Content uploaded by William J. Platt
Author content
All content in this area was uploaded by William J. Platt on Aug 02, 2018
Content may be subject to copyright.
MARCH
2003
Restoration Ecology Vol. 11 No. 1, pp. 91–102
91
©
2003 Society for Ecological Restoration International
Effects of Differences in Prescribed Fire Regimes
on Patchiness and Intensity of Fires in Subtropical
Savannas of Everglades National Park, Florida
Matthew G. Slocum,
1
William J. Platt,
2
and Hillary C. Cooley
3
Abstract
We investigated effects of fire frequency, seasonal timing,
and plant community on patchiness and intensity of pre-
scribed fires in subtropical savannas in the Long Pine Key
region of Everglades National Park, Florida (U.S.A.). We
measured patchiness and intensity in different plant com-
munities along elevation gradients in “fire blocks.” These
blocks were prescribed burned at varying times during the
lightning season and at different frequencies between 1995
and 2000. Fire frequency, seasonal timing, and plant com-
munity all influenced the patchiness and intensity of pre-
scribed fires. Fires were less patchy and more intense, proba-
bly because of drier conditions and pyrogenic fuels, in higher
elevation plant communities (e.g., high pine savannas) than
in lower elevation communities (e.g., long-hydroperiod prai-
ries). In all plant communities fires became increasingly
patchy and less intense as the wet season progressed and
moisture accumulated in fuels. Frequent prescribed fire
resulted in increased patchiness but a wider range of inten-
sities; higher intensities appeared to result from regrowth
of more flammable vegetation. Our study suggests that
frequent early lightning season prescribed fires produce a
wider range of post-fire conditions than less frequent late
lightning season prescribed fires. Our study also suggests
that natural early lightning season fires readily carried
through pine savannas and short-hydroperiod prairies, but
lower elevation long-hydroperiod prairies functioned as
firebreaks. Natural fires probably crossed these firebreaks
only during drier years, potentially producing large land-
scape-level fires. Knowledge of how patchily and intensely
fires burn across a savanna landscape should be useful for
developing landscape-level fire management.
Key words:
Everglades National Park, fine fuel consump-
tion, fire regime, landscape level, lightning season, long-
hydroperiod prairie, Long Pine Key, patchiness, pine sa-
vanna, restoration, seasonal timing, short-hydroperiod
prairie.
1
Wetland Biogeochemistry Institute, Louisiana State University, Baton Rouge,
Louisiana 70803, U.S.A. E-mail: mateo457@yahoo.com.
2
Department of Biological Sciences, Louisiana State University, Baton Rouge,
Louisiana 70803, U.S.A.
3
Department of Environmental Studies, Florida International University,
University Park, ECS 344, Miami, Florida 33199, U.S.A.
Introduction
Historically, lightning-initiated fires occurred frequently in
pine savannas in the coastal plain of the southeastern
United States (Harper 1927; Chapman 1932; Komarek
1964; Platt 1999). These savannas became fragmented by
the early 20th century; thus natural fire regimes were al-
tered and never described scientifically (Frost 1993). This
lack of information has produced poorly conceived fire
management policy, which has emphasized both fire sup-
pression and prescribed fires that did not mimic natural
fires or their effects (Doren et al. 1993; Platt & Peet 1998;
Platt 1999; Drewa et al. 2002b). As a result, less fire-
adapted native and exotic woody plant species have fre-
quently displaced species-rich herbaceous ground cover
(Heyward 1939; Walker & Peet 1983; Streng et al. 1993;
DeCoster et al. 1999; Platt 1999). Recently, management
efforts have involved more ecologically sound prescribed
fires (e.g., Huffman & Blanchard 1991), but lack of knowl-
edge of natural fire regimes has hampered such attempts
(Glitzenstein et al. 1995; Olson & Platt 1995). Such efforts
are further complicated by small differences in prescribed
fire regimes causing substantial effects on the vegetation
(Platt & Schwartz 1990; Streng et al. 1993; Glitzenstein et
al. 1995; Platt 1999; Provencher et al. 2001; Mulligan &
Kirkman 2002).
Why small changes in fire regimes often produce large
effects is beginning to be understood (Johnson 1992). One
model proposes that vegetation in southeastern savannas
evolved under conditions involving frequent predictable
fires (Platt 1999). Large fires in the southeast historically
appear to have occurred primarily during the transition
from the nonlightning/dry season (hereafter “nonlightning
season”) to the lightning/wet season (hereafter “lightning
season”) (Komarek 1964; Doren et al. 1993; Platt 1999).
At these times lightning strikes increase in frequency
(Maier et al. 1979; Snyder 1991; Goodman & Christian
1993; Hodanish et al. 1997). The dried fuels at this time are
easily ignited, potentially producing large landscape-level
fires given certain synoptic weather conditions. Such light-
ning season fires probably occurred frequently because
herbaceous vegetation recovers quickly and thus can re-
burn within a few years (Platt et al. 1991; Brewer & Platt
Patchiness and Intensity of Prescribed Fires
92
Restoration Ecology
MARCH
2003
1994; DeCoster et al. 1999). This model of natural fires
(i.e., high frequency lightning season fires) in savanna
landscapes is supported by studies of responses of vegeta-
tion to prescribed fires (Streng et al. 1993; Glitzenstein et
al. 1995; Olson & Platt 1995; DeCoster et al. 1999). Only
when prescribed fires have followed the model have
species-rich herbaceous ground cover communities been
restored. In contrast, prescribed fires that continue to em-
phasize past policies (i.e., infrequent and/or nonlightning
season fires) over long time intervals have resulted in
ground cover communities dominated by less fire-adapted
woody species (Platt et al. 1991; Glitzenstein et al. 1995;
Drewa et al. 2002a).
Seasonal timing and frequency should affect patchiness
and intensity of fires and therefore the structure and com-
position of vegetation (Sparks et al. 2002). Most experi-
mental studies, however, have been conducted in small
plots, in single communities, and under less extreme envi-
ronmental conditions; as a result these studies have been
characterized by reduced variability in fire behavior (e.g.,
Snyder 1986; Streng et al. 1993; Glitzenstein et al. 1995;
Olson & Platt 1995; Hiers et al. 2000; Drewa et al. 2002b).
Therefore, experimental fires probably have not accu-
rately mimicked the variation in intensity and patchiness
of natural fires or the effect this variation could have had
on vegetation. Moreover, because pine savanna land-
scapes generally contain a number of plant communities
(e.g., Abrahamson & Hartnett 1990; Platt & Schwartz
1990), any understanding of natural fire regimes on a land-
scape level must also consider how fires carry across these
plant communities, especially when variation in elevation
is involved. If intensity and patchiness of natural fires are
to be mimicked by prescribed fires, studies need to be con-
ducted on how variation in timing and frequency affect
the intensity and patchiness of burn in the relevant plant
communities.
In this study we measured how fires burning at different
frequencies and times and in different plant communities
affected fire patchiness and intensity. We conducted our
study in different “upland” savanna plant communities oc-
curring along elevation gradients in the Long Pine Key re-
gion of Everglades National Park. At the park restoration
of natural fire regimes in a landscape context has been a
high priority because these savannas comprise the only re-
maining intact subtropical savannas in the United States.
Over the past decade prescribed fires have attempted to
mimic the understood natural pattern but have varied
enough in timing and frequency to allow for study on how
this variation affects fires. Moreover, these fires have been
allowed to burn unchecked through large management
blocks among which there are similar elevation gradients.
We tested three predictions regarding how plant com-
munity, fire frequency, and seasonal timing should influ-
ence patchiness and intensity of prescribed fires:
(1) Increased patchiness and lower intensity of prescribed
fires should occur in lower elevation plant communi-
ties compared with higher elevation plant communi-
ties, because these areas will be seasonally flooded and
also contain less pyrogenic vegetation.
(2) Increased patchiness and lower intensity should result
from increased frequency of prescribed fires because
accumulations of fuels (live and dead vegetation)
should be less at the time of fires.
(3) Increased patchiness and lower intensity should result
if prescribed fires are shifted later in the lightning sea-
son because of increased moisture in the vegetation
and the environment.
We examined landscape-level variation in patchiness and
intensity during prescribed lightning season fires. We also
explored differences in fire patchiness and intensity that
might result from differences in prescribed fire frequency
and seasonal timing. We propose that landscape-level vari-
ation in characteristics of some of the prescribed fires re-
sembled variation occurring in natural fires. Knowledge of
landscape-level variation in characteristics of prescribed
fires that are designed to mimic frequencies and seasonal
timing of natural fires should be useful in guiding restora-
tion and management efforts in these threatened habitats.
Methods
Study Site
Long Pine Key (25
o
22–24
N, 80
o
37–41
W) is located at the
southwestern tip of the Atlantic Coastal Ridge. This ridge
consists of oolitic limestone outcrops several meters above
sea level, which extend south along the Florida coast
through Miami to Homestead, where they bend westward
into the park (Hoffmeister et al. 1967). Long Pine Key is a
series of these outcrops crossed by lower elevation glades
that once were tidal channels (Fig. 1) (Hoffmeister et al.
1967; Olmsted et al. 1983). Elevations in Long Pine Key
reach a maximum of 3.4 masl in the outcrops to a mini-
mum of 1.5 masl in the glades (Olmsted et al. 1983). An-
nual weather patterns include a pronounced dry season
(
60 mm rainfall per month) that begins in the fall (October–
November) and is followed by a wet season (140 mm rain-
fall per month) that occurs May to mid-July (Hela 1952;
Chen & Gerber 1990).
Historically, the Atlantic Coastal Ridge was dominated
by subtropical savannas (Harshberger 1914; Harper 1927;
Davis 1943; Craighead 1971), but today more than 90% of
these savannas have been eliminated by urban develop-
ment (Snyder 1986; Snyder et al. 1990). Long Pine Key is
the last large remnant in a natural landscape. Its more ele-
vated outcrops contain savannas dominated by south Flor-
ida slash pine (
Pinus elliottii
Engelman var
densa
Little &
Dorman), with a distinctive species-rich calciphilic ground
cover that is dominated by warm season grasses, especially
firegrass (
Andropogon cabanisii
Hackel) at higher eleva-
tions and muhly grass (
Muhlenbergia filipes
M. A. Curtis)
at lower elevations (DeCoster et al. 1999; Platt 1999;
Schmitz et al. 2001). The lower elevation glades are sea-
Patchiness and Intensity of Prescribed Fires
MARCH
2003
Restoration Ecology
93
sonally flooded and contain short- and long-hydroperiod
prairies, which are species-rich almost treeless savannas
dominated by graminoids such as muhly grass, paspalum
(
Paspalum monostachyum
Vasey ex Chapman), and sawgrass
(
Cladium jamaicense
Crantz) (Porter 1967; Olmsted et al
1983; DeCoster et al. 1999; Schmitz et al. 2001).
Humans have disturbed Long Pine Key. Pine stands are
second growth, regenerating from cull trees left behind
after logging in the 1930s and 1940s (Robertson 1953; Tay-
lor 1981; Olmsted et al. 1983; Platt et al. 2000). In addition,
hydroperiods in the glades have probably been influenced
by artificial changes in amounts and timing of water flow
through the region (Parker 1974; DeGrove 1984). Long-
term changes in relative abundances of some dominant spe-
cies have been documented, but local extinctions of species,
at least in short-hydroperiod prairies of Long Pine Key, ap-
pear not to have occurred (Olmsted & Armentano 1997).
Fire Regimes of Long Pine Key
Prescribed fire regimes in Long Pine Key have been al-
tered over the past half century. Little was known about
natural fire regimes when the park was founded in 1947,
and at first fires were suppressed. However, as it became
clear that fires were important in maintaining Everglades
habitats (e.g., Robertson 1953), fires were reintroduced,
starting in the late 1950s (Wade et al. 1980; Taylor & Hern-
don 1981; Snyder 1991). These attempts were not success-
ful in maintaining savanna habitats because there was little
information on how differences in fire characteristics af-
fect vegetation (Doren et al. 1993). Gradually the natural
fire regime became better understood through the study of
climatology (Chen & Gerber 1990), carefully maintained
fire records (Taylor 1981; Doren & Rochefort 1984), and
responses of vegetation to prescribed fires (Doren et al.
1993; DeCoster et al. 1999; Platt et al. 2000; Schmitz et al.
2001). Beginning in 1989 an attempt was made to generate
more natural fire regimes by focusing on increasing fre-
quency and shifting the timing of burn to the transition be-
tween the nonlightning and lightning seasons. In 1989 and
1990 all of Long Pine Key was prescribed burned during
the early to middle lightning season to initiate a restora-
tion effort based on the hypothesized natural fire regime
(i.e., increased frequencies and correct timing). Since then,
fires have been ignited every 2 to 3 years, with most igni-
tions occurring in the early to middle lightning season
(Doren et al. 1993; DeCoster et al. 1999; Platt et al. 2000).
This policy has resulted in restoration of species-rich her-
baceous-dominated savannas. Rapid recovery of the vege-
tation generates new continuous fine fuels capable of car-
rying fires within 2 years after a fire (DeCoster et al. 1999),
especially in pine savannas where herbaceous species re-
grow rapidly and pine needles are shed almost continu-
ously (Herndon & Taylor 1985). The response of vegeta-
tion to lightning season fires is congruent with field
observations that lightning initiated fires occurred as often
as every 2 to 3 years in pine savannas (Harper 1927; Rob-
ertson 1953). Such frequent fires probably were most often
of low intensity; estimates of maximum fire temperatures
over the past decade average around 500
o
C, but maxima
Figure 1. Fire blocks within Long Pine Key at Everglades National Park. The blocks are delimited by old logging roads (thin lines) and main
roads (thicker lines). The different plant communities are indicated with shades of gray, including hardwood hammocks (dark), pine savanna (me-
dium), wet prairie (light), and exotic shrub thicket (i.e., Schinus terebinthifolia, in white). Transects (short thin lines with crossmarks) extend from
low elevations in long-hydroperiod prairie to high elevations in high pine savannas.
Patchiness and Intensity of Prescribed Fires
94
Restoration Ecology
MARCH
2003
more than 800
o
C have been recorded in herbaceous pine
savanna ground cover (W. J. Platt, unpublished data).
Currently, fires are ignited separately in different fire
blocks (Fig. 1), compartments of several hundreds to thou-
sands of hectares, most of which contain both pine savannas
and glades. Fires are contained within the blocks by burning
small strips (blacklining) along old logging roads to gener-
ate firebreaks. Fires are started with a combination of spot
and line ignitions, simulating lightning strikes and move-
ment from adjacent areas. Movement of the fires through
the blocks is not controlled, and as a result wide ranges of
fire behavior occur, depending on weather conditions,
fuels, and types of ignition patterns (Doren et al. 1993).
Experimental Design and Data Analysis
Eleven prescribed fires burned through eight fire blocks
(three blocks burned twice) during May through Septem-
ber in 1997 to 2000. Before these fires burned we estab-
lished permanent transects in the blocks. Each transect be-
gan in the lowest point in a glade and extended upslope,
approximately perpendicular to the long axis of the glade,
to the highest points of adjacent pine savannas. Five to
eight permanent plots (50
20 meters, with one 20-meter
side on the transect) were placed at approximately 10 cm
differences in elevation along each transect (DeCoster et
al. 1999; Schmitz et al. 2001). Each plot was divided into 10
modules of 10
10 meters, four of which contained two
submodules each (3.16
3.16 meters; 10-m
2
sampling
area) (Fig. 2). When the area containing a transect was
burned, all submodules in all plots along the transect were
examined. Each submodule of 10 m
2
was considered a
sampling unit for estimating fire effects. This sized sampling
unit was sufficient to define plant communities and to mini-
mize variability in fire characteristics within the unit. We
thus were able to assign single plant communities and fire
characteristics to this sized unit (Fig. 2). There were 384
submodules along all transects, but because three transects
burned twice, 520 submodule by fire events were used in
analyses.
The two dependent variables, fire patchiness and fire in-
tensity, were measured in each submodule after prescribed
fires burned through a transect. Fire patchiness was as-
signed as a “yes or “no” for each of the 520 submodules by
fire events, based on whether or not that submodule
burned (520 submodules).
Fire intensity was examined in the field using the 385
submodules that burned. We examined consumption of
vegetation and litter, using categories similar to those used
by Spier and Snyder (1998). Based on field experience a
priori categories of low, moderate, and high fire intensity
were constructed describing degrees of damage and con-
sumption of three components of the vegetation (grami-
noids and forbs, shrubs and palms, and fine fuel litter). We
examined each submodule within a week after fires. Crite-
ria for assigning low, moderate, or high fire intensity for
each of these vegetation components were as follows:
Herbaceous species. Low: plants consumed in the fire but
at least 5 cm of bases of graminoids and forbs re-
mained. Moderate: plants consumed, but less than 2 cm
of bases of graminoids and forbs remained. High:
plants consumed, with less than 1 cm of bases of grami-
noids and forbs still present.
Shrubs and palms. Low: Any shrubs present were not top-
killed; leaves sometimes browned, but not consumed.
Leaves of any palms browned, with perhaps tips con-
sumed. Moderate: leaves of shrubs and palms partially
or completely consumed. High: leaves of any shrubs
and palms consumed; stems of shrubs and petioles of
palm leaves partially or completely consumed.
Fine fuel litter. Low: Most litter consumed, but patches of
litter, especially of leaves and matted dead material,
were still present. Moderate: Most litter consumed, ex-
cept for small amounts matted on the ground or in de-
pressions. High: No litter remained, except in deep so-
lution holes in the limestone.
Although these assignments were made separately for
the three vegetation components, in more than 95% of the
submodules the assignments were the same for all three
components, and thus a single assignment of fire intensity
was possible per submodule. Spier and Snyder (1998)
demonstrated that use of consumption to estimate fire in-
tensity gave similar results as more direct measurements
based on fire temperature tablets. Although some subjec-
tive judgment was necessary to delimit where one category
of fire intensity ended and another began, these cut-offs
were biologically relevant—they delimited different re-
sponses involved in recovery. For example, after a low in-
tensity fire grasses regrew existing leaves, after a moderate
fire they replaced leaves, and after a high intensity fire
they replaced entire culms. Finally, our assignment tech-
nique offers advantages over other techniques (e.g., more
direct measurement of fuel consumption or of maximum
Figure 2. A diagram of a plot and the patchiness and intensity of a
typical fire (dark gray, moderate intensity; light gray, low intensity;
white, no fire). Submodules (black) are of a good size to estimate
patchiness and intensities of fires. Within this pattern of burn com-
munity types also vary. Two submodules (within circles of dashes) are
of a different vegetation community than the other six submodules.
Patchiness and Intensity of Prescribed Fires
MARCH
2003
Restoration Ecology
95
fire temperatures) in that they allow comparison across
different plant communities and enable many plots to be
sampled quickly using standard criteria.
The three independent variables that were assigned to
each submodule were seasonal timing, fire frequency, and
plant community. “Seasonal timing” was assigned based
on when prescribed fires occurred. We categorized each
fire and the submodules within that fire as early (before
July), middle (during July), or late lightning season (after
July). “Fire frequency” was assigned based on how often a
transect had burned since 1990 (when all blocks were
burned). If a transect had burned between 1990 and a fire
we were studying, then the submodules within that
transect were categorized as “frequently burned.” If a
transect had not burned, then its submodules were as-
signed as “infrequently burned.” “Plant community” was
assigned based on an ordination classification of abun-
dances of plant species in submodules. For every submod-
ule we estimated the cover of each species as the number
of 10
10–cm squares in which each species occurred
(maximum of 10,000). These data were used in a cluster
analysis to group submodules into discrete categories.
Though any delineation of a plant community requires
some subjective judgment, ordination techniques help in
keeping these judgments as objective as possible and may
also help suggest patterns. In the analysis we first gener-
ated a dissimilarity matrix using a Bray-Curtis distance
metric (Bray & Curtis 1957) using an SAS program writ-
ten by Drewa (1999; Drewa et al. 2002a). This matrix was
analyzed using the SAS CLUSTER procedure, and the re-
sults were converted into a dendrogram using the TREE
procedure. We then identified the most distinctive clusters
using the NCLUSTERS option. Note that because Bray-
Curtis measurements preferentially weigh common spe-
cies more than rare species, submodules were grouped
based on differences in relative cover of the more abun-
dant species (Drewa et al. 2002a, 2002b). Once the sub-
modules were assigned to clusters, the species composition
of each cluster was examined, compared with previous
work done on the species composition in the transects
(DeCoster et al. 1999; Schmitz et al. 2001), and then as-
signed a plant community (e.g., long-hydroperiod prairie).
Previous work defining plant communities in Long Pine
Key (DeCoster et al. 1999) was done on the plot scale and
thus was of insufficient resolution to use in this study. No-
menclature of species was taken from Avery and Loope
(1996). Figure 2 gives an example of how plant communities
could vary in a plot and how these interacted with fires.
We explored the relationship between the dependent
and independent variables using logistic regression analy-
ses (SAS LOGISTIC procedure; SAS Institute 1999). For
fire patchiness binary logistic regression analysis was used
because fire patchiness has two levels (“yes” and “no”).
For fire intensity logistic regression for ordered categories
was the most appropriate test because fire intensity is or-
dered (low, moderate, and high) (Allison 1999). For the
logistic regression for ordered categories the LOGISTIC
procedure provided a score test to test for the propor-
tional odds assumption.
Results
Plant Communities
Cluster analysis divided the submodules into eight distinct
groups (Fig. 3). Dominant species in each group are listed
in Appendix 1.
Five clusters were identified as belonging to plant com-
munities already defined by DeCoster et al. (1999). Three
of these occurred outside of pine overstory. Two of these
contained 127 submodules and represented different man-
ifestations of short-hydroperiod prairie, which had high
cover and were dominated by grasses and sedges (
Muhlen-
bergia
filipes
,
Cladium
jamaicense
,
Schizachyrium
rhi-
zomatum
,
Rhychospora
divergens
) and the forb
Centella
asiatica
. A third cluster (35 submodules) represented long-
hydroperiod prairie, containing mostly sawgrass (
Cladium
jamaicense
), a hydrophilic forb (
Phyla nodiflora
), and
a
grass (
Paspalum monostachyum
). The forth and fifth clus-
ters occurred beneath pine overstory. One of these (81
submodules) represented high pine savanna and was dom-
inated by firegrass (
Andropogon
cabanisii
), a parasitic
Figure 3. Dendrogram of the eight most important
clusters delineated by cluster analysis using the Bray-
Curtis distance measure. Plant communities delin-
eated include SHP (short-hydroperiod prairie, two
clusters), LHP (long-hydroperiod prairie), HPS (high
pine savanna), SPP (shrub/palm patches), LPS (low
pine savanna), and two “outlier” clusters (submod-
ules of less distinct character not in any of the other
six types).
Patchiness and Intensity of Prescribed Fires
96
Restoration Ecology
MARCH
2003
vine (
Cassytha
filiformis
), a forb (
Crotolaria pumila
), and
bracken fern (
Pteridium
caudatum
). The other cluster repre-
sented low pine savanna (56 submodules) and was repre-
sented by a high cover of species found in short-hydroperiod
prairie (including
Muhlenbergia
filipes
and
Schizachyrium
rhizomatum
) and moderate cover of species found in high
pine savanna (
Andropogon
cabanisii
and the two palms
Serenoa repens
and
Sabal palmetto
).
Three clusters did not fit into plant communities de-
fined by previous work (Fig. 3) (DeCoster et al. 1999),
which was probably a result of our finer scale of sampling.
Of these three clusters, one with 62 submodules was the
most important. This cluster was characterized by higher
elevation submodules that occurred beneath pine over-
story, had low overall cover, and was dominated by shrubs
(
Guettarda scabra
and
Myrica cerifera
) and the vine/shrub
(
Chiococca parviflora
). This cluster also had high relative
cover of the two palms. Submodules in this cluster, which
we designated as “shrub/palm patches,” were intermixed
with low and high pine savanna submodules dominated by
grasses. The other two clusters (23 submodules) were
“community outliers” and were characterized by unusual
assemblages of plants. We excluded these latter two clus-
ters from data analyses because their small numbers of
submodules prevented meaningful analysis.
Intensity and Patchiness of Fires
Plant communities tended to burn more thoroughly at
high rather than low elevations (Table 1). Almost all sub-
modules in high (91%) and low (89%) pine savannas
burned in prescribed fires. Despite being intermixed with
low and high pine savanna submodules, a significantly
smaller proportion of the submodules in shrub/palm patches
(80%) burned (Table 2). Smaller proportions of submod-
Table 1.
Numbers (percentages) of submodules (10 m
2
) in each of five types of Everglades “upland” savanna communities that
burned/did not burn (fire patchiness) and that burned at different intensities in prescribed fires between 1997 and 2000.
Fire Patchiness Fire Intensity
Community Type Burned Not Burned Total Low Medium High Total
High pine savanna 108 (91%) 11 (9%) 119 23 (22%) 74 (68%) 11 (10%) 108
Low pine savanna 62 (89%) 8 (11%) 70 22 (36%) 38 (61%) 2 (3%) 62
Shrub pine savanna 79 (80%) 20 (20%) 99 33 (42%) 45 (57%) 1 (1%) 79
Short-hydroperiod prairie 126 (68%) 58 (32%) 184 48 (38%) 77 (61%) 1 (1%) 126
Long-hydroperiod prairie 9 (19%) 39 (81%) 48 3 (33%) 6 (67%) 0 (0%) 9
Total no. of plots 384 136 520 129 240 15 384
Table 2.
Logistic regression analyses of the effects of plant community on fire patchiness and intensity.
Fire Patchiness Fire Intensity*
Estimate Standard Error Estimate Standard Error
Community Type
High pine savanna (HPS) 3.75 0.49 0.86 0.71
Low pine savanna (LPS) 3.51 0.53 0.00 0.72
Shrub/pine patches (SPP) 2.84 0.45 0.29 0.71
Short-hydroperiod prairie (SHP) 2.47 0.41 0.26 0.70
Long-hydroperiod prairie (LHP) 0 0 0 0
Odds Ratio
2
p Odds Ratio
2
p
Comparison
HPS vs. LPS 1.27 0.23 0.63 2.37 6.39 0.01
HPS vs. SPP 2.48 5.09 0.02 2.53 13.22 0.0003
HPS vs. SHP 3.58 12.32 0.0004 2.46 14.94 0.0001
HPS vs. LHP 42.5 59.37 0.0001 2.37 1.46 0.23
LPS vs. SHS 1.96 2.23 0.13 1.34 0.74 0.39
LPS vs. SPP 2.82 6.24 0.01 1.30 0.68 0.41
LPS vs. LHP 33.6 44.44 0.0001 1.00 0.00 1.00
SPP vs. SHP 1.44 1.40 0.23 0.96 0.01 0.91
SPP vs. LHP 17.1 40.45 0.0001 0.75 0.17 0.68
SHP vs. LHP 11.9 36.38 0.0001 0.77 0.14 0.71
Estimates for patchiness and intensity are based on relative chances that a submodule in a given type of plant community will burn or will experience a more intense
fire than a submodule in long-hydroperiod prairie (the lowest elevation community). In pairwise comparisons, the odds ratio indicates the higher incidence of fire or of
more intense fires of the first community compared with the second community. The
2
test and probability indicate if the odds are likely to be significantly different
from 1.0 (equally likely).
* The score test for the proportional odds assumption was not significant (chi-square
4.94,
df
4,
p
0.29), indicating that it was valid to use the test.
Patchiness and Intensity of Prescribed Fires
MARCH
2003
Restoration Ecology
97
ules burned in the lower elevation prairies than in pine sa-
vannas. Sixty-eight percent of the submodules in short-
hydroperiod prairies and only 19% of the submodules in
long-hydroperiod prairies burned in prescribed fires (Ta-
ble 1). Differences between high and low pine savannas
and short- and long-hydroperiod prairies were statistically
significant (Table 2). Fire patchiness in the shrub patches
did not differ significantly from that in short-hydroperiod
prairies but did differ from that in long-hydroperiod prai-
ries. A significantly larger proportion of submodules burned
in short-hydroperiod prairies than long-hydroperiod prairies
(Table 2).
High pine savanna submodules tended to burn more in-
tensely than submodules in the other plant communities.
Ten percent of the submodules in high pine savanna
burned at high intensities, but less than 3% of the submod-
ules in the other plant communities burned at high intensi-
ties (Table 1). In addition, fewer submodules in high pine
savanna burned at low intensity than in the other plant
communities (22% compared with 33–42%). These differ-
ences between high pine savanna and the other plant com-
munities were highly significant (Table 2), except for the
comparison with long-hydroperiod prairie, which had so
few submodules burn that it did not have enough repli-
cates for a powerful test (Table 1). All other comparisons
of fire intensities between plant communities were not sig-
nificant (Table 2).
Fire frequency affected the likelihood that submodules
would burn (Table 3A). Frequently burned submodules
burned significantly less often (73%) than infrequently
burned submodules (82%). Frequently burned submod-
ules tended to burn more intensely, however, than infre-
quently burned submodules. Infrequently burned submod-
ules were 51% likely to have a moderate-intensity fire,
compared with 65% for frequently burned submodules. In
addition, frequently burned submodules were 6% likely to
burn with high intensity, whereas infrequently burned sub-
modules never burned with high intensity. These differ-
ences were statistically significant (Table 3A).
Seasonal timing of fires affected both the patchiness and
intensity of prescribed fires (Table 3B). Fires in the late
lightning season tended to burn both more patchily and
less intensely than fires set earlier in the lightning season.
About 44% of the submodules in prescribed fires of the
late lightning season did not burn, whereas significantly
fewer submodules (20%) did not burn in early or middle
lightning season fires. Similarly, early and middle lightning
season submodules were more likely to have intense fires,
having over 65% chance for any given submodule to burn
in the moderate to high intensity category. Late lightning
Table 3.
Logistic regression analysis of the effect of fire frequency (A) and fire season (B) on fire patchiness and intensity.
Fire Patchiness Fire Intensity
a
A. Fire Frequency Estimate Standard Error Estimate Standard Error
Frequency
Infrequent 0.50 0.26
0.92 0.23
Frequent 0 0 0 0
Odds Ratio
2
p Odds Ratio
2
p
Comparison
Infrequent vs. frequent 1.65 3.8 0.05 0.40 15.4
0.0001
Fire Patchiness Fire Intensity
b
B. Fire Season Estimate Standard Error Estimate Standard Error
Fire Season
Early 1.15 0.32 1.53 0.37
Middle 1.12 0.26 1.04 0.32
Late 0 0 0 0
Odds Ratio
2
p Odds Ratio
2
p
Comparison
Early vs. Middle 1.0 0.0 0.93 1.6 3.3 0.07
Early vs. Late 3.1 12.9 0.0003 4.6 17.0 0.0001
Middle vs. Late 3.1 19.0 0.0001 2.8 10.9 0.001
Estimates for patchiness and intensity are based on relative chances that an infrequently burned submodule will burn or will experience a more intense fire than a fre-
quently burned submodule or that a submodule burned in the early or middle of the lightning/wet season will burn or experience a more intense fire than a submodule
burned late in the lightning/wet season. In pairwise comparisons, the odds ratio indicates the higher incidence of fire or of more intense fires compared with fires at dif-
ferent frequencies or fire seasons. The
2
test and probability indicate if the odds are likely to be significantly different from 1.0 (equally likely).
a
Score test for the proportional odds assumption:
2 2.6, df 1, p 0.11.
bScore test for the proportional odds assumption: 2 3.4, df 2, p 0.18.
Patchiness and Intensity of Prescribed Fires
98 Restoration Ecology MARCH 2003
season submodules, however, burned significantly less in-
tensely (42% in the moderate intensity category and never
in the high intensity category). There was some tendency
for early lightning season fires to burn more intensely than
middle lightning season fires, with 78% of the early light-
ning season submodules to burn with moderate to high in-
tensities, whereas 65% of the middle lightning season sub-
modules burned in the moderate to high fire categories (a
marginally significant difference). There was no significant
difference between the early and middle lightning seasons
in fire patchiness.
Discussion
Over the past decade substantial variation has occurred in
patchiness and intensity of prescribed fires ignited in the
ground cover of savannas in the Long Pine Key region of
Everglades National Park. Our study indicates that this
variation was influenced by fire frequency, fire season, and
fuel characteristics of different plant communities present
within the Everglades landscapes.
As predicted higher elevation plant communities
burned less patchily and more intensely than lower eleva-
tion plant communities. We propose that the patchiness of
prescribed fires was affected by differences in moisture
content of fine fuels along the topographic gradient. The
first thunderstorms of the lightning season moisten the
peaty substrates in lower areas and fill the lowest sites (i.e.,
long-hydroperiod prairies) with standing water within a
few weeks, thus resulting in patchy fires (W. J. Platt, per-
sonal observation). Higher elevation sites (often only 10–
20 cm higher) remained dry enough to burn more thor-
oughly (see also Snyder 1986; Negrón-Ortiz & Gorchov
2000). Everglades hydrological records indicate that pre-
scribed fires we studied occurred during exceptionally wet
years; we suggest that during drier years lower elevation
plant communities might burn much less patchily. Such
longer-term differences in moisture content at the times of
fires might influence the tendency to burn over large
areas.
Fire intensity also was affected by the types of fuels in
the different plant communities. High intensity fires oc-
curred about 10% of the time in high pine savannas but
less than 3% in the other plant communities. Although
abundant fine fuels occurred in all communities, only high
pine savanna fuels appear particularly flammable and py-
rogenic. In particular, firegrass (Andropogon cabanisii)
tends to be much more abundant in high pine savannas
than other communities (DeCoster et al. 1999; this study).
This species, like warm season grasses of other pine savan-
nas (Streng et al. 1993; Platt 1999; Platt & Gottschalk
2001), produces abundant leaves that when dry provide
fuels for high intensity fires. In addition, high densities of
slash pines (Doren et al. 1993; Platt et al. 2000) shed abun-
dant quantities of needles that form continuous layers of
flammable fuels (Herndon & Taylor 1985). Reduced
patchiness and increased intensity of lightning season fires
have also been documented in other savannas with pines
and warm season grasses (Platt et al. 1988a, 1991; Streng et
al. 1993; Glitzenstein et al. 1995; Olson & Platt 1995;
Drewa et al. 2002b). In these savannas lightning season
fires favor pines and herbaceous species in the ground
cover but do not completely eliminate trees and shrubs
(Walker & Peet 1983; Platt et al. 1988a, 1988b; Rebertus et
al. 1989a, 1989b; Streng et al. 1993; Glitzenstein et al. 1995;
Olson & Platt 1995; Drewa et al. 2002b).
Patches of shrubs occurred within larger areas of pine
savannas dominated by herbs and grasses. Ground cover
of resprouting shrubs (e.g., Guettarda scabra, Myrica cer-
ifera, Chiococca parvifolia, Morinda royoc) and palms
(Serenoa repens and Sabal palmetto) burned more patchily
and less intensely than the surrounding vegetation. Be-
cause frequent early lightning season fires shift dominance
away from woody species toward grass and herbs (De-
Coster et al. 1999; Platt 1999), these residual patches of
shrubs may occur where local edaphic conditions, such as
water-filled solution holes that buffer local microclimates
(Olmsted et al. 1983, 1993; Schmitz et al. 2001), increase
patchiness, and decrease intensity of prescribed fires.
Long Pine Key contains discrete patches of different hard-
wood-dominated communities that range from shrub
thickets to subtropical forests dominated by large trees
(Robertson 1953; Olmsted et al. 1983; Snyder et al. 1990).
These forests only burn during large-scale fires in dry
years (Robertson & Platt 1992, 2001). Patches of shrubs
that burn less completely and intensely than surrounding
herbaceous-dominated savannas might represent one end
of a continuum of hardwood-dominated patches that burn
at different frequencies and intensities.
Changing seasonal timing produced predicted effects on
patchiness and intensity of prescribed fires. Early and mid-
lightning season fires were considerably more likely than
late fires to burn a given area and to burn that area at high
intensity. These results resembled others from the Ever-
glades region (e.g., Negrón-Ortiz & Gorchov 2000), as
well as those from studies of warm-temperate longleaf
pine savannas (e.g., Streng et al. 1993; Glitzenstein et al.
1995) and fires in a tropical savanna (Williams et al. 1998).
Fine fuels, driest in April and May before the onset of the
lightning season in both subtropical Everglades savannas
and warm-temperate longleaf pine savannas, become pro-
gressively more moist during the lightning season, result-
ing in increased patchiness and decreased intensity of pre-
scribed fires. Because water levels reached seasonal highs
by September in short- and long-hydroperiod prairies, nat-
ural fires occurring after the middle lightning season prob-
ably almost never burned large areas, even within pine
savannas. In fact, igniting prescribed fires late in the light-
ning season required daily analysis of local weather condi-
tions to identify those infrequent conditions under which
prescribed fires might be expected to burn (Everglades
National Park fire records).
Differences in fire frequency produced both expected
and unexpected effects on characteristics of prescribed
Patchiness and Intensity of Prescribed Fires
MARCH 2003 Restoration Ecology 99
fires. As predicted frequently burned sites burned more
patchily than infrequently burned sites, probably because
the infrequently burned sites had more fine fuels and
evenly distributed fuels, especially litter, than frequently
burned sites. However, these larger amounts of fine fuels
also suggest that infrequently burned sites should burn
more intensely, but we did not find this to be the case. We
attribute this to differences in quality of fuels. The ground
cover in more frequently burned sites, especially in the
pine savannas, may have had drier and finer fuels, consist-
ing mostly of warm season grasses that regrow rapidly
after fire (DeCoster et al. 1999) and produce abundant dry
fine fuels near ground level (Platt & Gottschalk 2001).
Less frequently burned sites, however, may have had
thicker litter layers, which trap moisture, as well as shrubs,
which produce less fuel and also cast shade that raises
moisture levels in soils and litter.
Conclusions
Prescribed fires ignited in the early lightning season pro-
duced a wide range of patchiness and intensity within
Everglades savanna communities. Within 3 years of an
early lightning season prescribed fire, fine fuels were suffi-
cient for subsequent fires to reburn large areas of Ever-
glades savanna. When timing was shifted later into the
lightning season or when frequency was decreased, the
amount of unburned area increased and fewer high inten-
sity patches occurred in prescribed fires. Thus, our study
suggests that shifting prescribed fires away from the timing
and frequencies that occurred in presettlement times re-
sults in less area burned and a smaller range of post-fire
environmental conditions resulting from differences in lo-
cal fire intensities. Continued use of non-natural timing
and frequency in prescribed fires may eventually lead to
the establishment of extensive shrub communities and the
decline of species-rich herbaceous vegetation (DeCoster
et al. 1999).
Frequent early lightning season prescribed fires can
generate a savanna-like aspect of plant communities (Platt
1999) and increase plant species biodiversity in the ground
cover (DeCoster et al. 1999). Effects of these fires resem-
ble those occurring during natural lightning-initiated fires,
such as the lightning-initiated Ingraham Fire that burned
about 50,000 ha and 10% of Long Pine Key in 1989. In that
fire patchiness was least in pine savannas and increased
downslope, with long-hydroperiod prairies burning most
patchily (W. J. Platt, personal observation).
Our study suggests ways that prescribed fires might be
manipulated to mimic natural fires in Everglades savanna
landscapes. In drier years natural ignitions may have oc-
curred in a variety of savanna habitats, as indicated by
Everglades fire records. Once pine savannas, especially high
pine savannas with pyrogenic vegetation, were ignited
they probably would have carried fires considerable dis-
tances, burning large areas more thoroughly and intensely
than lower elevation communities. Nonetheless, such fires
should generate patches of shrubs that burn at different
frequencies and intensities, producing a heterogeneous
landscape. Long-hydroperiod prairies probably functioned
in most years as natural fire breaks and may have had a
substantial effect on the likelihood of fires being transmit-
ted to adjacent pine savannas. Prescribed fires that burn
large areas, mimicking fires like the Ingraham Fire, may
be possible only in years that these natural fire breaks are
dry enough to carry fire. In the intervening years it should
be possible to design large-scale prescribed fires in which
recently burned areas and long-hydroperiod prairies act to
contain prescribed fires without using artificial breaks
such as roads.
Acknowledgments
We thank Jim DeCoster, Lisu Derungs, Martin Schmitz,
Sarah Riley, Heather Ducharme, Steve Newland, Steve
McMann, Chris Thibodeaux, Billy Platt, and David Baker
for assistance in establishing plots and sampling vegetation.
Ecological approaches to prescribed fire management of
Everglades habitats have been developed by Bob Doren,
Sue Husari, David Lentz, Bill Kaage, Bob Panko, Janet
Passek, John Segar, and Nate Benson and were first insti-
tuted by Everglades National Park superintendent Michael
Findlay. Support for this study was provided by the Na-
tional Park Service of the U.S. Department of the Interior
and by Everglades National Park.
LITERATURE CITED
Abrahamson, W. G., and D. C. Hartnett. 1990. Pine flatwoods and dry
prairies. Pages 103–149 in R. L. Myers and J. J. Ewel, editors. Eco-
systems of Florida. University of Florida Press, Orlando.
Allison, P. D. 1999. Logistic regression using the SAS system: theory and
application. SAS Institute Inc., Cary, North Carolina.
Avery, G. N., and L. L. Loope. 1996. Plants of Everglades National Park:
a preliminary checklist of vascular plants. 3rd edition edited by R. G.
Reimus. South Florida Research Center Report T-574. Everglades
National Park, Homestead, Florida.
Bray, J. R., and J. T. Curtis. 1957. An ordination of the upland forest com-
munities of southern Wisconsin. Ecological Monographs 27:325–349.
Brewer, J. S., and W. J. Platt. 1994. Effects of fire season and herbivory on
reproductive success in a clonal forb, Pityopsis graminifolia (Michx.)
Nutt. Journal of Ecology 82:665–675.
Chapman, H. H. 1932. Is the longleaf type a climax? Ecology 13:328–334.
Chen, E., and J. F. Gerber. 1990. Climate. Pages 11–34 in R. Myers and J.
Ewel, editors. Ecosystems of Florida. University of Florida Press,
Orlando.
Craighead, F. C. 1971. The trees of south Florida. Vol. 1. The natural envi-
ronments and their succession. University of Miami Press, Coral
Gables, Florida.
Davis, J. H. 1943. The natural features of southern Florida. Florida Geo-
logical Survey, Bulletin 25. Florida Geological Survey, Tallahassee,
Florida.
DeCoster, J., W. J. Platt, and S. A. Riley. 1999. Pine savannas of Ever-
glades National Park: an endangered ecosystem. Pages 81–88 in D. T.
Jones and B. W. Gamble, editors. Florida’s garden of good and evil:
proceedings of the joint symposium of the Florida Exotic Pest Plant
Council and the Florida Native Plant Society, June 1998, Palm
Beach Gardens, Florida. South Florida Water Management District,
West Palm Beach, Florida.
Patchiness and Intensity of Prescribed Fires
100 Restoration Ecology MARCH 2003
DeGrove, J. M. 1984. History of water management in south Florida.
Pages 22–27 in P. J. Gleason, editor. Environments of south Florida:
present and past. 2nd edition. Miami Geological Society, Coral
Gables, Florida.
Doren, R. F., W. J. Platt, and L. D. Whiteaker. 1993. Density and size
structure of slash pine stands in the everglades region of south Flor-
ida. Forest Ecology and Management 59:295–311.
Doren, R. F., and R. M. Rochefort. 1984. Summary of fires in Ever-
glades National Park and Big Cypress National Preserve, 1981.
South Florida Research Center Report SFRC-84/01. Homestead,
Florida.
Drewa, P. B. 1999. Community structure and the effects of experimen-
tal fires on hardwood shrub species in southeastern longleaf
pine savannas. PhD dissertation. Louisiana State University, Ba-
ton Rouge.
Drewa, P. B., W. J. Platt, and E. B. Moser. 2002a. Community structure
along elevation gradients in southeastern longleaf pine savannas.
Plant Ecology 160:61–78.
Drewa, P. B., W. J. Platt, and E. B. Moser. 2002b. Fire effects on resprout-
ing of shrubs in southeastern longleaf pine savannas. Ecology 83:
755–767.
Frost, C. C. 1993. Four centuries of changing landscape patterns in the
longleaf pine ecosystem. Proceedings of the Tall Timbers Fire Ecol-
ogy Conference 18:17–44.
Glitzenstein, J. S., W. J. Platt, and D. R. Streng. 1995. Effects of fire re-
gime and habitat on tree dynamics in north Florida longleaf pine sa-
vannas. Ecological Monographs 65:441–476.
Goodman, S. J., and H. J. Christian. 1993. Global observations of light-
ning. Pages 191–219 in R. J. Gurney, J. L. Foster, and C. L. Parkin-
son, editors. Atlas of satellite observations related to global change.
Cambridge University Press, Cambridge.
Harper, R. M. 1927. Natural resources of southern Florida. 18th Annual
Report, Florida Geological Survey, Tallahassee, Florida.
Harshberger, J. W. 1914. The vegetation of south Florida, south of 2730
north, exclusive of the Florida Keys. Transactions of the Wagner
Free Institute of Science 7:49–189.
Hela, I. 1952. Remarks on the climate of south Florida. Bulletin of the
Marine Society of the Gulf of Mexico and Caribbean Sea 2:438–
447.
Herndon, A., and D. L. Taylor. 1985. Litterfall in pinelands of Everglades
National Park. South Florida Research Center Report SFRC-85/01.
Everglades National Park, Homestead, Florida.
Heyward, F. 1939. The relation of fire to stand composition of longleaf
pine forests. Ecology 20:287–304.
Hiers, J. K., R. Wyatt, and R. J. Mitchell. 2000. The effects of fire regime
on legume reproduction in longleaf pine savannas: is season selec-
tive? Oecologia 125:521–530.
Hodanish, S., D. Sharp, W. Collins, C. Paxton, and R. E. Orville. 1997. A
ten year monthly lightning climatology of Florida: 1986–1995.
Weather Forecasting 12:427–446.
Hoffmeister, J. E., K. W. Stockman, and H. G. Multer. 1967. Miami lime-
stone of Florida and its recent Bahamian counterpart. Bulletin of the
Geological Society of America 78:175–190.
Huffman, J. M., and S. W. Blanchard. 1991. Changes in woody vegetation
in Florida dry prairie and wetlands during a period of fire exclusion
and after dry-growing season fire. Pages 75–83 in S. C. Nodvin and
T. A. Waldrop, editors. Fire and the environment: ecological and
cultural perspectives. USDA Forest Service, General Technical Re-
port SE-69, U.S. Forest Service, Washington, D.C.
Johnson, E. A. 1992. Fire and vegetation dynamics: studies from the North
American boreal forest. Cambridge University Press, Cambridge.
Komarek, E. V., Sr. 1964. The natural history of lightning. Proceedings of
the Tall Timbers Fire Ecology Conference 3:139–183.
Maier, M. W., A. G. Boulanger, and R. I. Sax. 1979. An initial assessment
of flash density and peak current characteristics of lightning flashes
to ground in south Florida. U.S. Nuclear Regulatory Commission
Report CR-1024.
Mulligan, M. K., and L. K. Kirkman. Burning influence on wiregrass
(Aristida beyrichiana) restoration plantings: natural seedling recruit-
ment and survival. Restoration Ecology 10:334–339.
Negrón-Ortiz, V., and D. L. Gorchov. 2000. Effects of fire season and
post-fire herbivory on the cycad Zamia pumila (Zamiaceae) in slash
pine savanna, Everglades National Park, Florida. International Jour-
nal of Plant Sciences 161:659–669.
Olmsted, I. C., and T. V. Armentano. 1997. Vegetation of Shark Slough,
Everglades National Park. South Florida Natural Resources Center
Technical Report 97-001. National Park Service, Everglades Na-
tional Park, Homestead, Florida.
Olmsted, I. C., W. B. Robertson Jr., J. Johnson, and O. L. Bass. 1983. The
vegetation of Long Pine Key, Everglades National Park. South Flor-
ida Research Center Report SFRC-83/05, National Park Service, At-
lanta, Georgia.
Olmsted, I., H. Dunevitz, and W. J. Platt. 1993. Effects of freezes on tropi-
cal trees in Everglades National Park, Florida, USA. Tropical Ecol-
ogy 34:17–34.
Olson, M. S., and W. J. Platt. 1995. Effects of habitat and growing season
fires on resprouting of shrubs in longleaf pine savannas. Vegetatio
119:101–118.
Parker, G. G. 1974. Hydrology of the pre-drainage system of the Ever-
glades in southern Florida. Pages 18–27 in P. J. Gleason, editor. En-
vironments of south Florida: present and past. Memoir 2. Miami
Geological Society, Miami, Florida.
Platt, W. J. 1999. Southeastern pine savannas. Pages 23–51 in R. C. Ander-
son, J. S. Fralish, and J. Baskin, editors. The savanna, barren, and
rock outcrop communities of North America. Cambridge University
Press, Cambridge.
Platt, W. J., R. F. Doren, and T. Armentano. 2000. Effects of Hurricane
Andrew on stands of slash pine (Pinus elliottii var. densa) in the
Everglades region of south Florida (USA). Plant Ecology 146:43–60.
Platt, W. J., G. W. Evans, and M. M. Davis. 1988a. Effects of fire season
on flowering of forbs and shrubs in longleaf pine forests. Oecologia
76:353–363.
Platt, W. J., G. W. Evans, and S. L. Rathbun. 1988b. The population dy-
namics of a long-lived conifer (Pinus palustris). The American Natu-
ralist 131:491–525.
Platt, W. J., J. S. Glitzenstein, and D. R. Streng. 1991. Evaluating pyroge-
nicity and its effects on vegetation in longleaf pine savannas. Pro-
ceedings of the Tall Timbers Fire Ecology Conference 17:143–161.
Platt, W. J., and R. M. Gottschalk. 2001. Effects of exotic grasses on po-
tential fine fuel loads in the groundcover of south Florida slash pine
savannas. International Journal of Wildland Fire 10:155–159.
Platt, W. J., and R. K. Peet. 1998. Ecological concepts in conservation bi-
ology: lessons from southeastern ecosystems. Ecological Applica-
tions 8:907–908.
Platt, W. J., and M. W. Schwartz. 1990. Temperate hardwood forests.
Pages 194–229 in R. Myers and J. Ewel, editors. Ecosystems of Flor-
ida. University of Florida Press, Orlando.
Porter, C. L., Jr. 1967. Composition and productivity of a subtropical prai-
rie. Ecology 48:937–942.
Provencher, L., B. J. Herring, D. R. Gordon, H. L. Rodgers, K. E. M. Gal-
ley, G. W. Tanner, J. L. Hardesty, and L. A. Brennan. 2001. Effects
of hardwood reduction techniques on longleaf pine sandhill vegeta-
tion in northwest Florida. Restoration Ecology 9:13–27.
Rebertus, A. J., G. B. Williamson, and E. B. Moser. 1989a. Fire-induced
changes in Quercus laevis spatial pattern in Florida sandhills. Jour-
nal of Ecology 77:638–650.
Rebertus, A. J., G. B. Williamson, and E. B. Moser. 1989b. Longleaf pine
pyrogenicity and turkey oak mortality in Florida xeric sandhills.
Ecology 70:60–70.
Robertson, K. M., and W. J. Platt. 1992. Effects of fire on bromeliads in
Patchiness and Intensity of Prescribed Fires
MARCH 2003 Restoration Ecology 101
subtropical hammocks of Everglades National Park, Florida. Selby-
ana 13:39–49.
Robertson, K. M., and W. J. Platt. 2001. Effects of multiple disturbances
(fire, hurricane) on epiphyte-host tree associations in a subtropical
forest, Florida, USA. Biotropica 33:573–582.
Robertson, W. B., Jr. 1953. A survey of the effects of fire in Everglades
National Park. Mimeographed Report, U.S. Department of the Inte-
rior, National Park Service, Atlanta, Georgia.
SAS Institute, Inc. 1999. SAS/STAT user’s guide. Version 8. 4th edition.
Cary, North Carolina.
Schmitz, M., W. J. Platt, and J. DeCoster. 2001. Small-scale geomorpho-
logical heterogeneity and numbers of plant species in pine savannas
and short-hydroperiod prairies of Everglades National Park, Flor-
ida. Plant Ecology 160:137–148.
Snyder, J. R. 1986. The impact of wet season and dry season fires on
Miami Rock Ridge Pineland, Everglades National Park. South Flor-
ida Research Center Report SFRC-86/06, National Park Service, At-
lanta, Georgia.
Snyder, J. R. 1991. Fire regimes in subtropical south Florida. Proceedings
Tall Timbers Fire Ecology Conference 17:303–319.
Snyder, J. R., A. K. Herndon, and W. B. Robertson, Jr. 1990. South Flor-
ida rockland ecosystems: tropical hammocks and pinelands. Pages
230–274 in R. Myers and J. Ewel, editors. Ecosystems of Florida.
University of Florida Press, Orlando.
Sparks, J. C., R. E. Masters, D. M. Engle, and G. A. Bukenhofer. 2002.
Season of burn influences fire behavior and fuel consumption in re-
stored shortleaf pine-grassland communities. Restoration Ecology
10:714–722.
Spier, L. P., and J. R. Snyder. 1998. Effects of wet- and dry-season fires on
Jacquemontia curtisii, a south Florida pine forest endemic. Natural
Areas Journal 18:350–357.
Streng, D. R., J. S. Glitzenstein, and W. J. Platt. 1993. Evaluating season
of burn in longleaf pine forests: a critical literature review and some
results from an ongoing long-term study. Proceedings Tall Timbers
Fire Ecology Conference 18:227–263.
Taylor, D. L. 1981. Fire history and fire records for Everglades National
Park, 1948–1979. South Florida Research Center Report T–619, Na-
tional Park Service, Atlanta, Georgia.
Taylor, D., and A. K. Herndon. 1981. Impact of 22 years of fire on under-
story hardwood shrubs in slash pine communities within Everglades
National Park. South Florida Research Center Report T-640, Na-
tional Park Service, Atlanta, Georgia.
Wade, D., J. Ewel, and R. Hofstetter. 1980. Fire in South Florida ecosys-
tems. U.S. Department of Agriculture, Forest Service, General
Technical Report SE 17, Southeastern Forest Experiment Station,
Asheville, North Carolina.
Walker, J., and R. K. Peet. 1983. Composition and species diversity of
pine-wiregrass savannas of the Green Swamp, North Carolina. Veg-
etatio 55:163–179.
Williams, R. J., A. M. Gill, and P. H. R. Moore. 1998. Seasonal changes in
fire behavior in a tropical savanna in northern Australia. Interna-
tional Journal of Wildland Fire 8:227–239.
Patchiness and Intensity of Prescribed Fires
102 Restoration Ecology MARCH 2003
Appendix 1. Mean frequency of the ten most common species in each of the eight groups identified by cluster analysis.
Plant Community
Species, with Authors and Family HPS LPS SPP SHP1 SHP2 LHP OUT1 OUT2 Total
Cladium jamaicense Crantz Cyperaceae — 116 — 631 770 615 — 200 2,331
Muhlenbergia filipes M.A. Curtis Poaceae 87 423 124 836 588 — — 192 2,250
Rhynchospora divergens Chapman ex M.A. Curtis Cyperaceae — — — 262 829 — — 127 1,217
Schizachyrium rhizomatum (Swallen) Gould Poaceae — 384 —300 380 — — — 1,064
Andropogon cabanisii Hackel Poaceae 711 163 — — — — — — 874
Dichanthelium dichotomum (L.) Gould Poaceae — — — 112 — — 93 489 694
Serenoa repens (W. Bartram) Small Arecaceae 149 185 158 — — — — 171 663
Cassytha filiformis L. Lauraceae 392 128 131 — — — — — 651
Centella asiatica (L.) Urban Apiaceae — — — 205 440 — — — 645
Panicum tenerum Beyrich ex Trinius Poaceae — — — 85 229 — 218 — 531
Sabal palmetto (Walter) Loddiges ex Schultes & Schultes f. Arecaceae 188 131 141 — — — — — 459
Paspalum monostachyum Vasey ex Chapman Poaceae — — — 161 — 235 — — 396
Borreria terminalis Small Rubiceae — 149 — — 231 — — — 380
Phyla nodiflora (L.) Greene — — — — — 359 — — 359
Evolvulus sericeus Swartz Convolvulaceae — — — 86 256 — — — 342
Iva microcephala Nuttall Asteraceae — — — — — — 322 — 322
Elytraria caroliniensis (J.F. Gmelin) Persoon Acanthaceae — — — — 297 — — — 297
Mikania scandens (L.) Willdenow Asteraceae — — — — — 147 — 149 295
Chiococca parvifolia Wullschlaegel ex Grisebach Rubiceae — 117 171 — — — — — 289
Guettarda scabra (L.) Vent RubiceaE 91 — 188 — — — — — 278
Phyllanthus species Euphorbiaceae — — 115 — — — — 160 275
Pluchea rosea (L.) Cassini Asteraceae — — — — — 148 — 119 267
Morinda royoc L. Rubiceae — 120 115 — — — — — 234
Dyschoriste angusta (A. Gray) Small Acanthaceae — — — — 220 — — — 220
Myrica cerifera L. Myricaceae — — 195 — — — — — 195
Pteridium caudatum (L.) Maxon Dennstaedtaceae 192 — — — — — — — 192
Andropogon glomeratus (Walter) Britton et al. Poaceae — — — — — — — 191 191
Crotolaria pumila Ortega Fabaceae 183 — — — — — — — 183
Salix caroliniana Michaux Salicaceae — — — — — 175 — — 175
Stenandrium dulce (Cavanilles) Nees Acanthaceae — — — — — — — 172 172
Proserpinaca palustris L. Haloragaceae — — — — — 167 — — 167
Ludwigia microcarpa Michaux Onagraceae — — — — — 164 — — 164
Rhynchospora microcarpa Baldwin ex A. Gray Cyperaceae — — — 100 — — 47 — 147
Cephalanthus occidentalis L. Rubiceae — — — — — 146 — — 146
Hyptis alata (Rafinesque) Shinners Lamiaceae — — — — — 140 — — 140
Dodonaea viscosa (L.) Jacquin Sapindaceae 137 — — — — — — — 137
Erigeron quercifolius Lamarck Asteraceae — — — — — — 131 — 131
Paspalum species Poaceae — — 115 — — — — — 115
Dichromena colorata (L.) Hitchcock Cyperaceae — — — — — — 109 — 109
Myrsine floridana A. DC. Myrsinaceae 100 — — — — — — — 100
Sabatia grandiflora (A. Gray) Small Gentianaceae — — — — — — 61 — 61
Heliotropium leavenworthii (A. Gray) Small Boraginaceae — — — — — — 54 — 54
Eupatorium leptophyllum DC. Asteraceae — — — — — — 41 — 41
Melanthera angustifolia A. Richard Asteraceae — — — — — — 41 — 41
Average frequency/species 223 192 145 278 424 230 112 197 409
No. of submodules 81 56 62 99 28 35 8 15 384
Species that are most abundant within a given group are indicated by bold fonts. Plant communities are as follows: HPS, high pine savanna; LPS, low pine savanna;
SPP, shrub/palm patches; SHP1 and SHP2, two manifestations of short-hydroperiod prairie; LHP, long-hydroperiod prairie; OUT1 and OUT2, groupings of species
that were outliers in the analysis (i.e., having low numbers of submodules).
