ArticlePDF Available

Scale-dependent responses of longleaf pine vegetation to fire frequency and environmental context across two decades

Authors:

Abstract and Figures

1.Disturbance is an important driver of plant community structure in many grasslands and woodlands, and alteration of disturbance regimes can have large consequences for species richness and composition. However, the response of vegetation to disturbance may change with environmental context. We re-sampled a unique, nested permanent vegetation plot data set in the longleaf pine ecosystem of the southeastern USA after 20 years to determine how environmental context and fire frequency jointly influence vegetation change across multiple spatial scales (0.01 to 1000 m²).2.The magnitude of vegetation change was quantified using two different, yet complementary metrics of beta-diversity (beta turnover measured as the proportion of species turning over and Bray-Curtis dissimilarity) and by documenting changes in species richness. We used null model analysis to explore whether communities were more dynamic over time at small spatial scales relative to larger scales.3.Changes in species richness, beta turnover, and Bray-Curtis dissimilarity were greatest on silty, frequently-burned sites, whereas sandy, less frequently-burned sites remained relatively stable. The amount of change detected was scale-dependent: species richness increased at larger spatial scales over time, but decreased at the two smallest spatial scales. Null model analysis revealed that beta turnover standardized effect sizes (SES) were negative and significantly different from random expectation at all spatial scales except the smallest. Thus, the magnitude of compositional change across most scales was small, despite substantial changes in species richness across time. We attribute this initial contradiction to the turnover of infrequent, low-abundance species amidst a matrix of dominant grasses.4.Synthesis. In contrast to previous longleaf pine studies, we found fire frequency to be less important than environmental site conditions in predicting vegetation change. Thus, future work in this ecosystem and in other fire-dependent grasslands and woodlands should consider not only disturbance, but also environmental context. Since species richness and beta-diversity patterns were scale-dependent, we recommend sampling vegetation across multiple spatial scales in order to comprehensively quantify changes in community structure over time. We believe this study lays the groundwork for understanding how fire and environmental filtering jointly influence vegetation dynamics across space and time in fire-dependent grasslands and woodlands.This article is protected by copyright. All rights reserved.
Content may be subject to copyright.
Scale-dependent responses of longleaf pine vegetation
to re frequency and environmental context across
two decades
Kyle A. Palmquist
1
*, Robert K. Peet
2
and Stephen R. Mitchell
3
1
Department of Botany University of Wyoming, Laramie, WY 82071, USA;
2
Department of Biology University of North
Carolina at Chapel Hill, Chapel Hill, NC 27599, USA; and
3
Nicholas School of the Environment Duke University,
Durham, NC 27708, USA
Summary
1. Disturbance is an important driver of plant community structure in many grasslands and wood-
lands, and alteration of disturbance regimes can have large consequences for species richness and
composition. However, the response of vegetation to disturbance may change with environmental
context. We resampled a unique, nested permanent vegetation plot data set in the longleaf pine eco-
system of the southeastern USA after 20 years to determine how environmental context and re fre-
quency jointly inuence vegetation change across multiple spatial scales (0.011000 m
2
).
2. The magnitude of vegetation change was quantied using two different, yet complementary metrics
of beta-diversity (beta turnover measured as the proportion of species turning over and BrayCurtis dis-
similarity) and by documenting changes in species richness. We used null model analysis to explore
whether communities were more dynamic over time at small spatial scales relative to larger scales.
3. Changes in species richness, beta turnover and BrayCurtis dissimilarity were greatest on silty,
frequently burned sites, whereas sandy, less frequently burned sites remained relatively stable. The
amount of change detected was scale dependent: species richness increased at larger spatial scales
over time, but decreased at the two smallest spatial scales. Null model analysis revealed that beta
turnover standardized effect sizes (SES) were negative and signicantly different from random
expectation at all spatial scales except the smallest. Thus, the magnitude of compositional change
across most scales was small, despite substantial changes in species richness across time. We attri-
bute this initial contradiction to the turnover of infrequent, low-abundance species amidst a matrix
of dominant grasses.
4. Synthesis. In contrast to previous longleaf pine studies, we found re frequency to be less impor-
tant than environmental site conditions in predicting vegetation change. Thus, future work in this
ecosystem and in other re-dependent grasslands and woodlands should consider not only distur-
bance, but also environmental context. Since species richness and beta-diversity patterns were scale
dependent, we recommend sampling vegetation across multiple spatial scales in order to comprehen-
sively quantify changes in community structure over time. We believe this study lays the ground-
work for understanding how re and environmental ltering jointly inuence vegetation dynamics
across space and time in re-dependent grasslands and woodlands.
Key-words: beta-diversity, disturbance, environmental gradient, plant population and community
dynamics, savanna, spatial grain, species richness, vegetation change
Introduction
Fire shapes plant community structure in many terrestrial eco-
systems (Bowman et al. 2009; Turner 2010). In re-dependent
ecosystems, particularly grasslands and woodlands, re often
increases species richness through the release of resources
(e.g. nutrients and light), while reducing competitive exclu-
sion (Kirkman et al. 2004; Overbeck et al. 2005; Smith et al.
2013). However, re regimes are changing throughout the
world (Westerling et al. 2006; Bowman et al. 2009),
with potentially large and lasting consequences for species
*Correspondence author: E-mail: kpalmqu1@uwyo.edu
©2015 The Authors. Journal of Ecology ©2015 British Ecological Society
Journal of Ecology 2015, 103, 9981008 doi: 10.1111/1365-2745.12412
distributions, species richness patterns and community compo-
sition, especially where re is necessary for the maintenance
of community structure (Leach & Givnish 1996; Johnstone &
Chapin 2003). Here, we dene altered re regimes to repre-
sent any changes in re attributes, whether re exclusion or
increased re frequency. Since the effects of altered re
regimes may change over time, with spatial scale or with
environmental context, it is critical to quantify those impacts
over longer temporal extents (i.e. decades), across multiple
spatial scales and along environmental gradients.
Vegetation change may proceed differently in response to
altered re regimes depending on the spatial scale in question.
For example, re exclusion may initially result in reduced
small-scale species density due to the loss of individuals, with
no detectable effects at larger spatial scales (Palmquist, Peet
& Weakley 2014). However, given continued re exclusion,
the loss of species may accelerate at larger scales due to
extinction caused by smaller population sizes. There have
been relatively few opportunities to explore spatiotemporal
vegetation change because of a paucity of permanently
marked, nested vegetation plots that have been resampled
multiple times over the course of decades (but see Lep
s 2014;
Palmquist, Peet & Weakley 2014). This is despite the fact
that ecologists have developed a strong theoretical framework
for understanding patterns of species richness and turnover
across space and time (e.g. Preston 1960; Adler & Lauenroth
2003; Soininen 2010; White et al. 2010).
In addition, vegetation change over time may be strongly
dependent on environmental context. Sites that differ in key
environmental attributes (e.g. soil moisture) may have differ-
ent rates of change because of underlying variation in
resource availability (Gibson & Hulbert 1987; Gauthier et al.
2010; Amatangelo et al. 2011). Furthermore, the effects of
re and other disturbances can change with environmental
context, which has direct consequences for plant species rich-
ness and composition (Kirkman et al. 2001; Harrison, Inouye
& Safford 2003; Pausas & Ribeiro 2013; Smith et al. 2013;
Dantas et al. 2015).
Here, we assess how prescribed re frequency and environ-
mental context (e.g. soil properties, community type) have
affected patterns of species richness and community composi-
tion in longleaf pine (Pinus palustris) plant communities over
time and across multiple spatial scales. The longleaf pine eco-
system is a re-dependent system located in the species-rich
Coastal Plain of the southeastern United States and is charac-
terized by an open tree canopy and an herbaceous layer domi-
nated by graminoids and forbs (Noss 2013). Tree density and
canopy cover vary according to re history, land-use history
and environmental conditions and can range from 0% to 50%
canopy cover (Platt, Evans & Rathbun 1988). Here, we
describe these systems as re-dependent grasslands and wood-
lands to express the variation in tree density in this ecosystem
and to put them into context with other grass-dominated
ecosystems in the world. Frequent re (every
15 years) is necessary for the maintenance of plant species
richness, plant species composition and vegetation structure
(Walker & Peet 1984; Kirkman et al. 2004). Without frequent
re, species richness declines, woody components and litter
increase, and the understorey becomes dense and closed
(Heyward 1939; Glitzenstein, Streng & Wade 2003; Hiers
et al. 2007). Longleaf pine ecosystems are of high conserva-
tion concern because they are the most species rich in North
America at small scales (52 species in 1 m
2
; Walker & Peet
1984; Palmquist, Peet & Weakley 2014) and have high levels
of endemism (Sorrie & Weakley 2001; Noss et al. 2014).
Furthermore, the North American Coastal Plain in which the
longleaf pine ecosystem is embedded was recently proposed
as a global biodiversity hotspot (Noss et al. 2014).
The longleaf pine ecosystem is an ideal system to explore
spatiotemporal vegetation change with respect to re and
environmental context for several reasons. First, this ecosys-
tem is structured by local and regional environmental gradi-
ents (e.g. soil properties and climate; Walker & Peet 1984;
Carr et al. 2009; Peet, Palmquist & Tessel 2014), and thus, it
is possible to examine the effects of environmental context on
vegetation change over time in response to re frequency. In
particular, soil properties (e.g. moisture, texture and base cat-
ion availability) are the most important environmental drivers
of plant community structure in longleaf pine woodlands
(Peet 2006; Carr et al. 2009; Peet, Palmquist & Tessel 2014).
Second, similar to other re-dependent ecosystems in North
America, longleaf pine woodlands experienced a period of
re exclusion in the early to mid-20th century (Frost 2006).
However, in the last few decades, there has been an increased
effort by land managers to implement prescribed re, and
consequently, re frequencies in our study area have
increased since the 1990s relative to the previous re regime.
Despite general increases in re frequency over the last
20 years, the land managers in this ecosystem have varied
goals and resources with the consequence that the frequency
of re implemented since the 1990s varies across sites. Here,
we explore the effects of differences in re frequency across
sites, while examining the impacts of reinstating generally
more frequent re over the last 20 years by resampling a per-
manently marked vegetation plot data set that spans the envi-
ronmental gradient in this ecosystem.
We quantied several aspects of community structure in
longleaf pine woodlands to address three main questions.
First, how do patterns of beta-diversity (e.g. turnover in spe-
cies composition) and species richness across time vary with
spatial scale? Secondly, how have re regime and environ-
mental context inuenced the above metrics and do those
effects change with spatial scale? Lastly, are beta-diversity
and species richness responding in similar ways to re fre-
quency and environmental context? We expected re history
would inuence the types of species turning over, with
increases in woody species and loss of small-statured herba-
ceous species (e.g. rosette herbs) on re-suppressed sites. In
addition, we expected that certain types of species (e.g.
annual species, species sensitive to changes in soil moisture
availability) would inherently turn over more frequently than
others due to life-history characteristics. Our study is unique
in that we examined vegetation change over decades in per-
manent vegetation plots at scales of 0.011000 m
2
in
©2015 The Authors. Journal of Ecology ©2015 British Ecological Society, Journal of Ecology,103, 9981008
Scale dependence of change in longleaf pine vegetation 999
calculated dissimilarity matrices for each growth form at 1000 m
2
and
then calculated a mean dissimilarity value for each growth form.
Results
SPECIES RICHNESS
Species richness increased signicantly over time at scales
of 11000 m
2
, remained constant at 0.1 m
2
and decreased
at 0.01 m
2
(Table 1). Silt % and quadratic transformed re
frequency were the best predictors of changes in species
richness at 100, 400 and 1000 m
2
(unique variance
explained by silt %: R
2
=0.250.32, by re frequency:
R
2
=0.160.18; Table 2, Fig. 1). In contrast to larger spa-
tial scales, at 0.0110 m
2
community type was a better pre-
dictor of changes in species richness than silt %, whereas
quadratic transformed re frequency continued to be impor-
tant but secondarily to community type (Table 2). Silt %
and community type are correlated, but not equivalent, as
silt % is one of two key axes along with soil moisture that
largely predict longleaf pine community types (see g.
A10.7 in Peet, Palmquist & Tessel 2014). Thus, soil mois-
ture, in addition to soil texture, appears to be an important
driver of species richness at small spatial scales. In addi-
tion, community type may have emerged as a better predic-
tor of species richness than silt % at smaller spatial scales
because one community type in particular (i.e. savannas)
lost species over time at those scales (see below for further
discussion; Appendix S3).
Across all scales, increases in species richness were gener-
ally greatest on sites with more frequent re and higher silt
content (e.g. silty woodlands, Appendix S3, Fig. 1). In con-
trast to silty, frequently burned sites, species richness on
sandy, less frequently burned sites remained relatively stable
over time (e.g. sandhills, subxeric woodlands, Appendix S3).
At all spatial scales, environmental context (e.g. silt %, com-
munity type) was a better predictor of changes in species
richness than re frequency, although this may reect the
fact most sites experienced relatively frequent re (see Dis-
cussion section). Although species richness generally
increased with re frequency, we detected species richness
declines at small spatial scales on the site with the highest
re frequency (Big Island Savanna in the Green Swamp Pre-
serve, Fig. 1). We attribute this species loss to reduced re
frequency in the last 15 years, perhaps compounded with
ongoing, long-term drought (see Palmquist, Peet & Weakley
2014). The species richness declines in Big Island Savanna
explain why quadratic transformed re frequency was a bet-
ter t to the data than untransformed re frequency: species
richness had a positive linear relationship with re frequency,
but then decreased in the most frequently burned site
(Fig. 1).
BETA TURNOVER
Mean raw species turnover was high, especially at larger spa-
tial scales (15.7 species gained and/or lost at 100 m
2
, 20.4
Table 1. Changes in species richness and beta-diversity from the early 1990s to 2009 across multiple spatial scales
Area
(m
2
)
Mean
richness DRichness
Mean
turnover
Turnover
range
Mean
obs b
Mean
sim bSES
Mean obs
dissim
Mean sim
dissim SES
1000 59.1 6.3** 20.4** 050 0.16 0.62 8.04 0.21 0.36 2.13
400 51.6 5.2** 18.5** 047 0.17 0.66 8.50 0.21 0.38 2.20
100 36.3 4.7** 15.7** 045 0.19 0.67 7.62 0.25 0.57 4.24
10 20.9 2.8** 12.4** 045 0.26 0.70 5.73 –– –
1 11.9 0.9* 8.9** 037 0.35 0.74 3.88 –– –
0.1 5.6 0.03* 5.8** 028 0.46 0.94 2.76 –– –
0.01 2.0 0.3* 3.3** 014 0.74 0.83 0.39 –––
Mean richness is the mean species richness in 2009, whereas Drichness is the mean change in richness across time. Mean turnover is the mean
number of species lost and gained over time, and turnover range is the range of mean turnover. Mean obs bis the mean beta turnover (mean
turnover divided by two times the mean species richness), and mean sim bis the mean simulated beta turnover across 1000 null communities.
Mean obs dissim and mean sim dissim are the observed and simulated mean BrayCurtis dissimilarity of each plot to itself over time, respec-
tively. Standardized effect size (SES) values above 2 indicate values that are greater than random expectation, whereas values below 2 are lower
than random expectation. Signicant SES values are highlighted in grey. **P<0.001, *P<0.05.
Table 2. Variance partitioning results for species richness, beta turn-
over and dissimilarity across all scales (0.011000 m
2
)
Spatial
Scale
(m
2
)
Species richness
Beta turnover Dissimilarity
Environment Fire Shared Environment Environment
R
2
R
2
R
2
R
2
R
2
1000 0.27*0.17 0.06 0.25 0.31
400 0.32*0.18 0.05 0.22 0.16
100 0.25*0.16 0.02 0.18 0.11
10 0.20 0.06 0.11 0.11
1 0.18 0.15 0.22 0.14
0.1 0.25 0.08 0.21 0.10
0.01 0.21 0.06 0.12 0.10
Reported values include the unique variance explained by environ-
mental context (Environment) and the unique variance explained by
quadratic transformed re frequency (Fire), and the shared variance
explained by both predictors (Shared). Asterisks denote when silt %
was the best predictor, whereas no asterisks indicates community type
was the best predictor. Variance partitioning results are not reported
for dissimilarity at scales below 100 m
2
because we lacked abundance
data at those scales. All associated P-values for each predictor in the
models are <0.01.
©2015 The Authors. Journal of Ecology ©2015 British Ecological Society, Journal of Ecology,103, 9981008
1002 K. A. Palmquist, R. K. Peet & S. R. Mitchell
number of res =3) than they were during 19912009 (mean number
of re =2; Appendix S1). Despite this fact, re frequency since 1991
varied across different sites according to land management agency
goals and resources (112 res since 1991, see Appendix S1).
ANALYSIS
Prior to analysis, all taxonomic names were standardized across the
two sampling periods to ensure that changes in nomenclature and tax-
onomic resolution of the ora across time did not affect the magni-
tude of vegetation change detected. We created three separate
standardized species lists: one for calculating species richness, one for
calculating beta turnover and one for calculating BrayCurtis dissimi-
larity (see below for further discussion of metrics, see Appendix S5
for further discussion of standardization).
Several metrics were used to quantify the magnitude and direction
of vegetation change over time. Species richness in 19911993 and
20092010 and raw species turnover (dened here as the number of
species lost over time plus the number gained) were calculated for each
spatial scale for each plot. Thereafter, we calculated the mean change
in species richness and the mean raw turnover at each spatial scale. To
quantify the magnitude of compositional change over time, we used
two different metrics of beta-diversity that capture somewhat different
aspects of species turnover. Since different metrics of beta-diversity
can yield slightly different results, using complementary metrics is rec-
ommended (Anderson et al. 2011). The rst metric was Wilson &
Shmidas (1984) beta turnover metric: (bT)=(g+l)/2a, which sums
the number of species gained (g) and lost (l) over time and divides by
two times the mean species richness (a). This metric describes how
many species have been lost and gained over time and does not con-
sider species abundance. The second beta-diversity metric, BrayCurtis
dissimilarity, considers how many species are shared across two sites
scaled by their abundance (Bray & Curtis 1957). We use cover class
code (110) as our metric of abundance (see Peet, Wentworth & White
1998), which gives a more equal balance to rare and common species
than per cent cover. We calculated BrayCurtis dissimilarity (hereafter
dissimilarity) at spatial scales 100 m
2
(i.e. those for which we had
estimates of species abundance). The dissimilarity of each plot to itself
20 years later was extracted and used in analyses below.
To examine whether dissimilarity and beta turnover values changed
across spatial scales, we used null model analysis. This was necessary
because beta-diversity values are expected to be larger at small scales
due to chance alone; as the size of the sample approaches the size of
the pool, species composition becomes more similar (Kraft et al.
2011). First, 1000 null communities were generated using the swap
method (Gotelli & Entsminger 2003), which held constant species
richness per site and species occupancy across all sites (e.g. the total
number of sites each species occurred in). We then calculated a
dissimilarity matrix on each null community and extracted the dissim-
ilarity of plots to themselves over time. We also calculated beta
turnover for each null community. To compare results across spatial
scales, we calculated a standardized effect size (SES) for each scale,
SES =(I
obs
I
sim
)/S
sim
, where I
obs
is the observed mean dissimilarity
of all plots to themselves over time or mean observed beta turnover,
I
sim
is the mean simulated dissimilarity of all plots to themselves over
time or mean simulated beta turnover, and S
sim
is the standard devia-
tion of the simulated indices (Gotelli & McCabe 2002). SES values
above two indicate dissimilarity and beta turnover values that are
greater than random expectation, whereas SES values below two indi-
cate dissimilarity and beta turnover values that are less than random
expectation. We examined how SES values changed across spatial
scales with the expectation that SES values would be more positive at
small scales relative to larger scales if vegetation change was more
rapid at small scales. Null model analysis for dissimilarity was imple-
mented in Rversion 2.15.2 using the vegan and bipartite packages (R
Development Core Team 2012), while null model analysis for beta
turnover was conducted by modifying the code provided in Kraft
et al. (2011).
Linear models were used to quantify how environmental context
(e.g. soil properties, community type) and re history inuenced spe-
cies richness, beta turnover and dissimilarity over time. Model selec-
tion using AIC was employed to identify which soil (e.g. nutrients,
texture, organic matter, bulk density), site (elevation) and re (re fre-
quency, average re-return interval, time since re, re season) attri-
butes should remain in the model for species richness and beta
turnover at scales of 0.011000 m
2
and for dissimilarity at scales of
1001000 m
2
(Burnham & Anderson 2002). In addition, we explored
whether community type, a categorical variable that maps onto soil
orders and reects broader-scale environmental variation within the
longleaf pine ecosystem (see Peet 2006; Peet, Palmquist & Tessel
2014), was a better predictor of vegetation change than individual soil
variables. During model selection, we explored all soil, site, commu-
nity type and re variables and removed from the nal models those
that were not signicant predictors and/or did not explain additional
variation in species richness, beta turnover and dissimilarity. Fe, Ca,
Mg and Al were log-transformed before analysis due to large and
asymmetric data ranges for these variables across plots. After model
selection, variance partitioning analyses were conducted to determine
the unique variance explained by each predictor in the best-t model,
along with the shared and unexplained variance in each model
(Legendre & Legendre 2012). The explained variances we report are
the unique variance components of each predictor in the nal models.
Non-metric multidimensional scaling (NMS) ordination was used to
examine the magnitude and direction of compositional change visu-
ally. NMS displayed all 1000 m
2
plots from both sampling events in
ordination space, and vectors were drawn from each plot during the
19911993 sampling to the same plot during 20092010 (Fig. 2).
Environmental overlays of soil, site and re variables were used to
identify the environmental attributes and disturbance regimes of plots
in ordination space. These graphics helped illustrate whether sites with
certain re regimes or environmental conditions had experienced more
or less change than other plots. Additionally, NMS ordination was
used to examine whether the ordination space during 19911993 had
expanded or contracted over time (Fig. 3). Contraction in ordination
space indicates that plots are becoming more similar to one another in
community composition over time. NMS was performed in Rv.2.15.2
using the LABDSV packages (R Development Core Team 2012).
To determine if particular groups of taxa were consistently being
lost or gained over time, we assigned each species to a growth form
category (caulescent herb, matrix graminoid, fern, geophyte, hemipar-
asite, insectivore, legume, rosette herb, shrub, single-culm graminoid,
sub-shrub, tree and vine; see Appendix S2) and then examined how
the frequency of each growth form changed over time (Table 3). We
summarized vegetation change by growth form in two ways. First, we
calculated the total number of times species in each growth form that
were gained and lost over time at both 1 and 1000 m
2
. We then cal-
culated a ratio that reects the number of species gained versus lost,
by dividing the total number of gains for each growth form by the
total number of losses. Values greater than one indicate gains exceed
losses, whereas values less than one indicate losses exceed gains. Sec-
ond, to examine whether particular growth forms were more dynamic
over time, irrespective of whether they were lost or gained, we
©2015 The Authors. Journal of Ecology ©2015 British Ecological Society, Journal of Ecology,103, 9981008
Scale dependence of change in longleaf pine vegetation 1001
calculated dissimilarity matrices for each growth form at 1000 m
2
and
then calculated a mean dissimilarity value for each growth form.
Results
SPECIES RICHNESS
Species richness increased signicantly over time at scales
of 11000 m
2
, remained constant at 0.1 m
2
and decreased
at 0.01 m
2
(Table 1). Silt % and quadratic transformed re
frequency were the best predictors of changes in species
richness at 100, 400 and 1000 m
2
(unique variance
explained by silt %: R
2
=0.250.32, by re frequency:
R
2
=0.160.18; Table 2, Fig. 1). In contrast to larger spa-
tial scales, at 0.0110 m
2
community type was a better pre-
dictor of changes in species richness than silt %, whereas
quadratic transformed re frequency continued to be impor-
tant but secondarily to community type (Table 2). Silt %
and community type are correlated, but not equivalent, as
silt % is one of two key axes along with soil moisture that
largely predict longleaf pine community types (see g.
A10.7 in Peet, Palmquist & Tessel 2014). Thus, soil mois-
ture, in addition to soil texture, appears to be an important
driver of species richness at small spatial scales. In addi-
tion, community type may have emerged as a better predic-
tor of species richness than silt % at smaller spatial scales
because one community type in particular (i.e. savannas)
lost species over time at those scales (see below for further
discussion; Appendix S3).
Across all scales, increases in species richness were gener-
ally greatest on sites with more frequent re and higher silt
content (e.g. silty woodlands, Appendix S3, Fig. 1). In con-
trast to silty, frequently burned sites, species richness on
sandy, less frequently burned sites remained relatively stable
over time (e.g. sandhills, subxeric woodlands, Appendix S3).
At all spatial scales, environmental context (e.g. silt %, com-
munity type) was a better predictor of changes in species
richness than re frequency, although this may reect the
fact most sites experienced relatively frequent re (see Dis-
cussion section). Although species richness generally
increased with re frequency, we detected species richness
declines at small spatial scales on the site with the highest
re frequency (Big Island Savanna in the Green Swamp Pre-
serve, Fig. 1). We attribute this species loss to reduced re
frequency in the last 15 years, perhaps compounded with
ongoing, long-term drought (see Palmquist, Peet & Weakley
2014). The species richness declines in Big Island Savanna
explain why quadratic transformed re frequency was a bet-
ter t to the data than untransformed re frequency: species
richness had a positive linear relationship with re frequency,
but then decreased in the most frequently burned site
(Fig. 1).
BETA TURNOVER
Mean raw species turnover was high, especially at larger spa-
tial scales (15.7 species gained and/or lost at 100 m
2
, 20.4
Table 1. Changes in species richness and beta-diversity from the early 1990s to 2009 across multiple spatial scales
Area
(m
2
)
Mean
richness DRichness
Mean
turnover
Turnover
range
Mean
obs b
Mean
sim bSES
Mean obs
dissim
Mean sim
dissim SES
1000 59.1 6.3** 20.4** 050 0.16 0.62 8.04 0.21 0.36 2.13
400 51.6 5.2** 18.5** 047 0.17 0.66 8.50 0.21 0.38 2.20
100 36.3 4.7** 15.7** 045 0.19 0.67 7.62 0.25 0.57 4.24
10 20.9 2.8** 12.4** 045 0.26 0.70 5.73 –– –
1 11.9 0.9* 8.9** 037 0.35 0.74 3.88 –– –
0.1 5.6 0.03* 5.8** 028 0.46 0.94 2.76 –– –
0.01 2.0 0.3* 3.3** 014 0.74 0.83 0.39 –– –
Mean richness is the mean species richness in 2009, whereas Drichness is the mean change in richness across time. Mean turnover is the mean
number of species lost and gained over time, and turnover range is the range of mean turnover. Mean obs bis the mean beta turnover (mean
turnover divided by two times the mean species richness), and mean sim bis the mean simulated beta turnover across 1000 null communities.
Mean obs dissim and mean sim dissim are the observed and simulated mean BrayCurtis dissimilarity of each plot to itself over time, respec-
tively. Standardized effect size (SES) values above 2 indicate values that are greater than random expectation, whereas values below 2 are lower
than random expectation. Signicant SES values are highlighted in grey. **P<0.001, *P<0.05.
Table 2. Variance partitioning results for species richness, beta turn-
over and dissimilarity across all scales (0.011000 m
2
)
Spatial
Scale
(m
2
)
Species richness
Beta turnover Dissimilarity
Environment Fire Shared Environment Environment
R
2
R
2
R
2
R
2
R
2
1000 0.27*0.17 0.06 0.25 0.31
400 0.32*0.18 0.05 0.22 0.16
100 0.25*0.16 0.02 0.18 0.11
10 0.20 0.06 0.11 0.11
1 0.18 0.15 0.22 0.14
0.1 0.25 0.08 0.21 0.10
0.01 0.21 0.06 0.12 0.10
Reported values include the unique variance explained by environ-
mental context (Environment) and the unique variance explained by
quadratic transformed re frequency (Fire), and the shared variance
explained by both predictors (Shared). Asterisks denote when silt %
was the best predictor, whereas no asterisks indicates community type
was the best predictor. Variance partitioning results are not reported
for dissimilarity at scales below 100 m
2
because we lacked abundance
data at those scales. All associated P-values for each predictor in the
models are <0.01.
©2015 The Authors. Journal of Ecology ©2015 British Ecological Society, Journal of Ecology,103, 9981008
1002 K. A. Palmquist, R. K. Peet & S. R. Mitchell
at 1000 m
2
, Table 1). Beta turnover was moderate to high,
representing turnover of between 16% and 74% of plant
species over time, but was greatest at small spatial scales
(<1m
2
, Table 1). Thus, there are more species turning over
at larger scales, but they make up a smaller proportion of the
total ora than the amount turning over at small scales. At all
spatial scales, community type was the only signicant pre-
dictor of beta turnover (Table 2, P-values <0.01). Similar to
species richness, beta turnover was greater on silty, mesic
sites (e.g. silty woodlands and savannas, Appendix S3).
We used null model analysis to explore whether beta turn-
over SES values increased as spatial scale decreased. SES
values across most scales were negative and signicant
(Table 1), indicating observed beta turnover was consistently
smaller than simulated beta turnover. The one exception was
at 0.01 m
2
, where beta turnover was no different than the
mean simulated beta turnover (SES =0.39, Table 1).
Although most SES values associated with beta turnover were
negative and signicantly different than random expectation,
they became less negative as spatial scale decreased, suggesting
a trend in beta turnover SES values across scales, albeit a
non-signicant one.
To summarize the types of species turning over across
time, we calculated a ratio of gains to losses, along with dis-
similarity for each growth form (Table 3). At 1000 m
2
, the
ratio of gained versus lost species revealed most growth
forms have increased in frequency over time (i.e. gains have
exceeded losses, >1; Table 3). Notable exceptions include
geophytes and hemiparasites, which have consistently been
lost over time (ratio =0.4, 0.5, respectively). At 1 m
2
, insec-
tivores (ratio =0.3), along with geophytes (ratio =0.4) and
hemiparasites (ratio =0.3), have been lost disproportionately.
Many types of species turned over across time, but dissimilar-
ity was especially high for hemiparasites (0.80; e.g. Seymeria
cassioides), geophytes (0.70; e.g. Calopogon spp.), single-
culm graminoids (0.62; e.g. Scleria spp.), insectivores (0.60;
e.g. Drosera capillaris), rosette herbs (0.49; e.g. Eurybia pa-
ludosa) and caulescent herbs (0.48; e.g. Polygala lutea).
Small-statured herbaceous species and species sensitive to
changes in moisture (e.g. insectivores) have been gained and
2 4 6 8 10 12
0510–10
–5
15 20
0 5 10 15
–5–10 05
–5–10–15 0
–5–10 0
5
10 15
20
25
Fire frequency
2 4 6 8 10 12
Fire frequency
2 4 6 8 10 12
Fire frequency
2 4 6 8 10 12
Fire frequency
2 4 6 8 10 12
Fire frequency
2 4 6 8 10 12
Fire frequency
Change in richness
R² = 0.17
1000 m²
R² = 0.16
100 m²
R² = 0.06
10 m²
Change in richness
R² = 0.15
1 m²
R² = 0.08
0.1 m²
–4 –2 0 2
R² = 0.06
0.01 m²
Fig. 1. Change in species richness over time at 0.011000 m
2
versus re frequency. In general, frequently burned sites have gained a greater
number of species over time. However, plots that have historically experienced the greatest re frequency (12, located on the Green Swamp Pre-
serve) have not gained as many species because of recent changes in the re management regime, compounded with long-term drought. R
2
values
represent the unique variance explained by quadratic transformed re frequency.
©2015 The Authors. Journal of Ecology ©2015 British Ecological Society, Journal of Ecology,103, 9981008
Scale dependence of change in longleaf pine vegetation 1003
lost most frequently. In contrast, ferns, matrix graminoids,
shrubs and sub-shrubs have remained relatively stable
(Table 3).
BRAYCURTIS DISSIMILARITY
As with beta turnover, we used null model analysis to exam-
ine whether the observed dissimilarity was greater or less than
null expectation. We found the observed dissimilarity was
signicantly lower than the simulated dissimilarity at 100
1000 m
2
, suggesting greater vegetation stability at those
scales (Table 1). At 1000 m
2
, the best predictor of dissimilar-
ity over time was community type, with moist, silty sites (e.g.
savannas) experiencing the greatest compositional change
over time (R
2
=0.31, P<0.001; Table 2). Community type
was also the best predictor of dissimilarity at 400 m
2
(R
2
=0.16, P<0.05) and 100 m
2
(R
2
=0.11, P<0.05). In
general, silty, frequently burned sites have experienced greater
changes in dissimilarity over time, in addition to greater
changes in species richness and beta turnover.
Non-metric multidimensional scaling conrmed that com-
positional change for most 1000 m
2
plots has been relatively
modest. In Fig. 2, vectors connect the same plot to itself over
time, and the length of the vectors relates to the magnitude of
vegetation change. Only a few plots show substantial change
over time, these sites are re-suppressed sandhills and subxer-
ic woodlands (Fig. 2). NMS also revealed that there has been
slight homogenization of the vegetation, indicated by a small
constriction in the amount of ordination space occupied by all
plots over time (Fig. 3). Thus, plots have become slightly
more similar to other plots in their community composition.
We attribute this to increased burning efforts over the last
20 years by land management agencies within the study area,
thus shifting the vegetation to a composition more typical of
re-maintained vegetation.
Discussion
We examined how re frequency and environmental context
have inuenced vegetation structure in longleaf pine commu-
nities over ~20 years and across multiple spatial scales. We
detected a scale-dependent response in the vegetation over
time; most notably, species richness increased signicantly at
11000 m
2
, remained constant at 0.1 m
2
and decreased at
0.01 m
2
. In general, the magnitude of vegetation change
increased as re frequency and silt percentage increased.
Thus, on average, silty, frequently burned sites have experi-
enced greater changes in species richness, beta turnover and
dissimilarity than frequently burned sandy sites (Appendix
S3), likely owing to greater biomass accumulation on silty
sites and greater asymmetric competition for light. Small-sta-
tured herbaceous species and species sensitive to changes in
soil moisture represent the bulk of species turning over across
time. Environmental context (e.g. silt % or community type)
was consistently the most important predictor of vegetation
change, while re frequency explained additional variation,
albeit less. This is consistent with work from other re-depen-
dent grasslands and woodlands, which suggests soil texture is
the primary driver of plant community structure when re fre-
quency is high (Sankaran et al. 2005; Dantas et al. 2015).
Other work in frequently disturbed grasslands and wood-
lands suggests that environmental ltering is an important
factor structuring plant species richness and composition
Table 3. Species gained and lost over time summarized by growth
form at 1 and 1000 m
2
Spatial scale 1000 m
2
1m
2
1000 m
2
Growth form
Gained/Lost
ratio
Gained/Lost
ratio Dissim
Caulescent herb 1.8 1.1 0.48
Clubmoss 1.7 2.2 0.43
Matrix graminoid 3.3 1.5 0.20
Fern 6.0 1.6 0.19
Geophyte 0.4 0.4 0.70
Hemiparasite 0.5 0.3 0.80
Insectivore 1.6 0.3 0.60
Legume 1.4 1.5 0.46
Rosette herb 2.1 1.2 0.49
Shrub 3.7 1.8 0.35
Single-culm graminoid 1.2 1.1 0.62
Sub-shrub 1.8 1.3 0.19
Tree 2.3 1.6 0.46
Vine 4.5 4.4 0.51
Gained/Lost ratio is the number of species in each growth form
gained over time/the number of species in each growth form lost over
time. Ratios of >1 indicate that gains exceed losses, whereas ratios
of <1 indicate more species have been lost than gained. Dissim at
1000 m
2
is the mean BrayCurtis dissimilarity of plots to themselves
over time for each growth form.
Fig. 2. Non-metric multidimensional scaling ordination showing
changes in BrayCurtis dissimilarity at 1000 m
2
over time for different
community types, represented by the length of vectors that connect the
same plot over time. Compositional change is modest, except for two
plots in the lower left corner. Environmental vectors indicate the direc-
tions of variation in organic matter (om), silt %, sand %, re frequency
and pH across compositional space. Compositional change over time
has been greatest on sites with high sand content and high re-return
intervals (i.e. infrequently burned sandhills and subxeric woodlands).
©2015 The Authors. Journal of Ecology ©2015 British Ecological Society, Journal of Ecology,103, 9981008
1004 K. A. Palmquist, R. K. Peet & S. R. Mitchell
over time (e.g. Biondini, Patton & Nyren 1998; Prober, Thi-
ele & Speijers 2013; Smith et al. 2013; Dantas et al. 2015).
Prober, Thiele & Speijers (2013) found that rainfall
explained 3160% of the variation in plant species richness
and plant cover in a grazing experiment in Australian wood-
lands, whereas grazing frequency explained only 38%. Glit-
zenstein, Streng & Wade (2003) examined changes in
species richness and composition in relation to re fre-
quency in two separate multidecadal experiments in the
longleaf pine ecosystem. In both experimental sites, soil
moisture explained more variation in species composition
than re frequency (28.4% vs. 17.7% in experiment 1;
30.9% vs. 21.8% in experiment 2). These ndings are con-
sistent with our work, which suggests local environmental
parameters (e.g. soil properties) may be better predictors of
vegetation change over time than re history in disturbance-
adapted grasslands and woodlands. This in part reects the
fact that environmental context can inuence re frequency
itself, with generally more re on moist sites that have
greater fuel accumulation (Smith et al. 2013). Sites with
higher re frequency have a greater likelihood of experienc-
ing vegetation change as each re event creates space and
resources for emergence of plants from the seed and bud
bank, or from seed dispersal from offsite (Overbeck et al.
2005). Furthermore, environmental context may be a better
predictor of vegetation change in longleaf pine woodlands
because several factors that inuence vegetation dynamics
vary with environmental site context. For example, mesic,
silty longleaf pine types (e.g. savannas) have larger species
pools (K. A. Palmquist, unpubl. data) and typically stronger
competition due to the presence of a few competitively
dominant grasses. In savannas, higher competition for light
may increase local extinction events, whereas larger species
pools may result in a greater inux of propagules leading to
both higher colonization and extinction rates, and hence
greater vegetation change (Kirkman et al. 2014). In contrast,
sites with smaller species pools (e.g. sandhills) may have
reduced colonization rates and consequently less vegetation
change over time (Oesterheld & Semmartin 2011; Lezama
et al. 2014). Lastly, environmental context may be more
important in re-dependent grasslands and woodlands that
experience very consistent and frequent re (Dantas et al.
2015).
Changes in species composition over time were fairly mod-
est, as indicated by the length of vectors in Fig. 2, despite the
fact that raw turnover and changes in species richness were
substantial (Table 1). We believe this initial contradiction can
be explained by the frequency and abundance of species in
the community. Longleaf pine woodlands are comprised of
many infrequent, low-abundance species and a small number
of frequent, abundant species (Kirkman et al. 2001; Keddy
et al. 2006; Clark, Siegrist & Keddy 2008). Frequent, abun-
dant species in this ecosystem are long-lived perennials with
substantial below-ground storage, allowing them to persist for
years or even decades in the same location. In contrast, infre-
quent species are typically shorter lived or susceptible to
year-to-year environmental change (e.g. soil moisture, Droser-
aspp.) or changes in re frequency (e.g. re-followers,
Lemon 1949). Thus, community composition at larger scales
has been relatively stable over time because the species that
are frequent and abundant stay constant, whereas there is high
turnover of the infrequent species, which contribute little to
BrayCurtis dissimilarity. This phenomenon has been docu-
mented in other grasslands and woodlands (Glenn & Collins
1993; Herben et al. 1993; Overbeck et al. 2005).
Previous studies in the longleaf pine ecosystem have sug-
gested that species composition has remained relatively stable
in sites with a long history of frequent re (Glitzenstein et al.
2008). However, the stability of vegetation over time is also a
function of edaphic properties. Glitzenstein, Streng & Wade
(2003) suggested that longleaf pine vegetation at the wet end
of the environmental gradient might change more rapidly with
reductions in re frequency, due to competitive exclusion of
herbaceous species by woody plants, which is consistent with
our ndings. This is in contrast to drier sites, which may be
more stable in response to decreases in re frequency
(Beckage & Stout 2000; Glitzenstein, Streng & Wade 2003).
Several other studies have also documented increases in
species richness with increases in re frequency (Glitzenstein,
Streng & Wade 2003), suggesting that frequent re is neces-
sary for the maintenance of species richness, particularly at
the wet end of the gradient.
–10
(a) (b)
50 510
–5 0 5 10 15
NMDS 1
NMDS 2
1990
2009
0.55 0.65 0.75 0.85
0.50 0.60 0.70 0.80
Average dissimilarity 1990
Average dissimilarity 2009
Fig. 3. (a) Non-metric multidimensional
scaling (NMS) ordination highlighting plots
by sampling year (1990, 2009). Lines outline
the amount of ordination space occupied by
plots in each year. The total NMS area
occupied by plots in 2009 has contracted
slightly from the early 1990s. (b) The
average dissimilarity for each plot to all other
plots in the early 1990s versus the average
dissimilarity of each plot to all other plots
during 2009. The grey line indicates the 1-1
line. Most plots fall below the 1-1 line,
indicating that there has been slight
convergence of community composition over
time.
©2015 The Authors. Journal of Ecology ©2015 British Ecological Society, Journal of Ecology,103, 9981008
Scale dependence of change in longleaf pine vegetation 1005
Vegetation change was also highly dependent on the spatial
scale of observation. Species richness increased at large spa-
tial scales over time, but declined at the smallest two scales
(Table 1). Using null model analysis, we documented that
there has been signicantly less change in dissimilarity over
time relative to null expectation at larger spatial scales. Simi-
larly, we found that the observed beta turnover at most spatial
scales was consistently smaller than the simulated beta turn-
over (SES values less than 2). The one exception was at
0.01 m
2
where the mean observed beta turnover was not sig-
nicantly different than random expectation (SES =0.39).
This suggests longleaf pine vegetation has been stable over
time at all spatial scales, but relatively less stable at the very
smallest scale. In fact, several 0.01 m
2
plots experienced com-
plete replacement of species over time. Turnover at small
scales occurs at a faster rate than turnover of entire subpopu-
lations at larger scales (1000 m
2
), due to the small number of
individuals present (Glenn & Collins 1993). In addition, envi-
ronmental heterogeneity generally increases with spatial scale,
with the consequence that environment lters on species com-
position become more important as spatial scale increases
(Shmida & Wilson 1985; Crawley & Harral 2001; Field et al.
2009). Furthermore, the total variance explained by re fre-
quency and/or environmental context consistently decreased
as spatial scale decreased for species richness, beta turnover
and dissimilarity (Table 2), which is perhaps suggestive that
other processes (e.g. stochastic recruitment) are structuring
beta-diversity patterns at the smallest scales.
Sampling at multiple spatial scales was crucial for identi-
fying how longleaf pine communities had changed over time
because the magnitude of vegetation change varied across
spatial scales. If we had only examined richness patterns
over time at 1000 m
2
, we would have missed the signal of
species loss at the two smallest spatial scales. Monitoring at
multiple spatial scales is not only important for understand-
ing ecological pattern and process, but also for conservation
planning and informing management agencies about best
practices (Boyd et al. 2008). This study is unique in that we
examined turnover in plant species composition simulta-
neously in space and time. Examining vegetation patterns
over time across multiple spatial scales helped to reveal the
generality of processes structuring community patterns in this
ecosystem.
Our work also helps to inform land managers on the effec-
tiveness of their re management regimes for maintaining
plant species richness over time. Across most sites, species
richness has increased over time at most or all spatial scales,
likely due to increased prescribed re frequency during the
last 20 years. This recent change to more frequent re by
many managers has resulted from a realization that frequent
re is necessary to maintain the ecological integrity of long-
leaf pine communities (Costanza 2010). In addition, increased
burning efforts over the last 20 years have resulted in a slight
homogenization of the vegetation over time, which we think
reects a state of re-maintained vegetation. However, not all
land management agencies have increased burning efforts in
the last 20 years. We have documented species richness
declines at small scales on one site, the Green Swamp Pre-
serve (Palmquist, Peet & Weakley 2014), where the average
re-return interval has changed from nearly annual re for
most of the 20th century to re every 23 years during the
last fteen years. This change in the re management regime
has resulted in greater and more prolonged litter accumula-
tion, which appears primarily responsible for the substantial
loss of biodiversity at small spatial scales in the Green
Swamp (Palmquist, Peet & Weakley 2014).
Additional factors that we were unable to account for could
have affected the amount of vegetation change we detected.
First, we did not have access to re history data for most sites
prior to the original sampling in 1991. Particularly, we could
not access whether the time since re was comparable
between the original sampling and the sampling in 2009
2010. Thus, some of the change we detected over time could
be due to differences in time since re. However, the sites
chosen during the original sampling represented the highest
quality, most re-maintained vegetation that remained in the
landscape; thus, it is likely that most sites were sampled
5 years after re or less. Second, although this study encom-
passed a range of re frequencies, nearly all sites have been
burned at least once within the last 15 years (see Appendix
S1). Thus, re frequency might have been a better predictor
of vegetation change over time if our data set had included a
balance of re-suppressed sites and frequently burned sites.
Although our study suggests environmental context is more
important than re frequency in predicting vegetation change
over time, additional studies are needed that span a broader
range of re frequencies to further tease apart the effects of
re history and environmental context. Third, changes in
climate over the last 20 years may have inuenced species
richness and beta-diversity patterns over time in our study
area. The Coastal Plain of North Carolina, in which these
sites are located, has experienced more frequent and severe
drought over the last 30 years (Palmquist, Peet & Weakley
2014). If drought were primarily responsible, we would
expect to see strong directional changes in species composi-
tion (e.g. Savage & Vellend 2015). However, NMS-revealed
plots did not change consistently in one direction (Fig. 2),
suggesting changes in water availability and/or temperature
are not the primary drivers of vegetation change over the last
20 years in this ecosystem. Furthermore, the frequency and
intensity of drought events have increased throughout our
entire study area, so drought should not be driving differences
in vegetation change across sites (Palmquist, Peet & Weakley
2014). Fourth, although we made efforts to standardize survey
time, we were unable to quantify differences in sampling
effort across the original participants. Thus, some of the dif-
ferences detected across time could be due to observer bias.
However, we do not expect observer bias to result in large
differences in the number of species detected at small spatial
scales (0.0110 m
2
) as there is a nite number of individuals
and hence species detectable at these scales. Lastly, while
most of the original subplot markers were relocated (Appen-
dix S1), for some subplots, our ability to resample the exact
same physical location at the smallest spatial scales was
©2015 The Authors. Journal of Ecology ©2015 British Ecological Society, Journal of Ecology,103, 9981008
1006 K. A. Palmquist, R. K. Peet & S. R. Mitchell
hindered. Hence, some caution is in order when interpreting
our results at 0.1 m
2
and below.
Our documentation of decadal changes in plant community
composition in relation to environmental context and re fre-
quency contributes to a growing pool of knowledge docu-
menting how disturbance inuences community structure in
grasslands and woodlands over decadal temporal extents
(Glitzenstein, Streng & Wade 2003; Spasojevic et al. 2010;
Smith et al. 2013). Our work also reveals that in addition to
disturbance, environmental ltering is a key process that
shapes plant diversity and composition over time in re-
dependent grasslands and woodlands. Furthermore, the magni-
tude of vegetation change we detected varied with spatial
scale, suggesting that temporal vegetation change may vary
with the spatial scale of observation.
Acknowledgements
This research was conducted under the Department of Defense Coastal/Estua-
rine Research Program (DCERP, Project RC-1413) funded by a grant from the
Strategic Environmental Research and Development Program (SERDP) to Nor-
man L. Christensen. In 2010, this work was also supported by an Alma Beers
Fellowship to KAP from the Department of Biology at the University of North
Carolina. We thank Allen Hurlbert and William Lauenroth for their valuable
feedback on an earlier version of this manuscript, along with two anonymous
referees whose comments greatly improved this manuscript. We thank T.R.
Wentworth, A.S. Weakley, M.S. Schafale and other participants in the Carolina
Vegetation Survey, who conducted the original vegetation survey in 1991
1993. Courtney Colwell, Jan Goodson, Jean Lynch and Susan Gale helped with
data collection in 2009. We thank land managers at Camp Lejeune Marine
Corps Base, Croatan National Forest, North Carolina Wildlife Resource Com-
mission, The Nature Conservancy, Cedar Island National Wildlife Refuge and
North Carolina State Parks for granting access to study sites.
Data accessibility
All plot data from both 19911993 and 20092010 have been uploaded to a
data set called Southeastern NC Longleaf re-sample on Vegbank (vegbank.org):
http://vegbank.org/cite/VB.ds.200278.SENCLLResample.
References
Adler, P.B. & Lauenroth, W.K. (2003) The power of time: Spatiotemporal scal-
ing of species diversity. Ecology Letters,6, 749756.
Amatangelo, K.L., Fulton, M.R., Rogers, D.A. & Waller, D.M. (2011) Con-
verging forest community composition along an edaphic gradient threatens
landscape-level diversity. Diversity and Distributions,17, 201213.
Anderson, M.J., Crist, T.O., Chase, J.M., Vellend, M., Inouye, B.D., Freestone,
A.L. et al. (2011) Navigating the multiple meanings of bdiversit y: a road-
map for the practicing ecologist. Ecology Letters,14,1928.
Beckage, B. & Stout, I.J. (2000) Effects of repeated burning on species richness
in a Florida pine savanna: a test of the intermediate disturbance hypothesis.
Journal of Vegetation Science,11, 113122.
Biondini, M.E., Patton, B.D. & Nyren, P.E. (1998) Grazing intensity and eco-
system processes in a northern mixed-grass prairie, USA. Ecological Applica-
tions,8, 469479.
Bowman, D.M.J.S., Balch, J.K., Artaxo, P., Bond, W.J., Carlson, J.M., Cochra-
ne, M.A. et al. (2009) Fire in the earth system. Science,324, 481484.
Boyd, C., Brooks, T.M., Butchart, S.H.M., Edgar, G.J., Da Fonseca, G.A.B.,
Hawkins, F., Hoffmann, M., Sechrest, W., Stuart, S.N. & van Dijk, P.P.
(2008) Spatial scale and the conservation of threatened species. Conservation
Letters,1,3743.
Bray, J.R. & Curtis, J.T. (1957) An ordination of upland forest communities of
southern Wisconsin. Ecological Monographs,27, 325349.
Burnham, K.P. & Anderson, D.R. (2002) Model Selection and Multimodel
Inference: A Practical Information-theoretic Approach. Springer-Verlag,
New York, NY.
Carr, S.C., Robertson, K.M., Platt, W.J. & Peet, R.K. (2009) A model of geo-
graphical, environmental and regional variation in vegetation composition of
pyrogenic grasslands of Florida. Journal of Biogeography,36, 16001612.
Christensen, N.L. (2000) Vegetation of the southeastern Coastal Plain. North
American Terrestrial Vegetation (eds M.G. Barbour & W.D. Billings), pp.
397448. Cambridge University Press, Cambridge, UK.
Clark, M.A., Siegrist, J. & Keddy, P.A. (2008) Patterns of frequency in spe-
cies-rich vegetation in pine savannas: effects of soil moisture and scale. Eco-
science,15, 529535.
Costanza, J.K. (2010) Conservation of the Longleaf Pine Ecosystem: Ecology
and Social Context. PhD Thesis, University of North Carolina at Chapel Hill,
Chapel Hill, NC.
Crawley, M.J. & Harral, J.E. (2001) Scale dependence in plant biodiversity.
Science,291, 864868.
Dantas, V.L., Batalha, M.A., Franca, H. & Pausas, J.G. (2015) Resource avail-
ability shapes re-ltered savannas. Journal of Vegetation Science,26, 395
403.
Field, R., Hawkins, B.A., Cornell, H.V., Currie, D.J., Diniz-Filho, J.A.F.,
Gu
egan, J.-F., Kaufman, D.M., Kerr, J.T., Mittelbach, G.G., Oberdorff, T.,
OBrien, E.M. & Turner, J.R.G. (2009) Spatial species-richness gradients
across scales: a meta-analysis. Journal of Biogeography,36, 132147.
Frost, C. (2006) History and future of the longleaf pine ecosystem. The Long-
leaf Pine Ecosystem: Ecology, Silviculture, and Restoration (eds S. Jose, E.
Jokela & D. Miller), pp. 948. Springer, New York, NY.
Gauthier, S., Boucher, D., Morissette, J. & De Grandpr
e, L. (2010) Fifty-seven
years of composition change in the eastern boreal forest of Canada. Journal
of Vegetation Science,21, 772785.
Gibson, D.J. & Hulbert, L.C. (1987) Effects of re, topography and year-to-
year climatic variation on species composition in tallgrass prairie. Vegetatio,
72, 175185.
Glenn, S.M. & Collins, S.L. (1993) Experimental analysis of patch dynamics in
tallgrass prairie plant communities. Journal of Vegetation Science,4, 157
162.
Glitzenstein, J.S., Streng, D.R. & Wade, D.D. (2003) Fire frequency effects on
longleaf pine (Pinus palustris P. Miller) vegetation in South Carolina and
northeast Florida, USA. Natural Areas Journal,23,2237.
Glitzenstein, J.S., Streng, D.R., Masters, R.E. & Platt, W.J. (2008) Clarifying
the long-term impacts of re frequency and re season in southeastern
Coastal Plain pine savannas. Managing an Ecosystem on the Edge, Proceed-
ings 6th Eastern Native Grass Symposium (eds W.C. Stringer, J. Andrae &
G. Yarrow), pp. 1318. Columbia, SC.
Gotelli, N.J. & Entsminger, G.L. (2003) Swap algorithms in null model analy-
sis. Ecology,84, 532535.
Gotelli, N.J. & McCabe, D.J. (2002) Species co-occurrence: a meta-analysis of
J. M. Diamonds assembly rules model. Ecology,83, 20912096.
Harrison, S., Inouye, B.D. & Safford, H.D. (2003) Ecological heterogeneity in
the effects of grazing and re on grassland diversity. Conservation Biology,
17, 837845.
Herben, T., Krahulec, F., Hadincov
a, V. & Sk
alov
a, H. (1993) Small-scale vari-
ability as a mechanism for large-scale stability in mountain grasslands. Jour-
nal of Vegetation Science,4, 163170.
Heyward, F. (1939) The relation of re to stand composition of longleaf pine
forests. Ecology,20, 287304.
Hiers, J.K., OBrien, J.J., Will, R.E. & Mitchell, R.J. (2007) Forest oor depth
mediates understory vigor in xeric Pinus palustris ecosystems. Ecological
Applications,17, 806814.
Johnstone, J.F. & Chapin, F.S. (2003) Non-equilibrium succession dynamics
indicate continued northern migration of lodgepole pine. Global Change
Biology,9, 14011409.
Keddy, P.A., Smith, L., Campbell, D.R., Clark, M. & Montz, G. (2006) Pat-
terns of herbaceous plant diversity in southeastern Louisiana pine savannas.
Applied Vegetation Science,9,1726.
Kirkman, K.P., Collins, S.L., Smith, M.D., Knapp, A.K., Burkepile, D.E., Burns,
C.E. et al. (2014) Responses to re differ between South African and North
American grassland communities. Journal of Vegetation Science,25, 793804.
Kirkman, L.K., Goebel, P.C., Palik, B.J. & West, L. (2004) Predicting plant
species diversity in a longleaf pine landscape. Ecoscience,11,80
93.
Kirkman, L.K., Mitchell, R.J., Helton, R.C. & Drew, M.B. (2001) Productivity
and species richness across an environmental gradient in a re-dependent
ecosystem. American Journal of Botany,88, 21192128.
Kraft, N.J.B., Comita, L.S., Chase, J.M., Sanders, N.J., Swenson, N.G., Crist,
T.O. et al. (2011) Disentangling the drivers of bdiversity along latitudinal
and elevational gradients. Science,333, 17551758.
Leach, M.K. & Givnish, T.J. (1996) Ecological determinants of species loss in
remnant prairies. Science,273, 15551558.
©2015 The Authors. Journal of Ecology ©2015 British Ecological Society, Journal of Ecology,103, 9981008
Scale dependence of change in longleaf pine vegetation 1007
Legendre, P. & Legendre, L. (2012) Numerical Ecology. Elsevier, Amsterdam,
The Netherlands.
Lemon, P.C. (1949) Successional responses of herbs in the longleaf-slash pine
forest after re. Ecology,30, 135145.
Lep
s, J. (2014) Scale- and time-dependent effects of fertilization, mowing and
dominant removal on a grassland community during a 15-year experiment.
Journal of Applied Ecology,51, 978987.
Lezama, F., Baeza, S., Altesor, A., Cesa, A., Chaneton, E.J. & Paruelo, J.M.
(2014) Variation of grazing-induced vegetation changes across a large-scale
productivity gradient. Journal of Vegetation Science,25,821.
Mehlich, A. (1984) Mehlich 3 soil test extractant: a modication of Mehlich 2
extractant. Communications in Soil Science and Plant Analysis,15, 1409
1416.
Noss, R.F., Platt, W.J., Sorrie, B.A., Weakley, A.S., Means, D.B., Costanza, J.
& Peet, R.K. (2014) How global biodiversity hotspots may go unrecognized:
lessons from the North American Coastal Plain. Diversity and Distributions,
21, 236244.
Noss, R.F. (2013) Forgotten Grasslands of the South. Island Press, Washing-
ton, D.C..
Oesterheld, M. & Semmartin, M. (2011) Impact of grazing on species composi-
tion: adding complexity to a generalized model. Austral Ecology,36, 881890.
Overbeck, G.E., M
uller, S.C., Pillar, V.D. & Pfadenhauer, J. (2005) Fine-scale
post-re dynamics in southern Brazilian subtropical grassland. Journal of
Vegetation Science,16, 655664.
Palmquist, K.A., Peet, R.K. & Weakley, A.S. (2014) Changes in plant species
richness following reduced re frequency and drought in one of the most
species-rich savannas in North America. Journal of Vegetation Science,25,
14261437.
Pausas, J.G. & Ribeiro, E. (2013) The global reproductivity relationship.
Global Ecology and Biogeography,22, 728736.
Peet, R.K., Palmquist, K.A. & Tessel, S.M. (2014) Herbaceous layer species
richness of southeastern forests and woodlands: patterns and causes. The
Herbaceous Layer in Forests of Eastern North America, 2nd edn. (ed. F.S.
Gilliam), pp. 255276. Oxford University Press, New York, NY.
Peet, R.K., Wentworth, T.R. & White, P.S. (1998) A exible, multipurpose
method for recording vegetation composition and structure. Castanea,63,
262274.
Peet, R.K. (2006) Ecological classication of longleaf pine woodlands. Long-
leaf Pine Ecosystems: Ecology, Management, and Restoration (eds S. Jose,
E. Jokela & D. Miller), pp. 5194. Springer, New York, NY.
Platt, W.J., Evans, G.W. & Rathbun, S.L. (1988) The population dynamics of a
long-lived conifer (Pinus palustris). The American Naturalist,131, 491525.
Preston, F.W. (1960) Time and space and the variation of species. Ecology,41,
611627.
Prober, S.M., Thiele, K.R. & Speijers, J. (2013) Management legacies shape
decadal-scale responses of plant diversity to experimental disturbance
regimes in fragmented grassy woodlands. Journal of Applied Ecology,50,
376386.
R Development Core Team. (2012) R: A Language and Environment for Statis-
tical Computing. R Foundation for Statistical Computing, Vienna, Austria.
Raunkiaer, C. (1934) The Life Forms of Plants and Statistical Plant Geogra-
phy. Clarendon Press, Oxford, UK.
Ruffner, J.A. (1985) Climates of the States: National Oceanic and Atmospheric
Administration Narrative Summaries, Tables, and Maps for Each State, with
Overview of State Climatologist Programs, 3rd edn. Gale Research Com-
pany, Detroit, MI.
Sankaran, M., Hanan, N.P., Scholes, R.J., Ratnam, J., Augustine, D.J., Cade,
B.S. et al. (2005) Determinants of woody cover in African savannas. Nature,
438, 846849.
Savage, J. & Vellend, M. (2015) Elevational shifts, biotic homogenization and
time lags in vegetation change during 40 years of climate warming. Ecogra-
phy, doi:10.1111/ecog.01131.
Shmida, A. & Wilson, M.V. (1985) Biological determinants of species diver-
sity. Journal of Biogeography,12,120.
Smith, M.D., van Wilgen, B.W., Burns, C.E., Govender, N., Potgieter, A.L.F.,
Andelman, S., Biggs, H.C., Botha, J. & Trollope, W.S.W. (2013) Long-term
effects of re frequency and season on herbaceous vegetation in savannas of
the Kruger National Park, South Africa. Journal of Plant Ecology,6,7183.
Soil Survey Staff, Natural Resources Conservation Service, United States
Department of Agriculture. (2012) Web Soil Survey. Available at http://web
soilsurvey.nrcs.usda.gov/ [accessed 15 Jan 2012].
Soininen, J. (2010) Species turnover along abiotic and biotic gradients: patterns
in space equal patterns in time? BioScience,60, 433439.
Sorrie, B.A. & Weakley, A.S. (2001) Coastal Plain vascular plant endemics:
phytogeographic patterns. Castanea,66,5082.
Spasojevic, M.J., Aicher, R.J., Koch, G.R., Marquardt, E.S., Mirotchnick, N.,
Troxler, T.G. & Collins, S.L. (2010) Fire and grazing in a mesic tallgrass
prairie: impacts on plant species and functional traits. Ecology,91, 1651
1659.
Turner, M.G. (2010) Disturbance and landscape dynamics in a changing world.
Ecology,91, 28332849.
Walker, J. & Peet, R. (1984) Composition and species diversity of pine-wire-
grass savannas of the Green Swamp, North Carolina. Vegetatio,55, 163
179.
Westerling, A.L., Hidalgo, H.G., Cayan, D.R. & Swetnam, T.W. (2006) Warm-
ing and earlier spring increase western U.S. forest wildre activity. Science,
313, 940943.
White, E.P., Ernest, S.K.M., Adler, P.B., Hurlbert, A.H. & Lyons, S.K. (2010)
Integrating spatial and temporal approaches to understanding species rich-
ness. Philosophical Transactions of the Royal Society B: Biological Sciences,
365, 36333643.
Wilson, M.V. & Shmida, A. (1984) Measuring beta diversity with presence-
absence data. Journal of Ecology,72, 10551064.
Received 14 December 2014; accepted 16 April 2015
Handling Editor: Charles Canham
Supporting Information
Additional Supporting Information may be found in the online ver-
sion of this article:
Appendix S1. Fire history, community type, and landowner data for
59 permanent vegetation plots.
Appendix S2. Growth form assignments to all taxa in the data set.
Appendix S3. Mean species richness, raw turnover, and beta turnover
for community types across scales.
Appendix S4. Map of the study area.
Appendix S5. Detailed description of taxonomic name standardiza-
tions for species richness, beta turnover, and Bray-Curtis dissimilarity.
©2015 The Authors. Journal of Ecology ©2015 British Ecological Society, Journal of Ecology,103, 9981008
1008 K. A. Palmquist, R. K. Peet & S. R. Mitchell
... 15 species per 1 m 2 ; Hedman et al., 2000), Florida (ca. 22 species per 1 m 2 ; Orzell & Bridges, 2006) and North Carolina (ca. 12 and > 40 species per 1 m 2 ; Palmquist et al., 2015, J. Walker & Peet, 1984. At the local scale (25 m 2 ), our sites supported an average of ca. ...
... At the local scale (25 m 2 ), our sites supported an average of ca. 28 species, which is similar to what has been found at other sites in North and South Carolina Palmquist et al., 2015;Peet, 2006), but at a larger scale (100 m 2 ) than was used in this study. However, further down the latitudinal gradient of the LLP ecosystem, in Louisiana, richness at the 100-m 2 scale (100 species per 100 m 2 ; Platt et al., 2006) is actually similar to that found at larger scales in other areas. ...
... Species area relationships within the LLP ecosystem have been addressed, with patterns of species richness being documented across a broad range of spatial scales, and as spatial scale increases, so does richness (Keddy et al., 2006;L. Kirkman & Myers, 2017;Palmquist et al., 2015;Peet, 2006). Although not studied here, richness at very small scales (0.01 and 0.1 m 2 ) averages 2 and 5.6, respectively, and increases to an average of 65 species at larger spatial scales (1,000 m 2 ; Peet, 2006;Palmquist et al., 2015). ...
Article
Full-text available
Restoration of ecosystems is complex, with multiple targets that can work in concert or conflict with each other, such as biodiversity, species dominance, and biomass. When properly managed, longleaf pine (LLP) savannas are among the most biologically diverse habitats in the world. However, anthropogenic influences, such as fire suppression, have decimated this ecosystem and its biodiversity, making restoration a priority. Here, we describe the biodiversity and community dynamics seen in the understory layer across xeric LLP savannas in North Carolina (NC) and then answer the following questions: What are the predictors of (1) biodiversity, (2) dominance, and (3) biomass at multiple spatial scales? Fifteen observational study sites in North Carolina spanning from the Sandhills to the Coastal Plain. At each of the fifteen sites, twenty‐five sampling plots were established where aboveground herbaceous biomass, species presence and abundance, soil characteristics, and light availability were measured along with numerous other environmental variables. Considerable variation exists across study plots within and across sites, with plant species richness ranging from 1–17 per m2. The relative cover of the dominant grass species, Aristida stricta (wiregrass), also varied greatly within and across sites, with a median of ~30% relative cover per plot. Wiregrass was a significant predictor of biomass and biodiversity at small scales. With increasing wiregrass abundance, richness decreases, with 25% relative wiregrass cover leading to the highest levels of biodiversity. Likewise, as wiregrass abundance is one of the stronger predictors of aboveground biomass, we also found a unimodal richness‐biomass relationship. Our results indicate that at lower ends of the productivity and richness gradients, land managers can increase all three restoration targets in the understory at the same time; however, at more diverse and productive sites, restoration practitioners may need to prioritize one target or find a balance between all three.
... Aujourd'hui encore, on manque de données temporelles pour mieux appréhender l'effet de la variabilité des milieux dynamiques (Palmquist et al. 2015, Datry et al. 2016d). Étant donné que l'occurrence des perturbations extrêmes des écosystèmes se fait de plus en plus fréquente , il devient urgent de palier à ce manque dans un contexte de changement climatique. ...
... En conséquence, il semble pertinent de ne plus considérer les processus de dispersion et de filtre environnemental comme nécessairement stables Kolasa 2008, Ricklefs 2008) et d'autant plus en milieu dynamique où un relevé à un temps donné sera très peu représentatif du milieu(Palmquist et al. 2015. Au contraire, il apparaît nécessaire d'évaluer conjointement les variabilités spatiales et temporelles de la structure des communautés(Heino and Mykrä 2008, Tonkin et al. 2016b, les étudier séparément risquant d'obscurcir la compréhension des mécanismes à l'oeuvre(Nuvoloni et al. 2016). ...
Thesis
L’écologie vise à comprendre comment les espèces et les individus sont distribués dans l’espace et dans le temps. De nombreux progrès ont récemment été réalisés dans l’analyse spatiale des communautés mais les mécanismes ont souvent été considérés comme à l’équilibre. Cela pourrait affecter la compréhension du rôle de processus déterministes (comme l’influence de facteurs environnementaux locaux) et stochastiques (comme une variation aléatoire du taux de mortalité) dans le maintien des écosystèmes, et d’autant plus pour les milieux dynamiques régulièrement soumis à des changements drastiques des conditions environnementales. Mon objectif dans cette thèse était ainsi d’améliorer la connaissance de la structure spatiale et temporelle des communautés en milieu dynamique en s’appuyant sur le cas de rivières intermittentes, sujettes à des cessation d’écoulement et/ou des assèchements. Le Chapitre 3 a contribué à l’amélioration d’un outil largement utilisé dans les analyses spatiales de communautés. En effet, les tests de Mantel, visant à mesurer la corrélation entre deux matrices de distance et tester sa significativité, sont un outil courant dans l’étude des métacommunautés pour mesurer la contribution du filtre environnemental et de la dispersion. Nous avons montré que la Moran Spectral Randomization permettait de corriger la surestimation de la corrélation entre deux matrices de distance présentant une autocorrélation spatiale. Nos résultats suggèrent que les analyses de communautés fondées sur ces tests ont pu surestimer la part accordée au filtre environnemental dans la structuration des communautés et que notre méthode permettra à de futures analyses de s’affranchir de ce biais. Les tests de Mantel étant utilisé dans d’autres domaines tels que la génétique, l’amélioration que nous proposons pourrait avoir des implications plus larges que le cadre de l’écologie des communautés. Dans le Chapitre 4, j’ai montré que la distribution spatiale et temporelle des assecs pouvait influencer la dynamique des communautés en rivière intermittente. L’influence de la localisation d’une perturbation, définie ici comme un changement extrême des conditions environnementales, sur la structure de communautés avec une dispersion contrainte dans une direction de l’espace avait été testée avec des simulations et sur des mésocosmes mais cela n’avait encore jamais été testé sur des données in situ. Les bassins versants avec des assèchements en amont présentaient des communautés plus dissimilaires entre elles que les communautés de bassins versants avec des assèchements en aval. Les assèchements en amont du réseau ont pu freiner la recolonisation par les organismes aquatiques, conduisant à moins d’homogénéisation des communautés. La fréquence et la durée des assecs avaient également une influence sur la dynamique temporelle des communautés, menant globalement à une plus grande variabilité dans le temps des taxons et des traits biologiques observés à un site, cette relation dépendant par ailleurs de la configuration spatiale d’assèchement. En comparant des rivières françaises exposées à des assèchements depuis des siècles et des rivières tchèques exposées à des assèchements depuis moins d’une décennie et dont l’intermittence est due à des perturbations anthropiques, j’ai montré des différences de réponse des communautés aquatiques à l’intermittence (Chapitre 5). Considérer la diversité bêta spatiale et temporelle a permis de mettre en évidence la plus forte variabilité de composition taxonomique qu’induisent les assèchements sur les communautés des rivières tchèques. De plus, une forte sensibilité des taxons résilients à l’augmentation de l’intermittence ayant été mise en évidence dans le Chapitre 5 pour les rivières qui ne sont pas naturellement intermittentes, l’augmentation de la durée et la longueur des assèchements attendue avec le changement climatique pourraient menacer le rétablissement des communautés après un assec dans ces rivières
... Among these factors, fire suppression following European settlement has significantly reshaped the historical vegetation/fuel structures and fire regimes (with a return interval of two to ten years) and is recognized as the primary cause of decline of longleaf pine ecosystems (Christensen 1981; Glitzenstein et al. 2001). Fire can not only help reduce litter and understory woody components, but also stimulate fire-dependent species, with frequent surface fires particularly benefiting longleaf (Landers et al. 1995;Palmquist et al. 2015;Hammond et al. 2015). There has been a growing interest in the use of prescribed fire in combination with vegetation management practices (e.g., removal of overstory and/or midstory trees) to restore longleaf pine ecosystems through natural regeneration (Brockway and Outcalt 2000;Jose et al. 2006), although it has been proven difficult to achieve in the field (Croker and Boyer 1975). ...
Article
We assessed natural regeneration of longleaf pine (Pinus palustris Mill.) using the data col-lected from the Escambia Experimental Forest in southern Alabama. Fifteen years following the regeneration control, natural regeneration of longleaf remained patchy across a wide range of site/stand conditions, with slightly more than half of all plots containing regeneration, but the den-sity of seedlings and saplings varied significantly. The abundance of seedlings ≤ 1-year-old was positively related to stand age and time since last fire, but negatively related to overstory basal area. The abundance of seedlings and saplings was positively related to stand age, but negatively related to time since last fire and overstory basal area. The probability of achieving ≥ 15,000 seedlings > 1-year-old but ≤ 1-m-tall ha-1 and ≥ 1,250 saplings > 1-m-tall ha-1 was, respectively, positively related to the ratio of time since last fire to overstory basal area and the ratio of quadratic mean diameter to site index. A longer fire interval (> two to three years) should be adopted to naturally regenerate longleaf. We did not find clear zones of exclusion present in natural regeneration even though over-story trees, seedlings and saplings tended to be repulsive spatially and > 80% grass stage seedlings and saplings occurred outside tree crowns.
... Our linear model analyzing understory woody species diversity provides further evidence regarding the influence of canopy species identity on tree regeneration. Environmental heterogeneity has also been documented as an important driver of understory herbaceous diversity in longleaf pine systems (Palmquist et al., 2015). We did observe a negative effect of light availability on diversity, which is contrary to our sapling abundance models. ...
Article
Longleaf pine once dominated much of the forested area of the Coastal Plain of the southeastern United States and is a focal forest type for restoration efforts. In these forests, two species dominate the canopy which may influence tree regeneration. Ultimately, the outcomes of habitat filtering, competition, and disturbance manifest in spatial patterns of tree regeneration, including in longleaf pine ecosystems. Understanding regeneration and establishment patterns can aid in restoration efforts. We ask how the dominant species in an established longleaf pine forest are spatially arranged to provide insight into the mechanisms that may be driving tree establishment in longleaf pine forests. We found that longleaf pine saplings were more likely to be found near the other dominant tree species in this forest, turkey oak trees, than in gaps or near to longleaf pine trees. Similarly, turkey oak saplings clustered around turkey oak trees but were dispersed in relation to longleaf pine trees. These findings point towards the interplay between canopy tree composition, leaf litter, and fire behavior as driving mechanisms in the successful establishment of both pines and oaks in this forest.
... Fire is an ancient and recurrent disturbance factor of Earth's terrestrial ecosystems including those of western and southeastern United States (Palmquist et al. 2015, Noss 2018, Pausas and Keeley 2019. Fire influences the ecology of ecosystems and alters which species survive and flourish. ...
Article
Full-text available
Worldwide, humans are altering the fire regimes (fire‐return intervals, severity, seasonality) of fire‐prone ecosystems, fragmenting natural landscapes, and altering climates. Efforts to restore fire regimes in natural areas are usually guided by fire management plans (FMP) that rely on prescribed burning. Despite the common use of FMPs, limited efforts have gone to assessing vegetative and faunal responses. While some insights into responses to fire from short‐term studies are available, there is less knowledge about FMP outcomes applied over multiple decades. Peninsular Florida hosts many fire‐prone communities including globally threatened Florida scrub. We repeatedly sampled species composition and abundance during a 38‐yr. period (1977‐2015) at 11 upland stands including wet prairies, flatwoods, and Florida scrub (i.e., oak scrub, rosemary scrub, hickory scrub). A total of 22 fires impacted the stands with individual stands experiencing as few as 2 and up to 11 fires. There was no invasion of non‐native plants following fires and stands showed remarkable ecological resilience with limited change in species composition and modest abundance shifts due to differential recovery rates. Importantly, ecological resilience was not eroded with repeated fire. Species richness, evenness, and diversity were significantly higher at the end vs. beginning of our study, suggesting that current fire regimes are within the range of those with which species evolved. While resprouting shrubs and trees persevered under a range of fire‐return intervals, an obligate‐seeding shrub was adversely impacted by fires more frequent than FMP‐recommended intervals. Several lessons are apparent from our findings. Best practice should: use the full range of fire‐return variation at evolutionarily appropriate intervals; utilize natural fire seasons to the extent possible; compare outcomes and modify prescriptions as necessary to meet FMP goals; give special consideration to burn units with embedded associations of widely differing fire regimes than principal associations in order to maintain diversity of both associations; and account for climate change in FMPs since fire behavior, frequency, and community responses to fire will likely change in the coming decades. Outcomes of FMPs must be carefully evaluated to ensure prescriptions are evolutionarily appropriate and to ameliorate the impacts of altered climates.
... Forest Service photograph by Don C. Bragg. relatively stable understory communities over time under a regime of frequent fire (Palmquist et al., 2015). When trees need to be regenerated and recruited under a frequent, low-intensity fire regime, uneven-aged silviculture methods including single-tree and group selection can be applied. ...
Article
Full-text available
Fire-sustained open oak and pine forests were once widespread across eastern North America, but are now comparatively scarce. To regain the goods and services of these open forests, managers are increasingly looking to restore them with the silvicultural systems and tools best suited to meet their objectives. Hence, we synthesized a number of research efforts and case studies from open pine, mixedwood, and oak-dominated forests in eastern North America to demonstrate the silvicultural options available and recognized knowledge gaps. The silvicultural treatment options and tools available are very similar to those applied in closed-canopy forests, even if the objectives are fundamentally different. For instance, while conventional practices in naturally regenerated forests concentrate on managing closed tree canopies to increase periodic yields and encourage new tree recruitment, open forest silviculture focuses on the maintenance of a vertically simple and understocked canopy to facilitate a robust herbaceous groundflora and limit woody plant regeneration. To achieve and sustain this understocked condition, open forest management applies multiple tools (e.g., prescribed fire, periodic harvests or deadenings, and herbicide use and planting if and when needed) along with other understory enhancement and maintenance treatments. This review demonstrates that while we have learned much about open forest silviculture over the decades, many information gaps and challenges for managers remain.
... In the fire-adapted Longleaf woodlands, in contrast, we observe that the constraints on species richness are highly scale-dependent. At large scales, species richness for shade-intolerant coastal species is highly dependent on soil moisture (including soil texture and depth to water table) and light resources (Palmquist et al. 2015). However, with decreasing spatial scale, the effects of SF and CC decrease substantially. ...
Article
Full-text available
The goal of elucidating the primary mechanisms constraining the assembly and distribution of biodiversity remains among the central unresolved challenges facing the field of ecology. Simulation studies and experimental manipulations have focused on how patterns in community assembly result from bivariate relationships along productivity or environmental gradients. However, the joint influence of multiple resource gradients on the distribution of species richness in natural communities remains under‐studied. Using data from a large network of multi‐scale vegetation plots across forests and woodlands of the southeastern US, we find significant evidence for the scale dependent, joint constraints of forest structure and soil resources on the distribution of vascular plant species richness. In addition to their significant partial effects on species richness, understory light levels and soil fertility positively interact, suggesting a trade‐off between the two limiting resources with species richness peaking both in high‐light, low‐fertility conditions as well as low‐light, high‐fertility settings. This finding provides a novel perspective on the biodiversity‐productivity relationship that suggests a transition in limiting resources from soil nutrients to light availability when enhanced productivity results in reduced light resources for subordinate individuals. Results likewise have meaningful implications for our understanding of scale dependent community assembly processes as size‐asymmetric competition replaces environmental filtering as the primary assembly mechanism structuring temperate forest communities along an increasing soil fertility gradient.
... It is important to note that the Ultisol sites in this study were all well drained, whereas other studies in the region have examined plant community change on wetter Ultisol sites, which may be more vulnerable to shrub encroachment (Palmquist et al., 2014). Within-site edaphic heterogeneity is another known correlate of whole-site plant species richness in NACP savannas (Costanza, Moody, & Peet, 2011;Palmquist, Peet, & Mitchell, 2015;Provencher, Litt, & Gordon, 2003) and was likely important on our study sites. ...
Article
Full-text available
Aim To assess the drivers of plant functional group richness and beta diversity in fire‐maintained North American Coastal Plain (NACP) savannas. Location The southern portion of the NACP, a global biodiversity hotspot. This region is characterized by fire‐dependent pine savanna fragments that are isolated within a matrix of agriculture, urban development, non‐pyrogenic plant communities and plantation forestry. Methods We used nested quadrats to sample plant species on 30 fire‐maintained savanna preserves in Florida and Georgia, USA. We analysed between‐site Sørenson dissimilarity, a measure of beta diversity, using NMDS and PerMANOVA. We measured nestedness using NODF, and we used Betapart to partition Sørenson dissimilarity into nestedness and turnover components. We used linear and generalized linear mixed models to explore drivers of functional group richness and composition, including fire regime (return interval, number of fires, time since fire and seasonality), vegetation structure (herbaceous cover, woody cover and tree density) and spatial factors (surrounding landscape and geographic distance). Results We found turnover‐dominated beta‐diversity patterns in all functional groups. Turnover was explained partly by spatial and environmental gradients, but roughly half of the turnover between sites was unexplained. Species richness was higher on sites where fire and fire surrogates had been used longer and more consistently, and these effects were partly independent of current vegetation structure. Fire regimes containing more growing season fire and more diversity of burn seasons promoted higher species richness. Relationships between small‐scale and large‐scale species richness varied by soil type and functional group. Main conclusions Fire‐maintained savannas in the southeastern NACP vary greatly in their plant functional group richness, but high beta diversity resulting primarily from species turnover suggests that even species‐poor sites can harbour less‐common members of the regional plant metacommunity. Prescribed fire regimes that include growing season fire as well as a diversity of burn seasons may best promote species and functional group richness.
Article
Although the influence of mechanical sand control techniques on vegetation on degraded sand dunes has received much attention for a long time, our understanding of the underlying processes of plant community formation is not yet strong. Analyses of trends in the turnover and nestedness components of beta diversity can be informative of the process response to sand dune stabilization. In this study, we applied a field investigation of sand dunes at different stabilization stages in northern China. We then used Baselga's method to calculate total beta diversity and partition it into spatial turnover and nestedness to clarify the underlying community change processes. Our study indicated that the implementation of straw checkerboard techniques significantly increased the beta diversity of the community, leading to species heterogeneity. Species turnover was the main contributor to species beta diversity, and it was caused by high environmental heterogeneity with sand dune stabilization. The contribution of the nestedness component to beta diversity decreased with dune stabilization, indicating that with stabilization, dissimilarity was largely driven by variation in community composition rather than differences in taxonomic richness. This study provides insights into the underlying variation in species composition among sites where mechanical sand fixation engineering has been implemented and important guidelines for combating desertification. This article is protected by copyright. All rights reserved.
Chapter
Southeastern pineland ecosystems fall into two broad groups: ecosystems with a grassy understory that depend on frequent low-intensity fires to maintain structure and biodiversity [longleaf (Pinus palustris), south Florida slash (P. densa), shortleaf (P. echinata)], and ecosystems with a shrubby understory that burn less frequently with higher intensity [pond pine (Pinus serotina) pocosin, sand pine (Pinus clausa) scrub]. We review the fire ecology and management of these ecosystems, covering weather, climate, fuels, and fire; historical fire regimes; fire associated tree mortality; fire dependency and postfire recovery; wildlife response to fire; and fire and ecological restoration. Finally, we discuss likely impacts of future climate changes and mitigation strategies. Maintaining fire frequency appropriate to each ecosystem and utilizing all burn opportunities will be even more critical in an era of climate change.
Article
Full-text available
Biodiversity hotspots are conservation priorities. We identify the North American Coastal Plain (NACP) as a global hotspot based on the classic definition, a region with > 1500 endemic plant species and > 70% habitat loss. This region has been bypassed in prior designations due to misconceptions and myths about its ecology and history. These fallacies include: (1) young age of the NACP, climatic instability over time and submergence during high sea-level stands; (2) climatic and environmental homogeneity; (3) closed forest as the climax vegetation; and (4) fire regimes that are mostly anthropogenic. We show that the NACP is older and more climatically stable than usually assumed, spatially heterogeneous and extremely rich in species and endemics for its range of latitude, especially within pine savannas and other mostly herbaceous and fire-dependent communities. We suspect systematic biases and misconceptions, in addition to missing information, obscure the existence of similarly biologically significant regions world-wide. Potential solutions to this problem include (1) increased field biological surveys and taxonomic determinations, especially within grassy biomes and regions with low soil fertility, which tend to have much overlooked biodiversity; (2) more research on the climatic refugium role of hotspots, given that regions of high endemism often coincide with regions with low velocity of climate change; (3) in low-lying coastal regions, consideration of the heterogeneity in land area generated by historically fluctuating sea levels, which likely enhanced opportunities for evolution of endemic species; and (4) immediate actions to establish new protected areas and implement science-based management to restore evolutionary environmental conditions in newly recognized hotspots.
Article
Full-text available
In this study, we used a hierarchical, multifactor ecological classification system to examine how spatial patterns of biodiversity develop in one of the most species-rich ecosystems in North America, the fire-maintained longleaf pine-wiregrass ecosystem and associated depressional wetlands and riparian forests. Our goal was to determine which landscape features are important controls on species richness, to establish how these constraints are expressed at different levels of organization, and to identify hotspots of biological diversity for a particular locality. We examine the following questions: 1) How is the variance in patterns of plant species richness and diversity partitioned at different scales, or classification units, of the hierarchical ecosystem classification developed for the study area? 2) What are the compositional similarities among ecosystem types? 3) For our study area, what are the sites expected to harbour highest species richness? We used a spatially explicit map of biodiversity to project abundance of species-rich communities in the landscape based on a previously developed ecological classification system for a lower Gulf Coastal Plain landscape. The data indicate that high species richness in this ecosystem was found in sites with frequent fire and high soil moisture. Sites in fire-maintained landscapes with lower frequency of fire were associated with geomorphological characteristics, suggesting a dependence of the diversity-disturbance relationship with soil type. With more frequent fire on some sites, high diversity shifts from canopy component to ground flora, with an overall increase in total species richness. Our approach demonstrates how potential species richness can be identified as a restoration goal and that multiple vegetation endpoints may be appropriate vegetation objectives. We identify basic management needs for the maintenance of biodiversity in this ecosystem that can be derived from an understanding of the combination of factors that most strongly predict diverse plant communities.
Article
Forgotten Grasslands of the South" is a literary and scientific case study of some of the biologically richest and most endangered ecosystems in North America. Eminent ecologist Reed Noss tells the story of how southern grasslands arose and persisted over time and addresses questions that are fundamental for conserving these vital yet poorly understood ecosystems. The author examines: the natural history of southern grasslandstheir origin and history (geologic, vegetation, and human)biological hotspots and endangered ecosystemsphysical determinants of grassland distribution, including ecology, soils, landform, and hydrology fire, herbivores, and ecological interactions.The final chapter presents a general conservation strategy for southern grasslands, including prioritization, protection, restoration, and management. Also included are examples of ongoing restoration projects, along with a prognosis for the future. In addition to offering fascinating new information about these little-studied ecosystems, Noss demonstrates how natural history is central to the practice of conservation. Natural history has been on a declining trajectory for decades, as theory and experimentation have dominated the field of ecology. Ecologists are coming to realize that these divergent approaches are in fact complementary, and that pursuing them together can bring greater knowledge and understanding of how the natural world works and how we can best conserve it. "Forgotten Grasslands of the South" explores the overarching importance of ecological processes in maintaining healthy ecosystems, and is the first book of its kind to apply natural history, in a modern, comprehensive sense, to the conservation of biodiversity across a broad region. It sets a new standard for scientific literature and is essential reading not only for those who study and work to conserve the grasslands of the South but also for everyone who is fascinated by the natural world.