Gap-scale disturbance processes in secondary hardwood
stands on the Cumberland Plateau, Tennessee, USA
Justin L. Hart Henri D. Grissino-Mayer
Received: 29 January 2008 / Accepted: 4 August 2008 / Published online: 23 August 2008
Springer Science+Business Media B.V. 2008
Abstract Disturbance regimes in many temperate, region. However, gap size was smaller in the
old growth forests are characterized by gap-scale developing stands, indicating that secondary forests
events. However, prior to a complex stage of contain a higher density of smaller gaps. The majority
development, canopy gaps may still serve as mech- of canopy gaps were projected to close by lateral
anisms for canopy tree replacement and stand crown expansion rather than height growth of
structural changes associated with older forests. We subcanopy individuals. However, canopy gaps still
investigated 40 canopy gaps in secondary hardwood provided a means for understory trees to recruit to
stands on the Cumberland Plateau in Tennessee to larger size classes. This process may allow over-
analyze gap-scale disturbance processes in develop- topped trees to reach intermediate positions, and
ing forests. Gap origin, age, land fraction, size, shape, eventually the canopy, after future disturbance
orientation, and gap maker characteristics were events. Over half of the trees located in true gaps
documented to investigate gap formation mechanisms with intermediate crown classi cations were Acer
and physical gap attributes. We also quanti ed saccharum, A. rubrum, or Liriodendron tulipifera.
density and diversity within gaps, gap closure, and Because the gaps were relatively small and close by
gap-phase replacement to examine the in uence of lateral branch growth of perimeter trees, the most
localized disturbances on forest development. The shade-tolerant A. saccharum has the greatest proba-
majority of canopy gaps were single-treefall events bility of becoming dominant in the canopy under the
caused by uprooted or snapped stems. The fraction of current disturbance regime. Half of the gap maker
the forest in canopy gaps was within the range trees removed from the canopy were Quercus;
reported from old growth remnants throughout the however, Acer species are the most probable replace-
ment trees. These data indicate that canopy gaps are
important drivers of forest change prior to a complex
stage of development. Even in relatively young
forests, gaps provide the mechanisms for stands to
develop a complex structure, and may be used to
J. L. Hart explain patterns of shifting species composition in
Department of Geography, University of North Alabama,
secondary forests of eastern North America.
Florence, AL 35632, USA
e-mail: abpnv4@r.postjobfree.com
Keywords Canopy gaps Disturbance
H. D. Grissino-Mayer
Forest development Mixed hardwoods
Department of Geography, The University of Tennessee,
Succession Tennessee
Knoxville, TN 37996, USA
123
132 Plant Ecol (2009) 201:131 146
1998; Yamamoto and Nishimura 1999). Because
Introduction
canopy gaps are generally larger in old growth
remnants, many of the gaps in these forests close by
All forest ecosystems are subject to natural distur-
bance events that shape species composition and the height growth of subcanopy individuals rather than
lateral crown expansion of perimeter trees (Runkle
stand structure. In many forest types, gap-scale
1982). This gap-replacement process creates forests
disturbance processes are the dominant disturbance
mechanisms. Thus, canopy gap characteristics and with complex age and size structures, and patchy
species composition in the canopy (Lorimer 1980;
forest response have been studied in forests through-
out eastern North America to elucidate patterns, and Runkle 1982; Yetter and Runkle 1986; Runkle and
processes of gap-scale disturbances and forest vege- Yetter 1987). Although canopy gaps in secondary
forests are hypothesized to be smaller in size, they may
tation dynamics. The overwhelming majority of
canopy gap studies, however, have been conducted still act as a mechanism for canopy tree replacement,
in old growth remnants (e.g., Lorimer 1980; Barden and stand structural changes associated with older
forests (Clebsch and Busing 1989; Wilder et al. 1999;
1981; Runkle 1982; Cho and Boerner 1991; Runkle
2000). Throughout the Eastern Deciduous Forest Taylor and Lorimer 2003; Cole and Lorimer 2005).
Region, most forested land supports secondary stands The overarching goal of our study was to docu-
(secondary referring to all non-primeval forests prior ment the in uence of localized, natural disturbance
to a complex stage of development) composed of events on the development of secondary hardwood
mixed hardwood species (Cowell 1998; Rebertus and stands during the understory reinitiation stage of
Meier 2001). Few studies have analyzed gap-scale development. Our research was driven by four major
disturbances and forest response in secondary forests questions. Question 1: What are the patterns and
(but see Clebsch and Busing 1989; Dahir and Lorimer processes of canopy gap formation prior to a complex
1996; Wilder et al. 1999; Yamamoto and Nishimura stage of forest development? We hypothesized that
1999), and no such research has been conducted in most canopy gaps would be created by uprooted
mixed hardwood stands on the Cumberland Plateau. stems, as windthrow has been widely reported from
Undoubtedly, forest disturbance dynamics differ many old growth stands and visual observation of the
between old growth remnants and mature secondary forest revealed uprooted trees. Question 2: What
stands. Differences in disturbance characteristics are percentage of the forest is occupied by canopy gaps
attributed to variations in species composition, bio- and what are the shape, size, and age distributions for
mass arrangement, and tree-age distribution. As gaps in developing stands? We hypothesized that the
forests mature, the distance between large individuals land fraction of the forest in gaps would be within the
increases. Tree crowns separate into distinct catego- range of variability reported from old growth stands,
ries, creating a more complex vertical structure, and but the forest would contain a higher density of
species composition shifts to favor later-successional smaller gaps relative to older stands. Question 3: Do
species (Goebel and Hix 1996; Oliver and Larson small canopy disturbances in uence density and
1996; Goebel and Hix 1997). Forest response to diversity patterns in secondary stands? We hypoth-
disturbance events likely differs between old growth esized that larger gaps would support a higher
and secondary stands, because of differences in stand number of individuals as well as higher levels of
structure and species composition, and also because diversity because they should contain more microsite
of the ages of the oldest trees, as older trees are less heterogeneity, and the likelihood of documenting rare
able to respond to increase in available resources species should increase by sampling a larger spatial
resulting from disturbance events (Fritts 2001). area. Question 4: How do the gaps close, and what
In old growth forests, the spacing between large effects do they have on composition and structure in
individuals is greater than in secondary forests. Thus, developing stands? We hypothesized that most gaps
when a canopy tree is removed from an old growth would close by lateral crown expansion rather than
stand, the size of the canopy gap created should be height growth of subcanopy individuals and would
larger than a comparable disturbance during earlier cause the structure of the forest to move from a high
stages of forest development (Clebsch and Busing density of small trees to a lower density of larger
1989; Spies et al. 1990; Tyrell and Crow 1994; Runkle individuals, more typical of older stands.
123
Plant Ecol (2009) 201:131 146 133
Methods acidic, highly leached, and low in fertility (Francis
and Loftus 1977; Smalley 1982; USDA 1995; Hart
Study area 2007). Depth to bedrock varies from 1 to 1.8 m and
slope gradients range from 15% to 60%. The
The study was conducted in the Pogue Creek Natural elevation of the study plots ranged from 260 to
Area (PCNA) located in Fentress County, Tennessee, 490 m amsl.
in the north-central portion of the state (Fig. 1). The Climate is classi ed as humid mesothermal with
PCNA is a 1,505 ha reserve managed by the State of moderately hot summers and short-mild to moder-
Tennessee, Department of Environment and Conser- ately cold winters (Thornthwaite 1948). Local
vation, Division of Natural Areas. The PCNA is topography strongly in uences microclimatic condi-
located on the Cumberland Plateau section of the tions. The average frost-free period is 160 days (from
Appalachian Plateaus physiographic province (Fenn- early-May to late-October) and the mean annual
temperature is 13 C. The July average is 23 C and
eman 1938). The underlying geology consists of
the January average is 2 C (USDA 1995). The area
Pennsylvanian sandstone, conglomerate, siltstone,
shale, and coal of the Crab Orchard and Crooked receives steady precipitation during the year with no
Forked Groups (Smalley 1986). The area has irreg- distinct dry season. Mean annual precipitation is
ular topography (Fenneman 1938) characterized by 137 cm and mean annual snowfall is 50 cm (USDA
long, narrow to moderately broad ridges and narrow 1995). Late spring and summer are characterized by
to moderately broad valleys (Smalley 1986). Soils are heavy rains that are often accompanied by moderate
Fig. 1 Map of the Pogue
Creek Natural Area,
Fentress County,
Tennessee. Shaded portion
of the Tennessee inset map
is the Cumberland Plateau
physiographic section
123
134 Plant Ecol (2009) 201:131 146
above) canopy gaps were documented by recording
to severe thunderstorms and strong winds (Smalley
the number of paces across each. The fraction of land
1982).
area in canopy gaps was calculated by dividing the
Braun (1950) classi ed the area as part of the Cliff
transect distance in gaps by total transect length
Section of the Mixed Mesophytic Forest Region, but
(Runkle 1985; Runkle 1992). At each gap, physical
local topography in uences forest composition and
site characteristics, including percent slope, aspect,
true mesophytic species only dominate on protected
and elevation, were recorded . When walking tran-
sites. Regionally, forests are intermediate between
sects through a forest, large gaps are more likely to be
mixed mesophytic and Quercus Carya types (Hinkle
encountered than relatively small gaps, and sampling
1978; Hinkle 1989; Hinkle et al. 1993). Forest
estimators have been created to correct for sampling
vegetation patterns of the PCNA were quanti ed by
bias (see De Vries 1974; Pickford and Hazard 1978).
Hart and Grissino-Mayer (2008). The forest was
However, values obtained with the use of estimator
dominated by Carya ovata, Quercus rubra, Q. alba,
and Q. montana. The sparse sapling layer was equations and those obtained by simply dividing
dominated by Acer saccharum. The forest was transect distance in gaps by total transect length are
established in the late 1920s after the cessation of similar (Runkle 1985).
local logging operations. From eld observations and Gap area was determined for expanded and true
investigation of 17 tree cross sections from a previous gaps by, measuring length (largest distance from gap
study, no signs of re or other large-scale disturbance edge to gap edge) and width (largest distance
events were evident since the anthropogenic distur- perpendicular to the length). These measurements
bances of the 1920s (Hart 2007). Castanea dentata were tted to the formula of an ellipse (Runkle 1985;
Marsh was a forest component prior to the arrival of Runkle 1992). Although gap shapes can be highly
Cryphonectria parasitica (Murrill) M.E. Barr (chest- variable (Ferreira de Lima 2005), most gaps at the
nut blight). The blight reached the Cumberland PCNA had elliptical shapes, which is common for
Plateau in the 1920s, and by the end of the 1930s, forests of the southern Appalachian Highlands (Run-
most C. dentata in the region were dead. Thus, the loss kle 1982; Runkle 1992; Clinton et al. 1994). Thus,
of the species roughly coincided with stand initiation. tting the measurements to the formula of an ellipse
was appropriate for this study.
Field sampling Canopy gaps can be created by several different
mechanisms that remove overstory trees. Biotic and
Canopy gaps (n = 40) were located along transects abiotic forest conditions can be modi ed differently
throughout the reserve using the line intersect method by canopy disturbances that are caused by different
gap formation mechanisms. Differences between gap
(Runkle 1982; Runkle 1985; Veblen 1985; Runkle
1992). Gaps were de ned as environments where a origins may also in uence forest response. In order to
visible void space existed in the main forest canopy, analyze these patterns, gap formation mechanisms
were classi ed into one of the three categories (snag,
leaf height of the tallest stems was less than three-
fourths the height of the adjacent canopy, and gap uprooted stem, or snapped stem) according to gap
makers were present. We did not use a minimum gap origin (Clinton et al. 1993). The number of trees
involved in gap formation was also recorded to
size threshold to document the full range of canopy
gaps. Transects were established parallel to slope document the abundance of single-tree versus multi-
contour beginning at randomly selected points tree events.
Gap maker trees were taxonomically classi ed to
throughout the forest. All transects were located
along mid-slope positions. We sampled at mid-slope quantify any species-speci c overstory mortality
patterns and possible composition changes associated
positions, because the mid-slope forests of the reserve
are indicative of slope forests of the greater Cum- with small canopy disturbances. We measured gap
berland Plateau region and the majority of forested maker diameter at breast height (dbh, ca. 1.4 m above
the surface or root collar for downed individuals) and
land in the reserve occurs along mid-slopes. Total
length. Basal area (m2) was calculated for all gap
transect length and transect length in expanded
(boundary de ned by the base of surrounding canopy makers that could be accurately measured and totaled
by gap, to determine the amount of basal area lost per
trees (Runkle 1981)) and true (area unrestricted from
123
Plant Ecol (2009) 201:131 146 135
gap maker death dates, tree rings were measured to
disturbance event. This information may be used to
the nearest 0.001 mm using a Velmex measuring
document the amount of biomass naturally removed
stage interfaced with Measure J2X software for all
from a stand through gap-scale processes. Direction
sampled gap makers. The measurement series were
of gap maker fall relative to slope (i.e., down, across,
visually compared to a reference Quercus chronology
or up slope) was also recorded and all gap makers
developed by Hart and Grissino-Mayer (2008) for the
were placed into one of four decay classes (1 4, with
site. We con rmed the graphical crossdating of all
4 being the most decayed) following criteria adapted
gap maker tree-ring series using the computer
from McCarthy and Bailey (1994).
software COFECHA, a quality-control program that
In each gap, we recorded species, crown class, and
uses segmented time series correlation analyses to
diameter of all trees C5 cm dbh to characterize forest
con rm the placements of all tree rings (Holmes
gap vegetation. Crown class categories (dominant,
1983; Grissino-Mayer 2001). In COFECHA, we
codominant, intermediate, and overtopped) were
tested consecutive 50-year segments (with 25-year
visually assessed based on the amount and direction
overlaps) on each undated gap maker series to the
of intercepted light (Oliver and Larson 1996). The
reference Quercus chronology. Once statistically
location of each of these individuals was also
con rmed, we assigned calendar years to all tree
recorded as being in either an expanded or true
rings in each individual undated measurement series.
canopy gap. All saplings (woody stems C1 m height,
\5 cm dbh) in the expanded gap area were tallied by All gap ages were con rmed using gap maker decay
classi cations.
species to characterize gap regeneration patterns. The
Canopy gaps can be caused by the removal of a
number of perimeter trees with dominant or codom-
single tree or a small cluster of trees. Because single-
inant positions in the canopy was documented for
tree gaps may result from the death of a large canopy
each gap, to analyze the number of trees required to
tree and multi-tree gaps may result from the deaths of
complete the canopy surrounding gaps, and the
relatively small trees, the amount of basal area lost
number of canopy individuals with the potential to
between single- and multi-tree gaps was statistically
close the void space through lateral crown expansion.
analyzed using a two-tailed t-test. This information
Tree core samples were collected to aid in the
may be useful to analyze the quantity of basal area
documentation of gap age. A minimum of nine trees
lost by small canopy disturbance events and applied
were cored (mean = 18.6) per gap resulting in the
to harvesting techniques that may mimic natural
collection of 742 cores. Tree core samples or cross
disturbance processes.
sections were also collected from all gap makers that
The rate of gap formation and closure may be
were not in an advanced state of decay (intact bark
balanced or may vary through time. Non-parametric
and no sapwood degradation), to aid with gap age
correlation techniques were used to analyze the
determination and to document the seasonal timing of
relationship between land fraction in gaps and gap
gap events, based on the amount of xylem produced
age. Gaps may be caused by a variety of formation
during the last year of growth. Dating the seasonality
mechanisms that differ in the way overstory vegeta-
of tree death and gap formation illustrates a new
tion is removed, and the mechanism of canopy tree
approach in dendroecology.
removal may in uence gap size. In order to deter-
mine, if a relationship existed between gap size and
Data analyses
gap origin, data were analyzed using a one-way
ANOVA. A Tukey honestly signi cant difference
Tree core and cross section samples were prepared
(HSD) test was used to compare mean expanded and
and processed for dating using the methods outlined
true gap sizes across origin categories to determine if
in Stokes and Smiley (1996). The samples were air-
gap size varied by gap formation mechanism.
dried, glued to wooden mounts, and sanded to reveal
Length and width of gaps were measured in the
the cellular structure of the wood (Orvis and
eld. Ratios were calculated for length to width
Grissino-Mayer 2002) before tree rings were dated
(L:W) of expanded and true gaps to document gap
with the aid of a stereo zoom microscope. All tree
shape characteristics. This information is useful to
cores were visually analyzed for radial growth
understand the variation in the shape of gaps created
releases to document gap age. In order to document
123
136 Plant Ecol (2009) 201:131 146
subsequently blown down, was classi ed incorrectly.
by the disturbance and has implications for forest
However, measures were taken to avoid this issue,
response and microenvironmental changes within the
such as documenting the decay class of gap makers
gap environment.
For each gap, density and diversity (H0 ) measures and noting the position of the gap maker relative to
other downed logs. The number of gap maker trees
were calculated for saplings, trees, and total stems
involved with opening the canopy ranged from one to
(all woody stems C1 m height) to document forest
four. The majority (78%) of the canopy gaps involved
response to canopy disturbances. Gap size is believed
the death of only one individual. Of the nine multi-
to in uence stem density and diversity. Correlation
tree gaps, six (66%) resulted from uprooted stems
coef cients were calculated to determine if a rela-
including the gap that consisted of the removal of
tionship existed between gap size and density of
four canopy individuals, while the three other multi-
individuals in gaps. Regression techniques were then
tree gaps resulted from snapped boles.
used to model gap size and density relationships. In
We identi ed 50 gap maker trees in the 40 canopy
order to analyze the relationship between expanded
gaps studied. Most gap makers (n = 36, 72%) could
gap area and diversity patterns, correlation coef -
be identi ed to the species level; however, 4 (8%)
cients were calculated for sapling, tree, and total stem
could only be identi ed to genus and 10 (20%) were
diversity.
too decayed to be taxonomically classi ed. Of the 36
Canopy gaps can close by crown expansion of
gap makers that could be identi ed to species, 12
perimeter trees at canopy level or by the height
different species were represented. The most com-
growth of understory individuals. The likely closure
mon species that caused canopy gap formation was
mechanism, either by height growth or lateral crown
Quercus montana (n = 8). At the genus level, 50% of
expansion, of each gap was recorded in the eld to
all gap makers were Quercus.
document changes in forest structure following the
Diameter was measured at ca. 1.4 m above the
removal of canopy trees. Probable gap successors,
surface or root collar for 46 gap makers. Diameter
which are the individuals that will likely ll the
measurements could not be collected for four gap
canopy void, can often be determined in the eld
makers that were in a state of advanced decay.
(Barden 1979; Barden 1980; White et al. 1985;
Average gap maker diameter at breast height was
Yamamoto and Nishimura 1999). The documentation
38.38 cm 11.6 (SD). The minimum diameter of a
of replacement trees is useful to project the future
gapmaker was 19.5 cm and the maximum was 70 cm.
composition of the stand and to analyze the in uence
The gap maker with a diameter of 19.5 cm was
of canopy gaps on forest succession. In order to
involved in a multi-tree uprooting event that also
quantify recruitment following overstory removal,
included the death of an individual with a diameter of
crown class distributions were constructed for all
28 cm. Average basal area lost per gap was
trees located in true gap environments for the 15 most
0.16 m2 0.10 (SD). The minimum removed was
dominant species with canopy potential. These mea-
0.05 m2 and the maximum was 0.52 m2. Multi-tree
sures may be used to document future canopy trees
gaps (mean = 0.24 m2 0.13 (SD)) resulted in a
and recruitment patterns associated with gap-scale
larger amount (P \ 0.01) of basal area lost compared
disturbance processes.
to single-tree events (mean = 0.14 m2 0.08 (SD)).
Age was determined for all canopy gaps by the
identi cation of radial growth releases, crossdating
Results
the gap makers to document death dates, eld
Gap formation patterns and processes observation, and gap maker decay classi cation.
Gap ages ranged from 1 to 17 years with a mean of
Of the 40 gaps sampled, 8 (20%) were created by 7 years. Multiple gaps occurred in 13 years. The
snags, 16 (40%) were created by uprooted stems, and highest frequency of gap events during any one year
16 (40%) were created by snapped stems. Eventually, was ve, which occurred during 3 years (1999, 2002,
snag trees will fall, generally during mild to severe and 2003).
wind events, possibly causing further disturbance to Gap seasonality was determined for 17 gaps by
the forest. It is possible that a gap created by a snag, examining the amount of xylem produced during the
123
Plant Ecol (2009) 201:131 146 137
last year of growth. Other gapmakers were too
decayed for this analysis. Of these 17 events, only
one occurred during the dormant season. For the
dormant season gap, the latewood portion of the last
ring was complete and buds were still present on the
tree. All other gap makers had partial rings, indicat-
ing that the gap events occurred during the growing
season. Because the majority of these individuals had
already completed the production of earlywood prior
to death, we surmise that these events occurred in the
middle or later part of the growing season.
Gap fraction and physical characteristics
Total transect length was 4.47 km, with 15% of the
total length in expanded gaps and true gaps, and 6% in
true canopy gaps only. When percentage values were
standardized at the hectare level, 1,500 m2/ha were in
expanded gaps and 600 m2/ha were in true gap Fig. 3 Mean sizes of expanded and true canopy gaps by gap
origin with standard deviations. Solid bar and different letter
environments. Total transect length in true canopy
indicate a signi cant (P \ 0.05) difference between gap
gaps was plotted by gap age to analyze patterns of gap
origins as detected by ANOVA and Tukey s post-hoc testing
formation and closure (Fig. 2). The largest amount of
The average L:W ratio of expanded gaps was
land area in true canopy gaps occurred in gaps that
1.58:1, with a maximum of 3.60:1 and a minimum of
were 2 years of age and no gap area occurred in gaps
1.01:1. Thus, the average expanded gap was 58%
aged 5, 6, 14, 15, or 16 years. A signi cant negative
longer than it was wide. Similar patterns were
relationship existed, where older gaps occupied a
observed for true gap areas, for which the mean ratio
smaller amount of land area relative to younger gaps.
Average expanded gap area was 213.34 m2 was 2.58:1. The maximum length of true gaps was
475% the width. The minimum L:W patterns of
108.44 (SD). The maximum expanded gap area was
587.91 m2 and the minimum was 47.10 m2. Average expanded and true gaps revealed circular over
true gap area when sampled was 42.78 m2 40.16 ellipsoidal shapes.
(SD), with a maximum of 157.84 m2 and a minimum of
1.14 m2. The size of expanded gaps created by Density and diversity within gaps
uprooted stems was signi cantly larger than that of
The mean number of canopy trees that bordered gaps
gaps created by snags (Fig. 3). No other size differ-
was 6.38 1.79 (SD). The maximum number of
ences between gap origins were signi cant.
perimeter trees was 12, and minimum number of trees
required to complete the canopy around a gap was 4.
In general, larger canopy gaps were bordered by a
higher number of canopy trees relative to smaller
gaps.
The average number of saplings in expanded gaps
was 54.48 28.47 (SD) with a maximum of 144 and
a minimum of 13 (Fig. 4). The mean number of trees
in expanded gaps was 22.73 7.99 (SD) with a
maximum of 44 and minimum of 11 individuals. The
average number of all stems C1 m height in
expanded gaps was 74.20 34.14 (SD). The highest
Fig. 2 Relationship between land fraction in true canopy gaps
number of stems in an expanded gap was 188 and the
and gap age in the Pogue Creek Natural Area in Tennessee
123
138 Plant Ecol (2009) 201:131 146
Table 1 Density of saplings (C1 m height, \5 cm dbh) in
expanded canopy gaps in the Pogue Creek Natural Area in
Tennessee
Species Density/ Relative
ha density
Acer saccharum Marsh. 863.63 35.70
Fagus grandifolia Ehrh. 474.70 19.62
Acer rubrum L. 327.83 13.55
Asimina triloba (L.) Dunal 168.03 6.95
Magnolia acuminata (L.) L. 158.63 6.56
Fraxinus americana L. 88.13 3.64
Fig. 4 Mean number of saplings (C1 m height, \5 cm dbh),
trees (C5 cm dbh), and total stems (all stems C1 m height) Liriodendron tulipifera L. 49.35 2.04
with standard deviations in expanded canopy gaps in the Pogue Oxydendrum arboreum (L.) DC. 48.18 1.99
Creek Natural Area in Tennessee
Cornus orida L. 37.60 1.55
Ulmus rubra Muhl. 31.73 1.31
lowest number of individuals was 28. The highest
Nyssa sylvatica Marsh. 30.55 1.26
values for saplings and trees occurred in the same gap
Cercis canadensis L. 29.38 1.21
that was 10 years old and caused by the uprooting of
Tilia heterophylla Vent. 12.93 0.53
four trees.
Aesculus ava Ait. 11.75 0.49
The sum of all saplings in all expanded gaps was
Carpinus caroliniana Walt. 8.23 0.34
calculated by species and standardized at the hectare
Ilex opaca Ait. 8.23 0.34
level to document sapling establishment, and possible
Magnolia tripetala L. 8.23 0.34
species recruitment in gap environments. The most
Quercus montana Willd. 8.23 0.34
abundant species in the sapling layer of expanded
Carya ovata (P. Mill.) K. Koch 5.88 0.24
gaps was Acer saccharum followed by Fagus gran-
Ostrya virginiana (P. Mill.) K. Koch 5.88 0.24
difolia and Acer rubrum (Table 1). Together these
Sassafras albidum (Nutt.) Nees 5.88 0.24
three species comprised almost 69% of all saplings in
Ailanthus altissima (Mill.) Swingle 4.70 0.19
expanded gaps.
Betula lenta L. 4.70 0.19
Acer saccharum represented 29.18% of all trees in
Diospyros virginiana L. 4.70 0.19
true canopy gaps followed by A. rubrum and
Liriodendron tulipifera (Table 2). Collectively, these Quercus alba L. 4.70 0.19
three species represent over half of all trees in true Ulmus alata Michx. 3.53 0.15
canopy gaps. Dominance (based on basal area) was Amelanchier laevis Weig. 2.35 0.10
also calculated for all canopy gap trees. The most Carya tomentosa (Poiret) Nutt. 2.35 0.10
dominant species were A. saccharum and A. rubrum Quercus rubra L. 2.35 0.10
(Table 2). The Acer species were followed by a Ulmus americana L. 2.35 0.10
second tier of species that included L. tulipifera and Hamamelis virginiana L. 1.18 0.05
Carya ovata. No other species represented more than Magnolia macrophylla Michx. 1.18 0.05
6% of the basal area. Species and diameter of all Morus rubra L. 1.18 0.05
snags in true canopy gaps were also recorded. A total Quercus velutina Lam. 1.18 0.05
of 40 snags were documented and mean snag Total 2419.33 100.00
diameter at breast height was 10.89 cm 6.21
(SD). Of the 40 snags within true gaps, 12 different
species were represented with A. rubrum, A. saccha- sapling layer diversity was 2.22 and the minimum
rum and Q. montana being the most common (n = 8 was 0.78. Total species richness of the tree layer was
for all species). 28. Average diversity of all trees in expanded gaps
Expanded canopy gaps contained 34 different was 1.90 0.35 (SD) with maximum and minimum
species in the sapling layer. Mean sapling diversity values of 2.44 and 1.20, respectively. Mean total
(H0 ) was 1.43 0.42 (SD) (Fig. 5). Maximum diversity of all stems C1 m height was 1.95 0.36
123
Plant Ecol (2009) 201:131 146 139
Table 2 Density and
Species Density/ Relative Dominance Relative
dominance measures for all
(m2/ha)
ha density dominance
trees (stems C5 cm dbh) in
true canopy gaps in the Acer saccharum 807.30 29.18 0.59 24.34
Pogue Creek Natural Area
Acer rubrum 391.95 14.16 0.32 13.15
in Tennessee
Liriodendron tulipifera 251.55 9.09 0.23 9.37
Carya ovata 146.25 5.29 0.21 8.68
Oxydendrum arboreum 146.25 5.29 0.14 5.91
Fagus grandifolia 175.50 6.34 0.13 5.48
Tilia heterophylla 122.85 4.44 0.10 4.15
Carya tomentosa 81.90 2.96 0.09 3.58
Carya glabra (P. Mill.) Sweet 64.35 2.33 0.08 3.38
Nyssa sylvatica 99.45 3.59 0.08 3.25
Fraxinus americana 70.20 2.54 0.07 2.95
Quercus alba 5.85 0.21 0.07 2.85
Cornus orida 111.15 4.02 0.06 2.59
Quercus montana 40.95 1.48 0.05 1.94
Magnolia acuminata 52.65 1.90 0.03 1.44
Quercus rubra 23.40 0.85 0.03 1.40
Ulmus rubra 40.95 1.48 0.03 1.16
Cercis canadensis 35.10 1.27 0.02 0.89
Carya cordiformis (Wangenh.) K. Koch 17.55 0.63 0.02 0.67
Ostrya virginiana 23.40 0.85 0.02 0.66
Diospyros virginiana 11.70 0.42 0.01 0.51
Sassafras albidum 11.70 0.42 0.01 0.49
Prunus serotina Ehrh. 5.85 0.21 0.01 0.31
Aesculus ava 5.85 0.21 0.01 0.25
Ulmus alata 5.85 0.21 0.00 0.21
Betula lenta 5.85 0.21 0.00 0.17
Magnolia tripetala 5.85 0.21 0.00 0.13
Ulmus americana 5.85 0.21 0.00 0.10
Total 2767.05 100.00 2.41 100.00
(SD). The highest total diversity value was 2.46 and
the lowest was 1.17. Interestingly, diversity patterns
differed by category (i.e., sapling, tree, and total). For
example, the gap with the lowest sapling diversity
was not the same gap with the lowest tree diversity.
However, the gap with the highest sapling and
highest total woody stem diversity values was an
exception.
Signi cant positive relationships were found for
the number of saplings (r = 0.54, P = 0.0003), trees
Fig. 5 Mean diversity for saplings (C1 m height, \5 cm dbh),
(r = 0.73, P \ 0.0001), and total stems (r = 0.62,
trees (C5 cm dbh), and total stems (all stems C1 m height)
P \ 0.0001) (Fig. 6). However, the largest gap did
with standard deviations in expanded canopy gaps in the Pogue
not contain the highest number of stems, which
Creek Natural Area in Tennessee
123
140 Plant Ecol (2009) 201:131 146
Fig. 7 Relationships between diversity values for saplings
Fig. 6 Relationships between the number of saplings (C1 m
(C1 m height, \5 cm dbh), trees (C5 cm dbh), and total stems
height, \5 cm dbh), trees (C5 cm dbh), and total stems (all
(all stems C1 m height) and expanded gap area in the Pogue
stems C1 m height) and expanded gap area in the Pogue Creek
Creek Natural Area in Tennessee
Natural Area in Tennessee
expanded area for all 40 gaps (213.34 m2). The gap
occurred in a gap of an intermediate size class (188
with the largest expanded area (587.91 m2) was
individuals/231.97 m2). A weak negative relationship
projected to close by understory height growth.
existed between sapling diversity and gap size
However, a relatively small gap (153.59 m2) was also
(r = -0.33, P = 0.04) (Fig. 7). A similar pattern
projected to close by height growth of a subcanopy
was also observed for total stem diversity (r = -
individual.
0.39, P = 0.01). Tree diversity showed no relation-
ship to expanded gap size. Shannon diversity (H0 ) is a Of the 10 successor trees documented, ve species
were represented (A. saccharum, A. rubrum, C. ovata,
dimensionless index such that gap size would not bias
Q. montana, and Quercus alba). Acer rubrum was the
the diversity measure.
most common gap successor (n = 3) followed by A.
saccharum (n = 2), C. ovata (n = 2), Q. montana
Gap closure and recruitment
(n = 2), and Q. alba (n = 1). Acer saccharum
represented 28.7% of trees with intermediate posi-
Of the 40 gaps studied, 10 were projected to close by
tions of all 15 selected species within true gap
height growth of understory individuals and the
environments (Table 3). Acer saccharum was fol-
remaining 30 gaps were projected to close by lateral
lowed by A. rubrum (13.45%) and L. tulipifera
branch growth of canopy trees surrounding the voids.
(13.45%), a noted gap-phase species. Collectively,
Mean expanded area of gaps likely to close via the
height growth of understory trees was 285.13 m2 these three species represented 55.6% of the inter-
mediate trees from the 15 selected species. A similar
137.58 (SD), which was ca. 34% greater than the mean
123
Plant Ecol (2009) 201:131 146 141
blowing down snag trees. Standing dead trees are
Table 3 Crown class distributions for all trees (stems C5 cm
dbh) in 40 true canopy gaps in the Pogue Creek Natural Area in often removed by mild to severe wind events, but the
Tennessee
potential for snags to be blown down varies by site
conditions (Jans et al. 1993). Further, snags that
Species Overtopped Intermediate
eventually fall likely alter the forest differently than
Density Relative Density Relative
gaps that are caused rapidly (Franklin et al. 1987;
density density
Krasny and Whitmore 1992; Clinton et al. 1994). The
Acer saccharum 72 36.36 64 28.70
eventual fall of a snag may cause additional forest
Acer rubrum 37 18.69 30 13.45
disturbance, possibly with a greater magnitude than
Liriodendron tulipifera 13 6.57 30 13.45
the initial event. Also, the bole and branches of
Carya ovata 4 2.02 21 9.42 standing dead trees may block sunlight from reaching
Tilia heterophylla 11 5.56 10 4.48 the understory, thereby, facilitating gap closure by
Oxydendrum arboreum 16 8.08 9 4.04 perimeter trees rather than subcanopy individuals.
Fraxinus americana 3 1.52 9 4.04 The percentage of single-tree gaps (78% of gaps
Carya glabra 1 0.51 9 4.04 sampled) was within the range of what has been
Fagus grandifolia 21 10.61 8 3.59 reported from old growth forests of the eastern USA
Carya tomentosa 5 2.53 8 3.59 (Runkle 1990). Of the multi-tree disturbance events,
Quercus alba 1 0.51 8 3.59 most were caused by uprooted stems. Windthrow
Quercus montana 1 0.51 6 2.69 gaps have the potential to cause more site modi ca-
Nyssa sylvatica 12 6.06 5 2.24 tion than gaps caused by other mechanisms, because
Quercus rubra 0 0.00 4 1.79 as the root network is lifted, microtopography (pits
and mounds) and soil characteristics are also modi-
Carya cordiformis 1 0.51 2 0.90
ed (Clinton et al. 1994; Beckage et al. 2000).
Total 198 100.00 223 100.00
Average diameter of gap maker trees was
38.38 cm at breast height and the average diameter
pattern was observed for overtopped positions, with of canopy trees (dominant and codominant crown
A. saccharum being the most abundant (36.36%) classes) that surrounded gaps was 38.83 cm 6.04
followed by A. rubrum (18.69%) and F. grandifolia (SD). This result is contrary to what has been
(10.61%). reported for old growth forests of the southern
Appalachians, where gap makers were signi cantly
larger than border trees (Runkle 1998). This pattern
may be related to the age of the forest. In second
Discussion
growth forests, canopy trees are within a narrower
Gap formation patterns and processes diameter range as their age (and diameter) structure is
not complex. Thus, in mature second growth forests,
The majority (80%) of the gaps documented origi- size does not indicate that one individual is more
nated from uprooted or snapped stems. Other studies likely to be removed from the canopy than another.
have also found these mechanisms to be the most