Content uploaded by Kyle A. Palmquist
Author content
All content in this area was uploaded by Kyle A. Palmquist on Nov 13, 2017
Content may be subject to copyright.
Journal of Vegetation Science
&&
(2014)
Changes in plant species richness following reduced
fire frequency and drought in one of the most species-
rich savannas in North America
Kyle A. Palmquist, Robert K. Peet & Alan S. Weakley
Keywords
Biodiversity; Grassland; Longleaf pine
ecosystem; Prescribed fire; Spatial scale;
Species frequency; Water stress
Nomenclature
Weakley (2012)
Received 29 October 2013
Accepted 4 March 2014
Co-ordinating Editor: John Morgan
Palmquist, K.A. (corresponding author,
kapalmqu@yahoo.com): Curriculum for the
Environment and Ecology, University of North
Carolina at Chapel Hill, Chapel Hill, NC, 27599-
3275, USA
Peet, R.K. (peet@unc.edu) & Weakley,A.S.
(weakley@unc.edu): Department of Biology,
University of North Carolinaat Chapel Hill,
Chapel Hill, NC, 27599-3280, USA
Abstract
Questions: How has plant species richness changed over two decades in one of
the most species-rich savannas in North America? Is an altered disturbance
regime, environmental stress, or both, driving these changes? In what ways can
observations in this savanna inform management of other species-rich plant
communities?
Location: Longleaf pine savannas in southeast North Carolina, USA.
Methods: In 2011–2013, we re-surveyed permanent plots established in the
1980s and 1990s in a longleaf pine (Pinus palustris) savanna in North Carolina to
quantify changes in species richness at multiple spatial scales following 15 yr of
reduced fire frequency plus periodic drought. For comparison, we re-sampled
other longleaf pine savannas in the region that had not experienced reduced fire
frequency, but had experienced similar drought. In addition, we identified
which types of species were lost and gained, and summarized changes in species
frequency by growth form, plant height, and habitat affinity.
Results: We detected substantial declines in small-scale species richness and
species frequency from the 1980s to 2011, representing a loss of 33% to 41% of
the flora, depending on the spatial scale. Small herbaceous species had become
particularly scarce. Additional re-sampling in the wetter years of 2012 and 2013
after consecutive years of fire revealed that species richness had increased
slightly from 2011, but was still considerably lower than that in the 1980s. Other
savannas did not exhibit such dramatic declines in species richness, suggesting
reduced fire frequency in addition to drought contributed to species loss in Big
Island Savanna over time.
Conclusions: Our work suggests that nearly annual fire is necessary for the
maintenance of high plant species richness in mesic longleaf pine savannas, and
even a modest reduction in fire frequency can have dramatic negative impacts.
This study also suggests that drought is an important factor structuring grassland
ecosystems in the southeastern US, despite relatively high regional precipitation.
We believe these findings can be generalized to other species-rich grasslands that
are sensitive to changes in disturbance regimes and may require frequent distur-
bance to maintain plant species richness.
Introduction
In species-rich grassland ecosystems, natural disturbance
(e.g. fire, grazing) or processes that mimic natural distur-
bance through the removal of above-ground biomass (e.g.
mowing) are essential for the maintenance of species rich-
ness, community structure, and rare species (Collins et al.
1998; Glitzenstein et al. 2003; Fidelis 2010; Peet et al.
2014). Frequent disturbance generally increases species
richness by reducing the abundance of dominant species,
increasing resource availability in the form of light, space
and nutrients (Kirkman et al. 2004), resulting in a shift
from asymmetric competition for light to more symmetric
below-ground competition (Peet & Christensen 1988; Wil-
1
Journal of Vegetation Science
Doi: 10.1111/jvs.12186©2014 International Association for Vegetation Science
son et al. 2012; Peet et al. 2014). In addition to distur-
bance, grassland community structure is influenced by
seasonal, periodic, and multi-year drought events (Gibson
& Hulbert 1987), which can have prolonged consequences
(Haddad et al. 2002). At local scales, drought can result in
declines in plant biomass and species richness, as well as
shifts in community composition (Tilman & Haddi 1992;
O’Connor 1995; Haddad et al. 2002; Cheng et al. 2011).
Many species-rich grassland ecosystems are experienc-
ing anthropogenic alteration of disturbance regimes
(changes in frequency, timing, intensity or severity of dis-
turbance) that have long-term impacts on ecosystem struc-
ture and function. These impacts vary depending on the
magnitude and direction of alteration, but may result in the
loss of species richness and changes in community compo-
sition and stand structure (Heyward 1939; Belsky 1992;
Glitzenstein et al. 2003). In addition, changes in distur-
bance regimes are often compounded with habitat destruc-
tion, fragmentation or altered environmental conditions
(e.g. nitrogen deposition, climate change), resulting in
additional pressure on grassland plant communities (Leach
& Givnish 1996; Stevens et al. 2011; Potts et al. 2012).
Longleaf pine (Pinus palustris) savannas are fire-depen-
dent, species-rich grasslands located in the southeastern
US, currently influenced by multiple stressors (e.g. fire
suppression, drought, habitat destruction, and habitat frag-
mentation). Habitat conversion and long-term fire sup-
pression have collectively reduced the longleaf pine
ecosystem to only 2–3% of its acreage at the time of Euro-
pean settlement (Outcalt & Sheffield 1996; Frost 2006).
Plant species richness within the herbaceous layer can be
exceptionally high, and at small scales represents the high-
est values ever recorded in North America (52 species in
1m
2
; Walker & Peet 1983; Peet et al. 2012, 2014) and
approaches world-record levels (Wilson et al. 2012). Fire is
an important factor responsible for the maintenance of
species richness within longleaf pine savannas, and is
essential for the survival of small-statured species within
the dense grass matrix (Walker & Peet 1983; Glitzenstein
et al. 2003; Kirkman et al. 2004). Drought events occur
periodically in longleaf pine savannas, most often in early
spring (March–May; Noss 2013), although little research
has investigated how drought influences species richness
in these systems (but see Myers & Harms 2011), despite
recognition of the importance of drought in other grass-
land ecosystems (Cleland et al. 2013). In addition to peri-
odic water stress, the southeastern Coastal Plain of the US
in which these systems are embedded has been experienc-
ing ongoing, long-term drought over the last 25 yr
(Fig. 1). We believe that both periodic and multi-year
drought events may be under-appreciated, yet important
drivers of community structure in longleaf pine savannas.
We used a unique, long-term, multi-scale data set from
Big Island Savanna of the Green Swamp Preserve, NC, to
explore how drought and fire regime have shaped plant
species richness over time. Big Island Savanna has been
considered one of the most species-rich and high-quality
longleaf pine savannas on the Coastal Plain of the south-
eastern US (McIver 1981; Frost et al. 1986). This site is the
source of the North American records of 42 species in
0.25 m
2
and 52 species in 1 m
2
(see Peet et al. 2012,
2014), values which rival those reported for other species-
rich grasslands throughout the world (Kull & Zobel 1991;
Cantero et al. 1999; Klime
s et al. 2001; Dengler et al.
2009, 2012; Wilson et al. 2012). The existence of long-
term plot records that span multiple spatial scales presents
(a) (b)
Sampling years
Other years
Sampling years
Other years
Fig. 1. (a) Mean annual Palmer Drought Severity Index (PDSI) and (b) mean annual Palmer Z Index (PZI) from 1970–2013 for the southeastern Coastal Plain
of North Carolina. PDSI and PZI values below 0 indicate drought years, whereas values above zero indicate non-drought years. PDSI quantifies long-term
drought conditions, while PZI reflects short-term changes in water availability. The number of drought events has increased since 1980, as indicated by the
black linear best-fit lines and a larger proportion of years below the dotted line.
Journal of Vegetation Science
2Doi: 10.1111/jvs.12186 ©2014 International Association for Vegetation Science
Plant speciesloss in a species-rich savanna K.A. Palmquist et al.
a unique opportunity to document changes in richness
and composition. Big Island Savanna experienced a
planned shift in its fire management regime from nearly
annual fire from at least 1940 to 1997 (Kologiski 1977;
Rome 1988) to a fire-return interval of 2–3 yr during
1997–2011. Although fire frequency has changed only
modestly, this shift may have had significant impacts on
species richness as Big Island Savanna had for decades
before experienced a constant fire management regime.
To explore the impact of reduced fire frequency and
drought on plant species richness in Big Island Savanna,
we re-sampled two sets of permanent vegetation plots
established over 25 and 20 yr ago. For comparative pur-
poses and to explore the impact of drought on other long-
leaf pine savannas, we also re-sampled plots on other
species-rich longleaf pine savannas in the region that had
not experienced reduced fire frequency. Specifically, we
asked:
1. Have species richness and species frequency in Big
Island Savanna changed in response to reduced fire
frequency and drought?
2. What is the relative importance of reduced fire fre-
quency vs drought in driving changes in species rich-
ness and frequency in Big Island Savanna?
3. In what ways can the changes observed in this savanna
inform management of other species-rich grasslands?
It is critical that we assess how changes in long-standing
fire regimes, compounded with additional stress from
drought have influenced plant species richness in this
savanna, which, unlike most other longleaf pine sites, did
not experience post-colonial fire suppression. From a con-
servation perspective, Big Island Savanna is irreplaceable
within the greater landscape. Moreover, this work has
implications not only for fire managers in the longleaf pine
ecosystem, but for managers and researchers who study
other chronically disturbed, species-rich grassland ecosys-
tems.
Methods
Study area
The Green Swamp Preserve is located in Brunswick
County, in the southeastern corner of the North Carolina
Coastal Plain (34°50N, 78°180W) and covers ca.
6700 ha. The majority of the site consists of shrub-domi-
nated ombrotrophic peatland (pocosin), within which
occur scattered islands of savanna on mineral soil (Kolog-
iski 1977). Elevation ranges from 12 to 25 m a.s.l., with
very little topographic relief. However, small differences
in elevation (>0.2 m) have major consequences for
hydrology, soil properties, and hence vegetation (Rome
1988; Christensen 2000). The climate is humid sub-tropi-
cal, with an average mean annual temperature of
15.5 °C and an average annual precipitation of 160 cm,
most of which occurs during the growing season (Ruffner
1985; State Climate Office of North Carolina). Droughts
occur periodically in the region (mostly during the
months of March–May) and result in at least temporary
loss or dormancy of species dependent on moist soils
(Kologiski 1977; Christensen 1981). Savanna soils in the
Green Swamp are typically Leon series, which are
derived from acidic, fine-textured, nutrient-poor, marine
sediments. They are generally poorly drained, with the
water table within 25 cm of the soil surface for 1–4mo
of the year (Kologiski 1977).
This study focused primarily on a single, 30-ha savanna
in the Green Swamp: Big Island Savanna. During the
18th–19th centuries, Big Island Savanna likely burned
almost annually due to the flatness of the landscape, the
large size of the fire compartment, and the flammability of
the vegetation (Frost 2006). Written records indicate
annual, late-winter fire was implemented in Big Island
Savanna for much of the 20th century, from at least and
likely prior to 1940 through 1997 (Kologiski 1977; Rome
1988). Thereafter, the fire management regime shifted to a
mix of growing-season and dormant-season fires with a
return interval of 2–3 yr. Thus, over the 15 yr prior to this
study, fire frequency was lower than had been the case
with the traditional management strategy. This shift
reflected an effort by managers to return to what has
recently been perceived as a more ‘natural’ fire regime
with somewhat less frequent fires timed later in the grow-
ing season when natural ignition is more likely (Frost et al.
1986; Huffman 2006).
We also examined long-term changes in species richness
and composition on other sites besides Big Island Savanna,
including other savannas in the Green Swamp Preserve,
Holly Shelter Game Lands, Croatan National Forest and
Camp Lejeune Marine Corp Base, to determine how long-
term drought has impacted savanna vegetation on similar
sites in the region (Fig. S1.1, in Appendix S1). All of these
sites have experienced long-term drought, but not reduced
fire frequency. However, some sites have likely experi-
enced somewhat more frequent fire over the last 20 yr rel-
ative to the original fire management regime (e.g. Camp
Lejeune, Croatan National Forest), although long-term fire
history data for these sites are lacking. In contrast, fire fre-
quency has remained relatively constant over time on
other sites (other savannas in the Green Swamp Preserve,
Holly Shelter Game Lands). Here, we focus on comparing
changes in species richness and frequency patterns in Big
Island Savanna to other environmentally similar sites that
have experienced drought but not changes in fire fre-
quency (other savannas in the Green Swamp and Holly
Shelter Game Lands). We more thoroughly explore pat-
terns of species richness and composition over time in rela-
3
Journal of Vegetation Science
Doi: 10.1111/jvs.12186©2014 International Association for Vegetation Science
K.A. Palmquist et al. Plant species loss in a species-rich savanna
tion to fire frequency and environmental context on these
comparison sites in another study (Palmquist 2014).
Sampling design
Two different sampling protocols were used to examine
changes in species richness and composition over time in
Big Island Savanna, which reflect the designs of the two
separate studies that we re-sampled. The first study (study
1; see Sykes et al. 1994) was based on plots established and
surveyed annually during June of 1985–1989, with a sixth
sample in 1994. In June 2011, we re-sampled six of the ori-
ginal 2.5-m
2
plots (see Sykes et al. 1994 for details of plot
configuration), each of which contained 10 0.25 m
2
sub-
plots. Three of the six plots were control plots and three
had received sugar additions twice a year to reduce nutri-
ent availability during 1985–1989. The sugar additions had
no noticeable impact. Each plot was permanently marked
in 1985 with steel conduit and each 0.25 m
2
subplot was
delineated with steel nails, which ensured we sampled the
same physical location over time. Within each subplot,
vascular plant species presence was recorded in five per-
manent 0.01 m
2
and five permanent 0.001 m
2
plots. This
resulted in 300 observations of 0.001 m
2
, 300 observations
of 0.01 m
2
, 60 observations of 0.25 m
2
, 12 observations of
1m
2
and six observations of 2.5 m
2
. For consistency, plots
were re-sampled in June, 4–5mo after fire during all
sampling years.
The second study (study 2) was conducted in Big Island
Savanna in June of 1993 by the Carolina Vegetation Sur-
vey (CVS) using the protocol described in Peet et al. (1998,
2012). In June 2011, four CVS plots were re-located and
re-sampled. CVS plots are 1000 m
2
(20 950 m) with
smaller sub plots nested within. Similar to study 1, each
CVS plot was permanently marked with ten pieces of steel
conduit. Once the plot was re-located, presence of all vas-
cular plant species was recorded at seven spatial scales in
permanent subplots (0.01, 0.1, 1, 10, 100, 400 and
1000 m
2
; see Peet et al. 1998, 2012 for details of plot
layout). All plots in Big Island Savanna were located ca.
50–300 m from one another. In addition, twenty-two
1000 m
2
CVS plots established during 1991–1993 on other
sites with similar soils and species composition were
re-sampled in 2009–2010 (study 3). All 26 CVS plots are
archived in VegBank (http://vegbank.org/cite/VB.ds.
199852.Palmquist2014GreenSwamp).
Statistical analysis
Prior to analysis, all taxonomic names were standardized
across the sampling years to ensure that changes in
nomenclature, taxonomic resolution and taxonomic
understanding of the flora across time were not affecting
the number or identity of species detected. Species richness
was calculated at each spatial scale for all three data sets
(study 1: 0.001, 0.01, 0.25, 1 and 2.5 m
2
; study 2 and study
3: 0.01, 0.1, 1, 10, 100, 400 and 1000 m
2
). We used the
same analytical methods for all three data sets, but analy-
sed them separately. Linear models and linear mixed
effects models were used to detect significant changes in
species richness at each spatial scale in 2011 relative to
richness at each spatial scale during 1985–1994. Random
intercepts models, a type of mixed effects model, were used
to examine richness at all spatial scales, except the full plot
size, as multiple estimates of richness for these scales were
drawn from the same plot. This modelling approach
accounted for spatial auto-correlation caused by the nested
nature of the data (Zuur et al. 2009). In each random
intercept model, species richness in the 1980s was
regressed against species richness in 2011–2013 using an
offset function. The unique plot identifier for each subplot
was set as a random effect, to account for spatial autocorre-
lation between subplots in the same plot. Linear models
were used to examine changes in richness over time for
the full plot (2.5 and 1000 m
2
, respectively, for study 1
and studies 2 and 3). All statistical analyses were per-
formed in R v 2.15.2 using the nlme package (R Founda-
tion for Statistical Computing, Vienna, AT). Results
reported for Big Island Savanna at 2.5 m
2
and below were
calculated from study 1, whereas results reported at scales
>2.5 m
2
were from study 2.
To quantify drought over the long term and to identify
individual drought years, we obtained monthly Palmer
Drought Severity Index (PDSI) and monthly Palmer Z
Index (PZI) data for 1970–2013 from the Southeastern
Coastal Plain of North Carolina (National Climatic Data
Center 2013). PDSI quantifies the duration and strength of
long-term drought, whereas PZI is more sensitive to short-
term pulses of water and reflects whether moisture condi-
tions deviate from normal (short-term drought). We
identified the sampling years of 1985, 1986, and 2011 as
significant drought years (Fig. 1). To determine if changes
in richness across time were due to individual drought
years, we examined whether there were significant differ-
ences between richness values in drought and non-
drought years. We also explored whether species richness
values from 2011 were lower than richness in early
drought years, which would suggest that other factors (e.g.
reduced fire frequency) had influenced species richness
over time.
In addition to examining changes in species richness,
we investigated which species were lost and gained
over time by tabulating the total number of times a
species occurred at each spatial scale in every year. We
summarized this information as both the mean number
of species and as the percentage of subplots occupied by
Journal of Vegetation Science
4Doi: 10.1111/jvs.12186 ©2014 International Association for Vegetation Science
Plant speciesloss in a species-rich savanna K.A. Palmquist et al.
each species in each year. We compared the identity
and frequency of species lost in earlier drought years to
those lost in 2011. We expected insectivores in particu-
lar to decrease over time in response to drought, as
they have been shown to be particularly sensitive to
drought in other longleaf pine studies (e.g. Folkerts
1982) and became substantially less abundant during
the drought years of 1985 and 1986 in our data set.
We also expected ‘wet’ and ‘mesic’ species to be lost to
a greater extent if drought alone was responsible for
changes in species richness and identity, as ‘dry’ species
could likely tolerate and survive drought (see Debinski
et al. 2013). In addition, we expected shrubs, trees and
vines to increase over time in abundance and frequency
in response to reduced fire frequency, and rosette
herbs, geophytes and other small herbaceous species to
decrease, as these groups have previously been shown
to be sensitive to fire suppression (Glitzenstein et al.
2003, 2012). To identify which types of species became
more or less frequent over time and whether those
changes were related to drought or reduced fire fre-
quency, we classified species in three ways. First, we
assigned species to one of 12 mutually exclusive growth
form categories (caulescent herb, matrix graminoid,
fern, geophyte, hemiparasite, insectivore, legume,
rosette herb, shrub, single-culm graminoid, subshrub,
tree and vine; see Table S2.1 in Appendix S2 for
growth form definitions) and then examined how the
frequency of each growth form changed over time. Sec-
ond, we categorized species according to their maxi-
mum height (short =most plant growth below 4 dm;
tall =most plant growth above 4 dm) to examine
whether short-statured species were preferentially lost
over time, suggestive of competitive exclusion caused
by reduced fire frequency. Third, to determine whether
species with environmental optima in mesic or wet
environments were preferentially lost relative to species
with optima in dry environments, we assigned species
to a categorical habitat optimum (dry, mesic, wet) based
on Weakley (2012) and our own knowledge of the 96
species in the data set.
To further disentangle the impacts of drought and
reduced fire frequency on changes in species richness and
frequency over time, we used two approaches. First, we
quantified changes in richness over time at other sites on
the Southeastern Coastal Plain in North Carolina that have
also recently experienced long-term drought. These sites
are similar to Big Island Savanna in that they occur on Ulti-
sol soils and have similar hydrologic and soil properties,
but differ in that they have not experienced reduced fire
frequency in the last 20 yr. In contrast to Big Island
Savanna, they have generally been burned consistently or
somewhat more frequently (every 2–4 yr) over the last
20 yr relative to the previous fire management regime. If
changes in species richness over time in other sites were
similar to those in Big Island Savanna, that would suggest
long-term drought had strong effects on species richness in
the region. Second, we re-sampled all plots in Big Island
Savanna again in 2012 and 2013, which were wetter years
than 2011 (Fig. 1). In addition, annual fire was returned to
Big Island Savanna in 2011–2013, specifically to facilitate
this study. Thus, the re-sampling events in 2012 and 2013
allowed us to assess the extent of species richness recovery,
if any, after consecutive years of fire and increased water
availability.
Results
Species richness and frequency patterns over time
At all spatial scales, species richness in Big Island Savanna
was lower in 2011 compared to all other sampling years,
and significantly lower for small spatial scales (≤2.5 m
2
;
Fig. 2, Table 1). These declines at small scales are excep-
tional and represent a loss of between 32.7% and 40.8% of
the flora, depending on the spatial scale in question
(Table 1). In contrast to small scales, richness at larger spa-
tial scales (≥10 m
2
) declined less substantially over time,
representing losses of 1.2–14.7% of plant species (Table 1).
In 2011, most species had become less frequent in
subplots in Big Island Savanna, although these declines
were most extreme at ≤1m
2
(Tables S2.1, S2.2, S2.3 in
Appendix S2). In particular, small-statured, herbaceous
20 30 40 50 60
Year
Mean richness (m2)
1985 1990 1995 2000 2005 2010 2015
0510
0.001 m2
0.01 m2
0.25 m2
1 m2
2.5 m2
Fig. 2. Mean species richness at each spatial scale in Big Island Savanna
during 1985–1989, 1994, and 2011–2013. Richness has decreased
significantly from 1994 to 2011 at all scales below 10 m
2
.Although,we
plot the linear trajectory between 1989–1994 and 1994–2011 (denoted
with a dashed line), the variation in species richness within these intervals
is unknown.
5
Journal of Vegetation Science
Doi: 10.1111/jvs.12186©2014 International Association for Vegetation Science
K.A. Palmquist et al. Plant species loss in a species-rich savanna
species exhibited the largest decrease: insectivores (1.2 of
11.5 species lost on average at 0.25 m
2
), single-culm
graminoids (2.3 of 11.5 species lost on average), rosette
herbs (4.3 of 11.5 species lost on average), and caulescent
herbs (1.8 of 11.5 species lost on average; Tables 2, S2.2 in
Appendix S2). Geophytes, matrix graminoids and hemi-
parasites also decreased in frequency in 2011 relative to
earlier years, but less substantially (0.6, 0.8 and 0.2 species
lost on average, respectively; Tables 2, S2.2 in Appendix
S2). Legumes and shrubs increased slightly in 2011, while
tree species remained constant in frequency (Tables 2,
S2.1, S2.2, S2.3 in Appendix S2). However, P. palustris
seedlings shifted from occupying 21.7% of 0.25-m
2
sub-
plots in 1985 to 3.3% of subplots in 2011 (Table S2.2 in
Appendix S2). Both short (<4dm)andtall(>4 dm) plant
species became less frequent in 2011, but short plants
became particularly scarce (8.6 short species lost on aver-
age vs 3.5 tall species lost on average; Table 2). In contrast
to expectation, species with their habitat optimum in wet
environments decreased very little over time, whereas spe-
cies with their habitat optimum in mesic and dry habitats
became substantially less frequent (9.3 and 1.6 species lost,
respectively; Table 2).
Species richness and frequency indrought vs non-
drought years
To determine whether the loss of species across time was
due to the 2011 drought, we compared patterns of species
richness and frequency from 2011 to the earlier drought
years of 1985 and 1986. Mean species richness was sub-
stantially lower in 2011 than in either 1985 or 1986 at
small scales, 0.01 m
2
(5.1 vs 10.0), 1 m
2
(26.4 vs 40.8) and
2.5 m
2
(33.6 vs 47.2; Figs 2 and 3). In fact, mean richness
at small scales during 1985 and 1986 was more similar to
richness in non-drought years (1987, 1988, 1989, 1994)
than richness in 2011 (Figs 2 and 3). The percentage of
subplots occupied by most species in 2011 was consider-
ably lower than earlier drought years, especially for rosette
herbs and single-culm graminoids (Tables 2, S2.1, S2.2,
S2.3 in Appendix S2). Species lost during the droughts of
1985 and 1986 spanned several growth form categories
and included both tall shrub species (e.g. Ilex glabra,Morella
carolinensis) and smaller-statured herbaceous species (e.g.
Viola primulifolia, Polygala lutea). Insectivores decreased
dramatically during both the 2011 and 1985–1986
droughts, suggesting these species are more sensitive to
short-term changes in moisture availability than other
species in longleaf pine savannas.
Disentangling the effects of long-term drought and
reduced fire frequency
To parse the relative effects of long-term drought and
reduced fire frequency, we compared changes in species
richness over time at other sites in the Southeastern
Coastal Plain of North Carolina to the patterns we observed
in Big Island Savanna. Species richness did not decrease at
most spatial scales in environmentally similar sites that also
experienced long-term drought (Table 3); in fact many
sites gained species. Increases in species richness were gen-
erally larger on sites that had experienced slightly more
frequent fire over the last 20 yr relative to the original fire
management regime (Camp Lejeune Marine Corp Base,
Croatan National Forest; Table 3). However, species rich-
ness also increased or remained constant at most spatial
scales on sites with consistent fire frequency (other savan-
nas in the Green Swamp Preserve, Holly Shelter Game
Lands; Table 3). When species loss was detected on other
sites, the magnitude of loss was substantially less than that
Table 1. Average number of species lost and gained over time at each spatial scale inBig Island Savanna from1985 to 2013.
Area (m
2
)ΔRichness % Change ΔRichness % Change ΔRichness % Change ΔRichness % Change
1985–1994 1985–1994 1994–2011 1994–2011 2011–2012 2011–2012 2012–2013 2012–2013
1000 ––11.2 5.8 6.7 1.5 1.7
400 ––4.3 5.6 7.5*9.4 0.5 0.6
100 ––2.2 3.9 7.4** 12.1 0.9 1.4
10 ––5.3•14.7 7.6** 19.8 0.01 0.2
2.5 3.33 6.7 16.3** 32.7 6.3*15.8 3.8*8.7
11.253.015.5** 37 5.7** 17.6 2.2 6.3
0.25 1.02 3.5 11.5** 40.8 3.8** 18.6 2.1** 9.1
0.1 ––5*40.5 2.5** 25.5 0.4 3.9
0.01 0.36 3.3 4.8** 48.1 1.6** 23.3 0.71** 9.6
0.001 0.33 10.7 1** 38.3 0.03 0 0.24*13.6
ΔRichness 1985–1994 represents the mean change in species richness across scales between sequential observations inthe original sampling. ΔRichness
1994–2011 is the change in species richness from 1994 to 2011, while ΔRichness 2011–2012 and ΔRichness 2012–2013 are the change inspecies richness
from 2011 to 2012 and from 2012 to 2013, respectively. Double asterisks indicate P<0.0001, single asterisks indicate P<0.05 and single dots indicate
P<0.10. % Change is the percentage of species lostor gained over time relative to mean species richness.
Journal of Vegetation Science
6Doi: 10.1111/jvs.12186 ©2014 International Association for Vegetation Science
Plant speciesloss in a species-rich savanna K.A. Palmquist et al.
detected at Big Island (4.76 species lost at 0.01 m
2
in Big
Island Savanna, vs 1.08 in Holly Shelter; Table 3).
We re-sampled subplots in Big Island Savanna in 2012
to determine whether species richness had recovered with
two consecutive years of fire and somewhat wetter condi-
tions (Fig. 1). Richness in 2012 increased significantly rela-
tive to 2011 at all spatial scales except the smallest, which
remained stable (Table 1). In addition, most species
became more frequent in subplots in 2012, especially
insectivores and single-culm graminoids, which increased
by 15.4% and 8.8%, respectively, at 0.25 m
2
(Tables 2,
S2.2 in Appendix S2). Very few species decreased in fre-
quency between 2011 and 2012, except one species of club
moss (Lycopodiella appressa, 18.3% decrease at 0.25 m
2
),
one single-culm graminoid (Scleria minor, 8.3% decrease at
0.25 m
2
) and one rosette herb (Aletris farinosa,6.7%
decrease at 0.25 m
2
; Table S2.2 in Appendix S2). Although
species richness rebounded somewhat in 2012, the
increase at small spatial scales (2.5 m
2
and below) was not
nearly large enough for recovery to 1985–1994 levels (i.e.
species loss from 1994 to 2011 substantially exceeded
species gain from 2011 to 2012; Table 2).
We observed that 2013 was a significantly wetter year
than either 2011 or 2012 (Fig. 1). In response to higher
water availability and continued annual fire, species rich-
ness increased from 2012 to 2013 at most spatial scales,
although only slightly and often not significantly
(Table 1). At the smallest spatial scale (0.001 m
2
), more
species were gained from 2012–2013 than from
2011–2012, however at all other spatial scales the increase
Year
Mean richness (1 m2)
24 28 32 36 40 44
–1.2 –0.8 –0.4 0.0 0.4 0.8
1984 1988 1992 1996 2000 2004 2008 2012
Richness
PZ index
Fire event
PZ index
Fig. 3. Comparison of the variation in mean species richness at 1 m
2
and
mean annual Palmer Z Index (PZI) between 1985 and 2013. Prescribed fire
events are denoted with black boxes. The trajectory of richness in the
intervals 1989–1994 and 1994–2011 is indicated with a dashed line as
within-interval variation in richness is unknown.
Table 2. Mean number of species of different growth form, plant height (short <4 dm, tall >4 dm), and habitat affinity in 0.25-m
2
subplots in Big Island
Savanna.
Mean Drought Year Non-Drought Year Mean Mean Mean
1985–1994 1985–1994 1985–1994 2011 2012 2013
Growth form
Matrix graminoid 4.9 4.9 4.9 4.1 4.4 4.3
Insectivore 2.3 2.0 2.4 1.1 1.4 2.2
Single culm graminoid 5.7 5.5 5.8 3.4 4.8 5.7
Rosette herb 8.9 8.5 9.1 4.6 5.1 5.5
Hemiparasite 1.4 1.3 1.4 1.2 1.4 1.4
Clubmoss 1.0 1.0 1.0 1.0 1.0 1.0
Subshrub 1.4 1.4 1.4 1.3 1.4 1.3
Caulescent herb 3.5 3.4 3.5 1.7 1.9 2.3
Geophyte 1.6 1.6 1.6 1.0 1.2 1.3
Shrub 1.1 1.1 1.0 1.6 1.4 1.4
Tree 1.0 1.0 1.0 1.0 1.0 1.1
Legume 1.0 0.0 1.0 1.3 1.2 1.3
Vine 0.0 0.0 0.0 0.0 1.0 1.0
Plant height
Short 15.5 15.0 15.7 6.9 9.7 12.0
Tall 13.4 12.9 13.7 9.9 10.9 10.7
Habitat optimum
Mesic 17.9 17.3 18.2 8.6 11.0 12.9
Dry 8.4 8.1 8.6 6.0 6.7 6.3
Wet 2.6 2.4 2.7 2.5 3.0 3.4
Mean number of species in subplots is summarized for 1985–1994, for the drought years in the 1985–1994 interval (1985, 1986), non-drought years during
1985–1994, and 2011–2013. Small-statured herbaceous species (e.g. rosette herbs, geophytes, caulescent herbs) have decreased in frequency from 1985–
1994 to 2011–2013, while shrubs and vines have increased, suggestive of competitive exclusion caused by fire suppression.
7
Journal of Vegetation Science
Doi: 10.1111/jvs.12186©2014 International Association for Vegetation Science
K.A. Palmquist et al. Plant species loss in a species-rich savanna
in species richness from 2012–2013 was substantially less
than from 2011–2012 (Table 1, Figs 2 and 3), which is sur-
prising considering how much wetter 2013 was than 2012.
Figure 3 shows that species richness patterns across years
do not perfectly mirror changes in PZI over time. This is
particularly noticeable for 2012–2013 when PZI increased
dramatically, but richness did not. In contrast, richness
increased substantially from 2011–2012 despite only small
increases in PZI, suggesting that species richness patterns
are not solely being shaped by soil water availability and
that species recovery is likely to be a slow process. Alterna-
tively, it is possible that there is a time lag and species rich-
ness has yet to recover due to the wetter years of 2012 and
2013. A time lag may become apparent with continued
monitoring of these permanent plots, if increased fire fre-
quency is maintained. Most species increased slightly in
frequency or remained constant from 2012 to 2013. How-
ever, insectivores and single-culm graminoids both
increased dramatically (Tables 2, S2.1, S2.2, S2.3 in
Appendix S2). Despite some recovery of species richness in
2012 and 2013, species richness at small spatial scales is still
far below the levels documented in the 1980s.
Discussion
The 2011 sampling event revealed large declines in species
richness and species frequency in Big Island Savanna at
small spatial scales (≤2.5 m
2
; formerly 53 species in 2.5 m
2
and now 34 species, and formerly 42 in 1 m
2
and now 26;
see Fig. 2). Species loss was ubiquitous across most groups
and extremely high for the small-statured, herbaceous
species that constitute the bulk of plant species richness at
this site. Despite modest recovery in 2012 and 2013, small-
scale species richness remains far below the levels docu-
mented in 1985–1994.
Our data suggest that both reduced fire frequency and
drought have contributed to species loss in Big Island
Savanna, perhaps in a complex and interactive manner.
Determination of the degree to which reduced fire fre-
quency vs long-term drought is responsible for this loss will
only be fully clarified with continued monitoring of these
plots and future experiments manipulating fire frequency.
However, several lines of evidence suggest reduced fire fre-
quency during the last 15 yr is the primary factor driving
species richness declines in Big Island Savanna. First, the
declines in species richness and frequency in response to
drought in 1985–1986 were substantially smaller than dur-
ing the drought year of 2011, which suggests an additional
factor, such as reduced fire frequency, is responsible.
Although individual drought events may result in small
reductions in species abundance, short-term drought is
unlikely to result in local extinction of species or large
shifts in community composition (Grime et al. 2008). Sec-
ond, the frequency of species that have a habitat optimum
in wet environments has remained constant over time,
which is contrary to our expectation that ‘wet’ species
would be most sensitive to drought and would be lost pref-
erentially if drought were the major factor influencing spe-
cies richness patterns. This pattern is most likely a
consequence of the ‘wet’ species being almost exclusively
in the ‘tall’ group, reflecting the taller and lusher growth
on wet sites. Third, reduced fire frequency in longleaf pine
savannas and other fire-dependent grasslands results in
the loss of small-statured species, which are competitively
excluded as the abundance of woody species, ferns and
large grasses increases post-fire (Leach & Givnish 1996;
Glitzenstein et al. 2003; Overbeck et al. 2005). The docu-
mented loss in 2011 of mostly small herbaceous species
with dry to mesic habitat affinity is indicative of fire sup-
pression. Finally, several species that are known to be
weak competitors, and/or highly dependent on fire (i.e.
‘fire-followers’; Lemon 1949) decreased over time in Big
Island Savanna (e.g. Agalinis aphylla,Aletris farinosa,
Aristida virgata, Calopogon spp., Cleistesiopsis divaricata,
Dichanthelium strigosum,Drosera capillaris,Lycopodiella
appressa,Pinguicula spp., Xyris ambigua; Lemon 1949;
Wilson & Keddy 1986; Gaudet & Keddy 1995; Brewer
1999a,b; Keddy et al. 2006).
Further evidence suggests reduced fire frequency rather
than drought is the primary cause of species loss. For
example, other, environmentally similar sites within the
region have experienced little if any species loss, despite
having also been subjected to long-term drought. In fact,
species richness at these sites has, on average, increased at
most spatial scales, both on sites with somewhat increased
Table 3. Mean change in richness (Δrichness) at three spatial scales from
1993 to 2009 for Ultisol savannas in the southeastern Coastal Plain of NC.
Site ΔRichness
0.01 m
2
ΔRichness
1m
2
ΔRichness
10 m
2
Green Swamp –Big Island 4.8 15.8 5.1
Green Swamp –Other 0.8 0.3 3.1
Camp LeJeune 0.2 3.5 5.8
Croatan National Forest 1.3 6.3 9.1
Holly Shelter Game Lands 1.1 1.0 3.4
Negative values indicate sites that have lost species over time, while posi-
tive values indicate sites that have gained species. Species richness has
increased on Croatan National Forest and Camp LeJeune Marine Corp
Base, perhaps owing to somewhat more frequent fire in the last 20 yr rela-
tive to the original fire management regime. Species richness has
remained constant or increased at all but the smallest spatial scale on
other savannas in the Green Swamp Preserve and on Holly Shelter Game
Lands, which have experienced constant fire frequency over the last sev-
eral decades. These environmentally similar sites experiencing long-term
drought have not lost as many species as Big Island Savanna, suggesting
long-term drought is not the primary driver of species loss in Big Island
Savanna.
Journal of Vegetation Science
8Doi: 10.1111/jvs.12186 ©2014 International Association for Vegetation Science
Plant speciesloss in a species-rich savanna K.A. Palmquist et al.
fire frequency and those with consistent fire frequency
(Table 3). Additionally, two consecutive years of fire at Big
Island Savanna resulted in some recovery of species rich-
ness in 2012. Since 2012 was only slightly wetter than
2011, we attribute the increase in species richness in 2012
largely to two consecutive years of fire. However, long-
term drought has likely contributed to species loss in Big
Island Savanna and may explain why a few other sites in
southeastern North Carolina have lost species at small spa-
tial scales, albeit much less so than has been the case for
Big Island Savanna (Table 3).
SpecieslossesofthemagnitudeweobservedatBig
Island Savanna in response to alteration of disturbance
regimes, compounded with additional stressors (e.g.
drought, habitat fragmentation), have been reported in
other species-rich grassland systems (Leach & Givnish
1996; Glitzenstein et al. 2012). Some work suggests that
species richness may be slow to recover after stressful
events, such as drought or fire suppression, due to a loss of
local propagule sources, changes in the local environment
or shifts in vegetation structure in which often woody,
competitively superior species prevent the re-colonization
of herbaceous species (Tilman & Haddi 1992). Recovery
following stressful events may be especially challenging in
fragmented grasslands, such as Big Island Savanna, which
is embedded within a matrix of ombrotrophic peatland,
dominated by evergreen woody plants. For these reasons,
species richness at small spatial scales in Big Island
Savanna may take a significant amount of time to recover,
especially in the presence of ongoing drought, and likely
will not recover with continued reduced fire frequency.
The temporal and spatial breadth of our study was cru-
cial for detecting changes in species richness in Big Island
Savanna. Grasslands are structured by multiple processes
that vary over space and time (e.g. fire, grazing, drought;
see Collins & Smith 2006), and for this reason it is essential
to quantify plant species richness patterns across multiple
spatial and temporal scales. For example, previous work
has suggested that fire frequency has the largest impact on
species richness at small spatial scales (Glitzenstein et al.
2003; Collins & Smith 2006; Bowles & Jones 2013). Here,
we document a scale-dependent response to changes in
fire frequency and drought over time. Some combination
of drought and reduced fire frequency have reduced the
population sizes of most species in Big Island Savanna,
resulting in reduced species packing at small spatial scales,
as there are now fewer individuals of each species present.
If a nearly annual fire regime is not reinstated in Big Island
Savanna, population sizes are likely to continue to decline,
which may result in local extinction and declines in species
richness at larger spatial scales, especially for already infre-
quent and rare species. This hypothesis of species loss
trickling upward to larger spatial scales with continued fire
suppression has been suggested previously in another
longleaf pine study examining species richness patterns
over time in relation to fire frequency (Glitzenstein et al.
2012). Had we examined species richness patterns only at
1000 m
2
in Big Island Savanna, we would have concluded
that species richness had remained relatively stable over
time and that changes in the fire management regime and/
or long-term drought had not affected vegetation patterns
in this savanna (Table 1). Hence, monitoring at multiple
spatial and temporal scales is critical for understanding pat-
terns, identifying processes that drive those patterns, and
informing conservation and land management agencies
about best management practices.
One important finding from this study is that small
changes in fire management regimes can have large and
long-lasting consequences for plant species richness in
longleaf pine savannas. Our work suggests that very fre-
quent to annual fire is probably necessary to maintain
small-scale biodiversity and species packing in the most
species-rich, moist savannas, especially in the face of addi-
tional environmental stress. We believe this work can be
generalized to other species-rich grasslands that experience
chronic or continuous disturbance (e.g. alvar grasslands in
Northern Europe, oligotrophic mowed meadows of East-
ern Europe, cerrado in Brazil, Themeda triandra grasslands
in Australia, and mountain grasslands of central Argentina;
Wilson et al. 2012). Some evidence from other species-rich
systems also suggests that slight changes in disturbance
regimes can have large impacts on plant biodiversity (Mor-
gan 1999; Overbeck et al. 2005). In addition, our research
indicates that land managers should proceed cautiously
when making changes to long-standing management
regimes, despite how well intentioned such changes might
be, and assess impacts immediately after their implementa-
tion. Future work in other species-rich grasslands should
both explore whether chronic and nearly continuous dis-
turbance is necessary to maintain species richness and how
slight alteration of disturbance regimes can affect plant
species richness over time.
Both periodic and multi-year drought events are impor-
tant factors that shape species richness and community
composition patterns in the longleaf pine ecosystem and
potentially increase the risk of biodiversity loss with altered
disturbance regime. Drought events have been recognized
as important processes in many other grassland ecosystems
(e.g. Tilman & Haddi 1992; Knapp et al. 2002, 2006;
Anderson 2008; Evans et al. 2011; Cherwin & Knapp
2012). The severity and intensity of drought events in the
southeastern US have increased in the last 25 yr and are
predicted to continue increasing with ongoing climate
change (Klos et al. 2009). Thus, future research should
explore further the relative and interactive contributions
of drought and fire to changes in community structure and
9
Journal of Vegetation Science
Doi: 10.1111/jvs.12186©2014 International Association for Vegetation Science
K.A. Palmquist et al. Plant species loss in a species-rich savanna
composition in the longleaf pine ecosystem as such knowl-
edge will be critical for protecting these species-rich and
threatened communities.
Although longleaf pine savannas, among many other
grassland ecosystems, are dominated by long-lived peren-
nial species, species richness and frequency in these eco-
systems are surprisingly sensitive, both spatially and
temporally, to environmental changes, alteration of dis-
turbance regimes, and stochastic events (Sykes et al.
1994; Collins & Smith 2006). An unusually long history of
detailed vegetation sampling at multiple scales in Big
Island Savanna has enabled us to document complex
changes in species richness in response to drought and
reduced fire frequency. Additional studies in the longleaf
pine ecosystem and other grass-dominated ecosystems are
needed to more fully disentangle the complex and interac-
tive effects of environmental change and altered distur-
bance regimes on spatial and temporal patterns of species
richness. Moreover, understanding these complex
relationships will be necessary to provide critical guidance
to land managers responsible for conserving important
biodiversity sites.
Acknowledgements
We acknowledge funding from NSF grant BSR-8506098,
which made study 1 possible, and the US Forest Service,
which made study 2 possible. The 2013 re-sampling event
was made possible by a North Carolina Native Plant Society
grant to K.A. Palmquist. We acknowledge the collabora-
tion of Joan Walker, Tom Wentworth, Mike Schafale and
Richard Duncan in the original sampling, along with that
of several volunteer participants in the Carolina Vegetation
Survey. Whitney Brown, Bianca Lopez, Diane Palmquist,
Naomi Schwartz, Mike Turner, Jackie White and Brenda
Wichmann helped with data collection in the years of
2011–2013. Allen Hurlbert, Steve Mitchell, and an anony-
mous reviewer provided valuable comments on an earlier
version of the manuscript. We thank The Nature Conser-
vancy for access and sharing fire history data.
References
Anderson, T.M. 2008. Plant compositional change over time
increases with rainfall in Serengeti grasslands. Oikos 117:
675–682.
Belsky, A.J. 1992. Effects of grazing, competition, disturbance
and fire on species composition and diversity in grassland
communities. Journal of Vegetation Science 3: 187–200.
Bowles, M.L. & Jones, M.D. 2013. Repeated burning of eastern
tallgrass prairie increases richness and diversity, stabilizing
late successional vegetation. Ecological Applications 23: 464–
478.
Brewer, J.S. 1999a. Effects of fire, competition and soil distur-
bances on regeneration of a carnivorous plant (Drosera capil-
laris). American Midland Naturalist 141: 28–42.
Brewer, J.S. 1999b. Short term effects of fire and competition on
growth and plasticity of the yellow pitcher plant, Sarracenia
alata (Sarraceniaceae). American Journal of Botany 86: 1264–
1271.
Cantero, J.J., P€
artel, M. & Zobel, M. 1999. Is species richness
dependent on the neighbouring stands? An analysis of the
community patterns in mountain grasslands of central
Argentina. Oikos 87: 346–354.
Cheng, Y., Tsubo, M., Ito, T.Y., Nishihara, E. & Shinoda, M.
2011. Impact of rainfall variability and grazing pressure on
plant diversity in Mongolian grasslands. Journal of Arid
Environments 75: 471–476.
Cherwin, K. & Knapp, A. 2012. Unexpected patterns of sensitiv-
ity to drought in three semi-arid grasslands. Oecologia 169:
845–852.
Christensen, N.L. 1981. Fire regimes in southeastern ecosystems.
In: Mooney, H.A., Bonnicksen, T.M., Christensen, N.L.,
Lotan, J.E. & Reiners, W.A. (eds.) Fire regimes and ecosystem
properties: proceedings of the conference, pp. 112–136. USDA For-
est Service, Washington, DC, US.
Christensen, N.L. 2000. Vegetation of the southeastern Coastal
Plain. In: Barbour, M.G. & Billings, W.D. (eds.) North Ameri-
can terrestrial vegetation, pp. 397–448. Cambridge University
Press, Cambridge, UK.
Cleland, E.E., Collins, S.L., Dickson, T.L., Farrer, E.C., Gross,
K.L., Gherardi, L.A., Hallett, L.M., Hobbs, R.J., Hsu, J.S.,
Turnbull, L. & Suding, K.N. 2013. Sensitivity of grassland
plant community composition to spatial vs. temporal varia-
tion in precipitation. Ecology 94: 1687–1696.
Collins, S.L. & Smith, M.D. 2006. Scale-dependent interaction of
fire and grazing on community heterogeneity in tallgrass
prairie. Ecology 87: 2058–2067.
Collins, S.L., Knapp, A.K., Briggs, J.M., Blair, J.M. & Steinauer,
E.M. 1998. Modulation of diversity by grazing and mowing
in native tallgrass prairie. Science 280: 745–747.
Debinski, D.M., Caruthers, J.C., Cook, D., Crowley, J. & Wick-
ham, H. 2013. Gradient-based habitat affinities predict spe-
cies vulnerability to drought. Ecology 94: 1036–1045.
Dengler, J., Ruprecht, E., Szab
o, A., Turtureanu, P.D., Beldean,
M., U
gurlu, E., Pedashenko, H., Dolnik, C. & Jones, A. 2009.
EDGG cooperation on syntaxonomy and biodiversity of Fe-
stuco brometea communities in Transylvania (Romania):
report and preliminary results. Bulletin of the European Dry
Grassland Group 4: 13–19.
Dengler, J., Becker, T., Ruprecht, E., Szab
o, A., Becker, U.,
Beldean, M., Pedashenko, H., Biota-Nicolae, C., Dolnik, C.,
(...)&U
gurlu, E. 2012. Festuco-Brometea communities of the
Transylvanian Plateau (Romania) –a preliminary overview
on syntaxonomy, ecology, and biodiversity. Tuexenia 32:
319–359.
Evans, S.E., Byrne, K.M., Lauenroth, W.K. & Burke, I.C. 2011.
Defining the limit to resistance in a drought-tolerant
Journal of Vegetation Science
10 Doi: 10.1111/jvs.12186 ©2014 International Association for Vegetation Science
Plant speciesloss in a species-rich savanna K.A. Palmquist et al.
grassland: long-term severe drought significantly reduces
the dominant species and increases ruderals. Journal of Ecol-
ogy 99: 1500–1507.
Fidelis, A.2010. South Brazilian Campos grasslands: biodiversity,
conservation and the role of disturbance. In: Runas, J. &
Dahlgren, T. (eds.) Grassland biodiversity: habitat types, ecologi-
cal processes and environmental impacts, pp. 223–239. Nova Sci-
ence, New York, NY, US.
Folkerts, G.W. 1982. The Gulf Coast Pitcher Plant Bogs: one of
the continent’s most unusual assemblages of organisms
depends on a increasingly rare combination of saturated soil
and frequent fires. American Scientist 70: 260–267.
Frost, C. 2006. History and future of the longleaf pine ecosystem.
In: Jose, S., Jokela, E. & Miller, D. (eds.) The longleaf pine eco-
system: ecology, silviculture, and restoration, pp. 9–48. Springer,
New York, NY, US.
Frost, C.C., Walker, J. & Peet, R.K. 1986. Fire-dependent savan-
nas and prairies of the southeast: original extent, preserva-
tion status and management problems. In: Kulhavy, D.L. &
Conner, R.N. (eds.) Wilderness and natural areas in the eastern
United States,pp.348–357. Center for Applied Studies, School
of Forestry, Stephen F. Austin University, Nacogdoches, TX,
US.
Gaudet, C.L. & Keddy, P.A. 1995. Competitive performance and
species distribution in shoreline plant communities: a com-
parative approach. Ecology 76: 280–291.
Gibson, D.J. & Hulbert, L.C. 1987. Effects of fire, topography and
year-to-year climatic variation on species composition in
tallgrass prairie. Vegetatio 72: 175–185.
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: 22–37.
Glitzenstein, J.S., Streng, D.R., Masters, R.E., Robertson, K.M. &
Hermann, S.M. 2012. Fire frequency effects on vegetation in
north Florida pinelands: another look at the long-term Stod-
dard Fire Research Plots at Tall Timbers Research Station.
Forest Ecology and Management 264: 197–209.
Grime, J.P., Fridley, J.D., Askew, A.P., Thompson, K., Hodgson,
J.G. & Bennett, C.R. 2008. Long-term resistance to simulated
climate change in an infertile grassland. Proceedings of the
National Academy of Sciences of the United States of America 105:
10028–10032.
Haddad, N.M., Tilman, D. & Knops, J.M.H. 2002. Long-term
oscillations in grassland productivity induced by drought.
Ecology Letters 5: 110–120.
Heyward, F. 1939. The relation of fire to stand composition of
longleaf pine forests. Ecology 20: 287–304.
Huffman, J.M. 2006. Historical fire regimes in southeastern pine sav-
annas. PhD Thesis, Louisiana State University, Baton Rouge,
LA, US.
Keddy, P.A., Smith, L., Campbell, D.R., Clark, M. & Montz, G.
2006. Patterns of herbaceous plant diversity in southeastern
Louisianapinesavannas.Applied Vegetation Science 9: 17–26.
Kirkman, L.K., Goebel, P.C. & Palik, B.J. 2004. Predicting plant
species diversity in a longleaf pine landscape. Ecoscience 11:
80–93.
Klime
s, L., Dan
c
ak, M., H
ajek, M., Jongepierov
a, I. & Ku
cera, T.
2001. Scale-dependent biases in species counts in a grass-
land. Journal of Vegetation Science 12: 669–704.
Klos, R.J., Wang, G.G., Bauerle, W.L. & Rieck, J.R. 2009.
Drought impact on forest growth and mortality in the south-
east USA: an analysis using forest health and monitoring
data. Ecological Applications 19: 699–708.
Knapp, A.K., Fay, P.A., Blair, J.M., Collins, S.L., Smith, M.D.,
Carlisle, J.D., Harper, C.W., Danner, B.T., Lett, M.S. & Mc-
Carron, J.K. 2002. Rainfall variability, carbon cycling, and
plant species diversity in a mesic grassland. Science 298:
2202–2205.
Knapp, A.K., Burns, C.E., Fynn, R.W.S., Kirkman, K.P., Morris,
C.D. & Smith, M.D. 2006. Convergence and contingency in
production–precipitation relationships in North American
and South American C4 grasslands. Oecologia 149: 456–464.
Kologiski, R.L. 1977. The phytosociology of the Green Swamp, North
Carolina. [North Carolina Agricultural Experiment Station
Technical Bulletin 250]. North Carolina Agricultural Experi-
ment Station, Raleigh, NC, US.
Kull, K. & Zobel, M. 1991. High species richness in an Estonian
wooded meadow. Journal of Vegetation Science 2: 715–718.
Leach, M.K. & Givnish, T.J. 1996. Ecological determinants of
specieslossinremnantprairies.Science 273: 1555–1558.
Lemon, P.C. 1949. Successional responses of herbs in longleaf-
slash pine forest after fire. Ecology 30: 135–145.
McIver, H. 1981. Green swamp nature preserve. The Nature Conser-
vancy, Chapel Hill, NC, US.
Morgan, J.W. 1999. Defining grassland fire events and the
response of perennial plants to ann ual fire in temperate
grasslands of south-eastern Australia. Plant Ecology 144:
127–144.
Myers, J.A. & Harms, K.E. 2011. Seed arrival and ecological fil-
ters interact to assemble high-diversity plant communities.
Ecology 92: 676–686.
National Climatic Data Center. 2013. Available at http://www.
ncdc.noaa.gov/ Accessed July, 2013.
Noss, R.F. 2013. Forgotten grasslands of the south: natural history and
conservation. Island Press, Washington DC, US.
O’Connor, T.G. 1995. Transformation of a savanna grassland by
drought and grazing. African Journal of Range & Forage Science
12: 53–60.
Outcalt, K.W. & Sheffield, R.M. 1996. The longleaf pine forest:
trends and current conditions. [USDA Forest Service Southern
Research Station resource bulletin 9]. United States Depart-
ment of Agriculture, Washington, DC, US.
Overbeck, G.E., M€
uller, S.C., Pillar, V.D. & Pfadenhauer, J. 2005.
Fine-scale post-fire dynamics in southern Brazilian subtropi-
cal grassland. Journal of Vegetation Science 16: 655–664.
Palmquist, K.A. 2014. Vegetation dynamics and plant diversity pat-
terns across space and timein the longleaf pine ecosystem.Ph.Ddis-
11
Journal of Vegetation Science
Doi: 10.1111/jvs.12186©2014 International Association for Vegetation Science
K.A. Palmquist et al. Plant species loss in a species-rich savanna
sertation, University of North Carolina at Chapel Hill, Chapel
Hill, NC, US.
Peet, R.K. & Christensen, N.L. 1988. Changes in species diversity
during secondary forest succession on the North Carolina
piedmont. In: During, H.J., Werger, M.J.A. & Willems, J.H.
(eds.) Diversity and pattern in plant communities, pp. 233–245.
SPB Academic Publishing, The Hague, NL.
Peet, R.K., Wentworth, T.R. & White, P.S. 1998. A flexible, mul-
tipurpose method for recording vegetation composition and
structure. Castanea 63: 262–274.
Peet, R.K., Lee, M.T., Boyle, M.F., Wentworth, T.R., Schafale,
M.P. & Weakley, A.S. 2012. Vegetation plot database of the
Carolina Vegetation Survey database. Biodiversity and Ecology
4: 243–253.
Peet, R.K., Palmquist, K.A. & Tessel, S.M. 2014. Herbaceous
layer species richness of southeastern forests and woodlands:
patterns and causes. In: Gilliam, F.S. & Roberts, M.R., (eds.)
The herbaceous layer in forests of eastern North America, 2nd ed,
pp. 255–276. Oxford University Press, Oxford, UK.
Potts, D.L., Suding, K.N., Winston, G.C., Rocha, A.V. & Goulden,
M.L. 2012. Ecological effects of experimental drought and
prescribed fire in a southern California coastal grassland.
Journal of Arid Environments 81: 59–66.
Rome, A. 1988. Vegetation variation in a pine–wiregrass
savanna in the Green Swamp, North Carolina. Castanea
1988: 122–131.
Ruffner, J.A. 1985. Climates of the states: National Oceanic andAtmo-
spheric Administration narrative summaries, tables, and maps for
each state, with overview of state climatologist programs,3rded.
Gale Research Co., Detroit, MI, US.
State Climate Office of North Carolina, NC State University.
CRONOS [internet database] available at http://www.nc-
climate.ncsu.edu/cronos/ Accessed July 26, 2011.
Stevens, C., Dupr
e, C., Gaudnik, C., Dorland, E., Dise, N., Gow-
ing, D., Bleeker, A., Alard, D., Bobbink, R., (...)&Diek-
mann, M. 2011. Changes in species composition of
European acid grasslands observed along a gradient of nitro-
gen deposition. Journal of Vegetation Science 22: 207–215.
Sykes, M.T., van der Maarel, E., Peet, R.K. & Willems, J.H. 1994.
High species mobility in species-rich plant communities: an
intercontinental comparison. Folia Geobotanica et Phytotaxono-
mica 29: 439–448.
Tilman, D. & Haddi, A.E. 1992. Drought and biodiversity in
grasslands. Oecologia 89: 257–264.
Walker, J. & Peet, R.K. 1983. Composition and species diversity
of pine–wiregrass savannas of the Green Swamp, North Car-
olina. Vegetatio 55: 163–179.
Weakley, A.S. 2012. Flora of the southern and mid-Atlantic States.
University of North Carolina Herbarium, Chapel Hill, NC.
Wilson, S.D. & Keddy, P.A. 1986. Species competitive ability and
position along a natural stress/disturbance gradient. Ecology
67: 1236–1242.
Wilson, J.B., Peet, R.K., Dengler, J. & P€
artel, M. 2012. Plant spe-
cies richness: the world records. Journal of Vegetation Science
23: 796–802.
Zuur, A.F., Leno, E.N., Walker, N.J., Saveliev, A. & Smith, G.M.
2009. Mixed effect models and extensions in ecology with R.
Springer, New York, NY.
Supporting Information
Additional Supporting Information may be found in the
online version of this article:
Appendix S1. Map of plot locations in southeastern
North Carolina, USA.
Appendix S2. Species frequency and percentage of
subplots occupied for all species at 0.01, 0.25 and 1 m
2
.
Journal of Vegetation Science
12 Doi: 10.1111/jvs.12186 ©2014 International Association for Vegetation Science
Plant speciesloss in a species-rich savanna K.A. Palmquist et al.
