lable at ScienceDirect

Quaternary Science Reviews 90 (2014) 90e105

Contents lists avai

Quaternary Science Reviews

journal homepage: www.elsevier .com/locate/quascirev

Late Holocene vegetation, climate, and land-use impacts on carbondynamics in the Florida Everglades

Miriam C. Jones*, Christopher E. Bernhardt, Debra A. WillardU.S. Geological Survey, Eastern Geology and Paleoclimate Science Center, Reston, VA 20192, USA

a r t i c l e i n f o

Article history:Received 9 December 2013Accepted 7 February 2014Available online 19 March 2014

Keywords:PeatlandCarbonHoloceneEverglades

* Corresponding author. Tel.: þ1 703 648 6936.E-mail address: miriamjones@usgs.gov (M.C. Jone

http://dx.doi.org/10.1016/j.quascirev.2014.02.0100277-3791/Published by Elsevier Ltd.

a b s t r a c t

Tropical and subtropical peatlands are considered a significant carbon sink. The Florida Everglades in-cludes 6000-km2 of peat-accumulating wetland; however, detailed carbon dynamics from differentenvironments within the Everglades have not been extensively studied or compared. Here we presentcarbon accumulation rates from 13 cores and 4 different environments, including sawgrass ridges andsloughs, tree islands, and marl prairies, whose hydroperiods and vegetation communities differ. We findthat the lowest rates of C accumulation occur in sloughs in the southern Everglades. The highest rates arefound where hydroperiods are generally shorter, including near-tails of tree islands and drier ridges.Long-term average rates of 100 to >200 g C m�2 yr�1 are as high, and in some cases, higher than ratesrecorded from the tropics and 10e20 times higher than boreal averages. C accumulation rates wereimpacted by both the Medieval Climate Anomaly and the Little Ice Age, but the largest impacts to Caccumulation rates over the Holocene record have been the anthropogenic changes associated withexpansion of agriculture and construction of canals and levees to control movement of surface water.Water management practices in the 20th century have altered the natural hydroperiods and fire regimesof the Everglades. The Florida Everglades as a whole has acted as a significant carbon sink over the mid-to late-Holocene, but reduction of the spatial extent of the original wetland area, as well as the alterationof natural hydrology in the late 19th and 20th centuries, have significantly reduced the carbon sinkcapacity of this subtropical wetland.

Published by Elsevier Ltd.

1. Introduction

Global peatlands are the largest reservoir of soil carbon, coveringonly 3% of the Earth’s surface but storing roughly 450e500 Gt, orone-third of the terrestrial soil carbon (Gorham, 1991; Yu et al.,2010). The waterlogged nature of these wetland soils slowsdecomposition, which has allowed carbon to accumulate slowlyover thousands of years. Boreal and arctic peatlands cover thelargest land area and have been the most extensively studied (i.e.,Turunen et al., 2002; Gorham et al., 2003; Smith et al., 2004;MacDonald et al., 2006; Yu et al., 2009; Jones and Yu, 2010).Several studies have advanced understanding of long-term carbondynamics in tropical peatlands in Indonesia (Neuzil, 1997; Pageet al., 2004; Dommain et al., 2011; Page et al., 2011), Africa(Kivinen and Pakarinen, 1981; Joosten and Clarke, 2002; Page et al.,2011), and South America (Lähteenoja et al., 2009, 2013), whichcollectively comprise 11% of the global peatland area (Page et al.,

s).

2011). The Florida Everglades is an expansive low-latitude peataccumulating system, but information on the long-term carbondynamics, especially as it relates to climate, hydrology, and vege-tation, is sparse. Here we examine long-term rates of carbonaccumulation from four different types of wetland communitiesdistributed throughout the Everglades to understand the role ofvegetation and hydrology on carbon dynamics, and to assess howpast climatic perturbations and the more recent land-use changeshave impacted the ability of the Everglades to act as carbon sinks.

1.1. Regional setting

The Everglades ecosystem today occupies roughly 6000 km2 insouthern Florida, although the original extent was w12,000 km2

(Lodge, 2005). The wetland consists of a matrix of tree islands,mangrove swamps, cypress domes,marl prairies, sawgrassmarshes,sawgrass ridges, and sloughs. Pliocene and Pleistocene limestonebedrock underlies the Holocene peat and marls that have accumu-lated on the surface (Gleason and Stone, 1994). Peat began accu-mulating in the Everglades w7000 years ago in topographic low

M.C. Jones et al. / Quaternary Science Reviews 90 (2014) 90e105 91

spots (Gleason and Stone,1994), and hydrologic fluctuations relatedto global- to regional- scale changes in sea level and climate haveinfluenced vegetation patterns. Themid-Holocene expansion of theEverglades is coeval with the establishment of many peatlands andswamps on the southeastern Atlantic and Gulf Coastal plains andcan be attributed to the combined consequences of sea-level rise,orbitally driven changes in atmospheric circulation, and a stabili-zation of higher water tables (Willard and Bernhardt, 2011). Land-use change since the late 19th century, primarily through theinstallation of canals, levees, and otherwater-control structures, hasaltered the hydrology and impacted distribution of native plantcommunities as well as the frequency and severity of wildfires(McAvoyet al., 2011). Nearly half of the original Everglades land areahas been lost to agriculture and development that began in the late19th century, and the drainage of these areas has caused land sub-sidence of more than 1.5 m in some areas (Snyder and Davidson,1994). An effort is underway to restore the natural hydrologic pat-terns in the Everglades by removing many of the canals and levees(CERP, 2012). Understanding past long-term carbon dynamics ofthe Everglades is important to anticipating the consequences ofimplementing the Comprehensive Everglades Restoration Plan(CERP, 2012) on the Everglades carbon storage capacity.

The subtropical climate of the Florida Everglades is character-ized by hot, humid summers and mild winters. Much like tropicalenvironments, the wet/dry cycle is more important than winter/summer temperature in controlling Everglades vegetation patterns(Richardson, 2010). The majority (70%) of the rain falls during thewarmest portion of the year, from mid-May to November (average86 cm; range 58e135 cm) (Duever et al., 1994). This seasonality islargely responsible for Everglades peat accumulation, because 81%of the Everglades water budget was historically derived fromrainfall, with 8% from lake overflow and 10% from marginal over-flow, and only 1% from groundwater (Harvey and McCormick,2009). Recent changes to the hydrology have resulted in an in-crease to 33% of total inputs coming from surface-water inflow(Harvey and McCormick, 2009). This pattern of precipitation isdriven by themovement of the Bermuda High from its location overBermuda during thewet season to the Azores during the dry season(Stahle and Cleaveland, 1994). The El Niño Southern Oscillation(ENSO) can significantly impact rainfall patterns in South Florida. ElNiño results in greater than average rainfall, while La Niña increasesthe prevalence of droughts across South Florida (Abtew et al.,2006). The hydrology of the region also is impacted by the multi-decadal variability associated with Atlantic Multidecadal Oscilla-tion (AMO) (Enfield et al., 2001) and the North Atlantic Oscillation(NAO). Average temperatures are greater than 27 �C from April toOctober in the northern Everglades and March to November in thesouthern Everglades, while winter temperatures average above10 �C (Richardson, 2010).

Prior to impoundment, overland and sheet flow from LakeOkeechobee and the Kissimmee River created a mosaic of habitats,including sloughs, sawgrass ridges, tree islands, and marl prairie.Sloughs are dominated by Nymphaea (waterlily), Nymphoides(floating heart), and Nuphar (spatterdock) in relatively deep waterand>10months of inundation (Gunderson,1994), while ridges (.1e.2 m higher than sloughs) are dominated by Cladium jamaicense(sawgrass) and are only inundated 6e9months. Tree islands occupythe highest ground and are scattered throughout the ridge andslough landscape, comprising 14% of the Everglades today (Sklarand van der Valk, 2002). Tree islands are teardrop-shaped and ori-ented parallel to flow. Tree island heads are the most elevated andprovide refuge for subtropical hardwood trees and shrubs (e.g.,Bursera simaruba, Rivina humilis, Eugenia axillaris, Annona glabra,Chrysobalanus icaco, Persea borbonia, Ilex sp., Salix carolina, Myrsinefloridana, andMorella cerifera) and ferns (Acrostichum danaeifolium,

Blechnum serrulatum, Thelypteris kunthii), while the near-tails(downstream from the heads) experience longer hydroperiodsand the vegetation is dominated by water-tolerant hardwoods(Morella and Cephalanthus), ferns (Osmunda regalis), sedges, andother marsh taxa (Sagittaria and Pontederia) (Heisler et al., 2002).The strand islands of the Loxahatchee NationalWildlife Refuge havea hydroperiod up to 9 months and today are largely dominated bydahoon holly (Ilex cassine) (Lodge, 2010), although communitieshistorically have included ferns (Osmunda regalis, B. serrulatum), andshrubs and herbs (M. cerifera, Ilex, Amaranthaceae, Asteraceae)(Bernhardt et al., 2013).

Marl prairies have the shortest hydroperiods of all Evergladesecosystems, ranging from 2 to 9 months. The shorter hydroperiodand shallow water allow marl to precipitate, whereas the longerhydroperiods result in peat accumulation. Marl prairies have highspecies diversity despite being sparsely vegetated (Lodge, 2005).Those locations with 1e2 month hydroperiods are dominated bygrasses such as Schizachyrium rhizomatum, those with 3e5 monthhydroperiods are dominated by Muhlenbergia, and those with 6e8month hydroperiods are dominated by C. jamaicense (Olmsted andLoope, 1984; Davis et al., 2005). Periphyton assemblages aredominated by filamentous cyanobacteria such as Scytonema andSchizothrix (Davis et al., 2005).

2. Methods

2.1. Core collection and sampling

Analyses were performed on 13 cores that were collected from arange of wetland communities within the Everglades, includingnear tails of tree islands and at least one tree island head, ridges andsloughs, and the marl prairie from 1998 to 2008 (Fig. 1). A pistoncorerwith a 10-cmdiameter barrelwas used for collecting the cores.All coreswere taken tobedrock, except for the one strand island corefrom the Loxahatchee National Wildlife Refuge, 00-3-7-1, for whichbasal peat was not reached. For all cores, sediment lithology wasdescribed and sediment was sampled for pollen and microscopiccharcoal in 1-cm increments for the top 20 cm and 2-cm incrementsfrom 20 cm to the base of the core. All samples were oven dried at50 �C and the dried samples were stored at room temperature.

2.2. Geochronology

The chronologies of these cores are based on radiocarbon dates(14C), lead-210 (210Pb), and pollen biostratigraphy. Radiocarbondates were obtained from bulk sediments picked clean of roots byBeta Analytic and converted to calendar years using Calib 6.0 usingthe IntCal09 calibration curve (Reimer et al., 2011). Bulk sampleswere used for radiometric dating because identifiable plant macros,like seeds, could not be found, as the peat was very decomposed.Using bulk dates in such a setting does invite the potential that thepeat could incorporate older carbon and consequentially yield ananomalously older date. For most records, the chronology for thelast 100 years is based on the first occurrence of the pollen of theinvasive species Casuarina equisetifolia, which was introduced toSouth Florida at about AD 1900 (Langeland, 1990). Where possiblebased on the amount of available material, we also used 210Pb(lead-210) to constrain the last century of deposition. Lead-210accumulation rates were calculated making the assumption of aconstant initial excess 210Pb concentration (CIC model).

2.3. Pollen

Most pollen records associated with the cores analyzed forcarbon analysis have been previously published. The tree island

Fig. 1. Satellite image of the Florida Everglades, showing the locations of the cores analyzed in this study.

M.C. Jones et al. / Quaternary Science Reviews 90 (2014) 90e10592

cores are published in Willard et al. (2006) and Bernhardt (2011)and the ridge and slough pollen records are published inBernhardt et al. (2004), and Bernhardt (2011). The marl prairie(08-08-2-7A and 08-5-19-2) pollen records are presented here forthe first time. Palynomorphs were processed using standardprocedures (Willard et al., 2001; Traverse, 2007). Each .5e1.5 gsample of dry sediment was spiked with one tablet of exoticLycopodium spores to calculate palynomorph concentration(grains/g). Samples were processed with HCl and HF to removecarbonates and silicates, respectively, acetolyzed (1 part sulfuricacid: 9 parts acetic anhydride) in boiling water bath for 10 min,neutralized, and treated with 10% KOH for 10 min in a water bathat 70 �C. After neutralization, the residual samples were sievedwith 149 mm and 10 mmnylon mesh to remove the coarse and clayfractions, respectively. When necessary, samples were swirled ina watch glass to remove additional mineral matter. Samples werestained with Bismarck Brown and mounted on microscope slidesin glycerin jelly. At least 300 pollen grains and spores werecounted for each sample to determine percent abundance and

concentration. Spores, excluding bryophyte spores, are includedin the pollen sum. Pollen data are archived on the SOFIA (SouthFlorida Information Access http://sofia.usgs.gov) and NorthAmerican Pollen Database at the World Data Center for Paleocli-matology in Boulder, CO (http://www.ngdc.noaa.gov/paleo/pollen.html).

2.4. Charcoal analysis

Charcoal analysis was performed on ridge and slough cores (02-05-21-2 and 02-05-21-5) and marl prairie cores (08-08-2-7A and08-5-19-2). The metric of charcoal area per pollen grain (C/P) wasused to quantify changes in local fire regime (Pederson et al., 2005).This method has previously been published in the Everglades coresfrom tree islands (05-7-27-9, 05-7-27-10) (Bernhardt, 2011).Microscopic charcoal particles greater than 50 by 10 mm (500 mm2)were counted on slides prepared for pollen analysis. This sizefraction is considered to represent fire events within 20 km (Clark,1988; Pederson et al., 2005).

M.C. Jones et al. / Quaternary Science Reviews 90 (2014) 90e105 93

2.5. Carbon analysis

Analyses were performed on cores that had previously beensubsectioned into 1-cm or 2-cm intervals and dried at 50 �C. Thedried samples were subsampled to obtain volumetric measure-ments and weighed. Loss-on-ignition (LOI) analysis was performedat 550 �C using standard methods (Dean, 1974). To calculate carbondensity, the bulk density of the sample was multiplied by the LOI toobtain the ash-free bulk density. This value was multiplied by theassumed average amount of carbon in the organic fraction of thesediment, which in this case is best explained by the followingquadratic relationship:

Organic C ¼ð0:4� 0:01Þ*LOIþ ð0:0025� 0:0003Þ*LOI2;�r2 ¼ 0:990; n ¼ 250

�:

This relationship was determined from ten marshes in easternNorth Carolina, containing both mineral and organic soils (Craftet al., 1991). To determine carbon accumulation rates, first theradiometric dates obtained from each core and accretion rates weredetermined from the Bayesian age model program Bacon (Blaauwand Christen, 2011). These rates were multiplied by the carbondensity of each sample to obtain carbon accumulation rates(g C m�2 yr�1) for each core.

2.6. Statistical analysis

To measure statistical relationships between changes in carbonaccumulation rates and environmental changes in the cores, weused a one-way analysis of variance (ANOVA) between key pollentaxa and carbon accumulation. For the tree islands, this parameterwas the percent abundance of Osmunda regalis, which indicatestree island establishment, and for the marl prairie, this parameterwas the percent abundance of Cladium. In both cases, an increase inabundance represents the growth of ridges.

3. Results

3.1. Ridge and slough

Ridges are generally more effective at C accumulation thansloughs, though both seem to cycle C quickly, as rates are lowthroughout the cores until about 150e200 years ago when Caccumulation rates increase nearly exponentially (Fig. 2).

The cores 02-05-21-5 (s) and 02-05-21-2(r) are geographicallyfarther south and consequently wetter than 02-05-20-14 (r) and02-05-20-13 (s) and than 00-8-10-13 (r) and 00-8-10-14 (s). Theridge core 02-05-21-2 experiences increasing rates of carbonaccumulation (14.65 g C m�2 yr�1 from 3.37 g C m�2 yr�1) associ-ated with a transition to ridge (w500 cal yr BP, 24 cm) (Fig. 3), asexpressed by increasing Cladium and ferns (Bernhardt et al., 2004).The C accumulation rates slow in the upper 5 cm (w75e100 yearsago) of the core, in conjunction with a decrease in fern spores.Charcoal concentrations remain relatively low untilw200 cal yr BP,when they peak to nearly 300 mm2/grain, before declining again tothe present, similar in timing to the increase in C accumulationrates in the upper 5 cm of the core.

The adjacent slough core (02-05-20-13) experiences rates ofcarbon accumulation below or near 8 g C m�2 yr�1 until the upper5 cm, where C accumulation rates rapidly rise from 16 g C m�2 yr�1

to over 45 g C m�2 yr�1. The ridge core 02-05-20-14 experienceslow overall rates of C accumulation (w10 g C m�2 yr�1), which arelowest near the base of the record (w1 g C m�2 yr�1), but increasein the upper 5 cm to >200 g C m�2 yr�1. The pollen record shows

that the modern ridge was not in place until the upper 3e4 cm,where the average C accumulation rates are 168 g C m�2 yr�1

(Figs. 2 and 4). The charcoal peat in the slough core occurs earlierthan the ridge core w500 cal yr BP, before declining again towardpresent.

The sites farthest to the north and west, 00-8-10-13 (r) and 00-8-10-14 (s), experience average C accumulation rates an order ofmagnitude higher than the other four ridge and slough cores(Fig. 2), with C accumulation rates for the ridge averaging238 g C m�2 yr�1 and for the slough core 136 g C m�2 yr�1. In bothcores a dramatic increase in C accumulation occurs w1950, goingfrom an average of 26.6 g C m�2 y�1 pre-1950 to 403 g C m�2 yr�1

after 1950 for 00-8-10-13 and 43 g Cm�2 yr�1 pre-1950 and 323 g Cm�2 y�1 after AD 1950 for 00-8-10-14. In the slough core 00-8-10-14 C accumulation remains low in the slough stage of the core, but Caccumulation rates increase slightly when Amaranthaceae pollenincrease (Fig. 4). When the slough transitions to slough/sawgrassmarsh/edge of tree island, as indicated by the increase in fernspores, Morella, Amaranthaceae, Asteraceae, and Sagittaria(Bernhardt et al., 2004, Fig. 4) from AD 1950 to w AD 1980, Caccumulation rates increase sharply. Marl exclusively comprises thelithology of this section of the core. The pollen indicates a transitionback to slough vegetation in the upper 5 cm, which is associatedwith a decreasing rate of C accumulation and is accompanied by areturn to organic-rich silt and an increase in abundance of Nym-phaea pollen and an increase in Cladium (Fig. 4). For the associatedridge core, 00-8-10-13, the lowest rates of C accumulation areobserved in the slough phase of the ridge core, but the C accumu-lation increases after 1000 cal yr BP, similar to the increasesobserved in tree island cores 00-8-7-1 and 05-7-27-9. Anotherincrease occurs w100 years ago along with more abundantOsmunda regalis (Fig. 4), similar in timing to the tree island cores05-7-27-9, 98-4-23, and 00-3-7-1, as well as to the ridge and sloughcores to the south, when Nymphaea and Amaranthaceae increase inabundance (Bernhardt et al., 2004). A decrease in the carbonaccumulation rate occurs in the upper 5 cm (w1980s), when thesite returned to a slough or lower ridge environment, as Osmundaregalis decreases and Cladium increases. In the one ridge core 00-8-10-13, the results of a one-way ANOVA on the abundance ofOsmunda regalis spores and C accumulation rates indicates a strongand statistically significant relationship, where Osmuda regalis in-creases are also accompanied by increases in C accumulation rates(F ¼ 67.9, p < .0001). Both of these cores accumulated mass atroughly linear rates, but the ridge core 00-8-10-13 accumulatesmass faster in the lower 20 cm, while C mass accumulation occursmost rapidly in the upper 20 cm, coincident with the timing ofridge establishment (wAD 1950). The remaining cores showroughly linear rates of organic and dry mass accumulation thatincrease in slope in the last 50e100 years. Organic mass accumu-lation shows a slightly more convex pattern, suggesting an increasein accretion rates (Fig. 5).

3.2. Tree island

The near-tail core from Gumbo Limbo Hammock (00-8-7-1)experiences low carbon accumulation rates (w10e15 g C m�2 yr�1)from base of the core (3000 cal yr BP) to w1200 cal yr BP in anenvironment that is characterized as marsh, with abundantAmaranthaceae, Polygonaceae, Asteraceae, and Pinus pollen (Fig. 6).A marked increase in C accumulation rates (w30e85 g C m�2 yr�1)occurred after 1000 cal yr BP and continued to 80 cal yr BP inconjunction with a switch to a tree island environment, as exem-plified by a sharp increase in Osmunda regalis spores, a slight in-crease in other fern spores, and a decline of Amaranthaceae andPinus pollen. Another rapid increase in the C accumulation rates

Fig. 2. C accumulation rates for three ridge-slough core pairs in the Florida Everglades. Note that cores 00-8-10-13 (r) and 00-8-10-14 (s) are plotted on a log scale, while theremaining cores are plotted on a numeric scale. All cores show the low C accumulation rates occurring until the last 50e100 years when they increase markedly. The first increase inC accumulation of 00-8-10-13 occurs after 1000 cal yr BP, similar in timing to the increase associated with tree island establishment.

M.C. Jones et al. / Quaternary Science Reviews 90 (2014) 90e10594

(160e310 g C m�2 yr�1) occurs in the last several decades, inconjunction with a decline in Osmunda regalis and an increase inmonolete spores, Morella, and Cephalanthus pollen. Despite anoverall elevated C accumulation rate in the last 1000 years (relativeto 3000e1200 cal yr BP), a downward trend with lower rates occursfrom w700 to w80 cal yr BP, which generally coincides with thetiming of the Little Ice Age. A dip in the C accumulation rate atw600 cal yr BP also corresponds to a decrease in Osmunda regalisand an increase in Amaranthaceae (Fig. 6a).

The near-tail core from Skinners Island (98-4-23), exhibitsoverall high C accumulation rates (core average¼ 207 g Cm�2 yr�1),but the lowest occur near the base (w80 g C m�2 yr�1) from 700 to375 cal yr BP (Fig. 6). The transition to a tree island environment as

early as 350 cal yr BP corresponds to an increase in C accumulationrates that continued as the tree island became established afterw150 cal yr BP (Fig. 6b). Tree island establishment is indicated inpollen records by an increase in monolete spores and Osmundaregalis, corresponding with a steady increase in C accumulationrates to w420 g C m�2 yr�1 (average 207.5 g C m�2 yr�1). A steadydecline in C accumulation began after AD 1950, which also corre-sponds with a slight decrease in Osmunda regalis spores andAmaranthaceae, as well as a small increase in Cladium and Morella,suggesting a recent change in local hydrology.

Loxahatchee Strand Island (00-3-7-1) experienced high carbonaccumulation rates throughout the core (average 168 g Cm�2 yr�1),and the entirety of this core represents tree island vegetation

Fig. 3. Pollen diagram and carbon accumulation rates for ridge and slough pair 00-8-10-13 and 00-8-10-14.

M.C. Jones et al. / Quaternary Science Reviews 90 (2014) 90e105 95

(Fig. 7). Elevated rates (150e300 g C m�2 yr�1) from 1150 to1000 cal yr BP are associated with an increase in Osmunda regalisspores. The lowest rates (40e73 g C m�2 yr�1) occur in associationwith a decrease in Osmunda regalis (820e312 cal yr BP), while the

highest rates (w280e430 g C m�2 yr�1) occur 175e38 cal yr BP(40e13 cm), where the percentage of Osmunda regalis increases inabundance. Rates decline again in the last w80 years (w70e195 g C m�2 yr�1;12e0 cm), in association with an increase in Ilex,

Fig. 4. Pollen diagram and carbon accumulation rates plotted against C/P (mm2/grains) for the ridge and slough pair, 02-05-21-2 and 02-05-21-5, showing a charcoal peak coin-cident with drier conditions associated with the Little Ice Age. The charcoal peaks are followed by higher C accumulation rates.

M.C. Jones et al. / Quaternary Science Reviews 90 (2014) 90e10596

Morella, and Blechnum and an overall decrease of all ferns, butparticularly Osmunda regalis.

The tree island core from the Duck Club Hammock head (5-7-27-9) (supplementary info.) experienced moderate rates of C accumu-lation w20 g C m�2 yr�1 near the base, where sawgrass marsh andtree island vegetation prevailed, including ferns, Amaranthaceae,

Cyperaceae,Cladium, and Pinus. Themiddleof the core is interruptedby Native American midden deposits with no pollen preservationor carbon accumulation. C accumulation resumed 140 cal yr BP(w20 cm from the top) with relatively high average carbon accu-mulation rates (w65 g Cm�2 yr�1), increasing tow185 g Cm�2 yr�1.The near tail core from the same tree island (5-7-27-10) experiences

Fig. 5. Organic mass accumulation in a) ridges and sloughs, b) tree islands, and c) marl prairies. Organic mass accumulation for the ridge and slough cores 00-08-10-13 and 00-08-10-14 in panel (a) are plotted on the secondary y-axis on the right because of much higher accumulation rates.

M.C. Jones et al. / Quaternary Science Reviews 90 (2014) 90e105 97

low rates of carbon accumulation (4 g C m�2 yr�1) during the initialmarsh stage but begins to increase as the site transitions to a treeisland w650 cal yr BP as indicated by an increase in ferns, particu-larly Osmunda regalis.

The results of the one-way ANOVA from each core indicate thatOsmunda regalis abundance and carbon accumulation are stronglylinked (Table 2) and significant (p < .0001). This suggests that theassociated shifts in hydroperiod and vegetation are largelyresponsible for the system’s carbon sink capacity. Cumulative drymass accumulation occurs at roughly the same rate in all tree islandcores, but the mass accumulation rates increase with tree islanddevelopment. The change in slope with tree island development ismore pronounced in the organic mass accumulation curves, sug-gesting that most of the mass accumulation with tree islanddevelopment is organic. The flat nature of the organic mass accu-mulation curve for core 05-7-27-9 is attributable to a midden layer.The near-tail core from that same tree island, 05-7-27-10 iscomposed of peat throughout, but experiences the same low-accumulation pattern, and the pollen suggests it is still a marshenvironment for the period analyzed. Unfortunately, the upper-most portion of this core could not be analyzed for C due to a lack ofremaining sample material.

3.3. Marl prairie

The marl prairie core 08-08-2-7a found to the southeast ofTaylor Slough has accumulated w11.4 g C m�2 yr�1 over the last3300 years. The site exhibits higher C accumulation rates associatedwith Cladium, fern, and Asteraceae and Poaceae communities(Fig. 8a). The highest rates of carbon accumulation at this site occurfrom 3200 cal yr BP to 2800 cal yr BP (w55e37 cm;15.3 g C m�2 yr�1, up to 32 g Cm�2 yr�1) in associationwith slightlyelevated Morella, Asteraceae, Cladium, and fern. Low carbonaccumulation rates occur from 2800 to 2400 cal yr BP(w6.8 g Cm�2 yr�1) as very little wetland pollen is recorded. HigherC accumulation rates occur from 2300 to 1500 cal yr BP(w12.4 g C m�2 yr�1). The lowest rates of carbon accumulationoccur from 645 to 155 cal yr BP (2.1 g C m�2 yr�1), which coincideswith the timing of the Little Ice Age and also coincides with adecrease in Morella pollen. Carbon accumulation increases againaround AD 1950 to 11.5 g C m�2 yr�1, which coincides with an in-crease in Cladium, Asteraceae, Poaceae, Amaranthaceae, and ferns.Similar to core 08-5-19-2, charcoal concentrations begin to increasew600 cal yr BP, but the C/P remains relatively low (w20 mm2/grain)until w AD 1950 when concentrations increase to w140 mm/grain.

Fig. 6. Pollen diagram and carbon accumulation rates of tree island cores a) 98-4-23, showing a tree island transitionw150 cal yr BP associated with increased C accumulation ratesand b) 00-8-7-1, showing a transition from slough habitat to tree island, w1000 cal yr BP, coincident with an increase in C accumulation rates.

M.C. Jones et al. / Quaternary Science Reviews 90 (2014) 90e10598

Fig. 7. Pollen taxa and carbon accumulation rates for the Loxahatchee strand island core 00-3-7-1, showing increased rates of C accumulation associated with tree islanddevelopment and an abrupt decrease in the last w100 years.

M.C. Jones et al. / Quaternary Science Reviews 90 (2014) 90e105 99

Cumulative dry mass and organic mass accumulation both followlinear accumulation rates, except for a period from 1975 to780 cal yr BP (w30e20 cm), when the core transitions to peat withLOI 60e65% from 20 to 30%, before transitioning back to marl from20 cm to the top.

The marl prairie core 08-5-19-2 (Figs. 1 and 8b), located on theeastern edge of the RocklandMarl, has accumulated carbon at a rateof 15.5 g C m�2 yr�1 from w1650 years to 700 years. The sitetransitions from peat tomarlw700 cal yr BP causing a 75% decreasein C accumulation rates from an average of 24.1 g C m�2 yr�1 to5 g Cm�2 yr�1, which also roughly coincides with an increase in theabundance of charcoal in the record. This shift from peat to marl iscaused by a decrease in hydroperiod. The pollen record (Fig. 8b)shows a transition to slightly greater Amaranthaceae and Aster-aceae abundance, suggesting a drier, more disturbance-proneenvironment after w700 cal yr BP. This core accumulates bothdry mass and organic mass at a slight convex rate, owing to thefaster accumulation during the part of the core where peat accu-mulates and slower rate when the core transitions to marl.

Table 1Carbon accumulation rates (g C m�2 yr�1) calculated using the relationship identified by

Environment Core ID Long-term average Caccumulation rate(g C m�2 yr�1)

C accumulationsince 50 cal BP(g C m�2 yr�1)

Cb(g

Ridge andslough

00-8-10-14 (s) 136.61 302.6700-8-10-13 (r) 238.13 402.6302-05-20-14 (r) 18.97 72.7602-05-20-13 (s) 6.65 31.4102-05-21-5 (s) 8.89 24.9002-05-21-2 (r) 8.36 11.60

Marl Prairie 08-08-2-7A 11.45 11.7408-5-19-2 15.56 3.26

Tree Island 05-7-27-9 (head) 23.89 44.5205-7-27-10 (near tail) 1.8100-8-7-1 (near tail) 55.78 127.2198-4-23 (near tail) 207.45 248.77 1

Loxahatchee 00-3-7-1 (strand island) 168.04 145.25 1

In the marl prairie cores, the statistical relationship betweenpercent Cladium abundance and carbon accumulation rates wasvery strong. It was significant (F ¼ 67.09; p < .0001) in 08-08-2-7aand less strong but still significant (F ¼ 29.8; p < .0001) in 08-5-19-2, where peat deposition changes to marl deposition.

Both dry mass and organic mass accumulation display slightlyconvex patterns (Fig. 5), suggesting a slowdown of accumulation ofboth inorganic and organic sediments through time toward thepresent. In both cores, the most pronounced change in slope occursw1000 cal yr BP, coincident with the switch from peat to marl in08-05-19-2.

4. Discussion

4.1. Geographic setting and C accumulation

C accumulation rates in the Florida Everglades vary significantlydepending on local vegetation community and site-specific envi-ronmental history. The sites studied encompass the last 2000e4000

Craft et al. (1991).

accumulationefore 50 cal BPC m�2 yr�1)

LIA averageC accumulationrate (g C m�2 yr�1)

MWP averageC accumulationrate (g C m�2 yr�1)

Pre-MWP averageC accumulation rate(g C m�2 yr�1)

25.89 33.77 20.33 18.9526.63 28.39 7.69 7.973.90 6.22 2.23 1.032.19 2.72 2.71 1.035.81 9.48 2.88 3.01

11.91 14.15 3.99 2.7311.38 3.81 8.63 13.6916.98 3.53 24.64 19.9623.75 66.62 4.261.81 .85 1.69

28.69 44.97 25.87 7.7242.04 163.6675.90 225.84 166.89

Table 2Results of the one-way ANOVA, showing the covariance of specific pollen taxa, Osmunda regalis and Cladium, and carbon accumulation rates.

Core ID ANOVA Source DF Sum of squares Mean square F Significance

00-8-10-13 Osmunda vs C accumulation Total 44 6659851 151360.3A 1 4078456.3 4078456.3 67.94 <.0001Error 43 2581394.7 60032.4

00-8-7-1 Osmunda vs C accumulation Total 31 139137.38 4488.3A 1 57945.1 57945.2 21.41 <.0001Error 30 81192.2 2706.4

00-3-7-1 Osmunda vs C accumulation Total 37 1516377.6 40983.2A 1 847279.7 847279.7 45.59 <.0001Error 36 66908 18586.1

98-4-23 Osmunda vs C accumulation Total 53 2019357.7 38101.1A 1 883063 883063.1 40.41 <.0001Error 52 1136294.6 21851.8

08-08-2-7A Cladium vs C accumulation Total 61 9346.7 153.2A 1 4934.2 4934.2 67.09 <.0001Error 60 4412.5 73.5

08-5-19-2 Osmunda vs C accumulation Total 39 8108.2 207.9A 1 3565.3 3565.3 29.82 <.0001Error 38 4542.8218 119.6

08-5-19-2 Cladium vs C accumulation Total 39 8107.1 207.9A 1 3564.3 3564.3 29.82 <.0001Error 38 4542.8 119.6

M.C. Jones et al. / Quaternary Science Reviews 90 (2014) 90e105100

years of peat accumulation and long-term average C accumulationrates range from 8.4 g C m�2 yr�1 (ridge and slough landscape) toover 200 g C m�2 yr�1 in specific ridges and some near tail settingsin tree islands. These high rates are up to two to four times higherthan long-term C accumulation rates calculated from peat cores inIndonesia (Neuzil,1997; Page et al., 2004; Dommain et al., 2011) and10 to 20 times higher than average northern high latitude rates(Gorham et al., 2003; Yu et al., 2009). The rates fall within the ratesfrom mangrove swamps and saltwater marshes, which were notstudied here, but accumulate on averagew80 g Cm�2 yr�1 (Chmuraet al., 2003). This lies in contrast to the study by Glaser et al. (2012)who found that the long-term rate of carbon accumulation from asingle core from Shark River slough was only 12.1 g C m�2 y�1,leading the authors to conclude that the Florida Everglades accu-mulate carbon at a much lower rate than boreal or tropical peat-lands. The rate of 12.1 g C m�2 yr�1 is, however, similar to the long-term average rates we calculated in the southernmost ridge andslough cores from this study (w8.4e8.9 g C m�2 yr�1 Table 1). TheFlorida Everglades exhibit broad hydrological, ecological, andmicro-scale heterogeneity, leading to great heterogeneity in Caccumulation. Overall, our data show that sloughs accumulated theleast amount of carbon, although the ridges and sloughs to thenorth accumulate farmore carbon. The high rates of C accumulationin the northernmost sites are in agreement with previous obser-vations of high peat accumulation and greater peat thicknessesoccurring in the northern Everglades (Gleason and Stone, 1994;Richardson and Huvane, 2008; Richardson, 2010).

4.2. Wetland community type and C accumulation

C accumulation in the Everglades is sensitive to vegetationcommunity, which is driven by hydroperiod and nutrient condi-tions. The largest shifts in past C accumulation rates are associatedwith changes in hydroperiod and associated shifts in vegetation. Inthe peat-accumulating systems of the Everglades, a shift to ashorter hydroperiod and the associated shift in vegetation, perhapscounter-intuitively, results in higher carbon accumulation rates,although if a site becomes too dry, C accumulation slows, likelybecause of greater peat oxidation (Table 3). The onset of tree islandestablishment resulted in a doubling to 10e11� higher rate ofcarbon accumulation than themarsh environment that preceded it.The shorter hydroperiod promotes growth of the more productive

and higher biomass producing Cladium and Osmunda-dominatedenvironment. The shorter hydroperiod also allows for Cladium togrow more robustly, which has been found in previous studies.Inundation results in stomatal closure, decreasing photosynthesis,because these plants have inefficient aerenchyma (Chabbi et al.,2000; Schedlbauer et al., 2010). Consequently, taller ridges showCladium plants growing twice as high as shorter ridges (Givnishet al., 2008). In general, the near-tail environment of the treeislands exhibits the highest rates of carbon accumulation. High Caccumulation rates following tree-island development is closelylinked with an increase in abundance in Osmuda regalis, which is amarker for tree island initiation and expansion (Willard et al., 2001,2006). These results suggest that the vegetation and hydroperiodassociated with tree island growth and/or expansion is an optimalsetting for peat accumulation in the Florida Everglades.

The ridge site with anomalously high C accumulation rates (00-8-10-13) to the north also experiences a sharp increase in abun-dance of Osmunda regalis (Fig. 4), suggesting an interval with aslightly shorter hydroperiod, before reverting to a Cladium-domi-nated ridge, accompanied by decreasing, though still high, Caccumulation rates. Of the tree islands studied, the low rates of peataccumulation at depth coincide with slough or marsh vegetationcommunities, consistent with the low rates observed in themodernsloughs.

4.3. Climate and C accumulation

From the cores analyzed, as well as other paleoecological studiesof tree islands (Willard et al., 2006; Bernhardt, 2011), tree islandestablishment appears to be closely linked with drier conditionsassociated with the onset of the Medieval Climate Anomaly (MCA)(1000e700 cal yr BP) when the Intertropical Convergence Zone(ITCZ) migrated southward (Bernhardt, 2011). The long-termchange in hydrology and vegetation associated with a drier MCAclimate can therefore be indirectly lead ascribed to an increasedcapacity of the Everglades to store carbon as fixed tree islandsexpanded. The Loxahatchee strand island core did not capture itstransition from marsh to tree island (Fig. 7), but the bottom of thecore has a C accumulation rate of 167 g C m�2 yr�1, much higherrates than other tree islands developing at this time, from 1100to w850 cal yr BP. A decrease in C accumulation rates(w56 g C m�2 yr�1) occurs with a decrease in Osmunda regalis and

Fig. 8. a.) Pollen diagram, carbon accumulation rates, and charcoal for marl prairie cores a.) 08-08-2-7a and b.) 08-05-19-2. CONISS analysis in both cores is based solely on pollenand spores.

M.C. Jones et al. / Quaternary Science Reviews 90 (2014) 90e105 101

an increase in monolete spores and Amaranthaceae. This shiftsuggests increasingly dry conditions that deleteriously impacted Caccumulation, likely the combination of a shift in vegetation com-munity, lower primary productivity, and greater peat oxidationassociated with a sustained lower hydroperiod. The subsequentlyhigher rates (290e430 g C m�2 yr�1) are most closely associatedwith an increase in Osmunda and Cladium abundance, which isconsistent with other sites that experience higher peat accumula-tion with the establishment of these taxa. The further drying,exemplified by a decrease in Osmunda abundance and substantial

increase in Ilex, that occurred 50e100 years ago resulted in a dra-matic drop in C accumulation, suggesting the threshold for opti-mum C accumulation and hydroperiod was crossed.

Similar to the tree islands, ridge expansion, and consequently,increases in C accumulation rates, occurred with precipitationminima resulting from a southward shift of the ITCZ during theMedieval Climate Anomaly and Little Ice Age (w600e100 cal yr BP)(Haug et al., 2001; Bernhardt and Willard, 2009). The first of theseshifts occurs after 1000 cal yr BP, and themost pronounced increasein the northernmost ridge is centered around 700 cal yr BP (Fig. 3),

Table 3Summary of hydroperiod, vegetation, and average carbon accumulation rates in the cores studied.

Environment Water depth(cm)

Hydroperiodlength (days)

Dominant vegetation Average C accumulationrate (g C m�2 yr�1)

Notes

Head <30 Subtropical hardwoods, Morella cerifera,Acrostichum danaeifolium, Blechnum serrulatum,Thelypteris kunthii

64.66

Near Tail 7.4 � 8.6 226 Morella cerifera, Cladium jamaicense,Cephalanthus occidentalis, Salix caroliniana

165.52

Ridge 48.5 � 1.3 357 Cladium jamaicense 131.05Slough 67.1 � 1.7 363 Nymphaea, Nuphar, Utricularia, periphyton 31.18Marl Prairie w10 90e270 Cladium jamaicense, Poaceae, Cyperaceae,

periphyton, sparse vegetation9.92 Short hydroperiod results in marl

precipitation, longer promotes peataccumulation

M.C. Jones et al. / Quaternary Science Reviews 90 (2014) 90e105102

which coincides with a precipitation minimum associated with theend of the MCA (Haug et al., 2001). Several tree island sites (05-7-27-9, 98-4-23, 00-3-7-1) and ridge sites (00-8-10-13, 02-05-21-2)show elevated C accumulation rates beginning 200e150 cal yr BP,coincident with the last precipitationminimum associatedwith theLIA (Haug et al., 2001), but the resolution in this part of the record isnot high enough to distinguish between climatic, anthropogenic,and autogenic influences.

In contrast to tree island expansion, a shorter hydroperiod in themarl prairie resulted in a switch from peat to marl that decreased Caccumulation rates (Fig. 8). The marl is composed of submergedperiphyton microalgae and macrophytes that precipitate calciumcarbonate. These marl systems have C accumulation rates that rivalaverage boreal peatland C accumulation rates (Yu et al., 2009), butthey accumulated more C during the longer hydroperiod peatphases than the marl phases. In the case of the core 08-05-19-2, theshift from peat to marl w700 cal yr BP can likely be attributed to ashift to a drier climate associated with a southward shift in the ITCZ(Haug et al., 2001;Willard and Bernhardt, 2011) associatedwith theend of theMCA and continuing into the LIA (Willard and Bernhardt,2011). Although the clear change from peat to marl is not apparentin the marl prairie site 08-08-2-7a, a period of lower C accumula-tion rates occurs fromw700 to 100 cal yr BP coincident with a shiftto drier conditions and several precipitation minima associatedwith the MCA and LIA (Haug et al., 2001). The higher C accumula-tion rates from 3300 to 2700 cal yr BP in 08-08-2-7a coincidewith aperiod of high precipitation variability, interspersed with severalprecipitation minima (Haug et al., 2001), suggesting C accumula-tion in this core is closely linked to changes in precipitation.

4.4. The impact of nutrient dynamics on C accumulation

The Everglades are considered an oligotrophic system, limitedby phosphorous. The highest phosphorous concentrations in thefreshwater Everglades can be found on tree islands heads, withlower but still elevated phosphorous concentrations found in neartail sediments (Wetzel et al., 2009). Data suggest that phosphorousredistribution from marshes to tree islands occurs through threemain mechanisms: evapotranspirational pumping of surface andgroundwater toward the islands (Ross et al., 2006), dry falloutdeposition (Wetzel et al., 2005), and increased deposition of animalfeces, especially guano from birds (Frederick and Powell, 1994). Thisincreased productivity allows for peat to accumulate at a muchfaster rate than the surrounding marsh, a positive feedback thatallows tree islands to continue to emerge above the marsh surface.The result is that from the tree islands studied, C accumulation rateswere an order of magnitude higher than the ridges and sloughs,marshes, and marl prairie cores. This agrees with recent findingsfrom Givnish et al. (2008) who found that peat accumulation rateswere nearly 4 times higher in the near tail of Manatee Hammock in

the southern Shark River Slough than the adjacent marsh. All of thecores were taken from the near tails of the tree islands, except for05-7-27-9, which was taken from the head. It accumulated carbonat a rate much slower than the other near tail cores, but becauseonly one core from a tree island head was examined, it is difficult todraw conclusions about whether heads accumulate carbon at aslower rate than near tails. However, the heads are the driest of allthe sites studied and are likely subject to greater aerobic decom-position, leading to lower apparent C accumulation rates.

Ridges and sloughs appear to cycle carbon quickly, as C accu-mulation rates are low and relatively invariable throughout each ofthese cores with the exception of the two northernmost sites(02-5-20-13 and 02-5-20-14). Rates increase in the uppermostportions of these cores, which may primarily reflect the fact thatthese sediments have had less time to be subjected to decompo-sition. While ridge expansion appears to be triggered by a drierclimate (Bernhardt andWillard, 2009), peat accretion in ridges andsloughs is sensitive to P concentrations and soil redox potential inaddition to hydroperiod (Larsen et al., 2007). Enhanced P helpspromote peat accumulation, but only when coupled with a lowerredox potential, since otherwise decomposition would also in-crease (Larsen et al., 2007). However, Larsen et al. (2007) modelingefforts found that a reduced hydroperiod would result in flatterridges and slough infilling, while a greater hydroperiod wouldpromote higher ridge development, which is not what our coreanalysis seems to suggest.

4.5. Fire and C accumulation

In cores where charcoal was quantified, higher C/P ratios,indicative of greater local fires, correspond with lower C accumu-lation rates, suggesting that fire not only has the ability to directlyinfluence C accumulation rates likely through peat consumption,but that increased fire prevalence also coincides with drier climaticconditions, as evidenced by the increase in dry indicator macro-phytes in ridge and slough cores, such as Amaranthaceae pollen(Fig. 3). Extended periods of aridity would result in lower primaryproductivity and greater peat oxidation that would lower peataccumulation. In both marl prairie cores, the drier climate condi-tions in the Everglades increased the occurrence of fires. Further-more, an increase in local fires negatively impacts C accumulationrates through direct peat consumption.

4.6. Water management impacts in the Florida Everglades

Beginning in the early 20th century with the construction of asystem of canals and levees built by the U.S. Army Corps of Engi-neers, the flow of water in south Florida has been controlled toprovide flood protection, water supply, and to increase agricul-tural development (McAvoy et al., 2011). These changes impacted

M.C. Jones et al. / Quaternary Science Reviews 90 (2014) 90e105 103

the hydrology significantly enough across much of the Evergladesto alter the native vegetation and the natural topographic andmicro-scale heterogeneity. Even though coring sites were selectedto minimize the impacts of nutrient enrichment associated withagricultural run-off and proximity to canals, the change in hy-drology not only had obvious impacts to the vegetation but also tothe rates of carbon accumulation in most of the cores studied. In1945, water conservation areas (WCAs) were established througha construction of levees and canals to help meet the water needs oflocal agriculture and a growing population. They were intended tonot only control water distribution, but also to help in flood pro-tection and water supply storage. Three WCAs were established,including WCA-1 that surrounded the Loxahatchee WildlifeRefuge. The Loxahatchee tree island core show a dramatic 3-folddecline in C accumulation rates in the last 80 years of the record.This timing coincides with the construction of the Hoover Dikeand Ocean Canal in the early 20th century, which disrupted sheetflow (Brandt et al., 2002) and increased abundance of shrubby taxaon tree islands (Willard et al., 2006). By 1961 the entire Refugewasencircled by canals and levees, effectively turning the system fromaminerotrophic to an ombrotrophic wetland (Willard et al., 2006).The resolution of our core is not high enough during this timeperiod to accurately assess the impact on carbon accumulation forthe last several decades, but significant vegetational changes,notably an increase in Ilex pollen and decrease in Osmunda spores,over the last 50e100 years of the record coincide with a precipi-tous decline in C accumulation as conditions on this tree islandbecame drier (Bernhardt et al., 2013). These water managementpractices alone caused subsidence and oxidation of the 4.3-m ofpeat formerly found throughout the Loxahatchee National Wild-life Refuge prior to drainage (Dachnowski-Stokes, 1930), resultingin a major loss of the Everglades Holocene carbon reservoir. Asimilar decline in C accumulation rates is observed in the Skinner’sHead tree island core 98-4-23 found in WCA-3A, beginning w AD1950, which coincides with the construction of the Miami, NorthNew River, and Hillsborough Canals, and the Bolles and CrossCanals and Hoover Dike to the north, increasing Osmunda abun-dance (Willard et al., 2006). The construction of the leveesdemarcating WCA-3A appears to have resulted in the most dra-matic decreases in C accumulation rates in the southern part ofWCA-3A, since it resulted in a lengthening of the hydroperiod,driven by the pooling of water along the southern boundary(Willard et al., 2006).

From the Everglades National Park, the near tail tree islandGumbo Limbo Hammock core (00-8-7-1) experiences a sharp 2- to3-fold increase in C accumulation rates in the early 20th century.Despite its designation as a National Park, the area is impacted byall of the water control structures to the north, as well as theconstruction of Tamiami Trail. The pollen assemblages from thisand other Everglades NP cores show an increase in monolete sporescharacteristic of tree island heads and decreasing Osmunda abun-dance characteristic of near tails, suggesting that tree island headsexpanded as a function of decreased sheet flow (Willard et al.,2006). This suggests that tree island head expansion decreasesoverall C accumulation, most likely related to higher aerobicdecomposition with decreased hydroperiod.

The implementation of water management practices not onlyhas resulted in an alteration of composition and carbon dynamicswithin tree islands, but has also resulted in the reduction in theoverall number of tree islands by 50% between 1940 and 1995(Lodge, 2005). Because tree islands store a significantly higherproportion of carbon than the surrounding marsh and slough,this decline in tree island abundance alone has resulted in a loss ofa disproportionally large amount of carbon through physicaldestruction and oxidation.

Ridges and sloughs have accumulated carbon at slow rates overtheir Holocene history, but nearly exponential increases in Caccumulation rates occurs in all cores since AD 1950, which maysimply be a function of rapid carbon cycling in these systems.However, in several of the cores analyzed, C accumulation rateswere an order of magnitude higher after AD 1950 than before. Thisis especially true of the northernmost core locations, where Caccumulation rates are 12e27 times higher after 1950 than before,suggesting they were impacted by the hydrological changesassociated with water management practices and nutrientenrichment related to agricultural runoff. The rates in the ridgecore 00-8-10-13 are even 2e3 times higher after AD 1950 than thehighest rates in tree island cores (400 g C m�2 yr�1 vs. 145e250 g C m�2 yr�1), which suggests additional anthropogenic in-fluence on C accumulation rates. These sites are in close proximityto Alligator Alley (I-75), north of which are several stormwatertreatment areas and the Everglades agricultural area, thus some ofthe phosphorus originating from these areas to the north are likelyimpacting the northernmost coring locations. A study of surfacesediments from 1992 to 2007 found that 21e30% of WCA-3 hadmore than 500 mg kg�1 of P in the upper 10 cm of the sediment,indicating P enrichment above historic concentrations (Brulandet al., 2006, 2007). Although P was not directly measured in thisstudy, Reddy et al. (1993) found a direct relationship betweencarbon accumulation and phosphorus accumulation, suggestinghigher phosphorus inputs in these sites promoted primary pro-ductivity in the otherwise oligotrophic, P-limited system, allowingfor rapid peat accumulation in both the ridges and sloughs. Thisrelationship was also found in a transect of cores taken near theHillsboro canal, where the highest rates of C accumulation sincethe 1970s occurred in closest proximity to the canal and decreasedwith distance from the canal (Craft and Richardson, 1993). Theauthors also suggest that the elevated rates of net primary pro-ductivity caused by the additional phosphorous more than com-pensates for any increase in decomposition, which agrees with theresults of this study. Furthermore, the increased phosphorous in-puts since water management practices and agriculture beganhave the potential to release phosphorous from the surface soilsfor the next 50e120 years (Reddy et al., 2011), especially in theEverglades agricultural area to the north. The unusually high Caccumulation in the slough core 00-8-10-14 corresponds to aninterval where an algal mat comprises the peat, suggesting that theanthropogenic influence resulted in a shift from detrital organicmaterial and floc that often makes up slough peat to a primarilyalgal community. The ridge and slough pair a few kilometers to thesoutheast (02-5-2-13/14) also have elevated C accumulation rates,but they are w6e8 times lower than the two northernmost sites.This is in agreement with the hypothesis that elevated phospho-rous concentrations closest to Alligator Alley drop off with dis-tance from the input source, as found in numerous other studiesacross the Everglades system (DeBusk et al., 1994, 2001; Newmanet al., 1997).

Because of early agricultural drainage of the northern sawgrassplains extending south of Lake Okeechobee, this study has notincluded this region of the Florida Everglades. However, some ofthe oldest, thickest peats have been recorded from this part of theEverglades wetland complex (McDowell et al., 1969; Gleason andStone, 1994), indicating that a significant portion of the originalEverglades carbon reservoir has been lost to agricultural drainage.

4.7. Everglades in the context of the global carbon cycle

The Florida Everglades initiated as sea level rose during theHolocene, and peat began accumulating in depressions as early as7000 years ago, and the oldest freshwater peat dates to nearly

M.C. Jones et al. / Quaternary Science Reviews 90 (2014) 90e105104

6300 cal yr BP from the southern end of Lake Okeechobee (Gleasonand Stone, 1994). The majority of the Everglades wetland, however,did not form until the mid-Holocene 5000e3000 cal yr BP(McDowell et al., 1969; Gleason and Stone, 1994). The expansion ofthe Everglades is coeval with the establishment of peatlands andswamps on the Atlantic and Gulf Coastal plains, including theDismal Swamp, Virginia and North Carolina (Cocke et al., 1934;Whitehead and Oaks, 1979) and the Okefenokee Swamp in Geor-gia (Spackman et al., 1976), and can be attributed to the combinedconsequences of sea-level rise, orbitally driven changes in atmo-spheric circulation, and a stabilization of higher water tables(Willard and Bernhardt, 2011). The mid-Holocene expansion ofthese subtropical to temperate peatlands is much later thannorthern high latitude peatlands (Smith et al., 2004; MacDonaldet al., 2006; Jones and Yu, 2010), which began expanding duringthe Lateglacial (Jones and Yu, 2010) and the early Holocene (Smithet al., 2004; MacDonald et al., 2006). The timing of Indonesianpeatland expansion, corresponding to nearly 74% of tropical peat-land carbon (Page et al., 2011), peaks w6000 cal yr BP (Page et al.,2004; Dommain et al., 2011). This peat expansion is slightly earlierthan the peak in the Florida Everglades, but the expansion in bothlocations is closely linked to Holocene sea level stabilization. Themid-Holocene expansion of the 12,000-km2 pre-anthropogenicallyinfluenced Everglades as well as other large subtropical, mid-latitude, and tropical peat-accumulating wetlands, likely had animpact on not only atmospheric CO2 levels, but also released sig-nificant quantities of CH4 to the atmosphere. These new wetlandswere likely contributors to the mid- to late-Holocene increase inatmospheric CH4 concentrations after 4ka, which has been pro-posed based on the isotopic signature of dDCH4 and d13CCH4 ofmethane (Sowers, 2010). A decrease in the interpolar gradient ofCH4 signifies a decrease in high latitude sources and an increase inlow latitude sources at that time (Sowers, 2010).

5. Conclusions

Long-term rates of C accumulation in the Florida Everglades areon average higher than the northern high latitude rates, and insome cases rival C accumulation rates in tropical peatlands such asIndonesia. Ridges and sloughs accumulate the least amount ofcarbon, while tree-island near tails accumulate most. In general,higher C accumulation is associated with lower hydroperiod, suchas in ridges and near tails of tree islands, but decreases whenconditions become too dry, such as on heads of tree islands. Caccumulation rates were impacted by paleoclimate events, mostnotably the Medieval Climate Anomaly and the Little Ice Age, butthe changes were not uni-directional. The MCA resulted in ridgeexpansion and tree island development, which increased C accu-mulation rates, whereas MCA and LIA drying in the marl prairielargely resulted in decreased C accumulation rates, since thedecrease in hydroperiod resulted in marl precipitation. The largestimpacts to C accumulation rates over the Holocene record havebeen the anthropogenic changes associated with expansion ofagriculture and the construction of canals and levees as part of thewater management practices in the 20th century. The Florida Ev-erglades as a whole have acted as a significant carbon sink over themid- to late-Holocene, but loss of the original wetland area, as wellas the alteration of natural hydrology in the late 19th and 20thcenturies have significantly reduced the carbon sink capacity of thissubtropical wetland.

Acknowledgments

Field work for core collection was supported by USGS GreaterEverglades Priority Ecosystems Science (GEPES). The research is

supported by the USGS Climate and Land Use Mission AreaResearch and Development Program. Any use of trade, product orfirm names is for descriptive purposes only and does not implyendorsement by the U.S. government.

Appendix A. Supplementary data

Supplementary data related to this article can be found at http://dx.doi.org/10.1016/j.quascirev.2014.02.010.

References

Abtew, W., Huebner, R.S., Ciuca, V., 2006. Hydrology of the South Florida environ-ment. In: South Florida Water Management District (Ed.), South Florida Envi-ronmental Report. South Florida Water Management District, West Palm Beach.

Bernhardt, C.E., Willard, D.A., 2009. Response of the Everglades ridge and sloughlandscape to climate variability and 20th-century water management. Ecol.App. 19, 1723e1738.

Bernhardt, C.E., Willard, D.A., Marot, M., Holmes, C.W., 2004. Anthropogenic andNatural Variation in Ridge and Slough Pollen Assemblages. U.S. Geol. SurveyOpen-File Report 2004e1448.

Bernhardt, C., 2011. Native Americans, regional drought and tree island evolution inthe Florida Everglades. Holocene. http://dx.doi.org/10.1177/0959683611400204.

Bernhardt, C.E., Brandt, L.A., Landacre, B., Marot, M.E., Willard, D.A., 2013. Recon-structing vegetation response to altered hydrology and its use for restoration,Arthur R. Marshall Loxahatchee National Wildlife Refuge, Florida. Wetlands,1e11.

Blaauw, M., Christen, J.A., 2011. Flexible paleoclimate age-depth models using anautoregressive gamma process. Bayesian Anal. 6, 457e474.

Brandt, L.A., Silveira, J.E., Kitchens, W.M., Sklar, F.H., Valk, A., 2002. Tree islands ofthe Arthur R. Marshall Loxahatchee National Wildlife Refuge. In: Sklar, F.H., VanDer Valk, A. (Eds.), Tree Islands of the Everglades. Kluwer Academic Publishers,pp. 311e335.

Bruland, G.L., Grunwald, S., Osborne, T.Z., Reddy, K.R., Newman, S., 2006. Spatialdistribution of soil properties in Water Conservation Area 3 of the Everglades.Soil. Sci. Soc. Am. J. 70 (5), 1662e1676.

Bruland, G.L., Osborne, T.Z., Reddy, K.R., Grunwald, S., Newman, S., DeBusk, W.F.,2007. Recent changes in soil total phosphorus in the Everglades: water con-servation area 3. Environ. Monit. Assess. 129 (1e3), 379e395.

Central Everglades Restoration Project (CERP), 2012. http://www.evergladesplan.org/index.aspx.

Chabbi, A., McKee, K.L., Mendelssohn, I.A., 2000. Fate of oxygen losses from Typhadomingensis (Typhaceae) and Cladium jamaicense (Cyperaceae) and conse-quences for root metabolism. Am. J. Bot. 87 (8), 1081e1090.

Chmura, G.L., Anisfield, S.C., Cahoon, D.R., Lynch, J.C., 2003. Global carbon seques-tration in tidal, saline wetland soils. Glob. Biogeochem. Cycles 17. http://dx.doi.org/10.1029/2002GB001917.

Clark, J.S., 1988. Particle motion and the theory analysis: source area, transport,deposition, and sampling. Quat. Res. 30, 67e80.

Cocke, E.C., Lewis, I.F., Patrick, R., 1934. A further study of dismal swamp peat. Am. J.Bot. 21, 374e396.

Craft, C.B., Seneca, E.D., Broome, S.W., 1991. Loss on ignition and Kjeldahl digestionfor estimating organic carbon and total in nitrogen in estuarine marsh soils:calibration with dry combustion. Estuaries 14, 175e179.

Craft, C.B., Richardson, C.J., 1993. Peat accretion and phosphorous accumulationalong a eutrophication gradient in the northern Everglades. Biogeochemistry22, 133e156.

Dachnowski-Stokes, A.P., 1930. Peat profiles of the Everglades in Florida: thestratigraphic features of the “Upper” Everglades and correlation with envi-ronmental changes. J. Wash. Acad. Sci. 20, 89e107.

Davis, S.M., Gaiser, E.E., Loftus, W.F., Huffman, A.E., 2005. Southern marl prairiesconceptual ecological model. Wetlands 25 (4), 821e831.

Dean, W.E., 1974. Determination of carbonate and organic matter in calcareoussediments and sedimentary rocks by loss on ignition: comparison with othermethods. J. Sed. Petrol. 44, 242e248.

DeBusk, W.F., Reddy, K.R., Koch, M.S., Wang, Y., 1994. Spatial patterns of soilphosphorus in Everglades Water Conservation Area 2A. Soil. Sci. Soc. Am. J. 58,543e552.

DeBusk, W.F., Newman, S., Reddy, K.R., 2001. Spatio-temporal patterns of soilphosphorus enrichment in Everglades Water Conservation Area 2A. J. Environ.Qual. 30, 1438e1446.

Duever, M.J., Meeder, J.F., Meeder, L.C., McCollom, J.M., 1994. The climate in SouthFlorida and its role in shaping the Everglades ecosystem. In: Davis, S.M.,Ogden, J.C. (Eds.), Everglades: the Ecosystem and its Restoration. St. Lucie Press,Delray Beach, FL, pp. 225e248.

Dommain, R., Couwenberg, J., Joosten, H., 2011. Development and carbon seques-tration of tropical peat domes in south-east Asia: links to post-glacial sea-levelchanges and Holocene climate variability. Quat. Sci. Rev. 30, 999e1010.

Enfield, D.B., Mestas-Nunez, A.M., Trimble, P.J., 2001. The Atlantic multidecadaloscillation and its relation to rainfall and river flows in the continental US.Geophys. Res. Lett. 28, 2077e2080.

M.C. Jones et al. / Quaternary Science Reviews 90 (2014) 90e105 105

Frederick, P.C., Powell, G.V., 1994. Nutrient transport by wading birds in the Ever-glades. In: Davis, S.M., Ogden, J.C. (Eds.), Everglades: the Ecosystem and itsRestoration. St. Lucie Press, Delray Beach, pp. 571e584.

Givnish, T.J., Volin, J.C., Owen, V.D., Volin, V.C., Muss, J.D., Glaser, P.H., 2008. Vege-tation differentiation in the patterned landscape of the central Everglades:importance of local and landscape drivers. Glob. Ecol. Biogeogr. 17, 384e402.

Glaser, P.H., Volin, J.C., Givnish, T.J., Hansen, B.C.S., Stricker, C.A., 2012. Carbon andsediment accumulation in the Everglades (USA) during the past 4000 years:rates, drivers, and sources of error. J. Geophys. Res. 117, G03026. http://dx.doi.org/10.1029/2011JG001821.

Gleason, P.J., Stone, P., 1994. Age, origin, and landscape evolution of the Evergladespeatland. In: Davis, S.M., Ogden, J.C. (Eds.), Everglades: the Ecosystem and itsRestoration. St. Lucie Press, Delray Beach, pp. 149e197.

Gorham, E., 1991. Northern peatlands: role in the carbon cycle and probable re-sponses to climatic warming. Ecol. Appl. 1, 182e195.

Gorham, E., Janssens, J.A., Glaser, P.H., 2003. Rates of peat accumulation during thepostglacial period in 32 sites from Alaska to Newfoundland, with specialemphasis on northern Minnesota. Can. J. Bot. 81, 429e438.

Gunderson, L.H., 1994. Vegetation of the Everglades: determinants of communitycomposition. In: Davis, S.M., Ogden, J.C. (Eds.), Everglades: the Ecosystem andits Restoration. St. Lucie Press, Delray Beach, FL, p. 848, 323e340.

Harvey, J.W., McCormick, P.V., 2009. Groundwater’s significance to changing hy-drology, water chemistry, and biological communities of a floodplainecosystem, Everglades, South Florida, USA. Hydrogeol. J. 17, 185e201.

Haug, G.H., Hughen, K.A., Sigman, D.M., Peterson, L.C., Röhl, U., 2001. Southwardmigration of the intertropical convergence zone through the Holocene. Science293, 1304e1308.

Heisler, L., Towles, D.T., Brandt, L.A., Pace, R.T., 2002. Tree island vegetation andwater management in the central Everglades. In: Sklar, F.H., van der Valk, A.G.(Eds.), Tree Islands of the Everglades. Kluwer Academic, Boston, pp. 283e309.

Jones, M.C., Yu, Z., 2010. Rapid deglacial and early Holocene expansion of peatlandsin Alaska. Proc. Nat. Acad. Sci. U. S. A. 107 (16), 7347e7352.

Joosten, H., Clarke, D., 2002. Wise use of mires and peatlands e background andprinciples, including a framework for decision-making. In: International MireConservation Group/International Peat Society, Jyväskylä, Finland.

Kivinen, E., Pakarinen, P., 1981. Geographical distribution of peat resources andmajor peatland complex types in the world. Ann. Acad. Sci. Fennicae (Geol.Geogr. III 132, 1e28.

Lähteenoja, O., Ruokolainen, K., Schulman, L., Oinonen, M., 2009. Amazonianpeatlands: an ignored C sink and potential source. Glob. Change Biol. 15 (9),2311e2320.

Lähteenoja, O., Flores, B., Nelson, B., 2013. Tropical peat accumulation in CentralAmazonia. Wetlands, 1e9.

Langeland, K., 1990. Exotic Woody Plant Control Florida. Florida CooperativeExtension Service. Circular 868.

Larsen, L.G., Harvey, J.W., Crimaldi, J.P., 2007. A delicate balance: ecohydriologicalfeedbacks governing landscape morphology in a lotic peatland. Ecol. Mon. 77,591e614.

Lodge, T.E., 2005. The Everglades Handbook: Understanding the Ecosystem. CRCPress, Boca Raton, p. 302.

Lodge, T.E., 2010. The Everglades Handbook: Understanding the Ecosystem, thirded. CRC Press, Boca Raton, Florida.

MacDonald, G.M., Beilman, D.W., Kremenetski, K.V., Sheng, Y., Smith, L.C.,Velichko, A.A., 2006. Rapid early development of circumarctic peatlands andatmospheric CH4 and CO2 variations. Science 314, 285e288.

McAvoy, C.W., Said, W.P., Obeysekera, J., Arman, J.V., Dreschel, T., 2011. Landscapesand Hydrology of the Predrainage Everglades. University Press of Florida,Gainesville, p. 576.

McDowell, L.L., Stephens, J.C., Stewart, E.H., 1969. Radiocarbon chronology of theFlorida Everglades peat. Soil. Sci. Soc. Am. Proc. 33, 743e745.

Neuzil, S.G., 1997. Onset and rate of peat and carbon accumulation in four domedombrogenous peat deposits, Indonesia. In: Rieley, J.O., Page, S.E. (Eds.), Biodi-versity and Sustainability of Tropical Peatlands. Samara Publishing, Cardigan,pp. 55e72.

Newman, S., Reddy, K.R., DeBusk, W.F., Wang, Y., Fisher, M.M., Shih, G., 1997. Spatialdistribution of soil nutrients in a northern Everglades Marsh: water conserva-tion area. Soil. Sci. Soc. Am. J. 61 (4), 1275e1283.

Olmsted, I.C., Loope, L.L., 1984. Plant communities of everglades National Park. In:Environments of South Florida: Past and Present II. Miami Geological Society,Coral Gables, FL, USA, pp. 167e184.

Page, S.E., Wust, R.A.J., Weiss, D., Rieley, J.O., Shotyk, W., Limin, S.H., 2004. A recordof Late Pleistocene and Holocene carbon accumulation and climate change froman equatorial peat bog (Kalimantan, Indonesia): implications for past, presentand future carbon dynamics. J. Quat. Sci. 19, 625e635.

Page, S.E., Rieley, J.O., Banks, C.J., 2011. Global and regional importance of thetropical peatland carbon pool. Glob. Change Biol. 17, 798e818. http://dx.doi.org/10.1111/j.1365-2486.2010.02279.

Pederson, D.C., Peteet, D.M., Kurdyla, D., Guilderson, T., 2005. Medieval warming,Little Ice Age, and European impact on the environment during the last mil-lennium in the lower Hudson Valley, New York, USA. Quat. Res. 63, 238e249.

Reddy, K.R., DeLaune, R.D., DeBusk, W.F., Koch, M., 1993. Long-term nutrient accu-mulation rates in the Everglades wetlands. Soil. Sci. Soc. Am. J. 57, 1145e1155.

Reddy, K.R., Newman, S., Osborne, T.Z., White, J.R., Fitz, H.C., 2011. Phosphorouscycling the greater Everglades Ecosystem: legacy phosphorous implications formanagement and restoration. Crit. Rev. Env. Sci. Tech. 41, 149e186.

Reimer, P.J., Baillie, M.G.L., Bard, E., Bayliss, A., Beck, J.W., Blackwell, P.G.,Ramsey, C.B., Buck, C.E., Burr, G.S., Edwards, R.L., Friedrich, M., Grootes, P.M.,Guilderson, T.P., Hajdas, I., Heaton, T.J., Hogg, A.G., Hughen, K.A., Kaiser, K.F.,Kromer, B., McCormac, F.G., Manning, S.W., Reimer, R.W., Richards, D.A.,Southon, J.R., Talamo, S., Turney, C.S.M., van der Plicht, J., Weyhenmeyer, C.E.,2011. IntCal09 and Marine09 radiocarbon age calibration curves, 0e50,000years cal. BP. Radiocarbon 51, 1111e1150.

Richardson, C.J., Huvane, J.K., 2008. Ecological status of the Everglades: environ-mental and human factors that control the peatland complex on the landscape.In: Richardson, C.J. (Ed.), The Everglades Experiments: Lessons for EcosystemRestoration. Springer, New York, pp. 13e58.

Richardson, C.J., 2010. The everglades: North America’s subtropical wetlands. Wet.Ecol. Man. 18, 517e542.

Ross, M.S., Mitchell-Bruker, S., Sah, J.P., Stothoff, S., Ruiz, P.L., Reed, D.L.,Jayachandran, K., Coultas, C.L., 2006. Interaction of hydrology and nutrientlimitation in the Ridge and Slough landscape of the southern Everglades.Hydrobiologia 569 (1), 37e59.

Schedlbauer, J.L., Oberbauer, S.F., Starr, G., Jimenez, K.L., 2010. Seasonal differencesin the CO2 exchange of a short-hydroperiod Florida Everglades marsh. Agric.For. Meteorol. 150 (7), 994e1006.

Smith, L.C., MacDonald, G.M., Velichko, A.A., Beilman, D.W., Borisova, O.K., Frey, K.E.,Kremenetski, K.V., Sheng, Y., 2004. Siberian peatlands a net carbon sink andglobal methane source since the Early Holocene. Science 303, 353e355.

Snyder, G.H., Davidson, J.M., 1994. Everglades agriculture: past, present, and future.In: Davis, S.M., Ogden, J.C. (Eds.), Everglades: the Ecosystem and its Restoration.St. Lucie Press, Delray Beach, pp. 85e115.

Sowers, T., 2010. Atmospheric methane isotope records covering the Holoceneperiod. Quat. Sci. Rev. 29, 213e221.

Spackman, W., Cohen, A.D., Give, P.H., Casagrande, D.J., 1976. A Field Guidebook toAid in the Comparative Study of the Okefenokee Swamp and the Everglades-Mangrove Swamp-marsh Complex of Southern Florida. Coal Research Section,The Pennsylvania State University, University Park.

Stahle, D.W., Cleaveland, M.K., 1994. Tree-ring reconstructed rainfall over thesoutheastern USA during the Medieval Warm Period and Little Ice Age. Clim.Change 26, 199e212.

Traverse, A., 2007. Paleopalynology, second ed. Springer, Dordrecht.Turunen, J., Tomppo, E., Tolonen, K., Reinikainen, A., 2002. Estimating carbon

accumulation rates of undrained mires in Finlandeapplication to boreal andsubarctic regions. Holocene 12 (1), 69e80.

Wetzel, P.R., van der Valk, A.G., Newman, S., Gawlik, D.E., Troxler Gann, T., Coronado-Molina, C.A., Sklar, F.H., 2005. Maintaining tree islands in the Florida Everglades:nutrient redistribution is the key. Front. Ecol. Environ. 3 (7), 370e376.

Wetzel, P.R., van der Valk, A.G., Newman, S., Coronado, C.A., Troxler-Gann, T.G.,Childers, D.C., Orem, W.H., Sklar, F.H., 2009. Heterogeneity of phosphorusdistribution in a patterned landscape, the Florida Everglades. Plant Ecol. 200,83e90.

Whitehead, D.R., Oaks, R.Q., 1979. Developmental history of the dismal swamp. In:Kirk, P.W. (Ed.), The Great Dismal Swamp. University Press of Virginia, Char-lottesville, pp. 25e43.

Willard, D.A., Weimer, L.M., Riegel, W.L., 2001. Pollen assemblages as paleoenvir-onmental proxies in the Florida Everglades. Rev. Palaeobot. Palynol. 113 (4),213e235.

Willard, D.A., Bernhardt, C.E., Holmes, C.W., Landacre, B., Marot, M., 2006. Responseof Everglades tree islands to environmental change. Ecol. Mon. 76, 565e583.

Willard, D.A., Bernhardt, C.E., 2011. Impacts of past climate and sea level change onEverglades wetlands: placing a century of anthropogenic change into a late-Holocene context. Clim. Change 107, 59e80.

Yu, Z., Beilman, D.W., Jones, M.C., 2009. Sensitivity of northern peatland carbondynamics to Holocene climate change. In: Baird, A.J., Belyea, L.R., Comas, X.,Reeve, A.S., Slater, L.D. (Eds.), Carbon Cycling in Northern Peatlands, Geophys-ical Monograph Series, vol. 184. http://dx.doi.org/10.1029/2008GM000872.

Yu, Z., Loisel, J., Brosseau, D.P., Beilman, D.W., Hunt, S.J., 2010. Global peatland dy-namics since the Last Glacial Maximum. Geophys. Res. Let. 37 (13), L13402.