Dynamic Responses of Multidiameter-class Fine Roots at Different Soil Depths to Thinning Measures in a Secondary Forest in China


 Background: Fine roots make critical contributions to carbon stocks and terrestrial productivity, and multidiameter-class fine roots exhibit functional heterogeneity. However, the dynamic characteristics of multidiameter-class fine roots at different soil depths following thinning disturbances are poorly understood. We investigated the biomass, production, mortality and turnover rate of < 0.5 mm, 0.5–1 mm and 1–2 mm fine roots at 0-20 cm, 20-40 cm and 40-60 cm soil depths under five thinning intensities (0%, 15%, 30%, 45%, and 60%) in a secondary forest in the Qinling Mountains. Results: The biomass, production and turnover rate of < 0.5 mm fine roots fluctuated with increasing thinning intensity, while 0.5-1 mm and 1-2 mm fine root biomass significantly decreased. Thinning measures had no effects on fine root necromass (except for T4) or mortality. The fine root dynamic characteristics in deeper soils were more sensitive to thinning measures. Principal component analysis results show that increased < 0.5 mm fine root biomass and production resulted from increased shrub and herb diversity and biomass and decreased soil nutrient availability, stand volume and litter biomass, whereas 0.5-1 mm and 1-2 mm fine root biomass showed the opposite trends and change mechanisms. Conclusions: Our results provide evidence of the positive effect of thinning on very fine root (< 0.5 mm) biomass and production and the negative effect on thicker fine roots (0.5-1, 1-2 mm) or all fine root (< 2 mm) biomass. From the perspective of fine root biomass and productivity, T2 (30%) is recommended for use in secondary forests of the Qinling Mountains. Moreover, our results suggest that thinning practices have varied effects on the dynamic characteristics of multidiameter-class fine roots.

Characteristics of stand and soil (mean ± SE; n = 4).   (Figure 2d, h, l and Table 3). The very ne root (< 0.5 mm) biomass and total mass uctuated with increasing thinning intensity. The very ne root biomass (F = 10.91, p < 0.01) and total mass (F = 10.62, p < 0.01) in T1 and T3 were signi cantly lower than those in CK, while those in T2 and T4 exceeded those in CK (although not signi cant) (Figure 2d, l and Table 3). The 0.5-1 mm (F = 4.42, p < 0.05) and 1-2 mm (F = 9.01, p < 0.01) ne root biomass and 1-2 mm (F = 4.15, p < 0.05) total mass decreased following thinning intensity, reaching signi cance at T3 and T4 (Figure 2d, l and Table 3). The 0.5-1 mm and 1-2 mm ne root necromass mirrored the overall ne root system necromass dynamics, whereas very ne root necromass was rarely observed (Figure 2h). Table 3 The effects of thinning intensity (T) on mean annual biomass, necromass, and total biomass of ne roots over the entire soil pro le and in speci c depth (D Fine root biomass, necromass, and total mass decreased with soil depth, and the values in the topsoil (0-20 cm) accounted for more than 70% of the total values for all soil depths ( Figure 2). Biomass and the total mass of very ne roots were mainly distributed in topsoil and accounted for approximately 50% of the values in the topsoil and 40% of the total values. The biomass and total mass of 0.5-1 mm and 1-2 mm roots dominated the deep soil depths, while the necromass of 0.5-1 mm and 1-2 mm roots was dominant at all soil depths ( Figure 2). The very ne root biomass and total mass were more sensitive to thinning measures than those of other size classes across all soil depths (Table 3 and Figure 2). The ne root biomass, necromass, and total mass at middle and deep soil depths exhibited greater percentage changes than those in topsoil following thinning, and these dynamics in the deep soil increased again in T3 and T2 ( Figure 2 and Table 4). Note: NA: no data was detected in the control group.
The ne root biomass, necromass, and total mass exhibited strong seasonal variations within the sampling year ( Figure 3, Figure  S1-2 and Table S1). The ne root biomass and total mass peaked in spring as determined by very ne root biomass levels ( Figure 3 and Figure S2). The ne root necromass peaked in the autumn, as did the 0.5-1 mm and 1-2 mm ne root necromass ( Figure S1).
Fine root production, mortality and turnover rate in different diameter classes The whole ne root system (∅ ≤ 2 mm) production and mortality of the entire soil pro le (0-60 cm) and the averaged turnover rate did not differ among different thinning intensities (all p > 0.05) (Figures 4d, h, l and Table 5). Very ne root production also uctuated following thinning and signi cantly exceeded that of the CK in T2 and T4 (F = 5.7, p < 0.01) (Figure 4d and Table 5). The very ne root turnover rate in T3 was signi cantly lower than that at the other thinning intensities (F = 2.71, p < 0.05) (Figure 4l and Table 5). Thinning practices did not affect the production and turnover rate of 0.5-1 mm and 1-2 mm ne roots (all p > 0.05) (Figure 4d, l and Table 5). The 0.5-1 mm ne root mortality value in T1 was higher than that under the other treatments (F = 5.27, p < 0.01).
The 0.5-1 mm and 1-2 mm ne root mortality levels mirrored the total mortality levels ( Figure 4h and Table 5). Table 5 The effects of thinning intensity (T) on annual production, mortality, and turnover rate of ne roots over the entire soil pro le and in speci c depth (D The production and mortality of ne roots decreased with soil depth, and the values in topsoil accounted for approximately 69% of total production and 71% of the total mortality ( Figure 4). The production of very ne roots also occupied the topsoil and accounted for approximately 51% in the topsoil and 35% of the total value. Fine root mortality of 1-2 mm was greater at all soil depths ( Figure  4e, f, g). The ne root turnover rate generally increased with deepening soil depths (Figure 4i, j, k). The very ne root production and turnover rate were more sensitive to thinning measures than those of other size classes among the soil depths (Table 5 and Figure   4). The percentage changes in ne root production, mortality and turnover rates in deep soil layers were also generally higher than those in topsoil following thinning, and these dynamics in the deep soil were enhanced again in T2 (Figures 4 and Table 4).

The linkages between ne root dynamics and stand and soil attributes
The PCA of the entire soil pro le according to standardized data shows that the rst two trait axes accounted for 32.3% and 14.3% of the total variation, respectively ( Figure 5). We found that the 0.5-1 mm and 1-2 mm ne root biomasses were highly positively related to soil properties (SOC, AN, AP and AK) and stand characteristics (volume and litter biomass) but negatively correlated with the diversity index and biomass of shrubs and herbs. Conversely, the biomass, production, and turnover rate of very ne roots and the necromass of 0.5-1 mm and 1-2 mm ne roots were negatively correlated with soil nutrient availability, stand volume and litter biomass, whereas they exhibited strong positive correlations with the diversity index and the biomass of shrubs and herbs. The PCA of individual soil depths shows that the association between the understory vegetation characteristics and the very ne root portion was stronger in the topsoil, whereas this correlation disappeared with the thicker ne root portion appearing at deeper soil depths ( Figure S3).

Discussion
Effects of thinning on ne root biomass, production and turnover rate Consistent with our hypothesis, we found that very ne root biomass, production and turnover rates were more sensitive to thinning practices than thicker ne roots. The results align with a recent observation that forest cutting has more signi cant effects on ne roots than on thicker roots [23]. These results may be attributed to the fact that ne roots of different diameter classes exhibit heterogeneous physiological functions and structural compositions, leading to discrepant responses following thinning practices [26]. Moreover, a similar study found that very ne roots with a higher ratio of nonstructural to structural mass are more sensitive to changes in abiotic or biotic factors caused by thinning practices [29].
The very ne root biomass, production and turnover rate levels uctuated (positive or negative effect) with increasing thinning intensity, which is inconsistent with positive or negative results obtained following thinning in previous studies [12,33]. Regarding positive effects, our PCA results suggest that the increase in the biomass and productivity of very ne roots resulted from increases in herb and shrub layer species diversity and biomass (especially in topsoil) ( Figures 5 and S3 a), which compensated for decreased ne root biomass and production resulting from cutting canopy trees, consistent with early studies [19,32]. In addition, previous research has demonstrated that low-nutrient conditions can stimulate the growth of ne root biomass and productivity [54]. In the present study, our whole-tree harvesting measures (reduced stand volume and liter biomass) increased the export of nutrients and reduced soil nutrient availability (Table 1), consistent with an early study [55]. Furthermore, the PCA results show that very ne root biomass and production were negatively correlated with soil nutrient availability, stand volume and liter biomass, supporting this view. Regarding negative effects, a possible explanation is that thinning practices reduce resource competition pressure, and low ne root biomass, production and turnover rate values may satisfy vegetation resource absorption and utilization requirements [56].
In contrast, we found that thinning treatments signi cantly reduced the biomass of 0.5-1 mm and 1-2 mm ne roots, supporting the negative effect of thinning found in previous studies [57]. On the one hand, our PCA results indicate that thinning practices reduced stand volume and litter biomass, which may decrease photosynthate partitioning to the root system and ultimately lead to the reduced biomass and productivity of 0.5-1 mm and 1-2 mm ne roots [23]. Early studies reported that the nutrient acquisition strategy for thicker ne roots may be achieved by increasing their lifespan and extending the period of nutrient absorption, which is strongly dependent on nutrient availability [25,58,59]. Our PCA results show that 0.5-1 mm and 1-2 mm ne root biomass was positively correlated with soil nutrient availability and negatively correlated with herb and shrub species diversity and biomass. Therefore, decreased stand volume, litter biomass and nutrient availability following thinning measures together reduced 0.5-1 mm and 1-2 mm ne root biomass. Overall, our ndings provide a reasonable explanation for the inconsistent impacts of thinning practices on ne root biomass and production observed in previous studies [12,33]. The positive effects of thinning on ne roots are due to signi cantly stimulated growth in very ne root biomass and productivity. Regarding negative effects, on the one hand, thinning reduces all diameter classes of ne root biomass. On the other hand, thinning signi cantly reduces 0.5-1 mm and 1-2 mm ne root biomass, which is not compensated for by the stimulation of very ne roots ( Figure 2) [23,60].
A previous study found the recovery and growth of remaining stand productivity after thinning to be determined by ne root productivity [32]. Furthermore, very ne roots (< 0.5 mm) are mainly responsible for the acquisition and uptake of soil resources [27,61]. In this study, very ne root biomass and productivity exceeded CK in the T2 and T4 treatments, and productivity reached a signi cant level before biomass. This result corresponds with the results of Yuan and Chen (2012) [62], who found ne root production to peak much sooner than ne root biomass, indicating that stands under the T2 and T4 treatments would present higher productivity and increased biomass in subsequent development. However, compared to T2, T4 removed excessive aboveground organic components, which were di cult to recover in a short period time [63]. In terms of ne root biomass and productivity, our results reveal that a 30% approach (T2: moderately reduced aboveground biomass) may be a more suitable thinning strategy for promoting productivity and increasing stand heterogeneity in forest ecosystems. Previous studies have also reported that root gaps in forests recover faster and are more ephemeral than canopy gaps after thinning  [19], who found thinning practices to decrease or increase necromass. On the one hand, previous studies have demonstrated that the biomass and necromass of very ne roots account for a greater proportion of the ne root system than other root size classes [29], and our very ne root biomass results support this conclusion, which may theoretically determine the necromass. However, very ne roots also possess higher nonstructural carbohydrate concentrations and decompose more easily; thus, these roots are rarely observed [47,68,69]. On the other hand, thinning measures may improve soil conditions (e.g., soil temperature, Figure S4) and may also further increase the ne root decomposition rate [33]. Thus, ne root necromass levels did not differ between treatments, potentially due to a large amount of ne root necromass decomposing and disappearing [70,71]. We note that, however, T4 exhibited signi cantly increased necromass levels. This nding may be attributable to the fact that high-intensity thinning and species replacement produced a large amount of ne root necromass [35,37], especially for 0.5-1 mm and 1-2 mm ne roots, which had not completely decomposed, resulting in a signi cant increase in necromass in T4.
A previous study showed that the thinning of stands leads to relief from root competition for soil resources and increased average root longevity, resulting in a reduction in root mortality [72]. However, we found that thinning practices did not in uence the ne root mortality of the entire soil pro le. It may be that whole-tree harvesting measures caused a high degree of resource loss (especially nutrient loss) [55], which counteracted the resource competitive pressure relieved by thinning and led to mortality in thinning stands comparable to that of undisturbed forest stands. The signi cantly higher mortality of 0.5-1 mm ne roots observed in T1 supports this explanation (Figure 4h). Given the high variability in root responses observed, further experiments are needed to measure ne root necromass and mortality in different diameter classes and to determine the factors that drive the disappearance of very ne root necromass following thinning [63].

Response of deeper soil ne root dynamic characteristics to thinning
As hypothesized, we found thinning effects on ne root dynamic characteristics to be stronger at deep soil depths, and ne root dynamics exhibited uctuating patterns. These ndings are consistent with the results of a recent study showing dramatically altered ne root biomass and necromass levels at deeper soil depths following thinning [23]. This response may re ect the resource acquisition strategy of the root system following thinning. In the topsoil layer, the ne root systems of shallow root understory plants and the remaining trees quickly colonize and recover in the area liberated by the disturbance; thus, these roots are less affected by thinning practices [22,63]. It is well known that trees and understories jointly determine the dynamics of roots in the surface soil layer due to belowground niche partitioning, while trees determine such dynamics at deep soil depths ( Figure S3) [73,74]. In our study, all four thinning intensities reduced ne root biomass and productivity at deep soil depths (Figures 2 and 4), which may be because these practices reduced the ne root densities of trees and consequently alleviated resource competition pressure at this depth [72]. Compared to low-and high-intensity thinning measures at deep soil depths, higher ne root biomass and productivity levels were observed in the suitable thinning treatment (Figures 2 and 4). This change could be attributed not only to thinning interventions but also to the regeneration of understory plants [15]. In suitable thinning treatments, the faster regeneration of understory plants excessively consumes resources, which results in greater resource competition pressure in the surface soil and requires trees to adjust their rooting depth and increase their ne root growth at deeper soil depths with less root competition [19]. Thus, we observed the uctuating phenomena of ne root dynamics in deep soil layers. However, the ne root characteristics at deep soil depths could not recover in a short period of time after thinning (Figures 2 and 4), re ecting their more sensitive response to thinning practices. Moreover, 0.5-1 and 1-2 mm ne roots dominated ne root dynamic characteristics at the deep soil depths and exhibited an increased turnover rate, indicating that thicker ne roots could better mirror potential carbon pools at deeper soil depths in forest ecosystems.
Some studies have reported that ne roots can be categorized according to their functions (e.g., the 1st order of the root system plays a role in absorption) [47,75]. Furthermore, ne root dynamics may vary among different functional plant groups [19].
Although the current study uses a more nuanced classi cation method that diverges from the traditional de nition (≤ 2 mm) to study stand-level dynamic characteristics of ne roots, future studies could build on root functional approaches and plant functional group distinctions to better understand how root function speci city and species diversity impact belowground processes at the ecosystem level.

Conclusions
Our study suggests that thinning practices have substantial effects on the dynamic characteristics of multidiameter-class ne roots at different soil depths. The positive effect of thinning on very ne roots and the negative effect on thicker ne roots and all diameter classes of ne roots provide reasonable explanations for the inconsistent effects of thinning practices. Here, 30% (T2) thinning intensity moderately reduced the aboveground biomass and yielded increased biomass and productivity among very ne roots compared to the other treatments, suggesting that a 30% approach is a more suitable thinning strategy for promoting productivity and increasing stand heterogeneity in forest ecosystems. Fine root dynamic characteristics at deeper soil depths are more sensitive to thinning measures. The sampled 0.5-1 and 1-2 mm ne roots dominated ne-root dynamic characteristics at deep soil layers and exhibited a higher turnover rate, indicating that thicker ne roots could better mirror potential carbon pools of deeper soils in forest ecosystems. Collectively, our ndings provide important insights into the effects of forest management on ecosystem functions and into the climate change mitigation potential of the sequestration of belowground biomass carbon.

Experimental design and treatments
Thinning measures in secondary forest stands (dominated by Pinus armandii, Betula albosinensis, and Picea asperata ) were carried out from July to September 2013. All of the selected plots were of the same stand age (35 years), occupied similar topography, and had no history of fertilization. A randomized complete block design was used in the study. Five 20 × 20 m plots were randomly installed within each secondary forest block. The whole-tree harvesting method was used, and the following thinning intensities were applied: (1) no thinning (CK), (2) 15% removal of the stand volume (T1), (3) 30% removal of the stand volume (T2), (4) 45% removal of the stand volume (T3), and (5) 60% removal of the stand volume (T4). To avoid potential edge effects, each plot was surrounded by a 5-m-wide buffer zone. All harvesting materials were removed from the plots. Each of the ve thinning treatments was replicated into four blocks, totaling 20 sampling plots ( ve thinning intensities × four blocks). The layout of the experimental design is shown in Figure 1 (including blocks and plots).

Soil, litter and vegetation survey
Soil sampling was conducted at three soil depths (0-20 cm, 20-40 cm and 40-60 cm) using a soil auger (40 mm diameter) in August 2018. We collected nine replicate soil samples following an "S"-shaped pattern at three depths in each plot ( Figure 1). Then, the collected soil samples were fully homogenized from the same depth to form a composite soil sample. In total, 60 composite soil samples (5 treatments × 4 blocks × 3 depths) were collected. Plant and fauna residues were manually removed, and the soil was then passed through a 2 mm screen. The soil samples were then divided into two portions: the rst part was air-dried to measure soil organic carbon (SOC), pH, available nitrogen (AN), available phosphorus (AP) and available potassium (AK). The second portion was used to measure the water content after oven-drying at 105°C for 48 h. Soil samples of three duplicates were collected at depths of 0-20 cm, 20-40 cm and 40-60 cm by volumetric rings (100 cm 3 ) after continuously sunny conditions to measure soil bulk density [39,43].
The tree height (H) and diameter at breast height (DBH ≥ 5 cm, 1.3 m) in each plot were measured. Understory species diversity was investigated in ve shrub subplots (2 × 2 m) and ve herb subplots (1 × 1 m) established along the diagonals in each plot. Whole plant sampling techniques were used to determine shrub and herb biomass [38]. For litter sampling, all organic material (undecomposed and decomposed parts on the ground) in ve 1 × 1 m subplots was collected. Herb and litter subplots were located on larger shrub subplots, and all vegetation surveys were carried out in August at peak vegetation coverage as previously described [45].

Fine root sampling
The sequential soil coring method was used to collect ne root biomass, production, mortality and turnover rate data using a previously described method [46]. Because ne root dynamic processes exhibit strong seasonal variations, ne roots were sampled throughout the year [47]. Furthermore, we expanded the soil depth interval to 0-60 cm based on the average soil layer thickness in the study area.
In each sampling plot, we randomly collected eight soil cores (90 mm inner diameter) over the rst three days of September were collected over the four seasons.
The composite ne root samples were transported to the laboratory in an icebox. To separate the roots from the soil, we rst soaked the ne root samples in water. Then, three diameter classes of ne roots (<0.5, 0.5-1, and 1-2 mm determined using electronic calipers) were carefully washed and sorted into living and dead groups according to their status using the method described by Brassard et al. (2013) [48]. Live roots were classi ed as having a pale exterior, as elastic and exible, and as free of decay with a whitish cortex, while dead roots were brown or black in color and in exible. Finally, all of the live and dead ne roots of the three diameters were oven-dried at 65°C to a constant mass.

Chemical and biochemical analyses
All soil chemical indicators were determined following a previously described method [43]. The SOC content of soils was measured using the K 2 Cr 2 O 7 oxidation method. Soil available nitrogen (AN) was identi ed by alkaline hydrolysis diffusion, and available phosphorus (AP) was measured by colorimetry after extraction with NaHCO 3 . Soil available potassium (AK) was extracted in ammonium acetate (pH 7.0) and identi ed on a ame photometer. The soil pH was determined in a 1:2.5 soil:water suspension.
The soil bulk density was obtained by calculating the ratio of soil mass to total volume (g·cm −3 ) after oven drying at 105°C to a constant weight [44].

Data calculation and analysis
Fine root biomass (g m −2 ) and necromass (g m −2 ) were calculated for each sampling season in each plot by summing the dry weight of live and dead ne roots in each soil core. Fine root production (g m −2 year −1 ) and mortality (g m −2 year −1 ) were determined using a simpli ed decision matrix method (Table 2) [19]. The ne root turnover rate (year−1) was de ned as the ratio of annual ne root production (g m −2 year −1 ) to the mean biomass (g m −2 ) of ne roots over a year [49].
In the present study, the biomass characteristics of ne roots were repeatedly measured across sampling seasons by soil depth within each plot. Therefore, we performed a linear mixed model analysis with three xed effects (thinning intensity (T), sampling season (S), and soil depth (D)) and random effects (plot and block) as described by Y ijkl = T i + S j ( l ) + D k ( l ) + T i × S j ( l ) + T i × D k ( l ) + S j ( l ) × D k ( l ) + T i × S j ( l ) × D k ( l ) + π l (1) where Y ijkl is ne root biomass, necromass, or total mass (g m −2 ); T i (i = 0, 15, 30, 45, 60) is the thinning intensity; S j(l) is the sample season (i.e., autumn, winter, spring, and summer); D k(l) is the soil depth (0-20 cm, 20-40 cm, and 40-60 cm); and π l is a random plot or block effect (l = 1, 2, …, 20).
The ne root characteristic percentage change (compared to CK) of the average for all thinning treatments at a certain soil depth could be seen as an indicator for evaluating ne root characteristic responses at different soil depths to thinning intervention. A higher percentage change of one soil depth indicates a more sensitive response to thinning measures, which was calculated as follows: where C p denotes the ne root biomass, necromass, total mass, production, mortality or turnover rate percentage change value; T n (n = 1, 2, 3, 4) is the thinning measure; C Tn and C CK are the ne root characteristic mean values for the thinning and control conditions, respectively; and | C Tn -C CK | represents the absolute value of C Tn -C CK .
The effects of thinning intensities on soil properties (water content, bulk density, SOC, AN, AP AK and pH) and stand characteristics (tree density, height, DBH and volume, understory vegetation biomass and species diversity index, and litter biomass) were also tested using a linear mixed-effects model ANOVA. For all models, the signi cance of xed effects was assessed using Satterthwaite approximations for degrees of freedom. When xed effects or interactions were signi cant, the least square means differences test was performed for multiple comparisons (main effect or simple effect analysis). The statistical value F was used to evaluate the sensitivity differences of three diameter classes of ne roots to thinning measures. The linear mixed-effects model was obtained with the 'lmerTest' and 'lme4' packages [50,51]. Multiple comparisons were draw using the 'emmeans' package. Principal component analysis (PCA) was performed to determine the relationships between ne root dynamics and stand characteristics and soil properties, using the 'FactoMineR' package [52]. All analyses were implemented using R for Windows version 4.1.1 statistical software [53].

Declarations
Ethics approval and consent to participate Not applicable Consent for publication Not applicable Availability of data and materials The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request. Location of the study area and the layout of experimental plots. CK, T1, T2, T3, T4 representing 0%, 15%, 30%, 45% and 60% thinning intensity, respectively.

Figure 2
The effects of thinning intensities on the ne root biomass, necromass, and total mass (average over 4 seasons) among different diameter classes in different soil depths (a-c, e-g and i-k) and entire soil pro le (d, h and l). Values are the mean of 4 replicates ± SE. CK, T1, T2, T3, T4 representing 0%, 15%, 30%, 45% and 60% thinning intensity, respectively. Different lowercase letters (a, b, c, d) indicate signi cant differences for same diameter class ne roots among thinning intensities (p < 0.05).

Figure 4
The effects of thinning intensities on ne root production, mortality and turnover rate among different diameter classes in different soil depths (a-c, e-g and i-k) and entire soil pro le (d, h and l). Values are the mean of 4 replicates ± SE. CK, T1, T2, T3, T4 representing 0%, 15%, 30%, 45% and 60% thinning intensity, respectively. Different lowercase letters (a, b, c, d) indicate signi cant differences for same diameter class ne roots among thinning intensities (p < 0.05).

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download.