Landscape Ecology 19: 153–165, 2004. © 2004 Kluwer Academic Publishers. Printed in the Netherlands.
153
Research article
Anthropogenic influences on potential fire spread in a pyrogenic ecosystem of Florida, USA Brean W. Duncan* and Paul A. Schmalzer Dynamac Corporation, Mail Code:DYN-2, Kennedy Space Center, FL 32899, USA; *Author for correspondence (e-mail:
[email protected]) Received 31 December 2002; accepted in revised form 21 August 2003
Key words: Fire modeling, Fuel fragmentation, FARSIDE, Historic, Pyrogenic, Southeast U.S.
Abstract Fire has historically been an important ecological factor maintaining southeastern U.S. vegetation. Humans have altered natural fire regimes by fragmenting fuels, introducing exotic species, and suppressing fires. Little is known about how these alterations specifically affect spatial fire extent and pattern. We applied historic 共1920 and 1943兲 and current 共1990兲 GIS fuels maps and the FARSITE fire spread model to quantify the differences between historic and current fire spread distributions. We held all fire modeling variables 共wind speed and direction, cloud cover, precipitation, humidity, air temperature, fuel moistures, ignition source and location兲 constant with exception of the fuel models representing different time periods. Model simulations suggest that fires during the early 1900’s burned freely across the landscape, while current fires are much smaller, restricted by anthropogenic influences. Fire extent declined linearly with patch density, and there was a quadratic relationship between fire extent and percent landscape covered by anthropogenic features. We found that as little as 10 percent anthropogenic landcover caused a 50 percent decline in fire extent. Most landscapes 共conservation or non-conservation areas兲 are now influenced by anthropogenic features which disrupt spatial fire behavior disproportionately to their actual size. These results suggest that land managers using fire to restore or maintain natural ecosystem function in pyrogenic systems will have to compensate for anthropogenic influences in their burn planning.
Introduction Anthropogenic influences now play a major role influencing fire regimes throughout the world 共Minnich 1983; Myers and White 1987; Baker 1992; Davis and Burrows 1994; Lunt 1998; Keely et al. 1999; Duncan et al. 1999; Ramos-Neto and Pivello 2000; Fernandes 2001兲. The fire regime of an area is defined by its fire type, fire intensity 共severity兲, fire size, return interval, seasonality, and spatial pattern 共Christensen 1985; Agee 1993兲. A fire regime is the result of many interactions with physical and biophysical variables, which influence vegetative characteristics, distribution of species, biogeochemical cycling, and ecosystem function within an area. Clearly, fire is governed by and influences many naturally complex landscape
processes, which have been further complicated by anthropogenic influences. Fire is one of the most intensely studied disturbances acting on the landscape, but relatively few studies directly quantify anthropogenic influences on the modern fire regime and even fewer quantify spatial aspects of this relationship. Observing and studying spatial behavior of unrestrained natural fire on the modern landscape is difficult and even more so for historic landscapes. Advancements in computer modeling have made it possible to simulate fire behavior including its spatial components over the landscape 共Mladenoff and Baker 1999兲. One of the primary advantages of spatial modeling is that it can provide the opportunity to perform simulations that give perspectives that humans cannot perceive in time and space 共Baker 1999兲. Mecha-
154 nistic modeling procedures are particularly useful for studying spatial fire behavior because they allow the evaluation of how specific environmental factors and spatial and temporal dependencies affect fire patterns 共Finney 1999兲. Spatial models of varying type have been used to explore the theoretical nature of anthropogenic effects on fire behavior. Turner et al. 共1989兲 used a probabilistic, cellular automata-modeling approach to explore how disturbances such as fire were altered by the spatial arrangement of fuels. They found that fragmentation could constrain the propagation of disturbances such as fire on the landscape. A physical and stochastic modeling approach simulated the longterm effects of fuel patterns including anthropogenic fuel fragmentation in California chaparral and found anthropogenic features reduced potential fire size and recurrence 共Davis and Burrows 1994兲. Miller and Urban 共2000兲 used a spatially explicit, probabilistic forest simulation model to explore the effects of fire suppression on fuel connectivity. They found fire suppression increased fuel connectivity, which altered fire behavior primarily under moderate burning conditions. In the Boundary Water Canoe Area, a statistical, spatial modeling approach indicated that fire suppression has caused smaller, less frequent fires 共Baker 1999兲. In this study we use historic and current fuels maps with a mechanistic fire model to quantify the effects of anthropogenic influences 共fuel fragmentation, fire suppression, exotic species兲 on potential fire spread through time. Mechanistic models, also known as process models, represent systems as a set of fundamental processes that each describe cause and effect relationships between physical variables. The mechanistic approach allowed us to isolate the effects of increasing anthropogenic influences by holding all fire modeling variables constant, except the fuel models representing a time series of landscape change in a southeastern pyrogenic community. Landscape indices, particularly patch density, are used to assess changes in fuel structure, and quantitative methods are used to describe changes in spatial fire behavior through time due to anthropogenic features 共buildings, industrial facilities, transportation, and agriculture兲 replacing native fuels. Our study focuses on fire extent and burn pattern alteration, which links our results to fire regime dynamics in a larger context. Previous studies in this area of research have been more theoretical, leaving out quantitative detail, making in-
tegration of this important information with land management practices difficult.
Study site and background Kennedy Space Center 共KSC兲/Merritt Island National Wildlife Refuge 共MINWR兲 comprises 57,000 ha in Brevard and Volusia counties located along the east coast of central Florida, USA 共Figure 1兲. Portions not directly used by the U.S. National Aeronautics and Space Administration 共NASA兲 for space operations support are managed by the U.S. Fish and Wildlife Service as MINWR. Urban development on KSC consists of industrial facilities and infrastructure to support launch operations. KSC/MINWR occupies a barrier island complex comprised of a diverse assemblage of fire-adapted terrestrial vegetative communities. Upland xeric sites are dominated by oak scrub vegetation 共Quercus spp.兲, while mesic sites are dominated by flatwoods 共e.g., Serenoa repens, Lyonia spp., Ilex sp., and an overstory of Pinus elliotii兲 共Schmalzer and Hinkle 1992a,b兲. Because the landscape is comprised of relict dunes forming ridge swale topography, there are interleaving swale marshes and hammocks on hydric soils between the xeric ridges. The swales are dominated by Spartina bakeri and Andropogon spp., while the hardwood hammocks are dominated by Quercus virginiana and Quercus laurifolia that have a structure that is much less flammable than surrounding communities. There is still remnant citrus agriculture on KSC/MINWR. Many groves planted before NASA ownership are currently leased and farmed. Cape Canaveral is a large, geologically-recent 共formed during the Quaternary period兲 barrier island with predominantly well-drained soils. Most of Cape Canaveral, totaling about 6,475 ha, has been within Cape Canaveral Air Force Station since the 1950s. Coastal strand and coastal scrub are the predominant vegetation types with oak scrub inland 共Schmalzer et al. 1999兲. Coastal scrub occurs on neutral to alkaline sandy soils; a shrub form of live oak 共Quercus virginiana兲 is the dominant species along with saw palmetto 共Serenoa repens兲. Oak scrub occurs on older, more leached soils. Typical scrub oaks 共Quercus geminata, Q. myrtifolia, Q. chapmanii兲, saw palmetto, and ericaceous shrubs 共e.g., Lyonia spp., Vaccinium spp.兲 predominate. Much scrub has been unburned for ⬎ 30 years and has developed xeric hammock features 共Schmalzer et al. 1999兲. Mesic
155
Figure 1. Map showing the geographic position of both Kennedy Space Center and Cape Canaveral Air Force Station 共CCAFS兲 in Florida, USA. Geographic barriers to fire are indicated.
hammocks occur near the Banana River. An extensive network of industrial infrastructure and facilities supporting launch operations are present on CCAFS.
Methods We used the Fire Area Simulator model 共FARSITE兲 version 3.0 共Finney 1998兲 for all spatial fire modeling with input directly from ARC/INFO GRID software 共Environmental Systems Research Institute 1997兲. We selected FARSITE because it is both based
on the robust and widely used Rothermal equations 共Rothermel 1972兲, and it provides a mechanism for producing spatial maps of fire spread and distribution. FARSITE requires all spatial inputs to be formatted as ASCII grid 共raster兲 maps. We modeled the landscape at 10 m resolution to include narrow, linear anthropogenic features such as roads. To increase modeling efficiency at this resolution, we split the study area into north, central, and south regions by using natural geographic barriers to fire. We converted 1920, 1943, and 1990 landcover maps 共Duncan et al. 2000, Duncan et al. 2003 in re-
156 Table 1. Landscape inputs for FARSITE simulations on Kennedy Space Center and Cape Canaveral Air Force Station, Florida.
Surface fuel model Canopy cover north and south regions 共%兲 Canopy cover central region 共%兲 Elevation 共m兲 Slope 共deg兲 Aspect 共deg兲
Freshwater marsh
Disturbed freshwater marsh
Oak scrub/ flatwoods
Hammocks/ hardwoods
Non-fuels
1 7 3 3 0 0
2 15 7 3 0 0
7 15 7 3 0 0
8 100 100 3 0 0
98 0 0 3 0 0
view兲 into fuel maps by assigning each landcover type to one of the 13 standard fire behavior fuel models 共Table 1兲 共Anderson 1982兲. Fuel models are used to help represent fire behavior potential and are based on vegetation communities with associated fuel characteristics. Freshwater marshes were assigned to Fuel Model 1 共short grass兲. We considered using the standard fuel model 3 to model freshwater marshes because it is a tall grass model that is well suited for modeling many of our Spartina bakeri marshes. The main problem with its use is that this standard model has no live biomass component, making it unrealistic to use for growing season fires. A secondary problem with its use is that many of our marshes are dominated by shorter grass species such as Calamolvilfa curtissi. Using model 3 caused simulated fires to burn unrealistically fast through these marshes. Disturbed freshwater marshes were assigned to Fuel Model 2. The disturbed freshwater marsh areas were identified on the landcover maps as being anthropogenically disturbed. Many of these marshes have moderate amounts of invasive shrubs and hardwood saplings, altering the physical fuel composition. Oak scrub and flatwoods communities were assigned to Fuel Model 7 共palmetto and shrub understory with sparse pine overstory兲. Hammocks, mixed hardwoods, and wetland hardwoods were assigned to Fuel Model 8 共ground litter fuels under forest canopy兲, and categories such as salt marsh or others associated with water or non-flammable features were assigned to Fuel Model 98 共non-fuel兲. Because our study concentrated on the effects of anthropogenic change and its effect on terrestrial flammability, and because it is difficult to separate flammable from non-flammable salt marsh, we modeled salt marsh as a non-flammable type. The FARSITE model’s limited spotting capabilities made it necessary to model narrow dirt roads as Fire Behavior Fuel Model 8. Experience suggests that
even under moderate meteorological conditions at least one spotting event per fire occurs in this region 共Frederic Adrian, MINWR, personal communication兲. The model has a provision for spotting from “running head fires” 共essentially forest crown fires that rarely develop in the sparse overstory on KSC/MINWR and CCAFS兲 but does not have a provision for spotting from understory fuels, such as the fuels in east central Florida. For this reason it was not necessary to specify crown fuel parameters for our model simulations. Spotting does occur however, most frequently caused by combusting cabbage palms 共Sabal pametto兲 sending embers aloft, causing spot fires a short distance 共5 to 25 m兲 away from the main fire. Under moderate meteorological conditions, hot embers rarely travel far enough to cross major paved roads, making these roads barriers to fire. To accurately model fire spread consistent with our observations during average meteorological conditions, we assigned narrow dirt roads to the slow burning Fuel Model 8. This compensated for FARSITE’s limited spotting capabilities in our study area. The fuel models were then converted to ASCII grid maps for input into FARSITE. FARSITE requires other landscape elements that are also covered in Table 1. The north and south portions of KSC/MINWR are dominated by pine flatwoods; therefore, the canopy coverage FARSITE uses for shading and wind reduction factors had to vary based on geographic location. The central region of KSC/MINWR and all of CCAFS are dominated by scrub and flatwoods vegetation communities with very little pine canopy, explaining the lower canopy cover values in Table 1. The elevation grid was given an elevation of 3 m for the entire study site, while the slope and aspect layers were assigned a value of zero degrees. This is because the topography of this barrier island complex is one of ridge-swale topography with only slight relief 共1 to 2 m兲. The subtle relief is
157 Table 2. Meteorological inputs for FARSITE fire simulations on Kennedy Space Center 共KSC兲 and Cape Canaveral Air Force Station 共CCAFS兲, Florida. These Data are for eight days in July of 1999 and represent typical July weather. Data were collected using the network of meteorological collection sites on KSC. MonthDayDaily Precip. 共in兲
Hour min. temp.
Hour max. temp.
Min. temp. 共°F兲
Max. temp. 共°F兲
Min. humid. 共%兲
Max. humid. 共%兲
Elev. of readings 共ft兲
07 07 07 07 07 07 07 07
0600 0600 0600 0600 0600 0600 0600 0600
1600 1600 1600 1600 1600 1600 1600 1600
70 76 76 75 75 77 71 76
90 94 92 92 93 95 93 91
51 46 53 51 50 43 53 60
99 95 92 93 92 91 94 96
0030 0030 0030 0030 0030 0030 0030 0030
20 21 22 23 24 25 26 27
1.03 0.00 0.00 0.00 0.00 0.50 1.42 0.00
Table 3. Initial fuel moistures for FARSITE simulations on Kennedy Space Center and Cape Canaveral Air Force Station, Florida. Fuel moistures are expressed in percent and may exceed 100. Fuel model
1 hour
10 hour
100 hour
Live herbaceous
Live Woody
1 2 7 8
10 10 10 10
12 12 12 13
16 14 14 16
90 90 85 85
100 100 110 115
unlikely to influence fire directly but does influence the distribution of vegetation types. The relief of just one meter makes the difference between a location having a xeric or mesic vegetation community. This variation is captured in the fuels map; therefore, the subtle topography of the area is factored into the model results. Wind data are required by FARSITE on an hourly basis with wind speed, wind direction, and cloud cover all being required by the model. Daily inputs for precipitation, minimum temperature, hour of minimum temperature, maximum temperature, hour of maximum temperature, minimum humidity, maximum humidity, and elevation of meteorological collection site are also required by FARSITE. All meteorological inputs 共Table 2兲 were gathered using KSC’s network of 21 meteorological sites. The model also required initial fuel moistures 共Table 3兲. Fuel moistures for each of these categories were generated by either averaging moisture values for summer samples of individual species for each category or by consultation with USFWS fire personnel at MINWR. We selected July 20th through 27th 1999 for our modeling window, because it represented typical meteorologic conditions during the summer when lightning strike probabilities were high. Because lightning initiates most fires during summer in this region, we
selected lightning as our ignition source. Ignition points were located in areas that currently contain large amounts of flammable native fuels and were somewhat central to each region. The output fire and fire enclaves 共unburned islands within burned areas兲 were exported directly to ArcView within FARSITE. Maps of each simulated fire were then generated and overlaid to produce composite fire maps of each region. Areas for each fuel type and extent burned were output from ARC/INFO by region and year for analysis. The software package FRAGSTATS was used to compare spatial configuration of fuels on each historic landscape 共McGarigal and Marks 1995兲. Statistical analyses were then conducted to test the relationship between percent burned and fuel fragmentation with null hypotheses of no relationship between amount burned and fuel fragmentation.
Results Anthropogenic features that directly remove flammable native fuels 共shown as non-fuel兲 and forest features increased throughout the study, fragmenting flatwoods and scrub fuels 共Figure 2兲. Linear anthropogenic features were few during 1920 but increased
Figure 2. The spatial distribution of fuel types by year on northern, central, southern Kennedy Space Center, and Cape Canaveral Air Force Station 共CCAFS兲. See Figure 1 for location.
158
159 Table 4. Area 共ha兲 of each fuel type by year and region on Kennedy Space Center 共KSC兲 and Cape Canaveral Air Force Station 共CCAFS兲, Florida. Percent change is calculated by subtracting 1990 area from 1920 area divided by 1920 area and then multiplied by 100. NA ⫽ not applicable. Fuel type
1920
1943
1990
% Change
North KSC 1 Short grass 2 Sparse trees/grass 7 Palmetto/shrubs 8 Forest/litter 98 Non-fuel
318 25 2,540 646 15,303
307 25 2,192 921 15,383
134 100 1,695 1,884 15,017
⫺ 58 ⫹300 ⫺ 33 ⫹192 ⫺2
Central KSC 1 Short grass 2 Sparse trees/grass 7 Palmetto/shrubs 8 Forest/litter 98 Non-fuel
2,470 161 6,644 336 14,237
2,404 162 6,278 1,234 13,772
2,344 305 4,516 3,268 13,418
⫺5 ⫹89 ⫺ 32 ⫹873 ⫹6
South KSC 1 Short grass 2 Sparse trees/grass 7 Palmetto/shrubs 8 Forest/litter 98 Non-fuel
2,937 142 7,902 408 9,142
2,856 142 7,447 714 9,371
1,788 571 4,493 3,545 10,133
⫺ 39 ⫹302 ⫺ 43 ⫹769 ⫹11
CCAFS 1 Short grass 2 Sparse trees/grass 7 Palmetto/shrubs 8 Forest/litter 98 Non-fuel
732 0 5,015 158 5,236
732 0 4,887 159 5,362
57 161 3,192 604 7,127
⫺ 92 NA ⫺ 36 ⫹282 ⫹36
in size and frequency in 1943 and 1990. Anthropogenic features have increased by 187% in the northern region of KSC, 3,397% in the central region, 1,166% in the southern region, and 2,129% on CCAFS between 1920 and 1990. The dynamics between fuel types show a transition from flammable marshes, flatwoods, and scrub to dramatic increases in less flammable forest types 共Table 4兲. For each region 共Figure 3兲, the largest simulated fire occurred in 1920, and fires became smaller for each successive modeling date 共Figure 4兲. Fire pattern changed with the reduction in fire extent; many of the 1943 and most of the 1990 fire boundaries coincided with anthropogenic features 共Figure 4兲. The largest decreases in fire size occurred between 1943 and 1990 in all regions except the southern region 共Table 5兲. In the northern region, 81% of the area burned in 1920 also burned in 1943, and 46% of that area burned in 1990. In the central region, 77% of the 1920 area burned in 1943, and 11% burned in 1990.
In the southern region, 40% of the 1920 area burned in 1943 and 16% burned in 1990. On Cape Canaveral Air Force Station, 79% of the 1920 area burned in 1943 with 8% burning in 1990. The linear regression between percent area burned and patch density was significant 共F ⫽ 21.05, R2 ⫽ 0.678, P ⬍ 0.001兲 共Figure 5兲, indicating a strong relationship between habitat fragmentation and percent area burned. For a one-unit increase in patch density our regression model states that there will be a corresponding 3.15 percent decrease in fire extent. The two leading landscape measures of fragmentation are patch density and mean patch size 共see Appendix for other measures兲 共McGarigal and Marks 1995兲. These are inversely related so that when mean patch size is large, patch density is low. In our study, there was a strong negative relationship between them 共r ⫽ ⫺ 0.848, P ⬍ 0.001兲, so we report patch density results only. Patch density is calculated as the number of patches in the landscape divided by total landscape area expressed as number per hundred hectares. There was a significant negative quadratic regression between percent burned and percent anthropogenic features 共F ⫽ 8.1, R2 ⫽ 0.643, P ⬍ 0.01兲 共Figure 6兲. The delta burn to anthropogenic feature ratio is another way to describe the relationship between anthropogenic features that have directly removed flammable fuels and fire extent 共Table 6兲. The delta burn to anthropogenic feature ratio 共DBAF兲 is defined by: DBAF ⫽
a⫺b c⫺d
where a is the 1990 burn area, b is the 1920 burn area, c is the 1990 anthropogenic feature area, and d is the 1920 anthropogenic feature area. DBAF is the average incremental difference between anthropogenic features that directly remove native flammable fuels and fire extent. This value represents the amount of fire reduction caused by one unit of anthropogenic feature development. The quadratic regression and these ratios both indicate that anthropogenic landcover had a disproportionate effect in reducing fire size on KSC. Past development does not necessarily predict the incremental effect of future development.
160
Figure 3. Overlaid fire simulation maps for A兲 northern B兲 central C兲 southern Kennedy Space Center 共KSC兲 and D兲 Cape Canaveral Air Force Station 共CCAFS兲. Simulated 1920 fire scars are on the bottom with 1943 fire scars in the middle and 1990 fire scars on top. This arrangement shows the spatial trend of simulated fire extends with time under average meteorological conditions on KSC and CCAFS. All simulated fire areas are mutual except where visible differences occur. That is the 1920 fire scars directly underlie 1943 boundaries except where visibly larger. The same applies for 1943 and 1990 simulation maps. Enclaves are islands within the burned fire extent and not all enclaves are visible due to overlays. Simulated lightning ignition locations are shown as black triangles. Fire simulations were performed using FARSITE fire modeling software.
161
Figure 4. Fire size distributions by time period and region on Kennedy Space Center 共KSC兲 and Cape Canaveral Air Force Station 共CCAFS兲, Florida. Areas derived by simulating fires in the FARSITE fire modeling software.
Figure 5. Regression plot showing the statistical relationship between the percent of landscape burned and fuel fragmentation 共patch density兲 on Kennedy Space Center and Cape Canaveral Air Force Station, Florida.
Discussion Flammability on the KSC/MINWR/CCAFS landscapes has been reduced markedly since the early 1900’s. Scrub and flatwoods communities 共the primary terrestrial fuel兲 in Florida are well adapted to fire and other natural disturbances 共Abrahamson 1984a,b; Myers 1990; Schmalzer and Hinkle 1992a,b兲. Historically, frequent lightning ignitions maintained natural scrub in a low, open structure and flatwoods as open savannas 共Robbins and Myers 1989兲. Vegetation patterns began to change as European settlement increased during the early 1900’s and brought open range management for ranching and
Figure 6. Quadratic regression plot showing the statistical relationship between percent of landscape burned and percent of landscape with anthropogenic features on Kennedy Space Center and Cape Canaveral Air Force Station, Florida. Anthropogenic features include buildings, industrial facilities, transportation, and agriculture.
clearing for citrus groves. The clearing for agriculture and supporting road network began fragmenting the landscape. Burning may have decreased when open range management formally ended in Brevard County in 1925 and in 1947 in Volusia County 共Davison and Bratton 1986兲. Fire was viewed as a threat to timber resources which lead timber industries to propose wildland fire suppression in the early 1940’s. Organized wildland fire suppression in Brevard County began in the 1950’s 共Larson 1952兲. Development of Cape Canaveral Air Force Station beginning in the early 1950’s and of Kennedy Space Center beginning in 1962 resulted in increased fire suppression. Fire suppression continued until 1981 on KSC and 1990 on CCAFS when prescribed burning was implemented to reduce dangerous fuel levels and restore habitat for native species 共Adrian and Farinetti 1995; Schmalzer et al. 1999兲. Facilities and infrastructure for the space program substantially increased landscape fragmentation. Flammability of the different vegetative communities has been affected in different ways by the change in landuse accompanying development of space launch capabilities. In the pine flatwoods communities, soil disturbance allowed dense pine forests to establish 共Duncan et al. 1999兲. The disturbance either directly or indirectly removed the native flammable understory that supported fire. This greatly reduced the flammability of many flatwoods areas under all but the most extreme fire weather. Fire suppression and hydrologic alteration have allowed mesic forest
162 Table 5. Fire size and potential fire size information for fire simulations on north, central, southern Kennedy Space Center 共KSC兲, and Cape Canaveral Air Force Station 共CCAFS兲, Florida. Fire areas were the actual extent burned, and enclaves were unburned islands within burned areas. Burnable area consisted of flammable fuel types excluding anthropogenic features, and percent burnable burned was the amount of burnable area actually burned in fire simulations. Anthropogenic features include buildings, industrial facilities, transportation, and agriculture. Simulations were performed using the FARSITE fire model. Fire area 共ha兲
Location/yr
Enclave area 共ha兲
Burnable area 共ha兲
% burnable burned
North KSC 1920 North KSC 1943 North KSC 1990
2,269 1,830 1,047
75 297 11
3,530 3,446 3,815
64 53 27
Central KSC 1920 Central KSC 1943 Central KSC 1990
6,429 4,949 744
1,839 1,734 238
9,613 10,078 10,433
67 49 7
South KSC 1920 South KSC 1943 South KSC 1990
7,926 3,217 1,264
1,996 1,017 375
11,389 11,160 10,399
70 29 12
CCAFS 1920 CCAFS 1943 CCAFS 1990
4,203 3,326 323
454 1,199 24
5,905 5,779 4,015
71 58 8
Table 6. The simulated contribution of anthropogenic features on fire extent reduction on Kennedy Space Center 共KSC兲 and Cape Canaveral Air Force Station 共CCAFS兲, Florida . The delta burn to anthropogenic feature ratio 共DBAF兲 is derived by dividing the difference between 1990 fire size and 1920 fire size 共delta burn 1990-1920兲 by the difference between 1990 and 1920 anthropogenic features. Anthropogenic features include buildings, industrial facilities, transportation, and agriculture. The ratio indicates that for every one unit of anthropogenic feature development there was a larger but varying amount of fire reduction. Region
Delta burn1990-1920 共ha兲
Delta anthropogenic 1990-1920共ha兲
DBAF
North KSC Central KSC South KSC CCAFS
⫺ 1,222 ⫺ 5,685 ⫺ 6,662 ⫺ 3,880
170 1,053 1,668 1,469
⫺ 7.2:1 ⫺ 5.4:1 ⫺ 3.9:1 ⫺ 2.6:1
species to spread into marshes where frequent fire previously hindered tree establishment 共Duncan et al. 1999兲. These forests replace the native flammable grasses and act as firebreaks because of the linear ridge-swale topography that run the entire length of many landscapes. Fire suppression has altered vegetation structure of long unburned scrub communities so they no longer burn readily except under the most extreme fire conditions 共Schmalzer et al. 1994; Schmalzer and Adrian 2001兲. Anthropogenic features such as roads and industrial areas have increased since 1920. These features, while not as dense and widespread as outside the federal property boundaries, still fragment remaining fuels. We found that a small amount of anthropogenic development significantly reduced fire extent in all regions of KSC/MINWR/CCAFS. As little as 10 percent anthropogenic landcover reduced fire extent by 50 percent from 1920 levels. The delta burn to
anthropogenic feature ratios suggest that there is a range of potential fire extent reduction that follows anthropogenic development. This range is between ⫺ 2.6:1 to ⫺ 7.2:1 for KSC/MINWR and CCAFS. This indicates that anthropogenic features have an inordinate effect on fire size, reducing fire size in excess of their actual area. The fire extent reduction observed between 1920 and 1943 was primarily driven by agricultural landuse changes and its infrastructure, while the 1943 to 1990 fire extent reduction can mostly be attributed to the increase in industrial landuse. Anthropogenic effects on Florida fire regimes are not limited to the alteration of fire extent but also affect other fire regime elements. Perhaps one of the most profound changes is in burn seasonality, as there has been a shift in peak acreage burned from the historic mid-summer to current spring maximum 共Florida Division of Forestry 2003兲. It would be interesting
163 to study the effect of seasonality and meteorology on spatial fire behavior to understand the ramifications of this seasonal shift. We focused, however, on the effects of anthropogenic influences on fire extent because the majority of fires 共frequency兲 on KSC/ MINWR are natural summer lightning ignitions 共F. Adrian, MINWR, personal communication兲, and KSC is a secured area so arson fires are not a major factor in its fire regime. This study was possible because historic fuels data were available and our study location was a controlled environment retaining summer lightning ignitions with a high ratio of native fuels to anthropogenic development. The results of the study would likely have been different if it was conducted outside of the Federal property boundaries where urbanization is predominant. We tried to make our results as realistic as possible; however, the strength of our study is in its relative results.
influences causing fragmentation of native fuels, altering fire extent and fire patterns. We found that anthropogenic features directly replacing flammable fuels have had a disproportionate effect on reducing fire size. This has profound implications for land managers unaware that relatively few anthropogenic features can greatly alter spatial fire behavior. Most modern landscapes are influenced by at least some anthropogenic features, and this influence is likely to increase in the future 共Gallagher and Carpenter 1997兲. This, coupled with the fact that fire is still one of the most important tools available to the land manager, implies that land managers will have to compensate better for the presence of anthropogenic features, if their goal is to maintain natural ecosystem function via prescribed fire.
Acknowledgements Conclusion This study used empirical historic and current fuel maps combined with spatial fire modeling techniques to perform a change detection of landscape flammability and fire extent from baseline historic conditions. We conclude that flammability of a pyrogenic system in the southeastern United States has decreased through time. Anthropogenic factors such as agricultural and industrial development have been critical
This study was conducted under NASA contract numbers NAS10-02001, NAS10-11624, and NAS1012180. We thank Kelly Gorman, Burton Summerfield, and Dr. W.M. Knott, III of NASA for their assistance and support. We would also like to thank Frederic W. Adrian at Merritt Island National Wildlife Refuge, Tammy E. Foster, David R. Breininger, William H. Romme and an anonymous reviewer for assistance with this manuscript.
Appendix Table A1. Select landscape indices of fuels on Kennedy Space Center 共KSC兲 and Cape Canaveral Air Force Station 共CCAFS兲. Indices were generated using the FRAGSTATS software and are useful for evaluating spatial composition and fragmentation of fuels. LANDSCAPE INDICES North KSC Total Area 共ha兲 Number of patches Patch Density 共#/100 ha兲 Mean Patch Size 共ha兲 Patch Size Standard Dev 共ha兲 Patch Size Coeff of Variation 共%兲 Total Edge 共m兲 Edge Density 共m/ha兲 Landscape Shape Index Central KSC Total Area 共ha兲 Number of patches Patch Density 共#/100 ha兲
1920
1943
1990
18833.2 983 5.21 19.1 296.5 1548.0 941163.3 49.9 20.83
18833.2 1014 5.38 18.5 291.1 1567.3 978480.7 51.9 21.60
18833.2 2539 13.48 7.4 179.7 2423.9 1638706.7 87.0 35.17
23851.1 1755 7.35
23851.1 1609 6.74
23851.1 5332 22.35
164 Table A1. Continued. LANDSCAPE INDICES
1920
1943
1990
Mean Patch Size 共ha兲 Patch Size Standard Dev 共ha兲 Patch Size Coeff of Variation 共%兲 Total Edge 共m兲 Edge Density 共m/ha兲 Landscape Shape Index
13.5 137.2 1009.9 1622600.8 68.0 31.06
14.8 123.4 832.5 1663927.6 69.7 31.82
4.4 57.7 1290.0 3420546.7 143.4 63.91
South KSC Total Area 共ha兲 Number of patches Patch Density 共#/100 ha兲 Mean Patch Size 共ha兲 Patch Size Standard Dev 共ha兲 Patch Size Coeff of Variation 共%兲 Total Edge 共m兲 Edge Density 共m/ha兲 Landscape Shape Index
20532.0 1693 8.24 12.1 125.2 1032.4 1492889.6 72.7 31.03
20532.0 1437 6.99 14.2 131.1 917.8 1529915.1 74.5 31.76
20532.0 4408 21.46 4.6 52.0 1118.3 2834220.2 138.0 57.44
CCAFS Total Area 共ha兲 Number of patches Patch Density 共#/100 ha兲 Mean Patch Size 共ha兲 Patch Size Standard Dev 共ha兲 Patch Size Coeff of Variation 共%兲 Total Edge 共m兲 Edge Density 共m/ha兲 Landscape Shape Index
11141.9 394 3.53 28.2 210.4 744.3 588367.8 52.8 17.15
11141.9 367 3.29 30.3 218.0 718.1 587675.4 52.7 17.13
11141.9 1409 12.64 7.9 68.9 872.2 1115518.2 100.1 31.24
References Abrahamson W.G. 1984a. Post-fire recovery of Florida Lake Wales Ridge vegetation. American Journal of Botany 71: 9–21. Abrahamson W.G. 1984b. Species responses to fire on the Florida Lake Wales Ridge. American Journal Botany 71: 35–43. Adrian F.W. and Farinetti R. 1995. Fire Management Plan Merritt Island National Wildlife Refuge. USFWS/MINWR, Titusville, Florida, USA. Agee J. 1993. Fire ecology of the Pacific Northwest forests. Island Press, Washington, USA. 493 p. Anderson H.E. 1982. Aids to determining fuel models for estimating fire behavior. United States Department of Agriculture Forest Service, General Technical Report INT-122. 22 p. Baker W.L. 1992. Effects of settlement and fire suppression on landscape structure. Ecology 73: 1879–1887. Baker W.L. 1999. Spatial simulation of the effects of human and natural disturbance regimes on landscape structure. In: Mladenoff D.J. and Baker W.L. 共eds兲, Spatial modeling of forest landscape change. pp. 277-308. Cambridge University Press, Cambridge, United Kingdom. 352 p. Christensen N.L. 1985. Shrubland fire regimes and their evolutionary consequences. In: Pickett S.T.A. and White P.S. 共eds兲, The ecology of natural disturbance and patch dynamics, pp. 85-100. Academic Press, Orlando, USA. 472 p.
Davis F.W. and Burrows D.A. 1994. Spatial simulation of fire in Mediterranean-climate landscapes. In: Moreno J.M. and Oechel W.C. 共eds兲, The role of fire in Mediterranean-type ecosystems, pp. 117-139. Springer-Verlag, New York, New York, USA. 201 p. Davison K.L. and Bratton S.P. 1986. The vegetation history of Canaveral National Seashore, Florida. CPSU Technical Report 22. U.S. National Park Service Cooperative Park Studies Unit, Institute of Ecology, University of Georgia, Athens, Georgia, USA. 75 p. Duncan B.W., Boyle S., Breininger D.R. and Schmalzer P.A. 1999. Coupling past management practice and historic landscape change on John F. Kennedy Space Center. Landscape Ecology 14: 291–309. Duncan B.W., Larson V.L. and Schmalzer P.A. 2000. Modeling historic landcover: an evaluation of two methodologies for producing baseline reference data. Natural Areas Journal 20: 308– 316. Duncan B.W., Larson V.L. and Schmalzer P.A. 2003. Historic landcover and recent landscape change in the north Indian River Lagoon watershed, Florida. Submitted to Natural Areas Journal. Environmental Systems Research Institute, Inc. 1997. ARC/INFO command reference and users guides 7.0. The geographic information systems software, Environmental Systems Research Institute, Inc., Redlands, California, USA. Fernandes P.A.M. 2001. Fire spread prediction in shrub fuels in Portugal. Forest Ecology and Management 144: 67–74.
165 Finney M.A. 1998. FARSITE: Fire area simulator-Model development and evaluation. United States Department of Agriculture Forest Service, Research paper RMRS-RP-4. 47p. Finney M.A. 1999. Mechanistic modeling of landscape fire patterns. In: Mladenoff D.J. and Baker W.L. 共eds兲, Spatial modeling of forest landscape change, pp. 186-209. Cambridge University Press, Cambridge, United Kingdom. 352 p. Florida Division of Forestry. 2003. http://flame.fl-dof.com/ Env/enso.html. Gallagher R. and Carpenter B. 1997. Human-dominated ecosystems. Science 277: 485. Keeley J.E., Fotheringham C.J. and Morias M. 1999. Reexamining fire suppression impacts on brushland fire regimes. Science 284: 1829–1832. Larson R.W. 1952. The timber supply situation in Florida. USDA Forest Service, Forest Resource Report No. 6. Washington, DC, USA. 60 p. Lunt I.D. 1998. Two hundred years of land use and vegetation change in a remnant coastal woodland in southern Australia. Australian Journal of Ecology 46: 629–647. McGarigal K. and Marks B.J. 1995. FRAGSTATS: Spatial pattern analysis program for quantifying landscape structure. United States Department of Agriculture Forest Service, General Technical Report PNW-GTR-351. 122 p. Miller C. and Urban D.L. 2000. Connectivity of forest fuels and surface fire regimes. Landscape Ecology 15: 145–154. Minnich R.A. 1983. Fire mosaics in southern California and northern Baja California. Science 219: 1287–1294. Mladenoff D.J. and Baker W.L. 共eds兲, 1999. Spatial modeling of forest landscape change. Cambridge University Press, Cambridge, United Kingdom. 352 p. Myers R.L. 1990. Scrub and high pine, pp. 150-193. In: Myers R.L. and Ewel J.J. 共eds兲, Ecosystems of Florida. University of Central Florida Press, Orlando, Florida, USA. 765 p.
Myers R.L. and White D.L. 1987. Landscape history and changes in sandhill vegetation in North-Central and South-Central Florida. Bulletin of the Torrey Botanical Club 114: 21–32. Ramos-Neto M.B. and Piello V.R. 2000. Lightning fires in a Brazilian Savanna National Park: Rethinking management strategies. Environmental Management 26: 675–684. Robbins L.E. and Myers R.L. 1989. Seasonal Effects of Prescribed Burning in Florida: A Review. The Nature Conservency, Tallahassee, Florida, USA. 119 p. Rothermel R.C. 1972. A mathematical model for predicting fire spread in wildland fuels. United States Department of Agriculture Forest Service Research Paper INT-115. Schmalzer P.A. and Adrian F.W. 2001. Scrub restoration on Kennedy Space Center/Merritt Island National Wildlife Refuge, 1992-2000, pp. 17-20. In: Zattau D. 共ed.兲, Proceedings of the Florida Scrub Symposium 2001. U.S. Fish and Wildlife Service. Jacksonville, Florida, USA. 63 pp. Schmalzer P.A., Boyle S.R. and Swain H.M. 1999. Scrub ecosystems of Brevard County, Florida: a regional characterization. Florida Scientist 62: 13–47. Schmalzer P.A. and Hinkle C.R. 1992a. Recovery of oak-saw palmetto scrub after fire. Castanea 57: 158–173. Schmalzer P.A. and Hinkle C.R. 1992b. Species composition and structure of oak-saw palmetto scrub vegetation. Castanea 57: 220–251. Schmalzer P.A., Breininger D.R., Adrian F.W., Schaub R. and Duncan B.W. 1994. Development and Implementation of a Scrub Habitat Compensation Plan for Kennedy Space Center. NASA Technical Memorandum 109202. John F. Kennedy Space Center, Florida, USA. 54 p. Turner M.G., Gardner R.H., Dale V.H. and O’Neill R.V. 1989. Predicting the spread of disturbance across heterogeneous landscapes. Oikos 55: 121–129.
