Environmental controls on soil respiration across a southern US climate gradient: a meta-analysis

    Environmental controls on soil respiration across a southern US climate gradient: a meta-analysis Kristofor R. Brye, Richard L. McMullen, Maria L. Silveira, Jill M.D. Motschenbacher, S. Faye Smith, Edward E. Gbur, Michele L. Helton PII: DOI: Reference:

S2352-0094(16)30010-4 doi: 10.1016/j.geodrs.2016.02.005 GEODRS 79

To appear in: Received date: Revised date: Accepted date:

2 January 2016 9 February 2016 10 February 2016

Please cite this article as: Brye, Kristofor R., McMullen, Richard L., Silveira, Maria L., Motschenbacher, Jill M.D., Smith, S. Faye, Gbur, Edward E., Helton, Michele L., Environmental controls on soil respiration across a southern US climate gradient: a meta-analysis, (2016), doi: 10.1016/j.geodrs.2016.02.005

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

ACCEPTED MANUSCRIPT Environmental controls on soil respiration across a southern US climate gradient: a metaanalysis

PT

Kristofor R. Bryea*, Richard L. McMullena, Maria L. Silveirab, Jill M.D. Motschenbacherc,

a

SC

RI

S. Faye Smitha, Edward E. Gburd, and Michele L. Heltona

Department of Crop, Soil, and Environmental Sciences, University of Arkansas, Fayetteville, AR

NU

72701, USA; [email protected], [email protected], [email protected],

b

MA

[email protected]

Soil and Water Science, University of Florida, Institute of Food and Agricultural Sciences Range

Department of Soil Science, North Dakota State University, Fargo, ND 58108, USA;

TE

c

D

Cattle Research and Education Center, Ona, FL 33865, USA; [email protected]

[email protected]

Agricultural Statistics Laboratory, University of Arkansas, Fayetteville, AR 72701, USA;

AC CE P

d

[email protected]

* Corresponding Author: K.R. Brye Department of Crop, Soil, and Environmental Sciences 115 Plant Sciences Building University of Arkansas Fayetteville, AR, 72701, USA Email: [email protected] Phone: (479) 575-5742

ACCEPTED MANUSCRIPT Environmental controls on soil respiration across a southern US climate gradient: a metaanalysis ABSTRACT

PT

Carbon (C) cycling in the soil is intimately associated with soil respiration as organic matter is

RI

decomposed by microbes. Consequently, soil C stocks and sequestration potential are also intimately

SC

associated with soil respiration. Although many soils of the southern and southeastern United States (US) are generally more weathered and contain less C than those of the upper Midwest due to moist,

NU

and generally warmer climatic conditions, their soil C sequestration potential may be greater due to

MA

their ability to produce biomass year-round, which, in turn, results in greater C inputs. Identifying influential environmental factors that control soil respiration across a large geographic area and

D

climate gradient can improve understanding of soil C sequestration potential in the southern US.

TE

The objectives of this study were to evaluate the effects of i) physiographic region (i.e., Arkansas Ozark Highlands, Arkansas Delta, and Florida Flatwoods) and ii) soil moisture regime (i.e., udic and

AC CE P

aquic) on the relationship among soil respiration and combined soil moisture and soil temperature related environmental parameters. Despite some expected differences and generally low model predictiveness (R2 < 0.4), results showed numerous similarities among multiple regression model coefficient estimates across widely differing physiographic regions along a southern climate gradient. Results also showed the relationship among soil respiration and soil moisture and soil temperature related environmental parameters differed (P < 0.05) between soil moisture regimes within regions. Improving the ability to predict soil respiration from directly measured and/or indirectly calculated environmental parameters will increase the understanding of factors controlling soil C sequestration, and potential agronomic and ecological sustainability, in the weathered soils of the southern and southeastern US. Keywords: soil respiration; soil carbon; soil moisture; soil temperature; Luvisols; Podzols

ACCEPTED MANUSCRIPT 1. Introduction Integrating both autotrophic and heterotrophic processes, soil respiration is responsible for the greatest flux of carbon (C) from terrestrial ecosystems to the atmosphere (Bond-Lamberty and

PT

Thomson, 2010; Vicca et al., 2014). Consequently, soil respiration is a key component of global C

RI

cycling and the characterization of the partitioning of C among the atmosphere, biosphere, and

SC

pedosphere is vital to furthering our understanding of the constraints on the global C budget. It is well-established that climate is a main factor controlling soil organic matter (SOM)

NU

decomposition and C cycling; thus climate differences contribute to differences in C stocks and sequestration potential among soils. The primary climatic or environmental factors influencing soil

MA

respiration are moisture (Brown et al., 2009; Gaumont-Guay et al., 2006; Davidson et al., 1998;

D

Wagai et al., 1998) and temperature (Reth et al., 2009; Brye et al., 2006b; Fierer et al., 2006; Fang

TE

and Moncrieff, 2001; Davidson et al., 1998; Lloyd and Taylor, 1994). Soil moisture often displays a quadratic relationship with soil respiration, whereas soil respiration commonly increases to an

AC CE P

optimum soil moisture level, typically around 50 to 60 % water-filled pore space, then decreases thereafter as the soil is too wet for optimal microorganism activity due to oxygen limitations (Taggart et al., 2012; Brye and Riley, 2009; Gaumont-Guay et al., 2006; Parton et al., 1993; Linn and Doran, 1984). In contrast, soil respiration typically increases exponentially as soil temperature increases (Wagai et al., 1998; Lloyd and Taylor, 1994; Parton et al., 1993) to a point before the temperature is too warm and microbial enzymatic functions begin to breakdown. Despite our general understanding of the effects of environmental factors on soil respiration, relationships among soil respiration and soil moisture and temperature independently have been quite variable and inconsistent (Lloyd and Taylor, 1994). Soil respiration has been reported to be positively correlated (Pingintha et al., 2010; Brown et al., 2009), negatively correlated (Brye et al., 2006b; Jones et al., 2006), and uncorrelated (McMullen et al., 2014; Ding et al., 2010; Brye et al., 2006a; Al-Kaisi and Yin, 2005) to soil moisture. Though soil respiration has generally been

ACCEPTED MANUSCRIPT reported to be positively correlated to soil temperature (McMullen et al., 2014; Ding et al., 2010; Ruehr et al., 2010; Brown et al., 2009; Brye et al., 2006a; Jones et al., 2006; Fang and Moncrieff, 2001), Davidson et al. (1998) suggested that soil moisture and temperature may act independently or

PT

be confounding, interdependent factors controlling soil respiration. In addition, soil respiration and

RI

soil temperature have demonstrated diel hysteresis (Gaumont-Guay et al., 2006; Oikawa et al.,

SC

2014), which further complicates the relationship between soil respiration and soil temperature. Consequently, the clear variations in field-observed relationships among soil respiration and soil

NU

moisture and temperature has led to uncertainty in the prediction of soil respiration beyond site-

MA

specific, local scales.

In the southern United States (US), where temperatures are greater and seasonal temperature

D

variations are generally less-pronounced than in the upper Midwest, soil respiration rates are also

TE

typically greater and elevated fluxes persist for a longer duration throughout the year (Motschenbacher et al., 2015; McMullen et al., 2014; Smith et al., 2014; Brye and Riley, 2009)

AC CE P

compared to the upper Midwest (Brye et al., 2002; Wagai et al., 1998). Consequently, temperature and moisture gradients throughout the US have contributed to southern soils often having lower SOM contents relative to northern soils. Although differences in soil properties, land use, soil management, and climate conditions influence C sequestration and cycling (Guo and Gifford, 2002; VandenBygaart et al., 2002, 2003; Lal, 2004; Bernsten et al., 2006; Davidson and Janssens, 2006; Laganière et al., 2010), VandenBygaart et al. (2002, 2003) suggested that the C sequestration potential was generally greater for soils with low SOM contents than soils containing greater C stocks. Consequently, highly weathered soils, such as the soils that are widespread throughout the southern US (i.e., Ultisols) and those with a long history of cultivated agriculture, particularly those of the Lower Mississippi River Delta region and other areas of intense agriculture throughout the southern US, may have a greater potential for further soil C sequestration than comparatively less weathered, upper-Midwestern soils (i.e., Alfisols and Mollisols) due to generally lower SOM

ACCEPTED MANUSCRIPT contents. In an effort to characterize the relationship among soil respiration and combined soil moisture and temperature in broiler-litter-amended, pasture soil in the Ozark Highlands of the

PT

southern US, McMullen et al. (2014) used multiple regression techniques to determine broiler litter

RI

rate effects on soil respiration using soil moisture and temperature measurements made concurrently

SC

with respiration measurements. It was concluded that a single, multiple regression model using combined environmental factors could predict soil respiration regardless of broiler litter rate for a

NU

managed grassland on a highly weathered Ultisol in the Ozark Highlands (McMullen et al., 2014).

MA

Similar approaches have also been used to evaluate residue and water management effects (Smith, 2013) and different bioenergy crop effects (Helton, 2014) on soil moisture-temperature-respiration

D

relationships in the Lower Mississippi River Delta region of eastern Arkansas.

TE

As long-term sustainability and improving soil health continue to be at the forefront of environmental and agricultural issues, understanding the complex interactions among environmental

AC CE P

factors and soil respiration becomes even more important in regions, such as the southern and southeastern US, where SOM contents are generally low due to the combination of moist and warm climatic conditions and historic annual cultivation for crop production. If regional, rather than geographically isolated, relationships among soil respiration, moisture, and temperature exist, or can be developed across broad climatic conditions, then soil management practices can be further refined to minimize C losses to the atmosphere. Furthermore, Vicca et al. (2014), who conducted a metaanalysis on the influence of altered precipitation patterns on soil respiration, recognized the need to evaluate relationships and establish functions across a broad range of soil moisture conditions. To our knowledge, there have been no meta-analyses conducted evaluating the relationships among soil respiration and combined soil moisture and temperature related environmental parameters over broad geographic/topographic and/or climatic gradients for soils of the southern and southeastern US. Therefore, the objectives of this study were to evaluate the effects of i)

ACCEPTED MANUSCRIPT physiographic region (i.e., Arkansas Ozarks, Arkansas Delta, and Florida Flatwoods) and ii) soil moisture regime (i.e., udic and aquic) on the relationships among soil respiration and combined soil moisture and soil temperature related environmental parameters across a climate gradient in the

PT

southern-southeastern US. It was hypothesized that the relationship among soil respiration and

RI

combined soil moisture and temperature differs greatly among physiographic regions representing a

SC

climate gradient. It was also hypothesized that the relationship among soil respiration and combined

NU

soil moisture and temperature would differ greatly between soil moisture regimes.

MA

2. Materials and methods

Over an approximate 11.5-year period between May 2002 and December 2013, eight

D

independent studies were conducted at a variety of locations, for a variety of durations, and included

TE

a variety of land uses and site-specific management practices across a climate gradient from the northwest Arkansas to east-central Arkansas to south-central Florida (Table 1; Fig. 1). All eight field

AC CE P

studies generated similar datasets that included simultaneous measurements of soil respiration, soil temperature, and soil moisture periodically over time. A total of 4511 soil respiration-temperaturemoisture observations were generated among these eight studies. These observations have been assembled into a single dataset to address and test the above-stated objective and hypotheses using a meta-analysis approach.

2.1. Site descriptions 2.1.1. Arkansas – Ozark Highlands Data from two field studies conducted in the Ozark Highlands of northwest Arkansas were included in this meta-analysis. One study, providing 326 observations for the meta-analysis, was conducted during the 2006 growing season on loam and silt-loam Alfisols and an Ultisol in a native tallgrass prairie in Rogers, AR and in a chronosequence of four tallgrass prairie restorations, ranging

ACCEPTED MANUSCRIPT in age at the time from 3- to 26-years old, at the Pea Ridge National Military Park near Garfield, AR (Brye and Riley, 2009). The second study, providing 594 observations, was conducted at the Arkansas Agricultural Research and Extension Center in Fayetteville, AR for three consecutive

PT

years between May 2009 and May 2012 in small plots of a managed grassland on a silt-loam Ultisol

RI

that received annual application rates of 0, 5.6, or 11.2 Mg ha-1 of non-pelletized broiler litter from a

SC

local source (McMullen et al., 2014). All three sites included in these two field studies conducted in northwest Arkansas were located in Major Land Resource Area (MLRA) 116A, the Ozark Highlands

NU

(Brye et al., 2013). Of the 920 total observations from northwest Arkansas, 198 measurements were

MA

conducted on soils with an aquic soil moisture regime, whereas the remaining 722 measurements were conducted on soils with a udic soil moisture regime. The 30-yr (1981-2010) mean annual

D

precipitation and air temperature range from 115.6 to 119.2 cm and from 14.5 to 14.6oC,

AC CE P

2.1.2. Arkansas – Delta

TE

respectively, throughout the region encompassing the Ozark Highlands field sites (NOAA, 2015).

Data from five field studies conducted in east-central Arkansas were included in this metaanalysis. One study, providing 576 observations, was conducted on silt-loam Alfisols at two locations (i.e., the Pine Tree Research Station near Colt, AR and the Cotton Research Station in Marianna, AR) in MLRA 134, the Southern Mississippi Valley Loess (Brye et al., 2013), over the course of two consecutive summer growing seasons (i.e., 2002 and 2003) in a wheat (Triticum aestivum L.)-soybean (Glycine max L.), double-crop system with a variety of residue management practices (i.e., high and low residue level, residue burning and non-burning, and conventional tillage and no-tillage) (Brye et al., 2006b). The second study, providing 480 observations, was conducted in 2004 on silt-loam Alfisols at two locations (i.e., the Rice Research and Extension Center near Stuttgart, AR and the Pine Tree Research Station near Colt, AR) in MLRA 131D, the Southern Mississippi River Terraces, and in MLRA 134, respectively (Brye et al., 2013). Both locations

ACCEPTED MANUSCRIPT received fresh or pelletized poultry litter amendments during the pre-flooded period of rice (Oryza sativa L.) grown in a rice-soybean rotation (Brye et al., 2006a). The third study, providing 600 observations, was conducted in 2009 and 2010 on a silt-loam Alfisol at the Rice Research and

PT

Extension Center near Stuttgart, AR (MLRA 131D) during the non-flooded period in a series of 10

RI

different rice rotations, including continuous rice and rice grown in various rotations with corn (Zea

SC

mays L.), soybean, and/or winter wheat, under optimal fertilization and managed using either conventional tillage or no-tillage (Motschenbacher et al., 2015). The fourth study, providing 960

NU

observations, was conducted over the course of two consecutive soybean growing seasons (2011 and

MA

2012) on a silt-loam Alfisol at the Cotton Research Station near Marianna, AR (MLRA 134) in a wheat-soybean, double-crop system with a variety of residue management practices (i.e., high and

D

low residue level, residue burning and non-burning, and conventional tillage and no-tillage) and

TE

under irrigated and dryland conditions (Smith et al., 2014). The Smith et al. (2014) study was a repeat of the Brye et al. (2006b) study conducted in 2002 and 2003, but with the addition of an

AC CE P

irrigated and dryland soybean production field treatment factor. The fifth study, providing 903 observations, was conducted over the course of two consecutive calendar years (2012 and 2013) on silt-loam Alfisols at the Pine Tree Research Station near Colt, AR (MLRA 134) in three bioenergyrelated agroecosystems: replicated 3-year-old monoculture switchgrass (Panicum virgatum) plots, replicated 3-year-old eastern cottonwood (Populus deltoides) plots, and replicated plots of a soybean-grain sorghum (Sorghum bicolor) crop rotation (Helton, 2014). Of the 3519 total observations from the Mississippi River Delta region of east-central Arkansas, 2271 measurements were conducted on soils with an aquic soil moisture regime, whereas the remaining 1248 measurements were conducted on soils with a udic soil moisture regime. The 30-yr (1981-2010) mean annual precipitation and air temperature range from 123.0 to 128.4 cm and from 16.0 to 16.6oC, respectively, throughout the region encompassing the Arkansas Delta field sites (NOAA, 2015).

ACCEPTED MANUSCRIPT

2.1.3. Florida Data from one field study conducted in MLRA 155, the Southern Florida Flatwoods, of

PT

south-central Florida were included in this meta-analysis. This study, providing 72 observations,

RI

was conducted at the Range Cattle Research and Education Center in Ona, FL between January and

SC

mid-August 2013 on a sandy Spodosol in low- (native rangeland), medium- (silvopasture), and highintensity (sown pasture) managed grasslands (Adewopo et al., 2015). All Florida measurements

NU

were conducted on a soil with an aquic soil regime. The 30-yr (1981-2010) mean annual

MA

precipitation and air temperature for the Florida field site are 130.0 cm and 23.2oC, respectively (NOAA, 2015). Table 1 summarizes the general characteristics of each of the field projects

TE

D

conducted.

2.2. Soil respiration, moisture, and temperature measurements

AC CE P

For all observations in Arkansas, the protocol for soil respiration measurements in a given plot was the same. At least one day prior to conducting measurements, a 10-cm diameter polyvinyl chloride (PVC) collar ~ 7.5 cm tall was either initially installed or moved to a new location within a plot and inserted approximately 2 cm into the soil. Immediately prior to conducting measurements, any green vegetation within the collars was clipped and removed so that only soil respiration was measured without any potential photosynthesis occurring. Soil respiration measurements were made using a LI-6400 infrared gas analyzer and a LI-6400-09 soil respiration chamber (Li-Cor Inc., Lincoln, NE). Collars were moved to a different location within a plot after roughly two consecutive measurement dates at the same location. Soil respiration measurements were generally conducted between 0800 and 1200 hours central daylight time. In addition to each soil respiration measurement, soil moisture in the top 6 cm was measured using a Theta Probe (Model TH20, Dynamax, Inc., Houston, TX) and soil temperature at the 10-cm depth using a pencil-type

ACCEPTED MANUSCRIPT thermometer. The protocol for soil respiration measurements in a given plot for the Florida field study was generally similar to that conducted for the Arkansas field studies, but with a few minor differences.

PT

Soil respiration was measured approximately weekly using an environmental gas monitor (EGM-2)

RI

portable infra-red gas analyzer (PP Systems, Amesbury, MA) equipped with a soil respiration

SC

chamber (SRC-1). Before each weekly measurement, 10-cm diameter by 2.5-cm tall PVC collars were installed for the soil respiration chamber to rest on while conducting measurements.

NU

Vegetation inside the collars was clipped and removed to avoid influence from shoot respiration.

MA

Immediately following each respiration measurement, soil moisture in the top 10 cm (VG-METER200 soil moisture meter, Vegetronix Inc., Riverton, UT) and soil temperature at the 10-cm depth (HI

D

98331 temperature probe, Hanna Instruments, Carrolton, TX) adjacent to the respiration collars were

AC CE P

2.3. Statistical analyses

TE

also measured.

The entire dataset, consisting of 4511 observations, was used to address the effects of physiographic region on the relationship among soil respiration and combined soil moisture and temperature (i.e., Objective 1). To address the effects of physiographic region and soil moisture regime on the relationship among soil respiration and combined soil moisture and temperature (i.e., Objective 2), only measurements from Arkansas were used (i.e., 4439 observations). Since the Florida field study did not have soil respiration measurements performed in a similar soil under udic soil moisture conditions, the Florida dataset was excluded from the analysis addressing potential regional and/or soil moisture regime differences in the relationship among soil respiration and combined soil moisture and temperature. For both objectives, linear correlation analyses were initially performed between soil respiration and the linear and quadratic terms for soil moisture and temperature and the product of

ACCEPTED MANUSCRIPT soil moisture and temperature separately among the three regional datasets and for the combined dataset. Subsequently, following procedures of McMullen et al. (2014), a complete-quadratic, multiple regression model was fit separately and compared among the single combined dataset and

PT

the three regional datasets. The complete-quadratic, full multiple regression model was represented

SC

RI

by the following equation:

[1]

NU

Rs = a(T10cm) + b(VWC) + c((T10cm)2) + d((VWC)2) + e((T10cm)(VWC)) + f

MA

, where Rs was soil respiration (μmol CO2 m-2 s-1), T10cm was soil temperature (oC) at the 10-cm depth mark, and VWC was the volumetric soil water content (cm3 cm-3). Soil temperature and

D

moisture parameters were systematically added to the multiple regression model in the order they

TE

appear in Equation 1. The terms a, b, c, d, and e represented model-fit regression coefficients for the five soil temperature and moisture related parameters, while f represented the regression constant.

AC CE P

Since it is well-known that both soil moisture (Linn and Doran, 1984; Skopp et al., 1990; Brown et al., 2009) and temperature (Fang and Moncrieff, 2001; Fierer et al., 2006; Reth et al., 2009) correlate well to the biological process of soil respiration and that those correlations are generally more complex than simple linear relationships, it stands to reason that both linear and quadratic soil moisture and temperature terms and their product (i.e., the complete-quadratic, full multiple regression model, Eq. 1) be included in a multiple regression model to consistently and systematically evaluate the combined effects of soil moisture and temperature on soil respiration. Using both linear and multiple regression approaches to evaluate the relationships between soil respiration and soil moisture and temperature dates back numerous decades (Witkamp, 1966; Froment, 1972; Gupta and Singh, 1981). Ninety-five percent confidence intervals (CI) were calculated for the resulting regression coefficient estimates (i.e., a, b, c, d, e, and f) for each of the regional datasets and the single,

ACCEPTED MANUSCRIPT combined dataset models. Confidence interval overlap was used to ascertain differences in coefficient estimates among the four datasets. In addition, best-fit multiple regression models were determined, based on the initial complete-quadratic model (Eq. 1), for each dataset by sequentially

PT

removing non-significant model parameters, starting with the product term then the quadratic terms,

RI

one at a time until only significant model parameters remained. The linear term was kept in the

SC

model regardless of its significance if its quadratic term was significant. Since the design structures of all field experiments varied widely, no formal design structures were retained for regression

NU

analyses. Since all respiration measurements conducted sequentially in time in any given plot were

MA

separated by at least 5 d, which allowed each plot to experience at least several diurnal temperature cycles and continued drying unless rainfall or irrigation occurred, it was reasonably assumed that all

D

measurements were independent and any temporal autocorrelation was negligible. Thus neither

TE

repeated measures nor temporal autocorrelation were explicitly addressed in this data analysis. All statistical analyses were conducted using Minitab (version 16, Minitab Inc., State College, PA).

AC CE P

Significance was judged at P < 0.05.

3. Results and discussion

3.1. General description of the regional datasets Field measurements made across all studies included in this meta-analysis captured a relatively wide range in soil respiration rates, soil moisture contents, and soil temperatures (Fig. 2). Soil respiration ranged from 0.6 to 61.4, 0 to 64, and 0.95 to 10.2 μmol CO2 m-2 s-1 for the AR – Ozark Highlands, AR – Delta, and Florida datasets, respectively. Soil moisture in the top 6 to 10 cm ranged from 0.01 to 0.65, from < 0.01 to 0.80, and from 0.09 to 0.81 cm3 cm-3 for the AR – Ozark Highlands, AR – Delta, and Florida datasets, respectively. The extremely wet soil conditions encountered at some times generally occurred during winter months and/or shortly after substantial rainfall, such that some ponding at the soil surface was present. The wide range of soil water

ACCEPTED MANUSCRIPT contents associated with the measured data included in this study, which spans both udic and aquic soil moisture regimes, at least partially addresses the recommendation by Vicca et al. (2014) to compile and analyze datasets across broad precipitation gradients and soil moisture regimes to

PT

capture as wide a range of soil water contents as possible. Soil temperatures at the 10-cm depth

RI

ranged from 3.4 to 35.8, from 2.4 to 46.8, and from 11.1 to 26.9 oC for the AR – Ozark Highlands,

SC

AR – Delta, and Florida datasets, respectively. The differing temperature maxima between the AR and FL sites was likely a result of the managed grassland landuse at the FL site, which provided a

NU

canopy cover for much of the growing season, while little canopy coverage in the agroecosystems of

MA

the AR sites was present at times during the growing season. The relatively wide range in 10-cm soil temperatures is consistent with the wide range in soil temperature regimes (i.e., mesic, thermic,

TE

D

and hyperthermic) encompassed by the various field studies included in this meta-analysis.

3.2. Physiographic Region Differences

AC CE P

3.2.1. Correlation Comparisons

Using the single, combined dataset (i.e., 4511 observations), soil respiration was significantly (P < 0.05) positively correlated with T10cm, VWC, (T10cm)2, and the product of T10cm and VWC, but was uncorrelated (P > 0.05) to (VWC)2 (Table 2). Results for a positive soil temperature relationship with soil respiration are similar to previous reports (McMullen et al., 2014; Ding et al., 2010; Ruehr et al., 2010; Brown et al., 2009; Brye et al., 2006a; Jones et al., 2006; Fang and Moncrieff, 2001). Results for a positive soil moisture relationship with soil respiration are also similar to previous reports (Pingintha et al., 2010; Brown et al., 2009). Despite the significant correlation of four of five environmental parameters evaluated in this study to soil respiration, correlation coefficients were all generally low (i.e., r < 0.3). With the exception of VWC for the AR – Ozark Highlands dataset, similar to that for the combined dataset, T10cm, VWC, (T10cm)2, and T10cm x VWC were significantly (P < 0.05) positively

ACCEPTED MANUSCRIPT correlated with soil respiration within the AR – Ozark Highlands, AR – Delta, and Florida datasets separately (Table 2). In addition, similar to that for the combined dataset, (VWC)2 was also uncorrelated (P > 0.05) with soil respiration for the AR – Delta dataset (Table 2). However, in

PT

contrast to the combined dataset, (VWC)2 was significantly (P < 0.05) negatively correlated with soil

RI

respiration for the AR – Ozark Highlands dataset, but significantly (P < 0.05) positively correlated

SC

with soil respiration for the Florida dataset (Table 2). Similar to the combined dataset, correlations with soil respiration were generally low (i.e., r ≤ 0.40) for the both AR datasets. However, in

MA

NU

contrast, correlations with soil respiration were larger (i.e., 0.43 < r < 0.78) for the Florida dataset.

3.2.2. Full multiple regression model comparisons

D

Using all five environmental parameters to explain the variation in soil respiration in the

TE

combined dataset resulted in a significant overall model (P < 0.001), where each of the five parameters plus the constant were also significant (P < 0.01) in the model (Table 3). However, the

AC CE P

full model only explained 14% of the variation in soil respiration (Table 3), indicating there are likely one or more factors, beyond those strictly related to soil moisture and temperature, that control soil respiration collectively across the southern climate gradient represented by the data from the physiographic regions included in this meta-analysis. Similar to the combined dataset, based on the same five environmental parameters, the full model explained only 16% of the variation in soil respiration for the AR – Delta dataset (Table 3). However, the full model explained 32 and 67% of the variation in soil respiration for the AR – Ozark Highlands and Florida datasets, respectively (Table 3), indicating that soil moisture and temperature related environmental parameters may be more closely linked to soil respiration in some regions than in others. In addition, these results suggest that soil moisture and temperature may have confounding effects on soil respiration as reported by Davidson et al. (1998). Alternatively, the AR – Ozark Highlands and Florida datasets could have simply had less variability in their environmental parameters. Nonetheless, it is clear

ACCEPTED MANUSCRIPT that, at a given soil moisture and soil temperature combination threshold, soil respiration becomes negligible (Fig. 2). However, it is also clear that the soil moisture-temperature threshold at which soil respiration becomes negligible can differ among physiographic regions. For example, at ~ 16oC,

PT

soil respiration in FL was ~ 2 μmol m-2 s-1, whereas soil respiration was four to seven times that for

RI

the AR sites (Fig. 2).

SC

Despite some widely differing characteristics among the three regions for which data were included in this meta-analysis, some similarities and differences in coefficient estimates resulted

NU

among datasets (Table 3). Coefficient estimates for the T10cm term from the two Arkansas datasets

MA

were both significant (P < 0.05) in the model and shared the same positive sign as that for the combined dataset, but that for the Florida dataset was non-significant (P > 0.05) in the model.

D

Regardless of sign, the CIs for the T10cm term for all four datasets overlapped, indicating the

TE

coefficients for the T10cm term were similar among all four datasets. Similar to that for the T10cm term, coefficient estimates for the VWC term from the two Arkansas datasets were both significant

AC CE P

(P < 0.05) in the model and shared the same positive sign as that for the combined dataset, but that for the Florida dataset was non-significant (P > 0.05) in the model. However, based on CIs, the coefficient estimate for the VWC term for the two Arkansas datasets, which did not differ, were greater than that for the Florida dataset (Table 3). Coefficient estimates for the (T10cm)2 term from the other three datasets were all non-significant (P > 0.05) in the model and all were similar to that for the combined dataset. In contrast to the VWC term, coefficient estimates for the (VWC)2 term from the other three models were all significant (P < 0.05) in the model and shared the same negative sign as that for the combined dataset. Similar to the (T10cm)2 term, the coefficient estimate for the T10cm x VWC term from the other three datasets were all non-significant (P > 0.05) in the model and all were similar to that for the combined dataset based on overlapping confidence intervals (Table 3). Considering that some characteristics differed widely among the three regions, it was not

ACCEPTED MANUSCRIPT surprising that there were statistical differences in full models and their coefficient estimates among the three regions’ datasets and the combined dataset. However, what was more surprising was the number of similarities among full models and their coefficient estimates across the three regions’

PT

datasets and the combined dataset. The number of statistical similarities among datasets across the

RI

southern climate gradient from northwest Arkansas to south-central Florida suggests that perhaps a

SC

robust (i.e., widely applicable) regression model, possibly a random coefficient regression model, can be ascertained to reasonably predict soil respiration across regions because environmental

NU

controls on soil respiration likely behave in a similar direction, but with varying magnitudes, in the

MA

relatively warm and wet climate of the mid-southern, southern, and southeastern US.

D

3.2.3. Best-fit regression model comparisons

TE

Best-fit multiple regression models, based on the complete-quadratic model, varied among datasets. All five environmental parameters were significant (P < 0.05) in the model for the

AC CE P

combined dataset, but the predictability of soil respiration was quite low (R2 = 0.14; Table 4). Both Arkansas datasets resulted in best-fit multiple regression models with the same four of five environmental parameters except for the T10cm x VWC term, but, similar to that for the combined dataset, the predictability of soil respiration was quite low (R2 = 0.16 and 0.32 for the AR – Delta and AR – Ozark Highlands, respectively). However, over 60% of the variation in soil respiration was explained by the model for the Florida dataset, which only included the constant and the T10cm and (T10cm)2 terms (R2 = 0.62) and no term related to soil moisture. The model fits for the Florida dataset were only slightly lower than the model fits reported by Gaumont-Guay at al. (2006) that included only soil temperature. Similar to the full, five-parameter, complete-quadratic models, some differences in best-fit models and coefficient estimates were expected due to the differences in characteristics among the regions. However, the numerous similarities in best-fit models and coefficient estimates among the

ACCEPTED MANUSCRIPT regions’ datasets and compared to the combined dataset (Table 4) suggest that, with more data to refine the relationships and/or more complex functions, one or only a few multiple regression models may be ascertained to predict soil respiration instead of many site-specific models. If

PT

achievable, a single or only a few models would assist regional- or larger-scale C cycling models,

RI

such as Century (Parton et al., 1988) and/or the Community Land Model (Levis et al., 2014), to

SC

predict changes in C stocks and sequestration over time and into the future as a result of climate

NU

change.

MA

3.3. Soil moisture regime differences

Excluding the Florida dataset, five separate datasets were prepared, analyzed, and compared

D

among Arkansas field studies only. Those datasets included the Ozark Highlands under udic

TE

conditions (Ozarks – Udic; n = 722), Ozark Highlands under aquic conditions (Ozarks – Aquic; n = 198), Delta region of eastern Arkansas under udic conditions (Delta – Udic; n = 1248), Delta region

AC CE P

of eastern Arkansas under aquic conditions (Delta – Aquic; n = 2271), and all data combined across regions and soil moisture regimes (n = 4439).

3.3.1. Correlation comparisons

All five soil moisture and temperature related environmental parameters were significantly (P < 0.05) correlated with soil respiration in one or more of the five datasets analyzed (Table 5). The Delta – Udic dataset had all five environmental parameters significantly (P < 0.05) correlated with soil respiration. Volumetric water content was the only parameter unrelated (P > 0.05) to soil respiration in the Delta –Aquic dataset. Both VWC and (VWC)2 were unrelated (P > 0.05) to soil respiration in the Ozarks – Udic dataset. When significant, T10cm, (T10cm)2, and T10cm x VWC were always positively correlated with soil respiration. When significant, VWC was positively correlated in two datasets, but was negatively correlated with soil respiration in the Ozarks – Aquic dataset.

ACCEPTED MANUSCRIPT Despite the number of significant correlations between environmental parameters and soil respiration among the various datasets, all correlation coefficients were generally low (i.e., r ≤ 0.48;

PT

Table 5).

RI

3.3.2. Full multiple regression model comparisons

SC

Using all five environmental parameters in a complete-quadratic, multiple regression model to explain the variation in soil respiration in the combined dataset from Arkansas field studies only

NU

resulted in a significant overall model (P < 0.001; Table 6). Each of the five parameters were

MA

significant (P < 0.05) in the model, but only 14% of the variation in soil respiration was explained by the full model using the combined dataset (Table 6). As previously noted, these results indicate

D

there are likely one or more factors, beyond those strictly related to soil moisture and temperature,

TE

which control soil respiration in the relatively warm and moist Arkansas environment. Similar to the combined dataset, based on the same five environmental parameters, the full model explained only

AC CE P

15% of the variation in soil respiration for the Delta - Aquic dataset, while 38, 36, and 31% of the variation in soil respiration was explained for the Ozarks – Udic, Ozarks – Aquic, and Delta – Udic datasets, respectively (Table 6).

Similar to the comparison among physiographic regions including Florida, and similar to that hypothesized, several differences existed among regression coefficient estimates for the five datasets prepared from Arkansas only field studies (Table 6). The resulting model for the Delta – Udic dataset was unique among the datasets evaluated. Coefficient estimates for T10cm, VWC, and T10cm x VWC for the Delta – Udic dataset all differed from those for the combined, Ozarks – Udic, and Delta – Aquic datasets and the T10cm x VWC coefficient estimate also differed from that for the Ozarks – Aquic dataset. With only 198 observations, the Ozarks – Aquic dataset had large variability associated with the multiple regression coefficient estimates. Coefficient estimates for T10cm, VWC, and (T10cm)2 for the Ozarks – Aquic dataset were similar to the estimates for each of the

ACCEPTED MANUSCRIPT other three dataset models. With the exception of those for the Delta – Udic dataset, coefficient estimates for the constant, T10cm, VWC, and T10cm x VWC for the other three region-soil-moistureregime datasets did not differ from those for the combined dataset or among themselves.

PT

The relationship among soil respiration and the five environmental parameters evaluated in

RI

this meta-analysis differed somewhat between soil moisture regimes within physiographic regions

SC

(Table 6). Coefficient estimates for T10cm, VWC, and T10cm x VWC differed between soil moisture regimes in the Delta region of eastern Arkansas, but did not differ in the Ozark Highlands regions of

NU

northwest Arkansas. Coefficient estimates for (VWC)2 differed between soil moisture regimes in the

MA

Ozark Highlands regions of northwest Arkansas, but did not differ in the Delta region of eastern Arkansas. Coefficient estimates for (T10cm)2 were unaffected by physiographic region or soil

D

moisture regime. In general, the single model from the combined dataset was equally effective

TE

describing the relationship between soil respiration and soil moisture and temperature related environmental parameters for three of the four region-soil-moisture-regime datasets. However, the

AC CE P

single model from the combined dataset did not adequately describe the relationship between soil respiration and soil moisture and temperature related environmental parameters for the Delta – Udic dataset. In addition, none of the other five dataset models provided a large degree of predictive confidence (i.e., 0.13 < R2 < 0.39).

3.3.3. Best-fit regression model comparisons As previously noted, best-fit multiple regression models, based on the complete-quadratic model, varied somewhat among Arkansas only datasets (Table 7). Similar to the full model tested and described above (Table 6), the predictive ability of all best-fit models remained low (i.e., R2 < 0.39). Similarities and differences in coefficient estimates existed among the region-soil-moistureregime dataset models and in comparison to the combined dataset model (Table 7). Coefficient

ACCEPTED MANUSCRIPT estimates for VWC for the Ozarks – and Delta – Aquic models were similar to that for the combined dataset model, while those for the Ozarks – and Delta – Udic models differed from that for the combined dataset model. Only the coefficient estimate for T10cm for the Delta – Udic model differed

PT

from that for the combined dataset model. When included in the final, best-fit model, coefficient

RI

estimates for (T10cm)2 were all similar to that for the combined dataset model. Only the coefficient

SC

estimate for (VWC)2 for the Ozarks – Udic model differed from that for the combined dataset model. When included in the final, best-fit model, the coefficient estimate for T10cm x VWC for the Delta –

NU

Udic dataset differed from that for the combined dataset model.

MA

Similar to the full model tested and described above (Table 6) and to that hypothesized, the relationship among soil respiration and the combined environmental parameters differed somewhat

D

between soil moisture regimes within physiographic regions and between physiographic regions

TE

within soil moisture regimes (Table 7). Results show that the relationship among soil respiration and combined soil moisture and temperature related environmental parameters is likely more complex

AC CE P

than what a single, complete-quadratic model from a combined dataset across physiographic regions and soil moisture regimes in Arkansas can adequately characterize. Comparing between physiographic regions within a soil moisture regime, results showed that coefficient estimates for the constant, T10cm, VWC, and (VWC)2 did not differ between regions in the aquic soil moisture regime, but all four terms’ coefficient estimates differed between regions in the udic soil moisture regime. However, the coefficient estimate for (T10cm)2 was similar between regions in the udic soil moisture regime, but was only significant (P < 0.05) in the model for the Delta – Aquic dataset model. Differences in the relationship among soil respiration and combined soil moisture and temperature related environmental parameters between soil moisture regimes may be related to less-limiting soil moisture under aquic compared to udic conditions during times of increasing and peak soil respiration during the growing season. In addition, during times outside of the typical growing season, particularly in Arkansas, many soils in aquic moisture regimes are too

ACCEPTED MANUSCRIPT wet to support much biological activity, thus soil respiration is essentially non-existent for extended periods of time (Helton, 2014). The well-established, general quadratic relationship between soil respiration and soil moisture supports this observation (Linn and Doran, 1984; Parton et al., 1993;

PT

Gaumont-Guay et al., 2006; Brye and Riley, 2009; Taggart et al., 2012)

RI

Physiographic region and soil moisture regime differences in the relationship among soil

SC

respiration and combined soil moisture and temperature related environmental parameters identified in this meta-analysis were consistent with similar differences in near-surface soil properties reported

NU

by Brye et al. (2004) and Brye and Gbur (2010). Brye et al. (2004) assessed the effects of climate on

MA

soil physical, chemical, and biological properties in the top 10 cm of silt-loam soils and linear relationships among several variables from multiple native prairies sampled between August 2001

D

and April 2002 in the Ozark Highlands and in the Grand Prairie region of eastern Arkansas, which

TE

encompasses the RREC field studies used in this meta-analysis (Brye et al., 2006a; Motschenbacher et al., 2015). Results showed that bulk density and total N and C concentrations did not differ

AC CE P

between regions, but that SOM concentration and the soil C:N ratio were both greater in the warmer/wetter Grand Prairie region of eastern Arkansas compared to those in the Ozark Highlands (Brye et al., 2004). The linear relationships between total N and C concentration, electrical conductivity, and extractable soil Mn and P and SOM concentration also differed between regions representing a climate gradient. Brye and Gbur (2010) compared SOM and soil organic carbon (SOC) content changes over time (i.e., 2001 to 2007) in the top 10 cm of silt-loam soils under native prairie and agricultural landuse from a single location in the Ozark Highlands and the Grand Prairie and from udic and aquic soil moisture regimes in both regions. Results showed that total soil N contents were similar between udic and aquic conditions in the Ozark Highlands, but differed between soil moisture regimes in the Grand Prairie and that differences in bulk density and SOM and SOM contents between physiographic region-soil moisture regime combinations were also affected by landuse (i.e.,

ACCEPTED MANUSCRIPT native prairie and agriculture) in 2001 (Brye and Gbur, 2010). In addition, averaged over time, results showed that SOM and SOC contents and the soil C:N ratio differed among physiographic region-soil moisture regime-landuse combinations.

PT

The results of this meta-analysis evaluating the effect of soil moisture regime on the

RI

relationship among soil respiration and soil moisture and temperature related environmental

SC

parameters can aid large-scale C cycling models, like the Century model (Parton et al., 1988), with prediction of SOM and C cycling in aquic soils often used for rice production, which is somewhat of

NU

a limitation for Century (Milne et al., 2008). The anaerobic conditions that occur in rice-producing

MA

soils for extended periods of time (i.e., at least the flooded portion of a growing season) alter the more well understood aerobic oxidation processes and biochemical reactions. Consequently, SOM

D

and C cycling predictions are generally more difficult without established relationships such as those

AC CE P

4. Conclusions

TE

developed and evaluated in this meta-analysis.

Though much is known regarding the independent effects of soil moisture and temperature on soil respiration, the combined and interactive effects of soil moisture and temperature on respiration are less clear. Furthermore, little research has been conducted evaluating the relationships among soil respiration and combined soil moisture and temperature related environmental parameters over regional scales for soils of the southern and southeastern US. In this study, the ability of a complete-quadratic, multiple regression model to predict soil respiration across a southern climate gradient from the Ozark Highlands of northwest Arkansas to the Southern Flatwoods of Florida using combined soil moisture and temperature related environmental parameters was generally low (R2 < 0.4). However, results suggested that the relationship among soil respiration and combined soil moisture and temperature related environmental parameters may be somewhat stable across a large geographic area and across some large site-specific differences,

ACCEPTED MANUSCRIPT such as soil moisture regime. Similar regression coefficient estimates among physiographic regions demonstrated some degree of robustness (i.e., wide applicability) of certain environmental parameters, but the inclusion of other soil-respiration-controlling factors and/or more complex

PT

functions will likely improve model predictability. However, results also indicated that the

RI

relationship among soil respiration and combined soil moisture and temperature related

SC

environmental parameters from loam- to silt-loam-textured soils differ more between soil moisture regimes in the warmer and wetter Mississippi River Delta region of eastern Arkansas than in the

NU

Ozark Highlands region of northwest Arkansas. Consequently, unless a larger dataset can be

MA

constructed and/or additional interactive functions be identified to improve resulting relationships and model predictability, it appears that physiographic region and/or soil moisture regime need to be

D

accounted for when attempting to predict soil respiration from combined soil moisture and

TE

temperature related environmental parameters, such as those evaluated in this meta-analysis. Improving the ability to predict soil respiration from directly measured and/or indirectly calculated

AC CE P

environmental parameters will also increase understanding of the factors controlling soil C sequestration, and potential agronomic and ecological sustainability, in the weathered soils of the southern and southeastern US.

Acknowledgements The authors gratefully acknowledge the field assistance provided by Matt Cordell, Lee Riley, Shawn Clark, Matt Gregory, Taylor Adams, Kevin Rorex, Terry Sells, and Daniel McCarty. This manuscript is a direct result of research efforts made in association with the multistate regional project S-1048.

ACCEPTED MANUSCRIPT References Adewopo, J.B., Silveira, M.L., Xu, S., Gerber, S., Sollenberger, L.E., Martin, T., 2015. Management intensification effects on autotrophic and heterotrophic soil respiration in subtropical grasslands. Ecol. Indic. 56, 6-14.

PT

Al-Kaisi, M.M., Yin, X., 2005. Tillage and crop residue effects on soil carbon and carbon dioxide emission in corn-soybean rotations. J. Environ. Qual. 34, 437-445.

SC

RI

Bernsten, J., Peterson, B.M., Olesen, J.E., 2006. Simulating trends in crop yield and soil carbon in a long-term experiment – effects of rising CO2, N deposition and improved cultivation. Plant Soil 287, 235-245.

NU

Bond-Lamberty, B., Thomson, A., 2010. Temperature-associated increases in the global soil respiration record. Nature 464, 579-582.

MA

Brown, M., Whitehead, D., Hunt, J.E., Clough, T.J., Arnold, G.C., Baisden, W.T., Sherlock, R.R., 2009. Regulation of soil surface respiration in a grazed pasture in New Zealand. Agric. For. Meteorol. 149, 205-213.

D

Brye, K.R., Gbur, E.E., 2010. Regional differences in soil carbon and nitrogen storage as affected by landuse and soil moisture regime. Soil Sci. 175, 339-348.

TE

Brye, K.R., Riley, T.L., 2009. Soil and plant property differences across a chronosequence of humid-temperate tallgrass prairie restorations. Soil Sci. 174, 346-357.

AC CE P

Brye, K.R., Golden, B., Slaton, N.A., 2006a. Poultry litter decomposition dynamics as affected by litter form and rate prior to flooding for rice production. Soil Sci. Soc. Am. J. 70, 11551167. Brye, K.R., West, C.P., Gbur, E.E., 2004. Soil quality differences under native tallgrass prairie across a climosequence in Arkansas. Am. Midl. Nat. 152, 214-230. Brye, K.R., Longer, D.E., Gbur, E.E., 2006b. Impact of tillage and reside burning on CO2 flux in a wheat-soybean production system. Soil Sci. Soc. Am. J. 70, 1145-1154. Brye, K.R., Mersiovsky, E., Hernandez, L., Ward, L., 2013. Soils of Arkansas. Arkansas Agricultural Experiment Station, University of Arkansas System Division of Agriculture, Fayetteville, AR. 136 p. Brye, K.R., Gower, S.T., Norman, J.M., Bundy, L.G., 2002. Carbon budgets for a prairie and agroecosystems: Effects of land use and inter-annual variability. Ecol. Appl. 12, 962-979. Davidson, E.A., Janssens, I.A., 2006. Temperature sensitivity of soil carbon decomposition and feedbacks to climate change. Nature 440, 165-173. Davidson, E.A., Belk, E., Boone, R.D., 1998. Soil water content and temperature as independent or confounded factors controlling soil respiration in a temperate mixed hardwood forest. Global Change Biol. 4, 217-227.

ACCEPTED MANUSCRIPT Ding, W., Yu, H., Cai, Z., Han, F., Xu, Z., 2010. Response of soil respiration to N fertilization in a loamy soil under maize cultivation. Geoderma 155 381-389.

PT

Fang, C., Moncrieff, J.B., 2001. The dependence of soil CO2 efflux on temperature. Soil Biol. Biochem. 33, 155-165.

RI

Fierer, N., Colman, B.P., Schimel, J.P., Jackson, R.B., 2006. Predicting the temperature dependence of microbial respiration in soil: a continental-scale analysis. Global Biogeochem. Cycles 20, GB3026.

SC

Froment, A., 1972. Soil respiration in a mixed oak forest. Oikos 23, 273-277.

NU

Gaumont-Guay, D., Black, T.A., Griffis, T.J., Barr, A.G., Jassal, R.S., Nesic, Z., 2006. Interpreting the dependence of soil respiration on soil temperature and water content in a boreal aspen stand. Agric For. Meteorol. 140, 220-235.

MA

Guo, L.B., Gifford, R.M., 2002. Soil carbon stocks and land use change: a meta analysis. Global Change Biol. 8, 345-360.

D

Gupta, S.R., Singh, J.S., 1981. Soil respiration in a tropical grassland. Soil Biol. Biochem. 13, 261-268.

TE

Helton, M., 2014. Carbon dioxide emissions from switchgrass and cottonwood grown as bioenergy crops in the Lower Mississippi Alluvial Valley. M.S. thesis. Univ. Ark., Fayetteville.

AC CE P

Jones, S.K., Rees, R.M., Kosmas, D., Ball, B.C., Skiba, U.M., 2006. Carbon sequestration in a temperate grassland; management and climatic controls. Soil Use Manage. 22, 132-142. Laganière, J., Angers, D.A., Paré, D., 2010. Carbon accumulation in agricultural soils after afforestation: a meta-analysis. Global Change Biol. 16, 439-453. Lal, R., 2004. Soil carbon sequestration impacts on global climate change and food security. Science 304, 1623-1627. Levis, S., Hartman, M.D., Bonan, G.B., 2014. The Community Land Model underestimates land-use CO2 emissions by neglecting soil disturbance from cultivation. Geosci. Model Dev. 7, 613-620. Linn, D.M., Doran, J.W., 1984. Effect of water-filled pore space on carbon dioxide and nitrous oxide production in tilled and nontilled soils. Soil Sci. Soc. Am. J. 48, 1267-1272. Lloyd, J., Taylor, J.A., 1994. On the temperature dependence of soil respiration. Funct. Ecol. 8, 315-323. McMullen, R.L., Brye, K.R., Gbur, E.E., 2014. Soil respiration as affected by long-term broiler litter application to a Udult in the Ozark Highlands. J. Environ. Qual. 44, 115-126. Milne, E., Williams, S., Brye, K., Easter, M., Killian, K., Paustian, K., 2008. Simulating soil

ACCEPTED MANUSCRIPT organic carbon in a rice-soybean-wheat-soybean chronosequence in Prairie County, Arkansas using the Century model. J. Integr. Biosci. 6, 41-52.

PT

Motschenbacher, J., Brye, K., Anders, M., Gbur, E., Slaton, N., Evans-White, M., 2015. Daily soil surface CO2 flux during non-flooded periods in flood-irrigated rice rotations. Agron. Sustain. Dev. 35, 771–782.

RI

National Oceanic and Atmospheric Administration (NOAA). 2015. Data Tools: 1981-2010 Normals [online] (Available at http://www.ncdc.noaa.gov/cdo-web/datatools/normals (verified 25 August, 2015)).

NU

SC

Oikawa, P.Y., Grantz, D.A., Chatterjee, A., Eberwein, J.E., Allsman, L.A., Jenerette, G.D., 2014. Unifying soil respiration pulses, inhibition, and temperature hysteresis through dynamics of labile soil carbon and O2. J. Geophys. Res. Biogeosci. 119, 521-536.

MA

Parton, W.J., Scurlock, J.M.O., Ojima, D.S., Gilmanov, T.G., Scholes, R.J., Schimel, D.S., Kirchner, T., Menaut, J-C., Seastedt, T., Garcia Moya, E., Kamnalrut, A., Kinyamario, J.I., 1993. Observations and modeling of biomass and soil organic matter dynamics for the grassland biome worldwide. Global Biogeochem. Cycles 7, 785-809.

D

Parton, W.J., Stewart, J.W.B., Cole, C.V., 1988. Dynamics of C, N, P and S in grassland soils: a model. Biogeochem. 5, 109-131.

AC CE P

TE

Pingintha, N., Leclerc, M.Y., Beasley, Jr., J.P., Zhang, G., Senthong, C., 2010. Assessment of the soil CO2 gradient method for soil CO2 efflux measurements: comparison of six models in the calculation of the relative gas diffusion coefficient. Tellus 62B, 47-58. Reth, S., Graf, W., Reichstein, M., Munch, J.C., 2009. Sustained stimulation of soil respiration after 10 years of experimental warming. Environ. Res. Lett. 4, 1-5. Ruehr, N.K., Knohl, A., Buchmann, N., 2010. Environmental variables controlling soil respiration on diurnal, seasonal and annual time-scales in a mixed mountain forest in Switzerland. Biogeochem. 98, 153-170. Skopp, J., Jawson, M.D., Doran, J.W., 1990. Steady-state aerobic microbial activity as a function of soil water content. Soil Sci. Soc. Am. J. 54, 1619-1625. Smith, S.F., 2013. Long-term residue and water management effects on soil respiration and soil aggregate stability in a wheat-soybean, double-crop system in eastern Arkansas. M.S. thesis. Univ. Ark., Fayetteville. Smith, S.F., Brye, K.R., Gbur, E., Chen, P., Korth, K., 2014. Long-term residue management effects on soil respiration in a wheat-soybean, double-crop system. Soil Sci. 179, 118-129. Taggart, M., Heitman, J.L., Shi, W., Vepraskas, M., 2012. Temperature and water content effects on carbon mineralization for sapric material. Wetlands 32, 939-944. VandenBygaart, A.J., Gregorich, E.G., Angers, D.A., 2003. Influence of agricultural

ACCEPTED MANUSCRIPT management of soil organic carbon: a compendium and assessment of Canadian studies. Can. J. Soil Sci. 83, 363-380. VandenBygaart, A.J., Yang, X.M., Aspinall, J.D., 2002. Variability in carbon sequestration potential in no-till soil landscapes of southern Ontario. Soil Tillage Res. 65, 231-241.

NU

SC

RI

PT

Vicca, S., Bahn, M., Estiarte, M., van Loon, E.E., Vargas, R., Alberti, G., Ambus, P., Arain, M.A., Beier, C., Bentley, L.P., Borken, W., Buchmann, N., Collins, S.L., de Dato, G., Dukes, J.S., Escolar, C., Fay, P., Guidolotti, G., Hanson, P.J., Kahmen, A., Kröel-Dulay, G., LadreiterKnauss, T., Larsen, K.S., Lellei-Kovacs, E., Lebrija-Trejos, E., Maestre, F.T., Marhan, S., Marshall, M., Meir, P., Miao, Y., Muhr, J., Niklaus, P.A., Ogaya, R., Peñuelas, J., Poll, C., Rustad, L.E., Savage, K., Schindlbacher, A., Schmidt, I.K., Smith, A.R., Sotta, E.D., Suseela, V., Tietema, A., van Gestel, N., van Straaten, O., Wan, S., Weber, U., Janssens, I.A., 2014. Can current moisture responses predict soil CO2 efflux under altered precipitation regimes? A synthesis of manipulation experiments. Biogeosci. 11, 2991-3013.

MA

Wagai, R., Brye, K.R., Gower, S.T., Norman, J.M., Bundy, L.G., 1998. Land use and environmental factors influencing soil surface CO2 flux and microbial biomass in natural and managed ecosystems in southern Wisconsin. Soil Biol. Biochem. 30, 1501-1509.

AC CE P

TE

D

Witkamp, M., 1966. Decomposition of leaf litter in relation to environment, microflora and microbial respiration. Ecology 47, 194-201.

ACCEPTED MANUSCRIPT Figure captions

Fig. 1. Graphical depiction of the approximate location and distribution of the field study sites in

PT

Arkansas (left panel; asterisks represent northwest Arkansas locations and triangles represent eastern

SC

RI

Arkansas locations) and Florida (right panel) that generated the data included in the meta-analysis.

Fig. 2. Relationships among soil respiration and 0- to 6-cm (AR studies) or 0- to 10-cm (Florida

NU

study) soil moisture and 10-cm soil temperature for three regional datasets: Arkansas (AR) - Ozark

AC CE P

TE

D

MA

Highlands, AR – Delta, and Florida. Note the different y-axis range for the FL data panels.

ACCEPTED MANUSCRIPT

PT

* * ▲

RI

▲

MA

NU

SC

▲

AC CE P

TE

D

Fig. 1

■

AC CE P

TE

D

MA

NU

SC

RI

PT

ACCEPTED MANUSCRIPT

Fig. 2

ACCEPTED MANUSCRIPT

Typic Glossaqualf s Typic Albaqualfs Typic Albaqualfs

1 yr (2006)

Managed grassland

Broiler litter application rates of 0, 5.6, 11.2 Mg ha-1 yr-1

3 yr (20092012)

Agroecosyst em

Wheat-soybean, double-crop rotation; high and low residue inputs; residue burning and nonburning; conventional tillage and notillage Pre-flood period of rice in a ricesoybean rotation; fresh and pelletized poultry litter Seven different rice rotations with corn, soybean, and wheat; conventional tillage and notillage; optimum fertilization Wheat-soybean, double-crop rotation; high and low residue inputs; residue burning and nonburning; conventional tillage and notillage; irrigated and dryland Switchgrass, cottonwood, and soybean-grain

Silt loam Loam Loam

33 0

Brye and Riley (2009)

116A

Silt loam

60 6

McMullen et al. (2014)

2 yr (2002, 2003)

134

Silt loam

57 6

Brye et al. (2006b)

1 yr (2004)

134 131D

Silt loam

48 0

Brye et al. (2006a)

2 yr (2009, 2010)

131D

Silt loam

65 4

Motschenbac her et al. (2015)

2 yr (2011, 2012)

134

Silt loam

96 0

Smith et al. (2014)

2 yr (2012, 2013)

134

Silt loam

91 5

Helton (2014)

NU

D

Delta

MA

Highlands

116A

RI

Native prairie Prairie restoration

SC

Tallgrass prairie

TE

Typic Glossaqualf s Aquic Fraglossud alfs

Arkansas Ozark

Agroecosyst em

AC CE P

Oxyaquic Fragiudalfs Mollic Albaqualfs Typic Fragiudults Typic Fragiudults

PT

Table 1 Summary of study characteristics pertaining to each data set included in the meta-analysis. Soil State/Regi General Management Duratio MLR Soil nb Reference Taxonom on Landuse Practices n/ Aa Surfa y (Year) ce Textu re

Agroecosyst em

Aquic Fraglossud alfs

Agroecosyst em

Aquic Fraglossud alfs Typic

Agroecosyst em

ACCEPTED MANUSCRIPT

Managed grassland

Flatwoods

1 yr (2013)

D

MA

NU

Major land resource area (MLRA) Number of observations (n)

TE

b

AC CE P

a

Low intensity/native rangeland Moderate intensity/silvopas ture High intensity/sown pasture

155

Sand

PT

Florida Southern

RI

Typic Alaquods

sorghum rotation

SC

Fragiaqualf s Oxyaquic Fragiudalfs

72

Adewopo et al. (2015)

ACCEPTED MANUSCRIPT Table 2 Summary of correlations among soil respiration and 10-cm soil temperature (T10cm), 0- to 6-cm volumetric soil water content (VWC), the square of T10cm [(T10cm)2], the square of VWC [(VWC)2], and the product of T10cm and VWC (T10cm x VWC) for the Arkansas – Ozark Highlands, Arkansas – Delta, and Florida datasets and for all data combined.

Arkansas – Ozark Highlands

920

Arkansas – Delta

3519 0.29 *** 72

All Data Combined

4511 0.26 ***

0.04 **

D

MA

NU

Number of observations (n) ** (P < 0.01), *** (P < 0.001)

TE

b

0.57 ***

AC CE P

a

0.77 ***

0.26 ***

(VWC)2 T10cm x VWC - 0.11 *** 0.26 *** < 0.01

0.22 ***

- 0.01

0.78 ***

SC

Florida

0.06 ***

PT

Soil Respiration Correlations (r) T10cm VWC (T10cm)2 0.40 ***b - 0.04 0.36 ***

RI

na

Dataset

0.43 ***

0.33 *** 0.60 *** 0.29 ***

ACCEPTED MANUSCRIPT

Statistic

Estimate CI95 upper CI95 lower

Dataset All Data Combined -4.13***b -1.59 -6.67

AR – Ozark Highlands -8.34*** -3.28 -13.40

Constant

T10cm

Estimate CI95 upper CI95 lower SS (%)

0.38*** 0.54 0.22 7.0

VWC

Estimate CI95 upper CI95 lower SS (%)

19.01*** 26.94 11.08 2.7

(T10cm)2

Estimate CI95 upper CI95 lower SS (%)

(VWC)2

T10cm x VWC

AR - Delta

Florida

-6.01*** -2.61 -9.41

3.75 13.21 -5.71

0.57*** 0.90 0.24 16.4

0.33*** 0.53 0.13 8.4

-0.22 0.88 -1.32 60.0

46.56*** 64.75 28.37 1.8

25.48*** 36.14 14.82 5.0

-27.54 13.00 -68.08 < 0.1

-0.006*** 0.00 -0.01 1.7

-0.006 0.00 -0.01 0.6

-0.003 0.00 -0.01 1.0

0.008 0.04 -0.02 3.7

Estimate CI95 upper CI95 lower SS (%)

-38.14*** -30.27 -46.01 2.6

-86.31*** -70.56 -102.06 12.9

-40.27*** -29.21 -51.33 1.8

-17.39* -2.80 -31.98 1.4

Estimate CI95 upper CI95 lower SS (%)

0.35** 0.58 0.12 0.2

0.28 0.86 -0.30 < 0.1

0.19 0.48 -0.10 < 0.1

1.63 3.48 -0.22 1.6

3519 < 0.001 0.16

72 < 0.001 0.67

AC CE P

TE

D

MA

NU

SC

RI

Model Parameter

PT

Table 3 Summary of multiple regression coefficient estimates, upper and lower 95% confidence intervals (CI95), and the percentage of the sequential sums of squares (SS) explained by each coefficient relating 10-cm soil temperature (T10cm), 0- to 6-cm volumetric soil water content (VWC), the square of T10cm [(T10cm)2], the square of VWC [(VWC)2], and the product of T10cm and VWC (T10cm x VWC) to soil respiration for the Arkansas – Ozark Highlands, Arkansas – Delta, and Florida datasets and for all data combined.

Whole model Na 4511 920 P < 0.001 < 0.001 2 R 0.14 0.32 a Number of observations (n) b * (P < 0.05), ** (P < 0.01), *** (P < 0.001)

ACCEPTED MANUSCRIPT

-10.56 -8.48 -12.68

0.70 0.91 0.49

54.48 62.27 46.70

-0.007 0.00 -0.01

31.78 36.73 26.83

Estimate CI95 upper CI95 lower

-8.15 -7.02 -9.28

0.45 0.54 0.35

Florida

Estimate CI95 upper CI95 lower

4.68 12.83 -3.47

-0.54 0.31 -1.39

All Data Combined

Estimate CI95 upper CI95 lower

-4.13 -1.59 -6.67

0.38 0.54 0.22

AC

-0.005 0.00 -0.01

-

0.023 0.04 0.00

19.01 26.94 11.08

-0.006 0.00 -0.01

CE

AR – Delta

RI

(T10cm)2

(VWC)2

T10cm x VWC -90.25 -76.80 -103.71 -

SC

VWC

Whole-model P R2 < 0.001

0.32

-

< 0.001

0.16

-

< 0.001

0.62

0.35 0.58 0.12

< 0.001

0.14

NU

AR – Ozark Highlands Estimate CI95 upper CI95 lower

Model Parameter Constant T10cm

MA

Statistic

PT ED

Dataset

PT

Table 4 Summary of best-fit multiple regression coefficient estimates and upper and lower 95% confidence intervals (CI95) relating 10-cm soil temperature (T10cm), 0- to 6-cm volumetric soil water content (VWC), the square of T10cm [(T10cm)2], the square of VWC [(VWC)2], and/or the product of T10cm and VWC (T10cm x VWC) to soil respiration for the Arkansas – Ozark Highlands, Arkansas – Delta, and Florida datasets and for all data combined.

-43.63 -33.79 -53.47 -38.14 -30.27 -46.01

ACCEPTED MANUSCRIPT

PT

Table 5 Summary of correlations among soil respiration and 10-cm soil temperature (T10cm), 0- to 6-cm volumetric soil water content (VWC), the square of T10cm [(T10cm)2], the square of VWC [(VWC)2], and the product of T10cm and VWC (T10cm x VWC) for five datasets from Arkansas soils differing by soil moisture regime (i.e., udic or aquic). Dataset na Soil Respiration Correlations (r) T10cm VWC (T10cm)2 (VWC)2 T10cm x VWC Ozark Highlands - Udic

722

0.42***b

Ozark Highlands - Aquic

198

0.48***

- 0.28***

Delta – Udic

1248 0.34***

0.23***

0.32***

Delta – Aquic

2271 0.29***

0.00

0.25***

- 0.06**

0.30***

0.03*

0.22***

- 0.02

0.28***

AC CE P

TE

D

- 0.03

0.46***

- 0.35***

RI

0.38***

SC

NU

MA

All Data Combined 4439 0.26*** a Number of observations (n) b * (P < 0.05), ** (P < 0.01), *** (P < 0.001)

0.03

0.22***

0.39*** - 0.06 0.43***

ACCEPTED MANUSCRIPT

Model Parameter

Statistic

Dataset All Data Combined

Constant

Estimate CI95 upper CI95 lower

-4.91***a -2.12 -7.71

Ozark Highlands - Udic -11.14*** -5.38 -16.90

T10cm

Estimate CI95 upper CI95 lower SS (%)

0.42*** 0.58 0.25 6.8

0.66** 1.04 0.28 18.7

VWC

Estimate CI95 upper CI95 lower SS (%)

23.06*** 32.36 13.76 2.7

(T10cm)2

Estimate CI95 upper CI95 lower SS (%)

(VWC)2

T10cm x VWC

PT

Table 6 Summary of multiple regression coefficient estimates, upper and lower 95% confidence intervals (CI95), and the percentage of the sequential sums of squares (SS) explained by each coefficient relating 10-cm soil temperature (T10cm), 0- to 6-cm volumetric soil water content (VWC), the square of T10cm [(T10cm)2], the square of VWC [(VWC)2], and the product of T10cm and VWC (T10cm x VWC) to soil respiration for five datasets from Arkansas soils differing by soil moisture regime (i.e., udic or aquic).

Delta - Aquic

11.51*** 16.39 6.63

-10.71*** -6.14 -15.27

0.41 2.07 -1.25 22.9

-0.64*** -0.33 -0.95 11.5

0.63*** 0.89 0.36 8.5

63.70*** 85.59 41.81 6.5

20.69 67.96 -26.58 0.5

-30.16*** -16.04 -44.27 9.2

40.88*** 55.27 26.48 3.3

-0.007*** 0.00 -0.01 1.8

-0.006 0.00 -0.01 0.9

-0.004 0.02 -0.03 2.7

-0.008** 0.00 -0.01 0.7

-0.007*** 0.00 -0.01 1.1

Estimate CI95 upper CI95 lower SS (%)

-43.30*** -33.85 -52.75 2.6

-119.8*** -96.28 -143.32 11.8

-44.68** -17.49 -71.87 10.3

-55.73*** -38.96 -72.51 2.6

-51.44*** -37.06 -65.83 2.0

Estimate CI95 upper CI95 lower SS (%)

0.27* 0.53 0.02 0.1

0.28 0.95 -0.40 0.1

0.52 1.98 -0.95 0.2

2.46*** 2.90 2.03 6.9

-0.26 0.14 -0.66 0.1

198 < 0.001 0.36

1248 < 0.001 0.31

2271 < 0.001 0.15

TE

D

MA

NU

SC

RI

Delta - Udic

AC CE P

Ozark Highlands - Aquic -5.93 19.21 -31.07

Whole model nb 4439 722 P < 0.001 < 0.001 2 R 0.14 0.38 a * (P < 0.05), ** (P < 0.01), *** (P < 0.001) b Number of observations (n)

ACCEPTED MANUSCRIPT

-145.13

-

NU

0.41*** 38.46***

-

-54.87***

-

-5.07

0.54

50.56

-

-37.28

-

-12.72

0.28

26.35

-

-72.45

-

11.51***

0.64*** 30.16*** -0.008**

-55.73***

2.46***

16.39

-0.33

-16.04

0.00

-38.96

2.90

6.63

-0.95

-44.27

-0.01

-72.51

2.03

-7.93***

0.48*** 32.64*** 0.006*** -46.94***

-

-6.51

0.60

39.15

0.00

-34.39

-

-9.34

0.36

26.13

-0.01

-59.49

-

-4.91***

0.42*** 23.06*** 0.007*** -43.30***

0.27*

-2.12

0.58

32.36

0.00

-33.85

0.53

-7.71

0.25

13.76

-0.01

-52.75

0.02

Estimate

Estimate CI95 upper CI95 lower

a

-

-8.89***

Estimate

CI95 upper CI95 lower All Data Combined

-103.31

SC

CI95 upper CI95 lower

CI95 upper CI95 lower Delta – Aquic

-

MA

Estimate

D

Delta – Udic

124.22***

TE

Ozark Highlands - Aquic

Estimate 13.32***a 0.78*** 71.63*** -0.007* CI95 upper -11.02 1.01 82.09 0.00 CI95 lower -15.62 0.56 61.17 -0.01

AC CE P

Ozark Highlands – Udic

RI

PT

Table 7 Summary of best-fit multiple regression coefficient estimates and upper and lower 95% confidence intervals (CI95) relating 10-cm soil temperature (T10cm), 0- to 6-cm volumetric soil water content (VWC), the square of T10cm [(T10cm)2], the square of VWC [(VWC)2], and/or the product of T10cm and VWC (T10cm x VWC) to soil respiration for five datasets from Arkansas soils differing by soil moisture regime (i.e., udic or aquic). Dataset Statistic Model Parameter Wholemodel 2 2 Constant T10cm VWC (T10cm) (VWC) T10cm x P R2 VWC

* (P < 0.05), ** (P < 0.01), *** (P < 0.001)

38

< 0.001 0.38

< 0.001 0.36

< 0.001 0.31

< 0.001 0.15

< 0.001 0.14

ACCEPTED MANUSCRIPT Highlights

PT

RI

SC NU MA D TE



Identifying controls on soil respiration can improve understanding of soil C cycling. Physiographic region and soil moisture regime effect soil respiration. Numerous similarities among multiple regression model coefficient estimates were identified. Moisture regime within region affected relationships explaining soil respiration.

AC CE P

  

39