- Top of page
- Summary and Conclusions
 Canada's forests play an important role in the global carbon cycle through carbon (C) storage and C exchange with the atmosphere. While estimates of aboveground biomass have been improving, little is known about belowground C storage in root biomass. Here we estimated the contribution of roots to the C budget of Canada's 2.3 × 106 km2 managed forests from 1990 to 2008 using the empirical modeling approach of the Carbon Budget Model of the Canadian Forest Sector (CBM-CFS3) driven by detailed forestry data sets from the National Forest C Monitoring, Accounting and Reporting System. The estimated average net primary production (NPP) during this period was 809 Tg C yr−1 (352 g C m2 yr−1) with root growth and replacement of turnover contributing 39.8 % of NPP. Average heterotrophic respiration (Rh) was 738 Tg C yr−1 (321 g C m−2 yr−1), which resulted in a net ecosystem production (NEP) value of 31 g C m−2 yr−1(71 Tg C yr−1), and on average only 8.7% of NPP remained in the system as NEP. Estimated average root C stocks were 2.38 Pg (1235 g C m−2), mostly in coarse roots (≥ 5 mm diameter), and had an average root to shoot percentage (belowground to aboveground biomass) of 25.6%. Detailed monitoring of C exchange between forests and the atmosphere and an improved understanding of the belowground processes and their response to environmental changes are needed to improve our understanding of the terrestrial C budget.
- Top of page
- Summary and Conclusions
 Globally, forests are important carbon (C) sinks, sequestering ~30% of anthropogenic CO2 emissions annually [Canadell et al., 2007; Pan et al., 2011] but can also act as an important regional net CO2 sources when disturbed, deforested, or degraded [Kurz et al., 2008; van der Werf et al., 2009]. The obvious manifestation of this C sequestration is the accumulation of aboveground tree biomass, but a substantial proportion of forest net primary production (NPP) is directed toward maintenance and accumulation of belowground biomass [Cairns et al., 1997; Heath et al., 2005; Mokany et al., 2006; van der Werf et al., 2009]. Despite the importance of belowground C allocation to the functioning of plant and soil communities, controls on belowground C allocation and the release of C from roots into soil are relatively poorly understood [Jones et al., 2004]. Consequently, our ability to predict how belowground C flows will respond to environmental change remains rudimentary [Giardina et al., 2005].
 Belowground net primary productivity (NPPBG) cannot be measured over large scales using straightforward methods, and therefore it is typically estimated as a function of more readily observable aboveground vegetation dynamics [Running et al., 2004] using relatively simple quantitative relationships. Root turnover and respiration are equally challenging to measure [Kurz and Kimmins, 1987; Steele et al., 1997; Vogt et al., 1986], and representation of these processes is typically highly simplified in terrestrial ecosystem models. Root biomass is also simplified in models by using root to shoot ratios or more complex functional relationships to estimate belowground biomass C stocks as a function of more easily measured aboveground biomass C.
 Advances in Earth observation of land cover dynamics and terrestrial ecosystem C modeling are far outpacing advances in understanding of belowground C dynamics, despite the importance of belowground C flows to the terrestrial C budget [Chapin et al., 2009]. As this knowledge gap widens, it becomes increasingly important to report and evaluate model predictions of belowground C stocks and flows so that these may be critically examined and used to advance our knowledge of these important components of the terrestrial C budget.
 We evaluated national-scale estimates of root biomass C stocks and flows for Canada's managed forests using the Carbon Budget Model of the Canadian Forest Sector, version 3 (CBM-CFS3). The CBM-CFS3 is an inventory-based model of forest C dynamics that consists of a linked set of submodels for live biomass, plant detritus and soil C stocks, forest management, land-use change, and disturbance [Kurz et al., 2009]. This model is used to estimate greenhouse gas emissions and removals from Canada's managed forests, a subset of the entire forest area of Canada [Environment Canada, 2009; Stinson et al., 2011]. Root biomass and turnover parameters in the CBM-CFS3 were developed by Kurz et al.  and Li et al.  using empirical regressions to determine root biomass as a function of aboveground biomass and forest type, allocation of root biomass into fine and coarse roots, and fine root turnover rates.
 The objectives of this study were to (1) develop an updated inventory-based estimate of C in root biomass within Canada's managed forest, (2) examine root to shoot ratios for Canadian forest types, and (3) estimate the contribution of roots to NPP and heterotrophic respiration (Rh). These estimates are described in a manner that is intended to facilitate comparison with field measurements and encourage critical review with the aim of stimulating future advances in simulation modeling of belowground C flows. A final objective was to examine the sensitivity of predictions of dead organic matter (defined as plant detritus and humified organic matter pools including soil C) stocks and flows to changes in root-related model parameters and to assess the impact of altering root parameters on national-scale production and Rh estimates.
- Top of page
- Summary and Conclusions
 Using empirically derived equations of root biomass and turnover along with detailed stand-level inventories from provincial and territorial sources, we have estimated the contribution of roots to national-scale C stocks and flows. Our results have shown that roots have a relatively small contribution to the total biomass stocks, but have a larger contribution to NPP and Rh, mainly associated with replacement of ephemeral fine root biomass. The impacts of adjusting root turnover parameters using recently published values were assessed through national-scale simulations, and the effects on NPP, Rh, and NEP were found to be small (<7%), although estimates of mineral soil stocks were affected.
 In this discussion, we compare our modeling assumptions and stock and flux estimates to published experimental results and identify potential areas of improvement for modeling capabilities as well as data needs. Where possible, modeled results were compared to published studies, which included biome-level or national-scale results, but where these data are lacking, we relied on published case studies.
 Estimated root to shoot percentages compared favorably with values from the literature. For the Boreal region, Mokany et al.  found a slightly higher root to shoot percentage of 39% ± 6% versus the model's predicted value of 31.2% for low aboveground biomass (<7500 Mg m−2) but was consistent with the root to shoot percentage for high aboveground biomass (>7500 Mg m−2), 24% ± 2% versus the model's predicted value of 24.7%. Estimates of root to shoot percentages by Cairns et al.  and Gower et al.  for the boreal region were 27% ± 1% and 28 ± 11%, respectively, which were similar to the overall root to shoot percentage of 25.6% predicted by the national-scale run.
 Our results found that the fine root C biomass density was 168 g C m−2, or 2.7% of the total biomass C, which was consistent with Vogt et al. , who found that fine roots were less than 5% of the total biomass. The predicted coarse root biomass density was 1067 g C m−2, which was comparable to the lower bound of boreal forest estimates by Yuan and Chen  of 1090 g C m−2 (average 1230 g C m−2, standard error of 140, n = 128) assuming a 5 mm diameter threshold, and 16% smaller than the observed value of 1296 g C m−2 by Lavigne and Krasowski .
 The CBM-CFS3 estimate of NPP for the managed forest of Canada, 352 g C m2 yr−1, compared favorably with previous estimates by Chen et al. [2000a], Kang et al. , Li et al. , and Zheng et al.  (Table 4) (see Stinson et al. [2011, Table 4]). Our estimate was lower than the estimate of 422 ± 45 g C m2 yr−1 (standard error) in a review of nine boreal studies in 24 stands [Gower et al., 2001] and within the range of NPP values (271 g C m2 yr−1 to 536 g C m2 yr−1) for the boreal [Luyssaert et al., 2007]. The NPPBG estimate of 140 g C m2 yr−1 for the CBM-CFS3 was very close to the NPPBG by Gower et al.  of 144 ± 25 g C m2 yr−1(standard error) and within the range of boreal forest NPPBG values by Luyssaert et al.  of 69 g C m2 yr−1 to 166 g C m2 yr−1. The ratio of NPPBG/NPP estimated by the CBM-CFS3 was 39.8%, which was similar to the estimates by Gower et al.  of 33% ± 3% (Table 4). Fine root NPP was 31%, which was similar to a global estimate of 33% by Jackson et al. .
Table 4. NPPAG (Aboveground) and NPPBG (Belowground)
|(g C m−2 yr−1)||(%)|
|This study||Canada's managed forest||352||212||140||39.8|
|Luyssaert et al. ||Boreal||271 (n = 38)a 334 (n = 14) 536 (n = 6)|| ||69 (n = 36) Conifer/humid 166 (n = 14) Conifer/semiarid 112 (n = 5) Deciduous/semiarid|| |
|Li et al. ||Prairies|| || ||138||47|
|Gower et al. ||Boreal||422 ± 45b||278 ± 30||144 ± 25 (n = 24)||33 ± 3|
|Steele et al. ||Boreal|| || ||113 (n = 4) Conifer 57 (n = 2) Deciduous|| |
|Gower et al. c||Boreal||273 ± 23 474 ± 90|| ||102 ± 70 Conifer 105 ± 270 Deciduous||39 ± 3 21 ± 3|
 Most of the fine root NPP was associated with the growth of ephemeral fine roots. Roots are generally partitioned into ephemeral fine roots and structural coarse roots using a diameter threshold (e.g., < 5 mm), although consensus on a specific diameter threshold has not been reached [Ruark and Bockheim, 1987; Steele et al., 1997]. Gill and Jackson  have shown that turnover decreases with diameter, and the inconsistencies of defining “fine” make it difficult to compare turnover estimates across studies. In addition, the measurement method influences the estimate of fine root turnover [Hendricks et al., 2006; Steele et al., 1997; Strand et al., 2008] which further complicates study comparisons.
 Fine roots are sensitive to site and stand characteristics such as successional stage, stand composition, and nutrient and moisture availability [Ares and Peinemann, 1992; Brassard et al., 2009; Pinno et al., 2010; Pritchett, 1986; Vogt et al., 1986], and these conditions are inherently poorly known and difficult to model. There is some evidence to suggest that softwood species have a lower fine root turnover rate than hardwoods due to different strategies employed for moisture and nutrient uptake. For example, root longevity in softwood species has been reported to be increased by mycorrhizae due to the physical protection, better access to nutrients and water from the bulk soil, and protection from pathogens [Bauhus and Messier, 1999; Bloomfield et al., 1996; Trofymow and Lalumière, 2011]. A recent review of fine root turnover by Yuan and Chen  found that fine root turnover for roots less than 2 mm diameter was 51% larger for hardwood species than for softwood species.
 In comparison to recent fine root turnover estimates in the literature, the CBM-CFS3 value of 64.1% [Li et al., 2003] is high but is still within the range of recent measurements [Olesinski et al., 2012a; Olesinski et al., 2012b]. Reviews of fine root turnover studies have found lower fine root turnover values of 40% [Gill and Jackson, 2000] and 51% ± 11% [Yuan and Chen, 2010]. Given the lack of consistency in defining “fine”, the uncertainty in the fine root turnover estimate and the demonstrated small impact on national-scale NPP, Rh, and NEP estimates, the CBM-CFS3's fine root turnover value does not warrant adjustment at this time. Additional data are needed to refine the fine root turnover rate to be forest-type specific, and these turnover rates could then be employed in the CBM-CFS3.
 Determination of the appropriate value of the coarse root turnover to compare to the model's default value of 2% yr−1 was difficult, because data on coarse root necromass or turnover rate are seldom reported in the literature, and the few estimates that are available for boreal forests are quite diverse. An earlier review by Gill and Jackson  found an average coarse root turnover of 4.4% ± 0.7% yr−1 for 11 boreal forest estimates excluding peaty forests. A more recent review by Yuan and Chen  found a coarse root turnover of 30% ± 8% yr−1 for 20 boreal forest estimates. Coarse root turnover estimates in models range from 2% yr−1 [Newton, 2006; Rasse et al., 2001] to less than 5% yr−1 (Chen et al. [2000a] used 2.7% yr−1 for softwood and 4.5% yr−1 for hardwood), to 15% yr−1 [Hunt et al., 1999]. Other estimates include a coarse root turnover of 28% yr−1 based on the best fit between a fractal model and field observations of root architecture from two young trees [Nygren et al., 2009] and a turnover of 1.5% yr−1 assuming the turnover rate was similar to that of branches [Peltoniemi et al., 2006]. It is difficult to determine why the coarse root turnover rates are so diverse. Coarse root senescence over years may arise because of adverse growing seasons, herbivory, pathogens, and defoliation. However, most of the coarse root biomass is concentrated near the stump, and while branches die as the canopy lifts with increasing tree height, most coarse roots simply increase in diameter to provide structural support for the tree. Annual coarse root turnover as a proportion of coarse root biomass is thus likely at the lower end of the range of estimates in the literature. Given the variability in the measured coarse root turnover estimates and the limited impact on national-scale NPP and Rh estimates, the CBM-CFS3's coarse root turnover rate does not warrant adjustment, but it would be beneficial to have additional data on coarse root turnover rates.
 The sensitivity analysis revealed that carbon stocks were sensitive to reducing the C concentration of roots from 50% to 47.5%. The default value of the carbon percentage of biomass is 50% in the CBM-CFS3, which is consistent with the default value of 50% found in the Good Practice Guidance of the Intergovernmental Panel on Climate Change  and many other studies, but does not account for some of the observed range of C concentration [Thomas and Martin, 2012]. Both higher (52%) and lower (47.1%) carbon concentrations have been observed in roots [Green et al., 2005], and fine root C concentration has been shown to vary with stand age, soil depth, and ectomycorrhizal association [Trofymow and Lalumière, 2011]. Measured values of the root C concentration could be included in the model if data for Canadian species were available.
 The CBM-CFS3 allocates 50% of the dead roots to C pools within the soil organic horizon and 50% to C pools within the mineral soil, but there is insufficient information to evaluate this assumption. Studies of rooting depth in the boreal generally find shallow rooting depths, with 83% of the fine root biomass located in the upper 30 cm of the soil profile [Jackson et al., 1997]. Root profiles can vary by broad species groups [Finér et al., 1997], moisture and soil conditions [Ares and Peinemann, 1992], and successional stage [Gale and Grigal, 1987; Yuan and Chen, 2010]. Modeling forest C stocks at the national scale excludes such details as the soil moisture content and successional status, but future regional applications of the model could employ alternative allocations of the root C within the soil organic horizons and mineral soil.
 The decay rate of fine roots has recently been estimated from a long-term litterbag study by Currie et al. . Mass remaining time series of three types of fine roots had an asymptotic decay form, with an exponential decay rate of 0.81 yr−1 or 0.56 yr−1 when expressed as a power series decay. This decay rate was similar to the model's power series base decay rate of 0.5 yr−1 at 10°C for dead roots in the very fast pool within the mineral soil. Decay rates from litterbag studies may underestimate the actual decay rate because roots are removed from in situ decay conditions and rhizosphere associations [Dornbush et al., 2002] and because the litterbag mesh can reduce colonization by decay organisms and subsequent decay rates [Setälä et al., 1996]. The decay rate of fine roots has been found to be affected by age [Ruess et al., 2003], species [Finér et al., 1997], soil manganese concentration [Borken et al., 2007], and climate and initial nitrogen concentration [Currie et al., 2010].
 The decay of coarse roots has been estimated from chronosequence studies. Ludovici et al.  found that exponential decomposition rates were 0.053 yr−1 while Chen et al.  found that coarse root decomposition rates were 0.03 to 0.11 yr−1 for the woody portion of the root. These observed decay rates are smaller than the model's base decay rate of 0.1435 yr−1, even after adjustments for temperature and functional form are applied. Additional data on coarse root decay are needed to estimate a national-scale base decay rate and temperature response.
 The recalcitrant portion of dead fine root decay was found to be 32.7% in a long-term litterbag study by Currie et al. , which was higher than the 17% assumed by the CBM-CFS3. A national-scale run with the recalcitrant percentage increased to 32.7% found that the C stocks in the slow pool within the mineral soil were increased by 28.3%. This large change in C stocks would necessitate a recalibration of slow pool decay parameters because mineral soil stocks have been calibrated to reproduce observed C stocks in ground plot data. However, it is beyond the scope of the present study to recalibrate the model to quantitatively predict the change in mineral soil respiration. Based on the results of the present sensitivity study, and that of White et al. , it is clear that the recalcitrant percentage of decayed fine roots has broader impacts and requires further study when more data become available. However, despite the large impacts on estimates of C stocks, the impacts of changing the recalcitrant percentage on estimates of fluxes were very small, with NEP estimates increasing by 1.8%.
 Results presented in this study are national in scale and utilize relationships derived from root study compilations to represent stocks and turnover. However, there are several potential changes to the CBM-CFS3 which could improve predictions at a regional scale. Root productivity, turnover, and decay rates are site and species specific [Bauhus and Messier, 1999; Currie et al., 2010; Finér et al., 1997; González-Molina et al., 2011; Gower et al., 2001; Stover et al., 2007; Yuan and Chen, 2010], but these fine-scale dynamics are presently unaccounted for in the model. Root diseases have not been included in the model, but these could alter mortality, root to stem biomass proportions, and reduce long-term site potential [Cruickshank et al., 2011]. Additionally, the model does not consider impacts on roots from changes in atmospheric carbon dioxide [González-Molina et al., 2011; Heath et al., 2005; Iversen, 2010; Robert B. Jackson et al., 2009; Norby et al., 2004; Stover et al., 2007; Trueman and Gonzalez-Meler, 2005] or temperatures [Eissenstat and Volder, 2005; Gill and Jackson, 2000] or drought [W. Chen et al., 2000b; Hogg et al., 2008], nor does it consider survival of roots after a disturbance for clonal species [Frey et al., 2003].