Modeling the Effects of an Altered Fire Regime
on Vegetation Succession and Amphibian Habitat Associations
within the Okefenokee Swamp

Jon T. McCloskey and Cynthia S. Loftin
Department of Wildlife Ecology, University of Maine and
USGS-BRD Maine Cooperative Fish and Wildlife Research Unit

       Hamilton (1982) and Loftin (1998) have suggested that vegetation communities within the Okefenokee National Wildlife Refuge (ONWR) are progressing towards a fire intolerant, hardwood system due to historical logging practices and decades of wildfire suppression.  With changes in composition and distribution of vegetation assemblages, habitat quality for ONWR wetland-dependent wildlife, such as amphibians, is potentially altered.  The theory that landscape-scale vegetation change is resulting from historic wetland fire management and logging can only be tested through computer simulation, because the temporal and spatial scales of the variables are too great to be measured in a typical controlled experiment.  Furthermore, the nonlinear nature of the variables and the interactions among processes make such a dynamic system impossible to replicate without abstraction.

     Historical ecological research (e.g., Hamilton 1982, Loftin 1998) indicates that landscape vegetation patterns within ONWR develop primarily through the processes of fire, peat accumulation, and vegetation succession.  These processes are in turn affected by water level and hydroperiod (duration of inundation), which are driven primarily by climate. Other factors such as human activities, hurricanes, nutrient cycling, plant production, and peat accumulation also affect vegetation succession but will not be considered in this model.

     Landscape patterning processes in the ONWR operate at the mesoscale (i.e., areas of tens to hundreds of meters and time frequencies of decades to centuries). This model will project landscape change by decades over centuries, and will be based on a 14-class vegetation map developed using SPOT satellite data (10 m spatial resolution) collected during 2001.  The model will look at the relationship between vegetation structure (e.g., composition, patch size, patch connectivity, edge amount, and other measures of diversity) and environmental processes (i.e., hydrology and succession), and disturbances (i.e., fire and drought).  Our primary objective is to gain understanding of the two-way interaction between disturbance and vegetation structure (i.e., the effects of fire on landscape pattern and how landscape pattern affects the spread of fire), and how selected amphibian life-history types (represented by selected species) are potentially affected by the resulting change in ONWR vegetation distributions.

FIRE

     Fire behavior has three distinct components: ignition, spread, energy release. Only ignition and spread will be considered here (Figure 1). Lightning is the primary source of fires within ONWR and is most common during the summer.  Because the likelihood of fire ignition in ONWR varies seasonally, we have divided the year into three categories.  May-June indicates high likelihood of fire; July-November indicates moderate likelihood; and, December-April has a low likelihood due to the lack of thunderstorm-generated lightning.

     Fire propagation is influenced by fuel density, moisture content, atmospheric humidity, wind speed and direction, and soil moisture/water level (Rothermel 1983). Our model considers fuel moisture content, humidity, and soil moisture to be a function of water level and hydroperiod. Whether or not a plant within a vegetation class ignites and spreads is a function of many factors including species, age, physiological status at the time the fire occurs, plant competition, and hydrology. This model only considers tolerance and susceptibility of dominant species to fire, water level, and hydroperiod.  We are also analyzing historical wind data to determine the prevailing wind for each season to affect fire behavior.

     The most important factor in fire behavior is the amount of fuel and its moisture content (Rothermel 1983). These variables vary daily, seasonally, and between wet and dry years. In our model the amount of fuel will be a function of the time since last fire and the vegetation class (e.g., shrub communities have more fuel than herbaceous prairies).  Plant species that senesce during winter cause an increase in the amount of dead fuel. In addition, rare winter frosts can also increase the amount of dead fuel (Cypert 1973).

     Lower precipitation and increasing evapotranspiration due to leaf growth result in low water levels and litter moisture content in early spring (April-May). Therefore, optimal conditions for wildfire spread occur at the end of this dry season (May-June), when water levels are at their lowest, dead fuels are abundant and dry, and the first summer storms provide lightning as a source of ignition (Silveira 1996).  Fires at this time of the year propagate easily and may be moderate to severe. 

     Lightning is common during the wet season (June-November), and fires caused by lightning are often ignited. However, water levels are higher during this time, making fuels moist and less likely to burn extensively. Consequently, fires burn only a limited area around the strike location and are of low intensity. 

     During extended periods of drought, the peat surface itself may become dry enough to burn. In drought, the absence of standing water permits plants adapted to shorter hydroperiods to grow in the aquatic prairies (Silveira 1996).  Over time the dry fuel conditions coupled with increased fuel density create the potential for intense fires that spread quickly over large areas.

SUCCESSION

     The environmental conditions driving vegetation succession in ONWR are water level and hydroperiod.  Although we are focusing on the effects of fire on landscape pattern and how landscape pattern affects the spread of fire, hydrological conditions also influence ONWR vegetation dynamics and are the basis for the rules determining fire spread. In the absence of fire, lower water levels and shorter hydroperiods facilitate successsion from herbaceous to woody vegetation. Higher water levels and longer hydroperiods reset the progression; however, this "backwards" progression caused by extended hydroperiod will not be considered in this model.

     Vegetation itself influences water levels and hydroperiods by building peat. Peat accumulation moves succession forward by raising the peat surface elevation, reducing water levels and hydroperiod. However, peat formation depends on saturation to slow decomposition, so that fluctuations in water level and hydroperiod must be synchronized for the peat surface elevation to increase. In addition, as early successional species become more abundant, they are eventually replaced by shade-tolerant species. Thus, competitive interactions and peat formation will be abstractly considered in this model.

     Fire potentially reverses succession in ONWR.  Fire frequency and intensity determine the sequence of vegetation change in response to burning.  In turn, water level and hydroperiod influence this response, as species' tolerances to flooding vary.

     Herbaceous vegetation re-sprouts quickly after a burn of moderate intensity, but small woody seedlings with little energy stored in their root systems may not survive. Thus, succession of woody species may be set back with these fires. Without fire, peat will gradually accumulate and create conditions more favorable for flood-intolerant woody vegetation. Fires occurring later in the season when vegetation contains more moisture or when water levels are maintained by frequent precipitation are less likely to be severe or spread, unless drought conditions prevail. These fires are less likely to result in long-term vegetation change unless they occur with drought.

     Many of the wetland vegetation species in ONWR tolerate frequent, low-intensity fires. Most species re-sprout quickly following low intensity burns, unless the fire is followed by a long period of deep flooding.  Low-intensity fires maintain, rather than change, wetland vegetation patterns at the landscape level (Silveira 1996).

     Extremely intense fires can cause long-term changes in the vegetation pattern. The high heat penetrates into the soil, killing the roots of some plants. These fires can also burn into the peat and lower the surface elevation. The lower elevation increases hydroperiod following drought, leading to long-term shifts in composition and distribution of swamp vegetation assemblages.  Likewise, frequent fires may cause changes of longer duration.

AMPHIBIANS

     Recent studies have indicated that amphibian populations are declining worldwide at an alarming rate (Alford and Richards 1999).  Causes of these declines vary regionally and include ultraviolet radiation, predation, habitat loss, environmental toxins, disease, changes in climate, and interactions among these factors (Alford and Richards 1999).  There are approximately 38 species of amphibians found within ONWR and many travel long distances among various aquatic and terrestrial habitats and use different habitats at different stages of their life cycle (Alford and Richards 1999). If the altered fire regime within ONWR has drastically altered vegetation diversity and distributions, amphibian populations may be adversely effected.

     We will use multiple regression analysis to examine the relationships between landscape scale habitat variables (e.g., measures of composition, edge, patch, and diversity) and anuran abundance and richness. Anurans will be grouped into guilds based on preferred habitat (derived from the literature) during the breeding and non-breeding seasons. Amphibians with broad environmental tolerances should be less affected by changes in vegetation than those that have limited mobility or require specific land types for breeding.  A moderate number of models will be considered based on alternative hypotheses. We will use Akaike's Information Criterion to select the most parsimonious models. Anuran data were collected at ONWR during 2000-2001 by biologists at the USGS-BRD Florida Integrated Science Center.

SUCCESSION MODEL DATA AND SOFTWARE

Data used in development of the ONWR fire and vegetation succession model are from a variety of sources.  Fire distribution and occurrence data are from the ONWR fire management program and data complied by Loftin (1998).  Water level and weather data were complied from records at ONWR and from earlier studies by Loftin (1998).  The vegetation map was created during 2001-2002 from SPOT satellite imagery and digital aerial photography.  Vegetation species' responses to fire and hydrological conditions have been summarized from earlier studies by Loftin (1998).  We are using the Spatially Explicit Landscape Event Simulator (SELES) software for model development.

PROPOSED SCHEDULE FOR MODEL DEVELOPMENT

(JUNE 2003 – MAY 2005)

• Model Programming, Analysis, Report Preparation: April 2003 – May 2004
• Manuscript Preparation: May 2004 – May 2005

LITERATURE CITED

Alford, R. A., and S. J. Richards. 1999.  Global amphibian declines: a problem in applied ecology.  Annual Review of Ecology and Systematics 30:133-165.

Hamilton, D. B.  1984. Succession and the influence of disturbance in Okefenokee swamp. In The Okefenokee Swamp: its natural history, Geology, and Geochemistry.  A. J. Cohen, D. J. Casagrande, M. J. Andrejko, and G. R. Best, eds. Wetlands Surveys, Los Alamos, New Mexico, pp. 86-111.

Loftin, C. S.  1998. Assessing patterns and processes of landscape change in Okefenokee Swamp, Georgia. Ph.D. Dissertation, University of Florida, Gainesville, Florida, 864 pp.
 

Rothermel, R. C. 1983.  How to predict the spread and intensity of forest and range fires.  USDA Forest Service General Technical Report INT 143. USDA Forest Service Intermountain Rangeland Experiment Station, Ogden, UT.

Silveira, J. E.  1996. Landscape dynamics in the Everglades: Vegetation pattern and disturbance in water conservation area 1. Ph.D. Dissertation, University of Florida, Gainesville, Florida, 177 pp.

 

Figure 1.  Conceptual model of processes contributing to fire disturbance and vegetation succession dynamics within Okefenokee National Wildlife Refuge.