Fine root contribution to the soil carbon stock of an agroforestry system in a Caatinga-Atlantic Forest transition zone

The objective of this work was to evaluate the distribution of fine roots and its influence on the soil organic carbon stock, at a depth of 20 cm, in a Grevillea robusta and Coffea arabica agroforestry system. The study was conducted in an agroforestry system established 15 years ago in a transition area of Caatinga and Atlantic Forest biomes in Brazil. G. robusta trees representing the most frequent diameter class were selected, and three distances of these trees (0, 0.75 and 1.50 m) and two soil collection depths (0–10 and 10–20 cm) were defined. The root samples were scanned and quantified using a software program. There was a general predominance of roots with a diameter of 0.6 mm at the shortest distance from the surface layer, while there was a predominance of roots with a diameter of 0.4 mm in the 10–20 cm layer. The root carbon stock at a distance of 0.75 m was higher at a depth of 0–10 cm (0.60 Mg ha-1). The soil organic carbon stock also showed higher results in the 0–10 cm layer compared to the 10–20 cm layer, although with significant variation only in the distance of 1.5 m. There was a higher concentration of fine roots in the topsoil, probably influenced by a greater availability of water and nutrients from plant residues. The soil carbon stock is not closely related to root density or root carbon stock. The data presented in this study do not provide a definitive conclusion.


A B S T R A C T
The objective of this work was to evaluate the distribution of fine roots and its influence on the soil organic carbon stock, at a depth of 20 cm, in a Grevillea robusta and Coffea arabica agroforestry system. The study was conducted in an agroforestry system established 15 years ago in a transition area of Caatinga and Atlantic Forest biomes in Brazil. G. robusta trees representing the most frequent diameter class were selected, and three distances of these trees (0, 0.75 and 1.50 m) and two soil collection depths (0-10 and 10-20 cm) were defined. The root samples were scanned and quantified using a software program. There was a general predominance of roots with a diameter of 0.6 mm at the shortest distance from the surface layer, while there was a predominance of roots with a diameter of 0.4 mm in the 10-20 cm layer. The root carbon stock at a distance of 0.75 m was higher at a depth of 0-10 cm (0.60 Mg ha -1 ). The soil organic carbon stock also showed higher results in the 0-10 cm layer compared to the 10-20 cm layer, although with significant variation only in the distance of 1.5 m. There was a higher concentration of fine roots in the topsoil, probably influenced by a greater availability of water and nutrients from plant residues. The soil carbon stock is not closely related to root density or root carbon stock. The data presented in this study do not provide a definitive conclusion. Keywords: Grevillea robusta; Coffea arabica; root density; root diameter; organic matter.

Fine root contribution to the soil carbon stock of an agroforestry system in a Caatinga-Atlantic Forest transition zone Introduction
Agroforestry systems (AFS) are defined as any land use system that implies the deliberate introduction or maintenance of two or more plant species, where at least one of these is arboreal or other perennial species and is associated with agricultural crops, pasture and/or livestock to exploit the ecological and economic interactions of the different components (NAIR; NAIR, 2014). AFS are known for contributing to mitigating greenhouse gases (STOUT; LAL; MONGER, 2016; VI-CENTE; GAMA-RODRIGUES; GAMA-RODRIGUES, 2016), as they provide environmentally beneficial services and economic advantages to farmers (RODRIGUES et al., 2007).
Interspecific competition for water and nutrients between intercropped crops is a variable that interferes in the productivity of AFS. Tree roots can extend into the crop lines in the active crop competition zone, which occurs up to just over 1 m apart, but mainly in the surface layers (THEVATHASAN; GORDON, 2004). Thus, the choice of deeper-rooted tree species in AFS may reduce vertical competition due to their expansion by layers not exploited by crops, promoting better use of the soil profile (PADOVAN et al., 2015;THOMAS;ISAAC, 2017). On the other hand, even for deep-rooted species, it is observed that a large volume of roots is present in the surface layers, absorbing nutrients from the mineralization of plant residues. According to Defrenet et al. (2016), even though species in AFS exploit different soil niches, coffee roots often dominate the fine root population of the system in surface layers. It is due to higher root biomass in coffee AFS (MEIRELES et al., 2019).
The accumulation of organic carbon in the soil in AFS results from added and decomposing plant residues (from shoots and roots) coming from the different species that compose the system (CHATTERJEE et al., 2018). It is estimated that the soil organic carbon (SOC) stock in AFS differs across regions of the world and at different soil depths, ranging from 30-300 Mg C ha -1 (AGEVI et al., 2017).
Roots in general, and in particular fine roots, constitute an important carbon input into the soil (LIAO et al., 2014). The degree of fine root carbon (FRC) contribution to SOC depends on land use and management system, which determines the root architecture, cycling rate, root exudates and colonization by mycorrhizae (HERTEL; HARTEVELD; LEUSCHNER, 2009;POLLIERER et al., 2012). In addition, root distribution is influenced by edaphic characteristics. For example, Addo-Danso et al. (2020) reported a positive relationship between base saturation and specific root length and specific root area. Fine roots also show greater length in tropical sites where available phosphorus in the soil is low. Le Bissonnais et al. (2018) evaluated the effect of land use gradients (monocultures to AFS and forest) on soil aggregate stability, which was higher in surface layers than deeper layers. Aggregate stability was the main driver of SOC, cation exchange capacity and root traits.
The quantification of root biomass and its distribution in the soil may help to understand the relationship between root dynamics and SOC stock. Although many studies on this topic have already been performed with tree species, most of them are based on quantifying the total root system biomass, and the number of studies measuring the density of fine roots in different diameter classes is limited. In this work, fine roots were found in the surface layer and directly related to the higher concentration of organic matter and nutrients (WITSCHORECK; SCHUM-ACHER; CALDEIRA, 2003). Chatterjee et al. (2020) also reported that the SOC stock in coffee AFS was increased only to a depth of 10 cm after 17 years of establishment due to shade-pruned Erytrina spp.
Given the above, the present study aimed to evaluate the distribution of fine roots and its influence on the SOC stock, at a depth of 20 cm, in a Grevillea robusta and Coffea arabica agroforestry system. Our study is based on the fact that fine root turnover is the dominant form of below-ground carbon input (UPSON; BURGESS, 2013), with a cycle of less than one year (FREITAS; BARROSO; CARNEIRO, 2008).

Study site characterization
The study was carried out in an AFS formed by the G. robusta A. Cunn. ex. R. Br and C. arabica plants planted 15 years ago, spaced 3.5 m (between trees) × 1.5 m (between trees and coffee plants) × 2.5 m (between coffee plants). The area is located in Lucaia District, Planalto municipality, Bahia State (coordinates UTM X: 334277 and Y: 8368812). The AFS was located in an area with pasture naturally formed with predominance of genus Brachiaria grass. Other species were not present in the AFS.
The region has an average altitude of 943 m and a tropical altitude climate (Cwb type according to the Köppen classification), with an average temperature of 19.2ºC and rainfall of 641 mm year -1 (CLI-MATE-DATA, 2012). The study area is located in a transition section between the Caatinga and Atlantic Forest biomes and has soil classified as dystrophic yellow latosol (EMBRAPA, 2013). The chemical characteristics are presented in Table 1.

Fine root and soil collection
We selected six G robusta trees in the most frequent diameter measurements at a height of 1.3 m (DBH) (class center = 27.95 cm) to perform the root and soil collection. The DBH measurement distribution for trees considering an amplitude of 6.14 cm is shown in Figure 1. The selection was carried out in a total area of 1.5 ha with 132 G. robusta tree/ha and 3530 C. arabica plants/ha.
Soil sampling was performed on the G. robusta planting line at 0, 0.75 and 1.50 m from the trunk of each selected tree at depths of 0-10  and 10-20 cm ( Figure 2). Two undisturbed soil samples to evaluate fine roots and soil density and one disturbed sample for SOC determination were taken at each distance and depth, making a total of 108 samples. The disturbed samples were taken using a Dutch auger and the undisturbed samples using a cylindrical ring auger. It was not possible to identify the origin of the roots as to whether they came from trees or coffee.

Soil density
Soil density was determined by the volumetric ring method (EM-BRAPA, 2017) in samples with preserved structure and known volume (7 cm in height and 7 cm in diameter, totaling 269.3 cm 3 volume).

Fine root mass and diameter
The soil samples were placed in plastic containers and then washed with running water and collected on a 0.25 mm sieve to remove the soil mass to determine the mass and diameter of fine roots (KUMAR; UDAWATTA; ANDERSON, 2010). After washing, all roots were manually clamped and arranged on white-bottomed acrylic slides.
The roots of each sample were scanned ( Figure 3) and were distributed in ten root diameter classes (0.45, 0.63 0.81 1.0 1.18, 1.34, 1.52, 1.70, 1.87, 2.05 mm) with the aid of SAFIRA ® software (JORGE; RODRIGUES, 2008) for the different studied distances and depths, according to the method described by Costa et al. (2014).
After scanning, the root samples were placed in aluminum containers, which were then put in a forced-air oven at 65ºC for 72 hours. The samples were subsequently weighed on an analytical balance accurate to 0.001g to determine dry mass.

Fine root density
Root dry mass was used to determine soil root density by means of Equation 1: in which: D = density in g cm -3 ; m = root mass in g; v = ring volume (269.4 cm 3 );

Fine root and soil organic carbon stock
Root (after oven drying) and soil (after air drying and 2.0 mm sieving) samples were macerated in a mortar. Next, 0.02 g of roots and 0.2 g of soil subsamples were removed and submitted to chemical analysis to determine carbon content, using the wet oxidation method with K 2 Cr 2 O 7 in acid medium and titration with ammonium ferrous sulfate (EMBRAPA, 2017).
SOC was calculated on the basis of carbon content and soil density according to Equation 2: SOC = T SOC (g 100 g -1 ) × Ds × Slt (2)   in which: SOC = soil organic carbon stock in Mg ha -1 ; T SOC = total soil organic carbon content; Ds = soil density (g cm -3 ); Slt = soil layer thickness (cm).
FRC was calculated according to Equation 3: in which: FRC = fine root carbon stock in Mg ha -1 ; C FRC = fine root organic carbon content; Dr = root density (g cm -3 ); Slt = soil layer thickness (cm).

Statistical analysis
The root density, SOC stock and FRC values met the parametric criteria and were then submitted to analysis of variance (ANOVA) according to a 3 × 2 factorial scheme with 6 replications (3 distances and 2 depths). Student's t-test at 5% significance was adopted to compare means between distances and depths. The analyses were performed using STATISTICA ® v.10.0 software (StatSoft Inc., 1984-2011. A descriptive frequency analysis was performed for mean root diameter values using the SIGMAPLOT ® v.12.0 software program (Systat Software inc.) and the contour maps were produced using the Surfer ® v.8.0 program, considering a vertical Cartesian plane formed by the spatial distribution of the soil layers and distances of the G. robusta trees.

Fine root diameter
Root diameters ranged from 0.4 to 1.6 mm at the depth of 0-10 cm and from 0.45 to 1.34 mm at the depth of 10-20 cm (Figure 4). Roots in classes of 0.63 mm were more frequent at distances of 0 and 0.75 m and 0.45 mm at 1.5 m.
Overall, there was a predominance of roots with a diameter of 0.6 mm in the distances near the trees at the 0-10 cm depth, while there was a predominance of roots with a diameter of 0.4 mm at the distance of 1.5 m at the 0-10 cm depth, and in all distances of the 10-20 cm depth, which represented 60 to 80% of the total roots.

Root density and fine root and soil organic carbon stocks
The interaction between distance and depth produced a significant effect for the variables of root density and SOC and FRC stocks (Table 2). Significance was only observed for depth when evaluating the isolated effect of the considered factors. The results of the distance × depth interaction are presented in Table 3 and Figure 5.
The FRC and SOC stocks did not vary between the different distances studied. In the case of FRC, differences between depths were only found at a distance of 0.75 m, which showed the highest value in the 0-10 cm layer (Table 3). The SOC stock only showed variation between depths at a distance of 1.5 m, with higher results in the surface layer.
Higher root density was observed in the first soil layer (0-10 cm). However, there was only variation between distances at a depth of 10-20 cm ( Figure 5). Higher values were observed in the distance 0 m, although only with a significant difference at the distance of 0.75 m.

Spatial root distribution influencing soil organic carbon stock
The contours formed by the distribution of root density data and FRC and SOC stocks showed a decrease with increasing depth ( Figure  6A, 6B and 6C). This means that higher values are present in the surface layer, with a slight displacement to positions closer to the G. robusta tree line. It was noted that the distance influenced the root density contours and consequently the FRC stock. Higher root concentration occurred in the 0-10 cm layer and specifically at a distance of 0 and 0.75 m, decreasing vertically and horizontally moving away from this region.
SOC distribution ranged from 30 to 20 Mg ha -1 in depth, and showed higher values according to distance ( Figure 6C) in the surface layer at a distance of 1.5 m. On the other hand, lower SOC stock values were found at a distance of 0.75 m (25 Mg ha -1 ), increasing about 2 Mg ha -1 close to the G. robusta trees. It was also possible to notice a similar distribution pattern at greater depth for all evaluated indicators.

Fine root diameter
The predominance of roots with larger diameter in the 0-10 cm layer (Figure 4) suggests that roots close to G. robusta trees are associated with plant support, while the roots at increasing distance from the tree are more associated with absorption and therefore have smaller diameters. Greater amounts of subsurface roots are also associated with nutrient exploitation from plant waste mineralization (ISAAC; BORDEN, 2019). According to Mora-Garcés (2018), the distribution of fine roots generally decreases with increasing soil depth. The roots can be found in a non-standardized grouping concentrated in cracks or animal pits, showing a large amount of short branches (ZONTA et al., 2006). Fine roots in AFS can also be concentrated in locations with a large agglomeration of plant residues, absorbing nutrients directly from the litter after mineralization (THAKUR; KUMAR; KUNHAMU, 2015).

Root density and fine root carbon and soil organic carbon stocks
The highest FRC values observed in the surface soil layers (Table 3) were expected, as they showed higher root concentration (PADOVAN Table 3 ALBUQUERQUE et al., 2015). This results from the greater contact of the litter (leaves, branches and bark) with the soil, which promotes greater nutrient flow in the surface layer, stimulates the development of proteoid roots and the accumulation of carbon in the soil after its turnover (PULROLNIK et al., 2009). Morais et al. (2017) found that carbon stored in fine roots is more concentrated in the topsoil (0-10 cm). The authors also observed that fine roots store 40% more carbon than thick and medium roots in this layer. The absence of a difference in FRC stock at distances of 0 and 1.5 m (Table 3) suggested unevenness in G. robusta root development. Thus, it is likely that horizontal variability in fine root distribution is being more influenced by soil resource availability than by root architecture of the species in question. In studying the distribution of G. robusta roots in AFS, Smith et al. (1999) observed great unevenness in the distribution of fine roots and argued that the root distribution complementarity of the different components in the AFS may be com-promised by restrictions in the availability of water and nutrients for the tree component, which results in increased competition (BALJIT; PARAMPARDEEP; GILL, 2016).

-Fine root carbon (FRC) and soil organic carbon (SOC) stocks as a function of different distances and depths in a Coffea arabica and
The reduction in root concentration with increasing depth (Figure 5) can be attributed to the large amount of litter found in AFSs, which contributes to the development of fine roots in the topsoil and in the organic layer itself. Similar results were obtained by Defrenet et al. (2016) in evaluating the biomass and root dynamics of AFSs based on coffee planted in Costa Rica, finding higher amounts of fine roots in surface soil (12% of total roots). Fine root biomass was also twofold higher in the row compared with between rows.
The litter acts as a mulch, protecting the surface soil and providing nutrients. Freitas, Barroso and Carneiro (2008) point out that the growth of fine roots (≤ 2 mm) has a strong correlation with the availability of organic matter and soil moisture, being closely associated with litter, since it is a carbon source and favors water retention.

Spatial root distribution influencing soil organic carbon stock
The results obtained did not allow us to determine the origin of the evaluated roots (of C. arabica or G. robusta), which would indicate which species would predominantly be contributing to the carbon accumulation in the soil, since the roots of C. arabica can develop in the middle of the G. robusta lines and vice versa. Some mechanisms (still little known) may alter the root growth of the intercropped crop in an AFS. For example, Livsley, Gregory and Buresh (2000) reported that corn crops showed a greater amount of fine roots and root length in an AFS with Grevillea sp. when compared to a monoculture. A similar pattern was observed by Duan et al. (2019) for oat roots, which were influenced by the presence of walnut, which caused increased root length and decreased root diameter.
Regardless of the soil carbon origin, considering the components present in the AFS, it was observed that the FRC stock had little influence on the SOC stock. On the one hand, the SOC stock did not follow a similar distribution as FRC stock in the 0-10 cm layer, on the other hand there was a high correlation between SOC stock, root density and FRC stock in the 10-20 cm layer ( Figure 6). This indicated that the carbon accumulation in the 0-10 cm layer depended on more litter contribution than only the carbon originating from the fine root turnover, as in natural systems of Brazilian biomes (OZÓRIO et al., 2019). The influence of the FRC stock in the 10-20 cm layer was high due to a decreasing carbon incorporation rate from the surface to the deeper layer (CHATTERJEE et al., 2020). AFS are known to have complex re-lations between species which results in a heterogeneous environment, especially in the soil-plant transition.
AFSs also help to maintain the natural physical properties of the soil, especially because soil tillage is usually only done in pre-planting (FALCÃO et al., 2020). The conservationist character of AFS assists in natural root turnover, without harming the soil carbon accumulation.

Conclusion
There is a higher concentration of fine roots in the topsoil which decreases with increasing depth. The root density shows a homogeneous horizontal distribution from the base of G. robusta, probably being more influenced by litter and edaphic characteristics than by the fine roots. The SOC stock is not closely related to root density or root carbon stock. The data presented in this study do not provide a definitive conclusion. Thus, more investigations focusing on identifying the fine root origin of the different species in the AFS are necessary, including deeper layers (up to 100 cm) and evaluating other edaphic characteristics.