Incorporation of fine root detritus into forest soil organic matter

One of the principal inputs of organic matter to forest soils is turnover of tree fine roots, but the process of decomposition of fine root litter and its conversion into stable soil organic matter (SOM) has received limited study. We labeled fine roots of sugar maple (Acer saccharum Marsh.) with 13C and traced the label for 7 years into four contrasting soils to improve understanding of this process. After 7 years we recovered an average of 8.9% of the 13C label, with about two-thirds recovered as coarse particulate organic matter and one-third in microaggregates and on silt and clay particles. No differences in 13C recovery were detected between 1–2 and 3–4 order fine roots. Most of the 13C in microaggregates (53–250 µm, 58%) was occluded within macroaggregates, and the recovery in this fraction increased significantly from year 2 to 7, illustrating the role of fine root detritus in the formation of microaggregates. This process was most pronounced in the A horizon of a higher pH soil (pH = 5.5) with high iron oxide content. Conversely, the lowest 13C recovery in this fraction was observed in the A horizon of an acidic, fine-textured Inceptisol (Cambisol—World Reference Base). We estimate that annual input into relatively stable fractions of SOM represents about 14% of the total annual accumulation in these fractions; thus, our results support recent evidence that fine root litter is only a moderate contributor to stable SOM in acid temperate forest soils.


Introduction
Forest soil organic matter (SOM) is derived from aboveground and belowground detritus and rhizosphere carbon flux (root exudates; rhizodeposition, mycorrhizal fungi; Farrar et al. 2003).Although aboveground plant detritus usually comprises the largest proportion of total C input to forest soils, biochemical tracers indicate that roots are principal contributors to forest SOM (Nierop 1998;Angst et al. 2021), and evidence from soil food web studies suggests the dominant role of below-ground inputs in powering forest soil food webs (Pollierer et al. 2007;Gilbert et al. 2014).Fine root detritus typically comprises about one-fourth of organic matter input in temperate deciduous forest soils (Fahey et al. 2005) although considerable uncertainty remains because of poorly constrained estimates of root turnover (lifespan) and rhizosphere carbon flux.Moreover, the proportion of fine root detritus that is retained as relatively stable SOM, and factors influencing that stabilization, are not fully understood.
The concept that SOM is stabilized biochemically by the formation of recalcitrant detrital residues has been replaced by the evident role for biophysical mechanisms that protect even labile organic compounds from microbial decomposers (Dungait et al. 2012).The two prominent physical mechanisms are (1) sorption of specific organic compounds onto the surface of mineral soil colloids (Kogel-Knabner et al. 2008), and (2) detritus being occluded within soil aggregate structures (Six et al. 2004).These two mechanisms are not mutually exclusive.The mineral particles in soils-sand, silt, clay-that associate with organic compounds are ultimately bound into aggregate structures (Harris et al. 1966;Tisdall and Oades 1982;Six et al. 2004).In general, two size classes of aggregates are distinguished: microaggregates (< 250 µm in diameter) and macroaggregates (250 µm to 2 mm).Organic matter occluded within aggregates is less accessible to microbial exoenzymes and because microaggregates are thought to be more stable structures than macroaggregates, the organic matter within them is also expected to be more stable (Golchin et al. 1994).For example, stable and colloidal organic matter precipitated and adsorbed on the surface of clay mineral colloids (phyllosilicates and allophanic minerals) and (co-) precipitated on metal (oxyhydr-and hydr-) oxides, particularly aluminum, iron and manganese (Kleber et al. 2021), appears to be resistant to microbial utilization, being comparatively "older" based on 14 C analysis (turnover time of several centuries) than other fractions of SOM like low density particulate matter (Trumbore 2000;Cotrufo and Lavallee 2022).These relatively stable forms of SOM in microaggregates and mineral associated organic matter (MAOM; Jastrow 1996) play a key role in the global carbon cycle as large stocks of SOM.
Root systems of trees can be characterized in terms of size distribution, but functionally root order is a more useful distinction (McCormack et al. 2015).First and second-order fine roots, the most distal in the root system, are primarily involved in uptake of soil resources.They also turn over more rapidly (i.e., shorter lifespan) than higher order roots that are primarily involved in transport, storage, and anchorage.The lower order roots typically have higher nitrogen and lower lignin concentrations than the higher orders (Pregitzer et al. 2002) and based on these differences in organic matter quality they might be expected to decay more rapidly (Melillo et al. 1982).However, some comparative studies of fine root decay have concluded the opposite: faster or equal decay rates of higher order (3-4) compared to lower order (1-2) fine roots (Fan and Guo 2010;Goebel et al. 2011;Beidler and Pritchett 2017).
The objective of the present study was to evaluate the long-term decomposition of fine roots of sugar maple (Acer saccharum Marsh) and incorporation of this detritus into aggregate fractions in forest soil.We labeled fine root systems of sugar maple trees with 13 C, excavated the roots, and incubated the roots (without mesh bags) for 7 years within soil cores.We compared decay of two size classes of fine roots (order 1 and 2 vs order 3 and 4) and four soils with differing texture, structure, and chemical characteristics.We hypothesized that differences in root litter quality between lower and higher order fine roots would correspond with differences in decay rates and incorporation into the relatively stable SOM fractions (microaggregates and silt plus clay).We expected that root-derived carbon would accumulate through time in the microaggregate fraction, reflecting the role of fine root detritus in formation of these structures (Tisdall and Oades 1979;Rillig et al. 2015).Finally, we hypothesized that differences among the four soils, especially texture, pH and metal oxide content, would correspond with differences in the distribution of root litter-derived carbon among soil fractions after six to 7 years of root decomposition; in particular, we expected greater accumulation in stable SOM fractions for soils with higher clay, pH and iron oxide contents (Eusterhues et al. 2005;Wiesmeier et al. 2019).We hoped that this study would advance understanding of the mechanisms of incorporation of fine root detritus into forest SOM.

Study site and soils
The field incubations of decaying fine roots were conducted in a sugar maple forest at the Turkey Hill Plantations (THP) located in Dryden, New York USA.
Vol.: (0123456789) The THP were established in 1940 on abandoned fields by R.F.Chandler and associates to evaluate the effects of different forest trees on soil properties.The THP have been described in detail elsewhere (Phillips and Fahey 2006), and a brief description of the sugar maple plantation is provided here.At the time of study initiation, the trees were 72 years old with a stand basal area 33 m 2 ha.The stand is essentially pure sugar maple with over 99% of basal area comprised of this species over its 0.4 ha extent.The elevation at the site is 430 m with a very gentle slope.For nearby Ithaca, NY, for the period 1981-2010, the average annual precipitation is 990 mm and average annual temperatures are − 1 °C and 18.7 °C in January and June, respectively.Soils at the THP are developed from silt-enriched glacial till derived from local sandstone and siltstone.They are classified as coarseloamy, mixed mesic Typic Fragiudepts (Inceptisol), from the Bath and Mardin series, based on the U.S. Department of Agriculture, Seventh Approximation (Soil Survey Staff 1999) and a Cambisol in the World Reference Base (WRB) system (IUSS 2006).
Soils used in the fine root decay incubations were collected from four locations, including the A horizon from the THP sugar maple stand (hereafter designated IN A1) and three sites near Old Forge, New York in the Adirondack Mountains.The soils were sieved at 2 mm.They were chosen to span a range of soil properties: pH, organic matter content, structure, chemistry, and texture (Table 1).The soils used in the incubations from the three sites in the Adirondacks were distinguished by the soil parent materials and consequent soil horizons: (1) SP B1, the Bh horizon of a Typic Haplorthod (Spodosol, Essex series, Soil Survey Staff 1999;Podzol, WRB, IUSS 2006) developed in bouldery glacial till derived primarily from mixed gneisses; (2) IN A2, the A horizon of a Cryochrept (Inceptisol, Soil Survey Staff 1999;Brunisol, WRB, IUSS 2006) developed in glacial till derived from gneisses but including local marble; and (3) SP B2 the B2ir horizon of a Haplorthod (Spodosol, Colton series, Soil Survey Staff 1999;Podzol, WRB, IUSS 2006) developed in coarse glacial outwash and known colloquially as an "iron podzol".The composition of the mature forests at the Adirondack sites also differed; SP B1, mixed northern hardwood (maplebeech-yellow birch); IN A2, maple-basswood; and SP B2, white pine-red pine.

Isotope labeling of fine roots
Root systems of sugar maple trees were labeled with enriched 13 C.The procedures for isotope labelling have been described in detail elsewhere (Horowitz et al. 2009;Fahey et al. 2021).Briefly, sugar maple saplings growing in the understory of a thinned maple stand were enclosed in large chambers (3 m diameter).Labeled 13 CO 2 (40 atom % enriched) was added to each chamber on several sunny mornings in early September 2006 and again in September 2010, and the saplings assimilated the label.Thus, root systems of the maple saplings were strongly labeled with 13 C (Table 2).In mid-summer 2012 roots within the chambers were excavated by hand, cleaned of soil, and returned to the laboratory.Fresh fine roots were sorted manually into two fractions: order 1-2 feeder roots and order 3-4 transport roots (McCormack et al. 2015).Root samples were air dried, and four subsamples were collected for initial moisture and chemical determinations.Preparation and deployment of root decomposition cores Fine root decomposition and incorporation into SOM were quantified by adding isotope-labeled fine root (order 1-2 and order 3-4) detritus to soil and incubating in 5 cm diameter × 20 cm long PVC pipes with several 2 cm holes drilled in the sides to allow ingrowth of fungal hyphae and roots.These root decay cores were prepared by precisely weighing air-dry fine roots (ca.1-2 cm long fragments) and mixing into each of the four soil horizons described above at a ratio that was like that observed in regional sugar maple forests (i.e., 300 g m −2 for order 1-2 and 100 g m −2 for order 3-4; Fisk et al. 2004).Soil-root mixtures were added to each core to roughly match bulk density in these soils.Additional samples of cores with each soil but without roots added were prepared for reference measurement of 13 C natural abundance in each aggregate fraction (Table S1).
In the sugar maple stand at the THP, soil cores 5 cm diameter × 20 cm deep were removed to allow installation of the root decay cores.The cores were installed on a grid with the different soil horizons and root types (fine and coarse) arranged at random along several transects.A total of 180 root decay cores were installed in October 2012 (two root types x four soil horizons × 20 replicates), as well as five reference cores of each soil horizon.

Root decay core collection and processing
A sub-set of each root decay core type was collected periodically and returned to the laboratory for processing.In October 2013 (year 1), 2014 (year 2), and 2015 (year 3), three or four cores of each root and soil horizon were collected, and the samples were composited.In 2018 (year 6) and 2019 (year 7), four cores of each type were collected, and each core was processed and analyzed individually.One reference (no roots added) core of each soil horizon also was collected in year 0, 1, 2, and 7 and processed as for the labeled cores.In the laboratory soil samples were removed from the cores, thoroughly mixed, sieved through a 2 mm mesh and air-dried.
Water-stable aggregates were fractionated by wet sieving of an air-dried portion (80 g) of the < 2 mm fraction of soil following the scheme in Elliott (1986).Soil was spread evenly on a sieve with an opening of > 250 μm, and the sieve was immersed in water for 5 min ("slaking").The sieve was then moved up and down at a rate of 50 strokes in 2 min.Floatable material was decanted and saved since it was part of the < 2 mm fraction (see Methods Supplement); we refer to this fraction as low-density POM.Except during the first year a fine mesh wand was used to fully collect floatable particles.Material that passed through the 250 µm sieve was allowed to collect on or pass through a sieve with an opening of 53 µm.The three fractions were washed into pre-weighed pans, i.e., macroaggregates (> 250 µm), microaggregates (53-250 μm) and free silt plus clay fraction (< 53 μm).The three size fractions were oven-dried at 60 °C and 13 C was measured as detailed below.
A subsample of the macroaggregate fraction was further separated into component parts following the procedure in Bossuyt et al. (2004).Briefly, 10 g of oven-dried macroaggregates were slaked in deionized water for 5 min.These samples were then gently shaken with 50 stainless-steel ball bearings (4 mm diameter) while submerged on top of a 250 μm sieve until all macroaggregates were broken.A continuous stream of water flushed the < 250 μm material through the mesh to avoid disruption of microaggregates released from the macroaggregates.Further sieving of the < 250 μm fraction through a 53 μm sieve resulted in three size fractions isolated from the macroaggregates, i.e., coarse particulate matter (> 250 μm; this fraction includes both sand and coarse POM), microaggregates (53-250 μm) and silt plus clay (< 53 μm).Each of these fractions was rinsed into pre-weighed aluminum pans, oven-dried at 60 °C, and 13 C concentration was measured as detailed below.Using this procedure, the mass of soil and 13 C in each of these aggregate fractions was estimated.

Calculation of 13 C recovery
Percent recovery in each soil aggregate fraction of the 13 C added to each root decay core was estimated based on the 13 C enrichment of the fractions relative to the natural abundance values for each soil horizon and aggregate fraction in the reference soils; the average of three reference cores (year 1, 2, and 7) for each soil horizon was used as there were no temporal trends observed in natural abundance.Added 13 C was estimated for each root decay core as the product of the root mass added, %C and atom % 13 C of the added fine roots (Table 2).For each sample (or composited sample in years 1, 2, and 3) the mass proportion of each soil fraction was multiplied by the measured soil mass in each core to estimate the fraction mass.These fraction mass values were multiplied by %C and atom % 13 C enrichment to estimate the mass of 13 C in each reference and root decay core.The difference in this 13 C pool between root decay and reference (natural abundance) samples was used to estimate % recovery of the added 13 C in each fraction.Total recovery per core was estimated as the sum of the soil fractions: low-density POM, free microaggregates, free < 53 µm particles (silt plus clay), and the three components of macroaggregates, coarse particulate matter (CPM, mostly sand), occluded microaggregates, and occluded < 53 µm particles (silt plus clay).

Measurement of soil and root chemistry and soil properties
Total carbon, nitrogen and carbon isotope concentrations in soil and root tissue samples were determined on a Finnegan Isotope Ratio Mass Spectrometer at the Cornell Stable Isotope Laboratory, with appropriate standards for normalization correction, instrument linearity, and precision purposes.We determined acid detergent lignin (ADL) on an ANKOM fiber analyzer (ANKOM Technology, Macedon, NY, USA).We determined concentrations of metal oxides (Al, Fe, Mn) on separate subsamples of homogenized soil samples (< 2 mm) and ground to pass a 35-mesh sieve for each soil horizon by extraction in citrate-dithionite (Fe d ) and for a separate sample in ammonium oxalate (Fe o ) with detection by ICP-OES (Courchesne and Turmel 2008).Soil pH was determined in deionized water using a glass electrode.Soil particle size concentrations were determined by the pipette method following pretreatment to remove organic matter and iron oxides (Gee and Or 2002).Total carbon was determined by automated combustion and gas chromatography with thermal conductivity detection using a Thermo Flash 1112 analyzer (CE Elantech, Lakewood, NJ, USA).Base cations (Ca, K, Mg, Mn, and Na) were determined by extraction in Morgan's solution with detection by ICP-OES (McIntosh 1969).Soil properties were measured by the Cornell University Soil Testing Laboratory and are presented in Table 1.

Statistical analysis
The effects of soil horizon, root type and year of collection on total percent 13 C recovery were analyzed using repeated measures analysis of variance (RM ANOVA).Percentage values were log10 transformed to achieve normality of residuals.For collection years six and seven there were four replicates of each soil horizon and root order.A separate RM ANOVA was performed for those 2 years to allow further analysis of percent 13 C recovery in the different aggregate fractions among soil and root types.Because collection year had a significant effect (p = 0.001) on % recovery for year 6 and 7, analysis of soil horizon effects on % recovery in the different fractions was done separately for these years.A one-way ANOVA was used to evaluate soil horizon effects on 13 C recovery for each fraction.A log10 transformation was applied to 13 C recovery values to achieve normality of residuals.In cases where significant soil horizon effects were detected, post hoc mean separation was evaluated using Turkey's HSD.Finally, significant differences among the 5 years of collection for the mass of each aggregate fraction were evaluated using RM ANOVA on logit transformed values.

Aggregate fractions
Most of the soil particle mass was contained in aggregate structures as operationally defined by the slaking Vol:.( 1234567890) procedure described in Methods.The proportion of macroaggregates (> 250 µm) and microaggregates (53-250 µm) differed significantly (RM ANOVA, soil horizon effect; F 3, 159 = 4.43, p = 0.0052) among the four soils at the time of the last collection (Table 3).In particular, the abundance of free microaggregates (i.e., those not contained within macroaggregates) ranged from 209 mg/g in IN A2 to 246 mg/g in IN A1; the latter soil had especially high proportion of the fine fraction (< 53 µm, 182 mg/g) reflecting the high silt and clay content based on standard particle size analysis (Table 1).
The composition of the macroaggregate fraction also differed considerably among the soils (Table 3).For example, the abundance of microaggregates occluded within macroaggregates ranged from 87 mg/g for SP B2 to 171 mg/g for SP B1 (RM ANOVA, soil horizon effect; F 3, 39 = 3.43, p = 0.0271).Except for the high silt IN A1, most of the mass of macroaggregates was in coarse particles, mostly sand grains (Table 3).Carbon concentration in the different soils and their aggregate fractions also varied markedly (Table S-2).The highest C concentrations were generally found in the silt plus clay (< 53 µm) fractions, ranging as high as 9 to 10% for the spodic B horizon (SP B1) which had the highest overall SOC content (Table 1).The lowest values in this fraction were found in IN A1 (2.1 to 2.5%); much of the carbon in this soil was found in microaggregates occluded within macroaggregates (Table S-2).This fraction was especially enriched in SOC in soil IN A2 (10%).Finally, C concentrations of the coarse particles (mostly sand) in the macroaggregates were very low (average = 1.5%;Table S-2).
The proportional abundance of the various aggregate fractions did not change significantly across the 7 years of incubation (RM ANOVA, year effect; F 4, 287 = 0.41, p = 0.8042).In year 7 the abundance of the low-density POM remained similar to that observed initially, ranging from 2.1 mg/g (SP B2) to 8.2 mg/g (IN A1); these values were undoubtedly maintained by visually obvious ingrowth and subsequent turnover of fine roots in the soil cores.
Total recovery of 13 C Recovery of 13 C added to soil cores as strongly labeled fine roots of sugar maple declined through 7 years of in situ incubation (Fig. 1).After 1 year, total 13 C recovery was unexpectedly low, especially for the low-density POM.This low recovery was a methodological artifact (see Supplement 1, methodological notes) that limited recovery of this fraction; in subsequent years this problem was corrected by a revision of the procedure for the decanting step in the aggregate separations (see Methods).Note that in Fig. 1 the value for year 1 is based on measurements of fine root decay in the same site (Fahey, unpublished data) using an in situ root core method (Dornbush et al. 2002).Also, although the relatively stable fractions (microaggregates and silt plus clay) were a small proportion of total recovery (Fig. 1), there was considerable recovery in these fractions in all years (Fig. 2).
No significant difference was detected for total recovery of 13 C between order 1-2 and order 3-4 fine roots for years 2 through 7 (RM ANOVA, root order x year interaction; F 1, 73 = 1.69, p = 0.1976).For all subsequent analyses these two root orders were combined as no significant interactions of root type with soil horizon or aggregate fractions were observed (thus, n = 8 for each sample type in years 6 and 7).
The time course of total 13 C recovery followed the typical exponential pattern with one exception: a  48) 241 ( 20) 124 ( 37) 461 ( 65) 87 ( 7) 55 (10) significant decline in 13 C recovery was observed from year 6 to 7 (RM ANOVA, year effect; F 1, 71 = 11.30,p = 0.0012; Fig. 1).Differences in total % 13 C recovery among the four soil horizons were mostly nonsignificant.Initial fine root decay rates, after 2 years, were similar across all the soils.By year 6 total recovery tended to be slightly lower for soil SP B2 than the others, but this trend was not statistically significant (RM ANOVA, soil horizon effect; F 3, 31 = 1.05, p = 0.3866).However, by year 7 between soil horizon differences were significant (RM ANOVA, soil horizon effect; F 3, 31 = 3.19, p = 0.0371) driven primarily by lower % recovery for soil SP B2 (Table S-3).

Recovery in various aggregate fractions
Throughout the incubation period most of the added 13 C was recovered in the low-density POM fraction, and recovery in this fraction declined exponentially (Fig. 1).Percent recovery in the four fractions we designate as relatively stable fractions (i.e., free and occluded microaggregates and free and occluded silt plus clay) fluctuated through time, in a few cases exhibiting marked changes (Fig. 2).In particular, we observed an obvious peak in the recovery of these fractions in the first year, and then from year 1 to 2, total recovery in these fractions declined markedly (Fig. 2).Thereafter (year 2 to 7), temporal variation of recovery in the four relatively stable fractions was statistically significant (RM ANOVA, year effect F 4, 84 = 2.96, p = 0.0246).Recovery in microaggregates increased from year 2 to 7, especially for microaggregates occluded in macroaggregates which increased significantly through time (F 4, 81 = 3.86, p = 0.0082).Overall, the time trend for recovery in microaggregates from year 2 to 7 was best described by a second-order polynomial function (R 2 = 0.76).
Recovery in silt plus clay fractions remained roughly constant from year 2 to 7 despite some minor fluctuations (e.g., R 2 = 0.32 for polynomial function).By year 7 most of the 13 C that was not in low-density POM was recovered in microaggregates (Fig. 2).A small amount was also recovered in coarse particles in the macroaggregates, apparently somehow co-mingled with sand particles; 13 C values for this fraction were erratic, presumably owing to measurement error resulting from low C concentrations (Table S-2).
The distribution of recovery among the aggregate fractions varied significantly among soil horizons by year seven (RM ANOVA, soil horizon effect F3 4, 31 = 3.19, p = 0.0371).In particular, % recovery in occluded microaggregates was very high in soil IN A2 and very low in soil SP B2 (Fig. 3).Despite its coarse texture, the latter soil exhibited high recovery in the silt plus clay fraction.Finally, we estimated recovery in each of the stable fractions relative to the C stock in those fractions for soil IN A1 which is the only soil for which we have detailed bulk density and C stock values.In absolute terms, total recovery in the stable fractions in the upper 20 cm soil was a very small proportion of the C stock, averaging 0.05% and ranging from 0.02% (free silt plus clay) to 0.07% (free microaggregates).

Discussion
The role and mechanisms of fine root detritus in supplying stabilized forest SOM have received limited study.We traced 13 C from two fine root types (order 1-2 vs. 3-4) of sugar maple into several soil fractions in four acid forest soils over 7 years to better understand fine root detrital dynamics.We hypothesized that differences in tissue quality between lower and higher order roots would correspond with differences in fine root decay, but no significant differences were detected in 13 C recovery between order 1-2 and order 3-4 fine roots after the first year.Notably, the differences in substrate quality between these root types were modest (Table 2).Goebel et al. ( 2011) observed similar decay rates for order 1 to 4 fine roots whereas other studies have reported significant differences in decay rates between lower and higher order roots; however, the direction of these differences varied among studies, some reporting higher and others lower decay rates for lower order roots (Fan and Guo 2010;Biedler and Pritchard 2017).Our results indicate that any small differences in substrate quality probably do not significantly influence the longer term (7 year) decay and incorporation of fine root detritus into SOM in these temperate forest soils.
Our study was designed primarily to explore how differences in soil properties might influence stabilization of fine root detrital C. The four soil horizons chosen for study differed markedly in texture, structure, SOC, pH, and metal (Al, Fe) content (Table 1).After 7 years the only difference in overall 13 C recovery that we observed among these four soils was the significantly lower recovery in soil SP B2.This spodic horizon from an "iron podzol" (Spodosol) that developed on very coarse glacial outwash was distinctive in having the lowest content of occluded microaggregates (Table 3); however, the overall low recovery was mostly associated with low values for the low-density POM fraction.The cause of the more complete decay of POM in this soil horizon is not clear.The small differences in total recovery among the other three soils is suggestive of more subtle influences of physical and chemical properties on the overall stabilization of root litter-derived SOM.
Our study provides some insights into the process of incorporation of fine root detritus into SOM.The traditional concept that resistance to microbial decomposition was afforded biochemically has been replaced by recent work ascribing SOM stabilization primarily to physical mechanisms that protect inherently labile organic matter from degradative microbial Fig. 3 Percent recovery of 13 C, added as labeled fine root detritus of sugar maple, after 7 years of decomposition in four forest soil horizons with contrasting soil properties (see Table 1).Mean values and standard errors for four different fractions of stable soil organic matter based on pooled data for two root order classes (1-2, 3-4) exoenzymes (Dungait et al. 2012).Two interrelated mechanisms contribute to this stabilization of SOM: sorption on mineral colloid surfaces (phyllosilicates, metal oxides) (Kogel-Knabner et al. 2008), and protection within soil microaggregates (Six et al. 2004;Lehmann et al. 2007).The latter are thought to form because of binding by and between organic matter and soil minerals and may afford physical protection by retarding the entry of biota, enzymes, oxygen, and nutrients (Totsche et al. 2018;Biesgen et al. 2020).Resistance to oxidative degradation of organic matter sorbed to surfaces of mineral colloids may result from strong binding, especially within the confines of microaggregates (Churchman et al. 2020;Kleber et al 2021).
Soil aggregates are an intimate mixture of organic matter, soil minerals, and microorganisms and their by-products.Many studies have examined aggregate formation and stability, especially in agricultural soils, because of impacts of tillage and cropping on soil C (Six et al. 2004).The current paradigm is that microaggregates (< 250 µm in size) are a relatively stable mixture of secondary minerals, highly processed detritus and by-products of microbial decay.These microaggregates are bound together into less stable macroaggregates (> 250 µm) by fresher organic matter (Golchin et al. 1994), and disruption of macroaggregate structures can lead to more rapid microbial turnover of the organic components (Six et al. 2004).
After 1 year of fine root decomposition, recovery of the 13 C label in soil aggregate fractions probably represented mostly soluble organic C adsorbed on soil colloids as well as some POM, including microbes.Kaiser and Guggenberger (2003) argued that much of the sorbed SOM remains labile, and Kleber et al. (2021) emphasized the continued reactivity of organic matter adsorbed on mineral surfaces.Moreover, Saidy et al (2013) reported desorption of about 10% of sorbed DOC from various secondary clay minerals.In support of these concepts, we observed a decline in percent recovery in these fractions during the second year despite continuing supply of labeled root decay products.The initially high recovery in macroaggregates, followed by the reduction in years 2 and 3 also supports the role of fresh root litter in macroaggregate formation (Six et al. 2001) as well as the relatively low stability of these structures (Rabbi et al. 2014).
During the later stages of fine root decomposition (3 to 7 years) we observed steady or increasing percent recovery in relatively stable soil fractions (Fig. 2) Apparently, continuing decomposition of root-derived components of these fractions (microaggregates and silt plus clay) was counterbalanced by continuing supply to these fractions from microbial processing of detrital residue.Most striking was the large increase of recovery in microaggregates, especially those occluded within macroaggregates, in year 6 and 7.This result strongly supports the concept that microaggregate formation is influenced by organic matter derived from fine root decay.
Total recovery of fine root detrital 13 C declined significantly in year 7 across all the soils; this decline was primarily associated with declining recovery in the low-density POM fraction.At the same time, recovery in occluded microaggregates increased significantly.Although the cause of this striking pattern is uncertain, it is notable that warm season (May-October) rainfall in year 7 was by far the highest during the entire study, exceeding the long-term average by about 20%.We speculate that exceptional wetting caused natural slaking of macroaggregates along with incorporation of the label into new, waterstable microaggregates during this summer season (cf.Amézketa 1999;Menon et al. 2020).
Based on the differences in soil properties and % recovery in various fractions among the four soils, we can speculate on the mechanisms influencing the incorporation of fine root detritus into relatively stable SOM fractions in these acid forest soils.After 7 years, recovery of root-derived C in the stable fractions differed significantly among the four soils (Fig. 3).Most striking was the high recovery in occluded microaggregates in the A horizon of a higher pH soil (pH = 5.5) that coincidentally had the highest metal oxide content (IN A2; Table 1).The source of the metal oxides in this soil is uncertain, but periodically poor drainage may cause anoxic conditions, reduction of iron, and re-oxidation and re-precipitation as iron oxides under subsequent oxic conditions.This process is possible at the higher pH that facilitates re-oxidation of reduced iron (Cornell and Schwertmann 2003).Very high recovery was observed in multiple replicates; for example, five of eight replicates of this soil exceeded 3% recovery in this fraction in year 7.This high recovery was mirrored by high C concentration in this fraction (9.9%;Table S2).Significantly lower total 13 C recovery was observed in this fraction in the two spodic horizons (SP B1 and SP B2) with low pH and more moderate concentrations of metal oxides (Table 1).
Although macroaggregates were abundant in the B2ir horizon soil (SP B2; Table 3), by year 7 incorporation of root-derived C into these macroaggregates was very low, and especially for microaggregates occluded within them (Fig. 3).Notably, these occluded microaggregates had low C concentrations in this soil (2.7%) compared with the more typical spodic Bs horizon soil (SP B1, 5.8%).The SP B2 soil also had its highest recovery in the free silt plus clay fraction.Thus, the fate of root litter-derived C differed markedly between these spodic B horizons, with a much higher proportion added to silt plus clay in the B2ir and more going into microaggregates in the Bs horizon.
The Inceptisol A horizon collected from the incubation study plot (IN A1) had much higher clay content than the Adirondack soils (12% vs. 4%; Table 1), yet isotope recovery in the silt plus clay fractions of this soil was similar to the others (Fig. 3).On one hand, this is surprising because the dominant phyllosilicate mineral in the IN A1 soil is illite (M.McBride, pers.comm.), which has a higher capacity for DOC sorption than kaolinite (Saidy et al. 2013) and vermiculite, the dominant clay minerals in the Adirondack soils (Kahle et al. 2004).On the other hand, organic matter binding to phyllosilicates occurs via cation bridging and exchange (Kleber et al. 2021), and perhaps the low concentration of exchangeable Ca in this acidic soil limits sorption.
After 7 years of fine root decay about 4% of the added 13 C was recovered in low-density POM, representing over half of the total recovery; most of the remainder was recovered in the relatively stable fractions.Just how stable is the new C in these fractions?According to current concepts most MAOM, especially that contained in microaggregates, and strongly sorbed to metal oxides via ligand exchange, is probably very stable (Kleber et al. 2021).
To illustrate the magnitude of the contribution of fine root detritus to stable SOM we estimated the mass of carbon that accumulated in the four relatively stable fractions after 7 years of decomposition for the forest soil at the incubation site-IN A1-(approach detailed in Supplement).In the top 20 cm of soil, we estimate 1.9 g C/m 2 year based on measured fine root biomass (373 g/m 2 ), assumed root turnover of 0.5/ year and observed % recovery values for the stable fractions (Fig. 3).This value compares with an estimated total annual C accumulation in these stable fractions of 14 g C/m 2 assuming a turnover time of 300 years as estimated by C isotope analysis for Harvard Forest, Massachusetts, USA (Trumbore 2000).Thus, the contribution of fine root detritus to stable SOM fractions is estimated to be about 14% (i.e.1.9/14), the remainder coming from aboveground detritus, rhizosphere C flux and coarse root detritus.

Comparisons with living root inputs and leaf litter decay
Parallel studies of sugar maple fine roots, litter and SOM provide information relevant to our interpretations.At a nearby site on the same Mardin series soil horizon as IN A1, Yavitt et al. (2015) traced 13 C from roots of sugar maple saplings into soil aggregate fractions; note that this C would include some fine root turnover as well as lots of rhizosphere C flux (Phillips and Fahey 2005).After 3 years, 21.5% of the label from living tree roots was recovered in soil (compared with 13.1 ± 2.1% for root detritus in the present study (Fig. 1).The distribution of this root-derived 13 C among soil aggregate fractions also differed qualitatively from the root litter in the present study (Table 4).In particular, after 3 years a much lower proportion of total recovery was found in microaggregates derived from root litter (6.7%) compared with the sapling root system label (24.2%).Together with previous results, this observation accords with the conclusion of Sokol et al. (2019a) that C inputs from living roots (rhizosphere C flux) are most readily converted into stable SOM.
We can also compare the dynamics of decomposition and SOM formation between fine root detritus and leaf litter.Yavitt et al. (2015) measured recovery of 13 C labeled sugar maple leaf litter in the relatively stable fractions for similar soils as IN A1 (Mardin series).After 2 years, leaf litter 13 C was quite uniformly distributed among the four stable fractions (microaggregates = 0.51%; free silt plus clay = 0.66%; occluded microaggregates = 0.66%; and occluded silt plus clay = 0.42%); by comparison, the pattern for fine root litter was similar (free microaggregates = 0.75%; free silt plus clay = 0.39%; occluded microaggregates = 0.52%; and occluded silt plus clay = 0.24%).
Thus, the close proximity of fine root detritus to mineral surfaces and aggregates did not appear to have strong effects on short-term incorporation into these components of stable SOM in comparison with leaf litter.Leaching of soluble organic C from decaying leaf litter was the principal mechanism delivering 13 C label to mineral soil (Fahey et al. 2011), and the similarity of our results suggest the same mechanism is probably important for fine root detritus, at least in the initial stages of decay.Notably, however, recovery in mineral soil continued to increase in year 3 for leaf litter (Yavitt et al. 2015) whereas it levelled off for root litter before increasing later in year 6 and 7 (Fig. 2).Finally, the extent of decomposition of sugar maple fine roots after 7 years (8.9% remaining) greatly exceeds the "limit value" of about 20% remaining, measured for sugar maple leaf litter after 6 years of decay (in litter bags; Lovett et al. 2016).Thus, the extent of decay for sugar maple fine roots in mineral soil horizons seems to be greater than for leaf litter in forest floor organic horizons in these acid forest soils.
In conclusion, in contrast with aboveground plant litter, fine root detritus is generated in immediate proximity to soil minerals, thereby maximizing the opportunity for stabilization (Sokol et al. 2019b).Nevertheless, our observations suggest that a relatively small proportion of fine root litter is stabilized in these acid forest soils; after 7 years only about 2% of the 13 C added in fine root litter was recovered in relatively stable SOM fractions.Most of the recovery was in microaggregates, especially those contained within macroaggregate structures, a process that appears to be sensitive to soil pH.About 5% of the added 13 C was recovered in lowdensity POM, some of which could eventually be catabolized or converted to the stable fractions, as observed in the last year of incubation.Therefore, although fine root turnover constitutes a large C input to forest soils, it does not appear be the principal contributor to the large, stabilized SOM stock, and our results support the conclusion of Sokol et al. (2019a) that rhizosphere carbon flux greatly exceeds both aboveground and belowground detritus as a source of temperate forest SOM.
Table 4 Percent recovery (± S.E.) of 13 C label in various soil aggregate fractions 3 years after label addition for this study of sugar maple fine root litter decay in comparison with observations of Yavitt et al. (2015)

Fig. 1 Fig. 2
Fig.1Percent 13 C label recovered of sugar maple fine root detritus through 7 years of decomposition in forest soils in central New York, USA.Values represent mean and standard error for 1st to 4th order fine roots and all four soils used for incubations.Values in year 1 for particulate organic matter fraction are based on in situ core measurements for the same site(Fahey, unpublished)

Table 1
Selected physical and chemical characteristics of four forest soil horizons added to soil cores to which 13 C labeled fine roots of sugar maple were added for incubation in a maple plantation in central New York, USA Vol:. (1234567890)

Table 3
Aggregate fraction mass of four soil horizons after 7 years of field incubations (standard errors in parentheses, n = 8; cores with two different root sizes pooled) in a sugar maple plantation in central New York, USA.Soil horizons as for Table1LDPOM low-density particulate organic matter, < 53 µm is silt plus clay, CPM coarse particulate matter for living maple root 13 C inputs to the same soil horizon LDPOM low-density particulate organic matter, < 53 µm is silt plus clay, CPM coarse particulate matter