Skip to main content
Browse Subject Areas

Click through the PLOS taxonomy to find articles in your field.

For more information about PLOS Subject Areas, click here.

  • Loading metrics

Management Impacts on Carbon Dynamics in a Sierra Nevada Mixed Conifer Forest

  • Sabina Dore ,

    Affiliation Department of Environmental Science, Policy, and Management, University of California, Berkeley, California, United States of America

  • Danny L. Fry,

    Affiliation Department of Environmental Science, Policy, and Management, University of California, Berkeley, California, United States of America

  • Brandon M. Collins,

    Affiliations United States Department of Agriculture Forest Service, Pacific Southwest Research Station, Davis, California, United States of America, University of California Center for Fire Research and Outreach, College of Natural Resources, University of California, Berkeley, California, United States of America

  • Rodrigo Vargas,

    Affiliation Department of Plant and Soil Sciences, University of Delaware, Newark, Delaware, United States of America

  • Robert A. York,

    Affiliations Department of Environmental Science, Policy, and Management, University of California, Berkeley, California, United States of America, University of California Berkeley Center for Forestry, Georgetown, California, United States of America

  • Scott L. Stephens

    Affiliation Department of Environmental Science, Policy, and Management, University of California, Berkeley, California, United States of America


Forest ecosystems can act as sinks of carbon and thus mitigate anthropogenic carbon emissions. When forests are actively managed, treatments can alter forests carbon dynamics, reducing their sink strength and switching them from sinks to sources of carbon. These effects are generally characterized by fast temporal dynamics. Hence this study monitored for over a decade the impacts of management practices commonly used to reduce fire hazards on the carbon dynamics of mixed-conifer forests in the Sierra Nevada, California, USA. Soil CO2 efflux, carbon pools (i.e. soil carbon, litter, fine roots, tree biomass), and radial tree growth were compared among un-manipulated controls, prescribed fire, thinning, thinning followed by fire, and two clear-cut harvested sites. Soil CO2 efflux was reduced by both fire and harvesting (ca. 15%). Soil carbon content (upper 15 cm) was not significantly changed by harvest or fire treatments. Fine root biomass was reduced by clear-cut harvest (60–70%) but not by fire, and the litter layer was reduced 80% by clear-cut harvest and 40% by fire. Thinning effects on tree growth and biomass were concentrated in the first year after treatments, whereas fire effects persisted over the seven-year post-treatment period. Over this period, tree radial growth was increased (25%) by thinning and reduced (12%) by fire. After seven years, tree biomass returned to pre-treatment levels in both fire and thinning treatments; however, biomass and productivity decreased 30%-40% compared to controls when thinning was combined with fire. The clear-cut treatment had the strongest impact, reducing ecosystem carbon stocks and delaying the capacity for carbon uptake. We conclude that post-treatment carbon dynamics and ecosystem recovery time varied with intensity and type of treatments. Consequently, management practices can be selected to minimize ecosystem carbon losses while increasing future carbon uptake, resilience to high severity fire, and climate related stresses.


Forest ecosystems constitute a major reservoir of global terrestrial carbon [12] and have the potential to sequester anthropogenic carbon emissions. Understanding carbon cycling in forest ecosystems is therefore critical for estimating the global carbon budget. Today most forests have been altered by centuries of human impact. Management practices of timber harvesting, livestock grazing, and fire suppression/exclusion have altered tree density and species composition [3]. In many western North American forests these changes in tree density and species composition, coupled with increases in surface fuel loads, have increased vulnerability to uncharacteristically large, intense fires [45]. To reduce this vulnerability and to increase forest resilience, management practices often seek to remove biomass either by mechanical methods or prescribed fire [6]. However, these practices represent a disturbance to forest ecosystems, which reduce photosynthetic active surfaces, remove carbon stored on-site, and alter physical and chemical properties of soils [7]. Even if treatments cause short-term carbon losses, these losses need to be weighed against extended and extensive losses that can be caused by high severity fires [810]. It is also necessary to consider other possible benefits of treatments such as enhancement of forest resilience and resistance, increased biodiversity, improvement of hydrological benefits, and erosion protection.

Fuels reduction and harvesting treatments (i.e. prescribed fire, thinning, and clear-cut harvest) can affect ecosystem carbon pools and soil CO2 efflux. Soil CO2 efflux (Fs) is the flux of microbial and plant-respired CO2 that represent the largest source of CO2 from terrestrial ecosystems to the atmosphere [1]. Thinning and prescribed fire represent partial, or low severity forest disturbances, in that they are not stand-replacing. Partial disturbances are important because they affect more forested area than do stand replacing disturbances [1112], and because their occurrence is predicted to increase with climate change [8, 13].

Recent literature documenting the effects of thinning and prescribed burning in the southwestern U.S. region integrate the analysis of the effects of disturbances on commercial timber yield with different ecosystem components, including understory vegetation [14], carbon storage, and carbon fluxes [7, 1520]. These studies show that severe disturbances often switch forests from sinks to sources of carbon [16, 21]. Post-disturbance emission of carbon by decomposition can exceed assimilation of carbon by vegetation for decades [2122]. In addition, these studies demonstrate that the trajectory and the time needed for ecosystem recovery have implications for long-term and regional scale carbon balances [9, 23]. Therefore, it is important to monitor ecosystems for longer periods (>5 years) to fully capture the implications of management on ecosystem carbon dynamics [12, 22]. The present research is among few studies (see also [2324]) to include long-term measurements of carbon stocks and dynamics, starting seven years before treatments and continuing for 10 years post-treatment.

This study examines the effects of prescribed fire, thinning, and clear-cut harvesting on ecosystem carbon pools (tree biomass and soil pools), tree radial growth, and Fs in a mixed conifer forest in the Sierra Nevada. Our results increase the understanding of forest ecosystem carbon dynamics following management treatments and their dependence on treatment type, intensity and frequency. Our objectives were to: 1) quantify treatment effects on Fs. In particular separating the effects of changes in soil temperature and water content from changes in basal rates (Fs at a reference soil temperature); 2) compare relationships between Fs and soil temperature, CO2 concentrations and diffusion through the soil profile across different forest treatments; 3) assess the magnitude of treatment effects on carbon stocks, and their dependence on treatment intensity and type, and time since treatment; 4) quantify effects of thinning and fire on tree radial growth. Our ultimate goal is to inform how forest management practices can be selected to minimize forest carbon losses while increasing future carbon uptake, resilience to high severity fire, and climate related stresses.


Study site

The study was conducted at Blodgett Forest (38°54′N, 120°39′W), a University of California Research Station on the western slopes of the northern Sierra Nevada approximately 21 km east of Georgetown, California. Blodgett is a mixed-conifer forest composed of sugar pine (Pinus lambertiana), ponderosa pine (Pinus ponderosa), white fir (Abies concolor), incense-cedar (Calocedrus decurrens), Douglas-fir (Pseudotsuga menziesii), and California black oak (Quercus kelloggii). Elevation ranges from 1100 to 1410 m. Total annual precipitation averages about 1600 mm, falling mainly between September and May. The average minimum daily temperature in January is 0.6°C and the average maximum daily temperature in July is 28.3°C [25]. The loamy-sandy soils are underlain by Mesozoic, granitic material and are predominantly classified as the Holland and Musick series (fine-loamy, mixed, semiactive, mesic Ultic Haploxeralfs [26]).

The forest is an actively managed timberland and has been repeatedly harvested with both even- and uneven-aged methods. Wildfires have been suppressed at Blodgett in the last century, reflecting a management history common to many forests in California and elsewhere in the western United States [27]. Prior to fire suppression, the median composite fire interval at the 9–15 ha spatial scale was 4.7 years with a range of 4–28 years [28].

Forest management treatments

This study includes six treatment types: un-manipulated control (CTRL), prescribed fire (FIRE), thinning (THN), thinning plus prescribed fire (THN+FIRE), and clear-cut harvest with and without mechanical soil ripping (HARVRIP, HARVNO_RIP, respectively). The overall goal of the FIRE and THN practices were to reduce the vulnerability of these forests to high-severity fires [29]. Soil ripping is a common post-harvesting practice used in commercial forests to reduce soil compaction and prepare the soil for seedling planting.

The treatments were applied to forest stands with similar climatic conditions, as plots were established within a 10 km radius. Pre-treatment species composition was similar among stands (S1 Table). Calocedrus decurrens and Abies concolor were the most common species, accounting for about 37% (ranging 34%–40% across treatments) and 27% (ranging 24%–30% across treatments) of the trees, respectively. Pseudotsuga menziesii accounted for 13% of the trees (ranging 11%–17% across treatments) and the combination of Pinus lambertiana and Pinus ponderosa accounted for 8% of the trees (ranging 6%–13% across treatments). The FIRE, CTRL, THN and THN+FIRE treatments were part of the Fire and Fire Surrogate Study (FFS). This study began in 2000 and analyzed the ecological effects of fuel treatments on vegetation structure and other ecosystem characteristics across a network of 13 locations in the United States. Additional information about the study can be found elsewhere [30]. In addition to the FFS treatment sites, we included four sites (each between 2000 and 7000 m2) that were clear-cut harvested in the summer of 2010. In these sites all trees were harvested and the residual material was piled and burned. In half of the harvested sites a tractor mechanically ripped soils with a wing-tipped subsoiler. In fall 2010, one-year old ponderosa pine seedlings were planted in pairs, on a 5 x 5 m grid spacing (840 seedlings ha-1). For clarity, a schematic diagram showing locations and timing of treatments and measurements is included in Fig 1.

Fig 1. Study layout.

Map of the Blodgett Forest. The control, mechanical thinning, prescribed fire, thinning followed by prescribed fire treatments each had three replicates. Soil CO2 flux and soil carbon pools were measured in the clear-cut harvest treatment area and one of the replicates of the control and prescribed fire treatments. Timing of treatments (treat.) and measurements (meas.) during a period of 17 years, from 1995 to 2012, is also shown.

We analyzed tree-ring growth and stand biomass in the CTRL, FIRE, THN and THN+FIRE treatments. Measurements included three replicates per treatment (Fig 1) and continued until 2009, seven years after the prescribed fire and mechanical treatments. Because of practical limitations and instrumentation cost, the analysis on Fs and its driving variables (i.e., soil temperature and water content, fine roots, litter and soil carbon) was conducted in one, randomly selected replicate of the CTRL and FIRE treatments (CTRLFs and FIREFs) and the newly installed HARV sites (Fig 1). In these four areas soil CO2 fluxes, litter, soil carbon, and fine root biomass were measured in 2011 and 2012 (more details on timing in Fig 1).

Soil CO2 efflux measurements

Measurements of Fs were taken using the chamber technique [31] and the soil CO2 profile technique [32]. At each site (CTRLFs, FIREFs, and HARVRIP and HARVNO_RIP), we periodically measured Fs in multiple locations using the chamber technique and continuously monitored Fs with the profile technique. Our approach of complementing the chamber and soil CO2 profile techniques allowed us to quantify both spatial and temporal variability in Fs [33].

a. Soil CO2 chamber technique.

We measured soil CO2 efflux using a Li-6000, and later a Li-6400, with a 10 cm diameter soil chamber attachment (Li-Cor, Lincoln, USA) at 110 locations (total for the CTRLFs, FIREFs, and HARVRIP and HARVNO_RIP sites) bi-monthly during snow free periods (June 2011 to December of 2012). In the spring of 2012 the Li-6000 and Li-6400 were compared in the field obtaining good agreement (r2 = 0.95) and therefore, the Li-6000 data were increased for the 15% difference found on the regression slope (data not shown). Measurements across all plots were taken over a two-day period, between 0900 and 1700 hours. The order of the sites and plots measured was randomly changed during each sampling date to avoid a systematic bias based on potential temperature changes. During these manual measurements, the chamber was positioned on PVC soil collars installed 1 cm into the soil to avoid soil disturbance and to allow repeat measurements on the same locations. Soil CO2 efflux was calculated from the change of CO2 concentration over time and averaged for two cycles over a 10 ppm range encompassing the ambient CO2 concentration. Soil water content (SWC, integrated over the 0–5 cm depth using a HH2 and ML2x, DeltaT devices, Cambridge, UK) and soil temperature (Ts, measured at 10 cm depth using a 6000-09TC, Licor, Lincoln, USA) were measured adjacent to each Fs collar.

In both the FIREFs and CTRLFs sites, measurements were collected on 29 different locations per site distributed over 3 ha. In each the HARVRIP and HARVNO_RIP sites, we measured Fs in 20 different locations distributed over 6 ha due to the spatial arrangement of the harvest units. More information about measurement location and spatial variability can be found in Dore et al. [34].

b. Soil CO2 profile technique.

We used solid-state CO2 sensors (CARBOCAP model GMM 220, range 0–1%, Vaisala, Helsinki, Finland) installed at 2, 8, 16, and 24 cm soil depths to measure Fs at each site (CTRLFs, FIREFs, and HARVRIP and HARVNO_RIP). Ts was measured with thermocouples at the same depth used for the CO2 sensors, and SWC (ECH2O, Decagon Devices, Pullman, WA, USA) was measured at two depths, 8 and 16 cm. Site specific calibration factors of SWC sensors and their dependence on Ts were quantified in the lab using each site’s mineral soil sampled over a 20 cm depth. In winter 2012 sensors were calibrated against known CO2 concentrations in the lab, and specific parameters determined during calibration were applied to each probe at each site.

CO2 concentrations from the sensors were measured every 15 seconds and averaged over 30 minute intervals. The concentrations were corrected for soil temperature and atmospheric pressure and were used to calculate Fs (expressed in μmol m-2 s-1) based on Fick's first law of diffusion Fs = -Ds dC/dz [32, 35]. Ds is the diffusion coefficient in the soil and C is the CO2 concentration at depth z. Ds is calculated by the product of the CO2 diffusion coefficient in the free air Da, and the gas tortuosity factor ξ.

The factor ξ can be quantified empirically by quantifying its relationship with soil porosity (ε), which is determined by the SWC, bulk density, and particle density for the mineral soil [35]. We determined site and depth specific ξ in the lab using intact soil samples collected adjacent to the profiles (two replicates per site at 3 depths: 0–5 cm, 10–15 and 15–20 cm; 24 samples total). A sealed 100 ml volume above a 5 cm diameter, 5 cm long aluminum cylinder containing the undisturbed soil sample was initially flushed with nitrogen. Air was able to diffuse from the bottom of the cylinder through the soil. We measured the increase of oxygen with time in the volume above the soil (using a SO-200 Apogee, Logan, USA) for a period of 3–5 minutes after a 2-minute period necessary to reach a constant rate. For each site and depth, measurements were repeated at four different water contents (n = 96 measurements). Following each diffusion measurement SWC was determined using the fresh weight of the sample and bulk density obtained from its dry weight (102°C for 24 hours and until constant weight) at the end of all measurements. The resulting equation ξ = 0.0139e6.2889*ε (r2 = 0.7, p< 0.001) was common to all soils from different sites and depths (Fig 2). To calculate Fs, however, we used the relationship between the tortuosity factor ξ and SWC (to obtain an exponential trend we used 1- SWC). This relationship ξ = 0.007e4.094*(1-swc) had a higher coefficient of determination (r2 = 0.85, p< 0.001) and was able to quantify the diffusion coefficient independently from accurate site/depth specific measurements of soil bulk density and particle density for the mineral soil. The effects of these soil characteristics were included in the site/depth specific relationship between SWC and ξ.

Fig 2. Soil tortuosity and diffusion.

(a) Comparison of the empirically determined soil tortuosity factor at the control (CTRLFs), fire (FIREFs), and tree harvest with (HARVRIP) and without soil ripping (HARVNO_RIP) treatment sites. Values predicted from the exponential Moldrup curve are included for comparison [35]. (b) Seasonal trend of the diffusion coefficient at the treatment sites calculated from the specific relationship between tortuosity factor and soil water content ξ = 0.007e4.094*(1-swc) and the CO2 diffusivity in free air, which was affected by soil temperature, atmospheric pressure, and soil water content monitored at each site.

The CO2 profile data were adjusted to match the mean spatial values of the chamber data for 2011 and 2012 (S1 Fig). To achieve this, we used for each site and year the slope and intercept of the linear relationships between daily averages of Fs simultaneously measured by chambers and each site-profile. Thus, the adjusted values represented the treatment-scale Fs and had at the same time high temporal resolution due to continuous measurements. At each site, daily averages of the adjusted Fs were fitted with daily averages of Ts and SWC profiles for 2011 and 2012 using the equation: (1) where S10 is basal respiration at 10°C, Q10 is the parameter reflecting the temperature sensitivity of Rs, and a and b are the parameters of a simplified Gompertz function [36]. The measurement depth of Ts and SWC profiles which was able to explain most of the measured seasonal variability in Fs was used to parameterize the equation. We used Ts at 10 cm at the HARV and FIREFs sites and Ts at 24 cm at the CTRLFs site. SWC at 8 cm was used at all sites. The fitted equations explained 60% to 85% of the measured Fs variability (Table 1).

Table 1. Parameters of CO2 soil efflux Eq (1), annual total soil CO2 emissions, and soil temperature and water content.

Annual fluxes for 2011 and 2012 were obtained by adding daily Fs values derived from Eq 1. To calculate uncertainty for annual fluxes at each site we randomly selected 1000 sets of Eq (1) parameters, with each parameter drawn from normal distributions characterized by their fitted means and standard deviation. Each set of parameters was used to compute Fs for every day of the year applying the site specific Ts and SWC. The 1000 annual values calculated from each set of parameter were used to calculate uncertainty as 95% confidence interval.

Effects of treatments on Fs were analyzed using different methods. First, we compared chamber fluxes measured on the same day at the different sites, including with this approach both the effect of treatments on basal Fs and the effect due to post-treatment differences in Ts and SWC. Second, we used a general linear model of the form Fs = f(treatment, Ts, SWC) to assess the effect of treatments on Fs correcting for the effect of Ts and SWC, thus removing across-treatment differences in environmental conditions displayed during measurement campaigns. Third, we quantified the treatment effect over the different seasons, comparing the annual sums of Fs. To separate how much of the difference in annual fluxes among sites was due to the effects of treatments on Fs basal rates and how much was due to changed microclimatic conditions, we modeled Fs at the FIREFs and HARV sites using the CTRLFs site’s climatic inputs, as if treatments at these sites didn’t affect environmental conditions but only the Fs basal rate.

Soil and fine root carbon pools

Fine root biomass was measured in summer 2012. Soil samples were collected from 0–15 and 15–30 cm, with a 20 cm2 area auger at 10 locations adjacent to randomly selected Fs collars at each of the CTRLFs, FIREFs, HARVRIP and HARVNO_RIP sites (i.e., 20 samples per treatment). Both live and dead fine roots were hand-picked and separated in < 2mm and 2–5 mm diameter classes from each soil core. Fine root dry weight was determined after drying samples at 70°C for 24 hours. The litter layer was collected inside the 10 cm diameter Fs collars (27 plots at the FIREFs and CTRLFs sites, 18 plots at both HARV sites) at the end of 2012, and dried at 70°C for 4 days before dividing it into leaves and woody components.

Bulk density of the mineral soil layer (0–5 cm depth) was determined with 5 cm diameter cores collected inside the Fs collars (27 plots at both the FIREFs and CTRLFs sites, 18 plots at both the HARV sites) after removing the litter layer, at the end of the Fs sampling campaign. Bulk density of the deeper 5 to15 cm soil layer was determined on 8 locations (5 cm diameter, 10 cm long samples), adjacent to randomly selected Fs collars at each site. Soil carbon content was determined on 18 samples per site on a 5 cm diameter, 15 cm deep volume. Each sample was separated into two depths, 0–5 cm, and 5–15 cm. Samples were air dried, sieved (2 mm), ground, and finally their carbon content was determined using an elemental analyzer (Thermo Flash 2000, Thermo scientific, MA, USA).

Tree biomass and annual tree radial growth

Tree biomass and tree radial growth were measured in the CTRL, FIRE, THN, and THN+FIRE treatments. Each treatment had three replicates ranging from 12 to 28 ha (12 experimental plots total). Tree diameter at breast height (DBH), species, and status were measured in 20 circular, 0.04 ha sub-plots distributed on a systematic grid in each experimental plot (240 sub-plots total). Only trees with diameters larger than 5 cm were considered in this study. Measurements of the same plots were collected the year before treatments (2001), the first post-treatment year (2003), and the seventh year post-treatment (2009).

In 2010, in each of the experimental plots, 30–60 trees of the five major species (Pinus lambertiana, Pinus ponderosa, Abies concolor, Calocedrus decurrens, Pseudotsuga menziesii), distributed over five size classes (DBH < 35 cm, 35–55 cm, 55–75 cm, 75–95 cm, and DBH > 95 cm), were cored and tree-ring widths measured to 0.001mm using a sliding stage to quantify annual growth from 1995 to 2009. The chosen starting point for this analysis was 1995 to have the same length, a seven-year period, before and after treatments.

For each treatment we determined annually individual and mean tree radial growth. The individual tree radial growth is the radial growth of a hypothetical average tree. We calculated it annually for each treatment by averaging the radial growth of each size class and species (i.e. 20 increments per each treatment). The annual mean tree radial growth was calculated by averaging the annual radial growth of all trees present in each treatment. The mean tree radial growth included both the effect of treatments on the growth of each tree (a positive effect in case of reduced competition or a negative effect in case of treatment related injuries) and the effect of decreased tree density (number of surviving trees changing with treatment and year). We used for each tree the mean annual radial growth of its species, experimental plot, and size class. If a size class had no samples, its growth was estimated as the average of the two adjacent classes (same species, year and experimental plot) or as the growth of the closest class (in case of missing smallest or biggest classes). Only 6% of the trees were in size classes with no tree sampled, and of those, 67% had DBH< 35 cm.

We used the allometric equations provided in Jenkins et al. [37] to calculate aboveground tree biomass for each year. For the period 1996 to 2001, diameters were back calculated from all trees initially measured in 2001. We assumed mortality during this period was zero at all sites, as natural cumulative mortality at the CTRL treatment was only 4% between 2001 and 2009. For the period 2002 to 2005 diameters were calculated from all trees remaining after treatments in 2003. For the period 2006–2009 diameters were calculated from the trees recorded in 2009. Coarse root biomass was calculated from aboveground biomass using the equation for temperate forests in Cairns et al. [38]. Aboveground and coarse root biomass was converted to carbon assuming a carbon concentration of 48% [39]. The difference in stand biomass carbon stocks between two consecutive years was used to express the stand annual aboveground tree productivity.

In fall 2010, one-year old ponderosa pine seedlings were planted in pairs on a 5 x 5 m grid in all two HARV sites. To calculate seedling biomass and to analyze the effects of mechanical soil ripping on growth rates, diameter and height of 240 to 400 seedlings were measured in each of the HARVRIP and HARVNO_RIP treatment sites during the summer of 2012. Eighty-five seedlings were collected and dry weight of needles and wood determined. The relationship between diameter and dry weight of needles, wood, and the whole plant was determined separately. The resulting equations were bL = 1.5305 D2.1416 (r2 = 0.73) for needles, bs = 0.9741 D2.2691 (r2 = 0.87) for stems, and bP = 2.5502 D2.1834 (r2 = 0.82) for the whole plant, where bL, bS, and bP was the dry weight of needles, stems, and whole plant, respectively, and D was the basal diameter (in cm). The equation for the whole plants was used to calculate the carbon stored as tree biomass at the HARV sites.

Data Analysis

When Fs measured during the same measurement campaign was compared across treatment sites, effects of treatments on Fs, Ts, and SWC were calculated from the slope of the linear relationship between chamber mean at each treatment site against simultaneously measured values at the CTRLFs site. To evaluate the treatment effects we tested if mean values of Fs, Ts, and SWC at the treatment sites were significantly different using one way repeated measures analysis of variance (ANOVA). When comparing differences of annual Fs (and diffusivity) among treatments, we used one-way ANOVA based on daily means. To compare soil carbon, litter, and fine root biomass among CTRLFs, FIREFs and HARV sites we used one-way ANOVA.

To quantify the effects of the treatments on tree-ring growth, tree and stand biomass carbon stocks we calculated the effect size using the Before-After Control-Impact (BACI) approach [40]. The BACI effect size = (μca - μcb)—(μta - μtb) is the differential change in means (μ) between the control (c) and the treatment sites (t) before (b) and after (a) treatment. The effect was evaluated by testing the existence of an interaction between period and treatment using a two-way repeated measures ANOVA and using the seven year averages for pre and post-treatment periods when available. Differences were considered significant at α <0.05 level.

The Tukey post hoc test was used to make multiple comparisons among treatments if a significant difference was detected. Normality and equal variance tests were conducted and, if necessary, data were log transformed to meet these conditions. S1 Table summarizes and describes statistical tests used to compare stand characteristic, carbon pools, or fluxes among treatments.


Effect of treatments on microclimate

Both fire and tree harvesting reduced tree density of the stands (p < 0.001; Table 1 and S1 Table). Consequently the amount of energy reaching the ground increased, as the direct comparison of Ts simultaneously measured at the different sites showed. Soil temperature was 30% higher in the HARV site and 16% higher in the FIREFs site (p < 0.001; Fig 3b, S1 Table) compared to the CTRLFs site. SWC was 29% higher at the HARV site and 11% higher at the FIREFs site compared to the CTRLFs site (p < 0.001; Fig 3c, S1 Table). Thus, differences in both Ts and SWC with the CTRLFs site were highest at the HARV sites where all vegetation was removed. Seasonal variations of Ts and SWC were highest at the HARV sites and lowest at the CTRLFs site.

Fig 3. Effects of treatments on soil CO2 efflux and soil environmental conditions.

Comparison of (a) soil CO2 efflux, (b) soil temperature, and (c) soil water content (SWC) between fire (FIREFs) and harvest (HARV) treatment sites and the control site (CTRLFs). Each symbol represents the average of the 20–29 plots measured in 2012 using the chamber technique. The harvest data is the average of values from units with and without soil ripping. Slope and r2 of the linear regression are also shown.

Effect of treatments on soil CO2 efflux

Soil CO2 efflux (measured using the chamber technique) was reduced by treatment (p < 0.001; S1 Table). Comparing Fs measured at the different sites on the same measurement date, the FIREFs and HARV sites (average of HARVRIP and HARVNO_RIP) were 14% and 18% lower, respectively, compared to the CTRLFs site (Fig 3, S1 Table). Soil CO2 efflux was not significantly different between HARVRIP and HARVNO_RIP sites (p > 0.05).

CO2 concentrations at HARV sites were higher and showed greater seasonal variation (Fig 4) than at the FIREFs and CTRLFs sites, especially at deeper depths (16 and 24 cm). Concentrations of CO2 in a soil layer are the result of production and the capacity of CO2 to exit the soil (i.e. diffusion). The diffusion coefficient was highest at the CTRLFs site and lowest at the HARVRIP site (p < 0.001; Fig 2b, S1 Table).

Fig 4. CO2 concentration.

Concentration of CO2 measured at 2, 8, 16 and 24 cm depths at the control (CTRLFs), prescribed fire (FIREFs), and harvest (HARV) treatment sites in 2012.

FIREFs and HARV sites had reduced annual Fs compared to the CTRLFs site (p < 0.001; Table 1 and S1 Table). The annual emissions of treated sites decreased despite having higher Ts and SWC (Table 1 and S1 Table). Seasonally, sites were more similar during winter and spring, when differences between treated and control site were on average 0.5 μmolm-2 s-1 (Fig 5a) compared to the summer, when difference were 2.5 μmolm-2 s-1. In July-August, when soil water availability decreased (Fig 5c), Fs decreased at all sites. Decrease in Fs started 2–5 weeks earlier at the treated sites than at the CTRLFs site for both years (Fig 5a).

Fig 5. Seasonal trends of soil CO2 efflux, soil temperature and soil water content.

(a) Soil CO2 efflux, (b) soil temperature (10 cm depth), and (c) soil water content (8 cm depth) recorded during 2011–2012 from profiles installed at the control (CTRLFs), prescribed fire (FIREFs), and tree harvest with (HARVRIP) and without soil ripping (HARVNO_RIP) sites. (d) Comparisons of soil CO2 efflux at the control site (CTRLFs), and soil CO2 efflux at the fire (FIREFs) and harvest (HARV) sites modeled using climate data recorded at the control site.

Effect of treatments on soil carbon, fine roots, and litter

Soil carbon (up to a depth of 15 cm) was different among treatment sites when expressed as % content, however not when expressed as g C m-2, due to the different soil bulk densities of the sites (p = 0.03; Table 2 and S1 Table). Fine root biomass was similar at the FIREFs site compared to the CTRLFs site (583 and 517 g C m-2 respectively), but was reduced by 60–70% (to 143 g C m-2 at the HARVNO_RIP and 191 g C m-2 at the HARVRIP site) two years after the harvest. The litter layer was reduced at all treatment sites compared to the CTRLFs site, ranging from -40% in the FIREFs (p < 0.008) and HARVNO_RIP sites to –80% in the HARVRIP site (p < 0.001; Table 2 and S1 Table). If we consider carbon stored as mineral soil, fine roots, and litter layer, three years after the second fire (initial 2002, re-burn in 2009, Fig 1), the FIREFs site had 19% less carbon than the CTRLFs site, and two years after the clear-cut the HARV site had 26% less carbon than the CTRLFs site.

Effects of treatments on forest biomass and tree growth

Pre-treatment biomass and tree density did not differ significantly among sites (p >0.05; Table 1 and S1 Table, Fig 6a). All treatments mostly removed or killed small diameter trees (DBH < 30 cm, Fig 6b)

Fig 6. Stand characteristics.

(a) Distribution of trees (> 5 cm DBH) in 5 cm diameter classes in the prescribed fire (FIRE), control (CTRL), mechanical thinning (THN) and mechanical thinning followed by prescribed fire (THN+FIRE) treatments at Blodgett Forest (3 replicates per treatment). Tree diameter distribution was measured the year before the treatment (PRE) in 2001, the first post-treatment year in 2003 (POST_1), and seven years post-treatment in 2009 (POST_7). Each class is labeled with the lower interval limit. (b) Difference in number of trees between 2001 and 2003 (POST_1) and 2003 and 2009 (POST_7) by diameter class.

Over seven years after the treatments, individual tree annual radial growth decreased after FIRE (-10%; p < 0.001), increased 25% after THN (p < 0.001), and was not significantly different after THN+FIRE (p = 0.51 Fig 7a, Table 3 and S1 Table). For each species (Pinus lambertiana, Pinus ponderosa, Abies concolor, Calocedrus decurrens, Pseudotsuga menziesii) FIRE decreased the annual radial growth by 7–20%, whereas THN increased annual radial growth by 13–30%. The THN+FIRE treatment had a mixed effect, increasing annual radial growth in some species (Abies concolor + 33%) but decreasing it in others (Pseudotsuga menziesi—13%; Table 3 and S1 Table). Effect of treatments on individual tree annual radial growth depended on tree size. Annual radial growth increased most in smaller diameter trees (<35cm diameter) for all treatments, with the highest increase at the THN treatment (52% increase compared to a 3% decrease at the FIRE treatment, Table 3).

Fig 7. Tree growth and productivity.

(a) Annual individual tree radial growth (tree ring widths, in mm yr-1) for the control (CTRL), prescribed fire (FIRE), mechanical thinning (THN), and mechanical thinning followed by prescribed fire (THN+FIRE) treatments from 1986 to 2009 at Blodgett Forest. Samples were collected from each of the five tree species equally distributed among size classes. (b) Annual mean tree radial growth (mm yr-1), calculated as the average tree growth of all trees in each treatment, identifies both the effect of treatments on tree increment growth (a) and the effect of treatments on tree density. (c) Stand annual productivity was expressed as the difference in stand biomass (aboveground) between two consecutive years. The rectangles show the treatment year 2002. Data are the averages from three replicates.

Table 3. Tree biomass carbon stocks, growth and productivity.

When averaging the tree annual radial growth over the stand, because of the lower post-treatment tree density (Fig 7b), the before-after analysis showed mean annual radial growth decreased 27% after FIRE (p = 0.03) and was not affected by THN (p = 0.4) and THN+FIRE treatments (p = 0.42; Table 3 and S1 Table). All active treatments reduced stand aboveground productivity (expressed as difference between biomass of two consecutive years), however the THN+FIRE treatment had the lowest post-treatment average productivity of 3 t C ha-1 (2003–2009), approximately 50% lower than before treatment (p < 0.001; Fig 7c, Table 3).

The CTRL treatment was still accumulating carbon during the observed 14-year period. However, radial growth decreased between the periods 1995–2001 and 2003–2009 from 3.04 (±0.14 SD) to 2.65 (±0.31 SD) mm yr-1. From 2001 to 2009, carbon stored as tree biomass increased from 205 (±37 SD) to 240 (±39 SD) t C ha-1 at the CTRL treatment, was reduced ca. 10–15% by FIRE (p = 0.005) or THN (p = 0.012), and 36% by the THN+FIRE treatment (p <0.001; Table 3 and S1 Table). All active treatments decreased tree density, however the greatest decrease occurred in the THN+FIRE (54% effect size, p <0.001; Fig 6, Table 3 and S1 Table), and was higher after THN than after FIRE (43% decrease compared to 32% respectively). In the FIRE treatment the number of live trees decreased substantially (> 100 trees ha-1, p = < 0.001) between 2003 and 2009. This further decrease between 2003 and 2009 occurred also at the THN+FIRE treatments, compared to a more stable condition at the THN treatment (p = 0.88; Fig 6, Table 3 and S1 Table).

In the first year after treatment, the FIRE treatment did not lose any carbon stored as tree biomass compared to losses of 29 (±0.10 SD) t C ha-1 in the THN treatment and the 52 (±0.11 SD) t C ha-1at the THN+FIRE treatment (Table 3 and S1 Table). However, between the first and seventh post-treatment years, the FIRE and the THN+FIRE treatments both accumulated a total of ca. 5 (±3 SD) t C ha-1 compared to 30 (±3 SD) t C ha-1 accumulated by the THN treatment, and 27 (±3 SD) t C ha-1 at the CTRL. Both the FIRE and the THN treatments recovered to pre-treatment biomass carbon stocks levels by the seventh post-treatment year (p = 0.13; Table 3 and S1 Table).

At the HARV sites (Table 4), soil ripping did not benefit the seedlings 20 months after planting. No differences were found between HARVRIP and HARVNO_RIP in survivorship, height, and diameter and thus biomass carbon stocks, still ~1 g C m-2.

Table 4. Characteristics of Pinus ponderosa seedlings at Blodgett Forest.


Prescribed fire, thinning, and clear-cut harvest all affected measured pools of carbon in mixed conifer forests in the north-central Sierra Nevada. These treatments also affected ecosystem carbon fluxes, decreasing carbon sequestered annually as tree biomass, or released from the soil. Prescribed fire and thinning affected biomass carbon stocks and growth differently. Over seven post-treatment years, the THN treatment showed a stronger sink strength and reached higher biomass levels compared to treatments that include fire. Our results confirmed that biomass carbon stocks and growth depend on the intensity and type of treatment, and that it is possible to design forest restoration treatments that achieve both increased forest resiliency and carbon storage.

Treatments and soil CO2 efflux

The effect on Fs was in part due to the reduction of tree cover, which affected the main drivers of ecosystem processes: energy and water (Table 1 and S1 Table). In addition, forest management treatments altered basal Fs (Table 1 and S1 Table), which is related to biotic factors such as soil microbes, root biomass, and soil organic carbon content. The lower basal respiration at the HARV site can be explained by the lack of autotrophic CO2 production in the soil.

When we remove the effect of environmental conditions by modeling a hypothetical Fs at the FIREFs and HARV sites using the CTRLFs site’s climatic inputs (Fig 5d), we obtained annual Fs lower (circa 300 g C m-2 year-1) than the annual current rates. Thus effects of forest management treatments and effects on microclimate altered soil basal Fs in opposite directions: the reduced basal Fs lowered the annual Fs, but the increase in soil temperature and SWC increased Fs. Because at our treated sites Fs was decreased by treatment (Table 1 and S1 Table), the effect of basal Fs was stronger than the effect of microclimate. Complex effects of disturbances on Fs were also found in other studies, where factors acted independently on autotrophic and heterotrophic processes and compensatory mechanisms caused disturbances effects in part to cancel out [7, 41].

In several studies analyzing the effects of fire and thinning on Fs, researchers also found an increase of soil temperature and soil water content after forest management [17, 4143]. However, effects on Fs varied. Burning increased and thinning reduced Fs in forests similar to those studies here [42]. Previous work in young ponderosa pine forests demonstrated that Fs generally decreased only when thinning and burning were combined [41, 44]. Misson et al. [7] and Concilio et al. [45] concluded that the response to forest management treatments is proportional to their intensity. Disagreement in results could be due in part to the difficulty to detect treatment effects. In a previous study conducted at our research site we determined that disturbance increased spatial variability of Fs [34], mirroring post-disturbance increases in spatial variability of Ts and SWC and reducing the ability to accurately quantify Fs and to detect differences between treatments [34].

Differences in Fs among sites were not a result of large differences in maximum and minimum Fs rates, but to the timing of their fluctuations. For example, we measured a decline in Fs during summer drought, as observed in several other studies [7, 4648], and treatments changed the onset and duration of these periods (Fig 5a). During the low SWC summer periods, the CTRLFs site showed higher Fs than the FIREFs and HARV sites (Fig 5a), even if SWC was lower at this site (Fig 5b and 5c). This could be due to the difficulties in measuring SWC at the appropriate depth, so that the top 5 cm measurement does not represent the layer effectively controlling Fs. Summer SWC integrated over the root depth could be higher at CTRLFs than at the treated sites and should be used in future research. An alternative explanation is the prevailing contribution of the autotrophic component at this site, together with an independent control of Ts and SWC over this component. Tang and others [41] found that the heterotrophic component of Fs was more susceptible to seasonal drought than the autotrophic component in a nearby forest.

Our values of Q10, ranging from 3.8 (± 0.18 95% CI) to 2.2 (± 0.43 95% CI), showed a decreasing trend going from the CTRLFs, to the FIREFs and the HARV sites (Table 1), such that temperature effects were reduced where seasonal temperature variations were higher. Our values matched the values reported by Kobziar et al. [44] in a similar forest located about 100 km south from Blodgett, ranging from 1.9 to 5.38. A lower Q10 at the HARV sites, where autotrophic respiration is lowest, agrees with the lower Q10 for heterotrophic decomposition than for autotrophic respiration reported in Ngao et al. [36]. Our higher Q10 at the CTRLFs with older trees concurs with the higher Q10 for the older stand in a chronosequence of coastal Douglas-fir stands [49]. In our sites Q10 was higher at the CTRLFs, where basal respiration was highest, a finding also reported in Davidson et al [46]. Finally Q10 decreased at the HARV sites, where temperature was highest, showing a dependence on soil temperature also reported in Tjoelker et al. [50].

We measured seasonal peak daily Fs of 6–8 μmol m-2 s-1 and annual (control treatments) values ranging between 1200 and 1300 g C m-2 yr-1. Our values are in the range of fluxes measured in previous studies in similar forested ecosystems [25, 4142, 44]. The relatively high annual Fs (for example compared to the range of values for temperate coniferous forests in Subke et al. [51]) was in part explained by a Fs activity that persisted late in the fall. Soil temperature was > 10°C until November, and warm and wet conditions are favorable to both heterotrophic and autotrophic processes [46]. Our high winter Fs values (2–4 μmol m-2 s-1) are supported by Xu et al. [25] winter Fs values in a ponderosa pine plantation adjacent to Blodgett Forest (3 μmol m-2 s-1).

The diffusion coefficient of CO2 was lower at the HARV and FIREFs sites compared to the CTRLFs site (Fig 2, S1 Table). Differences were not due to modifications in soil physical properties because the relationship between tortuosity and soil water status were not different (Fig 2a). However SWC was higher at these sites compared to the CTRLFs site. Thus soil water content had a prominent role in preventing CO2 to exit from the soil surface [46] and could be the cause of the higher CO2 concentration recorded at the HARV site (Fig 4). Higher CO2 concentration could also be a result of higher CO2 production, due to the higher temperature and SWC favorable to decomposing processes, or more abundant substrate for decomposition (however we did not find this in terms of soil carbon, litter, or fine roots).

Fuel Treatment Impacts to Biomass and Tree Growth

After seven years, prescribed fire reduced tree density by 32% and resulted in a 28% reduction in tree biomass productivity and a 10% reduction in individual tree annual radial growth. These reductions were from a combination of direct mortality from fire and secondary mortality [10, 52]. Secondary mortality had a stronger influence on reducing post-fire stand biomass than increased growth rates of the surviving trees, which was also found in other research [17].

Trees that survived the prescribed fire were not able to make efficient use of the increased pool of inorganic nitrogen after fire consumed a large amount of forest floor [53]. A previous related study assessed the impacts of fire and thinning on tree vigor and vulnerability [54] and found negative impacts for fire and positive for thinning. The authors proposed that the negative fire impacts could be minimized by delaying prescribed fire 2–5 years following thinning/mastication, or removing whole-trees instead of mastication which left all activity fuels on site [54].

Over the seven post-disturbance years observed in this study, thinning had a smaller impact on ecosystem carbon dynamics than treatments that included fire. Post-thinning tree density was relatively stable compared to the FIRE treatment. The reduced competition for resources in the THN treatment had a positive impact on tree radial growth on both individual tree and mean tree growth of the treatment. Accumulated carbon was similar than the CTRL treatment both of which were much larger than the FIRE treatment. Nevertheless, if we limited our inventories to the first post-disturbance year, we would have found the opposite conclusion with a stronger impact of thinning than fire (THN reducing biomass carbon stocks 18% and tree density 41%, compared to FIRE reducing biomass carbon stocks 4% and the number of trees by 14%). Fire induced mortality persisted almost a decade, whereas when fuel and biomass were treated mechanically the impacts were concentrated in the first year post treatment.

When analyzing effects of thinning and fire on forests carbon dynamics it is important to include not only the carbon stored in trees, but also the nature and fate of the carbon removed during management operations. With a prescribed fire, most of the carbon removed is released during combustion directly to the atmosphere [8, 10]. Alternatively wood partially burned and standing dead (snags) can take decades to decompose. When biomass is removed by logging, only carbon stored in residual matter that is burned, or in wood used for energy production, is immediately released to the atmosphere. Carbon used as paper and shipping material is stored for longer periods (1–6 years), and for the longest period when wood is used as solid and wood composite products, particularly if used in home construction [55]. Litter, coarse woody debris, and understory vegetation may be only minimally affected by thinning [3, 11, 53]. However carbon is also emitted for transportation of wood products to final destinations and by logging equipment during thinning. At Blodgett forest, thinning did not significantly affect the litter layer or coarse woody debris, whereas both were reduced by prescribed fire [10, 53]. Wood logged by thinning at this forest was sold (with a positive revenue) mostly as construction lumber [55]. Fifty percent of wood removed had a life span of 50 years, and only circa 15% was released to the atmosphere in the first year [10]. In this forest, compared to prescribed fire, thinning caused smaller carbon losses during treatment, kept carbon stored for a longer period, did not remove carbon stored at the soil surface, and increased tree and stand growth.

Despite the decreased productivity and the differences in effects between FIRE and THN treatments, after seven years both of these treatments reached pre-treatment biomass carbon stocks levels, and still maintained a lower fire hazards because of the reduction or removal of surface and ladder fuels [6, 20, 31]. For the THN+FIRE treatment, where 54% of trees were removed or killed, we did not observe any recovery during the first 7 post treatment years. The THN+FIRE treatment displayed the beneficial effect of reduced competition among trees on radial tree growth (also observed at the THN treatment) for two species (A. concolor and C. decurrens). However, the positive effects on these species were attenuated with the negative effect caused on other species (especially P. menziesii).

In comparison with the partial disturbance due to fire or thinning, the clear-cut harvest had the strongest effect on carbon pools and fluxes. Two years after harvesting, carbon stored above ground and as tree biomass was reduced 80% and stand productivity was close to zero. At the same time soil fluxes were reduced by a similar amount as the fire site (ca. 15%). Soil ripping did not have a positive initial effect on seedling growth but it increased the impacts of harvesting on carbon dynamics, as indicated by the larger decrease, compared to the HARVNO_RIP site, of soil carbon and litter.

The intensity of the treatments impacted ecosystem carbon dynamics. A large decrease in tree density (55% reduction) reduced stand biomass carbon stocks and productivity for at least seven years in the thinning followed by fire treatment. When tree density was decreased to a smaller extent (40% reduction) by thinning alone, tree growth was higher than at the untreated stands, and biomass carbon stocks quickly recovered. The intensity of the treatment was not the only determinant factor, the type of disturbance was also important. Fire and thinning, applied singularly or combined, had different impacts on stand productivity and on the time needed to recover to pre-treatment carbon levels. The prescribed fire removed only a relatively small number of trees, with no change in biomass carbon stocks in the first post-treatment year, but tree injuries caused secondary tree mortality and low growth. Therefore, long-term studies are needed to assess the effects of management treatments on carbon dynamics, because studies limited to the first post-disturbance year will not capture the multiple and complex interactions involved. Finally, these results provide insights for managers pursuing forests carbon sequestration to select treatment types and intensities that minimize carbon losses, speed recovery time, and increases post-disturbance carbon uptake.

Supporting Information

S1 Fig. Seasonal trend of soil CO2 efflux measured during 2011 and 2012.

Soil CO2 fluxes measured at the control (CTRLFs), fire (FIREFs), and tree harvest with (HARVRIP) and without soil ripping (HARVNO_RIP) treatment sites. Fluxes measured with the profile technique (black line) were adjusted to match fluxes measured periodically at 20–29 locations per site using the chamber technique (black circles).


S1 Table. Statistical analysis.

a) Description of the statistical analysis of soil carbon pools and CO2 efflux (Fs) in a mixed conifer forest subject to fire (FIRE), clear cut harvesting with (RIP) and without (NO_RIP) soil ripping, and an undisturbed forest (CTRL) at Blodgett Forest Research Station in the Sierra Nevada. 1, 2 compares soil CO2 efflux measured using the chamber technique (chamber). 1 Test for total effect on Fs, where the general linear model 2 characterizes interactions among Fs, soil temperature, and soil water content. b) Statistical analysis of stand characteristics, biomass, growth, and productivity of a mixed conifer forest subject to fire (FIRE), thinning (THN), thinning plus burning (THN+FIRE) and an undisturbed forest (CTRL) at Blodgett Forest Research Station in the north-central Sierra Nevada.



We appreciate the help from Joe Verfaillie and Blodgett Forest Research Station staff, especially Russell Seufert. Field assistance from Kenton Stutz, Austin Jones, Ben Weise, Karla King, and Zev Balsen was invaluable. We thank Jen York for her field expertise.

Author Contributions

Conceived and designed the experiments: SLS DLF RV. Performed the experiments: SD DLF BMC RAY. Analyzed the data: SD. Contributed reagents/materials/analysis tools: RAY. Wrote the paper: SD DLF BMC RAY RV SLS.


  1. 1. Houghton JT, Jenkins GT, Ephraums JJ, editors. 1990. Climate Change: The IPCC Scientific Assessment. Cambridge University Press, Cambridge.
  2. 2. Pan Y, Birdsey RA, Fang J, Houghton R, Kauppi PE, Kurz WA, et al. A large and persistent carbon sink in the world’s forests. Science 2011; 333:988–93. pmid:21764754
  3. 3. Finkral AJ, Evans AM. The effects of a thinning treatment on carbon stocks in a northern Arizona ponderosa pine forest. Forest Ecology and Management 2008;255:2743–2750.
  4. 4. Hessburg PF, Agee JK, Franklin F. Dry forests and wildland fires of the inland Northwest USA: contrasting the landscape ecology of the pre-settlement and modern eras. Forest Ecology and Management 2005;211:117–39.
  5. 5. Stephens SL, Agee JK, Fulé PZ, North MP, Romme WH, Swetnam TW, et al. Managing forests and fire in changing climates. Science 2013; 342:41–2. pmid:24092714
  6. 6. Stephens SL, Moghaddas JJ, Hartsough BR, Moghaddas EEY, Clinton NE. Fuel treatment effects on stand-level carbon pools, treatment-related emissions, and fire risk in a Sierra Nevada mixed-conifer forest. Can J For Res 2009 08/01; 2014/04;39(8):1538–1547.
  7. 7. Misson L, Tang J, Xu M, McKay M, Goldstein AH. Influences of recovery from clear-cut, climate variability, and thinning on the carbon balance of a young ponderosa pine plantation. Agricultural and Forest Meteorology 2005;130(3–4):207–222.
  8. 8. Campbell JL, Harmon ME, Mitchell SR. Can fuel-reduction treatments really increase forest carbon storage in the western US by reducing future fire emissions? Frontiers in Ecology and the Environment 2011;
  9. 9. Hurteau MD, Brooks ML. Short- and Long-Term Effects of Fire on Carbon in US Dry Temperate Forest Systems. Bioscience 2011 02/01; 2012/02;61(2):139–146.
  10. 10. Stephens SL, Boerner REJ, Moghaddas JJ, Moghaddas EEY, Collins BM, Dow CB, et al. Fuel treatment impacts on estimated wildfire carbon loss from forests in Montana, Oregon, California, and Arizona. Ecosphere 2012 05/01; 2013/04;3(5):art38.
  11. 11. Smith TF, Rizzo DM, North M. Patterns of mortality in an old-growth mixed-conifer forest of the southern Sierra Nevada, California. Forest Science 2005;51:266–75.
  12. 12. Williams CA, Collatz GJ, Masek J, Goward SN. Carbon consequences of forest disturbance and recovery across the conterminous United States. Global Biogeochemical Cycles 2012;26, GB1005,
  13. 13. Restaino JC, Peterson DL. Wildfire and fuel treatment effects on forest carbon dynamics in the western United States. Forest Ecology and Management 2013;303:46–60.
  14. 14. Metlen KM, Fiedler CE, Youngblood A. Understory response to fuel reduction treatments in the Blue Mountains of Northeastern Oregon. Northwest Science 2004;78(3):175–84.
  15. 15. Campbell JC, Alberti G, Martin JG, Law BE. Carbon dynamics of a ponderosa pine plantation following fuel reduction treatment in the northern Sierra Nevada. Forest Ecology and Management 2009; 257:453–463.
  16. 16. Dore S, Montes-Helu M, Hart SC, Hungate BA, Koch GW, Moon JB, et al. Recovery of ponderosa pine ecosystem carbon and water fluxes from thinning and stand replacing fire. Global Change Biology 2012;18:3171–85.
  17. 17. Irvine J, Law BE, Hibbard KA. Post-Fire Carbon Pools and Fluxes in Semi-Arid Ponderosa Pine in Central Oregon. Global Change Biol 2007;13:1–13.
  18. 18. Grady KC, Hart SC. Influences of thinning, prescribed burning, and wildfire on soil processes and properties in southwestern ponderosa pine forests: A retrospective study. Forest Ecology and Management 2006;234(1–3):123–135.
  19. 19. Sullivan BW, Kolb TE, Hart SC, Kaye JP, Dore S, Montes-Helu M. Thinning reduces soil carbon dioxide but not methane flux from southwestern USA ponderosa pine forests. For Ecol Manage 2008 6/30;255(12):4047–4055.
  20. 20. North M, Hurteau M, Innes J. Fire suppression and fuels treatment effects on mixed-conifer carbon stocks and emissions. Ecological Applications 2009;19:1385–1396. pmid:19769088
  21. 21. Amiro BD, Barr AG, Barr JG, Black TA, Bracho R, Brown M, et al. Ecosystem carbon dioxide fluxes after disturbance in forests of North America. J Geophys Res 2010 10/27;115:G00K02.
  22. 22. Harmon ME, Bond-Lamberty B, Tang JW, Vargas R. Heterotrophic respiration in disturbed forests: A review with examples from North America. Journal of Geophysical Research-Biogeosciences 2011;116:G00K04.
  23. 23. Dore S, Kolb TE, Montes-Helu M, Eckert SE, Sullivan BW, Hungate BA, et al. Carbon and water fluxes from ponderosa pine forests disturbed by wildfire and thinning. Ecological Applications 2010(3):663–683. pmid:20437955
  24. 24. Miesel JR. Differential responses of Pinus ponderosa and Abies concolor foliar characteristics and diameter growth to thinning and prescribed fire treatments. Forest Ecology and Management 2012; 284:163–173.
  25. 25. Xu M, Qi Y. Soil-surface CO2 efflux and its spatial and temporal variations in a young ponderosa pine plantation in northern California. Global Change Biol 2001;7(6):667–677.
  26. 26. Olson CM, Helms JA. Forest growth and stand structure at Blodgett Forest Research station. In: Sierra Nevada ecosystem project: final report to Congress, Vol. III. Centers for Water and Wildland Resources, University of California, Davis, 1996; pp 681–732.
  27. 27. Laudenslayer WF, Darr HH. Historical effects of logging on forests of the Cascade and Sierra Nevada ranges of California. 1990 Transactions of the Western Section of the Wildlife Society. 1990; 12–23.
  28. 28. Stephens SL and Collins BM. Fire regimes of mixed conifer forests in the north-central Sierra Nevada at multiple spatial scales. Northwest Science 2004;78:12–23.
  29. 29. Stephens SL, Moghaddas JJ. Experimental fuel treatment impacts on forest structure, potential fire behavior, and predicted tree mortality in a California mixed conifer forest. For Ecol Manage 2005 8/25;215(1–3):21–36.
  30. 30. McIver J, Youngblood A, Stephens S. The National Fire and Fire Surrogate study: ecological consequences of fuel reduction methods in seasonally dry forests. Ecological Applications 2009;19:283–284. pmid:19323190
  31. 31. Davidson EA, Savage K, Verchot LV, Navarro R. Minimizing artifacts and biases in chamber-based measurements of soil respiration. Agricultural and Forest Meteorology 2002;113(1–4):21–37.
  32. 32. Vargas R, Baldocchi DD, Allen MF, Bahn M, Black TA, Collins SL, et al. Looking deeper into the soil: biophysical controls and seasonal lags of soil CO2 production and efflux. Ecological Applications 2010;20:1569–82. pmid:20945760
  33. 33. Savage KE, Davidson EA. A comparison of manual and automated systems for soil CO2 flux measurements: trade-offs between spatial and temporal resolution. Journal of Experimental Botany 2003 March 1;54(384):891–899. pmid:12598560
  34. 34. Dore S, Fry DL, Stephens SL. Spatial heterogeneity of soil CO2 efflux after harvest and prescribed fire in a California mixed conifer forest. For Ecol Manage 2014 5/1;319(0):150–160.
  35. 35. Jassal R, Black A, Novak M, Morgenstern K, Nesic Z, Gaumont-Guay D. Reltionship between soil CO2 concentrations and forest-floor CO2 effluxes. Agricultural and Forest Meteorology 2005;130(3–4):176–192.
  36. 36. Ngao J, Epron D, Delpierre N, Bréda N, Granier A, Longdoz B. Spatial variability of soil CO2 efflux linked to soil parameters and ecosystem characteristics in a temperate beech forest. Agric For Meteorol 2012 3/15;154–155(0):136–146.
  37. 37. Jenkins JC, Chojnacky DC, Heath LS, Birdsey RA. National-Scale Biomass Estimators for United States Tree Species. Forest Science 2003;49(1):12–35.
  38. 38. Cairns MA, Brown S, Helmer EH, Baumgardner GA. Root biomass allocation in the world's upland forests. Oecologia 1997 06/01;111(1):1–11.
  39. 39. Penman J, Gytarsky M, Hiraishi T, Krug T, Kruger D, Pipatti R, et al. 2003. Good Practice Guidance for Land Use, Land-Use Change and Forestry. IPCC National Greenhouse Gas Inventories Programme and Institute for Global Environmental Strategies, Kanagawa, Japan. Available:
  40. 40. Stewart-Oaten A, Bence JR. Temporal and Spatial Variation in Environmental Impact Assessment. Ecological Monographs 2001;71:305–39.
  41. 41. Tang J, Misson L, Gershenson A, Cheng W, Goldstein AH. Continuous measurements of soil respiration with and without roots in a ponderosa pine plantation in the Sierra Nevada Mountains. Agric For Meteorol 2005 10/3;132(3–4):212–227.
  42. 42. Ma S, Chen J, North M, Erickson HE, Bresee M, Le Moine J. Short-Term Effects of Experimental Burning and Thinning on Soil Respiration in an Old-Growth, Mixed-Conifer Forest. Environ Manage 2004 07/01;33(1):148–159.
  43. 43. Peng Y, Thomas SC, Tian D. Forest management and soil respiration: implications for carbon sequestration. Environmental Reviews 2008;16:93–111
  44. 44. Kobziar LN, Stephens SL. The effects of fuels treatments on soil carbon respiration in a Sierra Nevada pine plantation. Agric For Meteorol 2006 12/20;141(2–4):161–178.
  45. 45. Concilio A, Ma S, Ryu S, North M, Chen J. Soil respiration response to experimental disturbances over 3 years. For Ecol Manage 2006 6/15;228(1–3):82–90.
  46. 46. Davidson EA, Janssens IA, Luo Y. On the variability of respiration in terrestrial ecosystems: moving beyond Q10. Global Change Biol 2006;12(2):154–164.
  47. 47. Hirano T, Kim H, Tanaka Y. Long-term half-hourly measurement of soil CO2 concentration and soil respiration in a temperate deciduous forest. Journal of Geophysical Research: Atmospheres 2003;108(D20):4631.
  48. 48. Reichstein M, Tenhunen JD, Roupsard O, Ourcival J, Rambal S, Dore S, et al. Ecosystem respiration in two Mediterranean evergreen Holm Oak forests: drought effects and decomposition dynamics. Funct Ecol 2002;16(1):27–39.
  49. 49. Humphreys ER, Black TA, Morgenstern K, Cai T, Drewitt GB, Nesic Z, et al. Carbon dioxide fluxes in coastal Douglas-fir stands at different stages of development after clearcut harvesting. Agricultural and Forest Meteorology 2006;140(1–4):6–22.
  50. 50. Tjoelker MG, Oleksyn J, Reich PB. Modelling respiration of vegetation: evidence for a general temperature-dependent Q10. Global Change Biol 2001;7(2):223–230.
  51. 51. Subke J, Iinglima I, Cotrufo M. Trends and methodological impacts in soil CO2 efflux partitioning: A metaanalytical review. Global Change Biology 2006;12(6):921–43.
  52. 52. Stark DL, Wood DL, Storer AJ, Stephens SL. Prescribed fire and mechanical thinning effects on bark beetle caused tree mortality in a mid-elevation Sierran mixed-conifer forest. Forest Ecology and Management 2013;306:61–67.
  53. 53. Moghaddas JJ, Craggs L. A fuel treatment reduces fire severity and increases suppression efficiency in a mixed conifer forest. International Journal of Wildland Fire 2007;16:673–8.
  54. 54. Collins BM, Das AJ, Battles JJ, Fry DL, Krasnow KD, Stephens SL. Beyond reducing fire hazard: fuel treatment impacts on overstory tree survival. Ecological Applications 2014;24:1879–1886.
  55. 55. Hartsough BR, Abrams S, Barbour RJ, Drews ES, McIver JD, Moghaddas JJ, et al. The economics of alternative fuel reduction treatments in western United States dry forests: financial and policy implications from the National Fire and Fire Surrogate Study. Forest Economics and Policy 2008;10:344–35.