Soil carbon change factors for the Canadian agriculture national greenhouse gas inventory

Soil carbon change factors for the Canadian agriculture national greenhouse gas inventory A. J. VandenBygaart 1, B. G. McConkey 2, D. A. Angers 3, W. Smith 1, H. de Gooijer 4, M. Bentham 5, and T. Martin 1 1 Eastern Cereal and Oilseed Research Centre, Agriculture and Agri-Food Canada, K.W. Neatby Building, 960 Carling Avenue, Ottawa, Ontario, Canada K1A 0C6 (e-mail: vandenbygaarta@agr.gc.ca); 2 Semiarid Prairie Agricultural Research Centre, Agriculture and Agri-Food Canada, 1 Airport Rd, P.O. Box 1030, Swift Current, Saskatchewan, Canada S9H 3X2; 3 Soils and Crops Research and Development Centre, Agriculture and Agri-Food Canada, 2560 Hochelaga Blvd, Sainte-Foy, Quebec, Canada G1V 2J3; 4 Prairie Farm Rehabilitation Administration, #2 Government Road, Indian Head, Saskatchewan, Canada S0G 2K0; and 5 Saskatoon Research Centre, Agriculture and Agri-Food Canada, 107 Science Place, Saskatoon, Saskatchewan, Canada S7N 0X2. Received 16 February 2007, accepted 5 December 2007. VandenBygaart, A. J., McConkey, B. G., Angers, D A., Smith, W., de Gooijer, H., Bentham, M. and Martin, T. 2008. Soil carbon change factors for the Canadian agriculture national greenhouse gas inventory. Can. J. Soil Sci. 88: 671680. Canada annually reports on all of its annual greenhouse gas emissions to the United Nations Framework Convention on Climate Change (UNFCCC), including estimates of CO 2 emissions and removals from cropland management. Soil carbon (C) change in cropland resulting from management is estimated by using C change factors multiplied by the area of cropland subjected to a management change. In this paper we compare soil C change factors in Canadian cropland obtained using a C modelling approach (Century model) to both empirical estimates obtained from the scientific literature, and to default Intergovernmental Panel on Climate Change (IPCC) estimates. Factors were estimated for land management changes from annual to perennial cropping, tillage to no-tillage and from summer fallow to continuous cropping. Empirical data comparing C change between conventional tillage (CT) and no-tillage (NT) were highly variable, but the modelled factors were still within the range derived from the empirical data. Factors for changes from CT to NT varied from 0.06 to 0.16 Mg C ha 1 yr 1 across the country. When considering the change from annual to perennial cropping, the modelled factors ranged from 0.46 to 0.56 Mg C ha 1 yr 1, which is in the range of empirical values, and were slightly greater in the eastern than the western soil regions. For conversion of crop-fallow to continuous cropping, the modelled rate of C storage (0.33 Mg C ha 1 yr 1 ) was more than double the average rate of 0.1590.06 Mg C ha 1 yr 1 derived from two independent assessments of the literature. For each of the management changes considered, the modelled factors were generally lower than IPCC estimates, and this is partly attributable to differences in calculation methods and to the fact that C changes likely occur more slowly in the cold climate of Canada. Generally, the results show that the modelling approach used at present to derive C change factors for use in Canada s inventory is adequate. However, soil C change factors for cropland soils in Canada would be greatly improved by a reduction in the high variability usually associated with empirical data, and by improved simulation of the Century model under varying management conditions. Key words: Soil organic carbon, Canada, tillage, perennial cropping, cropping intensity, carbon change factors VandenBygaart, A. J., McConkey, B. G., Angers, D A., Smith, W., de Gooijer, H., Bentham, M. et Martin, T. 2008. Coefficients de l e volution des stocks de carbone du sol pour l Inventaire canadien des gaz à effet de serre en agriculture. Can. J. Soil Sci. 88: 671680. Chaque année, le Canada signale ses e missions de gaz a` effet de serre aux termes de la Conventioncadre des Nations Unies sur les changements climatiques (CCNUCC). On estime notamment les de gagements et les retraits de CO 2 résultant de l exploitation des terres agricoles. La variation des stocks de carbone du sol (C) re sultant de la gestion des terres agricoles est estime e en multipliant les coefficients de l e volution des stocks de C par la superficie des cultures qui connaissent un changement de gestion. Les auteurs comparent les coefficients pour les terres agricoles canadiennes obtenus par modélisation (mode` le Century) a` ceux issus des estimations empiriques, releve s dans la documentation scientifique, et aux estimations par défaut du Groupe d experts intergouvernemental sur le changement climatique (GEIC). Ils ont estime les coefficients associe s à une modification de la vocation des terres, soit le passage des cultures annuelles aux cultures vivaces, du travail du sol au non-travail du sol et de la jache` re d e té à la monoculture. Les donne es empiriques comparant les stocks de C pour le travail du sol (TS) et le non-travail du sol (NTS) varient considérablement, mais les coefficients obtenus par modélisation demeurent dans la fourchette de ceux issus des donne es empiriques. Au Canada, les coefficients pour le passage du TS au NTS varient de 0,06 à 0,16 Mg de C par hectare annuellement. Pour le passage des cultures annuelles aux cultures vivaces, les coefficients obtenus par mode lisation varient de 0,46 à 0,56 Mg de C par hectare annuellement, re sultat qui se situe dans la fourchette des valeurs empiriques. Les re sultats sont le ge` rement plus e leve s dans Abbreviations: CT, conventional tillage; IPCC, Intergovernmental Panel on Climate Change; LMC, land management changes; NT, no-tillage 671

672 CANADIAN JOURNAL OF SOIL SCIENCE l Est que dans l Ouest. Pour le passage de la jache` re a` la monoculture, les stocks de C estime s par mode lisation (0,33 Mg de C par hectare annuellement) correspondent à plus du double du taux moyen de 0,15 9 0,06 Mg de C par hectare annuel rapporte dans deux e valuations inde pendantes, tire es de la documentation. Pour chacun des changements examine s, la mode lisation donne ge ne ralement des coefficients inférieurs aux estimations du GEIC. On le doit en partie aux méthodes de calcul diffe rentes et au fait que les réserves de C e voluent sans doute plus lentement dans le climat plus froid du Canada. En ge ne ral, les résultats indiquent que la technique de modélisation employée actuellement pour de river les coefficients de l e volution des stocks de C dans le cadre de l inventaire canadien sont ade quats. Néanmoins, ces coefficients profiteraient d une re duction de la variabilite habituellement associe e aux valeurs empiriques et d une meilleure simulation des diverses conditions de gestion par le modèle Century. Mots clés: Carbone organique du sol, Canada, travail du sol, cultures vivaces, rendement des cultures, coefficients de l e volution des stocks de carbone In Canada a number of management practices are known to influence soil organic C in agricultural land. Reducing tillage intensity, increasing C input via cropping system regimes, and conversion from annual to perennial crops all can significantly increase soil C in agricultural soils (Janzen et al. 1998; Bruce et al. 1999). Adoption of practices with reduced tillage or no-till can result in significant accumulation of soil C compared with conventional tillage practices (Janzen et al. 1998; McConkey et al. 2003; VandenBygaart et al. 2003). Eliminating or reducing the frequency of summerfallow and the greater production of perennial forage crops are means of increasing carbon input within cropping systems (Biederbeck et al. 1984; Bremer et al. 1994; Campbell et al. 1998, 2000, 2005; McConkey et al. 2003). Conversely, increasing tillage intensity, reducing carbon input and switching to more annual crops rather than perennial crops will reduce soil C (VandenBygaart et al. 2003). Under the United Nations Framework Convention on Climate Change (UNFCCC), Canada is required to report on all of its annual greenhouse gas emissions including estimates of CO 2 emissions and removals due to cropland management. The Canadian Agriculture Greenhouse Gas Monitoring Accounting and Reporting System (CanAG-MARS) was developed to annually provide estimates of changes in soil C in agricultural land for Canada s National Greenhouse Gas Inventory Report [National Inventory Report (NIR 2006)]. For reporting purposes Canada derives estimates of carbon (C) change in cropland from management using C change factors multiplied by the area of cropland subject to management change derived from activity data provided from the Census of Agriculture. Due to the large area of cropland in Canada, these estimates can only feasibly be determined at broad scales, and, as such, reporting of C change from cropland in Canada is calculated at the scale of the Soil Landscapes of Canada polygons (Agriculture and Agri-Food Canada 2006). The Intergovernmental Panel of Climate Change (IPCC) guidelines for national greenhouse gas inventories describes three approaches for using mineral soil C change factors for national inventories for cropland (IPCC 2006). Tier 1 uses default C stock change factors for general land management effects of tillage and inputs of organic materials to the soil which were derived from an international body of studies. Tier 2 involves using a similar linear approach to deriving soil C change as Tier 1, but involves incorporating country specific C factors, reference C stocks, soil types etc. Tier 3 uses dynamic simulation models for computing annual C stock changes due to land management changes. Furthermore, it is considered good practice to verify soil organic C changes using data and methods that are independent of those used to prepare the inventory (IPCC 2006). Our objective in this paper was to compare the soil C change factors derived for Canada s NIR using a modelling methodology versus those derived from empirical data extracted from the refereed literature, and those provided by the IPCC s Guidelines for National Greenhouse Gas Inventories (IPCC 2006). MATERIALS AND METHODS Land Management Changes Represented in the Canadian Greenhouse Gas Inventory There were data-availability restrictions which put boundaries on which land management changes (LMC) could be adequately considered in the inventory. Requirements for including a LMC in the C accounting system for croplands included the availability of the land management activity data: (a) activity data had to be available over a time series at least from 1990 to 2004; (b) activity data had to be available at, at minimum, a national or regional scale; (c) the LMC had to produce significant changes in C stocks either because the change in C per unit area of the activity was large, or that the aerial extent of the activity was large, or both. The LMCs that met the inclusions criteria were:. Change in area of perennial crops.. Change in the area of annual crops.. Change in tillage practices.. Change in area of summerfallow. Many LMC that potentially affect soil C on agricultural land were excluded from the greenhouse gas inventory because they did not meet one or more of the criteria requirements. Potentially important LMC that were excluded from CanAG-MARS were animal

VANDENBYGAART ET AL. * CANADA CARBON CHANGE FACTORS 673 manure application to cropland, fertility management, change in the types or nature of the annual crops grown, use of cover and green manure crops, management of tame pastures and hayland, irrigation, and effect of soil erosion management on soil C. More specific justifications for exclusion of these LMCs are further discussed in McConkey et al. (2007). Canada has a significant history of observations of long-term changes in soil C as a result of adoption of NT, reduction of summer fallow and increasing perennial cropping (VandenBygaart et al. 2003; Campbell et al. 2005), and thus it would appear attractive to utilize C change factors derived from these studies. However, these empirical data are limited spatially such that there are some regions with few or no data. Furthermore, most experiments yield few measurements of C change over time, so it is difficult to determine how C change varies with time since management change. Other limitations include difficulty in estimating the full effect of the range of interactions with the initial soil state and with combination of different practices. There is also difficulty determining the variability of C change that might occur without management change. Due to all the above limitations, the decision was made to derive C factors through modelling of soil C dynamics. Soil C Dynamics and the Century Model In this study, we assumed that the dynamics of soil C after a management change followed first-order kinetics and, consequently, we fit the following equation to the soil C difference: DSOC(t)DSOC max (1e kt ) (1) where DSOC(t) is the soil C change with time, t, since initial adoption of the management change; DSOC max is the maximum eventual soil C change expected for management change; and k is the rate constant for soil C change. Using Eq. 1, we derived an effective linear soil C change factor that, when multiplied by time since adoption of the new management, gives the same total change as Eq. 1. This factor, f (t), varies in value with time, t, since adoption of new management and is simply Eq. 1 divided by t: f (t)dsoc max (1e kt )=t (2) Furthermore, C factors derived from exponential curves are not a single value for an activity, but rather a timedependent equation. The value of the carbon factor depends on how many years have passed since the onset of the specific activity. Equation 3 was used to estimate an annual factor (year t 1 to year t): f (t)dsoc max (e k(t1) e (kt) ) (3) The Century model (Parton et al. 1987), version 4 (Natural Resources Ecology Laboratory 2007), was used to estimate the C change factors, similar to the method developed by Smith et al. (2001), by setting the factor to fit the slope of a linear regression for the first 20yr of the management-induced change in soil C. However, our approach differed from that of Smith et al. (2001) in two ways: First, the C factors were estimated from each distinct soil unit (i.e., component) within the Soil Landscapes of Canada (SLC) polygon data base (Agriculture and Agri-Food Canada 2006) rather than from generalized averages within soil groups and, second, the factor of soil C change was derived from the fit of relative C stock change to Eq. 1. The Century model was calibrated to ensure that crop yields were in line with known values for agricultural crops in Canada. Several of the publicly distributed parameter values for Century 4 (Natural Resource Ecology Laboratory 2007) were modified during this calibration including those for maximum potential production, harvest index, the amount of evaporation per unit of potential evapotranspiration, and some of the parameters concerning N uptake, maximum and minimum C:N ratios in crop and soil organic matter, and rate of N losses. The resulting Century parameter set defining C and N dynamics and growth of each crop were the same for simulations throughout Canada. A standardized initialization procedure, similar to that used by Smith et al. (1997, 2000, 2001), was used. A generalized description was developed for land management practices on cropland from 1910 to 2000 derived from a mix of expert knowledge and agricultural statistics of land management (Statistics Canada 2007). After this initialization period, the soil C factors were then estimated from the difference in soil C stocks over time for the simulated generalized land use and management scenario, with and without the change in management practice (Smith et al. 2001). For example, the factor for the conversion of intensive tillage to no-tillage was determined from the difference between simulated C stock for the generalized land use and management scenario and that with no-tillage substituted for intensive tillage in that generalized scenario from year 2001 onwards. The initial soil C input pools in Century have a large effect on simulated C stocks (Falloon and Smith 2002) so the relatively long period (90 yr) of common initialization helped ensure that the initial C pools for factor estimation were reasonable for Century simulation given the general soil, climate, and land use and management for the location simulated. More specifics on how Century was used in the inventory are described in depth in McConkey et al. (2007). Factors were generated using the Century model for land management changes occurring on all Canadian cropland. The k and DSOC max values for Eq. 3 were derived from non-linear least-squares fitting of Eq. 1 to the estimated DSOC (t). The weighted mean k and DSOC max were calculated by region and soil textural category, and weighted by the extent of each soil allocated to annual cropland. The mean k and the mean of absolute values of DSOC max for LMC in the

674 CANADIAN JOURNAL OF SOIL SCIENCE opposite direction were then calculated to have reversibility (e.g., soil C change from annual crops shifting to perennial crops was made equal to the negative of soil C change from perennial crops shifting to annual crops). Empirical Data Compilation We validated the factors by comparing them with empirical data derived from appropriate published studies in Canada. To better facilitate the comparison with the spatially-sporadic empirical data, Canada was divided into five agro-ecological regions based on climate/soil/management criteria (Table 1). The two largest regions represented semiarid and subhumid portions of the Canadian prairies. East-central Canada consisted of Ontario and Quebec, the Pacific region (i.e., West Pacific) was composed of agricultural land west of the Rocky Mountains and Eastern Canada represented land in the Atlantic Provinces. We also compared the C change factors with Tier 1 (default) factors from the IPCC 2006 Revised Guidelines (IPCC 2006), which, in turn, were derived from an analysis of worldwide published literature. For comparison with empirical data for the CT to NT factors, we used only replicated tillage experiments for which organic C and bulk density data were available and for which a CT control treatment was maintained. Conventional tillage was defined as two or more tillage operations per year for each region, with the exception of the semiarid prairie where it was defined as at least one tillage event per year. For Eastern Canada, several studies have shown no difference in C stocks between NT and CT (VandenBygaart et al. 2003; Angers et al. 1997). However, several of these studies were short-term (less than 10years) and some were also conducted on poorly-drained soils on which crop yields are generally lower under NT than under CT. Consequently additional criteria were used for the East-central region: Studies had to (i) be at least 10years in duration, (ii) be randomized with adequate replication, (iii) have a reference tillage treatment, and (iv) have soil sampling exceeding 30cm depth (VandenBygaart and Angers 2006). For western Canada we used the same criteria with the exception of soil sampling depth because tillage is generally not deeper than 15 cm, which limits the zone of influence on soil C dynamics to shallower depths than the mouldboard ploughed soils of Eastern Canada (Campbell et al. 1995). In comparison with other LMC, there are considerable empirical data for C change from change in fallow frequency and such empirical data are available for areas throughout the prairies, where essentially all area of fallow change occurs in Canada. The Century-derived temporal dynamics of C change were used but the magnitude of C change due to change in the area of fallow was scaled based on empirical data. We assumed that the SOC dynamics predicted by Century was appropriate so we set the k for Eq. 1 to the prairie average of 0.031 yr 1 from the Century simulations. However, because of the large amount of empirical data of C change after about 20yr of change in fallow Table 1. Rate constants (k), maximum carbon change (DSOC max ), and 20-yr average C change factors based on Century simulation for the Canadian greenhouse gas inventory and Intergovernmental Panel of Climate Change (IPCC) Tier 1 total carbon change, and 20-yr annual factors for converting from conventional tillage to no-tillage (CT to NT), decreasing fallow frequency and increasing proportion of perennial crops for various regions of Canada Region z LMC k (yr 1 ) Canadian Inventory IPCC Tier 1 DSOC max (Mg C ha 1 ) 20-yr average factor (Mg C ha 1 yr 1 ) DSOC (Mg C ha 1 ) 20-yr annual factor (Mg C ha 1 yr 1 ) East Atlantic CT to NT y 0.022 3.5 0.06 9.9912.4 x 0.4990.62 Decrease fallow 0.031 13.1 0.30 10.5913.6 0.5290.68 Increase perennials 0.022 43.4 0.77 42.897.8 2.1490.39 East Central CT to NT 0.025 5.0 0.10 9.9912.4 0.4990.62 Decrease fallow 0.031 13.1 0.30 10.5913.6 0.5290.68 Increase perennials 0.025 38.2 0.74 42.897.8 2.1490.39 Subhumid Prairie CT to NT 0.029 6.5 0.16 4.6913.2 0.2390.66 Decrease fallow 0.031 13.1 0.30 21.1914.3 1.0590.72 Increase perennials 0.023 29.4 0.55 37.599.2 1.8790.46 Semiarid Prairie CT to NT 0.024 3.7 0.07 3.296.3 0.1690.31 Decrease fallow 0.031 13.1 0.30 5.697.4 0.2890.37 Increase perennials 0.028 26.1 0.56 12.096.00.6090.30 West Pacific CT to NT 0.012 4.8 0.05 9.9912.4 0.4990.62 Decrease fallow 0.031 13.1 0.30 10.5913.6 0.5290.68 Increase perennials 0.016 34.4 0.46 42.897.8 2.1490.39 z For geographic extent of regions refer to text. y Conventional tillage refers to at least two tillage operations per year except in the semiarid prairies where it consists of at least one tillage operation per year; no-tillage has no soil disturbance except for annual seeding operation x Mean995% confidence limits.

VANDENBYGAART ET AL. * CANADA CARBON CHANGE FACTORS 675 frequency, we set the magnitude of DSOC max for Eq. 1 so it matched the average empirical C change over that period. Specifically, DSOC max per hectare of fallow loss in a crop rotation was calculated to be 13.1 Mg ha 1, which is equivalent to the average observed C change of 3Mgha 1 over 20yr after conversion from crop-fallow to continuous cropping (i.e., from Eq. 1, 0.5 ha of fallow change for each hectare of crop-fallow rotation results in C change per hectare of land of 3 Mg C ha 1 ). We also compared the factors derived by modelling with those provided by IPCC methodology (IPCC 2006). For the IPCC soil C change factors it is assumed the change occurs over 20years and, thus, we refer to this as an annual factor. The Century-derived factors were those for 0to 20yr derived using Eq. 2. To compare the Canada-specific soil C change factors, each of the five agro-ecological regions had to relate to those defined by the IPCC (2006). The semiarid prairies have a dry, cold temperate climate, with a default reference soil stock of 50Mg C ha 1 to 30cm depth, while all other agricultural areas of Canada have a moist, cold temperate climate with a default reference of 95 Mg C ha 1 (for Canada 2:1 lattice clays form the majority of clays). The factor used for comparison of the IPCC effect for fallow was derived based on conversion from frequent bare fallow with low residue return under frequent tillage (low carbon input; Table 5.5, Volume 4: Agriculture, Forestry and Other Land Use, p. 5.18) to annual cropping (medium carbon input) under noninverting tillage in the Canadian prairies (annual cropping under inverting tillage elsewhere in Canada). Frequency of fallow is most commonly every second year in the Canadian prairies (Campbell et al. 2005). Therefore, we required multiplying the empirical soil C change by two to be consistent with an assumed LMC to continuous cropping from fallow over 20yr as derived from IPCC. We also assumed that changing crop types associated with an increase in cropping frequency had little effect on the soil C change. The conversion from CT to NT was estimated as the C change from noninverting tillage to NT in the prairies and from inverting tillage to NT elsewhere in Canada, in all cases under medium carbon input. The increase in perennial crops was estimated as the change from low C input frequent tillage to managed grassland. The difference in total soil C between different management systems was calculated using the default IPCC methodology and divided by the default 20years to produce the 20-yr annual C change factors. RESULTS AND DISCUSSION Table 1 shows the carbon change factors for moving from CT to NT, reduction in fallow and increasing perennial cropping practices that were estimated using the Century model and using IPCC Tier 1 methodology. The factors used in this assessment were appropriate for use at a regional scale. Therefore, we compared the Century-based factors against regional mean factors derived from the experimental studies found in the literature and to the regionally appropriate IPCC Tier 1 factors. Carbon Change Factors for Conversion from CT to NT Table 2 presents a summary of the CT to NT factors derived from empirical data taken from long-term experiments in Canada using the selection criteria described above. The factors from the experimental data exhibit high variability having a standard deviation greater than the average value for each region (Table 3). The mean factor for experiments in the Subhumid Prairie Region was about two and a half times that of the semiarid prairie. The subhumid prairie has higher productivity owing to moister growing conditions. The mean Century-derived factors for both the Semiarid and Subhumid Prairie Region were about 30% lower than the factors from the experiments (Table 3). In a review of Canadian studies, VandenBygaart et al. (2003) calculated an average C change factor of 0.079 0.27 Mg C ha 1 yr 1 for conversion from CT to NT for soil of Eastern Canada, suggesting a limited potential for C sequestration under NT, but also a very high variability in the field data, with sites showing positive and negative C gains. Their analysis included sites from both the East-Central and East-Atlantic Regions. It also included many sites that were not included in the current study as they did not meet the selection criteria described above. In the current analysis, for the East- Central Region, one site showed a net loss and the other two studies showed an increase (Table 2). The broad variation in the empirical values makes it difficult to compare these to the Century-derived factors, but the modelled factor of 0.10 Mg C ha 1 yr 1 lies within the range of values for the empirical comparisons. Note also that the factors derived from the Century model include the crop residue C pool. This can amount to over 2 Mg Cha 1 in a long-term NT corn-based system (Yang and Wander 1999). When soils are sampled to derive the empirical data, residues that accumulate on the soil surface are typically removed before the total C is quantified (Yang and Wander 1999; Paustian et al. 1997). Carbon input to soils is one of the main factors influencing soil C content. The potential of NT to sequester soil C will largely depend on the effects of this practice on crop yields and subsequent C inputs to the soil. Lower crop yields under NT in heavy-textured soils and cool climates might partly explain the limited impact of NT on soil C in the East-Central Region. For instance, the Harrow, ON, experimental site showed lower soil C content with adoption of NT (Table 2), but also lower crop yields with NT than CT (Angers et al. 1997). Unfortunately, there are few studies for other regions. The results show that for the single site in the Atlantic Provinces, NT resulted in a loss of soil C (Table 2). In

676 CANADIAN JOURNAL OF SOIL SCIENCE Table 2. Rates of soil C change as a result of conversion of CT to NT agricultural management from long-term experiments (at least 10 yr) in Canada Location Region Duration (yrs) C storage rate (Mg C ha 1 yr 1 ) Reference Swift Current, SK Semiarid Prairie 11 0.02 z Campbell et al. (1995) Swift Current, SK Semiarid Prairie 12 0.07 Campbell et al. (1995) Swift Current, SK Semiarid Prairie 12 0.07 z McConkey et al. (2003) Stewart Valley, SK Semiarid Prairie 11 0.22 z McConkey et al. (2003) Lethbridge, AB Semiarid Prairie 16 0.10 Larney et al. (1997) Scott, SK Subhumid Prairie 16 0.28 McConkey et al. (2003) Ellersie, AB Subhumid Prairie 11 0.17 Nyborg et al. (1995) Ellersie, AB Subhumid Prairie 11 0.16 Nyborg et al. (1995) Ellersie, AB Subhumid Prairie 11 0.07 Nyborg et al. (1995) Melfort, SK Subhumid Prairie 25 0.48 McConkey et al. (2003) Breton, AB Subhumid Prairie 11 0.31 Nyborg et al. (1995) Breton, AB Subhumid Prairie 11 0.59 Nyborg et al. (1995) Breton, AB Subhumid Prairie 11 0.02 Nyborg et al. (1995) Charlottetown, PE East Atlantic 16 0.26 Carter (2005) St. Lambert, QC East Central 11 0.09 Angers et al. (1995) Elora, ON East Central 25 0.18 Deen and Kataki (2003) Harrow, ON East Central 11 0.09 In Angers et al. (1997) z One tillage per year compared with no-tillage. contrast, the Century-based regional factor for the Atlantic Provinces predicted a modest increase in C from adoption of NT (Table 3). There are no available NT studies for the Pacific Region. Clearly, more results from long-term experiments are needed, especially in the East-Central, East, and Pacific Regions. The results of this analysis show that although the empirical data are highly variable, the Century-derived factors still lie within the range of variability. The modelled factors for conversion of CT to NT generally followed the rankings of the empirical factors with the largest factors in the Subhumid Prairie Region and the lowest in the East Atlantic Region (Table 2 and 3). For the prairies, the IPCC Tier 1 factors for conversion from CT to NT were generally similar, both in terms of total difference in soil C and rate of C change over 20yr, to the Century-derived factors (Table 1). However, for regions outside of the prairies, the IPCC Tier 1 factors for conversion from CT to NT were higher than the Century derived factors, both in terms of total difference in soil C, and rate of C change over 20yr. In addition to the factors described above, this difference may also be related to the fact that total C change was Table 3. Summary of Century-derived and empirically-derived CT to NT factors for Canada assumed to occur over 20yr under Tier 1, whereas the total change occurred over many decades in the Century simulations. Total soil C change for Tier 1 is based on analysis of empirical data, and constraining that change to occur over 20yr is a simplifying assumption. Although, much of the soil C changes under NT can occur in the first decade, it is expected that under the cold conditions of Canada, slow but steady soil C change under varying management practices can take place over a period longer than 20yr (Izaurralde et al. 2001). Smith et al. (2008) summarized mitigation potentials for soil C storage under different management scenarios by thoroughly reviewing the literature for long-term experiments in the four main global climate zones. For tillage and residue management in cool-dry regions similar to the western provinces in Canada, Smith et al. (2008) determined a mean C storage rate of 0.0490.17 Mg ha 1 yr 1. Our estimates of 0.16 and 0.07 Mg ha 1 yr 1 for the subhumid and semiarid Prairies, respectively, (Table 1) are higher but well within their confidence limits. Smith et al. (2008) in cool-moist climate zones, state that tillage and residue management Average C factors Long-term experiments z Predicted Region n C change factor (Mg C ha 1 yr 1 ) Standard deviation C change factor (Mg C ha 1 yr 1 ) Semiarid Prairie 5 0.09 0.09 0.07 Subhumid Prairie 8 0.22 0.25 0.16 East Central 3 0.06 0.14 0.10 East Atlantic 1 0.26 NA 0.06 z Derived from studies listed in Table 2.

Table 4. Soil C change coefficients for conversion from cropland to permanent cover (reseeded grassland or perennial forage). (Note: Does not include native grassland or rotations with perennial forages) Location Soil type Type of comparison Alberta (Bow Island) Brown Fallow-wheat rotation vs grass [crested wheatgrass (Agropyron cristatum L.)] Saskatchewan. (12 sites) Dark Brown and Thin Black Various wheat-based rotations vs restored grasslands Saskatchewan (Scott) Dark Brown Bromegrass (Bromus inermis Leyss.)/alfalfa vs. wheat Saskatchewan (Melfort) Black Fallow wheat rotation vs. 12 yr restored grass Alberta (Lethbridge) Dark Brown Reseeded native grass vs. fallow-wheat rotation Saskatchewan (Swift Current) Brown Fallow-wheat-wheat rotation vs. crestedwheat grass Quebec (La Pocatière) Humic Gleysol Corn (Zea maysl.) vs. alfalfa Ontario (Harrow) Humic Gleysol Cont. corn vs. cont. grass (Poa Pratensis L.) Duration (yr) Soil C change (Mg C) Change rate (Mg C ha 1 yr 1 ) Comments Reference 6 3 0.50 Bremer et al. (2002) 512 0.600.80 12 sites (paired) Mensah et al. (2003) 30 42 1.40 Malhi et al. (2003) 12 00 Change in concentration offset by compaction? Wu et al. (2003) 7 6.6 0.94 Bremer et al. (1994) 10 0 0 Campbell et al. 14 2 0.14 (2000) 5 3 0.6 Angers (1992) 35 37.6 1.07 Gregorich et al. (2001) VANDENBYGAART ET AL. * CANADA CARBON CHANGE FACTORS 677

678 CANADIAN JOURNAL OF SOIL SCIENCE potentially can store 0.1490.14 Mg ha 1 yr 1. This is higher than, but comparable with, the rates of storage of 0.06 and 0.10 for East Atlantic and East-Central regions, respectively, in Canada (Table 1). Carbon Change Factor for Perennial Cropping More intensive pasture management with cultivated forages and areas planted with alfalfa (Medicago sativa L.) in grazing systems have increased in recent decades in the Great Plains of Canada and the United States of America (Entz et al. 2002) and in eastern Canada (Be langer et al. 2002). With regards to the perennial cropping modelled factors in the three western Canadian regions (Subhumid Prairie, Semiarid Prairie, and West Pacific Regions), the rates of C change (20-yr average factor) were considerably lower than in the two eastern Canadian regions (East Atlantic and East Central Regions) (0.460.56 vs. 0.740.77 Mg C ha 1 yr 1 ) (Table 1). Empirically derived factors were highly variable, ranging from 0to more than 1.4 Mg C ha 1 yr 1 across the five regions (Table 4). Reasons for the broad range could include large variation in experiment duration, variable treatments and interactions, different soil sampling strategies and different soil types. Nonetheless, for the western regions the mean of the empirical factors was 0.59 Mg C ha 1 yr 1 and compares favourably to the range of 0.460.56 Mg C ha 1 yr 1 in the modelled factors for these regions. For the two eastern Canada regions only two empirical change factors were available but appeared to also be in line with the modelled values (0.601.07 Mg C ha 1 yr 1 empirical versus 0.740.77 Mg C ha 1 yr 1 modelled). For the regions outside the semiarid and subhumid prairies, the changes in total C for increasing perennial cropping based on Century simulations were comparable to those from Tier 1 methodology (Table 1). However, within the two Prairie regions the annual C change factors from Tier 1 (20-yr annual factor) were considerably higher than those from Century because the total C change was assumed to occur over 20yr under Tier 1, whereas the total change occurred over many decades in the Century simulations, as discussed earlier. Changes in soil C induced by perennials can occur over periods longer than 20yr (Izaurralde et al. 2001). For the semiarid prairies, the total C change from the Century simulations was higher than that for Tier 1 (Table 1). The boundary between dry and moist temperate climates lies within the prairies, and approximately splits the agricultural proportion of the prairies into two equal parts. Therefore, the semiarid prairies are a relatively moist portion of the dry cold temperate climate and the subhumid prairies are a relatively dry portion of the moist cold temperate climate. Hence a more useful comparison is to compare the average total C changes for the two climates in the prairies. When this is done the average Tier 1 IPCC total C change for the prairies of 24.8 Mg C ha 1 is similar to the 27.8 Mg C ha 1 from Century simulations. Carbon Change Factor for Changes in Fallow Summer fallowing is a management strategy used in the Canadian prairies for conserving sufficient moisture in the soil to reduce the risk of crop failure due to drought (Janzen et al. 1998; Campbell et al. 2005). From 1990 to 2004 the area of summer fallow land has decreased by 40% in Canada (NIR 2006). Century simulations were conducted where fallow was replaced by spring wheat (Triticum aestivum L.) in mixed cropping rotations with the same frequency of crops as the average proportion of area in those crops according to the Census of Agriculture (Statistics Canada 2007). To standardize the results, they were expressed on the basis of unit of fallow reduction in the rotation. Thus, the simulated rate of 0.66 Mg C ha 1 yr 1 corresponds to a rate of change of 0.33 Mg C ha 1 yr 1 for conversion of cropfallow to continuous cropping. This was nearly double the average rate of 0.1590.06 Mg C ha 1 yr 1 derived for empirical results from 10studies with 19 comparisons (VandenBygaart et al. 2003) without considering the exact change in frequency of fallow. Campbell et al. (2005) analyzed data mostly from the same studies and also determined a mean rate of 0.15 Mg C ha 1 yr 1, but this was specifically for conversion of crop-fallow to continuous cropping, and these values were found to apply in both the semiarid and subhumid prairies. One important reason for the higher Century-predicted effects of reducing fallow was the large simulated C effect of eliminating fallow in cropping systems that were assumed to contain hay and pasture crops in rotation. Because of the large amount of empirical data for fallow effects, the magnitude of DSOC max from Eq. 1 was set so that the average change after 20years equalled 6 Mg C ha 1 yr 1 (equivalent to average change of 0.15 Mg C ha 1 yr 1 for 20yr after conversion from crop-fallow to continuous cropping). Bare fallow is a rare practice outside of the prairies. Nevertheless, for any changes in area of uncropped land that do occur elsewhere in Canada, in the absence of better data, the fallow reduction factors for the prairies were applied throughout Canada. The factors for regions of Canada outside of the prairies, the total C change and annual factors for decreasing fallow used in the Canadian inventory are similar to those from the IPCC Tier 1 factors (IPCC 2006) (Table 1). For the semiarid prairies, the total change and annual factors for decreased fallow used in the Canadian inventory were higher, but comparable with the Tier 1 values while, for the subhumid prairies, the Canadian inventory values were lower than the corresponding Tier 1 values. As was discussed for the increase in NT and in perennial crops, a more useful comparison between these approaches is to use the average values for the entire prairies. The average Tier 1 total C change for the prairies of 13.4 Mg C ha 1 was similar to the 13.1 Mg C ha 1 average of the empirically-scaled Century-based value. However, the Prairie average Tier 1 annual rate of C change of 0.67 Mg C

VANDENBYGAART ET AL. * CANADA CARBON CHANGE FACTORS 679 ha 1 yr 1 is higher than the 0.30 Mg C ha 1 yr 1 average of the empirically scaled Century-based value. Century predicts that total C change occurs at a lower rate but over a period many decades longer than the 20 yr assumed under the IPCC Tier 1 methodology. Given Canada s cold climate, C change for decreasing fallow is expected to occur over longer than 20yr as shown in long-term field experiments (e.g. Campbell et al. 2005). CONCLUSIONS According to IPCC s Guidelines for National Greenhouse Gas Inventories (IPCC 2006), verifying countryspecific soil C change factors by comparing to independently derived estimates is deemed good practice. The current modelling methodology for deriving soil C change factors in Canada adequately reflects empirical data derived from field experiments even though these data are spatially sporadic and highly variable. For each of the management changes considered, the modelled factors were generally lower than IPCC estimates, and this is partly attributable to differences in calculation methods and to the fact that C changes likely occur more slowly in the cold climate of Canada. Improvements in both the ability of Century to simulate C dynamics under varying management practices and the need to reduce high variability in empirical data should favour derivation of more accurate soil C change factors for cropland soils in Canada for future national inventory calculations. There is a need for initiating more experiments designed to provide measures of the impact of significant management factors on soil C changes across Canada. In particular, experiments conducted in eastern Canada and with sufficient durations were inadequately represented in the empirical data set. Although it is obvious that not all LMC situations in the country could be adequately represented, more reliable data on C change for CT to NT and perennial to annual in eastern Canada would be favourable for more reliable estimates of C change. Agriculture and Agri-Food Canada. 2006. Soil landscapes of Canada (SLC). [Online] Available: http://sis.agr.gc.ca/cansis/ nsdb/slc/intro.html [2006 Aug. 08]. Angers, D. A. 1992. Changes in soil aggregation and organic carbon under corn and alfalfa. Soil Sci. Soc. Am. J. 56: 1244 1249. Angers, D. A., Bolinder, M. A., Carter, M. R., Gregorich, E. G., Drury, C. F., Liang, B. C., Voroney, R. P., Simard, R. R., Donald, R. G., Beyaert, R. P., and Martel, J. 1997. Impact of tillage practices on organic carbon and nitrogen storage in cool, humid soils of eastern Canada. Soil Tillage Res. 41: 191 201. Angers, D. A., Voroney, R. P. and Côte, D. 1995. Dynamics of soil organic matter and corn residues affected by tillage practices. Soil Sci. Soc. Am. J. 59: 13111315. Belanger, G., Rochette, P., Castonguay, Y., Bootsma, A., Mongrain, D. and Ryan, D. A. J. 2002. Climate change and winter survival of perennial forage crops in Eastern Canada. Agron. J. 94: 11201130 Biederbeck, V. O., Campbell, C. A. and Zentner, R. P. 1984. Effect of crop rotation and fertilization on some biological properties of a loam in southwestern Saskatchewan. Can. J. Soil Sci. 64: 355367. Bremer, E., Janzen, H. H. and Johnston, A. M. 1994. Sensitivity of total, light fraction and mineralizable organic matter to management practices in a Lethbridge soil. Can. J. Soil Sci. 74: 131138. Bremer, E., Janzen, H. H. and McKenzie, R. H. 2002. Shortterm impact of fallow frequency and perennial grass on soil organic carbon in a Brown Chernozem in southern Alberta. Can. J. Soil Sci. 82: 481488. Bruce, J. P., Frome, M., Haites, E., Janzen, H., Lal, R. and Paustian, K. 1999. Carbon sequestration in soils. J. Soil Water Conserv. 54: 382389. Campbell, C. A., McConkey, B. G., Zentner, R. P., Dyck, F. B., Selles, F. and Curtin, D. 1995. Carbon sequestration in a Brown Chernozem as affected by tillage and rotation. Can. J. Soil Sci. 75: 449458. Campbell, C. A., Zentner, R. P., Selles, F., Biederbeck, V. O., McConkey, B. G., Blomert, B. and Jefferson, P. G. 2000. Quantifying short-term effects of crop rotations on soil organic carbon in southwestern Saskatchewan. Can. J. Soil Sci. 80: 193202. Campbell, C. A., Janzen, H. H., Paustian, K., Gregorich, E. G., Sherrod, L., Liang, B. C. and Zentner, R. P. 2005. Carbon storage in soils of the North American Great Plains: Effect of cropping frequency. Agron. J. 97: 349363. Carter, M. R. 2005. Long-term tillage effects on cool-season soybean in rotation with barley, soil properties and carbon and nitrogen storage for fine sandy loams in the humid climate of Atlantic Canada. Soil Tillage Res. 81: 10 9120. Deen, W. and Kataki, P. K. 2003. Carbon sequestration in a long-term conventional versus conservation tillage experiment. Soil Tillage Res. 74: 143150. Entz, M. H., Baron, V. S., Carr, P. M., Meyer, D. W., Smith, Jr., S. R. and McCaughey, W. P. 2002. Potential of forages to diversify cropping systems in the Northern Great Plains. Agron. J. 94: 240250. Falloon, P. and Smith, P. 2002. Simulating SOC changes in long-term experiments with RothC and CENTURY: model evaluation for a regional scale application. Soil Use Manage. 18: 101111 Gregorich, E. G., Drury, C. F. and Baldock, J. A. 2001. Changes in soil carbon under long-term maize in monculture and legume-based rotation. Can. J. Soil Sci. 81: 2131. IPCC. 2006. 2006 IPCC guidelines for national greenhouse gas inventories. Vol. 4. Agriculture, Forestry, and Other Land Use, Institute for Global Environmental Strategies, Hayama, Japan. [Online] Available: http://www.ipcc-nggip.iges.or.jp [2007 Jan. 17]. Izaurralde, R. C., McGill, W. B., Robertson, J. A., Juma, N. G. and Thurston, J. J. 2001. Carbon balance of the Breton classical plots over half a century. Soil Sci. Soc. Am. J. 675: 431441. Janzen, H. H., Campbell, C. A., Izaurralde, R. C., Ellert, B. H., Juma, N., McGill, W. B. and Zentner, R. P. 1998. Management effects on soil C storage on the Canadian prairies. Soil Tillage Res. 47: 181195. Larney, F. J., Bremer, E., Janzen, H. H., Johnston, A. M. and Lindwall, C. W. 1997. Changes in total, mineralizable, and light fraction soil organic matter with cropping and tillage

680 CANADIAN JOURNAL OF SOIL SCIENCE intensities in semiarid southern Alberta, Canada. Soil Tillage Res. 42: 229240. Malhi, S. S., Brandt, S. and Gill, K. S. 2003. Cultivation and grassland type effects on light fraction and total organic C and N in a Dark Brown Chernozemic soil. Can. J. Soil Sci. 83: 145 153. McConkey, B. G., Liang, B. C., Campbell, C. A., Curtin, D., Moulin, A., Brandt, S. A. and Lafond, G. P. 2003. Crop rotation and tillage impact on carbon sequestration in Canadian prairie soils. Soil Tillage Res. 74: 8190. McConkey, B. G., Angers, D. A., Bentham, M., Boehm, M., Brierley, A., Cerkowniak, D., Liang, B. C., Collas, P., de Gooijer, H., Desjardins, R. L., Gameda, S., Grant, B., Huffman, E., Hutchinson, J., Hill, L., Krug, P., Martin, T. Patterson, G., Rochette, P., Smith, W., VandenBygaart, A. J., Vergé, X. and Worth, D. 2007. Canadian agricultural greenhouse gas monitoring accounting and reporting system: Methodology and greenhouse gas estimates for agricultural land in the LULUCF sector for NIR 2006. Agriculture and Agri-Food Canada, Ottawa, ON. Mensah, F., Schoenau, J. J. and Malhi, S. S. 2003. Carbon changes in cultivated and excavated land converted to grasses in east-central Saskatchewan. Biogeochemistry 63: 8592. Natural Resources Ecological Laboratory 2007. Century 4. [Online] Available: http://www.nrel.colostate.edu/projects/century/ [2007 Nov. 07]. National Inventory Report. 2006. Greenhouse gas sources and sinks in Canada. 19902004 Greenhouse Gas Division, Environment Canada. Submission to the United Nations Framework Convention on Climate Change, April 2006. Nyborg, M., Solberg, E. D., Mahli, S. S. and Izaurralde, R. C. 1995. Fertilizer N, crop residue and tillage alter soil C and N content in a decade. Pages 93100 in R. Lal, J. Kimble, E. Levine, and B. A. Stewart, eds. Soil management and the greenhouse effect. Lewis Publishers, CRC Press, Boca Raton, FL. Paustian, K., Collins, H. P. and Paul, E. A. 1997. Management controls on soil carbon. Pages 1550 in E. A Paul, K. Paustian, E. T. Elliott, and C. V. Cole, eds. Soil organic matter in temperate agroecosystems: Long-term experiments in North America. CRC Press, Boca Raton, FL. Parton, W. J., Schimel, D. S., Cole, C. V. and Ojima, D. S. 1987. Analysis of factors controlling soil organic matter levels in Great Plains grasslands. Soil Sci. Soc. Am. J. 51: 11731179. Smith, W. N., Rochette, P., Monreal, C., Desjardins, R. L., Pattey, E. and Jaques, A. 1997. The rate of carbon change in agricultural soils in Canada at the landscape level. Can. J. Soil Sci. 77: 219229. Smith, W. N., Desjardins, R. L. and Pattey, E. 2000. The net flux of carbon from agricultural soils in Canada 19702010. Global Change Biol. 6: 558568. Smith, W. N., Desjardins, R. L. and Grant, B. 2001. Estimated changes in soil carbon associated with agricultural practices in Canada. Can. J. Soil Sci. 81: 221227. Statistics Canada. 2007. 2001 Census of Agriculture. [Online] Available: http://www.statcan.ca/english/agcensus2001/index. htm [2007 Jan. 16]. VandenBygaart, A. J., Gregorich, E. G. and Angers, D. A. 2003. Influence of agricultural management on soil organic carbon: A compendium and analysis of Canadian studies. Can. J. Soil Sci. 83: 363380. VandenBygaart, A. J. and Angers, D. A. 2006. Towards accurate measurements of soil organic carbon stock change in agroecosystems. Can. J. Soil Sci. 86: 465471. Yang, X. M. and Wander, M. M. 1999. Tillage effects on soil organic carbon distribution and estimation of C storage. Soil Tillage Res. 52: 19.