses were done using SAS (SAS 2002).

Results

Forest Floor Mass and Depth

There were no significant differences in forest floor mass (2.5 2.9 kg m-2) in cove

hardwood plots for both summer and winter periods (P > 0.61). There also were no

significant differences in forest floor mass (2.7 3.1 kg m-2) in oak hardwood plots

during summer and winter periods (P > 0.37). There were no significant differences in

forest floor depth (3.4 4.0 cm) in cove hardwood plots during summer and winter

periods (P > 0.39). There were also no significant differences in forest floor depth (3.5 -

3.8 cm) in the oak hardwood plots during summer and winter periods (P > 0.47).

Soil pH

In cove hardwood plots, mean forest floor pH ranged from 4.63 to 5.77 in the forest

floor and from 4.01 to 4.95 in the mineral soil (Table 4-1). In oak hardwood plots, mean

forest floor pH ranged from 4.01 to 5.40 and mean mineral soil pH ranged from 3.65 to

4.49 (Table 4-1). In cove hardwood plots, the mean pH of the forest floor increased









slightly during summer incubation on zero and low density plots (P < 0.01), but did not

significantly change on high density plots (P = 0.50). Mean mineral soil pH did not

change significantly for any density class during either incubation period (P > 0.43). In

oak hardwood plots, mean pH increased slightly in the forest floor and mineral soil of

zero density plots (P = 0.1 and P = 0.01) and in the forest floor of low density plots (P =

0.005). During the winter incubation, there was a significant increase in pH between the

initial and final samples in both forest types for all C.florida densities (P < 0.001; Table

4-1).

Initial Exchangeable Ca

Initial mean values for exchangeable Ca varied with C. florida density in both the

forest floor and mineral soil for both forest types. Mean forest floor values in the cove

hardwood plots ranged from 5.5 to 8.4 g kg-1, which was about ten times greater than

mean values found in the cove hardwood forest mineral soil 0.45 to 0.85 g kg-1 (P <

0.001; Figure 4-1). In oak hardwood plots, forest floor mean values ranged from 3.6 to

7.4 g kg-1, and mineral soil mean values ranged from 0.19 to 0.68 g kg-1 (Figure 4-1).

Initial exchangeable Ca mean values generally increased with C. florida density for both

forest types for summer and winter incubation periods and were significantly greater in

the high C. florida density plots compared to the zero density plots in both the forest floor

and mineral soil in both forest types (P < 0.01). Comparisons of initial exchangeable Ca

mean values made between the winter and summer incubations showed no significant

differences between the two periods (Figure 4-1).

Ca mineralization

Ca mineralization was greater in the forest floor than in the mineral soil for both

forest types (P < 0.0001). Mean values ranged from 2.09 to 9.16 mg kg-1 day-' for the









cove hardwood forest floor and from -1.10 to 0.20 mg kg-1 day-1 in the cove hardwood

mineral soil (Figure 4-2). For the oak hardwood forests, mean values ranged from 1.31 to

7.16 mg kg-1 day-1 in the forest floor and from -0.40 to 0.23 mg kg-1 day-1 in the mineral

soil (Figure 4-2). Mean values for the winter incubation period were significantly lower

than summer incubation period for most of the mineral soil comparisons, but were not

significantly different from the summer incubation periods for the forest floor

comparisons for both forest types (P > 0.21). Cornusflorida density had an effect on the

forest floor in both forest types (P < 0.1), with increasing C. florida density leading to

increased Ca mineralization (Figure 4-2).

Yearly Ca mineralization was greater in the high C. florida density plots compared

to the zero density plots for both forest types; cove hardwood, high density (3.3 g kg-1

yr-) versus zero density (0.6 g kg-1 yr1, P = 0.04) and oak hardwood, high density (2.4 g

kg-1 yr-) versus zero density (1.1 g kg-1 yr-1, P = 0.09; Table 4-2). In most cases, yearly

Ca mineralization in the mineral soil was negative, indicating Ca immobilization.

Discussion

Most of the yearly Ca mineralization in our study can be attributed to the forest

floor, which is similar to Dijkstra's (2003) findings in northwestern Connecticut that,

under most tree species, forest floor Ca mineralization far exceeded mineral soil Ca

mineralization. The two species that did have mineral soil exchangeable Ca inputs that

were comparable to the forest floor in Dijkstra's (2003) study were Acer saccharum

Marsh. and Fraxinus americana L., and this increase in mineral soil inputs was attributed

to high earthworm activity. Neither of these two species was in high abundance in either

forest type in our study. In addition, in glaciated areas, such as Connecticut, exotic

earthworm species tend to dominate over native species. The exotic species tend to break









down litter and duff at a much more rapid rate than natives. In non-glaciated areas, such

as the southern Appalachians, native earthworms dominate and the break down of coarse

organic matter is much slower (Hendrix and Bohlen 2002).

Ca mineralization differed significantly in the forest floor among the three C.

florida densities in both forest types and incubation periods. Within the forest floor, Ca

mineralization was significantly higher in the high density C. florida plots, except for the

winter incubation period in the oak hardwood forest type which did not show a

significant difference due to high plot variability. Increased mineralization in high density

C. florida plots could be attributed to the high Ca concentration (Thomas 1969, Blair

1988) and rapid decomposition of C. florida foliage (Thomas 1969, Blair 1988, Knoepp

et al. 2005). Mineralization in the forest floor did not differ between winter and summer

incubation periods for any density level or forest type. Dijkstra (2003) reported Ca

mineralized was greater in the summer incubation period compared to the winter

incubation period and attributed this to warmer temperatures during the summer

incubation period. The warmer winters of Tennessee compared to Connecticut may have

offset this difference and resulted in comparable winter and summer values. It should be

noted though, that mineral soil Ca mineralization values were significantly lower during

the winter incubation period of our study.

Initial exchangeable Ca values in the zero density C. florida plots were slightly

higher in the cove hardwood forest type than in the oak hardwood forest type in both the

mineral soil and forest floor (P < 0.001). This result could be attributed to the different Ca

concentrations found in the foliage of the dominant species in each forest type. Numerous

studies have shown how soil properties can be influenced by tree species (Boettcher and









Kalisz 1990, Finzi et al.1998, Dijkstra and Smits 2002, Fujinuma et al.2005). The cove

hardwood forest type was primarily dominated by L. tulipifera, which has an average of

1.74% Ca concentration in its foliage, compared to the dominant species in the oak

hardwood forest type, Q. alba, which has much lower average Ca foliar concentration

(0.73%) (Jenkins et al. 2006) (See Table 4-3 for a listing of species and corresponding Ca

concentration).

Despite the differences in initial values of exchangeable Ca between the two forest

types, values for initial exchangeable Ca were significantly greater in high density C.

florida plots compared to zero density plots in both the forest floor and mineral soil for

both forest types for both incubation periods. While previous research has focused on the

relationship between overstory trees and soil chemistry, our study shows that a single

understory woody species can have considerable influence on soil chemical properties.

One would assume that given the larger size of overstory trees, most of the forest floor

biomass comes from overstory trees and not from understory trees, therefore

overwhelming any effect understory trees might have on soil chemical properties.

However, in a study by Jenkins et al. (2006) in GSMNP, the authors reported that

understory foliar biomass contributed up to 49% of total stand foliar biomass, depending

on forest type and stand developmental stage.

Because of dogwood anthracnose, C. florida density has greatly declined across the

eastern United States (Anagnostakis and Ward 1996, Sherald et al. 1996, Hiers and Evans

1997, Jenkins and White 2002), dramatically reducing C. florida foliar biomass added to

the forest floor. In GSMNP, there has been a significant reduction in the amount of Ca

cycled to the forest floor in forest types containing C. florida. Typic cove forests









experienced a 86% decline in C. florida leaf litter over a 20 year period since 1977,

resulting in a corresponding decline (85%) in annual Ca inputs and oak-hickory forests

experienced a 78% reduction in C. florida leaf litter during the same period, resulting in a

78% reduction in annual Ca inputs (Jenkins et al. 2006). Throughout much of the Park,

and likely across the southern Appalachians as well, C. florida trees have largely been

replaced by T. canadensis, a species with more acidic litter that contains little Ca (Jenkins

and White 2002). Increased T. canadensis densities in the forest understory could further

disrupt Ca cycling in eastern forests. In a study that compared base cation levels beneath

three tree species (Acer saccharum Marsh., Tilia Americana L., and Tsuga canadensis) in

Ottawa National Forest in western Upper Michigan, the authors reported high levels of

base cation leaching underneath T. canadensis canopies which was attributed to the low

uptake of these cations by T. canadensis (Fujinuma et al. 2005). The hemlock woolly

adelgid (Adelges tsugae Annand), however, is spreading rapidly within GSMNP, and

forests across the southern Appalachians may experience heavy T. canadensis mortality,

similar to that observed in the northeastern United States (Johnson et al. 1999). If this

occurs, the importance of shade tolerant hardwood species, such as A. rubrum, may

increase in the forest types we sampled. While this and other hardwood species typically

contribute more calcium to annual cycling than T canadensis, their contributions are still

much lower than that of C. florida.

In a study of regional forest plant species diversity in central Europe, Cornwell and

Grubb (2003) reported the highest levels of plant species richness were found on nutrient

rich soils. Although it is not clear how loss of Ca inputs from C. florida has affected the

vigor and growth of other species and the overall stand dynamics in eastern hardwood









forests, several other common species co-occurring with C. florida (Q. coccinea, Q.

rubra L., and Robiniapseudoacacia L.) have been reported to show increased levels of

mortality in the southern Appalachian Mountains while dogwood anthracnose has been

dramatically reducing C. florida in eastern forests (Wyckoff and Clark 2002). Cornus

florida litter is a major source of Ca and a decline in foliar biomass could negatively

affect Ca mineralization in the soil, the primary source of uptake in eastern forests

(Dijkstra and Smits 2002), disrupting the Ca cycle (Figure 4-3). This negative impact,

combined with acid deposition, may eventually result in further Ca depletion (Hamburg

et al. 2003), which has been associated with canopy dieback in some eastern hardwood

forests (Wilmot et al. 1996).

Lack of Ca can affect other components of forest ecosystems, including soil fauna.

Decreased land snail abundance has been correlated with decreased exchangeable Ca

levels in Sweden (Wareborn 1992) and the central Appalachian Mountains (Hotopp

2002). In the Netherlands, poor reproductive success in Parus major L. (great tit, a

passerine bird) because of a lack of Ca in eggshells was attributed to reduced snail

abundance on soils depleted of calcium by acid deposition (Graveland et al. 1994,

Graveland 1996).

Because of the important role C. florida plays in the Ca cycle, preventing its loss

may be of critical importance in eastern forests. This is a difficult task due to the presence

of dogwood anthracnose. However, there has been some indication that prescribed

burning with proper frequency can help retain C. florida as a component of stands

infected with this disease (Holzmueller et al. 2006a).






76


Conclusion

These results suggest that C. florida density significantly affects Ca mineralization

in both cove hardwood and oak hardwood forest types, primarily in the forest floor. The

influence of C. florida on Ca mineralization may be attributed to the high Ca

concentration and rapid decomposition of its foliage. Because mineralized Ca in the

forest floor is the primary source of available Ca in eastern hardwood forests, loss of C.

florida may further alter Ca cycling in these forests with subsequent negative impacts on

associated flora and fauna.










Table 4-1. Mineral soil and forest floor mean pH (+ 1 SE) for summer and winter
incubations in the cove hardwood and oak hardwood forest types.

Summer Winter
Initial Final Initial Final
Cove hardwood
Forest Floor
None 4.63 (0.20) 5.31 (0.25)*** 4.71(0.15) 5.38 (0.21)***
Low 5.30 (0.19) 5.77 (0.15)** 5.30 (0.17) 5.72 (0.19)***
High 5.14 (0.17) 5.20 (0.25) ns 5.31(0.12) 5.71 (0.14)***

Mineral Soil
None 4.26 (0.16) 4.19 (0.16) ns 4.01(0.11) 4.95 (0.09)***
Low 4.29(0.12) 4.27 (0.15) ns 4.15 (0.09) 4.60 (0.09)***
High 4.26(0.11) 4.11 (0.07)ns 4.04(0.07) 4.63 (0.07)***

Oak hardwood
Forest Floor
None 4.13 (0.11) 4.35 (0.16)* 4.01 (0.08) 4.41 (0.07)***
Low 4.47 (0.19) 4.78 (0.20)** 4.41(0.12) 4.77 (0.15)**
High 5.23 (0.19) 5.40 (0.22) ns 5.11 (0.10) 5.48 (0.16)**
Mineral Soil
None 3.65 (0.05) 3.49 (0.08)** 3.65 (0.27) 4.14 (0.04)***
Low 3.74(0.14) 3.76 (0.19) ns 3.86(0.09) 4.45 (0.16)***
High 4.02 (0.12) 3.95 (0.18) ns 4.01(0.09) 4.49 (0.11)***
SStatistical comparisons were made between the initial and final samples for each density
in each forest type using paired t-tests, ns = P > 0.1, = P < 0.1, ** = P < 0.01, *** = P
< 0.001









Table 4-2. Mean yearly Ca mineralization (+ 1 SE) in forest floor, mineral soil, and
combined total (forest floor plus mineral soil) for the three Cornusflorida
sampling densities in the cove hardwood and oak hardwood forest types.

Forest Floor Mineral Soil Total
(g kg- yr-1) (g kg1 yr-) (g kg1 yr-1)
Cove hardwood
Zero 0.7 (0.8) al -0.1 (0.05) b 0.6 (0.8)a
Low 1.1 (0.9) a -0.2 (0.07)a 0.9 (0.9)a
High 3.3 (0.8) b 0.0 (0.04) 3.3 (0.8) b

Oak hardwood
Zero 1.1 (0.8)a -0.04 (0.01)a 1.1 (0.8)a
Low 2.6 (0.8) b -0.04 (0.04)a 2.5 (0.8) b
High 2.4 (0.9) b -0.03 (0.07) a 2.4 (0.9) b
1 Means with different letters in same column for each forest type are statistically
different (P < 0.1) using post-hoc pairwise comparisons among categories when
ANOVA P-value < 0.1









Table 4-3. Foliar calcium concentrations (%) from dominant species in the two forest
types (oak hardwood and cove hardwood) sampled in this study. Data from
trees sampled within Great Smoky Mountains National Park on long-term
vegetation plots (NPS unpublished data).

Cove hardwood Calcium Oak hardwood species Calcium
species concentration concentration
(%) (%)
Liriodendron tulipifera 1.74 Quercus spp. 0.73
Acer rubrum 0.82 Carya spp. 0.98
Tsuga canadensis 0.46 Pinus strobus 0.29
Betula lenta 0.95 Oxydendrum arboreum 0.91
Magnolia fraseri 1.07 Nyssa sylvatica 0.78
Cornusflorida 1.73 Cornusflorida 1.73
Data from study by Day and Monk (1977) in Coweeta Hydrologic Laboratory located in
southwestern North Carolina














10 10
ns b ns ns
8 Forest floor b 8 Forest floor b




None Low High None Low High
a as








SCornus florida stem density
-i 4 4

2 2
0 0



fore0 stMineral Soil b summer and winter collection times. Mineral Soil

were made for each density; all were nonsignificant (ns; P > 0.05). Bars for the same collection period in each panel with

5 different letters are significantly different (P < 0. 0.5) using post-hoc pairwise comparisons among categories when ANOVA
0P-valu3 0.3 a a0.05.
0.0 0.0 i
None Low High None Low High
Comus florida stem density

SInitial summer Initial winter

Figure 4-1. Mean initial exchangeable Ca levels (+ 1 SE) in the forest floor and mineral soil in the cove hardwood and oak hardwood
forest types during summer and winter collection times. Comparisons between summer and winter values in each panel
were made for each density; all were nonsignificant (ns; P > 0.05). Bars for the same collection period in each panel with
different letters are significantly different (P < 0.05) using post-hoc pairwise comparisons among categories when ANOVA
P-value < 0.05.


Cove hardwood forest type


Oak hardwood forest type













Cove hardwood forest type


T ns T
b | |b


High


Forest floor


ns
a


Zero


High Zero
Cornus florida stem density


SSummer Incubation


b ns T
b a


ns
bI a


Low High


Low High


0 Winter Incubation


Figure 4-2. Mean Ca mineralization ( 1 SE) for the forest floor and mineral soil in the cove hardwood and oak hardwood forest types
during summer and winter incubation periods. Comparisons between summer and winter values in each panel were made
for each density (ns = P > 0.05, = P < 0.05, ** = P < 0.001, *** = P < 0.0001). Bars from the same incubation period in
each panel with different letters are significantly different (P < 0.1), using post-hoc pairwise comparisons among categories
when ANOVA P-value < 0.1.


Forest floor


ns a


ns

a_ a


Zero


1.0
0.6
0.2
-0.2
-0.6
-1.0
-1.4


Mineral Soil
a -


ns
a


Zero


Low


Oak hardwood forest type


T b











Atmospheric deposition
and precipitation


Streamwater


Figure 4-3. Conceptual model of calcium (Ca) cycling in an eastern United States
hardwood forest. Arrow thickness indicates amount of Ca movement and box
size indicates size of available Ca pool based on data from Johnson et al.
(1985) and Yanai et al. (2005). Loss of Cornusflorida may decrease the size
of the forest floor Ca pool and therefore overall Ca availability may be less in
oak hardwood and cove hardwood forest types.














CHAPTER 5
SUMMARY AND CONCLUSION

This research project examined the influence of dogwood anthracnose and the

ecological role of Cornusflorida L. in Great Smoky Mountains National Park (GSMNP).

Specifically, the effects of past burning on C. florida survival and health (Chapter 2), the

effects of calcium (Ca), potassium (K), and magnesium (Mg) on dogwood density and

health (Chapter 3), and role of C. florida in Ca mineralization (Chapter 4) were examined

over a three year period. Findings from these three interrelated studies are briefly

summarized below.

In Chapter 2, we examined C. florida populations in burned and unburned oak-

hickory stands to determine if burning prior to anthracnose infection has reduced the

impacts of anthracnose. We hypothesized that fire has altered stand structure and created

open conditions less conducive to dogwood anthracnose, which is most virulent in moist

heavily shaded stands. We compared C. florida density, C. florida foliar infection and

crown dieback, stand structure, species composition, Tsuga canadensis (L.) Carr. density,

plot species richness, and plot diversity among four sampling categories: unburned

stands, and stands that had burned once, twice, and three times (single, double, and triple

burn stands, respectively) over a 20 year period (late 1960s to late 1980s). We also

analyzed community composition using multivariate analyses. Double burn stands

contained the greatest density of C. florida stems (770 stems ha-1) followed by triple burn

stands (233 stems ha-1), single bur stands (225 stems ha-1) and unburned stands (70

stems ha-1). While foliar infection ratings did not differ between categories, we observed









less crown dieback in small trees (< 5 cm dbh) in burned stands than in unburned stands

(P < 0.05). Total overstory density was greater in unburned stands (564 stems ha-1) than

in double and triple burned stand (317-436 stems ha-1, P < 0.0001), but understory stem

density was greater in burned stands (2851-5072 stems ha-1) than unburned stands (2292

stems ha-1, P = 0.024). However, the understory density and importance value of T.

canadensis, a coniferous species that creates heavy shading in forest understories, were

considerably greater in unburned stands than in burned stands. The results of our study

suggest that prescribed fire may offer a management tool to reduce the impacts of

dogwood anthracnose in eastern hardwood forests.

In Chapter 3 we found positive correlations between soil Ca, Mg, and K saturation

and C. florida stem density and basal area. We tested the effect of these cations at four

levels (0, 50, 100, and 200%) of a standard nursery fertilization rate on C. florida

seedling survival and resistance to dogwood anthracnose. Although most of the seedlings

died after one season of exposure to dogwood anthracnose, we found that seedlings that

had lower inputs of Ca and K cations showed higher levels of disease severity sooner

than seedlings in other treatments, suggesting these nutrients play a role in C. florida

survival from anthracnose. Magnesium treatment levels did not appear to have an effect

on C. florida disease severity or mortality.

In Chapter 4 we sampled sixty-eight 10 m x 10 m plots in two forest types, cove

hardwood and oak hardwood, to quantify the influence of C. florida density on initial

exchangeable Ca and Ca mineralization in the mineral soil and forest floor. Cornus

florida density was classified into three levels in both forest types (zero = 0 stems ha-1,

low = 200-300 stems ha-1 and high = > 600 stems ha-1). We found significantly greater









levels of initial exchangeable Ca on high density plots, compared to low density plots in

both forest types in the forest floor and mineral soil (P < 0.01). Calcium mineralization

occurred primarily in the forest floor and not in the mineral soil in both forest types.

Yearly Ca mineralization was greatest in the high density C. florida plots (cove

hardwood, high density 3.3 g kg-1 yr-1 versus zero density 0.6 g kg-1 yr-1, P = 0.04 and oak

hardwood, high density 2.4 g kg-1 yr- versus zero density 1.1 g kg-1 yr-1, P = 0.09). These

results indicate that the loss of C. florida from eastern United States forests will further

alter the Ca cycle and may negatively affect the health of eastern hardwood forests.

Overall, this project indicates that nutrient availability plays a role in C. florida

survival from dogwood anthracnose. Our results also indicate that prescribed burning

offers a management technique to maintain C. florida as a component in eastern

hardwood forests. Additionally, our project showed the importance of C. florida in the Ca

cycle in eastern hardwood forests. Although this project took place in Great Smoky

Mountains National Park, because of the large study area and wide distribution of the

forest types we sampled, we believe that our findings are applicable in forests across the

eastern United States where C. florida occurs.















LIST OF REFERENCES


Allen, S. C., S. Jose, P. K. R. Nair, B. J. Brecke, V. D. Nair, D. A. Graetz, and C. L.
Ramsey. 2005. Nitrogen mineralization in a pecan (Carya illinoensis K. Koch)-
cotton (Gossypium hirsutum L.) alley cropping system in the southern United
States. Biology and Fertility of Soils 41:28-37.

Ament, M. M., R. M. Auge, L. F. Grand, and M. T. Windham. 1998. An inoculation
technique for dogwood anthracnose. Journal of Environmental Horticulture 16:37-
41.

Anagnostakis, S. L. 2001. The effect of multiple importations of pests and pathogens on a
native tree. Biological Invasions 3:245-254.

Anagnostakis, S. L., and J. S. Ward. 1996. The status of flowering dogwood in five long-
term forest plots in Connecticut. Plant Disease 80:1403-1405.

Anderson, R. L. 1991. Background, pages 5-9. In R. L. Anderson (ed.), Results of the
1990 dogwood anthracnose impact assessment and pilot test in the southeastern
United States. U.S. Forest Service Southern Region. Protection report R8-PR 20.

Anderson, R. L., J. L. Knighten, S. E. Dowsett, and C. Henson. 1991. Effectiveness of
cultural techniques to control dogwood anthracnose, pages 39-46. In R. L.
Anderson (ed.), Results of the 1990 dogwood anthracnose impact assessment and
pilot test in the southeastern United States. U.S. Forest Service Southern Region.
Protection report R8-PR 20.

Anglberger, H., and E. Halmschlager. 2003. The severity of Sirococcus shoot blight in
mature Norway spruce stands with regard to tree nutrition, topography and stand
age. Forest Ecology and Management 177:221-230.

Arthur, M. A., R. D. Paratley, and B. A. Blankenship. 1998. Single and repeated fires
affect survival and regeneration of woody and herbaceous species in an oak-pine
forest. Journal of the Torrey Botanical Society 125:225-236.

Augusto, L., M. P. Turpault, and J. Ranger. 2000. Impact of forest tree species on
feldspar weathering rates. Geoderma 96:215-237.

Biondini, M. E., C. D. Bonham, and E. F. Redente. 1985. Secondary successional
patterns in a sagebrush (Artemisia tridentata) community as they relate to soil
disturbance and soil biological activity. Vegetatio 60:25-36.









Blair, J. M. 1988. Nutrient release from decomposing foliar litter of three tree species
with special reference to calcium, magnesium, and potassium dynamics. Plant and
Soil 110:49-55.

Blair, R. M. 1982. Growth and nonstructural carbohydrate content of southern browse
species as influenced by light intensity. Journal of Range Management 35:756-760.

Blankenship, B. A., and M. A. Arthur. 2006. Stand structure over 9 years in burned and
fire-excluded oak stands on the Cumberland Plateau, Kentucky. Forest Ecology and
Management 225:134-145.

Boettcher, S. E., and P. J. Kalisz. 1990. Single-tree influence on soil properties in the
mountains of eastern Kentucky. Ecology 71:1365-1372.

Boerner, R. E. J., J. A. Brinkman, and E. K. Sutherland. 2004. Effects of fire at two
frequencies on nitrogen transformations and soil chemistry in a nitrogen-enriched
forest landscape. Canadian Journal of Forest Research 34:609-618.

Britton, K. 0. 1993. Anthracnose infection of dogwood seedlings exposed to natural
inoculum in western North Carolina. Plant Disease 77:34-37.

Britton, K. 0. 1994. Dogwood anthracnose, pages 17-20. In C. Ferguson and P. Bowman
(eds.), Threats to forest health in the southern Appalachians. Southern Appalachian
Man and the Biosphere Cooperative. Gatlinburg, TN.

Britton, K. O., W. D. Pepper, D. L. Loftis, and D. O. Chellemi. 1994. Effect of timber
harvest practices on populations of Cornusflorida and severity of dogwood
anthracnose in western North Carolina. Plant Disease 78:398-402.

Britton, K. O., P. Berrang, and E. Mavity. 1996. Effects of pretreatment with simulated
acid rain on the severity of dogwood anthracnose. Plant Disease 80:646-649.

Brose, P., T. Schuler, D. Van Lear, and J. Berst. 2001. Bringing fire back: the changing
regimes of the Appalachian mixed-oak forests. Journal of Forestry 99:30-35.

Buell, J. H. 1940. Effect of season of cutting on sprouting of dogwood. Journal of
Forestry 38:649-650.

Carr, D. E., and L. E. Banas. 1999. Dogwood anthracnose (Discula destructive): effects of
and consequences for host (Cornusflorida) demography. American Midland
Naturalist 143:169-177.

Chellemi, D. 0., and K. O. Britton. 1992. Influence of canopy microclimate on incidence
and severity of dogwood anthracnose. Canadian Journal of Botany 70:1093-1096.

Chellemi, D. O., K. O. Britton, and W. T. Swank. 1992. Influence of site factors on
dogwood anthracnose in the Natahala Mountain Range of western North Carolina.
Plant Disease 76:915-918.









Clarke, K. R., and M. Ainsworth. 1993. A method of linking multivariate community
structure to environmental variables. Marine Ecology Progress Series 46:213-226.

Clinton, B. D., J. A. Yeakley, and D. K. Apsley. 2003. Tree growth and mortality in a
southern Appalachian deciduous forest following extended wet and dry periods.
Castanea 68:189-200.

Conway, W. S., C. E. Sams, R. G. McGuire, and A. Kelman. 1992. Calcium treatment of
apples and potatoes to reduce postharvest decay. Plant Disease 76:329-334.

Cornwell, W. K., and P. J. Grubb. 2003. Regional and local patterns in plant species
richness with respect to resource availability. Oikos 100:417-428.

Daughter, M. L., and C. R. Hibben. 1994. Dogwood anthracnose: a new disease
threatens two native Cornus species. Annual Review of Phytopathology 32:61-73.

Daughtrey, M. L., C. R. Hibben, K. O. Britton, M. T. Windham, and S. C. Redlin. 1996.
Dogwood anthracnose: understanding a disease new to North America. Plant
Disease 80:349-357.

Day, F. P., and C. D. Monk. 1977. Seasonal nutrient dynamics in the vegetation on a
southern Appalachian watershed. Journal of Botany 64:1126-1139.

Dijkstra, F. A. and M. M. Smits. 2002. Tree species effects on calcium cycling: the role
of calcium uptake in deep soils. Ecosystems 5:385-398.

Dijkstra, F. A. 2003. Calcium mineralization in the forest floor and surface soil beneath
different tree species in the northeastern US. Forest Ecology and Management
175:185-194.

Driscoll, C. T., G. B. Lawrence, A. J. Bulger, T. J. Butler, C. S. Cronan, C. Eager, K. F.
Lambert, G. E. Likens, J. L. Stoddard, and K. C. Weathers. 2001. Acid deposition
in northeastern United States: sources and inputs, ecosystem effects, and
management strategies. BioScience 51:180-198.

Dufrene M., and P. Legendre. 1997. Species assemblages and indicator species: the need
for a flexible asymmetrical approach. Ecological Monographs 67:345-366.

Elliott, K. J., L. R. Boring, W. T. Swank, and B. R. Haines. 1997. Successional changes
in plant species diversity and composition after clearcutting in a southern
Appalachian watershed. Forest Ecology and Management 92:67-85.

Elliott, K. J., R. L. Hendrick, A. E. Major, J. M. Vose, and W. T. Swank. 1999.
Vegetation dynamics after a prescribed fire in the southern Appalachians. Forest
Ecology and Management 114:199-213.









Elliot, K. J., L. R. Boring, and W. T. Swank. 2002. Above ground biomass and nutrient
accumulation 20 years after clear-cutting a southern Appalachian watershed.
Canadian Journal of Forest Research 32:667-683.

Eno, C. F. 1960. Nitrate production in the field by incubating the soil in polyethylene
bags. Soil Science Society of America Proceedings 24:277-279.

Epstein, E. 1972. Mineral nutrition of plants: principles and perspectives. John Wiley and
Sons Inc., New York. 412 p.

Erbaugh, D. K., M. T. Windham, A. J. W. Stodola, and R. M. Auge. 1995. Light intensity
and drought stress as predisposition factors for dogwood anthracnose. Journal of
Environmental Horticulture 13:186-189.

Finzi, A. C., C. D. Canham, and N. van Breemen. 1998. Canopy tree-soil interactions
within temperate forests: species effects on pH and cations. Ecological
Applications 8:440-446.

Fujinuma, R., J. Bockheim, and N. Balster. 2005. Base cation cycling by individual tree
species in old-growth forests of Upper Michigan, USA. Biogeochemistry 74:357-
376.

Galbraith, S. L., and W. H. Martin. 2005. Three decades of overstory and species change
in a mixed mesophytic forest in eastern Kentucky. Castanea 70:115-128.

Godman, R. M., and K. Lancaster. 1990. Tsuga canadensis (L.) Carr.- Eastern hemlock,
pages 604-612. In R. M. Bums and B. H. Honkala (eds.), Silvics of North
America. Volume 1. Conifers. Agricultural Handbook 654. U.S. Forest Service,
Washington, D.C.

Gould, A. B., and J. L. Peterson. 1994. The effect of moisture and sunlight on the severity
of dogwood anthracnose in street trees. Journal of Arboriculture 20:75-78.

Graveland, J. R., 1996. Avain eggshell formation in calcium-rich and calcium-poor
habits: importance of snail shells and anthropogenic calcium sources. Canadian
Journal of Zoology 74:1035-1044.

Graveland, J. R., r. van der Wal, J. H. van Balen, and A.J. van Noordwijk. 1994. Poor
reproduction in passerines from decline of small abundance on acidified soils.
Nature 368:446-448.

Hamburg, S. P., R. D. Yanai, M. A. Arthur, J. D. Blum, and T. G. Siccama. 2003. Biotic
control of calcium cycling in northern hardwood forests: acid rain and aging
forests. Ecosystems 6:399-406.

Harmon, M. 1982. Fire history of the westernmost portion of Great Smoky Mountains
National Park. Bulletin of the Torrey Botany Club 109:74-79.