Effects of soil texture on soil carbon and nitrogen dynamics after cessation of agriculture

Geoderma 136 (2006) 289 – 299 www.elsevier.com/locate/geoderma

Effects of soil texture on soil carbon and nitrogen dynamics after cessation of agriculture Kendra K. McLauchlan ⁎ Department of Ecology, Evolution, and Behavior University of Minnesota 1987 Upper Buford Circle St. Paul, MN 55108, USA Received 10 March 2005; received in revised form 22 March 2006; accepted 28 March 2006 Available online 19 May 2006

Abstract Soil organic matter (SOM) is an important ecosystem carbon (C) pool that is often depleted by agriculture. SOM content tends to be positively correlated with soil clay concentration among sites, but it is unknown how clay concentration affects the rate of SOM accumulation over time after cessation of agriculture. I used a 40-year chronosequence of 62 former agricultural fields in western Minnesota to determine the influence of clay concentration on the accumulation of soil C pools following agricultural abandonment. As time since cessation of agriculture increased, total soil organic carbon (SOC), unhydrolyzable C, microbial biomass C, labile C calculated from a laboratory soil incubation (Cl), and aggregate size all increased, while potential net nitrogen (N) mineralization and the decay constant of the labile C pool, kl, decreased. However, clay concentration had no effect on total soil C pool sizes or rate of accumulation. Clay concentration correlated positively with aggregate size and the rate of aggregate accumulation, and it correlated negatively with potential net N mineralization rates regardless of field age. These results indicate that on former agricultural fields converted to perennial grassland, soil texture may not be a significant factor influencing SOM accumulation rates on decadal time scales. © 2006 Elsevier B.V. All rights reserved. Keywords: Carbon cycle; Clay; Grain size; Land management; Nitrogen; Total organic carbon; Soil dynamics

1. Introduction Soil organic matter (SOM) is the repository for approximately 60% of the global terrestrial carbon (C) pool and is especially sensitive to agricultural land management (West and Post, 2002). An estimated 55 Pg C were released from the soil to the atmosphere during the 19th and 20th centuries because of agriculture (Paustian et al., 2000). SOM accumulates when perennial vegeta⁎ Present address. Environmental Studies Program, Dartmouth College, 6182 Steele Hall, Hanover, NH 03755, USA. Tel.: +1 603 646 0941; fax: +1 603 646-1682. E-mail address: [email protected]. 0016-7061/$ - see front matter © 2006 Elsevier B.V. All rights reserved. doi:10.1016/j.geoderma.2006.03.053

tion is established on former agricultural fields, creating a potential sink for atmospheric C (Post and Kwon, 2000; Follett, 2001). Yet, the factors that influence the rate, type, and magnitude of this accumulation are still unknown. Factors that influence the formation of soils over long periods of time–climate, organisms, relief, and parent material–may also affect the accumulation of soil organic carbon (SOC) on decadal timescales in former agricultural fields (Jenny, 1941). Variation in grassland vegetation (the “organisms” factor) has a minimal effect on SOC pools and rates of accumulation in the Great Plains of North America (Vinton and Burke, 1997; McLauchlan et al., 2006). Other research suggests that variation in the parent material state factor, leading to

290

K.K. McLauchlan / Geoderma 136 (2006) 289–299

differences in soil texture or clay concentration, may explain variation in the rate of SOC accumulation and C sequestration. There is some evidence that clay concentration may explain variation in rates of SOC accumulation. First, maximum and average SOC content increase with increasing soil clay content across several sites on the Great Plains (Nichols, 1984; Burke et al., 1989). However, this relationship is not global—sometimes SOC is better correlated with factors other than clay such as extractable aluminum, allophane content, or specific surface area (Percival et al., 2000; Krull et al., 2003). Nonetheless, the relationship between clay concentration and SOC content is sufficiently strong that SOM models such as Century (Parton et al., 1987) and RothC (Jenkinson, 1990) assume that SOM decomposition decreases as clay concentration increases, such that if all other factors are equal, SOC accumulates faster as soil clay concentration increases. The prediction that increasing clay concentration increases the rate of accumulation of SOC over decadal timescales has not been directly tested or empirically verified. Clay content may have distinct effects on the decomposition of different SOC pools (Franzluebbers et al., 1996). For example, respiration from soils during the early stages of a laboratory incubation, when labile SOC was being mineralized, was not influenced by clay (Wang et al., 2003). However, later in the same incubation, when more recalcitrant SOC pools were being mineralized, clay content slowed the rate of mineralization. Some SOM models are beginning to incorporate this heterogeneous clay effect on C decomposition (Muller and Hoper, 2004). In situ C mineralization rates generally decrease with increasing clay content (Hassink, 1997), although laboratory incubations do not always show this trend (Scott et al., 1996). These observations have led to the conclusion that clay particles protect some portion of SOC from decomposition. The same mechanisms that protect SOC from decomposition in clay-rich soils may cause them to accrue SOC more rapidly than sandy soils. Protection of SOC by clay particles has been postulated to occur through at least two separate mechanisms. First, as SOC becomes humified, it is chemically stabilized and adsorbed onto negatively charged clay minerals with high surface area. Second, SOC is physically protected from microbial mineralization through the formation of soil aggregates. The process of aggregate formation often occurs hierarchically, and the presence of clay particles enables this process (Tisdall and Oades, 1982; Six et al., 2000). Additionally, clay concentration may alter soil moisture, which in turn affects both decomposition of SOC and C inputs to soils via plant productivity.

Evidence for the role of clay content in soil nutrient cycling, especially the key step of nitrogen (N) mineralization, has been mixed. Some studies show that increasing clay content reduces net N mineralization (Cote et al., 2000) but in laboratory conditions, where temperature and moisture differences are controlled, clay content has little effect on net N mineralization rate (Giardina et al., 2001). Regardless of soil texture, net N mineralization decreases with increasing length of time since agriculture as the effects of below-ground disturbance become less pronounced and soil C content increases (Schimel, 1986; McLauchlan et al., 2006). The primary objective of this study was to determine the effect of soil clay concentration on the rate of change of SOM pools, N cycling, and soil structure over time following cessation of agriculture. To provide a range of soil textures, I studied soils from 62 grassland sites in western Minnesota, USA, that had similar topography, climate, and vegetation but differed in their parent material (either clay-rich glacial till or sandy glacial outwash) and length of time since the cessation of agriculture. I also sampled three native grassland sites that had never been plowed to constrain predictions of ecosystem variables. I tested three hypotheses relevant to the role of clay in the accumulation of SOM. First, I hypothesized that the accumulation of labile C, as measured by microbial biomass and by respiration from laboratory-incubated soils, would increase with increasing clay concentration because clay promotes the formation of aggregates which protect otherwise labile C from decomposition. Thus, the rate of increase in aggregate size should be positively influenced by clay concentration, as should the rate of increase in labile C pools over time since cessation of agriculture. Second, I hypothesized that the accumulation of recalcitrant C, as measured by unhydrolyzable C, would increase with increasing clay concentration because humified organic molecules, which are resistant to decomposition and have a relatively long turnover time, are chemically stabilized by clay minerals. Third, I hypothesized that soils with high clay concentration would have low rates of potential net N mineralization because organic N is protected from microbial activity either physically or through chemical bonds with clay minerals. 2. Methods 2.1. Study area and soil sampling To address these hypotheses, I identified a chronosequence of 62 former agricultural fields in Grant and Otter

K.K. McLauchlan / Geoderma 136 (2006) 289–299

Tail Counties in western Minnesota, and sampled soils and vegetation at 31 sites in 2001 and 31 sites in 2002. The vegetation before European settlement of the area in the mid-late 19th century was tallgrass prairie but now it is almost exclusively an agricultural landscape dominated by corn (Zea mays) and soybeans (Glycine max). The area is known as the “Prairie Pothole” region. These sites had been under agriculture but then converted back to perennial grassland at different times in the past 40 years as part of the Conservation Reserve Program (CRP) of the U.S. Department of Agriculture or the Waterfowl Production Area program of the Fish and Wildlife Service. The soils at all sites were originally formed under grassland vegetation, were then plow-tilled for row crop production for at least 20 years (during which time SOM levels were depleted), and were then converted to perennial grassland vegetation at a known time since 1960. In addition to the former agricultural fields, I sampled three sites with native prairie soils that had never been plowed and used them to constrain predictions of SOM content. The native prairie sites are similar to former agricultural fields in landscape position. The mean clay concentration of native prairie sites (16.4 ± 1.2%) is similar to the mean clay concentration of converted grassland sites (19.7 ± 7.3%). All sites are located in the same geographic area in western Minnesota within an 80-km radius and share the same climate (Staff, 1997). Sites differ primarily in type of soil parent material, either clay-rich calcareous glacial till or sandy glacial outwash, both from the Laurentide Ice Sheet. However, there is a range of soil textures associated with each type of parent material, making soil texture a continuous variable. The soils formed on glacial till range from loams to clay loams, encompassing three soil series complexes (Formdale, Sisseton, and Barnes), and they are classified as fine-loamy, mixed Calcic Hapludolls (Staff, 1997). The soils formed on outwash are sandy loams on two soil series (Arvilla and Dorset) and they are classified as sandy, mixed Udic Hapludolls (Staff, 1997). The claysized particles for all soils sampled in this study are predominantly montmorillonite. Current vegetation composition corresponds well with original seeding mix used to establish perennial vegetation at each site. Each site was planted with a mix of grasses after cessation of agriculture, either a cool-season grass, usually Bromus inermis, B. inermis and a legume (Medicago sativa L.), or native warm-season grasses such as Andropogon gerardii, Panicum virgatum, and Sorghastrum nutans. The native prairie sites had a mixture of cool-season and warm-season grassland vegetation native to western Minnesota. This amount of variation in grassland vegetation had previously been shown to have

291

negligible effects on SOM properties such as total SOC and labile SOC pools (McLauchlan et al., 2006). Sites differed in both soil texture and landscape relief, but topographic position was held constant during sampling to minimize variation among sites due to relief. Additionally, I sampled east- or west-facing slopes to minimize differences among sites caused by extreme aspect. Samples were taken at the end of the growing season over a 3-day period in September of 2001 and 2002, and stored at 4°C until they could be processed. Soils were sampled in two different years to provide a greater range of clay concentrations than had been sampled in 2001. On the shoulder slope position at each site, five 1.9-cm-diameter soil cores were taken to 10 cm depth in each of three 2 × 2-m plots and composited by plot. A portion of the fresh soil from each plot was passed through a 2-mm sieve to remove rocks and plant roots. Part of this sample was air-dried and ground for analyses of total C and N, unhydrolyzable C, and particle size while the fresh soil was used for determining microbial biomass C, soil respiration, and potential net N mineralization. The remainder of the fresh soil sample was passed through a 4-mm sieve and immediately air-dried for analysis of aggregate size structure. 2.2. Soil C methods On fresh soil samples, I measured microbial biomass C using the chloroform fumigation direct extraction technique (Anderson and Joergensen, 1997) and soil respiration from a 360-day laboratory incubation from which I mathematically determined the size (Cl)and decay constant (kl) of the labile C pool (Wedin and Pastor, 1993). On dried soil samples, total organic C and total N were determined by combustion with an elemental analyzer after pretreatment with phosphoric acid to remove carbonates (COSTECH Analytical, Model ECS 4010, Valencia, CA). Dried soil samples were refluxed in 6N hydrochloric acid at 110 °C for 16 h to isolate the chemically resistant residue, or unhydrolyzable fraction, which represents a recalcitrant soil C pool (Paul et al., 2001). Bulk density was used to convert these measurements to an areal basis. Further details of these methods are given in (McLauchlan and Hobbie, 2004). 2.3. Other soil properties I calculated potential net N mineralization rates as the difference in NO3− and NH4+ pools between the beginning and end of an aerobic 28-day incubation at 22°C (Hart et al., 1994). Inorganic N was measured by extraction with a 1 M potassium chloride solution (10 g

292

K.K. McLauchlan / Geoderma 136 (2006) 289–299

fresh soil: 25 mL extractant) followed by analysis on an Alpkem autoanalyzer fitted with a cadmium coil (FS 3000, OI Analytical, College Station, TX). The hydrometer method was used to determine particle size distribution after dispersal of soil particles with a 5% sodium hexametaphosphate solution (Gee and Bauder, 1986). Aggregate size distribution, a measure of soil structure, was determined using physical separation of aggregate size classes through wet sieving. To ensure that only water-stable aggregates were measured, an airdried soil sample of 50 g that had been passed through a 4-mm screen was wetted by submerging in water for 10 min, and soil was sieved mechanically for 10 min into five water-stable aggregate size classes: greater than 2000 μm, 1000–2000 μm, 500–1000 μm, 250–500 μm, and less than 250 μm (Cambardella and Elliott, 1992). An index of aggregate size, geometric mean diameter (GMD), was calculated with the formula Pw ∗logx  i i ws GMD ¼ e

centered around the mean prior to analysis for the interaction term of continuous predictor variables (field age and clay concentration). Non-significant terms were removed serially from the model in the order of increasing F values. Response variables associated with soil C included total SOC, unhydrolyzable C, microbial C, Cl, and

where wi is the weight of aggregates in a size class with average diameter xi and ws is the weight of the sample (Kemper and Chepil, 1965). The mass of sand particles collected on each sieve was included in the aggregate size calculations. 2.4. Plant properties Soil texture may influence plant productivity, which could influence soil properties, so I measured plant productivity to account for this effect. Aboveground net primary productivity (ANPP) was estimated by clipping maximum standing aboveground biomass in a 0.075 m2 area in each of three plots at each site, and drying and weighing live material. Belowground net primary productivity (BNPP) was estimated by measuring root production into a cylinder of root-free soil in each plot during the growing season (Cuevas and Medina, 1988). Root biomass produced during the season was separated from soil, dried, and weighed. 2.5. Data analysis A stepwise multiple least squares regression with backward elimination of predictors using the cutoff P b 0.05 was conducted using JMP 5.0 (SAS Institute). Field age, clay concentration, year sampled, and the pairwise interactions of year sampled and field age, year sampled and clay concentration, and field age and clay concentration were used as predictor variables. Data were

Fig. 1. The relationships between field age and several soil C pools: (a) total SOC, (b) unhydrolyzed C, a recalcitrant C pool, and (c) microbial C, a labile C pool. Bivariate linear regression lines are solid and are shown for clarity rather than the values for multivariate regression. The non-linear model fit is shown with a dashed line. Values for soils on native prairie sites are indicated with a star. The statistics for each model are (a) linear r2 = 0.24 and P b 0.0001, non-linear r2 = 0.25 and P b 0.0001; (b) linear r2 = 0.18 and P b 0.0006, non-linear r2 = 0.19 and P b 0.0004; and (c) linear r2 = 0.24 and P b 0.0001, non-linear r2 = 0.25 and P b 0.0001.

K.K. McLauchlan / Geoderma 136 (2006) 289–299

293

Table 1 Coefficients for multiple regression of each of seven response variables with predictors field age, % clay, year sampled, and all bivariate interactions (neither field age and year nor year sampled and % clay were ever significant)

GMD Net N mineralization SOC Microbial C Cl kl Unhydrolyzed C

Units

r2

Intercept

Field age

% clay

Year sampled

Field age × % clay

kg ha− 1 day− 1 kg ha− 1 kg ha− 1 kg ha− 1 day− 1 kg ha− 1

0.55 0.35 0.24 0.24 0.33 0.54 0.18

0.74⁎⁎⁎ 1.13⁎⁎⁎ 23916⁎⁎⁎ 535⁎⁎⁎ 888⁎⁎ 0.017⁎⁎⁎ 9614⁎⁎⁎

0.0029⁎⁎⁎ −0.021⁎⁎⁎ 516.8⁎⁎⁎ 12.8⁎⁎⁎ 32.9⁎⁎⁎ −0.0003⁎⁎⁎ 243.1⁎⁎⁎

0.0039⁎⁎ − 0.016⁎⁎

0.041⁎⁎⁎

0.0003⁎⁎

− 231⁎ 0.002⁎⁎⁎

Coefficients for predictors are slopes of the relationships between the predictor and response, except for the categorical year sampled, which represents the difference in intercept between samples taken in 2002 and 2001. The intercept for 2002 samples is calculated by subtracting the respective estimate from the original coefficient and adding it to 2001 samples. The coefficient for the interaction term is applied to data centered around the respective means for field age (17 years since cessation of agriculture) and % clay (19.7%). For predictors, ⁎ indicates Pb0.05, ⁎⁎ indicates Pb0.01, and ⁎⁎⁎ indicates Pb0.001.

kl. Other response variables were aggregate size (GMD) and potential net N mineralization. The addition of % silt to % clay as a predictor variable did not change the results of the multiple regression. A non-linear model between field age and three soil variables–total SOC, unhydrolyzable C, and microbial C–was also used to assess the linearity of these relationships as assumed in the multiple regression. The nonlinear model follows the equation Ct = Cm − Co⁎e(−kt), where Ct is the soil C content at time t, Co is the initial soil C content, Cm is the equilibrium or maximum soil C content, and k is the decomposition rate for the soil C pool (Jastrow, 1996). Parameter estimates were obtained with non-linear curve fitting in JMP 5.0. For illustrative purposes, to represent the independent effect of clay concentration on certain response variables, the residuals of two bivariate regressions between field age and GMD and potential net N mineralization were used as response variables in a simple linear regression with % clay. This analysis illustrates the results from the multivariate regression described above. As there was no significant relationship between response variables used in the multiple regression and either of the plant productivity variables ANPP or BNPP, these variables were excluded from further analysis.

There was also no relationship between ANPP and BNPP and the predictor variables used in the multiple regression. 3. Results 3.1. Field age Total, labile, and recalcitrant C pools increased with increasing time since cessation of agriculture (Fig. 1). Assuming a linear rate of increase, total SOC increased by 516.8 kg ha− 1 year− 1 while recalcitrant C, measured as unhydrolyzable C, increased by 243.1 kg ha− 1 year− 1 (Table 1). Microbial biomass C increased by 12.8 kg ha− 1 year− 1, and labile C (measured as Cl) increased by 32.9 kg ha− 1 year− 1 (Table 1). These rates of accumulation are different for each C pool in absolute terms but they are similar in relative terms to each C pool size. Annually, total SOC increased by 0.79%, microbial C by 0.88%, and unhydrolyzable C by 0.79% of their maximum pool sizes, respectively. These increases in soil C pools may be either linear or non-linear. The non-linear model assumes that the rate of change in soil properties slows over time and provides nearly an identical fit as the linear model during the first

Table 2 Average, minimum, and maximum values of soil response variables for sites on 62 former agricultural fields Property

Units

Average (s.e.)

Minimum

Maximum

GMD SOC Unhydrolyzed C Microbial C Cl kl Net N mineralization

kg ha− 1 kg ha− 1 kg ha− 1 kg ha− 1 d− 1 kg ha− 1 d− 1

0.862 (0.011) 32,734 (1434) 13,761 (779) 753 (35) 1994 (117) 0.012 (0.0006) 0.464 (0.052)

0.645 13,586 3889 330 667 0.004 −0.013

1.115 56,154 28,707 1603 4641 0.024 1.910

294

K.K. McLauchlan / Geoderma 136 (2006) 289–299

concentration, GMD (aggregate size index) increased by 0.039 and potential net N mineralization decreased by 0.16 kg ha− 1 day− 1 (Table 1). The differences among sites in GMD and potential net N mineralization due to clay concentration could have originated before, during, or after agriculture, but these data cannot be used for this analysis. The relationships between clay concentration, GMD, and potential net N mineralization were illustrated with a series of bivariate regressions. Clay concentration explained a portion of the variation in the residuals of a

Fig. 2. The relationship between field age and soil texture for each former agricultural field. Field age is the number of years since the site was converted from agriculture to perennial grassland. % clay is shown with a solid circle and % sand for the same site is shown with an open square.

40 years after cessation of agriculture. Thus, it is impossible to distinguish which model is more suitable to use during the time period of the chronosequence (Fig. 1). Regardless of which model is used during the first few decades after agriculture, the amount of variability in SOC contents among sites was large, ranging from 13,586 to 56,154 kg ha− 1, with only a quarter of this variation explained by field age (Tables 1 and 2). Not only did soil C pools increase over time, but soil structure (as indicated by GMD, an index of aggregate size) increased significantly with increasing field age. However, the decomposability of the labile C, as indicated by kl, decreased with increasing field age. Potential net N mineralization also decreased with field age by 0.021 kg ha− 1 day− 1 year− 1 (Table 1). The amount of variation explained by field age was also generally small for these response variables. 3.2. Clay Clay concentration across all 62 former agricultural fields averaged 19.7 ± 7.3% and did not vary systematically with field age (Fig. 2). In contrast to expectations, clay concentration did not significantly affect any soil C pools, either labile or recalcitrant (total SOC, unhydrolyzable SOC, microbial C, or Cl) (Table 1, Fig. 3). The amount of variation in soil C pools measured among the 62 sites was large, and was not explained by variation in clay concentration (Tables 1 and 2). However, as clay concentrations increased in soils, aggregate size increased significantly, and potential net N mineralization decreased significantly (Table 1). Regardless of field age, for every 10% increase in clay

Fig. 3. The relationships between % clay and several soil C pools: (a) total SOC, (b) unhydrolyzed C, a recalcitrant C pool, and (c) microbial C, a labile C pool. Native prairie sites are not shown.

K.K. McLauchlan / Geoderma 136 (2006) 289–299

295

sample was taken in 2002) the intercept. The relationships between field age and soil response variables were not influenced by sample year, and neither were the relationships between % clay and soil response variables. 3.4. Effects of clay on rate of change over time Clay concentration significantly affected the rate of increase of soil aggregate size, GMD, as indicated by the significant interaction between field age and clay concentration (Table 1). As field age increased, aggregates increased in size faster in soils with higher clay concentrations than in soils with lower clay concentrations. For each 10% increase in clay concentration, the rate of increase of the GMD size index increased by 0.003 year− 1 (Fig. 5). The interaction term for clay concentration and field age was not statistically significant as a predictor for any other response variables tested, including total SOC, microbial C, unhydrolyzable C, Cl, kl, and potential net N mineralization. Thus, these data did not show that clay concentration influences the rate of change of these soil properties over time after cessation of agriculture. 3.5. Native sites constraining changes over time

Fig. 4. Simple linear regressions between % clay and the residuals of linear regressions between field age and (a) GMD, an index of aggregate size, and (b) potential net N mineralization. Native prairie sites were not included in these analyses.

The rates of change in soil properties determined with the previous analyses will not continue indefinitely. Assuming that unplowed native grassland sites represent an equilibrium that the former agricultural fields will reach, soil properties of native prairie sites can be used to estimate when ecosystem properties of plowed fields

linear fit between field age and the two response variables affected by clay concentration: GMD and potential net N mineralization (Fig. 4). Despite the small r2 values of these bivariate regressions, they represent the relationship between clay concentration, GMD, and potential net N mineralization independent of field age. 3.3. Sampling year Because sites were sampled in two different years, I tested the effect of sampling year in the multiple regression. GMD, Cl, and kl were significantly different between sampling years, but the magnitude of this effect was small compared to the other effects (Table 1). These properties are known to vary temporally as they are sensitive to biological conditions and soil microclimate. Because year sampled is the only categorical variable in the multiple regression, the estimates for that predictor variable indicate the quantity that should be added to (if the sample was taken in 2001) or subtracted from (if the

Fig. 5. Modeled relationships between field age and GMD, an index of aggregate size, with different lines for different clay concentrations. The slopes of these lines were calculated using the multiple regression shown in Table 1 with values for clay concentrations and field ages representative of samples. Relationships shown are averages of 2001 and 2002 values.

296

K.K. McLauchlan / Geoderma 136 (2006) 289–299

Table 3 Values of soil C and N variables for native prairie sites Property

GMD SOC Unhydrolyzed C Microbial C Cl kl Net N mineralization

Units

kg ha− 1 kg ha− 1 kg ha− 1 kg ha− 1 d −1 kg ha− 1 d− 1

Average of unplowed site (s.e.)

Time until value of unplowed field is reached (year)

0.869 (0.011) 64,615 (4851) 30,707 (2699) 1372 (55) 3821 (225) 0.008 (0.0009) 0.264 (0.18)

40 80 85 71 89 32 45

The average value is calculated from three native prairie sites; length of time until that value is attained is based on linear rates of change; and native sites are assumed to represent maximum values. The length of time indicates number of years since cessation of agriculture. For these estimates, only field age was included in the regression model.

converted to perennial grasslands will stop changing. To provide some preliminary estimates, the simplest model is that changes in soil properties over time are linear. Interestingly, if a linear rate of change is assumed to project beyond the temporal extent of the data, different properties will take different lengths of time to attain their equilibrium values (Table 3). For example, total SOC pools will reach their maximum capacity approximately 80 years after conversion to perennial grassland, while aggregate size will only take half that time, 40 years, to reach its maximum value (Table 3). Different fractions of soil C pools will take over 70 years to reach maximum value, while kl and potential net N mineralization will take under 50 years to reach equilibrium values (Table 3). The non-linear model could also be used to estimate length of time until equilibrium soil properties are reached, but the derived parameter estimates for this model are extremely sensitive to the values for individual sites, and therefore, the linear fit is more robust (Fig. 1). 4. Discussion 4.1. Field age With increasing time since cessation of agriculture, both labile and recalcitrant SOC pools increased. The overall rate of SOC increase in the first few decades after cessation of agriculture was 516.8 kg C ha− 1 year− 1, which is comparable to other estimates for grasslands (200–600 kg C ha− 1 year− 1) (Post and Kwon, 2000) and forests (Richter et al., 1999) in the first few decades after agriculture. This study suggests that the rate of change in

soil properties could be either linear, with SOC increasing at a constant rate until an equilibrium value is reached, or non-linear, with the rate of increase slowing or saturating over time (Jastrow, 1996). Both of these models provide nearly identical fits for the first 40 years after cessation of agriculture and cannot be distinguished statistically during this time period (Fig. 1). To further resolve the important question of how to project the future trajectory of rates of change in C pool sizes in this system, further data collection, especially measurements of SOC from sites converted from agriculture more than 40 years ago, is necessary. In this study, a linear model suggests that most soil properties can achieve values found on native prairie sites in under a century, but these results should be interpreted with caution. In particular, the linear model predicts the earliest time at which the values found in native sites could be achieved. Labile C pool size, as measured by Cl, increased with field age, although the proportion of total SOC respired in the first 28 days of the laboratory incubation declined with field age, suggesting that a smaller portion of SOC was microbially mineralizable with increasing time since cultivation (data not shown). Recalcitrant C pool size, as measured by unhydrolyzable C, also increased as total SOC increased. However, the ratio of unhydrolyzable C to total SOC remained constant over time since cultivation, indicating that the proportion of C chemically stabilized by clay minerals or biochemically recalcitrant remained constant over time. The proportion of unhydrolyzable C within silt- and clay-sized fractions also stayed constant for samples taken across two different soil texture gradients in other locations in North America (Plante et al., 2006). There are two potential mechanisms that stabilize SOC and could be responsible for the increases in SOC over time observed in this study. Aggregate size increased over time, which was hypothesized to cause an increase in the protection of labile C. Cl, a measure of labile C, is derived from soil respiration in laboratory incubations where there is some physical protection due to aggregates. Nonetheless, labile C inside aggregates becomes available during the course of the incubation (Kristensen et al., 2003) and is included in the estimate of Cl. Previous work that directly examined the relationship of labile C pools to aggregate size on some of these sites in western Minnesota showed weak correlations, indicating that labile C may not become increasingly protected inside aggregates as time since agriculture increases (McLauchlan and Hobbie, 2004). In this study, aggregate size was positively correlated with total SOC and unhydrolyzable C. Acid hydrolysis destroys C with a short turnover time, and therefore, unhydrolyzed C approximates stable SOC

K.K. McLauchlan / Geoderma 136 (2006) 289–299

which is either chemically very resistant or is bound to clay minerals (Paul et al., 2003). The increase in unhydrolyzed C as field age increased indicates that stabilized SOC increased over time and that this pool can change quickly in response to a shift in management. N cycling was also affected by field age. Although total organic N pools increased with field age (data not shown), potential net N mineralization decreased with field age. This decrease has been observed in other studies and could be associated with cessation of belowground disturbance due to tillage (Baer et al., 2000). The ratio of net N mineralized to total SON decreased as fields aged, indicating increased resistance of organic N to decomposition over time. Pastor et al. (1987) found that potential net N mineralization relative to total SON decreased during secondary succession, and that pattern was also observed in this study. While field age was a statistically significant predictor of SOC, Cl, unhydrolyzable C, microbial C, GMD, and potential net N mineralization, there is much variation in these soil properties that remains unexplained by the predictors measured in this study. Several additional potential explanatory variables, such as % slope, base cation content, and vegetation characteristics, do not increase the predictive power of the multiple regression. Therefore, it is likely that the variation in management history (tillage practices, length of time in agriculture, cropping history) among former agricultural fields caused variation in soil properties that could not be accounted for by measurable predictor variables. This type of field variation is likely to add error to estimates of the C storage potential of croplands (Lal, 2004). 4.2. Clay Other studies have shown that clay concentration and SOC content are correlated (Burke et al., 1989; Schjonning et al., 1999), yet no link between clay concentration and SOC pool sizes or rates of formation was shown in this study. These results suggest that on decadal time scales in former agricultural fields with aggrading SOC content, soil texture has little effect on SOC formation. In these systems, which contain far less than their maximum SOC content because of agricultural depletion, clay concentration is less important for SOC content than other factors, particularly time since cessation of agriculture and possibly management history. Thus, the hypotheses that both labile and recalcitrant C pools would increase with increasing clay concentration were not supported. Although clay concentration did not affect SOC pools, it did affect aggregate size, with high concentrations of clay particles promoting the formation of large aggre-

297

gates. Other studies have shown that aggregates contain labile C that is physically protected from microbial decomposition and that large aggregates protect a larger quantity of labile C than does an equal mass of small aggregates (Amelung and Zech, 1999). However, clay concentration, while correlated with aggregate size, was not significantly correlated with C pool sizes in this study. The relationship between clay concentration and soil C pools may be obscured in dynamic former agricultural systems that are in the process of accumulating C, and the effects of clay on soil C pools are more likely to be seen once these soils reach C saturation (Six et al., 2002). As hypothesized, potential net N mineralization was lower in soils with more clay, at constant soil moisture conditions. Other studies have shown that N cycling, specifically net N mineralization, is retarded by clay particles in the soil (Cote et al., 2000). Two parsimonious explanations for this pattern are the physical protection of organic N through clay-enhanced aggregate formation and the direct chemical stabilization of humic substances which have high concentrations of organic N onto clay particles. The increases in aggregate size found here with increases in clay concentration may have been partially responsible for the reduced potential net N mineralization. 4.3. Effects of clay on rate of change over time The hypothesis that the rate of increase in aggregate size would increase with clay concentration was supported because the interaction of field age and % clay significantly increased GMD. However, this study did not support the hypotheses that the rate of accumulation of both labile and recalcitrant soil C pools would increase with increases in clay concentration. The interaction of clay concentration and field age did not significantly increase total SOC, microbial biomass C, Cl, or unhydrolyzable C. There are at least two possible explanations for this result: (1) the range of clay concentrations may have been too narrow to detect an effect on SOC accumulation rates, and (2) soil aggregate dynamics may have been influenced by the relatively low clay concentrations in these soils in western Minnesota. Because macroaggregates contain smaller aggregates (Six et al., 2000) and macroaggregates physically protect otherwise labile C from microbial decomposition, larger aggregates should be associated with greater labile C pools over time. In this study, although GMD and labile C pool sizes (Cl and microbial biomass C) are weakly but positively correlated, it is unclear how the effect of clay concentration on rate of increase in aggregate size may affect labile C pool sizes.

298

K.K. McLauchlan / Geoderma 136 (2006) 289–299

A quantitative estimate of how clay concentrations affect rates of SOC accumulation over decadal timescales is necessary to clarify predictions of SOC change with global land management changes and to correctly parameterize SOM models (Smith et al., 1997; Hurtt et al., 2002). Previous studies have suggested that in addition to affecting SOC content, clay concentrations may determine the capacity of soils to store SOC (Hassink, 1997). However, this study suggests that soil clay concentration in the first few decades after cessation of agriculture does not affect either SOC pool size or rate of accumulation, although clay concentration affects other soil properties. Replicated chronosequences, composed of several sites representing a particular age, have the potential to contribute more data about the effect of clay concentration on rates of SOM formation. Further study of native prairie sites on soil types similar to those used for agriculture can determine whether clay concentration determines equilibrium values of SOM and how long soil properties are likely to change after cessation of agriculture. Acknowledgements Sarah Hobbie, Joe Craine, Karen Walker, and Marissa Weiss provided field assistance. Dorian Hasselmann and Sam Stoxen provided laboratory assistance. I thank Sarah Hobbie, Ed Nater, Deborah Allan, Jennifer King, Peter Reich, Joe Craine, and two anonymous reviewers for extremely helpful suggestions on previous versions of the manuscript. Private landowners and the Fish and Wildlife Service generously allowed me to sample their soils. Financial support came from the Andrew W. Mellon Program for Conservation and the Environment and the Land Institute. References Amelung, W., Zech, W., 1999. Minimisation of organic matter disruption during particle-size fractionation of grassland epipedons. Geoderma 92, 73–85. Anderson, T.H., Joergensen, R.G., 1997. Relationship between SIR and FE estimates of microbial biomass C in deciduous forest soils at different pH. Soil Biology and Biochemistry 29, 1033–1042. Baer, S.G., Rice, C.W., Blair, J.M., 2000. Assessment of soil quality in fields with short and long term enrollment in the CRP. Journal of Soil and Water Conservation 55, 142–146. Burke, I.C., et al., 1989. Texture, climate, and cultivation effects on soil organic matter content in U.S. grassland soils. Soil Science Society of America Journal 53, 800–805. Cambardella, C.A., Elliott, E.T., 1992. Particulate soil organic-matter changes across a grassland cultivation sequence. Soil Science Society of America Journal 56, 777–783. Cote, L., Brown, S., Pare, D., Fyles, J., Bauhus, J., 2000. Dynamics of carbon and nitrogen mineralization in relation to stand type, stand

age and soil texture in the boreal mixedwood. Soil Biology & Biochemistry 32 (8–9), 1079–1090. Cuevas, E., Medina, E., 1988. Nutrient dynamics in Amazonian forest ecosystems II: fine root growth, nutrient availability, and leaf litter decomposition. Oecologia 76, 222–235. Follett, R.F., 2001. Soil management concepts and carbon sequestration in cropland soils. Soil and Tillage Research 61, 77–92. Franzluebbers, A.J., Haney, R.L., Hons, F.M., Zuberer, D.A., 1996. Active fractions of organic matter in soils with different texture. Soil Biology and Biochemistry 28, 1367–1372. Gee, G.W., Bauder, J.W., 1986. Particle size analysis, In: Klute, A. (Ed.), Methods of Soil Analysis, Part I: Physical and Mineralogical Methods, 2nd edition. American Society of Agronomy, Madison, WI, pp. 383–412. Giardina, C.P., Ryan, M.G., Hubbard, R.M., Binkley, D., 2001. Tree species and soil textural controls on carbon and nitrogen mineralization rates. Soil Science Society of America Journal 65, 1272–1279. Hart, S.C., Stark, J.M., Davidson, E.A., Firestone, M.K., 1994. Nitrogen mineralization, immobilization, and nitrification. Methods of Soil Analysis, Part II: Microbiological and Biochemical Properties. Soil Science Society of America, Madison, WI, pp. 985–1018. Hassink, J., 1997. The capacity of soils to preserve organic C and N by their association with clay and silt particles. Plant & Soil 191 (1), 77–87. Hurtt, G.C., et al., 2002. Projecting the future of the U.S. carbon sink. Proceedings of the National Academy of Sciences of the United States of America 99, 1389–1394. Jastrow, J.D., 1996. Soil aggregate formation and the accrual of particulate and mineral-associated organic matter. Soil Biology and Biochemistry 28 (4–5), 665–676. Jenkinson, D.S., 1990. The turnover of organic carbon and nitrogen in soil. Philosophical Transactions: Biological Sciences 329, 361–367. Jenny, H., 1941. Factors of Soil Formation: A System of Quantitative Pedology. McGraw Hill, New York, USA. Kemper, W.D., Chepil, W.S., 1965. Size distribution of aggregates. In: Black, C.A. (Ed.), Methods of Soil Analysis, Part I. American Society of Agronomy, Inc., Madison, WI, pp. 499–510. Kristensen, H.L., Debosz, K., McCarty, G.W., 2003. Short-term effects of tillage on mineralization of nitrogen and carbon in soil. Soil Biology & Biochemistry 35 (7), 979–986. Krull, E.S., Baldock, J.A., Skjemstad, J.O., 2003. Importance of mechanisms and processes of the stabilisation of soil organic matter for modelling carbon turnover. Functional Plant Biology 30 (2), 207–222. Lal, R., 2004. Soil carbon sequestration to mitigate climate change. Geoderma 123 (1–2), 1–22. McLauchlan, K.K., Hobbie, S.E., 2004. Comparison of labile soil organic matter fractionation techniques. Soil Science Society of America Journal 68, 1616–1625. McLauchlan, K.K., Hobbie, S.E., Post, W.M., 2006. Conversion from agriculture to grassland builds soil organic matter on decadal timescales. Ecological Applications 16, 143–153. Muller, T., Hoper, H., 2004. Soil organic matter turnover as a function of the soil clay content: consequences for model applications. Soil Biology and Biochemistry 36, 877–888. Nichols, J.D., 1984. Relation of organic carbon to soil properties and climate in the southern Great Plains. Soil Science Society of America Journal 48, 1382–1384. Parton, W.J., Schimel, D.S., Cole, C.V., Ojima, D.S., 1987. Analysis of factors controlling soil organic matter levels on Great Plains grasslands. Soil Science Society of America Journal 51, 1173–1179. Pastor, J., Stillwell, M.A., Tilman, D., 1987. Nitrogen mineralization and nitrification in four Minnesota old fields. Oecologia 71, 481–485.

K.K. McLauchlan / Geoderma 136 (2006) 289–299 Paul, E.A., Morris, S.J., Bohm, S., 2001. Determination of soil C pool sizes and turnover rates: biophysical fractionation and tracers. In: Lal, R. (Ed.), Assessment Methods for Soil Carbon. Lewis Publishers, Boca Raton, Florida, pp. 193–206. Paul, E.A., Morris, S.J., Six, J., Paustian, K., Gregorich, E.G., 2003. Interpretation of soil carbon and nitrogen dynamics in agricultural and afforested soils. Soil Science Society of America Journal 67 (5), 1620–1628. Paustian, K., Six, J., Elliott, E.T., Hunt, H.W., 2000. Management options for reducing CO2 emissions from agricultural soils. Biogeochemistry 48 (1), 147–163. Percival, H.J., Parfitt, R.L., Scott, N.A., 2000. Factors controlling soil carbon levels in New Zealand grasslands: is clay content important? Soil Science Society of America Journal 64 (5), 1623–1630. Plante, A.F., Conant, R.T., Stewart, C.E., Paustian, K., Six, J., 2006. Impact of soil texture on the distribution of soil organic matter in physical and chemical fractions. Soil Science Society of America Journal 70 (1), 287–296. Post, W.M., Kwon, K.C., 2000. Soil carbon sequestration and land-use change: processes and potential. Global Change Biology 6, 317–327. Richter, D.D., Markewitz, D., Trumbore, S.E., Wells, C.G., 1999. Rapid accumulation and turnover of soil carbon in a re-establishing forest. Nature 400 (6739), 56–58. Schimel, D.S., 1986. Carbon and nitrogen turnover in adjacent grassland and cropland systems. Biogeochemistry 2, 345–357. Schjonning, P., et al., 1999. Turnover of organic matter in differently textured soils: I. Physical characteristics of structurally disturbed and intact soils. Geoderma 89 (3–4), 177–198.

299

Scott, N.A., Cole, C.V., Elliott, E.T., Huffman, S.A., 1996. Soil textural control on decomposition and soil organic matter dynamics. Soil Science Society of America Journal 60, 1102–1109. Six, J., Elliott, E.T., Paustian, K., 2000. Soil macroaggregate turnover and microaggregate formation: a mechanism for C sequestration under no-tillage agriculture. Soil Biology & Biochemistry 32 (14), 2099–2103. Six, J., Conant, R.T., Paul, E.A., Paustian, K., 2002. Stabilization mechanisms of soil organic matter: implications for C-saturation of soils. Plant & Soil 241 (2), 155–176. Smith, P., et al., 1997. A comparison of the performance of nine soil organic matter models using datasets from seven long-term experiments. Geoderma 81 (1–2), 153–225. Staff, S.S., 1997. Soil Survey of Ottertail County, Minnesota, USA. USDA-NRCS, Washington, DC. Tisdall, J.M., Oades, J.M., 1982. Organic matter and water-stable aggregates in soils. Journal of Soil Science 33, 141–163. Vinton, M.A., Burke, I.C., 1997. Contingent effects of plant species on soils along a regional moisture gradient in the Great Plains. Oecologia 110 (3), 393–402. Wang, W.J., Dalal, R.C., Moody, P.W., Smith, C.J., 2003. Relationships of soil respiration to microbial biomass, substrate availability and clay content. Soil Biology and Biochemistry 35 (2), 273–284. Wedin, D.A., Pastor, J., 1993. Nitrogen mineralization dynamics in grass monocultures. Oecologia 96 (2), 186–192. West, T.O., Post, W.M., 2002. Soil organic carbon sequestration rates by tillage and crop rotation: a global data analysis. Soil Science Society of America Journal 66, 1930–1946.