Ecosystem resistance in the face of climate change

Florida Everglades

Editor's Introduction

Ecosystem resistance in the face of climate change: a case study from the freshwater marshes of the Florida Everglades

Climate change is a growing concern. The Florida everglades are made up of several wetland ecosystems that have varying abilities to capture and store carbon to slow down the climate change process. However as CO2 dynamics change due to fluctuating levels of precipitation and air temperature, this natural filter may be in jeopardy. How can this be fixed? Scientists in this paper report the results of a simulation study to help prepare for these expected changes and discuss the future of climate change within the everglade ecosystem.

NOTE: This article is part of a Collection of student-annotated papers that are the product of the SitC team’s research into best practices for using primary literature to support STEM education. For this reason, these papers have undergone an alternate review process and may lack educator guides. To learn more, visit the main Collection page: SitC Lab.

Paper Details

Original title
Ecosystem resistance in the face of climate change: a case study from the freshwater marshes of the Florida Everglades
Original publication date
Reference
Volume 6, Issue 4 Pages 1–23
Issue name
Ecosphere
DOI
10.1890/ES14-00404.1
Topics

Abstract

Shaped by the hydrology of the Kissimmee-Okeechobee-Everglades watershed, the Florida Everglades is composed of a conglomerate of wetland ecosystems that have varying capacities to sequester and store carbon. Hydrology, which is a product of the region's precipitation and temperature patterns combined with water management policy, drives community composition and productivity. As shifts in both precipitation and air temperature are expected over the next 100 years as a consequence of climate change, CO2 dynamics in the greater Everglades are expected to change. To reduce uncertainties associated with climate change and to explore how projected changes in atmospheric CO2 concentration and climate can alter current CO2 exchange rates in Everglades freshwater marsh ecosystems, we simulated fluxes of carbon among the atmosphere, vegetation, and soil using the DAYCENT model. We explored the effects of low, moderate, and high scenarios for atmospheric CO2 (550, 850, and 950 ppm), mean annual air temperature (+1, +2.5, and +4.2°C) and precipitation (−2, +7, and +14%), as predicted by the IPCC for the year 2100 for the region, on CO2 exchange rates in short- and long-hydroperiod wetland ecosystems. Under 100 years of current climate and atmospheric CO2concentration, Everglades freshwater marsh ecosystems were estimated to be CO2-neutral. As atmospheric CO2 concentration increased and under climate change projections, there were slight shifts in the start and length of the wet season (−1 to +7 days) and a small enhancement in the sink capacity (by −169 to −573 g C m−2century−1) occurred at both short- and long- hydroperiod ecosystems compared to CO2 dynamics under the current climate regime. Over 100 years, rising temperatures increased net CO2 exchange rates (+1 to 13 g C m−2 century−1) and shifts in precipitation patterns altered cumulative net carbon uptake by +13 to −46 g C m−2century−1. While changes in ecosystem structure, species composition, and disturbance regimes were beyond the scope of this research, results do indicate that climate change will produce small changes in CO2 dynamics in Everglades freshwater marsh ecosystems and suggest that the hydrologic regime and oligotrophic conditions of Everglades freshwater marshes lowers the ecosystem sensitivity to climate change.

Report

The Florida Everglades is composed of a conglomerate of wetland ecosystems that have been shaped by the complex hydrology of south Florida (Davis and Ogden 1994). Hydrology drives wetland community composition, structure, productivity, soil development, and influences decomposition rates across these ecosystems (Davis and Ogden 1994Mitsch and Gosslink 2007). The mosaic of seasonally fluctuating hydrologic patterns has led to a range of capacities to sequester and store carbon (C), and may be important for ecosystem sensitivity. We refer to ecosystem sensitivity as the magnitude of a response to change, where low sensitivity indicates that a response to an environmental perturbation is small. Climate change projections suggest shifts in both air temperature and precipitation over the next 50 to 100 years (Christensen et al. 2007Stanton and Ackerman 2007IPCC 2013), and when coupled with water management decisions and human population expansion in south Florida, these shifts may have serious implications for Everglades hydrology and ultimately the region's C dynamics (Stanton and Ackerman 2007).

The Everglades is expected to be vulnerable to the impacts of climate change that, along with anthropogenic controls, will initiate additional alterations in water levels and inundation periods (Stanton and Ackerman 2007IPCC 2013). Future climate is dependent on atmospheric CO2 concentration, which is predicted to increase in the range of 540 to 970 ppm by 2100 (Houghton et al. 2001Stanton and Ackerman 2007). In combination, wet season precipitation is projected to decrease by 5–10% (Christensen et al. 2007) while annual precipitation may be altered by −2 to +14% (IPCC 2013) and temperatures may warm 1 to 4.2°C (IPCC 2013). The change will likely include larger convective storms and greater intensity hurricanes (Allan and Soden 2008). The uncertainty and confidence in climate change projections depends on the quantity, quality, consistency of evidence (e.g., mechanistic understanding, theory, data, and models) and the degree of agreement among models (IPCC 2013). Although climate change is expected to alter vegetative communities and C dynamics within the Everglades (Davis and Ogden 1994Bush et al. 1998Todd et al. 2010), the range in climate projections is wide, thus making estimates of future C dynamics for the region even less certain.

Anthropogenic activities have been shown to reduce the capacity of ecosystems to cope with disturbance and change (Jackson et al. 2001Scheffer et al. 2001Elmqvist et al. 2003), which could have significant implications for ecosystem sensitivity to climate change and ultimately the system's C dynamics and sequestration capacity (Cao and Woodward 1998Riedo et al. 2000). The greater Everglades system is highly modified by water control structures that disconnect hydrological dynamics from precipitation patterns throughout the Kissimmee-Okeechobee-Everglades watershed (Perry 2004). Efforts are currently underway to improve water flows throughout Everglades National Park in areas suffering from chronically low water levels (Perry 2004). Fluctuations in hydric conditions that alter ecosystem C storage or emission rates might occur slowly, but can have a significant long lasting effect on C pools. Using the ecosystem model DAYCENT (Del Grosso et al. 2001), we aim to explore the effects of increasing atmospheric CO2 concentration, temperature and altered precipitation, independently and in combination, to determine the relative impact of each on current ecosystem CO2 exchange rates.

Limited experimental capabilities exist to evaluate the complicated interactive controls on ecosystem responses to multifactor drivers (Fuhrer 2003Luo et al. 2008), although these effects are critical to understand how climate change impacts terrestrial ecosystems (Luo et al. 2008). Ultimately, time and financial constraints limit multifactor experiments (Luo et al. 2008), and thus simulation models have been a useful tool to investigate the effects of rising atmospheric CO2 concentration and climate change scenarios on terrestrial ecosystems (Abdalla et al. 2010). Ideally, models that simulate long-term changes in C dynamics should link plant, soil and atmospheric processes, and account for interactions among effects (Riedo et al. 2000). The DAYCENT model (Del Grosso et al. 2001) meets these requirements, simulating ecosystem water, C and nutrient dynamics (Parton et al. 1987Parton et al. 1988) for various native and managed ecosystems (Del Grosso et al. 2002Del Grosso et al. 2009Abdalla et al. 2010).

DAYCENT simulates changes in C and nutrient dynamics within and through the soil-plant-atmosphere continuum in response to changes in environmental conditions (i.e., air temperature, precipitation and atmospheric CO2 concentration) and management practices (grazing, harvesting, burning, fertilizing, and irrigation; Del Grosso et al. 2000). DAYCENT has been used to successfully simulate ecosystem responses to changes in climate (Parton et al. 1995Luo et al. 2008Savage et al. 2013), and to model gas fluxes (CO2, CH4, N2O, NOx, N2). It has also been used to model C and nutrient dynamics (N, P, S) in shrublands (Li et al. 2006), forest (Hartman et al. 2007Parton et al. 2010), crops (Del Grosso et al. 2002Stehfest et al. 2007Del Grosso et al. 2009Chang et al. 2013Duval et al. 2013), and temperate wetlands and grasslands (Luo et al. 2008Morgan et al. 2004Parton et al. 2007). By capturing patterns in soil hydrology, soil thermal regimes, and C dynamics (Chimner et al. 2002Cheng et al. 2013Cheng et al. 2014), DAYCENT has been used to effectively model C dynamics in wetland ecosystems, which provides a strong basis for using DAYCENT in Everglades freshwater marsh ecosystems.

Alterations to the Everglades hydrologic cycle are expected with climate change, leaving C pools and sequestration potential highly uncertain. The goal of these model simulations is to determine how CO2 exchange rates in Everglades freshwater wetland ecosystems will change as a result of rising atmospheric CO2 concentration and associated climate projections. The model projections are important for understanding Everglades vulnerability to climate change, and to indicate how factors interact to influence the CO2 sequestering capacity of Everglades ecosystems. Wetland ecosystem sensitivity to climate change will have a significant impact on the global C cycle, considering that 535 Gt C are stored in wetland soils as peat (Mitsch and Gosslink 2007) and this sequestered C is vulnerable to changes in hydrologic cycles.

Ecosystem sensitivity to “future” conditions can be used as an indication of the vulnerability of Everglades C pools. Increased atmospheric CO2 concentration should enhance ecosystem productivity (Nowak et al. 2004Ainsworth and Long 2005) by reducing nutrient limitations (Hocking and Meyer 1991Drake and Gonzàlez-Meler 1997Ainsworth and Long 2005) and photorespiration in C3 species (Bowes 1993). Holding atmospheric CO2 concentration constant, wetland ecosystem productivity and CO2exchange rates are driven largely by hydrology (Schedlbauer et al. 2010), nutrient dynamics, light and temperature (Schedlbauer et al. 2010). Precipitation drives hydrological patterns (e.g., water levels) and the onset and start of seasons (Malone et al. 2014b). Water levels above the soil surface influence soil temperature, oxygen availability and respiratory processes, often leading to a decline in ecosystem respiration. Productivity increases with higher water availability until water levels interfere with exposed leaf area and oxygen availability in the soil. Low ecosystem sensitivity would suggest that C pools are less vulnerable to climate change and the future structure and function of ecosystems would be more heavily influenced by changes in disturbance regimes and sea level rise.

Materials and Methods

Model description

To determine the effect of projected climate change on Everglades freshwater marshes we used DAYCENT_Photosyn (Savage et al. 2013), a modified version of the DAYCENT model (www.nrel.colostate.edu/projects/daycent/). DAYCENT_Photosyn models key processes for evaluating wetland ecosystem CO2 exchange rates by simulating soil water and thermal dynamics, plant production and allocation of net primary production (NPP), decomposition of litter and soil organic matter, mineralization/immobilization and plant uptake of nutrients, and CH4 oxidation and methanogenesis. DAYCENT_Photosyn contains the SIPNET photosynthesis model (Braswell et al. 2005Sacks et al. 20062007), which is a simplified Farquhar photosynthesis and respiration submodel. In DAYCENT, C allocation in the NPP submodel is a function of plant phenology, water and nutrient stress (Parton et al. 2010), soil C and nutrient dynamics (Parton et al. 2001), trace gas flux (Del Grosso et al. 2000), and soil water and temperature (Parton et al. 1998Eitzinger et al. 2000Fig. 1). The key process influencing soil C dynamics is soil organic matter decomposition, which is controlled by soil moisture/oxygen concentrations, temperature and pH. DAYCENT assumes that NPP and organic matter decomposition rates increase as soil water content increases until optimum water content is reached, with the optimum higher for NPP than for decomposition. Analogously, NPP and decomposition are influenced by temperature, and the sensitivity of the temperature response is different for each process. Optimum, minimum, and maximum temperatures for NPP vary with vegetation type whereas no minimum or maximum temperatures for decomposition are assumed, although the rate at which decomposition increases declines as temperature increases. Both NPP and decomposition are also limited by mineral N availability. As CO2 concentrations increase, DAYCENT allows water-use efficiency to increase by reducing transpiration, so that water savings are immediately reflected in soil moisture. Increases in NPP under rising atmospheric CO2 concentration using DAYCENT arise from the combined direct effect of CO2 on NPP, indirect effects of soil moisture on NPP, and reduced N-limitation to NPP due to enhanced soil-N mineralization on moister soils.

 

Fig. 1.

 

Fig. 1. Conceptual diagram of the DAYCENT_Photosyn ecosystem model. DAYCENT incorporates a soil organic matter (SOM), net primary productivity (NPP) and land surface sub-model. In the SOM sub-model, decomposition is microbially mediated with an associated microbial respiration CO2 loss that depends on soil texture. The NPP sub-model uses relationships between climatic factors and available soil nutrients to calculate plant production and allocation of nutrients to live aboveground and belowground vegetation. The land surface sub-model simulates water flow through the plant canopy, litter, and soil profile (Parton et al. 1998Eitzinger et al. 2000).

DAYCENT software

DAYCENT_Photosyn software helps to model processes that are key for evaluating wetland ecosystem. The processes include CO2 exchange rates by simulating soil water and thermal processes, plant production, decomposition of soil organic matter, plant uptake of nutrients, and much more.

Soil organic matter (SOM)

SOM consists of decaying plant and animal at various stages of the decomposition process (organic matter). It also contains cells and tissues of soil organisms and substances synthesized by soil organisms. In the SOM sub-model, bacteria help with decomposition while depending on CO2 for respiration. As CO2 concentrations increase, organic matter decomposition rates increase as soil water content increases (to an optimal water content level).

Net primary productivity (NPP)

Net primary production (NPP) is the rate at which all the plants in an ecosystem produce net useful chemical energy.  It is equal to the difference between the rate at which the plants in an ecosystem produce useful chemical energy (GPP) and the rate at which they use some of that energy during respiration, see the following equation:

NPP = GPP - respiration [by plants]

Land surface

The land surface model looks at interactions of the biogeophysics and biogeochemistry of the land-atmosphere.

 

DAYCENT parameterization

DAYCENT was parameterized for two subtropical Everglades ecosystems with contrasting hydroperiods, Taylor Slough (TS) and Shark River Slough (SRS). The short-hydroperiod marsh, TS, is inundated 4 to 6 months each year (June to November) and is characterized by shallow marl soils (~0.14 m; Duever et al. 1978) and relatively uniform vegetation co-dominated by a C3 sedge (Cladium jamaicense Crantz.) and a C4 grass (Muhlenbergia capillaris Lam.). The long-hydroperiod marsh, SRS, is inundated ~12 months each year and is characterized by peat soils (~1 m thick) with ridge and slough microtopography (Duever et al. 1978). Ridges are dominated by C. jamaicense and sloughs are dominated by Eliocharis cellulosa and Nymphaea odorata. Periphyton also exists on submerged structures at both sites and as floating mats at SRS. Both sites are also P limited, have year-round growing seasons and experience wet and dry seasons that are produced by precipitation patterns in the south Florida region (Davis and Ogden 1994). Taylor Slough occupies ~14,398 ha while Shark River Slough covers 88,811 ha of the 609,447 ha Everglades National Park. It is important to note that freshwater marsh ecosystems do not dominate the entire area of TS and SRS. For a more detailed site description, see Davis and Ogden (1994). Vegetation and soil data used in the model were measured at Florida Coastal Everglades Long Term Ecological Research (FCE-LTER) sites TS-1b (25°26′16.5″ N, 80°35′40.68″ W) and SRS-2 (25°33′6.72″ N, 80°46′57.36″ W) and made available through the FCE-LTER data portal (http://fcelter.fiu.edu/data/) (Table 1).

Table1

Table 1. DAYCENT site characteristics for Taylor Slough (TS) and Shark River Slough (SRS). Site data was obtained from the Florida coastal Everglades Long-term Ecological Research (FCE LTER sites TS-1 and SRS-2), AmeriFlux and the literature.

Taylor slough (TS)

TS marshes are known for their short-hydroperiod (period of time during which a wetland is covered by water).TS floods 4 to 6 months each year (June to November) and is characterized by shallow marl soils. Marl, sometimes found under desert sands and used as fertilizer for lime-deficient soils, is a crumbly mixture of clays, calcium and magnesium carbonates, and remnants of shells.

Shark river slough (SRS)

SRS marshes are known for their long-hydroperiod. SRS floods ~12 months each year and is characterized by peat soils which are formed from partially decomposed plant material under wet and oxygen poor conditions. Since carbon accumulate in peat it is seen as a useful resource in fighting climate change.

 

Long-term weather data were obtained from the nearest weather station, NCDC Royal Palm Ranger Station, where NOAA Daily Surface Meteorological Data were available from 1963 to 2011 (Fig. 2). Data from TS and SRS in 2012 was added to the weather file to aid model validation, creating a 50-year weather file for each site. In the Everglades, complex annual, seasonal, and intra-seasonal precipitation patterns exist, which are caused by global climate cycles (Davis and Ogden 1994). In order to maintain the natural variability in observed weather data and prevent unrealistic patterns in precipitation, weather data variability in climate change simulations (2000 to 2100) was based on observed weather data 1963 to 2012.

 

Fig. 2.

 

Fig. 2. Long-term daily weather data from the NCDC Royal Palm Ranger Station from 1963 to 2012. In climate change simulations weather data variability during 2000 to 2100 was based on variability at the Royal Palm weather station during 1963 to 2011 and weather data from TS and SRS in 2012.

 

Long-term data

Long-term weather data was used in helping to generate a projected weather profile for the everglades. This data was obtained from the nearest weather station, NCDC Royal Palm Ranger Station.

Projected data

In the Everglades, unpredicted and variable rain patterns exist which are caused by global climate cycles. In order to create a reliable weather profile while factoring in this natural variability in observed weather data and prevent unrealistic patterns in precipitation, climate change simulations (2000 to 2100) was based on observed weather data 1963 to 2012 (long-term data).

 

To parameterize DAYCENT, vegetation and water dynamics were simplified. The DAYCENT model requires a representative species that is used to characterize the vegetative components of the model. In terms of biomass and density, C. jamaicense dominates both TS and SRS (Davis and Ogden 1994), and thus C. jamaicense was used to characterize the vegetation parameters. Current site leaf area and P conditions were also used in the model and assumed constant.

To capture the effects of inundation in the DAYCENT model we used the program's irrigation option, and indicated when and for how long sites were inundated. Changes in soil water content (available pore space; Kelly et al. 2000) affect O2 levels, which influence heterotrophic decomposition in the DAYCENT model. Simulations showed that increasing soil water content reduces decomposition rates 10–30% of the maximum rates when soils are saturated. The irrigation option keeps soils saturated, which influences soil oxygen levels, but does not include the effects of standing water on exposed leaf area. To incorporate the effects of standing water on the productivity of emergent vegetation (Jimenez et al. 2012), we adjusted the optimum and maximum temperatures for production and altered the coefficients used to calculate water stress on vegetation production. We matched historical seasonal water levels at each site; at TS, the system was inundated throughout the wet season, while SRS remained inundated during the entire year. The start and duration of the wet season were determined from fluctuations in the Bowen Ratio (β) at TS and SRS (2009–2012; Malone et al. 2014a). The length of the wet season was determined assuming a linear model between cumulative precipitation (January through March) and wet season length derived from β (Fig. 3A) at TS and SRS (separately) using 2009–2012 data (http://ameriflux.ornl.gov). We also determined the linear relationship between the wet season length and the first day of the wet season (Fig. 3B; Malone et al. 2014b). These correlations were then used to incorporate the effect of surface flows on ecosystem productivity and soil water availability. This method was effective in that it allowed seasons to fluctuate with precipitation patterns, and permitted seasons to vary with climate change. However it does not consider the effects of water depth or any changes in water levels that occur as a result of water management activities, and assumes that relationships between precipitation and season length and durations that occurred over a relatively short time frame (2009–2012) were appropriate in the past and in the future under climate change.

 

Fig. 3.

Fig. 3. Wet season length and the start of the wet season at TS and SRS were estimated with data from 2009 to 2012. First, the linear relationship between (A) wet season length and precipitation (mm m−2) January to March was estimated, and then the relationship between (B) wet season length and the start of the wet season was determined.

Graph a

On the x-axis, we see the amount of rain from January-March. On the y-axis, we see the length of the wet season in days. As we can see from the graph, as the wet season progresses, there seems to be an increase in rain.

Graph b

In this graph we see the length of the wet season in days on the x-axis. On the y-axis we see the start of the wet season by days of the year. As we can see, the earlier the wet seasons begins in the year, the longer the length in days it will last.

 

Although both TS and SRS experience similar weather, variations in soil water availability and the effect of temperature and moisture on potential productivity resulted in dissimilarities between model parameterization at the two sites. Using observed ecosystem CO2 exchange rates to parameterize temperature and moisture effects on potential productivity allowed us to account for the presence of co-dominant and subdominant species that also contribute to ecosystem CO2 exchange. Gross ecosystem exchange (GEE) and net ecosystem exchange (NEE) rates for TS (TS-1b) and SRS (SRS-2) from 2009 to 2011 were used to determine maximum potential production, and the effects of temperature and moisture on production. Maximum potential production, which is a function of solar radiation, was greater in the dry season compared to the wet season and greater for TS compared to SRS. Potential production (g C m−2 month−1) is a function of the maximum production defined, and the scalars for the effects of temperature, moisture, and shading. In Everglades freshwater marsh ecosystems the effect of temperature on potential production differed by site and seasonally. The optimum temperature for production was higher at SRS than at TS and higher during the summer-wet season (35°C and ~37.5°C at TS and SRS, respectively) than in the winter-dry season (27°C and 30°C at TS and SRS, respectively). The coefficients used to determine the effect of water stress on production were similar for both TS and SRS, though the effect of water stress was slightly greater at TS during the wet season.

DAYCENT validation

To examine the model's ability to adequately characterize the study sites, we ran DAYCENT, parameterized for each site, with weather data collected from TS-1b and SRS-2 in 2012 and compared modeled versus observed soil volumetric water content (VWC), soil temperature (Tsoil, °C) and CO2 exchange rates (i.e., NEE, ecosystem respiration [Reco], and GEE). Using methods similar to Veldkamp and O'Brien (2000), observed soil volumetric water content (VWC, %) was calculated from site-specific equations for soil conditions and from the dielectric constant using two soil moisture sensors (CS616, Campbell Scientific Inc.) buried between 0 and 20 cm soil depth at each site. Observed soil temperature (Tsoil, °C) was measured at 5 cm depths at two locations within each site using insulated thermocouples (Type-T, Omega Engineering, Stamford, CT). Observed VWC, Tsoil and CO2exchange rates (NEE, Reco, and GEE) were obtained from the eddy covariance tower sites at TS (TS-1b) (Schedlbauer et al. 2010Schedlbauer et al. 2011Jimenez et al. 2012) and SRS (SRS-2) (Jimenez et al. 2012). All observed data is available through AmeriFlux (http://ameriflux.ornl.gov).

Measured and simulated outputs were evaluated using the coefficient of determination (R2) and bias. Bias was quantified via a linear regression of simulated versus measured values. Average bias is small when slopes are near 1 and intercepts are near 0, but a thorough examination of modeled values must be made to appropriately evaluate patterns in over- and under-estimation for the range of data.

Climate change simulations

These climate change simulations do not define the future of freshwater marsh ecosystems, but describe how the current system would respond to climate change projections using simplified versions of Everglades short- and long- hydroperiod freshwater marsh ecosystems. A 2000-year equilibrium simulation, under recent climate (380 ppm of CO2, and using long-term weather data) was conducted before model validation and climate projection simulations so that simulations would start at quasi-equilibrium (Pepper et al. 2005). This approach allowed us to attribute ecosystem responses wholly to climate change and thereby avoid any confounding response from a non-equilibrium state.

Following parameterization and validation, we ran the model for 100 years (1) under climate change projections, and (2) maintaining recent CO2 concentration (380 ppm) and observed air temperature and precipitation patterns (1963 to 2012; NOAA Daily Surface Meteorological Data) in the Everglades region, for comparison. To examine the effects of rising atmospheric CO2 concentration and climate change projections we simulated climate change projections by increasing atmospheric CO2 concentration and temperature and altering precipitation patterns individually and in combination for each of the two sites. The same weather data used for the present-day simulation were altered by applying additive scalars to adjust atmospheric CO2 concentration and temperature and scalar multipliers to adjust precipitation seasonally and annually to simulate climate change (Fig. 4). Atmospheric CO2 concentration, temperature, and precipitation were altered gradually while conserving daily (high frequency) variability (Parton et al. 1995McMurtrie et al. 2001). We explored a low, moderate, and high scenario for atmospheric CO2 concentration (550, 850, and 950 ppm, respectively: EPA and IPCC 2007) and each climate driver: mean annual air temperature (+1, +2.5, and +4.2°C; IPCC 2013) and precipitation (−2, +7, and +14%; IPCC 2013) (Fig. 4). We also incorporated the seasonal shifts in temperature and precipitation that are expected for this region (Christensen et al. 2007Stanton and Ackerman 2007). Projected temperature change was applied to the daily minimum temperatures in the dry season (winter months), while temperature increases were applied to maximum temperatures in the wet season (Fig. 4D, E) (Christensen et al. 2007). The annual distribution of precipitation was altered by reducing wet season precipitation by 10% (Fig. 4D–F). When applied together, atmospheric CO2concentration, temperature, and altered precipitation patterns approximate the IPCC 2100 climate change projections for the Everglades region. While climate projections also suggest increased hurricane intensity (IPCC 2013) and more frequent heat waves (Stanton and Ackerman 2007), including these potentially important changes was beyond the scope of this study. We examined the effect of elevated atmospheric CO2 concentration and altered weather by comparing cumulative CO2 exchange rates (NEE, Reco, and GEE) over 100 years (2001 to 2100) to cumulative rates under current conditions.

 

Fig. 4.

 

Fig. 4. Climate change scalars for (A) elevated atmospheric CO2 concentration, (B) air temperature, and (C) precipitation. The distributions for (D) minimum temperature, (E) maximum temperature and (F) precipitation were modified to match projected seasonal change.

 

Results

DAYCENT validation

DAYCENT simulated daily Tsoil and VWC values did not closely match that of the observed data, although monthly average DAYCENT values were comparable to observed average values at TS and SRS (Fig. 5). At both sites the average difference between daily observed and modeled Tsoil and VWC was less than 0.97°C and 0.007, respectively. Monthly average fluctuations in Tsoil were slightly underestimated at both TS (R2 = 0.98) and at SRS (R2 = 0.99), likely the result of increasing water levels at each site. The soil VWC was underestimated at TS once the site was inundated (~0.025% on average), and slightly over-estimated at SRS in the wet season (~0.03% on average). Discrepancies in VWC at TS were a result of differences between observed and modeled inundation. Determined by precipitation patterns, simulated inundation occurred before the site was actually inundated at TS, resulting in higher simulated VWC than observed (Fig. 5C). Once the site actually became inundated, model performance improved. Discrepancies in VWC and Tsoilsuggest that DAYCENT could be improved in wetland ecosystems by incorporating water depth above the soil surface and the effect of water depth on soil water content and temperature.

 

fig. 5.

 

Fig. 5. Observed versus modeled soil temperature at (A) TS and (B) SRS, and soil volumetric water content (VWC) at (C) TS and (D) SRS.

DAYCENT validation

A consideration when using modeling software is that the simulations depend on inputted data from the user. However it is possible that factors that were not initially considered relevant could play a role later on. Similarly, it may be difficult to capture the effects of all of the different factors that need to be considered and so the resulting models and simulations may not match observed data.

Soil temperature (Tsoil)

Soil temperature daily projections did not closely match that of the observed data, although monthly average DAYCENT values were comparable to observed average values at TS and SRS. The average difference between daily observed and modeled Tsoil  was less than 0.97°C. Monthly average fluctuations in Tsoil were slightly underestimated at both TS and at SRS likely the result of increasing water levels at each site.

Volumetric water content

The volumetric water content daily projections did not closely match that of the observed data, although monthly average DAYCENT values were comparable to observed average values at TS and SRS. The average difference between daily observed and modeled VWC  was less than 0.007. The soil VWC was underestimated at TS once the site was flooded (~0.025% on average), and slightly over-estimated at SRS in the wet season (~0.03% on average).

 

DAYCENT weakly captured fluctuations in daily CO2 exchange rates, though monthly estimates were very similar to those observed and previously reported for NEE, Reco and GEE at TS and SRS for 2008 to 2009 (Jimenez et al. 2012). Atmospheric convention is used for CO2 exchange rates, where a positive value denotes a loss of C from the ecosystem. At TS, DAYCENT captured fluctuations in NEE (R2 = 0.80) and GEE (R2 = 0.94), though daily fluctuations in Reco (R2 = 0.65) were often slightly over-estimated (Fig. 6A). At SRS, NEE was over estimated in the dry season and during transition periods as a result of small overestimations in both Reco (R2 = 0.68; Fig. 6B) and GEE (R2 = 0.70; Fig. 6C). On a monthly basis, DAYCENT simulated NEE, Reco and GEE were realistic at both sites and DAYCENT captured fluctuation in GEE and NEE much better than it did Reco.

 

Fig. 6.

 

Fig. 6. Observed (solid) versus modeled (hollow) CO2 exchange rates (NEE, Reco and GEE) at TS (A) and SRS (B). Atmospheric convention is used here and positive numbers indicate a loss of C to the atmosphere.

 

Under recent atmospheric CO2 concentration and climate, NEE fluctuated around a null balance with a range of ±100 g C m−2 yr−1 at both sites. Reco ranged from ~200 to 500 g C m−2 yr−1 at TS and SRS, whereas GEE ranged from −200 to −500 g C m−2 yr−1, though rates at TS were higher on average. Cumulatively at 100 years, both TS (−30 g C m−2 century−1) and SRS (−5 g C m−2 century−1) were very small sinks for CO2, and represent a potential accumulation of 0.33 tons C ha−1 century−1 at TS and 0.05 tons C ha−1 century−1at SRS. These results suggest freshwater marsh ecosystems are near neutral for CO2 loss versus gain over 100 years.

Main effects of climate change scenarios

Rising atmospheric concentration of CO2 resulted in the greatest change in NEE, Reco, and GEE at both TS and SRS (Table 2Figs. 7 and 8). At TS cumulative NEE declined substantially with elevated CO2 concentration compared to simulations under current climate (380 ppm). Long-term average annual NEE was −2.0, −4.8 and −5.6 g C m−2 yr−1 in the low, intermediate and high scenarios, respectively, compared to −0.30 g C m−2 yr−1 under current climates at TS. Cumulative Reco at TS increased progressively by 1755.2, 7217.8 and 8245.8 g C m−2 century−1 under the low, moderate and high scenarios, and similar to Reco, cumulative GEE at TS declined (higher CO2 uptake) relative to current conditions by approximately equal and opposite amounts.

 

Fig. 7.

 

Fig. 7. The effect of elevated atmospheric CO2 concentration (550 ppm, 850 ppm and 950 ppm) on cumulative (A) NEE, (B) Reco and (G) GEE, at TS. The influence of rising temperatures (1°C, 2.7°C and 4.2°C) on cumulative (B) NEE, (E) Reco and (H) GEE and shifts in seasonal and annual precipitation patterns (−2%, +7% and +14%) on cumulative (C) NEE, (F) Reco and (I) GEE. Atmospheric convention is used here and positive numbers indicate a loss of C to the atmosphere. All simulations were compared to current weather and atmospheric CO2concentration (red line).

 

Fig. 8.

 

Fig. 8. The effect of elevated atmospheric CO2 concentration (550 ppm, 850 ppm and 950 ppm) on cumulative (A) NEE, (D) Reco and (G) GEE, at SRS. The influence of rising temperatures (1°C, 2.7°C and 4.2°C) on cumulative (B) NEE, (E) Reco and (H) GEE and shifts in seasonal and annual precipitation patterns (−2%, +7% and +14%) on cumulative (C) NEE, (F) Reco and (I) GEE. Atmospheric convention is used here and positive numbers indicate a loss of C to the atmosphere. All simulations were compared to current weather and atmospheric CO2concentration (red line).

 

Table2

 

Table 2. The change in cumulative NEE, Reco, and GEE (g C m−2 century−1) as a result of rising atmospheric CO2 concentration, increasing mean annual temperature, and variations in annual precipitation individually and in combination (IPCC 2013 projects).

 

At SRS elevated atmospheric concentration also enhanced net CO2 uptake. At 550, 850 and 950 ppm of CO2, cumulative NEE decreased by 176.3, 502.1 and 573.3 g C m−2century−1, respectively, compared to cumulative NEE at 380 ppm of CO2 over 100 years (Table 2Fig. 8A). Under these scenarios, the long-term average annual NEE was −1.8, −5.0 and −5.7 g C m−2 yr−1 at low, medium, and high, respectively. Similar to TS, at SRS cumulative Reco increased under the low, medium, and high scenarios while GEE decreased by nearly equal and opposite amounts.

Changes in temperature alone resulted in very small changes in CO2 exchange rates of Everglades ecosystems by causing slight shifts in the ratio of GEE: Reco. At TS, a 1, 2.5 and 4.2°C increase in mean annual temperature led to increased NEE (lower CO2 uptake) by 1.3, 3.7 and 6.6 g C m−2 century−1 (Table 2Fig. 7B). Cumulative Reco decreased progressively by 2.5, 6.7 and 11.7 g C m−2 century−1 with increased temperatures, while GEE increased by 3.8 to 18.3 g C m−2 century−1. At SRS, NEE decreased by 3.1 to 13.0 g C m−2 century−1 as a mean annual temperature increased by 1, 2.5 and 4.2°C, respectively (Table 2Fig. 8A). Cumulative Reco decreased by 17.2 to 73.9 g C m−2 century−1, and GEE increased by 20.3 to 86.9 g C m−2 century−1 as a mean annual temperature increased at SRS.

Changes in precipitation of −2, +7, and +14% altered the length of seasons by −1, 4, and 7 days on average, respectively. A 2% decrease in annual precipitation, which was weighted to reduce wet season precipitation more than in the dry season, increased cumulative NEE by 13.0 g C m−2 century−1 at TS, while no change in cumulative NEE was found at SRS (Table 2). As mean annual precipitation increased by 7 and 14%, net stored CO2 increased by 20.3 and 41.8 g C m−2 century−1, respectively, at TS (Fig. 7C). A change in precipitation of −2, +7 and +14% at TS altered cumulative Reco by −1105.0, −220.1, and 432.9 g C m−2century−1, respectively. At TS, cumulative GEE was altered by −1105.0, −220.1, and 432.9 g C m−2 with a −2 and +7 and +14% change in mean annual precipitation. At SRS, changes in mean annual precipitation had a smaller impact on CO2 exchange rates. A change in mean annual precipitation of −2, 7 and 14% shifted NEE by 0, −21.03, and −46.1 g C m−2century−1, respectively. Reco at SRS increased by 5.4, 147.6, and 423.3 g C m−2 century−1, and GEE shifted by 5.4, −115.8, and −469.4 g C m−2 century−1 with a −2, +7 and +14% increase in annual precipitation, respectively.

Interactive effects of climate change scenarios

Simultaneous changes in CO2 concentration, temperature, and precipitation modified ecosystem CO2 exchange rates (Table 2Fig. 9). An increase in CO2 concentration (550 ppm), combined with a 1°C rise in annual temperature and 2% decrease in annual precipitation led to 156.8 g C m−2 century−1 increase in cumulative NEE, 2092.4 g C m−2century−1 increase in cumulative Reco and a 2249.2 g C m−2 century−1 decrease in GEE at TS. At SRS, cumulative NEE decreased by 160.5 g C m−2 century−1 (Table 2Fig. 9B), while cumulative Reco increased 2001.4 g C m−2 century−1 and GEE declined by 2162.0 g C m−2century−1. In the second scenario, atmospheric CO2 concentration rose to 850 ppm, mean annual temperatures increased by 2.5°C, and mean annual precipitation increased by 7%. At TS cumulative NEE, Reco, and GEE changed by −468.1, 7323.5, and −7323.5 g C m−2century−1, respectively. At SRS, cumulative NEE, Reco and GEE were enhanced by −509.7, 5666.8, and −6176.5 g C m−2 century−1, respectively. The most extreme scenario explored included an atmospheric CO2 concentration of 950 ppm, a 4.2°C increase in mean annual temperature, and a 14% increase in annual precipitation. At TS, this high scenario led to a −544.4, 9279.4, and −9279.4 g C m−2 century−1 change in cumulative NEE, GEE, and Recorespectively. At SRS, this high scenario led to a −557.7, 6776.4, and −7334.0 g C m−2century−1 change in cumulative NEE, GEE, and Reco respectively. Climate change projections led to a change in potential C sequestration of 1.7, 5.2 and 6.0 tons C ha−1century−1 at TS and 1.8, 5.6 and 6.1 tons C ha−1 century−1 with the low, intermediate and high scenarios, respectively.

 

Fig. 9.

 

Fig. 9. The effect of climate change projections on cumulative CO2 exchange rates: (A) NEE, (C) Reco and (E) GEE at TS, and on (B) NEE, (D) Reco and (F) GEE on at SRS. Atmospheric convention is used here and positive numbers indicate a loss of C to the atmosphere.

Discussion

Climate change is considered a major threat to species survival and ecosystem integrity (Hulme 2005Erwin 2009). Occurring within the transition zone between aquatic and terrestrial environments, wetlands ecosystems are considered to be among the most vulnerable ecosystems to climate change (Burkett and Kusler 2000) as a result of its effects on wetland hydrology and temperature (Ferrati et al. 2005Erwin 2009). Everglades ecosystems are projected to experience large shifts in ecosystem structure and function by 2100 (Stanton and Ackerman 2007). While projections for temperature and precipitation are within the natural range of conditions observed in the region, ecosystem CO2 exchange rates were modified by higher winter minimum and summer temperatures, and by greater dry season precipitation. At both sites, increasing atmospheric CO2concentration caused the greatest changes in GEE and Reco rates; however, because increases in these fluxes were of similar magnitude but opposite in sign, the effect on net CO2 exchange was very small. These results suggest that climate change will lessen seasonal differences in precipitation patterns by reducing wet and increasing dry season precipitation and results suggest that Everglades ecosystems are more resilient to change than previously thought. This low ecosystem sensitivity is likely the result of hydrology and nutrient limitations.

Main and interactive effects of multifactor climate change

As an important limiting factor for the growth and productivity of many species (Vu et al. 1997), terrestrial ecosystems are currently and have been responding to rising atmospheric CO2 concentration (Gifford 1980Ciais et al. 1995Keeling et al. 1995Drake and Gonzàlez-Meler 1997). The biochemical basis of this response is well established (Farquhar et al. 1980) and indicates that below 600 ppm atmospheric CO2 is generally limiting (Nowak et al. 2004), and many ecosystems will respond to higher concentrations. This was the case for TS and SRS and led to greater C uptake and release rates. Photosynthesis (Long and Drake 1992) and transpiration (Heath 1948Drake and Gonzàlez-Meler 1997) have long been known to respond to changes in atmospheric CO2concentration. Elevated CO2 concentration reduces photorespiration rates (Bowes 1993) in C3 species and enhances light (Drake and Gonzàlez-Meler 1997Ainsworth and Long 2005), nutrient (Hocking and Meyer 1991Drake and Gonzàlez-Meler 1997) and water use efficiency (Drake and Gonzàlez-Meler 1997) in plants. Ecosystem respiration rates also respond to rising atmospheric CO2 concentration (Gonzàlez-Meler et al. 1996Drake and Gonzàlez-Meler 1997) through the direct inhibition of respiratory enzyme activity (e.g., cytochrome c oxidase and succinate dehydrogenase, Drake and Gonzàlez-Meler 1997). Rising atmospheric CO2 concentration generally reduces dark respiration rates (Drake and Gonzàlez-Meler 1997) compared to GEE rates.

A small increase in net CO2 uptake at both Everglades sites occurred in response to elevated atmospheric CO2, suggesting that ecosystem responses at both TS and SRS will provide a negative feedback to global warming and will maintain current soil C pools. Compared to SRS, TS had greater CO2 uptake rates, although SRS was most sensitive to changes in CO2 concentration. Variations in site sensitivity were due to differences in leaf area. At SRS, leaf area is greater than at TS, and studies have shown that while rising CO2concentration alters productivity rates, leaf area often remains unchanged (Drake and Gonzàlez-Meler 1997). Although both sites fluctuate between being a very small sink, source, or near neutral annually (Jimenez et al. 2012, Malone et al. 2014b), rising CO2concentration may cause an enhancement in the sink potential of Everglades freshwater marsh ecosystems by improving resource use efficiencies, though the response to CO2enrichment will be limited by low P levels.

Climate change is expected to have the strongest and most immediate effect on plant phenology (Forrest and Miller-Rushing 2010) and physiology. Significant changes in physiology have been observed in response to higher temperatures, affecting photosynthetic rates of C3 plants (Long 1991), which is a dominant mode of photosynthesis at both sites. Higher temperatures lower the activation state of Rubisco (Kobza and Edwards 1987Holaday et al. 1992) and both the solubility and the specificity for CO2 relative to O2 (Jordan and Ogren 1984Brooks and Farquhar 1985Long 1991). At the ecosystem level, variations of only a few degrees centigrade are sufficient to affect gas fluxes (Hirano et al. 2009). Compared to CO2 concentration, changes in temperature had a much smaller impact on CO2 exchange rates at both TS and SRS and sensitivity to changes in temperature was similar at both sites. An increase in temperature, up until optimal conditions, often leads to an increase in metabolic activity (Medlyn et al. 2002Lambers et al. 2008), although when temperatures rise beyond optimal growth and activity ranges higher temperatures lead to reductions in productivity (Lambers et al. 2008). Higher temperatures shifted the ratio of GEE to Reco and led to greater C release at TS and SRS, though this effect was very small. In Everglades ecosystems, inundation may buffer the ecosystem response to higher temperatures.

Precipitation affects both productivity and ecosystem respiration rates by influencing water and oxygen availability, altering exposed leaf area through its effect on water depth, and enhancing N inputs through wet N deposition. As water levels affect ecosystem CO2exchange rates differently at TS and SRS (Malone et al. 2014b), so do changes in precipitation patterns. At TS a 2% decrease in precipitation led to higher net exchange rates via a smaller decrease in ecosystem respiration relative to GEE. SRS was insensitive to the 2% reduction in precipitation, and at both TS and SRS, precipitation increased cumulative GEE more than Reco, leading to greater net C uptakes rates. In the precipitation scenarios, wet season precipitation declined more than in the dry season, and as a result of greater sensitivity to changes in water levels the short hydroperiod site, TS, was more sensitive to changes in precipitation.

Climate change scenarios represent the range in climate that would result from changes in greenhouse gas concentration in the atmosphere. At both sites, NEE, Reco and GEE responded to changes in atmospheric CO2 concentration, temperature and precipitation, though the magnitudes of change were small, less than a 20% increase in average annual CO2 exchange rates. Interactive effects of atmospheric CO2 concentration and climate change on ecosystem CO2 exchange rates were greater than the sum of the individual effects and this enhancement was likely due to the interaction between rising CO2concentration and air temperature. CO2 enrichment modifies the response to temperature (Drake and Gonzàlez-Meler 1997) in C3 species, like the dominant species at both sites, sawgrass. Enrichment reduces photorespiration rates, which moderates the adverse effects of high temperature on C3 photosynthesis and results in greater net photosynthesis as growth temperatures increase (Long 1991Vu et al. 1997). The amount of Rubisco required also declines with increasing temperature (Drake and Gonzàlez-Meler 1997) and the degree of C3 photosynthesis enhancement by higher CO2 concentration is influenced by the temperature optimum for the species (Vu et al. 1997). Considerable evidence supports the prediction that CO2 uptake will be greater in warm climates (Long 1991McMurtrie and Wang 1993), though this may not occur in systems with low sensitivity to climate change.

Ecosystem sensitivity

Ecosystem sensitivity may be an essential factor underlying the sustained C sequestering capacities of Everglades ecosystems. Ecosystem sensitivity to climate change is driven by water, nutrients, and the responses of C3 species relative to C4 species. At Both TS and SRS, nutrient limitations should constrain the response to climate change but at TS where hydroperiods are short, changes in water cycling and the response of C4 co-dominant species could increase the ecosystems sensitivity to climate change. Both TS and SRS exhibit low sensitivity to changes in rising CO2 concentration and climate change scenarios. Previous studies have indicated that wet systems might be more resilient to change than dry systems (Luo et al. 2008). In the Everglades, hydroperiods may serve as a barrier to climate variation, which links the importance of hydroperiod to ecosystem sensitivity. Water above the soil surface dampens low temperature effects, cools the system when temperatures are high, and reduces the temperature response of soil and macrophypte respiration by slowing gas exchange. In addition to hydroperiods, low nutrient (P) levels may also affect ecosystem sensitivity to change (Steward and Ornes 1975Curtis and Wang 1998). The Everglades is an oligotrophic system (Craighead 1971), and nutrient limitation can reduce the capacity of ecosystems to respond to rising atmospheric CO2 concentration (Stitt and Krapp 1999). Interacting environmental stresses can influence the response to elevated CO2 in plants (Lloyd and Farquhar 1994Curtis 1996Curtis and Wang 1998) and environmental stresses tend to reduce the CO2response in C3 species (Wand et al. 1999). Studies have shown that although vegetation may initially respond to elevated CO2, acclimation is reported to be more pronounced when plants are N limited (Wong 1979Oberbauer et al. 1986Bowes 1993Curtis 1996Stitt and Krapp 1999Ward and Strain 1999Isopp et al. 2000). Studies have also shown that acclimation to elevated CO2 rarely has reduced photosynthetic capacity enough to completely compensate for stimulated photosynthetic rates (Drake and Gonzàlez-Meler 1997Ward and Strain 1999). Although in these simulations Everglades freshwater marsh ecosystems exhibit low sensitivity to changes in atmospheric CO2 concentration and climate change, shifts in disturbance regimes (e.g., fire, tropical storms, drought) are also projected for the region and will significantly influence the future condition of the Everglades.

Disturbance regimes, C3 vs C4 plants, and invasive species

To understand Everglades ecosystem vulnerability and resistance to change, future research must consider changes in ecosystem structure, disturbance regimes, C3 versus C4 plants, and invasive species risk in the Everglades region. Although not considered in these model simulations, the magnitude and frequency of disturbance events may significantly alter Everglades ecosystem structure and function (Stanton and Ackerman 2007). A rise in hurricane intensity, more frequent heat waves, salt-water intrusion and sea-level rise are all projected for the Everglades region (Stanton and Ackerman 2007). Disturbance and altered hydrologic regimes promote the displacement of native vegetation by introduced or formerly restricted species (Groves and Burdon 1986Mooney and Drake 1986), and have already led to changes in vegetation community composition in the Everglades (Toth 19871988Herndon et al. 1991Urban et al. 1993). Sawgrass communities in the northern Everglades have already been replaced by dense stands of cattail (Typha domingensis Pers.; Richardson and Marshall 1990Rutchey et al. 1994Jensen et al. 1995Newman et al. 1998), which invade disturbed and nutrient rich environments (Dykyjová and Kvĕt 1978Grace and Harrison 1986Keddy 1990). Marl prairie landscapes comprising the outer regions of Everglades National Park (ENP) have also shown vulnerability to woody plant expansion (Jenkins et al. 2003Knickerbocker et al. 2009) due to alterations in natural disturbance mechanisms (flood and fire management, Hanan and Ross 2010). Since trees have the greatest response to elevated CO2 (Ainsworth and Long 2005), woody encroachment in TS may increase.

Soil disturbance has also caused encroachment by exotics (Dalrymple et al. 1993) in the Everglades. Bahia grass (Paspalum notatum Flüggé) and torpedo grass (Panicum repensL.), exotic species introduced for cattle forage, spread quickly and aggressively. Under current conditions, these C4 grasses are more metabolically efficient than C3 species, allowing them to encroach on sawgrass marshes. Invasion may increase in the future if these species are sensitive to elevated atmospheric CO2 concentration. It has been suggested that elevated CO2 may preferentially increase the abundance of invasive species (Dukes and Mooney 1999Weltzin et al. 2003), which may have already played a stimulatory role in plant invasions (Ziska 2003) and shifts in dominance. Although some studies have shown C4 species respond weakly to elevated CO2 (Ainsworth and Long 2005), others have found many C4 plants to exhibit enhanced photosynthetic and growth responses (Sionit and Patterson 1984Ziska et al. 1990Imai and Sato 1991Wand et al. 1999). This suggests that C4 photosynthesis is not necessarily saturated at current CO2levels (Sionit and Patterson 1984Imai and Sato 1991). C4 grasses have shown significant changes in gas exchange, leaf area development (Wand et al. 1999), and an average growth enhancement of ~22% (Poorter and Oberbauer 1993) following elevated CO2concentration. Previous research has also shown that CO2 saturation levels in C4 species may be altered by environmental conditions (Wand et al. 1999). Ziska et al. (1990) found that elevated CO2 concentration stimulated CO2 assimilation rates in a cordgrass (Spartina patens Aiton) dominated salt marsh, though the effect was seasonal. Although enhanced C assimilation rates in C3 species has been shown to decline when stressed (Wand et al. 1999), C4 species show less negative impacts of environmental stresses (Wand et al. 1999).

Model limitations

Shifts in disturbance regimes and species dynamics are very important factors that will impact the sensitivity of Everglades ecosystems. Although the DAYCENT model was appropriate for exploring the multifactor effects of rising atmospheric CO2 concentration and climate change on the current ecosystems CO2 exchange rates, to better understand the future condition of Everglades freshwater marsh ecosystems changes in species cover need to be incorporated into the model. In addition, incorporating changes in P dynamics would be a significant improvement in our confidence in model results in these P limited systems. Modeling water table dynamics (Dimitrov et al. 2014), beyond their effects on NPP, soil C dynamics and season length, will also be a very important component to determining how climate change will influence Everglades ecosystems. Furthermore, climate induced alterations to disturbance regimes (i.e., salt water intrusion, sea level rise and the frequency of hurricanes and drought) and the incorporation of data uncertainties into DAYCENT for both data and the model output should also be considered. Comparisons with results from simulations with other ecosystem models (Luo and Reynolds 1999Sitch et al. 2003Gerten et al. 2004Krinner et al. 2005) will be necessary before drawing firm conclusions related to the question of how hydroperiods aid ecosystem resilience and what thresholds for air temperature and precipitation patterns significantly alter ecosystem sensitivity in the Florida Everglades.

Conclusion

Climate change projections have created substantial uncertainty about the future structure and function of Everglades ecosystems. As shifts in atmospheric CO2concentrations, air temperature, and precipitation are all expected over the next 100 years, exploring these effects is important for determining the current systems' vulnerability, which may be an indication of the potential ecosystem response to changes in sea level rise and disturbance regimes. Our results indicate that the hydrologic regime (timing, length of inundation and water depth) is important for ecosystem vulnerability, aiding Everglades ecosystems in light of climate change. The hydrologic regime may prove vital for reducing vulnerability to invasive species and changes in disturbance regimes as well, suggesting that restoration efforts in the Everglades region will be important for ecosystem resilience. Overall, a future with greater C sequestration seems likely for the Everglades region, so long as hydroperiods are maintained. As climate change reduces seasonal fluctuations in temperature and precipitation, elevated CO2 exchange rates may enhance the sink strength of Everglades ecosystems. Larger sink strength, albeit small, may be enough to switch these near CO2 neutral sites to small sinks that negatively feedback to global warming.

Acknowledgments

This research is based in part on support from the Department of Energy's (DOE) National Institute for Climate Change Research (NICCR) through grant 07-SC-NICCR-1059, the US Department of Education Graduate Assistantships in Areas of National Need (GAANN) program, and the National Science Foundation (NSF) Division of Atmospheric and Geospace Sciences (AGS), Atmospheric Chemistry program through grant 1233006. This research was also supported by the NSF through the Florida Coastal Everglades Long Term Ecological Research program under Cooperative Agreements DBI-0620409 and DEB-9910514 and by the United States Forest Service Rocky Mountain Research Station (RMRS). Any opinions, findings, conclusions or recommendations expressed in this material are those of the authors and do not necessarily reflect the views of the NSF, DOE, or RMRS. Finally, the authors would like to recognize all those that have advanced our predictive understanding of ecology, past, present and in the future.

REFERENCES 

  1. M. Abdalla, M. Jones, J. Yeluripati, P. Smith, J. Burke, et al., Atmospheric Environment 44, 2961– 2970 (2010).
  2. E.A. Ainsworth, S. P. Long, New Phytologist 165, 351–372 (2005).
  3. R.P.R. Allan, B.J.B. Soden, Science 321, 1481–1484 (2008).
  4. G. Bowes, Annual Review of Plant Physiology and Plant Molecular Biology 44, 309–332 (1993).
  5. B.H. Braswell, W.J. Sacks, E. Linder, D.S. Schimel, Global Change Biology 11, 335–355 (2005).
  6. A. Brooks, G.D. Farquhar, Planta 1653, 397–406 (1985).
  7. V. Burkett, J. Kusler, Journal of the American Water Resources Association 362, 313–320 (2000).
  8. D.E. Bush, W.F. Loftus, W.F. Bass, Longterm hydrologic effects on marsh plant community structure in the southern Everglades 18, 230–241 (1998).
  9. M. Cao, F.I. Woodward, Nature 393, 249–252 (1998).
  10. K.H. Chang, J. Warland, P. Voroney, P. Bartlett, C. Wagner-Riddle, Soil Science Society of America Journal 77, 941–950 (2013).
  11. D. Childers, T. Troxler, Sawgrass above ground biomass from the Taylor Slough, Everglades National Park (FCE), South Florida from August 1999 to Present. (2011a). 
  12. D. Childers, T. Troxler, Sawgrass above ground biomass from the Taylor Slough, just outside Everglades National Park (FCE), South Florida from October 1997 to December 2006. (2008).
  13. D. Childers, T. Troxler, Sawgrass above ground biomass from the Shark River Slough, Everglades National Park (FCE), South Florida from November 2000 to Present. (2011b).
  14. D. Childers, T. Troxler, Sawgrass Above and Below Ground Total Nitrogen and Total Carbon from the Taylor Slough, Everglades National Park (FCE), South Florida for March 2002 to Present. (2011c).
  15. D. Childers, T. Troxler, Sawgrass above and below ground total phosphorus from the Shark River Slough, Everglades National Park (FCE), from September 2002 to Present. (2013a).
  16. D. Childers, T. Troxler, Sawgrass above and below ground total nitrogen and total carbon from the Taylor Slough, Everglades National Park (FCE), South Florida for March 2002 to Present. (2011d).
  17. D. Childers, T. Troxler, Sawgrass above and below ground total nitrogen and total carbon from the Shark River Slough, Everglades National Park (FCE), from September 2002 to Present. (2013b).
  18. K. Cheng, S.M. Ogle, W.J. Parton, G. Pan, Ecological Modelling 262, 19–31 (2013).
  19. K. Cheng, S.M. Ogle, W.J. Parton, G. Pan, 2014. Global Change Biology 20, 948–962 (2014).
  20. R.A. Chimner, D.J. Cooper, W.J. Parton, Wetlands 22, 100–110 (2002).
  21. J.H. Christensen, et al., Regional climate projections. In: Climate Change 2007: The Physical Science Basis. Contribution of Working Group I to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change. (Cambridge University Press, Cambridge, UK, 2007).
  22. P. Ciais, P. Tans, M. Trolier, J. White, R. Francey, Science 269, 1098–1102 (1995).
  23. F.C. Craighead, The trees of South Florida. Volume I. The natural environments and their succession. (University of Miami Press, Coral Gables, Florida, USA, 1971).
  24. P.S. Curtis, Plant, Cell and Environment 19, 127–137 (1996).
  25. P.S. Curtis, X. Wang, Oecologia 113, 299–313 (1998).
  26. N.K. Dalrymple, G.H. Dalrymple, K.A. Fanning, Restoration Ecology 1, 220–225 (1993).
  27. S.M. Davis, J.C. Ogden, editors, Everglades: the ecosystem and its restoration. (St. Lucie Press, Boca Raton, Florida, USA, 1994).
  28. S.J. Del Grosso, W.J. Parton, A.R. Mosier, M.D. Hartman, L. Brenner, et al., Simulated interaction of carbon dynamics and nitrogen trace gas fluxes using the DAYCENT model. Pages 303–332 in M. E. A. Schaffer, editor. Modeling carbon and nitrogen dynamics for soil management. (CRC Press, Boca Raton, Florida, USA, 2001).
  29. S.J. Del Grosso, W.J. Parton, A.R. Mosier, D.S. Ojima, A.E. Kulmala, et al., Global Biogeochemical Cycles 14, 1045–1060 (2000).
  30. S. Del Grosso, D.S. Ojima, W.J. Parton, A. Mosier, G. Peterson, et al., Environmental Pollution 116 Supplement 1, 75–83 (2002).
  31. S.J. Del Grosso, D.S. Ojima, W.J. Parton, E. Stehfest, Global and Planetary Change 67, 44–50 (2009).
  32. D.D. Dimitrov, J S. Bhatti, R.F. Grant, Ecological Modelling 274, 57–70 (2014).
  33. B.G. Drake, M.A. Gonzalez-Meler, Annual Review of Plant Physiology and Plant Molecular Biology 48, 609–639 (1997).
  34. M.J. Duever, J.E. Carlson, L.A. Riopelle, L.C. Duever, Ecosystem analysis at Corkscrew Swamp, in Cypress Wetlands for water management, recycling, and conservation: Fourth annual Report to the National Science Foundation and to the Rockefeller Foundation. (University of Florida, Gainesville, Florida, USA, 1978).
  35. J.S. Dukes, H.A. Mooney, Trends in Ecology and Evolution 14, 135–139 (1999).
  36. B.D. Duval, K.J. Anderson-Teixeira, S.C. Davis, C. Keogh, S.P. Long, et al., PLoS ONE 8, e72019 (2013).
  37. D. Dykyjova, J. Kvet, Pond littoral ecosystems. (Springer-Verlag, New York, New York, USA, 1978).
  38. J. Eitzinger, W.J. Parton, M. Hartman, Soil Science 165, 525 (2000).
  39. T. Elmqvist, C. Folke, M. Nystrom, Frontiers in Ecology 1, 488–494 (2003).
  40. K.L. Erwin, Wetlands Ecology and Management 17, 71–84 (2009).
  41. G.D. Farquhar, S. von Caemmerer, J.A. Berry, Planta 149, 78–90 (1980).
  42. R. Ferrati, G.A. Canziani, D.R. Moreno, Ecological Modelling 186, 3–15 (2005).
  43. J. Forrest, A.J. Miller-Rushing, Philosophical Transactions of the Royal Society B 365, 3101–3112 (2010).
  44. J. Fuhrer, Agriculture, Ecosystems and Environment 97, 1–20 (2003).
  45. D. Gerten, S. Schaphoff, U. Haberlandt, W. Lucht, S. Sitch, Journal of Hydrology 286, 249–270 (2004).
  46. R. Gifford, Carbon storage by the biosphere. Pages 167–181 in Carbon dioxide and climate: Australian research. (Australian Academy of Science, Canberra, Australia, 1980).
  47. M.A. Gonzalez-Meler, M. Ribas-Carbo, J.N. Siedow, B.G. Drake, Plant Physiology 112, 1349–1355 (1996).
  48. J.B. Grace, J.S. Harrison, Canada Journal of Plant Science 66, 361–379 (1986).
  49. R.H. Groves, J.J. Burdon, editors, Ecology of biological invasions. (Cambridge University Press, Cambridge, UK, 1986).
  50. E.J. Hanan, M.S. Ross, Landscape Ecology 25, 463–476 (2010).
  51. M.D. Hartman, J.S. Baron, D.S. Ojima, Ecological Modelling 200, 493–510 (2007).
  52. O.V.S. Heath, Nature 161, 179–181 (1948).
  53. A. Herndon, L. Gunderson, J. Stenberg, Wetlands 11, 12–28 (1991).
  54. T. Hirano, J. Jauhiainen, T. Inoue, H. Takahashi, Ecosystems 12, 873–887 (2009).
  55. P.J. Hocking, C. P. Meyer, Journal of Plant Nutrition 14, 571–584 (1991).
  56. A.S. Holaday, W. Martindale, R. Alred, A.L. Brooks, R.C. Leegood, Plant Physiology 98, 1105–1114 (1992).
  57. J.T. Houghton, Y. Ding, D.J. Griggs, M. Noguer, P.J. van der Linden, et al., Climate change 2001: the scientific basis. (Cambridge University Press, Cambridge, UK, 2001).
  58. P.E. Hulme, Journal of Applied Ecology 42, 7840794 (2005).
  59. K. Imai, O.M. Sato, Japanese Journal of Crop Science 60, 139–145 (1991).
  60. IPCC, Climate Change 2007: The Physical Science Basis. (Cambridge University Press, Cambridge, UK, 2007).
  61. IPCC, Climate Change 2013: The Physical Science Basis. (Cambridge University Press, Cambridge, UK, 2013).
  62. H. Isopp, M. Frehner, S.P. Long, J. Nosberger, Plant, Cell and Environment 23, 597– 607 (2000).
  63. J.B.C. Jackson, et al., Science 293, 629–637 (2001).
  64. J.C. Jenkins, D.C. Chojnacky, L.S. Heath, R.A. Birdsey, Forest Science 49, 12–35 (2003).
  65. J.R. Jensen, K. Rutchey, M.S. Koch, S. Narumalani, Photogrammetric Engineering and Remote Sensing 61, 199–209 (1995).
  66. K.L. Jimenez, G. Starr, C.L. Staudhammer, J.L. Schedlbauer, H.W. Loescher, et al., Journal of Geophysical Research 117, G04009 (2012).
  67. D.B. Jordan, W.L. Ogren, Planta 161, 308–313 (1984).
  68. P.A. Keddy, Competitive hierarchies and centrifugal organization in plant communities. Pages 265–290 in J. B. Grace and D. Z. Tilman, editors. Perspectives on plant competition. (Academic Press, London, UK, 1990).
  69. C.D. Keeling, T.P. Whorf, M. Wahlen, J. van der Plichtt, Nature 375, 666–670 (1995).
  70. R.H. Kelly, W.J. Parton, M.D. Hartman, L.K. Stretch, D.S. Ojima, et al., Journal of Geophysical Research: Atmospheres (1984–2012) 105, 20093– 20100 (2000).
  71. C.M. Knickerbocker, et al., Natural Areas Journal 29, 15–26 (2009).
  72. J. Kobza, G.E. Edwards, Plant Physiology 83, 69–74 (1987).
  73. G. Krinner, N. Viovy, N. de Noblet Ducoudre, Global Biogeochemical Cycles 19, GB1015 (2005).
  74. H. Lambers, F.S. Chapin, T.L. Pons, Plant physiological ecology. (Springer, New York, New York, USA, 2008).
  75. X. Li, T. Meixner, J.O. Sickman, A.E. Miller, J.P. Schimel, et al., Biogeochemistry 77, 217–245 (2006).
  76. J. Lloyd, G.D. Farquhar, Oecologia 99, 201–215 (1994).
  77. S.P. Long, Plant, Cell and Environment 14, 729–739 (1991).
  78. S.P. Long, B.G. Drake, Pages 69–107 in N. R. Baker and H. Thomas, editors. Topics in photosynthesis. Volume 2: Crop Photosynthesis Spatial and Temporal Determinants. (Elsevier, Amsterdam, The Netherlands, 1992).
  79. Y. Luo, J.F. Reynolds, Ecology 80, 1568–1583 (1999).
  80. Y. Luo, et al., Global Change Biology 14, 1986–1999 (2008).
  81. S.L. Malone, et al., Journal of Geophysical Research: Biogeosciences 119, 1487–1505 (2014a).
  82. S.L. Malone, et al., PLoS ONE 9, e115058 (2014b).
  83. R.E. McMurtrie, B.E. Medlyn, R.C. Dewar, Tree Physiology 21, 831–839 (2001).
  84. R.E. McMurtrie, Y.P. Wang, Plant, Cell and Environment 16, 1–13 (1993).
  85. B.E. Medlyn, et al., Plant, Cell and Environment 25, 1167–1179 (2002).
  86. S.L. Miao, F.H. Sklar, Wetlands Ecology and Management 5, 245–264 (1998).
  87. W.J. Mitsch, J.G. Gosslink, Wetlands. Fourth edition. (John Wiley and Sons, Hoboken, New Jersey, USA, 2007).
  88. H.A. Mooney, J.A.Z. Drake, editors, Ecology of biological invasions of North America and Hawaii. (Springer-Verlag, New York, New York, USA, 1986).
  89. J.A. Morgan, et al., Oecologia 140, 11–25 (2004).
  90. S. Newman, J. Schuette, J.B. Grace, K. Rutchey, Aquatic Botany 60, 265– 280 (1998).
  91. S.F. Oberbauer, N. Sionit, S.J. Hastings, W.C. Oechel, Canadian Journal of Botany 64, 2993–2998 (1986).
  92. W.J. Parton, M. Hartman, D.S. Ojima, D.S. Schimel, Global and Planetary Change 19, 35–48 (1998).
  93. W.J. Parton, D.S. Schimel, C.V. Cole, Soil Science Society of America Journal 51, 1173–1179 (1987).
  94. W.J. Parton, J.W.B. Stewart, C.V. Cole, Biogeochemistry 5, 109–131 (1988).
  95. D.A. Pepper, S.J. Del Grosso, R.E. McMurtrie, W.J. Parton, Global Biogeochemical Cycles 19, GB1004 (2005).
  96. L. Poorter, S.F. Oberbauer, Oecologia 96, 193–199 (1993).
  97. J.R. Richardson,  A. Marshall, An evaluation of refuge habitats and relationship to water quality, quantity and hydroperiod. (A.R.M. Loxahatchee National Wildlife Refuge, Boynton Beach, Florida, USA, 1990).
  98. K. Rutchey, L. Vilchek, South Florida Water Management District, Everglades Systems Research Division, Photogrammetric Engineering and Remote Sensing 60, 767–775 (1994).
  99. W.J. Sacks, D.S. Schimel, R.K. Monson, B.H. Braswell, Global Change Biology 12, 240–259 (2006).
  100. J.L. Schedlbauer, S.F. Oberbauer, G. Starr, K.L. Jimenez, Agricultural and Forest Meteorology 150, 994–1006 (2010).
  101. M. Scheffer, S. Carpenter, J.A. Foley, C. Folke, B. Walker, Nature 413, 591–596 (2001).
  102. S. Sitch, et al., Global Change Biology 9, 161–185 (2003).
  103. E. Stehfest, M. Heistermann, J.A. Priess, D.S. Ojima, J. Alcamo, Ecological Modelling 209, 203–219 (2007).
  104. M. Stitt, A. Krapp, Plant, Cell & Environment 22, 583–621 (1999).
  105. L.A. Toth, Technical Publication 87-6. (South Florida Water Management District, West Palm Beach, Florida, USA, 1987).
  106. N.H. Urban, S.M. Davis, N.G. Aumen, Aquatic Botany 46, 203–223 (1993).
  107. J. Vu, L.H. Allen, K.J. Boote, G. Bowes, Plant, Cell and Environment 20, 68–76 (1997).
  108. J.K. Ward, B.R. Strain, Tree Physiology 19, 211–220 (1999).
  109. S.C. Wong, Oecologia 44, 68–74 (1979).
  110. L.H. Ziska, B.G. Drake, S. Chamberlain, Oecologia 83, 469–472 (1990).