Impacts of Harvesting and Prescribed Burning on Forest Soil Carbon Dynamics: A Global Meta-Analysis

Abstract

Forest management practices such as clearcutting, thinning, and prescribed burning are widely implemented to achieve various ecological and silvicultural objectives, yet their effects on soil carbon dynamics and belowground processes remain uncertain. We conducted a global meta-analysis of 414 observations from 110 studies to quantify the impacts of these practices on total soil respiration (SR), its autotrophic (Ra) and heterotrophic (Rh) components, and associated biophysical and soil variables. Clearcutting and prescribed burning both reduced SR by an average of 11%, driven largely by Ra declines following reductions in live biomass, forest floor inputs, and microbial biomass. Thinning caused no significant change in SR, likely due to the limited belowground disturbance and residual vegetation compensatory growth, although impacts intensified when combined with post-treatments (e.g., residue removal or site-preparation burns), resembling those of clearcutting or repeated burns. In contrast, post-burn treatments following clearcutting did not substantially alter biological factors or SR components. Across practices, soil temperature increased due to the opening of the canopy, middle- and understory vegetation, and forest floor disturbance, but this warming showed no consistent relationship with Rh or SR. Instead, responses were primarily governed by substrate availability, highlighting its central role in soil carbon fluxes under management disturbances.

1. Introduction

Forests cover approximately 30% of the Earth’s land surface and store nearly 40% of terrestrial carbon [1,2], making them a critical component of the global carbon sink. Globally, ~7% of forests are managed plantations [3], and ~74% comprise secondary forests recovering from past disturbances [4]. Recent analyses have documented an increasing trend in soil heterotrophic respiration [5] and a concurrent decline in global forest carbon sink capacity [6,7], raising concerns about the resilience of forest carbon storage. Forest management practices are often designed to enhance tree growth, health, risk mitigation, or habitat quality [8], but they also represent disturbances that affect soil carbon fluxes and dynamics [9,10]. Quantifying their effects is therefore essential for improving predictions of terrestrial carbon feedback under global change [11].

Forest management activities, such as harvesting and prescribed burning, can alter key structural and environmental drivers of forest carbon cycling, such as light availability, microclimate, and the substrates and nutrients supply [10,12]. In response, trees may shift carbon allocation strategies to balance growth, maintenance, and stress tolerance, for example, by modifying biomass investment between above- and belowground components [13]. Changes in detrital input and rhizosphere substrates can constrain microbial activity, whereas shifts in microclimates, such as soil warming, may stimulate microbial decomposition and root growth [14]. Physical soil disturbance can also expose previously protected organic matter, thus accelerating mineralization [15]. These interacting factors contribute to the high variability in soil carbon responses across forest types, sites, and management regimes.

Recent meta-analyses have assessed the impact of disturbances on soil respiration in global forest ecosystems [16], and under specific management disturbances, such as forest fires [17,18,19], thinning [20,21,22], and harvesting [23]. These studies suggest that soil respiration (SR) tends to increase under light harvesting, driven by fine root biomass growth and increased soil nutrients, while both autotrophic (Ra) and heterotrophic (Rh) respiration generally decline with greater harvest intensity, like clearcutting, because of substantial loss in detritus inputs. In contrast, different studies have arrived at contrasting conclusions about the effect of fires on both Ra and Rh [17,18]. Fire effects can also depend on burn intensity, with prescribed burns typically being low-intensity compared to wildfires. As a result, syntheses focused on prescribed fire often reflect conditions that are not directly comparable to high-intensity fires.

Despite these advances, several information gaps remain. With accruing data, it is increasingly possible to resolve “harvesting” effects to those caused by thinning and clearcutting, and “fire” effects to prescribed burns and wildfires. At the same time, management practices are often applied in sequence or combination, such as clearcutting or thinning, followed by site-preparation burning [24,25,26], herbicide application [27], fertilization [28], or the use of repeated prescribed burning for fuel reduction [29]. However, the limited availability of such datasets has not allowed a thorough assessment of the modifying effects of these secondary treatments, and their long-term consequences for belowground carbon cycling remain poorly resolved [8,30]. Building on previous meta-analyses that primarily examined single management practices, this study will review their combined impacts on soil respiration components and associated soil and biophysical variables. A better quantitative understanding of these interactions will help balance climate mitigation, economic, and soil health considerations in land management decisions [11,31].

In this meta-analysis, we synthesized peer-reviewed studies on three major forest management practices—clearcutting, thinning, and prescribed burning—to quantify their effects on SR components (Ra and Rh) and associated biophysical and soil environmental variables. The objectives of this study were to (1) assess the effects of major forest management practices (clearcutting, thinning, and prescribed burn) on SR, its components, and soil carbon stocks, and (2) explore how soil carbon dynamics respond to variation in management strategies, climate, forest types, and recovery stages. We hypothesized that (i) clearcutting would substantially reduce Ra by increasing root mortality and limiting photosynthate supply from aboveground biomass and would concurrently reduce Rh by limiting both the photosynthate supply and the inputs of aboveground detritus; (ii) prescribed burning would reduce Ra through damage to fine root biomass and concurrently reduce Rh due to the subsequent decrease in root exudates, though the overall reduction would be smaller than with clearcutting; (iii) thinning effect would be minimal on Ra and Rh; and (iv) sequence or combined management effects would amplify the effects on Ra and Rh, resulting in greater cumulative impacts than individual practices alone.

2. Materials and Methods

2.1. Data Collection and Compilation

Peer-reviewed journal articles published before August 2024 were obtained by searching the ISI Web of Science database, using the search terms “(soil CO2 OR soil carbon dioxide OR soil carbon efflux OR soil carbon emission OR root respiration OR autotrophic respiration OR microb* respiration OR belowground respiration OR heterotrophic respiration) AND (harvest* OR thinn* OR log* OR understory OR litter* OR manag* OR clear cut* OR clearcut* OR burn* OR slash* OR fire) AND (boreal OR temperate OR tropical OR mediterranean) AND (forest)”. In addition to these search results, publications from previous meta-analyses on soil respiration [16,17,18] and the Soil Respiration Database (SRDB) Version 5 [32] that met the selection criteria were also included.

The retrieved articles were screened based on the following criteria: (1) field-based studies involving forest management practices; (2) inclusion of at least one soil respiration component (total, autotrophic, or heterotrophic respiration); (3) a study duration of more than one growing season; and (4) inclusion of both control and treatment groups. Regarding criterion (1), we excluded incubation studies, as this method does not fully account for the essential biophysical link between live plants and soil (e.g., photosynthate supply to rhizosphere), which may yield divergent results from field settings [33]. For chronosequence studies, the longest undisturbed stand was treated as the control. In total, 414 observations from 110 articles published between 1987 and 2025 met the selection criteria and were included in the meta-analysis (Figure 1, Table S2). The dataset includes 189 observations for clearcutting, 175 observations for thinning, and 50 observations for prescribed burning.

2.2. Data Extraction

Following the initial screening, relevant data were directly obtained from texts and tables, or extracted figures using PlotDigitizer (https://plotdigitizer.com/; accessed on 14 July 2025). Specifically, the following variables were extracted: (1) geographical and climate information, including latitude (°), longitude (°), climate zone (tropical, subtropical, temperate, Mediterranean, or boreal), mean annual temperature (MAT, °C), and mean annual precipitation (MAP, mm); (2) forest stand characteristics, including forest type (needleleaf, broadleaf, or mixed) and leaf habit (evergreen, deciduous, or mixed); (3) information on management practices, including the type of management (clearcutting, prescribed burn, or thinning), disturbance intensity, the time since application (years), whether harvest residues were retained, whether any additional practices were applied (e.g., site-preparation burn, thinning), and whether the practice was repeated; (4) soil respiration components, including total soil respiration (SR), autotrophic respiration (Ra), and heterotrophic respiration; (5) soil environmental and chemical properties of mineral soils, including soil temperature (Ts, °C), soil moisture (SM), bulk density (BD), pH, soil organic carbon and nitrogen (SOC and SON), total soil carbon and nitrogen (TC and TN; including both organic and inorganic), soil C:N ratio (CN), and (6) biological variables, including fine and coarse root biomass (FRB and CRB), microbial biomass carbon (MBC), stand leaf are index (LAI), forest floor mass (FF), aboveground litterfall (LF), and aboveground biomass (ABG). Dissolved organic carbon (DOC) and nitrogen (DON), and microbial biomass nitrogen (MBN) were also collected, but given the low number of studies that reported them, they could not be used in the meta-analysis.

Disturbance intensity and recovery time were categorized to evaluate the effects of forest management on soil carbon dynamics. Clearcutting intensity included stem-only harvesting, whole-tree harvesting, and whole-tree harvesting with either soil surface removal or soil compaction. Thinning intensity was classified as low (<25% tree or <20% basal area removal), medium (up to 50% trees or 20–35% basal area), and high (>50% trees or >35% basal area). Prescribed burn intensity was based on understory vegetation and damages to overstory trees: low (burned understory vegetation with unburned canopy), medium (partial canopy damage with 20–90% scorched), and high (complete canopy damage with >90% scorched). Recovery stages were grouped as early (<2 years), medium (2–5 years), and late (>5 years). The listed groupings were selected as the most common categories used in the literature.

For studies that reported multiple disturbance levels or site conditions (e.g., variations in tree density or species composition), each level was treated as a separate observation. Similarly, for multi-year studies with distinct annual measurement records, each year was considered as an independent observation. Data were extracted directly from the original publications whenever possible. When specific variables were unavailable, the data were sourced from studies conducted at the same location.

The compiled dataset had the greatest representation of studies from temperate forests (46%), followed by boreal (23%) and Mediterranean forests (17%), especially those located in Western Europe and North America. The geographic distribution of study sites ranged from 37.5° S to 62.3° N (Figure 1), with MAT ranging from −10.0 °C to 27.0 °C and MAP ranging from 146 mm to 3500 mm (Table S1). The time since the management disturbance ranged from 0 to 100 years (Table S1).

2.3. Data Analysis

We quantified the effect of forest management on soil respiration (SR), heterotrophic respiration (Rh), autotrophic respiration (Ra), and related variables using the natural logarithm of the response ratio ($lnRR$) as the effect size. The lnRR was calculated as the natural log of the ratio of the mean in the treatment group ($\overline{x_t}$) to that in the control group ($\overline{x_c}$), following the formula:

$$\mathrm{l}\mathrm{n}RR=\ln \left({\displaystyle \frac{\overline{x_t}}{\overline{x_c}}}\right)=\mathrm{l}\mathrm{n}\overline{x_t}-\mathrm{l}\mathrm{n}\overline{x_c}$$

(1)

Since standard errors or standard deviations were not consistently reported across studies, we used sample sizes to compute the weighting factors, following previous meta-analyses [16,23,34,35]. The weight $W_r$ for each observation was calculated as:

$$W_r={\displaystyle \frac{n_c\times n_t}{n_c+n_t}}$$

(2)

where $n_c$ and $n_t$ are the number of replicates in the control and treatment groups, respectively.

To evaluate the overall effects of management practices on soil respiration components, environmental conditions, and biophysical factors, we used weighted mixed-effects models with study ID included as a random effect. The model can be expressed as:

$$Y_{ij}={\beta }_0+u_j+{\epsilon }_{ij}$$

(3)

where $Y_{ij}$ is the response variable for observation $i$ in study $j$, ${\beta }_0$ is the overall intercept, and $u_j$ is the random effect of study $j$, with weight $W_r$ applied to each observation. Models were fitted using the “lmer” function in the “lme4” package [36]. To minimize overinterpretation, only variables reported in more than three independent studies were included. The grand mean of lnRR ($lnR{R}_{++}$) and its 95% confidence interval (CI) were estimated using a bootstrap approach implemented in the “boot” package [37]. The treatment effect was considered statistically significant if the 95% CI did not overlap with zero. For interpretability, lnRR estimates and CIs were back-transformed to represent the percentage change relative to the control group, calculated as:

$$\mathrm{E}\mathrm{f}\mathrm{f}\mathrm{e}\mathrm{c}\mathrm{t} \mathrm{S}\mathrm{i}\mathrm{z}\mathrm{e}\left(\mathrm{\%}\right)=\left(e^{lnR{R}_{++}}-1\right)\times 100\%$$

(4)

We also examined correlations between responses of soil respiration components and key soil or biological factors, as well as their variation across disturbance severity, season of implementation, forest type, climate zone, and the time since application for each management type. Differences among categorical variables were assessed using one-way analysis of variance (ANOVA), followed by Tukey’s post hoc test when the ANOVA results were significant. Results were considered statistically significant at p < 0.05. To identify influential moderators of soil respiration, we used the “glmulti” package, which generates all possible model combinations and ranks variable importance based on corrected Akaike information criterion (AICc) [38]. Relative importance was calculated as the sum of Akaike weights across all models. Variables with importance values ≥ 0.8 were considered essential. Due to inconsistent data coverages on responses of soil respiration components as well as key soil or biological factors, the moderator analysis was restricted to geographical and climatic variables (i.e., latitude, MAT, MAP, climate zone, and forest type) and management-related variables (i.e., time since treatment, severity level [for clearcutting and thinning], repetition [only prescribed burn], and the presence of post-harvest residues [for clearcutting and thinning], post-harvest management [for clearcutting and thinning]). The numerical values were standardized using z-score normalization, calculated as:

$$z=(x-\mu )/\sigma$$

(5)

where x is the original value, μ is the mean of the variable, and σ is the standard deviation.

Publication bias was assessed using a funnel plot and Egger’s regression for asymmetry [39]. No significant asymmetry was detected, indicating minimal publication bias in the dataset (Figure S1). All the data analyses were performed in R (version 4.3.3) [40] and implemented in RStudio (version 2023.12.1) [41].

3. Results

3.1. Management Effects on Soil Properties and Respiration Components

Forest management practices—clearcutting, thinning, and prescribed burning—altered key biological and environmental factors, which resulted in soil respiration dynamics and soil pools (Figure 2). These responses may also reflect post-disturbance vegetation recoveries, including understory vegetation, shrubs, and root sprouts, as well as post-treatment practices (e.g., residue removal, herbicide application, bedding). Among the 130 clearcutting observations reporting post-disturbance vegetation recovery, 47% documented natural regeneration of understory and woody vegetation, whereas 40% involved plantation establishment. These indicate that ecosystem responses (i.e., soil environmental conditions, soil respiratory factors) to management practices are shaped not only by the immediate effects of disturbance but also by the subsequent recovery of both overstory and understory vegetation.

Clearcutting led to the most pronounced shifts in forest structure and soil microclimate. Aboveground biomass (AGB) declined by 95% (95% CI: −99 to −77%), accompanied by reductions in leaf area index (LAI; −55%, 95% CI: −77 to −12%), coarse root biomass (CRB; −90%, 95% CI: −96 to −69%), and fine root biomass (FRB; −32%, 95% CI: −53 to −2.9%). These structural changes were coupled with significant increases in soil temperature (Ts; +13%, 95% CI: +7.0 to +19%) and soil moisture (SM; +13%, 95% CI: +2.1 to +28%). Thinning induced similar but largely non-significant trends, including ABG (−26%, 95% CI: −53 to +21%), and FRB (−9.5%, 95% CI: −26 to +11%), and a significant reduction in LAI (−59%, 95% CI: −77 to −23%). Associated changes in Ts (+4.5%, 95% CI: −8.2 to +18%) and SM (+2.9%, 95% CI: −3.6 to +10%) were not statistically significant. Prescribed burning caused comparatively modest shifts, with LAI reduced by 18% (95% CI: −30 to −5.5%) and a non-significant increase in AGB (+5.2%; 95% CI: −28 to +29%). Ts increased significantly (+9.2%, 95% CI: +4.7 to +13%), while SM increased more often than it decreased (+1.1%, 95% CI: −7.8 to +11%).

Aboveground litterfall (LF) significantly declined under clearcutting (−79%; 95% CI: −92 to −37%) and thinning (−31%; 95% CI: −46 to −12%), but exhibited a non-significant tendency toward increase under prescribed burning (+7.5%, 95% CI: −20 to +49%). Microbial biomass carbon (MBC) declined across treatments, with the greatest reduction observed under clearcutting (−18%, 95% CI: −26 to −10%), and to a smaller, non-significant decrease under thinning (−1.7%, 95% CI: −15 to +14%). For prescribed burning treatments, the number of available MBC observations was insufficient to support a quantitative analysis.

Soil organic carbon (SOC) and soil organic nitrogen (SON) exhibited no significant changes under any management treatment. SOC showed a slight decline under clearcutting (−6.5%; 95% CI: −23 to +14%) and thinning (−1.7%; 95% CI: −11 to +8.9%), but an increasing trend under prescribed burning (+11%; 95% CI: −6.6 to +28%). SON had increasing trends under clearcutting (+9.3%; 95% CI: −25 to +57%) and thinning (+0.7%; 95% CI: −14 to +17%), whereas data for prescribed burning were limited (n = 1).

Clearcutting, however, resulted in significant increases in both total carbon (TC; +31%, 95% CI: +13% to +51%) and total nitrogen (TN; +34%, 95% CI: +2.1% to +71%). Thinning induced a smaller, but significant increase in TC (+4.8%, 95% CI: +1.1% to +8.5%), without a significant change in TN (−9.1%, 95% CI: −29 to +19%). The soil C:N ratio (CN) significantly decreased by 5.2% following thinning (95% CI: −10 to −0.4%). In contrast, prescribed burning did not significantly alter TC (+1.9%; 95% CI: −18 to +21%) or TN (+6.8%; 95% CI: −1.3 to +15%).

These shifts in biological and soil microclimate data correlated with changes in soil respiration dynamics. Total soil respiration (SR) significantly decreased following clearcutting (−11%; 95% CI: −19 to −2.2%) and prescribed burning (−11%; 95% CI: −17 to −2.8%), while thinning had little effect (+1.2%, 95% CI: −5.0 to +7.7%). Though not statistically significant, decreases in both autotrophic (Ra) and heterotrophic (Rh) respiration appeared to contribute to the overall SR decline under clearcutting (Ra: −72%; 95% CI: −94 to +27% and Rh: −7.1%; 95% CI: −20 to +6.5%) and burning (Ra: −25%; 95% CI: −53 to +21% and Rh: −19%; 95% CI: −41 to +12%). In contrast, Ra and Rh increased slightly under thinning (Ra: +5.4%; 95% CI: −11 to + 23% and Rh: +4.5%; 95% CI: −4.5 to + 15%).

3.2. Responses of Soil Pools and Carbon Fluxes to Management Strategies, Climate, and Recovery Stage

3.2.1. Clearcutting

The decrease in soil respiration (SR) was pronounced immediately following clearcutting (−19%), but gradually recovered and shifted to a slight increase (+6.0%) at the late stage (>5 years). Among climate zones, a significant increase was only detected in Mediterranean forests (+18%), although this may be due to a selection bias (Figure 3a). A stepwise decline in SR was observed with increasing harvest intensity; however, this was not statistically significant due to limited sample size, particularly under more intensive treatments.

Autotrophic respiration (Ra) declined consistently across all treatment categories (Figure 3c). While Ra tended to recover over time, the changes were not statistically significant, again likely constrained by limited sample size. Heterotrophic respiration (Rh) also showed a general decreasing trend, though the magnitude of change was less pronounced than for Ra (Figure 3b). No consistent recovery trajectory was detected for Rh across stages. Although residue removal and fuel treatments (e.g., site-preparation burning [30 observations] or thinning [1 observation]) following clearcutting were generally associated with reductions in SR and Rh, these effects were not statistically significant.

Soil organic carbon (SOC) declined more substantially in boreal forests (−30%), whereas it increased by 14% in temperate forests (Figure 3d). Significant SOC decreases were also observed following harvest residue removal (−14%) and during the early recovery stage (−15%; <2 years). In contrast, total carbon (TC) generally increased across climate zones, forest types, and recovery stages, with significant gains detected in boreal forests (+63%). This increase in TC was also detected during both the early (+42%) and late (+37%) recovery stages (Figure 3e). Total nitrogen (TN) exhibited a similar pattern to TC, showing overall increases across the examined categories (Figure 3f).

FRB significantly increased in boreal forests (+30%), but decreased in temperate forests (−38%), although this latter pattern may be an artifact of the limited number of studies, particularly from boreal regions (n = 2). Across recovery stages, mean FRB was slightly positive in the late stage (+4.5%), but did not differ significantly from the early stage (−37%; <2 years). Litterfall (LF) exhibited a sharp decline during the early recovery stage (−94%), which partially recovered in the late stage (−49%; >5 years) (Figure 3i). Responses of the forest floor were variable, but overall, fuel reduction management did not produce significant changes (Figure 3h).

Among environmental variables, soil temperature (Ts) increased consistently across all treatment categories (Figure 3j). However, retention of harvest residues, implementation of less intensive harvesting (e.g., stem-only removal), and the absence of post-harvest interventions mitigated the extent of temperature increases. Furthermore, temperature effects began to diminish during the medium recovery stage (2 to 5 years). Soil moisture (SM) exhibited a general increasing trend, although the magnitude of change was smaller compared to Ts, and no statistically significant differences were detected among groups (Figure 3k).

3.2.2. Thinning

Thinning resulted in moderate changes in SR and its components. SR significantly increased in subtropical forests (+37%), while changes in other regions were not significant (Figure 4a). When thinning was followed by a prescribed burn, SR decreased more sharply (−18%) compared to thinning alone (+3.8%). No significant differences in SR were observed across forest types, thinning severity, residue treatments, or recovery stages. Rh exhibited a general increasing trend across thinning severity levels and recovery stages, though these changes were not statistically significant (Figure 4b). Ra significantly increased under moderate severity thinning (+14%), which was greater than the change observed under low severity thinning (−3.9%) (Figure 4c).

Soil organic carbon (SOC) did not change significantly across any categories, but was significantly higher under moderate thinning severity (+15%) compared with high severity (−10%) (Figure 4d). Although a stepwise decline in SOC was observed with increasing time since thinning, this trend was not statistically supported. Total nitrogen (TN) significantly declined in subtropical forests (−9.8%) (Figure 4e).

Fine root biomass (FRB) varied significantly across climate zones and thinning intensities, with higher values in temperate forests (−1.1%) compared to Mediterranean forests (−31%), and under high-intensity thinning (+1.6%) compared to low-intensity thinning (−27%) (Figure 4f). FRB declined more sharply when thinning was followed by site-preparation burning (−40%) than when no post-treatment was applied (−3.7%). The mean FRB during the early recovery stage was negative (−24%), but shifted to positive values at the medium (2–5 years; +5.2%) and late (>5 years; +6.5%) recovery stage, although these differences were not statistically significant. Forest floor mass (FF) generally declined, with the pronounced reductions in the early recovery stage (< 2 years; −45%) (Figure 4g). Similarly to FRB, although no significant differences were detected across recovery stages, mean FF during the medium stage showed a slight positive response (+1.6%). Site preparation burning following thinning effectively reduced FF by 64%, compared to no post-treatment (−4.3%). Litterfall (LF) generally declined following thinning, with a significant reduction under high-intensity thinning (−35%), but no significant difference compared with moderate thinning (−33%). In contrast to FRB and FF, LF did not show any signs of recovery, even in the late recovery stage (−32%; >5 years) (Figure 4h).

Soil warming following a thinning declined over time, with the early (+11%), the medium (+5.1%), and the late stage (−7.3%) (Figure 4i). Ts rose more in subtropical forests (+30%) than in temperate forests (−2.8%). SM showed an overall increasing trend across most categories, although the changes were not statistically significant across climate zones, forest types, thinning severity, or recovery stages (Figure 4j). Notably, residue retention after thinning mitigated the SM increase (+6.2%) compared to residue removal (−5.8%).

3.2.3. Prescribed Burning

Among the 29 observations that reported burn intensity, 27 were classified as low-intensity, commonly used for fuel reduction. SR declined significantly under repeated burning (−17%), a reduction that was substantially greater than that observed under single burns (−6.4%) (Figure 5a). No significant differences in SR were detected across climate zones, forest types, or recovery stages. Only 10% of prescribed-burn observations reported Rh and Ra, which limited further categorical analysis.

TN significantly increased by 12% in needleleaf forests, although the wide confidence intervals and limited sample sizes constrained the ability to detect a significant difference from broadleaf forests (Figure 5c). In contrast, TC did not show any significant changes or differences in category (Figure 5b). Forest floor mass (FF) significantly decreased, and the reduction in FF was greater in single burn (−87%) than repeated burns (−31%; Figure 5d). Litterfall (LF) showed insignificant increase in both single and repeated burns with no significant difference (Figure 5e).

Ts consistently increased across forest types and fire frequencies, but no significant differences were detected within categories (Figure 5f). In contrast, SM responses were more variable. SM significantly increased under repeated burns (+13%), in contrast to a slight decrease observed after single burns (−5.6%) (Figure 5g).

3.3. Relationship Between Soil Respiration Components and Biophysical Variables

Under clearcutting, SR declined with decreases in both Ts (R² = 0.08, p = 0.004; Figure 6a) and soil moisture (R² = 0.13, p < 0.001; Figure 6b). In contrast, change in microbial biomass carbon (MBC) was positively associated with the change in SR, with a relatively strong correlation (R² = 0.65, p = 0.016; Figure 6c). In thinning treatments, changes in soil respiration and its components were positively correlated with biological variables. SR increased with increasing forest floor mass (R² = 0.22, p = 0.005; Figure 6d), and Rh response was positively correlated with that in MBC (R² = 0.42, p = 0.041; Figure 6e) but negatively with that in litterfall (LF) (R² = 0.53, p = 0.011; Figure 6e). Change in Ra was positively related to that in fine root biomass (FRB) (R² = 0.79, p < 0.001; Figure 6g). No other variables under clearcutting or thinning, and none under prescribed burns, showed significant correlations with soil respiration components.

The key moderators of harvest and prescribed burning effects on SR were determined with model selection based on the sum of Akaike weights (Figure 7). For clearcutting, the retention of logging residue and the intensity of the disturbance were the most important and significant moderators of SR change (Figure 7a). For thinning, important moderators included thinning intensity, climate zone, post-treatment (i.e., prescribed burning), and mean annual temperature (MAT) (Figure 7b). For prescribed burning, MAT was the primary moderator of SR response (Figure 7c).

4. Discussion

4.1. Effects of Forest Management on Above- and Belowground Biomass and Soil Respiration Components

Clearcutting significantly reduced both fine and coarse root biomass (Figure 2a), whereas thinning and prescribed burning resulted in non-significant reductions (Figure 2b,c). Correspondingly, both Ra and Rh declined under clearcutting and prescribed burning, with the reduction being more pronounced in Ra than in Rh (Figure 2b,c). The stronger decline in Ra can be attributed to losses of live plant biomass, which directly suppresses substrate supply and root activity, and thus Ra. The reduction in Ra was clearly stronger in clearcutting (−72%), as the damage to both aboveground biomass and root systems was greater. Prescribed burning also reduced Ra (−25%), likely due to the mortality of fine roots in the surface soils, consistent with earlier reviews [17,18]. Such effects can arise even under low-intensity burns, especially when fires are applied repeatedly [42,43,44] or occur under unfavorable seasonal conditions [45,46]. Ra showed a recovery trend following a clearcut over time (Figure 3c), likely with the regrowth of understory vegetation and the regeneration of overstory species [47,48]. In contrast, the change in Rh was small, likely reflecting the offsetting effects of reduced root exudates but increased detritus both above- and belowground [49,50,51]. We should note that while the predominant effect of declining Rh reported here and in Akande et al. [16], the meta-analysis by Yang et al. [23] reported an overall increase. It is likely that their conclusion resulted from combining clearcutting and thinning into a general “harvesting” category. Furthermore, their finding of a negative relationship between Rh response ratio and harvest intensity is consistent with the current observation of the greatest Rh decline following clearcutting.

Unlike clearcutting and burning, thinning tended to maintain or even enhance SR components (Figure 2b). Non-significant increases in SR, Rh, and Ra were observed under thinning, consistent with previous meta-analyses reporting either significant or non-significant increases in SR or Ra following thinning [20,21,52]. Importantly, Ra increased significantly by 27% under moderate thinning intensity (Figure 4c), consistent with Zhao et al. [47], whereas SR and Rh did not respond significantly to thinning of any intensity (Figure 4a,b). The increase in Ra likely reflects enhanced fine root growth under moderate thinning (Figure 4f), supported by positive correlations between Ra (and SR) responses and fine root biomass (Figure 6g and Figure S2). Improved light availability and reduced competition for water and nutrients under moderate to high thinning intensity stimulate the growth of residual trees [22,53], promoting their root [54,55] and mycorrhizal development [56] and contributing to elevated Ra. Furthermore, earlier studies suggest that moderate thinning can also enhance understory vegetation growth, further increasing root biomass and Ra [55].

Harvesting, both thinning and clearcutting, reduced tree density and consequently decreased litterfall inputs, and this reduction persisted into the late recovery stage (>5 years; Figure 3i and Figure 4h), consistent with long-term declines in litter production observed under thinning [20]. Under clearcutting, MBC also declined significantly (Figure 2a), consistent with previous syntheses that reported a reduction in microbial biomass by 19% following harvesting [57]. Nevertheless, Rh did not exhibit a significant decline under either harvesting treatment; rather, under thinning, it even showed a slight, though non-significant, increase (Figure 2b and Figure 6b). As with clearcutting, the opposing effects of increased detritus production and reduced exudates may balance one another out, or one may temporarily dominate. For instance, Wang et al. [55] reported that thinning stimulated both fine root production and turnover in overstory and understory vegetation of a Chinese fir plantation, with the effect being more pronounced under intense thinning. López et al. [54] also reported that fine root mortality increased by 32% two years after a heavy thinning in the Mediterranean oak forest. In our analysis, signs of fine root biomass recovery were evident in both thinning and clearcutting treatments (Figure 3g and Figure 4f), likely contributing to the stability of Rh. Enhanced aboveground productivity following thinning [22] may further increase belowground carbon inputs; for example, Zhao et al. [58] observed increased root exudates during the growing season following a heavy thinning in a Chinese fir plantation. In addition to altering substrate inputs, thinning may also affect microbial community composition and functionality [52], thus affecting Rh. However, data on these effects are sparse in our dataset and were therefore not included in the current meta-analysis.

4.2. Effects of Forest Management on Soil Environment and Properties

Harvesting and prescribed burns reduced the forest floor and increased light availability by opening the canopy and reducing mid- and understory vegetation. These changes elevated soil temperature, particularly during the early recovery stage. Contrary to our hypothesis, however, higher temperature in response to all of these treatments, as well as increased soil moisture under clearcutting, did not enhance Rh or SR. The negative correlation of SR with Ts and SM under clearcutting (Figure 6a,b) aligns with earlier studies [16,59] and may be attributed to the restoration of carbohydrate supply and fine root biomass with understory regrowth and overstory regeneration over time that override correlations with temperature [48]. Furthermore, the loss of obligate symbiont microbes after clearcutting may also restrict soil C processing [51,60].

Our meta-analysis showed that overall effects on mineral soils were limited. The main exception was clearcutting, where TC and TN significantly increased, a pattern that persisted even during late recovery (>5 years), likely due to delayed mortality of belowground detritus [61,62]. In contrast, SOC declined under certain conditions, particularly when residue removal or site-preparation burning was applied, as these practices can extend depletion into deeper soil layers [63,64]. The magnitude and direction of these responses may also be shaped by climate and vegetation type [65,66]. Although the residue retention was similar in temperate (40%) and boreal forests (49%), SOC tended to increase more often in the temperate forests and decline in boreal forests. This pattern aligns with Mayer et al. [67], who reported the strongest SOC losses in cold-climate forests with large pre-harvest SOM pools. Nevertheless, SOC recovery is expected during early to mid-succession in boreal forests, supported by plant and mycorrhizal hyphal production, and the mycorrhizal suppression of saprotrophs (i.e., Gadgil effect) [68].

By contrast, prescribed burning led to slight, non-significant increases in SOC, TC, and TN, a pattern consistent with earlier reviews [30,69]. These modest changes may result from the incorporation of unburned or partially charred slash into the mineral soil, or from incomplete combustion of organic matter under low fire temperatures [70]. On the other hand, more severe fires can cause greater combustion of the forest floor and mineral soils, often resulting in lower soil C and N pools [71].

4.3. Effects of Repeated and Post-Treatment Burns on Soil Dynamics and Respiration

Repeated low-intensity prescribed burns are widely applied to manage forest structure, reduce wildfire risk, and promote ecosystem resilience [29]. In our dataset, 38% of observations involved repeated burns. These burns reduced SR by 17%, a decline greater than that observed for single burns (Figure 5a). Although low-intensity fires are often assumed to have negligible effects on soil processes [30], our results show that repeated applications generate cumulative impacts on both forest structure and belowground dynamics, thereby reducing SR. This meta-analysis detected that repeated burns, compared to single burns, led to smaller reductions in forest floor mass (Figure 5d), likely due to progressive fuel depletion, but greater increases in soil moisture (Figure 5g), potentially from reduced fine root density or reduced transpiration and increased throughfall following long-term canopy loss [70]. Yet, despite these shifts, SR declined more strongly under repeated burns, suggesting that limitations in substrate availability, and root and microbial activity were dominant controls. Interpretation of the full magnitude and temporal trajectory of this decline, however, remains constrained by the limited, short-term nature of many prescribed burning observations. Consistent evidence from prior work points to reductions in root biomass [44,72], losses of surface soil C and N [73,74], lowered litter quality [75], and altered microbial and mycorrhizal communities [42,43,76]. In a different study, we hypothesized that long-term frequent burning may have reduced fungal colonization of roots [44], amplifying declines in the autotrophic component of SR. Repeated fires may introduce pyrogenic C inputs [19,45,77], which can increase soil recalcitrance to decomposition [78,79], but its production under low-intensity fires is variable and less consistent than that from wildfires [30]. Collectively, these cumulative effects of organic matter loss, reduced root inputs, and altered microbial–plant interactions explain the stronger SR declines observed under repeated burns than single burns.

Harvesting combined with site-preparation burning is also widely implemented to suppress competing vegetation, reduce wildfire risk, and facilitate plantation establishment (e.g., [24,25]). Following clearcutting, logging residues can buffer soils against compaction and erosion, enhance moisture and nutrient retention, and provide organic inputs that support plant regrowth [80,81]. However, when residues are removed or combusted through site-preparation burning, long-term negative impacts may arise, including mineral soil carbon losses [82,83], shifts in understory plant compositions [84], and, in some cases, elevated mortality of remaining trees [8,85]. The effects of post-harvest burning were more pronounced under thinning than under clearcutting. After thinning, reductions in forest floor, fine root biomass, and SR were significantly greater with post-burn than without (Figure 4a,f,g), whereas post-burn following clearcutting did not substantially alter biological factors or SR components, although Ts increased (Figure 3j). Compared with clearcutting, thinning leaves more standing trees and residual root systems, which continue to supply carbon to the soil. Site-preparation burn, therefore, may disproportionately disrupt root and mycorrhizal activity and microbial processes by damaging residual fine roots, forest floor, and understory vegetation, cumulatively leading to reductions in SR, analogous to the effects of repeated prescribed burns. Additionally, among selected thinning studies, only 5 of the 67 observations (7%) explicitly reported that the forest floor was left intact; thus, it is possible that a site-preparation burning may directly affect soil surface roots and microbes. By contrast, clearcutting removes the majority of overstory trees, so subsequent fuel reduction fires did not significantly affect forest floor mass (Figure 3h). While SR and Rh showed trends of decrease when residues were removed or post-burn applied, these differences were not statistically significant (Figure 3a,b). As discussed earlier, recovery of SR and Ra appears more dependent on regrowth of understory vegetation and overstory tree regeneration, such that additional burning does not markedly intensify reductions in SR or Rh.

5. Conclusions

This study evaluated the effects of three major forest management practices—clearcutting, thinning, and prescribed burning—on soil respiration (SR), its component fluxes (Ra and Rh), and associated biophysical and soil environmental variables. Our results showed that clearcutting and repeated prescribed burning significantly reduced SR, particularly Ra, although for potentially different reasons. While after clearcutting the decline is clearly attributable to the removal of live crowns and thus the supply of carbon, prescribed burning may, on the one hand, combust detritus and, on the other hand, kill roots and microbes in the surface soil. While increased sink strength following prescribed burning may also be possible, the data evaluated here did not indicate this. In contrast, thinning caused non-significant changes in SR components, likely because the damage to the belowground compartment was minor and compensatory growth of remaining trees and understory vegetation offset long-term reductions in aboveground litterfall. However, this non-significant pattern may shift toward negative effects when thinning is combined with post-treatments, such as site-preparation burn or residue removal, which can further impair root systems and microbial communities, resembling impacts of clearcutting or repeated fire. Crucially, despite observed increases in soil temperature associated with canopy opening and/or reduced understory cover, SR and its components were not consistently correlated with warming, reinforcing the role of substrate availability rather than temperature as the dominant driver of soil carbon flux responses to management. We recommend that future research prioritize long-term field experiments, particularly under prescribed burn treatments, to better understand long-term soil carbon stabilization. Overall, the meta-analysis demonstrates that forest management decisions regulate not only aboveground biomass and stand structure but also fundamental belowground carbon processes, with important implications for soil health and climate mitigation.