HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS		QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

This Article Abstract Figures Only Full Text (PDF) Free Alert me when this article is cited Alert me if a correction is posted Services Similar articles in this journal Alert me to new issues of the journal Download to citation manager Google Scholar Articles by Leifeld, J. Articles by Oberholzer, H.-R. PubMed Articles by Leifeld, J. Articles by Oberholzer, H.-R. Agricola Articles by Leifeld, J. Articles by Oberholzer, H.-R. Related Collections Isotopes Soil Organic Matter Organic farming Agricultural Systems Soil Chemistry Animal Waste

Consequences of Conventional versus Organic farming on Soil Carbon: Results from a 27-Year Field Experiment

Agroscope Reckenholz-Tänikon Research Station ART, Reckenholzstrasse 191, 8046 Zurich, Switzerland

^* Corresponding author (jens.leifeld{at}art.admin.ch).

	ABSTRACT

TOP NOTES ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION CONCLUSIONS REFERENCES

 
Organic farming practices are regarded as being beneficial for the environment by promoting soil quality and sequestering soil organic carbon (SOC). We studied SOC dynamics in the long-term field experiment DOK in Switzerland. The experiment compares three organically fertilized treatments under conventional (CONFYM), bioorganic (BIOORG), and biodynamic (BIODYN) management, and two systems with (CONMIN) or without (NOFERT) mineral fertilizer. We analyzed measured SOC time series from 1977 to 2004 and applied soil fractionation, radiocarbon dating, and modeling with the carbon model RothC. The SOC declined significantly in most parcels, but was not systematically different between systems. Initial SOC contents correlated with soil texture and were identified as being important with respect to the change rate. The SOC loss was at the expense of mineral-associated carbon whereas the more labile fractions increased. The overall decline was explained by reduced carbon inputs since commencement of the experiment and was most pronounced in NOFERT and CONMIN. The model satisfactorily simulated the dynamics of most of the treatments for both initialization with equilibrium runs or measured SOC fractions. Carbon loss in CONFYM was not fully captured by the model. Composition of organic fertilizers depended on the particular management, and a model adjustment of their relative stability improved the match between model and measurements. Model runs without management effects indicated that the observed increase in temperatures at the experimental site does not induce a change in SOC. Overall, the study does not support a benefit of organic farming on SOC contents compared with conventional systems with manure.

Abbreviations: AMS, accelerator mass spectrometry • BIO, microbial biomass • BIODYN, biodynamic farming • BIOORG, bioorganic farming • CONFYM, conventional farming with farmyard manure • CONMIN, conventional mineral fertilizer • D_b, bulk density • DOC, dissolved organic carbon • DPM, decomposable plant litter pool • HUM, humified organic matter • IOM, inert organic matter • NOFERT, no fertilization • POM, particulate organic matter • RothC, Rothamsted carbon model • RPM, resistant plant material • rSOC, oxidation residue • SOC, soil organic carbon • S+A, sand and microaggregate fraction • s+c silt and clay fraction

Received for publication January 8, 2009.

	INTRODUCTION

TOP NOTES ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION CONCLUSIONS REFERENCES

 
ORGANIC FARMING SYSTEMS differ fundamentally from conventional ones in terms of fertilization, pest management, crop rotation, and the use of organic fertilizers (Elmaz et al., 2004). One goal of organic farming is to reduce the dependency of agricultural production on external inputs. The utilization of legumes to promote biological N inputs and of catch crops to reduce nutrient losses is quite common as is the use of manure for returning nutrients and organic matter to the soil as well as the absence of soluble mineral fertilizers and pesticides. Nitrogen fixation is an essential N input path in organic farming, and therefore arable organic farms in temperate regions often include animal husbandry with animals feeding on grass–clover mixtures or other N-fixing feed crops (Stockdale et al., 2001). Many of the organic practices are regulated by national policies and controlled by national or supranational organizations promoting organic agriculture.

It is frequently claimed that organic farming provides environmental benefits, many of which relate to soil properties. Adopted crop rotations may reduce soil erosion (Eltun et al., 2002). Replacement of mineral fertilizer and pesticides by organic fertilizers enhance soil biological activity, efficiency, and rate of microbial substrate use (Gunapala and Scow, 1998; Fließbach et al., 2007). Positive effects of organic farming on soil carbon have been reported repeatedly (e.g., Drinkwater et al., 1998; Liebig and Doran, 1999; Wells et al., 2000), whereas other groups found no or only inconsistent effects of organic vs. conventional farming on organic matter content in soil (Friedel, 2000; Shannon et al., 2002; Marinari et al., 2006). Some of the contradicting results may stem from fundamental differences in the experimental setup. Studies reporting on management effects often compare systems that differ in many respects from each other; for example, in crop rotation, rates of application of organic fertilizers or tillage (Robertson et al. 2000; Zaller and Köpke, 2004; Teasdale et al. 2007). This makes it difficult to ascribe consequences to causes. Ideally, identical rotations, equal use of organics, similar residue management and tillage are applied to decipher mechanisms behind observed differences in SOC that otherwise can only be discerned generally as management-induced.

One of the best-documented, longest-lasting planned comparisons of organic vs. conventional farming is the DOK (D: biodynamic, O: bioorganic, K: conventional) experiment in Switzerland (Mäder et al., 2002). In a 7-yr crop rotation study, conventional systems with or without organic fertilizer are compared with organic and biodynamic farming practices. Crop rotation, tillage, and residue management are the same and the application rates of organic fertilizers in systems mimicking animal husbandry are at similar levels. Management practices that represent the particular system in terms of fertilization, manure treatment, and pest control are performed according to the respective regulations and are considered as representative of Swiss agriculture. Treatment-induced effects related to the soil resource have been reported frequently for the DOK experiment. Eighteen years after the experiment commenced, Fließbach and Mäder (2000) measured significantly higher microbial biomass (BIO) for biodynamic (BIODYN) soils and a higher C concentration in particulate organic matter (POM) compared with CONFYM or organic treatments without biodynamic preparations (BIOORG). A higher BIO coincided with a faster mineralization of labeled residue in BIODYN after two rotation periods (Fließbach et al., 2000) and with smaller metabolic quotients. Those differences were considered to be indicative for a higher substrate use efficiency of the soil BIO in BIODYN, and were also confirmed by measurements after three rotations (Fließbach et al., 2007). Similarly, Widmer et al. (2006) reported significant effects for the three organically fertilized systems with respect to BIO and soil DNA content, whereas they found differences in community-level substrate utilization only between organically fertilized systems on the one hand, and the treatment with mineral fertilizer (CONMIN) or NOFERT on the other. Organic farming, both BIODYN and BIOORG, promoted root colonization by mycorrhiza, activity of exoenzymes, and also increased the aggregate stability relative to CONFYM or CONMIN (Mäder et al., 2002). Significant declines in SOC over three rotations (21 yr) were found for most treatments (Fließbach et al., 2007). However, BIODYN values declined not at all or significantly less than CONFYM and BIOORG; SOC contents decreased even more in CONMIN and the greatest in NOFERT.

The reasons for (i) the overall decline in SOC and (ii) differences in SOC among the management systems still have to be unraveled. Differences in microbial activity and community composition probably affect SOC in the long term. Also, treatment-specific preparation of the organic fertilizer, particularly manure composting, may influence soil organic matter due to changes in manure stability. Other important drivers include differences in initial conditions (SOC, texture) and residue inputs over the course of the experiment. In addition, climatic change has been held responsible for SOC declines elsewhere (Bellamy et al., 2005) and may also have an impact on SOC in the DOK trial.

The goal of this study is to acquire a deeper understanding of the SOC dynamics in the DOK experiment and of management effects on SOC in general. We hypothesize that organic farming is NOT beneficial over conventional farming in terms of carbon sequestration if crop rotation, tillage, and amount of organic fertilization are the same. We wanted to know how the initial conditions differed between treatments; how this may affect the change in SOC over time and examined management effects on the distribution of SOC among soil fractions of different stabilities. We paid particular attention to a sound estimate of the amount and quality of carbon inputs and explored whether the measured SOC dynamics in various management systems, differing mainly in fertilization and plant protection strategy, could be reasonably represented by a common soil carbon turnover model, that is, if the model reproduces management-induced differences. The model we used was the Rothamsted Carbon Model RothC 26.3 (Coleman and Jenkinson, 1999), which has been applied to long-term agricultural field experiments before (e.g., Smith et al., l997). Finally, we discuss how different procedures for model initialization affect the simulated time course of SOC in the DOK experiment and weigh the possible effects of climate change against those of management.

	MATERIALS AND METHODS

TOP NOTES ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION CONCLUSIONS REFERENCES

 
DOK Experiment: Setup, Treatments, and Sampling
The DOK experiment was started in 1978 in Therwil, Switzerland (7°33' E, 47°30' N). It compares management systems that differ mainly in type and intensity of fertilization and in methods of plant protection, but have identical crop rotations. The soil is a Typic Argalf (mesic); (haplic Luvisol according to WRB) on alluvial loess with an average texture of sand/silt/clay of 12:72:16. The mean annual temperature of the site was 9.7°C and the mean annual precipitation 791 mm (period 1864–2007). The preexperimental management included 3 yr of field vegetables, grain crops from 1973 to 1976, and a diverse arable rotation with leys and manure amendment back to 1957 (Fließbach et al., 2007). In 1976, the whole field of around 1.5 ha was under oats followed by grass–clover in 1977.

The DOK experiment comprises 96 parcels (eight treatments times four replicates times three crops planted simultaneously in each system every year) arranged in a split–split–block design. Here we report on results from five treatments, each replicated four times in the field and cultivated with the same crop in a given year. NOFERT has received neither organic nor inorganic fertilizer since 1978; CONMIN is fertilized with mineral fertilizer only at a rate recommended for integrated farming, but was not fertilized in the first rotation, BIODYN applies biodynamic farming (i.e., it includes application of biodynamic preparations and composting of manure) without addition of mineral fertilizer and pesticides, BIOORG applies organic farming without addition of mineral fertilizer and pesticides, and CONFYM is a conventional system receiving both mineral and organic fertilizers. Both BIODYN and BIOORG use mechanical weed control. Organic fertilizer in BIODYN, BIOORG, and CONFYM is applied as manure and slurry. The share (mean over four crop rotations) of manure to the total organic fertilizer applied is 72, 83, and 77% for BIODYN, BIOORG, and CONFYM and over four rotations organic fertilizer was applied in 59, 60, and 42 doses in BIODYN, BIOORG, and CONFYM, respectively.

All five treatments have the same type and frequency of tillage (moldboard plowing 18–20 cm deep and harrowing). Since 1985, CONMIN and CONFYM were farmed according to the Swiss national guidelines of integrated plant production. Pesticides were only applied if economic thresholds for infections were exceeded according to the integrated scheme of plant protection. Plant protection was conducted according to respective guidelines for the biodynamic and bioorganic systems in BIODYN and BIOORG (Lampkin, 1990). Plant protection in the unfertilized system NOFERT was the same as in BIODYN. An important characteristic of the three organically fertilized systems is their specific treatment of cattle manure. CONFYM uses stacked manure that is only slightly decomposed, BIOORG is amended with matured manure, and BIODYN uses composted manure. The rate of manure application corresponds to a stocking density of 1.2 during the first crop rotation or 1.4 livestock units in the following years.

Crop rotations were the same for all treatments and included potato (Solanum tuberosum L.), winter wheat I (Tristicum aestivum L.), white cabbage (Brassica oleracea L.), winter wheat II, winter barley (Hordeum vulgare L.), and grass–clover (2 yr; major species Poa pratensis L., Lolium perenne L., Festuca pratensis Huds., Dactylis glomerata L., Trifolium repens L., Trifolium pratense L.). White cabbage was replaced by beetroots (Beta vulgaris L.) after the first rotation, and soybean [Glycine max (L.) Merr.] followed by silage maize (Zea mays L.) were introduced in the fourth rotation after winter wheat II. The same species and varieties were cropped in all treatments. Green manure and fodder intercrops were frequently sewn after potato and winter wheat I. Straw from cereals was always removed. A detailed description of the DOK experiment can be found in Mäder et al. (2002) and Fließbach et al. (2007).

From a total of 96 parcels, 20 are studied here (five treatments times four replicates). The time course of bulk soil carbon was evaluated using 16 dates between 1977 and 2004. A detailed analysis included soil fractionation and was obtained using archived soil samples from 1977 (grass–clover cropped before the start of the DOK trial in 1978) and from 2004 when the selected parcels were again under grass–clover after completion of four crop rotations. The crop type was considered as critical possibly having an impact on the amount and distribution of carbon in the sampled soil, and therefore kept the same at both samplings. Results from 15 out of the 16 dates refer to the same parcels as sampled in 2004. However, the temporally shifted crop rotations forced us into using surrogate samples from the preexperimental year 1977, since some samples from the original parcels were already exhausted in our sample archive. Thus, samples from the respective adjacent parcels in the same block (distance between parcel centers is 5 m, parcel area is 5 by 20 m) were used as surrogate, which were considered to be equivalent in terms of the amount and distribution of SOC in the preexperimental state. We tested the equality of the surrogate sample parcelwise using a soil texture screening in 2006. A paired t test gave no significant difference in granular composition, and we therefore also consider effects of the shift by 5 m on soil carbon concentrations in 1977 to be negligible. A paired t test for carbon contents in 1977 was not possible because the preexperimental SOC contents, as for example reported by Fließbach et al. (2007), refer to block-wise composite samples, that is, they include a mixture of materials from six parcels belonging to the same block. Similarly, NOFERT and CONMIN were not differentiated on preexperimental sampling in 1977, but taken as composite samples from the same blocks. Therefore, the number of samples in 1977 was n = 16, but n = 20 in subsequent years. All soil data refer to the upper 20 cm because a consistent time series of measured soil carbon contents was available for that depth only and because it corresponds to the plowing depth.

Carbon Contents of Soils, Plant Residues, and Organic Fertilizers
The soil carbon content of the DOK experiment is commonly measured spectrophotometrically after wet oxidation (see Fließbach et al., 2007). For samples from 1977 and 2004, we additionally measured the C content by elemental analysis (oxidation in O₂, and quantification of CO₂ by GC-TCD, Hekatech, Germany) in the same way as soil fractions. Elemental analysis gave slightly higher concentrations (factor 1.036) than wet oxidation. A consistent dataset was obtained by applying this correction factor to the conversion of wet oxidation to total organic carbon for all bulk soil measurements between 1977 and 2004. A carbon fraction of 0.44 was assumed for dry plant litter.

At each application, organic matter contents of organic fertilizers were measured by loss-on-ignition after determination of the water content and total nitrogen by elemental analysis. Based on data presented by Petersen et al. (1998) and Ammann et al. (2007), we estimate the carbon content of the organic matter in organic fertilizer to be 50% for both slurry and manure. The different treatments of manure (see above) result in different organic matter contents and C/N ratios as indicators for manure stability (Table 1 ), which were used to shape the distribution of carbon among pools in the RothC model.

The surface area of the s+c fraction before and after oxidation was determined by N₂ adsorption using a Quantachrome Nova 2200 [Quantachrome, Odelzhausen (Germany)] surface analyzer and BET isotherm. Soil texture was measured using the standard pipette method after removal of organic matter by H₂O₂.

Radiocarbon Measurements
Radiocarbon contents in bulk soil samples and in rSOC fractions from three parcels sampled in 1977 and 2004, each representing CONMIN, BIODYN, or CONFYM, respectively, were measured in 2008 at the accelerator mass spectrometry (AMS) laboratory of ETH Zurich. All measurements were normalized to a ¹³C of –25 to compensate for possible isotopic fractionation, and the results are reported as percent modern carbon absolute after correcting for ¹⁴C decay since sample collection.

Modeling
We used the Rothamsted Carbon Model RothC 26.3 (Coleman and Jenkinson, 1999) for the simulations. The model simulates the turnover of SOC using a five-pool structure. Four of the pools (DPM, decomposable plant material; RPM, resistant plant material; BIO; and HUM) are defined by first-order decay with decreasing decomposability in the order DPM > BIO > RPM > HUM. In contrast, in the last pool, IOM is not transformed and represents a stable baseline. Decay rate constants are modified by temperature, soil moisture deficit, a simplified measure of residue quality, and indirectly by clay content, which modifies the substrate use efficiency of the BIO toward higher efficiencies at a higher clay content. The maximum soil moisture deficit is calculated for a given clay content (potential topsoil maximum deficit is 62.5 mm at 65% clay). The difference between monthly precipitation and evapotranspiration, the latter being different for bare and vegetated soil, gives the calculated actual topsoil moisture deficit with a maximum deficit as defined by the clay content.

Three different approaches were used to define the distribution of carbon among pools in the model (starting conditions) in 1977. In the first, we took measured SOC contents in 1977 and used the models' inverse mode to calculate the amount of plant inputs to reach the conditions in 1977. For these runs, we used averaged monthly climate data from 1901 to 1976. We assumed an annual input of 1 t manure-C per hectare and crop rotation without winter cover. These conditions best match the preexperimental site history as described above. The size of the IOM pool was estimated by the model according to Falloon et al. (1998) as a fixed share of the total. The first approach gives a distribution of C among pools at the end of 1976, which is referred to as equilibrium.

As a second approach, we defined the carbon in measured physical and chemical soil fractions as the starting value for the RothC pools in 1977. Briefly, the combined DPM and RPM pool of the RothC model correspond quantitatively to carbon in the combined POM and DOC fraction, the HUM pool to carbon in the combined S+A and s+c fraction, and the IOM pool to rSOC. The distribution of C between DPM and RPM is calculated as a function of the mean annual temperature. This method was introduced by Zimmermann et al. (2007), who revealed a good correspondence between the soil fractions as described above and the pools of the RothC model under equilibrium assumptions and argued that "measured fractions reflect better than any model the conditions under which SOC is accumulated," mainly because the preexperimental management history is often not or only inadequately known. We used the DOK experiment to validate this hypothesis.

Third, the model was again run in the inverse mode, but measured ¹⁴C values of the bulk soil organic matter (SOM) of three parcels from 1977 and 2004 were taken as an additional constraint. In the model, the radiocarbon content is used to adjust the size of the IOM pool to match the measured bulk soil ¹⁴C. The third approach was applied to only those three parcels for which ¹⁴C data were available. The original ¹⁴C time series of RothC was partially replaced by atmospheric measurements from the ‘Schauinsland’ station, which is relatively close to our site (distance 48 km), and a time series is available from 1977 to 2004 (Levin and Kromer 2004; Levin et al., 2008). The same three parcels were used to run simulations under the real climate from 1977 onward and under a reference climate referring to the period from 1901 to 1976 to distinguish possible effects of climate change on SOC from management-induced effects.

Based on the measurements in Table 1, we calculated average carbon losses during manure and slurry storage of 39, 32, and 21% for BIODYN, BIOORG, and CONFYM, respectively, assuming the initial organic matter (OM) content of plant litter was 95%. We estimated the corresponding pool distribution between DPM, RPM, and HUM to be 59, 39, and 2% for conventional organic fertilizer, and 29, 66, and 5% for composted organic fertilizer consisting of slurry and composted manure (i.e., BIODYN). We adopted the distribution of carbon among the three manure pools from the model default (i.e., 49, 49, 2%) for BIOORG.

All simulations were run parcelwise, that is, with the clay content and the initial SOC content of the respective parcel. The measured mean bulk density (D_b) from 32 parcels in 1977 (0–20 cm) was 1.32 (1 SE = 0.01). These parcels were not included in the present study but were situated adjacent to our parcels. The D_b at that time showed significant correlation with the soils' silt and clay content (r = +0.48, P < 0.01 and r = –0.42, P < 0.05, respectively). We used a multiple regression with percentage clay and silt as predictors to estimate bulk densities of all parcels of the present study and converted concentrations to carbon masses. For the 20 parcels of the current study, the estimated mean D_b based on the regression is 1.32 (range: 1.26–1.35), that is, the same as that for 32 measured parcels in 1977. We assumed constant D_b over time.

Climate data from the nearby weather station in Basel-Binningen (located about 7 km away from the experimental field and at the same elevation) from 1901 to 1976 were taken to simulate measured carbon stocks in 1977, which were considered to be in equilibrium. The actual monthly weather data were used (Table 2 ) from 1977 onward (i.e., commencement of the experiment plus the 1977 ley year).

We use the term pool for model pools of the RothC model, and fraction for physically available and measurable subunits of soil. Stock or carbon stock refers uniquely to the total carbon mass in the upper 20 cm either measured or modeled. In the latter case, we talk about modeled carbon stocks.

Statistics
Linear regression coefficients were computed for carbon stocks vs. time and were tested for significant deviation of the slope from zero. For this, the 16 measurements available for each parcel were analyzed and compared with the modeling results using simulated carbon stocks from the same months as soil samplings took place. Differences between treatments were evaluated by the Kruskal–Wallis test. In the tables, a number followed by a star indicates a significant level at the 0.05 probability level. The measured fractions for the years 1977 and 2004 were compared with model pools for the same years across treatments using the Mann–Whitney U test. Spearman's rank correlation coefficient was calculated to check for correlation between carbon and clay contents. Numbers followed by values in parentheses indicate one standard error of the mean.

	RESULTS

TOP NOTES ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION CONCLUSIONS REFERENCES

 
Initial Conditions and Carbon Inputs
Soil carbon stocks in 1977, before the experiment was started, were 42.9 (±1.1) t ha^–1 C (0–20 cm), ranged from 34.6 to 51.7 t (parcel level), and were highest in BIODYN and lowest in CONFYM (Table 3 ). Clay contents (16.3 ± 0.5) ranged from 12.7 to 25.7% (parcel level). SOC concentrations showed significant correlation with the clay content, both at the beginning of the experiment and after 27 yr (1977: r = 0.65, P < 0.01, n = 16; 2004: r = 0.60, P < 0.01, n = 20). The initial carbon and clay contents (assuming the latter was the same in 1977 and the year of measurement, 2006) did not differ significantly among parcels corresponding to treatments imposed subsequently.

View larger version (31K): [in this window] [in a new window]   Fig. 1. Distribution of carbon among measured soil fractions in 1977 (black) and 2004 (open). Error bars are 1 SE. x axis: NOFERT = no fertilization, CONMIN = conventional farming with mineral fertilizer only, BIODYN = biodynamic treatment, BIOORG = bioorganic treatment, CONFYM = conventional farming with mineral and organic fertilization. y axis: POC = particulate organic carbon, DOC = dissolved organic carbon, rSOC = carbon in silt and clay fraction resistant to oxidation with NaOCl, S+A = carbon in sand fraction after removal of POC, s+c = carbon in silt and clay-sized fraction.

The surface area of rSOC correlated significantly with the clay content (r = 0.79, P < 0.01) and did not change between 1977 and 2004.

Comparison of Initial Conditions in the Model: Equilibrium Runs vs. Initialization with Measured Fractions
The distribution of carbon among pools based on equilibrium runs (first approach) was systematically different from values based on soil fractionation (second approach; Table 5 ). Relative differences were biggest for DPM; however, DPM and BIO pools have fast turnover rates, making the simulation quite insensitive to their initial pool size. More important are the higher shares of C found in fractions that correspond with the stabilized HUM pool (i.e., the s+c plus S+A fraction). This difference was mostly at the expense of a much smaller share of C in the IOM and RPM fractions. When the model was initialized with measured SOC and radiocarbon values (third approach; n = 3), the correspondence between simulated (1.2 ± 0.1) and measured (1.3 ± 0.2 t ha^–1 C) IOM became very close.

View larger version (25K): [in this window] [in a new window]   Fig. 2. Change in SOC stocks over time (0–20 cm). Upper and lower limits as shaded areas or dashed and dotted lines display mean ± 1 SE based on simulations of single parcels (n = 4) and refer to different modes of model initialization and to modified stabilities of manure in biodynamic treatment BIODYN and conventional treatment with mineral and organic fertilizer CONFYM. Triangles show the measured carbon ± 1 SE error bar for 16 sampling dates.

View larger version (12K): [in this window] [in a new window]   Fig. 3. Measured annual soil carbon loss as a function of annual inputs averaged over 27 yr. Error bars are 1 SE. NOFERT = no fertilization, CONMIN = conventional farming with mineral fertilizer only, BIODYN = biodynamic treatment, BIOORG = bioorganic treatment, CONFYM = conventional farming with mineral and organic fertilization.

One plausible explanation for the slight over- and underestimation of change rates by the model in BIODYN and CONFYM, respectively, may be a difference in manure stability (i.e., degree of decomposition). Simulations marked in dark gray in Fig. 2 were run with fixed shares of the different manure pools that comprise the organic fertilizer in the model because the real distribution among pools was unknown. However, data on organic matter content and C/N ratios indicated that the composted manure in BIODYN was on average more stabilized compared with the manure in BIOORG (moderately decomposed) or CONFYM (stacked; only weakly decomposed). Manure in BIODYN contained less organic matter and had a smaller C/N ratio than manure in CONFYM, with manure in BIOORG situated in the middle (Table 5). We adjusted the pools of the manure for BIODYN and CONFYM (see Materials and Methods) to better match its real composition. For BIODYN, stabilized manure lead to higher and more realistic simulated carbon stocks (Fig. 2, dotted lines), whereas stocks simulated in CONFYM responded only weakly to the altered manure composition. After adjustment of the manure composition in BIODYN, the change rate of simulations and measurements became similar (Table 5); however, simulations of CONFYM still revealed no significant decline at the treatment level in contrast to the measurements.

Over time, the carbon content changed both in the four dynamic pools of RothC (i.e., simulated) and in the measured soil fractions isolated from bulk soil samples taken in 1977 and 2004. For three model pools, namely DPM, BIO, and HUM, the change across treatments (t ha^–1 C) correlated significantly with the change in the corresponding measured fraction (Table 7 ). The absolute change was, however, significantly different between pools and measured fractions for all but the HUM pool and its corresponding S+A and s+c fractions, where the change in carbon between 1977 and 2004 did not differ significantly between model pool and measured fractions (Table 7). The latter were also the only ones where the size of the model pool and the measured fractions across treatments were not significantly different in 2004 (P = 0.078; U-test) whereas differences were significant for 1977 and for all other pairs of model pool-measured fractions both in 1977 and in 2004.

View larger version (16K): [in this window] [in a new window]   Fig. 4. Comparison of soil carbon stocks simulated with (i) real climate and management (solid) and (ii) real climate but without management change after 1977 (dashed). Numbers at curve onsets indicate corresponding parcels (6, conventional farming with mineral fertilizer-only CONMIN; 24, conventional farming with mineral and organic fertilization CONFYM; 50, biodynamic treatment BIODYN).

View larger version (17K): [in this window] [in a new window]   Fig. 5. Simulated (lines) and measured (symbols) soil radiocarbon contents for single parcels of conventional farming with mineral fertilizer only (CONMIN), conventional farming with mineral and organic fertilization (CONFYM), and a biodynamic treatment (BIODYN). The atmospheric concentration (right scale) is plotted for comparison. Triangle CONMIN, open circle CONFYM, closed circle BIODYN. Runs refer to parcel numbers 1 (CONMIN), 1 (CONFYM), and 3 (BIODYN) in Table 6. Error bar for measured absolute percentage modern carbon represents the analytical error (1 SD) in the laboratory.

	DISCUSSION

TOP NOTES ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION CONCLUSIONS REFERENCES

In their study, Fließbach et al. (2007) reported significant differences in the change of SOC contents in BIODYN compared with BIOORG and CONFYM for the third rotation period (not significant for the first 14 yr); that is, they reported treatment effects. In contrast, we found no significant difference in the SOC change rate among treatments over 27 yr (Table 1). This difference may be ascribed to the higher sample number used by Fließbach et al. (2007), scatter in measured C contents, and to the longer measurement period in the present study. Over time, each system probably approaches a new equilibrium, thereby reducing the annual change rate.

Why is there a significant decline in soil carbon for many of the parcels? One clear indication can be found by comparing loss rates vs. carbon inputs (Fig. 3). Preexperimental inputs, as estimated by the model run in reverse mode, were higher than in any of the treatments thereafter, and loss rates correlated significantly with annual inputs, indicating that at least part of the SOC losses can be attributed to lower carbon inputs. Our input estimates are based on published allocation coefficients and measured yields. Both the temporal SOC dynamic and absolute C content are simulated reasonably well for many parcels and we consider the magnitude of inputs to be realistic, albeit they have an unknown uncertainty. For example, root-to-shoot ratios may differ between treatments and were shown to be slightly higher under nutrient depletion in a long-term experiment elsewhere (Wilts et al., 2004).

The annual input declined considerably by 0.8 (CONFYM) to 2.5 (NOFERT) t C ha^–1 relative to the period before 1977, and was smallest for the two systems without organic fertilization when plant residue input and organic fertilizer were taken together. In the conventional systems, CONMIN and CONFYM, yields of leys and winter wheat showed no trend over four crop rotations (Gunst et al., 2007), which may indicate that their productivity was close to the preexperimental situation when the field was also managed conventionally. Thus a yield depression was unlikely to be the cause of the reduced input in CONMIN and CONFYM after 1977. Yields in BIODYN and BIOORG (averaged over all crops) were 80% of those of the conventionally managed DOK systems over the same period (Gunst et al., 2007), tentatively indicating that the productivity of today's organic systems is smaller than it was before 1978 when the whole field was managed conventionally. Smaller yields may reduce inputs in the two organic systems relative to the preexperimental situation. During the experiment, however, estimated plant inputs were higher only for CONFYM, whereas they were similar for CONMIN, BIOORG, and BIODYN (Table 4). Yield differences were pronounced for crops whose input to soil is small (e.g., potatoes), whereas leys, which contribute the highest residue inputs, were often not significantly different in yield between organic and conventional farming over four crop rotations (Gunst et al., 2007). Therefore, productivity explains only some of the difference in C input in the fertilized systems. A conceivably smaller productivity in NOFERT probably caused the drop in SOC in that system, which can be considered to be illustrative only and without having much practical relevance.

Besides changes in productivity, changes in residue management (straw left on-site) or manure input may have contributed to reduced carbon inputs since 1977 for all treatments. During the experiment, all straw has been removed. The annual straw yield since 1978, averaged over all crops including those that do not produce straw, was 0.7 to 1.1 t C ha^–1. If straw yields were similar before 1978 and, in contrast with the experimental period, all straw has been left on site, such a change in management practice may explain a good deal of the reduced inputs. Straw accrual may have been even higher given that the current crop rotation is probably less grain-dominated than in earlier times. A second reason might be a shift in organic fertilizer allocation within the farm. Whereas organic fertilizer is spread in the DOK in accordance with the current rotation, the field was part of a whole farm before 1977. At that time, farm-internal manure redistribution from permanent grasslands to arable fields may have been common practice, thereby enhancing the SOM stock. A close relationship between inputs and carbon content agrees with observations made elsewhere (e.g., Buyanovsky and Wagner, 1998); however, changes in soil tillage and legacy of previous management changes may also contribute to the overall SOC dynamics. Based on available information, we consider these two latter factors to be of minor importance, although an effect on the observed time course cannot be excluded.

One objective of our study was to unravel possible impacts of initial soil conditions on the fate of SOC over time because both site conditions and the history of land-use or management may be a major driver for current SOC dynamics (Smith et al., 2007). Soil carbon stocks in 1977 ranged from 34.6 to 51.7 t. This range is partially explained by the highly significant correlation with the clay contents (spanning from 12.7 to 25.7%) due to the stabilizing effect of the clay fraction. For example, we found highest initial C-stock and a slightly heavier texture in BIODYN on the one hand and smallest initial C-stocks and a lighter texture in CONFYM on the other. Measurements of the soils' surface area corroborated this finding. However, most of the variability in carbon stocks is explained by differences in carbon inputs before 1977, which are probably related to productivity and thus indirectly also to soil chemical and physical attributes. Taken together, differences in SOC stocks at the beginning of the experiment were entirely driven by the spatial heterogeneity in soil attributes and productivity of the field, a conclusion which can be drawn from the uniform management of the field before 1977. We suggest that such site-specific initial differences in soil attributes may have had an effect on the subsequent carbon dynamics in this field experiment. Notably also in 2004, after 27 yr of different management, C contents were still significantly correlated to percentage clay, indicating that management practices did not fully mask textural effects. The variability of initial carbon and clay content may explain why we found significant trends over time for many parcels, but no significant differences among treatments.

Differences in microbial parameters as well as in manure preparation have been discussed as possible causes for management-induced differences in SOC, not only for the DOK experiment, but also for other field trials (Reganold et al., 1993; Pulleman et al., 2003, Melero et al., 2006). Our data lend no support to this. They rather suggest that the higher stability of the composted manure had no measurable effects on SOC dynamics because the decline was not significantly different among the three organically fertilized systems. Specifically, the data for BIODYN, BIOORG and CONFYM in Table 3, indicate almost identical differences in C stocks between 1977 and 2004; that is, the differences between parcels in 1977 were maintained over 27 yr of different management. It seems as if the nominally smaller carbon input into BIODYN, which is a result of carbon losses during manure composting, was compensated by the higher shares of stable organic matter in the composted manure. In other words, the easily degradable organic fraction in the manure used in BIODYN had been decomposed during stocking, whereas it was decomposed in the fresh manure only after application in BIOORG and CONFYM.

Carbon Dynamics over Time—Model Simulations
Accurate simulation of treatment effects requires accurate transformation of the degradability of the organic fertilizers when defining the input pools. The RothC model reproduces trends in SOC over time for biodynamic farming as well as for BIOORG, NOFERT, and CONMIN without the need to consider factors other than initial carbon pool distribution, meteorological conditions, soil texture, cropping system, and amount and type of organic fertilization. After changing the carbon shares in the simulated input pools toward higher fractions of stable carbon to account for the more stable fertilizer in BIODYN, the simulated carbon loss rate was close to the measured loss rate, whereas the loss rate was overestimated when the model default settings for carbon shares the input pools were used.

On the other hand, the significant decline of SOC in CONFYM was not reproduced by the model, even after adjustment of the organic fertilizer composition toward a higher share of labile compounds, and this system is the only one whose trend was only adequately simulated by RothC. Similar to BIODYN, spatial effects within the field may have an effect, given that the starting SOC contents were smallest in CONFYM across all treatments. CONFYM also has the highest metabolic quotients (Fließbach et al., 2007), that is, more CO₂ per unit BIO is respired. This effect, which has been attributed to a higher maintenance requirement of the BIO in CONFYM, may also result from priming effects caused by the much higher nutrient inputs in CONFYM compared with BIOORG and BIODYN. Higher nutrient loads in CONFYM were found by Mäder et al. (2002), who showed that inputs of soluble N exceeded those of the two other organically fertilized systems by a factor of three, and for K and P by a factor of 1.5 to 2.0. Khan et al. (2007) discussed a high-N input as a possible cause for accelerated SOM decomposition and consequently SOC decline in a long-term experiment in Illinois, USA. Similar mechanisms may be responsible for the observed soil carbon decline in CONFYM, which are not reproduced by RothC. Also rhizodeposition and thus belowground inputs may be smaller in CONFYM than in systems with lower nutrient input; however, experimental evidence is lacking. None of these processes is explicitly accounted for in the RothC model which may explain the difference in simulated and measured trends in CONFYM.

Warming may have contributed to the overall deterioration because the mean annual temperature at Basel-Binnigen increased from 9.5°C (1901–1976) to 10.3°C (1977–2004). Only those three parcels where ¹⁴C measurements were available for 1977 and 2004 as model validation (see below) were taken to identify possible effects of climate change. The results consistently show no effect of warming alone on simulated SOC (Fig. 4). In the model, the observed increase in the mean annual temperature may be too small to detect its effect on soil carbon against management-induced scatter. Increasing temperatures accelerate decomposition but coincide in the model with more frequent situations of soil moisture deficit that act negatively on the reaction rate and thus counteract the temperature effect. Though no field observations of soil moisture are available, the weather data render moisture deficits in soil a plausible explanation for offsetting temperature effects. Our data support previous studies in which climate effects were considered to be small compared with the consequences of management changes (Smith et al., 2007). The measured relative annual decline in SOC relative to the starting value is 0.94, 0.68, 0.27, 0.41, and 0.44% for NOFERT, CONMIN, BIODYN, BIORG, and CONFYM, respectively. These rates are much higher than the annual decline of 0.08% suggested by Smith et al. (2007) as a consequence of warming in England during approximately the same period (1978–2003), but similar to the annual 0.6% soil carbon loss that has been observed for the same period in England and Wales by Bellamy et al. (2005). Although a climatic effect cannot be excluded, it would have only small impacts, if at all, on soil carbon storage in the DOK experiment.

Soil Fractions, Model Initialization, and Radiocarbon
The similarity of internal shifts in C allocation in soil between 1977 and 2004 indicates the paramount meaning of those factors that are constant across treatments. This includes climate, soil type, crop rotation, and tillage. Differences in productivity and manure application only marginally affect distribution of C in measured soil fractions.

In most parcels, the carbon loss occurred mainly in the silt and clay-sized fraction and, to a minor extent, at the expense of soluble C (Fig. 1). On the other hand, the relative contribution of POM (a measured fraction that includes both free and occluded light carbon), carbon in stable microaggregates (S+A) and in NaOCl-residues increased. The increase in POM released after destruction of weak aggregates is rather unexpected because POM is considered to be labile and thus declines more rapidly than bulk SOC (Gregorich et al., 1995). Together with the increase in S+A, the data suggest a treatment-independent increased aggregation of soil due to the management change in 1978, probably caused by a higher share of leys and more frequent soil coverage. Management-independent effects were also shown for the incorporation of N15-labeled organic materials into physical soil fractions by comparing BIOORG and CONMIN parcels 25 yr after commencement of the DOK experiment (Bosshard et al., 2008).

Measured fractions and model pools were correlated but revealed systematic differences in the total amount of C for three out of the four dynamic pools considered as being more labile (DPM, RPM, BIO; Tables 5 and 7). In the model, average turnover rates of these pools range between months and <5 yr. Hence, a small bias in estimated input or site-specific turnover rates may induce significant differences between model pools and measured fractions, both of which are responding promptly to environmental conditions. In addition, turnover of labile C is affected by changes in soil structure, as discussed above, and such processes are not accounted for by RothC. The amount of carbon in the pool vs. the corresponding measured fraction was significantly different for DPM, RPM and BIO in both 1977 and 2004. Only the change in humified carbon, largely corresponding to the silt and clay fraction, was not different between measured fraction and model pool. Over time, initial differences in carbon stored in the humified pool HUM vs. the corresponding measured fractions (S+A and s+c) were equalized, leading to nonsignificant differences between them in 2004. Three factors, namely the trend between 1977 and 2004, a similar absolute pool size, and a close match of bulk soil radiocarbon values corroborates the usefulness of the fractionation to independently estimate the size of the HUM pool. The similar decline in measured s+c and modeled HUM tentatively indicates that the measured fraction containing the vast majority of C has mean residence times close to the HUM pool, which is 60 yr at the Therwil site.

Surprisingly, the relative and absolute C content in the chemically resistant rSOC increased. Based on earlier work (Zimmermann et al., 2007), we chose the chemically defined fraction that corresponds in size to the IOM pool of the model as a stable baseline. Its increase falsifies this hypothesis and indicates the nonuniqueness of this measured fraction according to the definition of Smith et al. (2002). The relatively high dynamic of rSOC is further corroborated by radiocarbon dating: the increase in percent modern carbon suggests input of recent, postbomb carbon. The suitability of chemical and physical treatments to isolate a meaningful stable and old fraction has recently been scrutinized by Bruun et al. (2007), and our results provide further support for their reasoning that chemical oxidation with NaOCl does not yield a consistently old carbon fraction.

We compared the modeling results from either model initialization with equilibrium runs or by means of measured fractions. In general, the fraction-based approach proved suitable to initialize the RothC model. Simulated rates of SOC change were almost the same for both approaches, but total C contents were always higher and, on average, closer to the measurements for simulations using measured fractions as starting conditions. This difference is attributable to a higher share of humified OM in the fraction-based approach, which seems to be more realistic than the amount of HUM simulated in the default mode. Measured IOM pools (i.e., NaOCl-residues) were of the same order of magnitude as those derived from model runs with measured ¹⁴C values. Although the increase in NaOCl-resistant C over time indicates that this measured fraction is not inert, it seems that, for the DOK experiment, sizing of the IOM pool by means of chemical fractionation is superior to using the models' default mode without ¹⁴C. The latter is based on a linear regression of data from a range of studies for which ¹⁴C contents were available (Falloon et al., 1998). The standard error of the slope calculated in that study was around 15%, and some sites deviated considerably from the line of least squares.

Overall, initialization with measured fractions slightly improved the match between measured and simulated soil carbon stocks, whereas it had no effect on the simulated change rate. It may therefore be argued that the analytical effort for fractionation is not worthwhile. However, the DOK experiment is a typical long-term agricultural study, and the model has been shown previously to simulate observed changes of similar experiments quite accurately (Smith et al., 1997; Ludwig et al., 2003). The evaluation may turn out differently at other sites with different environmental conditions (Leifeld et al., 2009) and the question remains unanswered as to whether the models' performance may be significantly improved by the use of measurable fractions as a proxy for model pools.

Radiocarbon measurements of bulk soils cannot only be used to adjust the models' size of the IOM pool, but also to validate the time course of SOC when more than one date for ¹⁴C measurements is available. The model accurately matches the observed radiocarbon values in 2004, although the three parcels used for ¹⁴C measurements differ in their SOC dynamics: the CONFYM parcel had a constant SOC, whereas CONMIN and BIODYN lost 8.7 and 4.4 t C ha^–1, respectively, during the experiment. The match of both ¹⁴C and SOC dynamics indicates that differences in SOC contents are mainly caused by different input rates, whereas decomposition rate constants are quite constant across treatments. Though the analysis is restricted to only three parcels and two points in time due to the high analytical costs of the ¹⁴C measurement, the result indicates that fundamental model assumptions for decomposition rate constants must be realistic. Bol et al. (2005) used a similar approach and also observed a reasonably good match between measured ¹⁴C and RothC simulations for the Askov long-term experiment in Denmark. However, their study with more frequent ¹⁴C sampling campaigns indicates that ¹⁴C measurements at only two points in time are not sufficient for an unbiased verification of the models' output on account of possible radiocarbon measurement errors or contamination.

Closed Carbon Budgets and System Sustainability
Carbon inputs to farming systems in the form of organic fertilizers are controlled by the productivity of the system. In more artificial situations, such as in the DOK experiment in which farm systems are simply mimicked, it is important to validate that chosen input rates are credible and correspond to the systems' capacity for assimilating C. On average, 0.89 (BIODYN), 1.04 (BIOORG), and 1.17 (CONFYM) t C ha^–1 yr^–1 are applied as organic fertilizer. These materials derive from annual ley and straw yields of 3.01, 2.92, and 3.38 t C ha^–1 in BIODYN, BIOORG, and CONFYM, respectively. We calculated the theoretical organic fertilizer (solid manure and slurry) outputs from the yields of leys and straw exports from the fields. We assume a conversion rate of 0.3 from grass-clover-C to dung-C (Minonzio et al., 1998), a loss of C during manure storage/composting of dung-straw mixtures of 39, 32, and 21% for BIODYN, BIOORG, and CONFYM, respectively (corresponding to OM measurements and model calculations above), and a C fraction of 0.44 for all plant material. With these conversion rates, average annual organic fertilizer inputs of 0.95, 1.04, and 1.38 t C ha^–1 for BIODYN, BIOORG, and CONFYM, respectively, can be achieved. The calculation suggests that the actual amendments are balanced (BIOORG) or even below the systems' productivity (BIODYN and CONFYM). Whereas the difference between applied and potential rates is small for BIODYN and may be explained by uncertainties in the overall calculation and measurement, rates in CONFYM are 0.21 t below the systems' capacity. Adopted organic fertilizer applications in CONFYM may therefore lead to smaller soil carbon loss rates than those reported here.

Is it possible that stabilization of almost the same amount of plant residues as a result of combined digestion and composting have a different effect on soil carbon in BIODYN compared with the application of equivalent carbon masses without composting in BIOORG? It seems likely that such an effect would only be apparent on account of the many uncertainties in the overall budget. First, carbon conversion rates during digestion and composting are variable and may differ from the calculations above. It cannot be excluded that organically fertilized treatments receive more or less manure input than sustained by the system. Second, a smaller organic fertilizer application rate in conjunction with a distinctly higher manure stability leads to a retarded decline in organic matter contents, and new steady-state conditions might be reached at a later point in time. The significant decline in two out of four BIODYN parcels may be an early indicator for such a trend. Finally, as shown above, the starting conditions determined much of the subsequent carbon dynamics in addition to management-induced effects. In this study of 20 parcels of the DOK trial, the treatment with the most stabilized manure is also the one with the highest starting C content and, among the organically fertilized parcels, the one with the highest clay content.

With an organic fertilizer conversion rate of CONFYM, similar amounts of organic fertilizer could be produced by the mineral treatment (1.1) and somewhat less in the null treatment (0.8 t C ha^–1 yr^–1), suggesting that amendment of that material to fields outside the experiment will lead to C sequestration elsewhere. Full carbon budgets, including C imports and exports, are therefore crucial for an unbiased management comparison (Schlesinger 2000), and the latter cannot be achieved by means of small-scale field experiments alone.

	CONCLUSIONS

TOP NOTES ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION CONCLUSIONS REFERENCES

 
In a 27-yr experiment that compared different management systems, an overall decline in topsoil SOC could be explained by reduced inputs and changed residue management practices, whereas model simulations showed that the observed increase in temperature had no effect. For a correct simulation of differences between conventional, bioorganic, and biodynamic farming systems, a realistic estimate of the composition of organic fertilizers is important because management types differ in their treatment of manure. Different soil conditions were identified as one important reason for a higher carbon stock in the biodynamic parcels, whereas stabilization of manure by composting seems to play a secondary role, if any. In the conventional treatment with organic fertilizer, the results indicate accelerated decomposition that is possibly induced by priming effects which cannot be reproduced by the model. Parcelwise measurements and simulations stressed the importance of preexperimental conditions on the change over time.

	ACKNOWLEDGMENTS

 
The DOK experiment is jointly managed by Agrocope Reckenholz-Tänikon Research Station ART in Zurich (Switzerland) and the Research Institute of Organic Agriculture FiBL in Frick (Switzerland). We gratefully acknowledge Peter Weisskopf and David Dubois for complementary funding, Lucie Gunst for providing the DOK management data, and Jochen Mayer for helpful comments on the manuscript. Kevin Coleman, Rothamsted (UK), is acknowledged for supplying the RothC 26.3 source code and Irka Hajdas, ETH Zurich, for measuring radiocarbon.

	NOTES

TOP NOTES ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION CONCLUSIONS REFERENCES

 
All rights reserved. No part of this periodical may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopying, recording, or any information storage and retrieval system, without permission in writing from the publisher.

	REFERENCES

TOP NOTES ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION CONCLUSIONS REFERENCES

 

• Ammann, C., C. Flechard, J. Leifeld, A. Neftel, and J. Fuhrer. 2007. The carbon budget of newly established temperate grassland depends on management intensity. Agric. Ecosyst. Environ. 121:5–20.[CrossRef]
• Bellamy, P.H., P.J. Loveland, R.I. Bradley, R.M. Lark, and G.J.D. Kirk. 2005. Carbon losses from all soils across England and Wales 1978–2003. Nature 437:245–248.[CrossRef][Medline]
• Bol, R., J. Eriksen, P. Smith, M.H. Garnett, K. Coleman, and B.T. Christensen. 2005. The natural abundance of C-13, N-15, S-34 and C-14 in archived (1923–2000) plant and soil samples from the Askov long-term experiments on animal manure and mineral fertilizer. Rapid Commun. Mass Spectrom. 19:3216–3226.[Web of Science][Medline]
• Bolinder, M.A., H.H. Janzen, E.G. Gregorich, D.A. Angers, and A.J. Vandenbygaart. 2007. An approach for estimating net primary productivity and annual carbon inputs to soil for common agricultural crops in Canada. Agric. Ecosyst. Environ. 118:29–42.[CrossRef]
• Bosshard, C., E. Frossard, D. Dubois, P. Mäder, I. Manolov, and A. Oberson. 2008. Incorporation of nitrogen-15-labeled amendments into physically separated soil organic matter fractions. Soil Sci. Soc. Am. J. 72:949–959.[Abstract/Free Full Text]
• Bruun, S.I., K. Thomsen, B.T. Christensen, and L.S. Jensen. 2007. In search of stable soil organic matter fractions: A comparison of methods applied to soils labelled with 14C for 40 days or 40 years. Eur. J. Soil Sci. 10.1111/j.1365-2389.2007.00985x.[CrossRef]
• Buyanovsky, G.A., and G.H. Wagner. 1998. Carbon cycling in cultivated land and its global significance. Glob. Change Biol. 4:131–141.[CrossRef]
• Coleman, K., and D.S. Jenkinson. 1999. RothC-26.3. A model for the turnover of carbon in soils. Herts, Rothamsted Research, Harpenden, Hertfordshire, UK.
• Drinkwater, L., P. Wagoner, and M. Sarrantonio. 1998. Legume-based cropping systems have reduced carbon and nitrogen losses. Nature 396:262–265.[CrossRef][Web of Science]
• Elmaz, Ö., H. Cerit, M. Özçelik, and S. Ulas. 2004. Impact of organic agriculture on the environment. Fresenius Environ. Bull. 13:1072–1078.
• Eltun, R., A. Korsaeth, and O. Nordheim. 2002. A comparison of environmental, soil fertility, yield, and economical effects in six cropping systems based on an 8-year experiment in Norway. Agric. Ecosyst. Environ. 90:155–168.
• Falloon, P., P. Smith, K. Coleman, and S. Marshall. 1998. Estimating the size of the inert organic matter pool from total soil organic carbon content for use in the Rothamsted Carbon Modell. Soil Biol. Biochem. 30:1207–1211.[CrossRef]
• Fließbach, A., and P. Mäder. 2000. Microbial biomass and size-density fractions differ between soils of organic and conventional agricultural systems. Soil Biol. Biochem. 32:757–768.[CrossRef]
• Fließbach, A., P. Mäder, and U. Niggli. 2000. Mineralization and microbial assimilation of C-14-labeled straw in soils of organic and conventional agricultural systems. Soil Biol. Biochem. 32:1131–1139.[CrossRef]
• Fließbach, A., H.R. Oberholzer, L. Gunst, and P. Mäder. 2007. Soil organic matter and biological soil quality indicators after 21 years of organic and conventional farming. Agric. Ecosyst. Environ. 118:273–284.[CrossRef]
• Friedel, J.K. 2000. The effect of farming system on labile fractions of organic matter in Calcari-Epileptic Regosols. J. Plant Nutr. Soil Sci. 163:41–45.
• Gregorich, E.G., B.H. Ellert, and C.M. Monreal. 1995. Turnover of soil organic matter and storage of corn residue carbon estimated from natural C-13 abundance. Can. J. Soil Sci. 75:161–167.
• Gunapala, N., and K.M. Scow. 1998. Dynamics of soil microbial biomass and activity in conventional and organic farming systems. Soil Biol. Biochem. 30:805–816.[CrossRef]
• Gunst, L., W. Jossi, U. Zihlmann, P. Mäder, and D. Dubois. 2007. DOK-Versuch: Erträge und Ertragsstabilität 1978–2005. Agrarforschung 14:542–547.[Web of Science]
• Jackson, R.B., J. Canadell, J.R. Ehleringer, H.A. Mooney, O.E. Sala, and E.D. Schulze. 1996. A global analysis of root distributions for terrestrial biomes. Oecologia 108:389–411.[Web of Science]
• Khan, S.A., R.L. Mulvaney, T.R. Ellsworth, and C.W. Boast. 2007. The myth of nitrogen fertilization for soil carbon sequestration. J. Environ. Qual. 36:1821–1832.[Abstract/Free Full Text]
• Lampkin, N. 1990. Organic farming. Farming Press, Ipswich, Suffolk, UK.
• Leifeld, J., M. Zimmermann, J. Fuhrer, and F. Conen. 2009. Storage and turnover of carbon in grassland soils along an elevation gradient in the Swiss Alps. Glob. Change Biol. 15:668–679.
• Levin, I., S. Hammer, B. Kromer, and F. Meinhardt. 2008. Radiocarbon observations in atmospheric CO₂: Determining fossil fuel CO₂ over Europe using Jungfraujoch observations as background. Sci. Total Environ. 391:211–216.[Medline]
• Levin, I., and B. Kromer. 2004. The Tropospheric (CO₂)-C-14 level in mid-latitudes of the Northern Hemisphere (1959–2003). Radiocarbon 46:1261–1272.[Web of Science]
• Liebig, M.A., and J.W. Doran. 1999. Impact of organic production practices on soil quality indicators. J. Environ. Qual. 28:1601–1609.[Abstract/Free Full Text]
• Ludwig, B., B. John, R. Ellerbrock, M. Kaiser, and H. Flessa. 2003. Stabilization of carbon from maize in a sandy soil in a long-term experiment. Eur. J. Soil Sci. 54:117–126.
• Mäder, P., A. Fließbach, D. Dubois, L. Gunst, P. Fried, and U. Niggli. 2002. Soil fertility and biodiversity in organic farming. Science 296:1694–1697.[Abstract/Free Full Text]
• Marinari, S., R. Mancinelli, E. Carnpiglia, and S. Grego. 2006. Chemical and biological indicators of soil quality in organic and conventional farming systems in Central Italy. Ecol. Indicators 6:701–711.
• Melero, S., J.C.R. Porras, J.F. Herencia, and E. Madejon. 2006. Chemical and biochemical properties in a silty loam soil under conventional and organic management. Soil Tillage Res. 90:162–170.
• Minonzio, G., A. Grub, and J. Fuhrer. 1998. Methan-Emissionen der schweizerischen Landwirtschaft. Schriftenreihe Umwelt No. 298, BUWAL, Bern, Switzerland.
• Petersen, S.O., A.M. Lind, and S.G. Sommer. 1998. Nitrogen and organic matter losses during storage of cattle and pig manure. J. Agric. Sci. 130:69–79.[CrossRef]
• Pulleman, M., A. Jongmans, J. Marinissen, and J. Bouma. 2003. Effects of organic versus conventional arable farming on soil structure and organic matter dynamics in a marine loam in the Netherlands. Soil Use Manage. 19:157–165.
• Reganold, J.P., A.S. Palmer, J.C. Lockhart, and A.N. Macgregor. 1993. Soil quality and financial performance of biodynamic and conventional farms in New Zealand. Science 260:344–349.[Abstract/Free Full Text]
• Robertson, G.P., E.A. Paul, and R.R. Harwood. 2000. Greenhouse gases in intensive agriculture: Contributions of individual gases to the radiative forcing of the atmosphere. Science 289:1922–1925.[Abstract/Free Full Text]
• Schlesinger, W.H. 2000. Carbon sequestration in soils: Some cautions amidst optimism. Agric. Ecosyst. Environ. 82:121–127.[CrossRef]
• Shannon, D., A.M. Sen, and D.B. Johnson. 2002. A comparative study of the microbiology of soils managed under organic and conventional regimes. Soil Use Manage. 18:274–283.[CrossRef]
• Smith, J.U., P. Smith, R. Monaghan, and R.J. Macdonald. 2002. When is a measured soil organic matter fraction equivalent to a model pool? Eur. J. Soil Sci. 53:405–416.
• Smith, P., S.J. Chapman, W.A. Scott, H.I.J. Black, M. Wattenbach, R. Milne, C.D. Campbell, A. Lilly, N. Ostle, P.E. Levy, D.G. Lumsdon, P. Millard, W. Towers, S. Zaehle, and J.U. Smith. 2007. Climate change cannot be entirely responsible for soil carbon loss observed in England and Wales, 1978–2003. Glob. Change Biol. 13:2605–2609.[CrossRef]
• Smith, P., J.U. Smith, D.S. Powlson, W.B. Mcgill, J.R.M. Arah, O.G. Chertov, K. Coleman, U. Franko, S. Frolking, D.S. Jenkinson, L.S. Jensen, R.H. Kelly, H. Klein-Gunnewiek, A.S. Komarov, C. Li, J.A.E. Molina, T. Müller, W.J. Parton, J.H.M. Thornley, and A.P. Whitmore. 1997. A comparison of the performance of nine soil organic matter models using datasets from seven long-term experiments. Geoderma 81:153–225.[CrossRef][Web of Science]
• Stockdale, E.A., N.H. Lampkin, M. Hovi, R. Keatinge, E.K.M. Lennartsson, D.W. Macdonald, S. Padel, F.H. Tattersall, M.S. Wolfe, and C.A. Watson. 2001. Agronomic and environmental implications of organic farming systems. Adv. Agron. 70:261–327.
• Teasdale, J.R., C.B. Coffman, and R.W. Mangum. 2007. Potential long-term benefits of no-tillage and organic cropping systems for grain production and soil improvement. Agron. J. 99:1297–1305.[Abstract/Free Full Text]
• Walther, U., J.-P. Ryser, and R. Flisch. 2001. Grundlagen für die Düngung im Acker- und Futterbau 2001. Agrarforschung 8:1–80.
• Wells, A.T., K.Y. Chan, and P.S. Cornish. 2000. Comparison of conventional and alternative vegetable farming systems on the properties of a yellow earth in New South Wales. Agric. Ecosyst. Environ. 80:47–60.
• Widmer, F., F. Rasche, M. Hartmann, and A. Fliessbach. 2006. Community structures and substrate utilization of bacteria in soils from organic and conventional farming systems of the DOK long-term field experiment. Appl. Soil Ecol. 33:294–307.[CrossRef]
• Wilts, A.R., D.C. Reicosky, R.R. Allmaras, and C.E. Clapp. 2004. Long-term corn residue effects: Harvest alternatives, soil carbon turnover, and root-derived carbon. Soil Sci. Soc. Am. J. 68:1342–1351.[Abstract/Free Full Text]
• Zaller, J.G., and U. Köpke. 2004. Effects of traditional and biodynamic farmyard manure amendment on yields, soil chemical, biochemical and biological properties in a long-term field experiment. Biol. Fertil. Soils 40:222–229.
• Zimmermann, M., J. Leifeld, M.W.I. Schmidt, P. Smith, and J. Fuhrer. 2007. Measured soil organic matter fractions can be related to pools in the RothC model. Eur. J. Soil Sci. 58:658–667.[CrossRef]

This Article Abstract Figures Only Full Text (PDF) Free Alert me when this article is cited Alert me if a correction is posted Services Similar articles in this journal Alert me to new issues of the journal Download to citation manager Google Scholar Articles by Leifeld, J. Articles by Oberholzer, H.-R. PubMed Articles by Leifeld, J. Articles by Oberholzer, H.-R. Agricola Articles by Leifeld, J. Articles by Oberholzer, H.-R. Related Collections Isotopes Soil Organic Matter Organic farming Agricultural Systems Soil Chemistry Animal Waste