The Role of Management on Methane EmissionsFrom Subtropical Wetlands Embeddedin Agricultural EcosystemsNicholas J. DeLucia1, Nuria Gomez‐Casanovas1,2,3, Elizabeth H. Boughton4 ,and Carl J. Bernacchi5,1,2

1Department of Plant Biology, University of Illinois at Urbana‐Champaign, Urbana, IL, USA, 2Carl R. Woese Institute forGenomic Biology, University of Illinois at Urbana‐Champaign, Urbana, IL, USA, 3Institute for Sustainability, Energy, andEnvironment, University of Illinois at Urbana‐Champaign, Urbana, IL, USA, 4Buck Island Ranch, Archbold BiologicalStations, Lake Placid, FL, USA, 5Global Change and Photosynthesis Research Unit, Agricultural Research Service, UnitedStates Department of Agriculture, Urbana, IL, USA

Abstract Wetlands are an important source of CH4 globally. However, uncertainty surrounding theimpact of anthropogenic activities on CH4 emissions from wetlands limits understanding of how theseecosystems will respond to management changes. Furthermore, by neglecting the potential for managementto influence CH4 emissions likely inflates error of CH4 emissions for regional and global CH4 models. Thisstudy employed a replicated factorial experimental design to investigate how management of theagricultural landscape, including grazing and/or management intensity, influences net CH4 emissions fromembedded, seasonal subtropical wetlands. This research further determined key mechanisms by whichmanagement decisions at the landscape scale modulate CH4 emissions from the embedded wetlands. NetCH4 exchange wasmeasured using a closed chamber system over two complete wet/dry seasonal cycles in 16wetlands embedded in either agronomically improved pastures (improved wetlands) or less intensivelygrazed unfertilized seminative pastures (seminative wetlands), as well as in grazed and ungrazed wetlands ineach treatment. Emissions of CH4 were higher from improved wetlands (2.82 μmol m−2 s−1) thanseminative wetlands (0.75 μmol m−2 s−1), particularly during the wet season. Enhanced CH4 emissions inimproved wetlands relative to seminative wetlands were caused by increased soil wetness and by higherbiomass in improved seminative wetlands. Unlike subtropical flooded pastures, our results showed thatgrazers do not alter CH4 emissions from subtropical wetlands. Current and future changes in managementintensity of pastures may cause shifts in net soil CH4 emissions from embedded subtropical wetlands, whichcould further enhance this emission source.

1. Introduction

Natural wetlands are the largest natural source of global CH4, a potent greenhouse gas with a warmingpotential to the atmosphere much greater than CO2 (Forster et al., 2007; Intergovernmental Panel onClimate Change, 2014) and responsible for between 55% and 77% of global natural CH4 (Ciais et al., 2013;Saunois et al., 2016). Wetlands are loosely defined as being inundated for part, if not most of the year, sup-port vegetation adapted to waterlogged conditions and have soil characteristics that are different than thesurrounding land (Environmental Protection Agency (EPA), n.d.). A large range of wetland types existsacross the globe, which results in varied characteristics related to ecosystem services, carbon cycling, andfunction. Wetlands in the subtropics and tropics are often characterized as marshes that are at the lowestdepression in a landscape, have an abundance of vegetation adapted to saturating conditions, are inundatedfor much of the year, and receive most of their water from groundwater sources and runoff creating a high‐nutrient, alkaline, or minerotrophic system. Historically, wetlands embedded in agriculture were oftendrained and reclaimed to accommodate development; however, efforts are continually pushing to protectthem from further development (Mitsch & Gosselink, 2015; Ramsar Convention Secretariat, 2016).

Land use in the area surrounding wetlands is likely to impact their structure and function. Because they aregenerally the lowest‐lying areas in a landscape, land management decisions at the landscape scale may dra-matically impact these systems. For example, management decisions in the surrounding land are shown to

©2019. American Geophysical Union.All Rights Reserved.

RESEARCH ARTICLE10.1029/2019JG005132

Key Points:• Wetlands are methane sources, but

little is known regarding howsurrounding agricultural practicesimpact emissions

• Wetlands embedded in intenselymanaged pastures have highermethane emissions than wetlandssurrounded by seminativelandscapes

• Whether a wetland was grazed orungrazed had no impact on theamount of methane released fromthe wetland

Supporting Information:• Supporting Information S1• Data Set S1• Data Set S2• Data Set S3• Data Set S4• Data Set S5• Data Set S6

Correspondence to:C. J. Bernacchi,carl.bernacchi@usda.gov

Citation:DeLucia, N. J., Gomez‐Casanovas, N.,Boughton, E. H., & Bernacchi, C. J.(2019). The role of management onmethane emissions from subtropicalwetlands embedded in agriculturalecosystems. Journal of GeophysicalResearch: Biogeosciences, 124. https://doi.org/10.1029/2019JG005132

Received 27 MAR 2019Accepted 23 JUL 2019Accepted article online 23 AUG 2019

Author Contributions:Conceptualization: Nicholas J.DeLucia, Nuria Gomez‐Casanovas,Carl J. BernacchiFormal analysis: Nicholas J. DeLucia,Carl J. BernacchiFunding acquisition: Elizabeth H.Boughton, Carl J. BernacchiInvestigation: Nicholas J. DeLucia,Nuria Gomez‐Casanovas, Carl J.Bernacchi(continued)

DELUCIA ET AL. 1

impact the plant biomass (Boughton et al., 2011), plant diversity (Boughton et al., 2016), soil characteristics(Ho et al., 2018; Verhoeven & Setter, 2010), and soil hydrology (Collins et al., 2014) of the wetlands them-selves. Previous work shows that conversion of wetlands to new ecosystem types can lead to significantchanges in CH4 emissions. For example, converting a marsh to shrimp aquaculture ponds was shown toincrease CH4 emissions tenfold (Yang et al., 2017). However, knowledge of how decisions in the surroundinglandscape affect fluxes of CH4 from the embedded wetlands remains uncertain as outlined previously(Bridgham et al., 2013; Petrescu et al., 2015; Turetsky et al., 2014).

The contribution of subtropical and tropical wetlands to annual CH4 emissions is poorly constrained inmod-els due to a dearth of available data and a lack of understanding of how land use decisions in surroundinglandscapes directly and indirectly impact wetland CH4 fluxes. Although the main drivers controlling CH4

dynamics in wetlands are well known (e.g., anaerobic conditions coupled with soil temperature, soil redoxpotential, substrate supply, etc.; Conrad, 1989), part of the variation in CH4 emissions across sites and timecould be explained by changes in landscape context, defined here as management of both wetlands and ofthe surrounding land. Land use surrounding wetlands may influence their CH4 dynamics, particularly iflandscape management alters wetland soil nutrients and plant biomass. As cultivated land is typically ferti-lized to promote and maintain productivity (Delucia et al., 2014), nutrient applications of the surroundingland will likely drain into lower lying wetlands and stimulate wetland plant growth (Boughton, Quintana‐Ascencio, Nickerson, & Bohlen, 2011). Increases in plant biomass in wetlands can increase CH4 emissionsas plants can act as conduits for CH4 (Blanc‐Betes et al., 2016; Whiting & Chanton, 1993) and may supplysubstrate for methanogenic microbes. However, fertilizers leached from cultivated land into embedded wet-lands can decrease CH4 emissions by stimulating methanotrophs and hence CH4 oxidation (Bangeret al., 2012).

Between 20% and 30% of pasture area in the subtropics and tropics is grazed by livestock (Asner et al., 2004;Ramankutty et al., 2008). Grazers not only can alter the exchange of CO2 and CH4 of pastures (Gomez‐Casanovas et al., 2018) but also can alter the production and oxidation of CH4 from wetlands as they affectsoil temperature, plant productivity, and nutrient inputs (Banger et al., 2012; Bridgham et al., 2013; Whiting& Chanton, 1993). As such, it is unclear how cattle grazing in and around wetlands can influence CH4 emis-sions. Grazers could increase CH4 emissions fromwetlands by increasing C input throughmanure and urinedeposition (Banger et al., 2012; Dong et al., 2006) or by increasing soil wetness and stimulating methanogen-esis over methanotrophy during nonflooded conditions as observed for the surrounding pastures (Gomez‐Casanovas et al., 2018). However, grazing could decrease soil CH4 emissions by decreasing the input of Cto soil due to biomass removal (Tanentzap & Coomes, 2012). It is important to resolve the uncertainty ofthe impact of grazing on methane emissions, especially in the highly grazed subtropics.

In the subtropical regions of Florida, grazed pastures cover >35% of the total land area (Ramankutty et al.,2008; U.S. Department of Agriculture, 2009). Embeddedwithin this larger agricultural landscape are isolated,seasonal wetlands (Boughton et al., 2011). These wetlands are nested within a mosaic of landscapes, whichinclude two types of grazed pastures: grassland ecosystems dominated by native grass species (hereafter semi-native pastures) and grasslands dedicated to maximizing beef cattle production per unit area (hereafterimproved pastures). The seminative and improved classification of these grasslands, consistent for studiesthroughout this region (Boughton et al., 2016), is limited to an agronomic, and not an ecological, perspective.Management regimes in improved pastures are characterized by regular additions of fertilizer, heavy cattlegrazing, drainage ditches, and conversion to nonnative forage grasses, whereas seminative pastures are lessintensively grazed, not fertilized, and composed of a mixture of native and introduced C4 forage grasses(Boughton et al., 2016).

The underlying objective of this study is to determine whether the surrounding landscape and grazing‐baseddisturbance within wetlands impact net CH4 emissions. We hypothesize that grazing disturbance and inten-sification of agriculture surrounding wetlands leads to higher CH4 fluxes. Archbold Biological Station's BuckIsland Ranch research center in Central Florida is an ideal facility to test this hypothesis given that the largeranch consists of a mosaic of land use types that allow for a full factorial land use and grazing experimentwith all other forcing variables remaining relatively constant. If the hypothesis is supported, it would indi-cate that modeling of subtropical wetlands needs to consider more than within‐wetland biogeochemistryand include the impact of the surrounding landscape.

10.1029/2019JG005132Journal of Geophysical Research: Biogeosciences

DELUCIA ET AL. 2

Methodology: Nicholas J. DeLucia,Elizabeth H. Boughton, Carl J.BernacchiProject administration: Elizabeth H.Boughton, Carl J. BernacchiResources: Elizabeth H. Boughton,Carl J. BernacchiSupervision:Nuria Gomez‐Casanovas,Carl J. BernacchiWriting – review & editing: NicholasJ. DeLucia, Nuria Gomez‐Casanovas,Elizabeth H. Boughton, Carl J.Bernacchi

2. Methods2.1. Site Description

This research was conducted between 2013 and 2015 at the Buck Island Ranch operated by ArchboldBiological Research Station in Lake Placid, Florida (27°09′N, 81°11′W). Buck Island Ranch is a 4,252‐hacommercial cattle ranch and ecological field station with over 600 isolated, seasonal wetlands. The climateis classified as humid‐subtropical with two distinct seasons, a wet, hot season fromMay through October anda relatively dry, cool season the rest of the year. Over the course of the study period, annual precipitation was1,534.4 mm on average, with 77% falling during the wet season. Mean temperature was 26.1 °C during thewet season and 19.4 °C during the dry season. Daily and historical temperature and rainfall was obtainedfrom the Midwestern Regional Climate Center (https://mrcc.illinois.edu/CLIMATE/).

Grazed pastures at Buck Island Ranch include intenselymanaged improved pastures and less intenselyman-aged seminative pastures. Improved pastures have been agronomically improved since the 1940s. They arecomposed primarily of introduced forage, Bahia grass (Paspalum notatum Flüggé), are typically fertilizedannually or semiannually with nitrogen (56 kg/ha; Swain et al., 2013), were historically fertilized withNPK (nitrogen, phosphorus, potassium; 1970s to 1986) at a rate of 56‐kg/ha NH4SO4 and NH4NO3 and34–90 kg/ha of P 2O5 and K2O, grazed more intensely during the summer wet season, and have numerousdrainage ditches coupled with water management (Boughton et al., 2016). Seminative pastures have mostlyintact natural vegetation and represent less intensely managed systems. They are composed of a mixture ofnative grasses (i.e., Andropogon spp., Axonopus spp., and Panicum spp.) and Bahia grass. Seminative pas-tures have never been fertilized, are moderately grazed during the winter dry season, and have fewer drai-nage ditches than improved pastures.

Soils at Buck Island Ranch are a mosaic of sandy, well‐drained Spodosols and Inceptisols and lower lyingAlfisols andmuck soils (Ho et al., 2018). Both pasture types contain high and low elevations. During the study(2013–2015) and for over 30 years the average stocking rate ranged from 0.57 to 1.7 animal units (AU)/ha inimproved pastures and 0.15–1.12 AU/ha in seminative pastures (Boughton et al., 2016). AU is a standard unitof measure where it is assumed that one cow is approximately 450 kg that consumes 12 kg · day · ha ofdry matter.

The legacy effects of soil organic carbon content within a wetland site are known to change over time atMEARC (Ho et al., 2018). In order to understand the change over time and over the course of this study,we used wetland soil organic carbon values published by Ho et al. (2018) that were taken from the same wet-lands used in our study. The changes to percent soil organic carbon content from 2007 to 2016 were as fol-lows: ungrazed improved wetlands (56.3–60.9% ± 5.4%, a change of Δ 8.2%), grazed improved wetlands(53.1–54.2% ± 9.7%, Δ 2%), ungrazed seminative wetlands (58.7–60.7% ± 21%, Δ 3.4%), and grazed semina-tive wetlands (47.5–51.1% ± 20%, Δ 7.6%).

We define the small seasonal wetlands (<1 ha−1) according to the surrounding land use and managementpractices. Despite differences in land use and management practices between improved and seminative wet-lands, they are still categorically considered minerotrophic wetlands that are high in nutrients, which pri-marily come from groundwater flow into lower lying depressions of the wetlands. They are embeddedacross the landscape and represent 15% of total areal land cover on the ranch, typical for this region(Gathumbi et al., 2005). Wetlands are typically flooded in the wet season, but hydroperiods follow seasonalrainfall patterns. Cattle roam between pasture and wetland and use wetlands for cooling and feeding. Thetiming and grazing intensity of grazed pastures were similar among treatments.

This experiment consisted of a fully replicated (n= 4) two‐way factorial design with treatments consisting ofwetlands surrounded by either improved or seminative pastures and with or without grazing. A total of 16wetlands (Figure 1) were selected in accordance with treatment type and similar size, shape, and hydroper-iod and are a subset of wetlands from a larger experiment to assess the interaction of fire and grazing on wet-land communities (Boughton et al., 2016; Ho et al., 2018). The wetlands for this experiment (Figure 1) wereselected among the nonburned treatments (Boughton et al., 2016). No wetland was intentionally burnedsince before 2006, but preceding the experiment, wetlands may have been exposed to irregular fire since pas-tures are typically burned every 3–5 years and pasture fires sometimes burn embedded wetlands if condi-tions are dry enough. No wetlands were burned during this experiment. Grazing exclosures for ungrazed

10.1029/2019JG005132Journal of Geophysical Research: Biogeosciences

DELUCIA ET AL. 3

Figure 1. Map of the study site, Buck Island Ranch in Florida, USA. Wetlands used for this study are outlined in black and are located in either improved pastures(dark gray) or seminative pastures (light gray). The dark boxes surrounding certain wetlands represent ungrazed treatments.

10.1029/2019JG005132Journal of Geophysical Research: Biogeosciences

DELUCIA ET AL. 4

wetlands were established using fences to prevent disturbance from cattlein early 2007. All pastures within a treatment were managed similarly.

2.2. Closed Chamber and Auxiliary Measurements and Design

Net CH4 fluxes between the land surface and atmosphere were mea-sured using a custom‐built vented canopy chamber as in Whitinget al. (1991). The volume of the chamber was 1 m3 and consisted ofan open path CH4 analyzer (LI‐7700, LI‐COR, Inc., Lincoln, NEUSA) and open path CO2/H2O infrared gas analyzer (LI‐7500A IRGA,LI‐COR, Inc.) mounted inside the enclosure. An interface unit (LI‐7550, LI‐COR, Inc.) mounted on the outside of the chamber made itpossible to simultaneously record humidity‐corrected CH4 concentra-tions over time at 5 Hz. The chamber was constructed using an alumi-num frame with transparent Lexan and Propafilm (ICI Americas,Chicago, IL, USA) sides. The chamber had vents along the top thatcan be opened to allow for airflow through the chamber after place-ment but before measurements started. The chamber had a 4‐cm alu-minum bottom edge to penetrate the soil and form a seal with thesoil. Fans (12V DC) were mounted in the corners of the chamber toensure proper gas mixing and to prevent large fluctuations in tempera-ture. Prior to closing the vents, the temperature in the chamberremained relatively stable and after the vents were closed temperaturesincreased <1 °C/min inside the chamber. CH4 fluxes were calculatedbased on the linear rate of change of CH4 concentration over ~60 sec-onds following an initial mixing period as described previously(Hutchinson & Livingston, 1993). The initial mixing period (~60 s)was not included in flux calculations to avoid artificially high flux ratesassociated with soil disturbance or changes in pressure. An a priori,acceptable lower threshold r2 was set at 0.8, and any measurementbelow this threshold (23% of flux measurements) was discarded, ensur-ing diffusive fluxes were measured.

To determine whether the surrounding landscape and grazing‐baseddisturbance within wetlands impact net CH4 emissions, eight sets ofmeasurements were made over two complete wet/dry seasons from

2013 until 2015 to capture temporal variation in wetland net CH4 fluxes (red arrows in Figure 2).Measurements taken between July and September represented wet season emissions and betweenNovember and March represented dry season emissions. For each sampling campaign, measurementswere made between 10 am and 3 pm over the course of 2–3 days, provided that similar weather condi-tions persisted. Wetlands for each treatment were sampled randomly at each sampling period to mini-mize confounding effects on fluxes resulting from daily variability. Measurements were completedwithin 1 hr for each wetland per sampling campaign. The chamber was randomly placed throughoutthe wetland without regard for canopy height or vegetation type. In all cases the canopy fits within thechamber, although efforts were made to prevent the vegetation from obscuring the open path of theCO2/H2O and/or CH4 sensor. The vegetation was never clipped. Within each wetland, six subplots of4–5 m in diameter were used to capture spatial variation: two from the outer edge (shallow), two froma midpoint between the outer edge and center (intermediate), and two from the center (deep). Eachspatial pair was averaged together for analysis.

In addition to CH4, CO2, and water vapor fluxes, auxiliary measurements including soil volumetric watercontent (VWC), soil and air temperature, and air humidity were collected for each measurement includingfor both flooded and nonflooded conditions. These parameters were measured at each wetland samplinglocation using handheld soil moisture and temperature sensors (HydroSense II attached to a CS658 probewith 20‐cm‐long rods, Campbell Scientific, Logan, UT, U.S., and a HH‐23 Handheld Thermometer

Figure 2. Daily averages of temperature and precipitation over the 3 yearsof this experiment and historical trends. Mean daily measured tempera-tures (black line) and daily temperature ranges (gray bars) overlay 30‐yearmean temperature ranges (red bands). Daily precipitation is denoted byblack bars. The graph areas between the two vertical lines denote thedefined wet season. Daily and historical temperature and rainfall areobtained from the Midwest Regional Climate Center (https://mrcc.illinois.edu/CLIMATE/). Flux measurements were made within 3 days of each redarrow.

10.1029/2019JG005132Journal of Geophysical Research: Biogeosciences

DELUCIA ET AL. 5

attached to a 20‐cmK‐type probe, Omega Engineering Inc., Stamford, CT, USA).We adjusted the sensor cali-bration based on soil type. For our particular soil type the sensor had a range of 0–40% VWC, where above40% indicated inundation. Air temperature was measured directly inside the chamber to ensure that tem-peratures did not increase >5 °C (109 Temperature Probe, Campbell Scientific) and humidity was accountedfor directly from the IRGA, although three installed fans helped to regulate humidity/H2O sensor.

2.3. Soil and Biomass Sampling and Analysis

Two soil cores were sampled to a depth of 50 cm from each measurement location (shallow, intermediate,and deep) and sample date for each wetland and divided into 0–25‐cm and 25–50‐cm layers for total mineralN (NH4 and NO3) analysis. Samples were first pushed through a 2‐mm sieve to remove roots and then ovendried at 105 °C for at least 24 hr. For each core, total N was measured on 10‐g subsamples for each layer usingan extractant prepared with 2‐MKCl and deionized water (Keeney &Nelson, 1982). Soil and extractant werecombined and shaken for 1 hr. Samples were gravity filtered through Whatman 42 filter paper (GEHealthcare, Buckinghamshire, UK). Nitrate (NO3) was measured via cadmium reduction on a LachatQuickchem 8000, and ammonium (NH4

+) was determined fluorometrically using TurnerDesigns Triolgyflourometer. Additionally, as part of routine wetland surveys conducted frequently at Buck Island Ranch(2006–2009), soil samples were taken from the same corresponding wetlands in this study and measuredfor total N (NO3 and NH4) and were used for comparison purposes (Ho et al., 2018).

Aboveground and belowground biomass and litter were collected from all wetlands in September 2015.Aboveground biomass was taken from each sampling point and wetland and determined by destructive har-vest using a 0.5‐m2 PVC quadrate. Biomass was sorted, and samples were dried at 60 °C until constant dryweight. Concurrently, within the same quadrate, three soil cores down to 50 cm were collected for below-ground biomass. Soil cores were separated into 0–25‐cm and 25–50‐cm increments and pushed through a2‐mm sieve. Roots were sorted into fine roots (<1‐mm diameter) and coarse roots (>1‐mm diameter) andoven dried at 60 °C until constant dry weight.

2.4. Statistical Analysis

Mean CH4 fluxes were calculated as the average of the three spatial locations (shallow, intermediate, anddeep) within each individual wetland and then averaged together per treatment or season (wet versusdry). For spatial differences in CH4 fluxes, two replicates for each sampling location were averaged togetherfor a total of three sampling locations (shallow, intermediate, and deep) per wetland. Differences in discreteCH4 fluxes, aboveground and belowground biomass, and nutrient content between treatments were testedby complete block two‐way repeated measures analysis of variance (ANOVA) with season and the interac-tion of treatments (pasture type/grazing type) as fixed factors. The relationship between mean soil tempera-ture, moisture, and CH4 fluxes was evaluated using the Spearman Rank Order Correlation, a nonparametrictest. All statistical tests were conducted using R (RStudio Team, 2015).

3. Results3.1. Meteorological and Environmental Conditions

This study took place between July 2013 andMarch 2015 and spanned two complete wet (May–October)/dry(November–April) seasonal cycles. The daily mean temperatures throughout this experiment fell within thehistorical, 30‐year mean daily range of temperatures, although the daily mean and maximum temperaturesoften deviated from the historic range (Figure 2).

Historically, cumulative precipitation during wet seasons (1,008.3 mm) exceeded that of the dry seasons(340.3 mm). Over the duration of this study mean cumulative precipitation in both the wet (1,168.4 mm)and dry (366mm) seasons was slightly wetter than average historical trends. For the wet seasons, cumulativeprecipitation was greatest in 2013 (1,376 mm) and lowest in 2015 (1,079 mm; Figure 2). For the dry seasons,cumulative precipitation was greatest in 2015 (488 mm) and lowest in 2013 (236 mm; Figure 2). Over theyears studied, cumulative precipitation during the wet season ranged between 85% and 68% of the totalannual rainfall.

Pasture type influenced both soil temperature andmoisture content of wetlands, particularly during wet sea-sons. During the wet season, seminative wetlands were warmer and drier than improved wetlands (p < 0.05;

10.1029/2019JG005132Journal of Geophysical Research: Biogeosciences

DELUCIA ET AL. 6

Figures 3a and 3b). During the dry seasons, soil moisture and temperaturewere similar between wetland type (Figures 3a and 3b). There was no dis-cernable treatment effect of grazing on either soil temperature or moist-ure. At a seasonal scale, soils were warmer and wetter during the wetthan the dry seasons (p < 0.05; Figures 3a and 3b).

3.2. Net CH4 Fluxes

CH4 emissions were higher in improved than in seminative wetlandsduring the wet season, and they were similar between wetland typeduring the dry season (p < 0.05; Figure 3c). Over the course of thestudy, average CH4 emissions from improved wetlands (2.82 μmol m2

s−1) were fourfold greater than that of seminative wetlands (0.75μmol m2 s−1; p < 0.05; Figure 3c). Differences in CH4 fluxes betweengrazed and ungrazed wetlands were not observed. Averaged withineach season, wetlands were sources of CH4 during both the wet season(1.77 μmol m2 s−1) and during the dry season but to a much lowerextent (0.12 μmol m2 s−1; Figure 3c).

The flux of CH4 to the atmosphere increased with both soil moisture(r2 = 0.56 using Spearmans Rank Order Correlation) and temperature(r2 = 0.49; Figure 4). Wetlands became a high source of CH4 whenthresholds in soil temperature and moisture were reached of ~24 °Cand ~38%, respectively (Figure 4). These thresholds and correlationsdid not vary between wetland type and grazing treatment (datanot shown).

Emissions of CH4 were greater in the deep and intermediate depthlocations of improved wetlands than of the corresponding locationsin seminative wetlands, but they were similar in the shallow samplinglocations during the wet season (Figure 5c). Grazing did not alter CH4

emissions across the hydrological gradient in wetlands (data notshown). During the wet season, the highest mean emissions werefrom the deep (2.90 μmol m2 s−1), followed by the intermediate (1.69μmol m2 s−1), and shallow (1.25 μmol m2 s−1) locations (Figure 5c).For all wetlands, spatial variability in CH4 emissions was lower duringthe dry season than during the wet season. Emissions from the deeplocations of improved wetlands were higher than seminative wetlands(Figure 5c). Wetland emissions were higher at the deep locations thanthe intermediate and shallow locations of wetlands within the samepasture type (p < 0.05; Figure 5c).

Spatially, soil temperature varied by season but remained constantfrom the shallow to the deep wetland locations regardless of treatment(Figure 5a). Weak spatial differences in mean soil moisture betweentreatments were detected during the wet season although the deeplocations within improved wetlands had the highest soil moisture con-tent during both the wet and dry seasons (Figure 5b). When treatmentvariables were excluded from analysis and all spatial values during thewet season were averaged based on location, strong differences arosebetween the deep versus intermediate and deep versus shallow loca-tions, but not the shallow versus intermediate locations (p < 0.05;Figure 5b). During the dry season, spatial differences were detectedwithin the footprint of all wetlands with the highest soil moisture inthe deep and lowest soil moisture in the shallow wetland locations(p < 0.05; Figure 5b).

Figure 3. Average (a) soil temperature (°C), (b) soil moisture (volumetricwater content, VWC), and (c) net CH4 flux (μmol m2 s−1) for each treat-ment, improved/grazed, seminative/grazed, improved/ungrazed, and semi-native/ungrazed, during the wet (May–October) and dry seasons(November–April) over the course of this study. Different letters denotesignificance between seasons (uppercase letters) and between treatmentswithin season (lowercase letters; p < 0.05). Error bars represent the standarderror of the mean.

10.1029/2019JG005132Journal of Geophysical Research: Biogeosciences

DELUCIA ET AL. 7

3.3. Wetland Characteristics3.3.1. Aboveground BiomassAverage green aboveground biomass was almost twofold higher inimproved (235.2 g DW m−2) than in seminative wetlands (128.8 g DWm−2; p < 0.05; Figure 6). Grazing had no detectable effect on green above-ground biomass. A weak effect of wetland type on litter biomass wasfound, with slightly higher litter biomass in improved than seminativewetlands. Surrounding pasture type and grazing had a strong influenceon the ratio of green to litter biomass, which was higher in grazedimproved wetlands (1.87) than grazed seminative wetlands (0.94; p< 0.05;Figure 6). Average aboveground green biomass was consistently higher inthe deep for both wetland types (improved and seminative) and decreasedmoving toward the shallow outer edges (p < 0.05; Figure 7).3.3.2. Belowground BiomassCoarse root biomass in the top 25 cm of the soil column was greater inimproved wetlands than seminative wetlands (p < 0.05; Figure 8) andwas similar between wetland types at 25–50‐cm soil depth. Fine root bio-mass in the top 25 cm and 25–50‐cm depths was similar between treat-ments. Across all treatments, belowground biomass was greater in thetop 25‐cm (125 mg DW) than the 25–50‐cm (28 mg DW) depths (P < 0.05;Figure 8). Grazing did not influence belowground biomass for either wet-land type or root size.3.3.3. Nutrients: NH4, NO3, and Inorganic NPasture type and grazing did not influence wetland total inorganic nitro-gen (IN), nitrate (NO3) values, or ammonium (NH4) in soils (p < 0.05;

Figure 9). Across all treatments, soil nitrogen contents including NH4, NO3, and IN were consistently higherin the 0–25‐cm soil depths than 25–50‐cm depths (p < 0.05; Figure 9).

4. Discussion

The main objective of this research was to assess the impact of intensive management decisions associatedwith subtropical pastures on embedded wetland CH4 emissions. Intensive management of surrounding landassociated with maximizing cattle production (improved pastures) increased net CH4 emissions fromembedded wetlands relative to wetlands nested in land less intensively managed (seminative pastures) dur-ing the wet season (Figure 3c). During the dry season, however, emissions of CH4 from improved and semi-native wetlands were similar, indicating the major role of soil hydrology driving CH4 dynamics ofsubtropical wetlands. Grazing of wetlands had no detectable impact on CH4 fluxes from this system(Figure 3c), although the direct impact of enteric methane emissions and short‐lived soil disturbance asso-ciated with cattle movement was not measured. Fluxes of CH4 had marked seasonality with higher emis-sions when soils were wet and warm, consistent with studies from the same region (Chamberlain et al.,2015; Gathumbi et al., 2005; Wang et al., 2006). Average net CH4 emissions from the wetlands were withinvalues reported for natural and artificially created wetlands from tropical and subtropical regions (0–0.76μmol CH4 m2 s−1; Boon & Mitchell, 1995; Kayranli et al., 2010; Marín‐Munez et al., 2015; Nahlik &Mitsch, 2011; Nicolini et al., 2013; Ortiz‐Llorente & Alvarez‐Cobelas, 2012; Turetsky et al., 2014), althoughwet season means were at the high end of the reported range.

The intensive management of improved pastures consists of fertilizing pastures to support high cattle stock-ing rates (Boughton, Quintana‐ascencio, & Bohlen, 2011). Long‐term addition of fertilizer to pasturesincreases total biomass (Figure 6) and alters plant species composition (see Boughton, Quintana‐ascencio,& Bohlen, 2011) in improved relative to seminative wetlands. Both changes in biomass and in vegetationcomposition could have major impacts on CH4 emissions from wetlands (Turetsky et al., 2014). Increasesin aboveground and belowground biomass can increase C inputs to soils via higher root exudation and litter,fueling CH4 production when methanogenisis is the predominant pathway for decomposition (Whitinget al., 1991; Whiting & Chanton, 1993).

Figure 4. Relationship of soil moisture content (VWC) and soil temperature(°C) on net CH4 fluxes (μmol m2 s−1) for improved and seminative wetlandsover the course of the study. The 3‐D plane represents a LorentzianRegression fitted to all data using a graphing package (SigmaPlot 14, SystatSoftware, Inc.).

10.1029/2019JG005132Journal of Geophysical Research: Biogeosciences

DELUCIA ET AL. 8

Figure 5. Within wetland spatial differences in average (a) soil temperature (°C), (b) soil moisture (VWC), and (c) net CH4fluxes (μmol m2 s−1) in seminative and improved wetlands from the outer edge (shallow), a midpoint between the outeredge and the center (intermediate), and the center (deep) during the wet (May–October) and dry seasons (November–April). Capital letters denote differences between wetland location regardless of wetland type, while lower case lettersdenote differences between wetland type and location (p < 0.05). Error bars represent the standard error of the mean.

10.1029/2019JG005132Journal of Geophysical Research: Biogeosciences

DELUCIA ET AL. 9

Vegetation can act as a conduit for transport of CH4 from the soil to the atmosphere (Dalal et al., 2008; Dinget al., 2005; King et al., 1998; Okazaki et al., 1998; Whiting et al., 1991). Plant functional groups differ in theirability to transport CH4 with graminoids exhibiting greater internal CH4 transport than forbs (Bhullar et al.,2013). At our study site, Juncus effusus (Juncus), a graminoid, is more dominant in improved/grazed wet-lands than seminative wetlands (Boughton, Quintana‐ascencio, & Bohlen, 2011; Ervin & Wetzel, 2002;Gathumbi et al., 2005; Tweel & Bohlen, 2008) and is known to transport CH4 through plant‐mediated trans-port directly from the production site to the atmosphere (Dalal et al., 2008; Henneberg et al., 2015;Henneberg et al., 2015; Petersen et al., 2012; Ström et al., 2015). While juncus is dominant in the improvedand grazed wetlands, it is neither clear which species it may be replacing nor whether it can explain thehigher CH4 emissions from these wetlands. Other productive graminoids in these wetlands have been shownto transport CH4, such as Panicum hemitomon (Ho et al., 2018; Rietl et al., 2017). The extent by which P.

Figure 6. Aboveground biomass including living green and litter biomass, as well as the ratio (green/litter) for each treat-ment. Wetland treatments are as in Figure 2. Different letters indicate significance (p < 0.05). Error bars represent thestandard error of the means.

Figure 7. Spatial differences in aboveground green biomass in improved and seminative wetlands from the outer edge(shallow), a midpoint between the outer edge and the center (intermediate), and the center (deep). Different lettersindicate significance (p < 0.05). Error bars represent the standard error of the means.

10.1029/2019JG005132Journal of Geophysical Research: Biogeosciences

DELUCIA ET AL. 10

hemitomon and Juncus offset each other over the different treatments, aswell as their relative influence in methane transport, is unknown; how-ever, changes in the relative richness or total biomass of these two speciesmay explain some of the observed results. The combination of increasedcarbon availability and species that promote plant‐mediated transport inimproved wetlands may also have contributed to increased CH4 emissionscompared to seminative wetlands supporting the emerging view that spe-cies composition can have a direct impact on CH4 dynamics (Turetskyet al., 2014).

The surrounding land uses may also impact the production of CH4

through nutrient dynamics. The application of fertilizers can stimulateboth biomass productivity and microbial activity, particularly methano-trophs, which would lead to a net reduction of CH4 emissions fromwetlands in improved pastures (Banger et al., 2012). While total soil nitro-gen was historically larger in improved than seminative wetlands (e.g.,Bohlen & Gathumbi, 2007), total soil nitrogen was similar across all treat-ments over the course of this study (e.g., from 2013 to 2015), perhaps due tohigher biomass in the higher‐nutrient fed improved pastures. Therefore, itis likely that the observed increase in CH4 emissions from the improvedwetlands constitutes a combination of factors that include interactions

between microbial populations, plant biomass, and micrometeorological conditions. Given this complexity,additional research on methanogenesis and methane oxidation is required (Megonigal et al., 2004).

Significant flooding, which leads to anaerobic soil conditions, is a major driver of CH4 fluxes (Altor &Mitsch, 2008; Chamberlain et al., 2016; Martikainen et al., 1995; Torn & Chapin, 1993) and could influenceboth soil anaerobic environment and plant biomass. Higher soil wetness is observed in improved wetlandsrelative to seminative wetlands (Figures 3a and 3c), likely as a function of more irrigation and artificialcanals facilitating flow into wetlands. In addition to the impact of fertilization, higher biomass in improvedwetlands could be explained by increased water availability, which could increase carbon inputs into thewetland and provide more labile substrate for methanogenic bacteria, driving higher CH4 emissions.

Flooded soils resulting in anaerobic conditions are requisite for CH4 production. While higher temperaturesgenerally lead to greater CH4 emissions from wetlands, the higher CH4 emissions were observed fromimproved, rather than seminative, wetlands despite seminative wetlands being warmer than improved wet-

lands (Figure 3a). This suggests that temperature is less critical than wateravailability and/or plant litter providing fuel for methanogenesis. The soiltemperatures when CH4 emissions were greatest were also well withinoptimum range for methanogenesis, typically from 20 to 30 °C(Dunfield et al., 1993; Inglett et al., 2012). These results suggest that ifthe two wetland types were similar in temperature, the differences inmethane fluxes would be greater than observed.

Minimum threshold values of soil moisture and soil temperature wererequired before emissions of CH4 were detected (Figure 4). The soil moist-ure threshold (VWC > 38%) was more defined, with a rapid increase inemissions after the threshold was reached, compared with soil tempera-ture, which showed a gradual increase between temperatures rangingfrom 24 to 32 °C, consistent with previous reports (Dunfield et al., 1993;Inglett et al., 2012). These thresholds were not affected by treatment typeand are consistent with previous findings that describe the relationshipbetween CH4 emissions and soil moisture and temperature in tropicaland subtropical regions (Teh et al., 2017; Turetsky et al., 2014).

The presence of cattle in wetlands showed no observable impact onCH4 emissions, soil wetness, or aboveground or belowground biomass.Previously reported methane emissions from grazed and ungrazed

Figure 8. Belowground biomass including coarse and fine roots at twodepth intervals (0–25 cm, 25–50 cm) in the soil profile for wetland type(improved and seminative wetlands). Different lowercase letters denotesignificance between depths in the soil profile, while different uppercaseletters denote significance between wetland types (p < 0.05). Error barsrepresent the standard error of the mean.

Figure 9. Mean soil nitrogen content including NH4, NO3, and IN at twodepths in the soil profile (0–25 cm, 25–50 cm) for each wetland type(improved and seminative wetlands). Capital letters denote significancebetween soil depths (p < 0.05), while lower case letters denote significancebetween wetland type and associated nitrogen species at. Error bars repre-sent the standard error from the mean.

10.1029/2019JG005132Journal of Geophysical Research: Biogeosciences

DELUCIA ET AL. 11

pastures at the same study site show cattle increase soil CH4 emissions and soil wetness while loweringbiomass (Gomez‐Casanovas et al., 2018). The different responses in CH4 emissions between pastures andwetlands, where no grazing responses were observed, can be potentially explained by the methodologyand on the relative impact of grazers on plant species and the microenvironment. Grazing significantlyreduces biomass in pastures, which leads to higher soil moisture (Gomez‐Casanovas et al., 2018). Highsoil moisture is a dominant feature of wetlands independent of grazing pressure. Thus, cattle presence inpastures can lead to higher soil moisture and thus higher CH4 emissions than nongrazed pastures,whereas cattle have little to no impact on soil moisture in wetlands. Differences may also be stronglydriven by the species present in pastures versus wetlands, which are functionally quite dissimilar (Hoet al., 2018). Grazing impacts the species richness and diversity in wetlands and pastures, but the domi-nant species are highly variable. Finally, the conclusions from the pasture research are based on eddycovariance measurements, which will include, to some extent, enteric CH4 emissions and the directimpact of soil disturbance from cattle movements (Gomez‐Casanovas et al., 2018), whereas thechamber‐based measurements exclude these factors.

The challenge of inferring wetland‐scale fluxes from chamber‐based measurements in a replicated field‐based design carries important limitations. As a large field experiment including 16 wetlands, there aresources of variation and confounding factors that should be considered when interpreting the results.First is that the measured fluxes only represent point‐based measurements and may not extrapolate overtime and space. Variation in fluxes was observed based on where measurements were made ranging fromthe shallow to the deep parts of each wetland (e.g., Figure 5c), which is why subsampling at each locationwas performed. However, each water depth is not represented equally within or between wetlands.Furthermore, measurements were collected over time to represent a range of environmental conditionsbut did not encompass the entirety of conditions that occurred over the duration of the experiment.Similar to previous studies focusing on point‐based measurements of key ecosystem fluxes (e.g., Gomez‐Casanovas et al., 2013), this experiment was not intended to pursue temporal integration of CH4 fluxesbut rather assess how point‐based measurements vary among treatments. There are also sources of variationfrom land history that need to be considered. The wetlands used in this experiment were, a priori, selectedbased on similar size, shape, and hydroperiod. Nevertheless, variation in environmental factors drivesexperimental error that can only be accounted through statistical estimates of variance. Such factors mayinclude fire as a dominant feature of the subtropical grassland biome, influence of nonbovine fauna, varia-tion in rainfall or flooding patterns, and a wide range of other factors. The wetlands in this study were notburned throughout the duration of this experiment. However, given the random assignment of the scatteredwetlands into treatments, any factors that impact the fluxes from the wetlands that cannot be directlyaccounted for will add to statistical error estimates, thereby suggesting a more conservative analysis.Finally, the measurement technique of placing a canopy chamber into the wetland may have forced ebulli-tion of methane from soils where the chamber bottom contacts the soil. However, the chambers wereallowed to vent to the atmosphere prior to the initiation of measurements and the technique quantifiesfluxes based on the slope of CH4 accumulation once the chamber is sealed. The fluxes were calculated onlywhen the accumulation of methane over time had an r2 greater than 0.8, which removed conditions whenstepwise changes in CH4, consistent with ebullition, were observed to not occur.

5. Conclusion

Wetlands are known to impact the global C and CH4 budget; however, a scarcity of data exists that quantifyfluxes particularly related to how changes to the surrounding landscape affect emissions. Here, intensivemanagement practices of surrounding land were shown to impact CH4 emissions of embedded wetlands,specifically during the wet season. Unlike subtropical pastures (Gomez‐Casanovas et al., 2018), the resultsshowed that grazers do not, at least indirectly, alter CH4 emissions from subtropical wetlands. At a dailyscale, soil CH4 emission was modulated by changes in soil temperature and moisture, with the largest emis-sions during periods of high soil wetness and temperature. Although the wetland type, agriculture, andmanagement intensity here may carry similarities with wetlands in many tropical and subtropical loca-tions, they are dramatically different from wetlands found in other biomes. Our results show that manage-ment of land can have implications for embedded wetlands; however, whether this is true for all wetland

10.1029/2019JG005132Journal of Geophysical Research: Biogeosciences

DELUCIA ET AL. 12

types and characterizations requires more study. Our experiment measured CH4 fluxes at multiple timepoints for all treatments, with each measurement campaign being measured over a very short time periodto minimize temporal variation. These measurement periods, however, represent only a snapshot in time,which raises a critical question of how these treatments may impose significant differences in CH4 fluxesduring periods of highly episodic CH4 emissions. Practically, these results suggest that the interactionbetween agricultural practices and wetland functioning needs to be represented in biogeochemical models.The lack of representation of this interaction limits ability to predict intensification effects on CH4

source/sink dynamics with agricultural, particularly with intensification, to meet growing agronomicdemands associated with a rising population (FAO et al., 2018). Furthermore, policy decision to protectwetlands, or to minimize their contribution to global CH4 emissions, may need to consider land manage-ment beyond the borders of the wetlands.

ReferencesAltor, A. E., & Mitsch, W. J. (2008). Pulsing hydrology, methane emissions and carbon dioxide fluxes in created marshes: A 2‐year eco-

system study. Wetlands, 28(2), 423–438. https://doi.org/10.1672/07‐98.1Asner, G. P., Elmore, A. J., Olander, L. P., Martin, R. E., & Harris, A. T. (2004). Grazing systems, ecosystem responses, and global change.

Annual Review of Environment and Resources, 29(1), 261–299. https://doi.org/10.1146/annurev.energy.29.062403.102142Banger, K., Tian, H., & Lu, C. (2012). Do nitrogen fertilizers stimulate or inhibit methane emissions from rice fields? Global Change Biology,

18(10), 3259–3267. https://doi.org/10.1111/j.1365‐2486.2012.02762.xBhullar, G. S., Edwards, P. J., & Venterink, H. O. (2013). Variation in the plant‐mediated methane transport and its importance for methane

emission from intact wetland peat mesocosms. Journal of Plant Ecology, 6(4), 298–304. https://doi.org/10.1093/jpe/rts045Blanc‐Betes, E., Welker, J. M., Sturchio, N. C., Chanton, J. P., & Gonzalez‐Meler, M. A. (2016). Winter precipitation and snow accumulation

drive the methane sink or source strength of Arctic tussock tundra. Global Change Biology, 22(8), 2818–2833. https://doi.org/10.1111/gcb.13242

Bohlen, P. J., & Gathumbi, S. M. (2007). Nitrogen cycling in seasonal wetlands in subtropical cattle pastures. Soil Science Society of AmericaJournal, 71(3), 1058. https://doi.org/10.2136/sssaj2005.00217

Boon, P. I., & Mitchell, A. (1995). Methanogenesis in the sediments of an Australian freshwater wetland: Comparison with aerobicdecay, and factors controlling methanogenesis. FEMS Microbiology Ecology, 18(3), 175–190. https://doi.org/10.1016/0168‐6496(95)00053‐5

Boughton, E. H., Quintana‐ascencio, P. F., & Bohlen, P. J. (2011). Refuge effects of Juncus effusus in grazed, subtropical wetland plantcommunities. Plant Ecology, 212(3), 451–460. https://doi.org/10.1007/s11258‐010‐9836‐4

Boughton, E. H., Quintana‐Ascencio, P. F., Bohlen, P. J., Fauth, J. E., & Jenkins, D. G. (2016). Interactive effects of pasture managementintensity, release from grazing and prescribed fire on forty subtropical wetland plant assemblages. Journal of Applied Ecology, 53(1),159–170. https://doi.org/10.1111/1365‐2664.12536

Boughton, E. H., Quintana‐Ascencio, P. F., Nickerson, D., & Bohlen, P. J. (2011). Management intensity affects the relationship betweennon‐native and native species in subtropical wetlands. Applied Vegetation Science, 14(2), 210–220. https://doi.org/10.1111/j.1654‐109X.2010.01116.x

Bridgham, S. D., Cadillo‐Quiroz, H., Keller, J. K., & Zhuang, Q. (2013). Methane emissions from wetlands: Biogeochemical, microbial, andmodeling perspectives from local to global scales. Global Change Biology, 19(5), 1325–1346. https://doi.org/10.1111/gcb.12131

Chamberlain, S. D., Boughton, E. H., & Sparks, J. P. (2015). Underlying ecosystem emissions exceed cattle‐emitted methane from sub-tropical lowland pastures. Ecosystems, 18(6), 933–945. https://doi.org/10.1007/s10021‐015‐9873‐x

Chamberlain, S. D., Gomez‐Casanovas, N., Walter, M. T., Boughton, E. H., Bernacchi, C. J., Delucia, E. H., et al. (2016). Influence oftransient flooding on methane fluxes from subtropical pastures. Journal of Geophysical Research: Biogeosciences, 121, 965–977.https://doi.org/10.1002/2015JG003283

Ciais, P., Sabine, C., Bala, G., Bopp, L., Brovkin, V., Canadell, J., et al. (2013). Carbon and other biogeochemical cycles. In T. F. Stocker,D. Qin, G.‐K. Plattner, M. Tignor, S. K. Allen, J. Boschung, A. Nauels, Y. Xia, V. Bex, & P. M. Midgley (Eds.), Climate change 2013: Thephysical science basis. Contribution of Working Group I to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change(Chap. 6, pp. 465–570). Cambridge, UK and New York: Cambridge University Press.

Collins, S. D., Heintzman, L. J., Starr, S. M., Wright, C. K., Henebry, G. M., & McIntyre, N. E. (2014). Hydrological dynamics of temporarywetlands in the southern Great Plains as a function of surrounding land use. Journal of Arid Environments, 109, 6–14. https://doi.org/10.1016/j.jaridenv.2014.05.006

Conrad, R. (1989). Control of methane production in terrestrial ecosystems. In M. Andreae & D. Schimel (Eds.), Exchange of trace gasesbetween terrestrial ecosystems and the atmosphere (pp. 39–58). Chichester: Wiley.

Dalal, R. C., Allen, D. E., Livesley, S. J., & Richards, G. (2008). Magnitude and biophysical regulators of methane emission and consumptionin the Australian agricultural, forest, and submerged landscapes: A review. Plant and Soil, 309(1‐2), 43–76. https://doi.org/10.1007/s11104‐007‐9446‐7

Delucia, E. H., Gomez‐Casanovas, N., Greenberg, J. A., Hudiburg, T. W., Kantola, I. B., Long, S. P., et al. (2014). The theoretical limit toplant productivity. Environmental Science and Technology, 48(16), 9471–9477. https://doi.org/10.1021/es502348e

Ding, W., Cai, Z., & Tsuruta, H. (2005). Plant species effects on methane emissions from freshwater marshes. Atmospheric Environment,39(18), 3199–3207. https://doi.org/10.1016/j.atmosenv.2005.02.022

Dong, H., Mangino, J., McAllister, T. A., Hatfield, J. L., Johnson, D. E., Lassey, K. R., et al. (2006). Emissions from livestock and manuremanagement. IPPC Guidelines for National Greenhous Gas Inventories, Chapter 10.

Dunfield, P., Knowles, R., Dumont, R., & Moore, T. (1993). Methane production and consumption in temperate and subarctic peat soils:Response to temperature and pH. Soil Biology and Biochemistry, 25(3), 321–326. https://doi.org/10.1016/0038‐0717(93)90130‐4

Environmental Protection Agency (EPA). (n.d.). Retrieved January 22, 2019, from Citation Online Website: https://www.epa.gov/wetlands/what‐wetland

10.1029/2019JG005132Journal of Geophysical Research: Biogeosciences

DELUCIA ET AL. 13

AcknowledgmentsAuthors are grateful to Julia Maki, EarlKeel, Melissa Wasson, Raoul Boughton,Stefan Pierre, and Keith Brinsko fortheir assistance in makingmeasurements and the rest of the staffat Buck Island Ranch for site access,lodging, transportation, and continuedsupport in the field. This research wasfunded by the U.S. Department ofAgriculture NIFA (Project 2016‐67019‐24988) and the Global Change andPhotosynthesis Research Unit of the U.S. Department of Agriculture—Agricultural Research Service. Datafrom Figure 1 are contained in theimage, the source of Figure 2 data isindicated in the figure legend, and alldata from Figures 3 are available asonline supporting information. Anyopinions, findings, and conclusions orrecommendations expressed in thispublication are those of the author(s)and do not necessarily reflect the viewsof the U.S. Department of Agriculture.Mention of trade names or commercialproducts in this publication is solely forthe purpose of providing specificinformation and does not implyrecommendation or endorsement bythe U.S. Department of Agriculture.USDA is an equal opportunity providerand employer.

Ervin, G. N., & Wetzel, R. G. (2002). Influence of a dominant macrophyte, Juncus effusus, on wetland plant species richness, diversity, andcommunity composition. Oecologia, 130(4), 626–636. https://doi.org/10.1007/s00442‐001‐0844‐x

FAO, IFAD, UNICEF,WFP,W. (2018). Food security and nutrition in the world, building climate resilience for food security and nutrition.Retrieved from www.fao.org/publications

Forster, P., Ramaswamy, V., Artaxo, P., & Berntsen, T. (2007). Changes in atmospheric constituents and in radiative forcing. Change,77(22), 129–234. https://doi.org/10.1103/PhysRevB.77.220407

Gathumbi, S. M., Bohlen, P. J., & Graetz, D. A. (2005). Nutrient enrichment of wetland vegetation and sediments in subtropical pastures.Soil Science Society of America Journal, 69(2), 539. https://doi.org/10.2136/sssaj2005.0539

Gomez‐Casanovas, N., Anderson‐Teixeira, K., Zeri, M., Bernacchi, C. J., & DeLucia, E. H. (2013). Gap filling strategies and error in esti-mating annual soil respiration. Global Change Biology, 19(6), 1941–1952. https://doi.org/10.1111/gcb.12127

Gomez‐Casanovas, N., Delucia, N. J., Bernacchi, C. J., Boughton, E. H., Sparks, J. P., Chamberlain, S. D., & Delucia, E. H. (2018). Grazingalters net ecosystem C fluxes and the global warming potential of a subtropical pasture. Ecological Applications, 28(2), 557–572.https://doi.org/10.1002/eap.1670

Henneberg, A., Brix, H., & Sorrell, B. K. (2015). The interactive effect of Juncus effusus and water table position on mesocosm methano-genesis and methane emissions. Plant Soil, 400(1–2), 45–54. https://doi.org/10.1007/s11104‐015‐2707‐y

Henneberg, A., Elsgaard, L., Sorrell, B. K., Brix, H., & Petersen, S. O. (2015). Does Juncus effusus enhance methane emissions from grazedpastures on peat? Biogeosciences Discussions, 12(11), 8467–8495. https://doi.org/10.5194/bgd‐12‐8467‐2015

Ho, J., Boughton, E. H., Jenkins, D. G., Sonnier, G., Bohlen, P. J., & Chambers, L. G. (2018). Ranching practices interactively affect soilnutrients in subtropical wetlands. Agriculture, Ecosystems and Environment, 254(2017), 130–137. https://doi.org/10.1016/j.agee.2017.11.031

Hutchinson, G. L., & Livingston, G. P. (1993).Use of chamber systems to measure trace gas fluxes. Agricultural ecosystem effects on trace gasesand global climate change 55, American Society of Agronomy, Crop Science Society of America, and Soil Science Society of America(pp. 63–78). https://doi.org/10.2134/asaspecpub55.c4

Inglett, K. S., Inglett, P. W., Reddy, K. R., & Osborne, T. Z. (2012). Temperature sensitivity of greenhouse gas production in wetland soils ofdifferent vegetation. Biogeochemistry, 108(1–3), 77–90. https://doi.org/10.1007/s10533‐011‐9573‐3

Intergovernmental Panel on Climate Change (2014). Summary for policymakers. Climate Change 2014: Synthesis Report. Contribution ofWorking Groups I, II and III to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change. https://doi.org/10.1017/CBO9781107415324

Kayranli, B., Scholz, M., Mustafa, A., & Hedmark, Å. (2010). Carbon storage and fluxes within freshwater wetlands: A critical review.Wetlands, 30(1), 111–124. https://doi.org/10.1007/s13157‐009‐0003‐4

Keeney, D. R., & Nelson, D. W. (1982). Methods of soil analysis: Part 2. Chemical and Microbiological Properties, American Society ofAgronomy, Method of Soil Analysis (pp. 648–649). https://doi.org/10.2134/agronmonogr9.2.2ed.c9

King, J. Y., Reeburgh, W. S., & Regli, S. K. (1998). Methane emission and transport by arctic sedges in Alaska: Results of a vegetationremoval experiment. Journal of Geophysical Research, 103(D22), 29,083–29,092. https://doi.org/10.1029/98JD00052

Marín‐Munez, J. L., Hernandez, M., & Moreno‐Casasola, P. (2015). Greenhouse gas emissions from coastal freshwater wetlands inVeracruz Mexico: Effect of plant community and seasonal dynamics. Atmospheric Environment, 107(26), 107–117. https://doi.org/10.1016/j.atmosenv.2015.02.036

Martikainen, P. J., Nykanen, H., Alm, J., & Silvola, J. (1995). Change in fluxes of carbon‐dioxide, methane and nitrous‐oxide due to forestdrainage of mire sites of different trophy. Plant and Soil, 168‐169(1), 571–577. https://doi.org/10.1007/bf00029370

Megonigal, J. P., Hines, M. E., & Visscher, P. T. (2004). Anaerobic metabolism: Linkages to trace gases and aerobic processes. Biogeochemistry,Smithsonian Environmental Research Center (pp. 317–424).

Mitsch, W. J., & Gosselink, J. G. (2015). Wetlands (5th ed., p. 744). Hoboken, NJ: John Wiley & Sons, Inc.Nahlik, A. M., & Mitsch, W. J. (2011). Methane emissions from tropical freshwater wetlands located in different climatic zones of Costa

Rica. Global Change Biology, 17(3), 1321–1334. https://doi.org/10.1111/j.1365‐2486.2010.02190.xNicolini, G., Castaldi, S., Fratini, G., & Valentini, R. (2013). A literature overview of micrometeorological CH4 and N2O flux measurements

in terrestrial ecosystems. Atmospheric Environment, 81, 311–319. https://doi.org/10.1016/j.atmosenv.2013.09.030Okazaki, M., Hadi, A., Inubushi, K., & Yonebayashi, K. (1998). Effect of converting wetland forest to sage palm plantations on methane gas

flux and organic carbon dynamics in tropical peat soil. Hydrological Processes, 12(13–14), 2073–2080.Ortiz‐Llorente, M. J., & Alvarez‐Cobelas, M. (2012). Comparison of biogenic methane emissions from unmanaged estuaries, lakes, oceans,

rivers and wetlands. Atmospheric Environment, 59, 328–337. https://doi.org/10.1016/j.atmosenv.2012.05.031Petersen, S. O., Hoffmann, C. C., Schäfer, C. M., Blicher‐Mathiesen, G., Elsgaard, L., Kristensen, K., & Greve, M. H. (2012). Annual

emissions of CH4 and N2O, and ecosystem respiration, from eight organic soils in Western Denmark managed by agriculture.Biogeosciences, 9(1), 403–422. https://doi.org/10.5194/bg‐9‐403‐2012

Petrescu, A., Lohila, A., & Tuovinen, J. (2015). The uncertain climate footprint of wetlands under human pressure. Proceedings of theNational Academy of Sciences, 122(15), 4594–4599. http://www.pnas.org/content/112/15/4594.short, https://doi.org/10.1073/pnas.1416267112

Ramankutty, N., Evan, A. T., Monfreda, C., & Foley, J. A. (2008). Farming the planet: 1. Geographic distribution of global agricultural landsin the year 2000. Global Biogeochemical Cycles, 22, GB1003. https://doi.org/10.1029/2007GB002952

Ramsar Convention Secretariat (2016). Ramsar convention on wetlands: Scope and implementation. The Ramsar Convention manual(pp. 1–9). The Wetland Book. https://doi.org/10.1007/978‐94‐007‐8172‐8

Rietl, A. J., Nyman, J. A., Lindau, C. W., & Jackson, C. R. (2017). Wetland methane emissions altered by vegetation disturbance: Aninteraction between stem clipping and nutrient enrichment. Aquatic Botany, 136, 205–211. https://doi.org/10.1016/j.aquabot.2016.10.008

RStudio Team (2015). RStudio: Integrated development for R. RStudio, Inc., Boston, MA. http://www.rstudio.com/Saunois, M., Bousquet, P., Poulter, B., Peregon, A., Ciais, P., Canadell, J. G., et al. (2016). The global methane budget 2000–2012. Earth

System Science Data, 8(2), 697–751. https://doi.org/10.5194/essd‐8‐697‐2016Ström, L., Mastepanov, M., & Christensen, T. R. (2015). Species‐specific effects of vascular plants on carbon turnover and methane emis-

sions from wetlands. Biogeochemistry, 75(1), 65–82.Swain, H. M., Boughton, E. H., Bohlen, P. J., & Lollis, L. O. (2013). Trade‐offs among ecosystem services and disservices on a Florida Ranch.

Rangelands, 35(5), 75–87. https://doi.org/10.2111/RANGELANDS‐D‐13‐00053.1Tanentzap, A. J., & Coomes, D. A. (2012). Carbon storage in terrestrial ecosystems: Do browsing and grazing herbivores matter? Biological

Reviews, 87(1), 72–94. https://doi.org/10.1111/j.1469‐185X.2011.00185.x

10.1029/2019JG005132Journal of Geophysical Research: Biogeosciences

DELUCIA ET AL. 14

Teh, Y. A., Murphy, W. A., Berrio, J., Boom, A., & Page, S. E. (2017). Seasonal variability in methane and nitrous oxide fluxes from tropicalpeatlands in the western Amazon basin. Biogeosciences, 14(15), 3669–3683. https://doi.org/10.5194/bg‐14‐3669‐2017

Torn, M. S., & Chapin, F. S. (1993). Environmental and biotic controls over methane flux from Arctic tundra. Chemosphere, 26(1–4),357–368. https://doi.org/10.1016/0045‐6535(93)90431‐4

Turetsky, M. R., Kotowska, A., Bubier, J., Dise, N. B., Crill, P., Hornibrook, E. R. C., et al. (2014). A synthesis of methane emissions from 71northern, temperate, and subtropical wetlands. Global Change Biology, 20(7), 2183–2197. https://doi.org/10.1111/gcb.12580

Tweel, A. W., & Bohlen, P. J. (2008). Influence of soft rush (Juncus effusus) on phosphorus flux in grazed seasonal wetlands. EcologicalEngineering, 33(3‐4), 242–251. https://doi.org/10.1016/j.ecoleng.2008.05.003

U.S. Department of Agriculture (2009). Rangeland in the Sunshine State? Fact Sheet. NRCS.Verhoeven, J. T. A., & Setter, T. L. (2010). Agricultural use of wetlands: Opportunities and limitations. Annals of Botany, 105(1), 155–163.

https://doi.org/10.1093/aob/mcp172Wang, K. H., McSorley, R., Bohlen, P., & Gathumbi, S. M. (2006). Cattle grazing increases microbial biomass and alters soil nematode

communities in subtropical pastures. Soil Biology and Biochemistry, 38(7), 1956–1965. https://doi.org/10.1016/j.soilbio.2005.12.019Whiting, G., Chanton, J. P., Bartlett, D. S., & Happell, J. D. (1991). Relationships between CH4 emissions, biomass, and CO2 exchange in a

subtropical Grassland. Journal of Geophysical Research, 96(D7), 13,067. https://doi.org/10.1029/91JD01248Whiting, G. J., & Chanton, P. (1993). Primary production control of methane emission from wetlands. Nature, 364(6440), 794–795.

https://doi.org/10.1038/364794a0Yang, P., Bastviken, D., Lai, D. Y. F., Jin, B. S., Mou, X. J., Tong, C., & Yao, Y. C. (2017). Effects of coastal marsh conversion to shrimp

aquaculture ponds on CH4 and N2O emissions. Estuarine, Coastal and Shelf Science, 199, 125–131. https://doi.org/10.1016/j.ecss.2017.09.023

10.1029/2019JG005132Journal of Geophysical Research: Biogeosciences

DELUCIA ET AL. 15