|UFDC Home||myUFDC Home | Help|
This item has the following downloads:
ECOLOGY OF FLOWERING DOGWOOD (Cornusflorida L.) IN RESPONSE TO
ANTHRACNOSE AND FIRE IN GREAT SMOKY MOUNTAINS NATIONAL PARK,
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
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
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
LIST OF TABLES
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
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,
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
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.
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
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
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 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
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
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
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.
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).
1-4.9 cm 5.0-9.9 cm 1-4.9 cm 5.0-9.9 cm
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
INFLUENCE OF FIRE ON THE DENSITY AND HEALTH OF Cornusflorida L.
(FLOWERING DOGWOOD) POPULATIONS IN GREAT SMOKY MOUNTAINS
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
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
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
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.
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 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.
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.
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
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
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.
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.
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.
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).
las (m) Single bum Double burn Triple bum Unburned P-value1
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
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.
564 (25) a
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.
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 from ANOVA
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
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
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
0.7 (0.3) a
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
5.1-10 cm class
P-value = 0.229
Double burn Triple burn
>10.1 cm class
P-value = 0.167
Single burn Double burn Triple burn Unburned Single burn
Double burn Triple burn
P-value = 0.0003
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.
E Triple l
-20 -1 0 0 0 0 20
Undersunderstory communities, showing the relative differences in community
SUnburdistinct compositional changes in overstory or understory with respect to burn
A AA SingleA
20 05 10 00 10'
15 A Single A A
1i 05 A AA ba A
~ yA Y
E Double burn
O Triple burn
0-2.5 2.6-5.0 5.1-10.0
Diameter class (cm)
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.
b t bb b b a
INFLUENCE OF CALCIUM, POTASSIUM, AND MAGNESIUM ON Cornusflorida
L. DENSITY AND RESISTANCE TO DOGWOOD ANTHRACNOSE
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
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).
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
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
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
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
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 <
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;
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;
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;
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).
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.
Table 3-1. Content of the base fertilizer mix.
Element Rate (%)
AgSul 90 6.2
Table 3-2. Inputs (%) added to the base fertilizer mix (separately) for each treatment.
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
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
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
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
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
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).
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
Corus florida density (stems ha-1)
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
S2 y = 0.0022x + 2.7002
R2 = 0.539
c, P < 0.001
0 500 1000 1500
Cornus florida density (stems ha1)
2 y = 2.2548x + 2.5135
SR2 = 0.2319
SP = 0.08
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.
,* y = 0.0043x + 3.7915
S R2 = 0.623
0 P < 0.0001
0 500 1000 1500
Corus florida density (stems ha-1)
p 3 y = 4.4378x + 3.5256
R2 = 0.4502
P = 0.008
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.
Calcium Treatments -
o#0 .0 ,- "
t 40 '"
0 O -50%
20 V -o 100%
Potassium Treatments .. ^_ / < .-- "
60 / -0%
20 -/ -- .100%
Magnesium Treatments .,,
t 40 -
20 -- o100%
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
Figure 3-5. Foliar calcium (Ca), potassium (K), and magnesium (Mg) concentrations of
Cornusflorida seedling foliage for the four treatment levels of each cation.
i 5' 2004 precipitation
0 Average precipitation
April May June July August September
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-
INFLUENCE OF Cornusflorida L. ON CALCIUM MINERALIZATION IN TWO
SOUTHERN APPALACHIAN FOREST TYPES
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
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).
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.
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.
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).
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).
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
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 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.
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
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,
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).
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.
Initial Final Initial Final
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)***
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)***
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)**
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
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)
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
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
ns b ns ns
8 Forest floor b 8 Forest floor b
None Low High None Low High
SCornus florida stem density
-i 4 4
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
Cornus florida stem density
b ns T
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.
Oak hardwood forest type
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.
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
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-
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
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
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
Carr, D. E., and L. E. Banas. 1999. Dogwood anthracnose (Discula destructive): effects of
and consequences for host (Cornusflorida) demography. American Midland
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.
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
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
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
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-
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,
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.
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.