Chemical, physical, and physicochemical soil attributes and their relationship with soil organic matter
The pH values ranged from 4.4 to 5.4 for H2O, 4.1 to 5.1 for KCl solution (1 mol L-1), 3.9 to 5.1 for CaCl2 solution (0.1 mol L-1) and 4.8 to 5.5 for SMP solution, decreasing in subsurface layers (Table 1). The pH values, regardless of the solution used, were generally lower in T and agroforests with younger ages (SAF1 and SAF3) compared to the seven-year-old agroforest (SAF7) and the reference system (Forest) (Table 1). This observation suggests that agroforestry practices influence soil pH. This effect is likely due to the replacement of chemical structures able to neutralize the acidity components (complexation of Al+3 by carboxylic and hydroxyl organic groups) and alkaline oxidized species that have the ability to neutralize H+ (such as CaCO3, CaO, MgCO3, and MgO) through litter decomposition. The production of organic acids, which can complex with Al+3 - slow down the hydrolysis of water and subsequent release of H+ ions into the environment - is related to aerobic decomposition conditions (Mikutta et al., 2007; MELO e WYPYCH, 2009), while alkaline oxides are commonly stored in cell vacuoles and released through cell lysis (Nelson e Cox, 2002).
The exchangeable base contents (Ca+2, Mg+2, K+, and Na+) and values of SB, CEC, and V% increased in the following order: Control (T) < SAF1 < SAF3 < SAF7 < Forest (reference), generally being higher in surface than in subsurface (Table 2 and 3). This effect is probably related to alkaline cations replacement through the input of litter, which tends to increase in soil surface over time. The increase V% in agroforestry systems indicates the potential of this intercropped production approach to enhance the availability of cationic nutrients for plants. In contrast, P content (Mehlich-1) was quite similar among the T, SAF1, and SAF3 samples, showing a slight increase in SAF7 and a significant increase in Forest. However, the observed trend for P differed from that of exchangeable bases, yet it reinforces the evidence that agroforestry practices have a significant effect on nutrient replenishment within the production system. Similar results were observed by other authors (Alfaia et al., 2004; Menezes et al., 2008; Boley et al., 2009). Possibly, the increase in interactions between plants and mycorrhizal fungi, as well as the expected greater population of phosphate-solubilizing microorganisms, over time are responsible for the adequate nutrition of plants during the development of the systems (Dobo et al., 2018).
The nutritional levels increase is associated with the enhanced mobilization and availability of elements due to the input of litter into the system (Guo et al., 2018; Oliveira et al., 2020). Over time, as mixed systems evolve, there is an increase in the amount of plant material deposited on the soil surface. As showed in Hanke et al (2023a), the calculated litter deposition in the different evaluated scenarios was approximately: T (1.84 kg biomass m-2 year-1); SAF1 (1.43 kg biomass m-2 year-1); SAF3 (2.27 kg biomass m-2 year-1); SAF7 (2.86 kg biomass m-2 year-1); and Forest (2.91 kg biomass m-2 year-1), supporting the observed increase in nutritional levels. Although root biomass was not quantified, a proportional increase in the incorporation of this material beneath the soil surface is expected, which could contribute to nutrient replenishment through its cycling. This material, which contains immobilized nutrients in biomass, undergoes progressive decomposition, leading to the release of these elements
Simultaneously, as tree plants grow and the system's complexity increases, there is also a progressive enhancement in the richness and diversity of multifunctional soil organismo. This microrganisms are responsible for decomposition and chemical elements remobilization in soil profile (Odum, 1988; Gliessmann, 2007; Notaro et al., 2014; Rocha et al., 2022). Examples of these ecological interactions include mycorrhizae, symbioses, and other types of mutualistic ecological relationships (Notaro et al., 2014; Dobo et al., 2019). Therefore, there is a greater potential for nutrient utilization, as well as a relative decrease in the loss of these elements from the system, resulting in an increase in their levels over time.
Significant positive correlations were observed between the carbon (C) content and the variables SB, effective CEC, potential CEC, and V%, as well as a significant negative correlation between C and Al+3 saturation (m%) (Figure 1). These correlations confirm that, in addition to the greater nutrient quantity promoted by the higher input of organic material in systems, SOM can contribute to retention of exchangeable bases, an increase in the density of exchange sites and a decrease in Al+3 saturation.
As expected, a behavior similar to that observed previously was also verified for exchange sites saturation by Al+3 (m%), which decreased in the following order: Control (T) < SAF 1 < SAF 3 < SAF 7 < Forest (Table X). This suggests that Agroforestry system longer succession time is capable to decrease the toxic effect of Al+3, coinciding with the increase in nutrient saturation of coloids exchange sites (Table X). The increase in exchangeable H+ levels observed (Table 2) can be explained by the following points: i) higher production of organic acids (carboxylic and hydroxyl), originating from biomass decomposition (litter); and ii) higher mineral hydrolysis, caused by electrolytic activity in solution, which can substantially increase mineral decomposition rates (Heckmann et al., 2013; Hanke et al., 2015). Similarly, the presence of Al+3 in systems with higher levels of C and N may imply that this ion is, in part, associated with SOM through the formation process of soluble metal-organic complexes "Al-SOM." Some of these complexes are extractable with a KCl solution (1 mol L-1) and can be determined, together with free Al+3, using flame atomic absorption spectrophotometry. Thus, the seemingly higher Al+3 levels under conditions of higher SOM content do not necessarily imply in its availability increase. It is due to the fact that “Al-SOM” complexes are virtually unfit for root system absorption, depending on the molecular size and reactivity of their chemical structure (Hanke e Dick, 2017a).
Over time a decrease in soil bulk density (Ds) and particle density (Dp) was also observed, along to increase in total porosity (Pt) and soil moisture (Table 4). This trend is possibily related to SOM content increase, as evidenced by the significant correlations between C and this set of physical soil variables (Figure 2). SOM can lead to a reduction in Ds and Dp and an increase in Pt and soil moisture due to its composition, consisting of low molecular weight structures (in comparison to minerals, for example), and its high density of structural pores. The SOM pores and surface charges can interact and compartmentalize the soil solution (Stumm, 1992; Dick et al., 2009; Hanke & Dick, 2017b).
Exchangeable bases stocks and its soil accumulation kinetics
As a substantial decrease in bulk density (Ds) was observed over time, stock data for alkaline bases, as well as Al+3, were calculated. In general, despite the Ds decrease, the increases in Ca+2, Mg+2, K+, and Na+ levels were sufficiently significant to lead to an increase in the stocks of these macronutrients over time. The stock values for Ca+2, Mg+2, K+, and Na+ ranged from 23.8 to 148.1 g m-2; from 6.4 to 35.3 g m-2; from 2.1 to 30.9 g m-2; and from 0 (below the detection limit of the method) to 1.2 g m-2, respectively (Table 5). Furthermore, the calculated stocks for the sum of these alkaline bases (SB) ranged from 31.9 to 204.1 g mm-2, and as expected, showed the same trend observed for individual elements (Table 5). These results confirm that the adoption of successional agroforestry practices tends to significantly increase the stocks of these alkaline bases in the soil, based on the development time of the intercropped agroforestry strata.
The increase in nutrientes stocks over time is attributed to increase in SOM levels, as evidenced by the significant positive regressions between C stocks and SB (Figure 3). This regression model (Y = 13.3178 + 0.0486X) shows that an increase of 20.6 g m-2 of soil organic C in over time corresponds to an increase of 1 g m-2 of the exchangeable bases (approximately: 0,68 g m-2 of Ca+2, 0,18 g m-2 of Mg+2, 0,14 g m-2 of K+ e 0,03 g m-2 of Na+ (Figure 3). The increase in SOM stocks leads to an expansion of electric charges produced by the deprotonation/protonation of chemical groups (carboxyls, phenolic hydroxyls, and amines, primarily). The electric charges formation controls the ion retention through sorption mechanisms (outer-sphere and inner-sphere complexes by the ligand exchange mechanism) (Stumm, 1992; Dick et al., 2009; Hanke & Dick, 2017a; Hanke & Dick, 2019). Additionally, in SOM structure there physical compartments that can store compounds with similar electrochemical behavior, potentially containing nutrients in a free solvated form or even in complexed forms (Dick et al., 2009). The linear relationship verified between C content with effective and potential CEC corroborates the positive effect of SOM on the nutrient retention in soil (Figure 1).
The Ca+2, Mg+2, K+, P stocks, and SB accumulated in soil also generally showed an increasing trend in the following order: Control (T) < SAF1 < SAF3 < SAF7 < Forest (Figure 4). For Ca+2, Mg+2, K+, and SB, a clear separation into two groups was observed for the accumulated stocks at different depths in the profile: i) a group consisting of T, SAF1, and SAF3; and ii) a group consisting of SAF7 and Forest. For P (Mehlich-1), although a similar behavior to the attributes and parameters presented earlier was observed, a distinct separation was noted among all treatments, with the stocks progressively increasing over time (T < SAF1 < SAF3 < SAF7 < Forest) (Figure 4).
Although the percentage of total stocks of Ca+2, Mg+2, P, and SB contained in the 0-20 cm layers was lower than that observed in 20-50 cm (Figure 5), it was verified that for K+, the surface stocks proportion is substantially higher than that verified in depth. This fact is probably related to the faster cycling of K+ in soil-litter interface. K+ does not make part molecular structures in plant metabolism, making it easier to become available in the early stages of plant decomposition. This rapid release implies two conditions: i) faster absorption by plants and organisms after its release into the soil, and ii) removal from soil through leaching, as this ion has low affinity for active sites on minerals and SOM according to lyotropic series, leading to lower retention and higher release into the solution. On the other hand, Ca+2 and Mg+2 are part of various structural compounds from primary and secondary plant metabolism (Nelson e Cox, 2002), and their availability is slower when compared to K+, reducing their relative saturation in surface layer (0-20 cm).
The Al+3 ranged from 15.4 to 227.9 g m-2 decreasing over time, possibly related to SOM complexing effect on this ion. This hypothesis is supported by the significant positive correlation observed between C content and the saturation of exchange sites with Al+3 (m%) (Figure X). SOM can complex Al+3 through its carboxylic groups, leading to the immobilization of this element (Hanke et al., 2015; Hanke e Dick, 2017a). Thus, SOM can be considered an important factor in controlling potential soil acidity components. In according to nutrients discussed earlier, a higher proportion of Al+3 stocks was observed in the subsurface (Figure 5).
To investigate the accumulation of Ca+2, Mg+2, K+, and SB stocks over time, the values founded were used in relation to the years of systems evolution. Thus, the following conditions and values were considered: i) T as the starting point (0 years); ii) SAF1 (1 year); iii) SAF3 (3 years); iv) SAF7 (7 years), and Mata (30 years). The stocks values were analyzed in surface (0-20 cm) and subsurface layer (20-50 cm), where these values were plotted (variable y) against time (variable x). The mathematical model used was the first-order exponential, described by: f(x) = y0 + a(1-e-bx). This exponential model assumes the elements accumulation rate is decreasing and their saturation tends toward a maximum, represented by the sum of the parameter "a" with the intercept (starting point - y0) of function when the system's evolution time tends towards infinity. The maximum time for system saturation (maximum stocks) was assumed as 99% of the value of "y0 + a." Figure 6 shows the behavior of the stocks of the main alkaline bases and their sum (SB) in soil samples.
The application of exponential model (f(x) = y0 + a(1-e-bx)) was significant for both the surface and subsurface layers, except for K+ - not significant for the 20-50 cm layer - and for P(Mehlich-1) - not significant for the 0-20 cm layer (Figure 6). Possibly, for these two elements, the accumulation dynamics differ from those observed for the other exchangeable bases, potentially being controlled by factors that were not possible to evaluate in this study (Figure 6).
Table 6 shows the model parameters, the range of variation in nutrient stocks during ecological succession and the time required to reach maximum stocks. As observed, the variation in the stocks of these elements - between the initial and maximum values - changed over time as follows: i) Ca+2 - 54% in 0-20 cm and 47% in 20-50 cm; ii) Mg+2 - 53% in 0-20 cm and 50% in 20-50 cm; iii) K+ - 77% in 0-20 cm; iv) SB - 57% in 0-20 cm and 48% in 20-50 cm; and v) P(Mehlich-1) - 22% in 20-50 cm (Table 6). This result emphasize the ecological succession's potential to increase nutrient retention and availability in the system. In turn, the time required to reach maximum stocks for these ions/parameters were: i) Ca+2 - 31 years in 0-20 cm and 33 years in 20-50 cm; ii) Mg+2 - 24 years in 0-20 cm and 26 years in 20-50 cm; iii) K+ - 22 years in 0-20 cm; iv) SB - 26 years in 0-20 cm and 31 years in 20-50 cm; and v) P(Mehlich-1) - 17 years in 20-50 cm (Table 6).
Differentiation of the analyzed systems through the application of Principal Component Analysis (PCA)
The PCA explained 76.0% of total variance distributed in the first two principal components (PCs), with PC1 (x-axis of the diagram) explaining a higher data variance proportion (50.2%) and PC2 (y-axis of the diagram) explaining 25.8% of the variance (Figure 7). The remaining 14.0% of the variance is explained by other aspects (soil and environmental attributes) that were not focused in this study or have little relevance to the analyzed variables. The dispersion pattern of the PCA remained stable when subjected to the "bootstrap" resampling test – considering ten thousand iterations with replacement and comparing the "bootstrap" sample to the pattern of the original sample (p < 0.05). Thus, the PCA ordering pattern was stable and significant, making it suitable for analyzing the sample units and soil variables.
Firstly, a clear separation between samples from the surface (quadrants 1 and 4) and subsurface layers (quadrants 2 and 3) can be observed in a highly similar organizational pattern in the diagram (Figure 7). The most influential variables for the description of PC 1 (x-axis of the diagram), that explains the higher proportion of data variance, were: Ca+2, Mg+2, K+, P, Al+3, SB, carboxylic groups, aromatic groups, SOM nitrogen compounds (extracted from Hanke et al., 2023b), pH KCl, pH CaCl2, pH SMP, V%, m%, Ds, Dp, Pt, Uv, % of C and N in SOM particulate fraction (>0.053 mm) and C/N ratio of SOM particulate (extracted from Hanke et al., 2023a). PCA revealed a gradient pattern among the evaluated conditions, ranging from the control (T) and SAF1 to the intermediate agroforestry condition (SAF3) and the more advanced systems (SAF7 and Forest). This highlights that, depending on the time of agroforestry system adoption, changes in soil attributes occur. These changes tend to move towards the reference system (forest) direction (Figure 7).
As previously mentioned, the increase in pH values, exchangeable base contents (variables Ca+2, Mg+2, K+, Na+, P Mehlich-1, CEC, and SB), and attributes of SOM occurs progressively from the T condition to the reference system (forest), with the chronological stages of agroforests in a transitional position between the extremes. This variation is generally more pronounced and evident in the layers closer to the surface than depth of all situations analyzed in the scatter plot (left of CP1 relative to the scatter plot origin). This is likely due to the higher dependence of subsurface environments in relation to anisotropic pedogenic processes that occur at the landscape level, whereas surface layers exhibit different interactions dynamics (exchange of energy and matter) with neighboring systems, especially with vegetation and litter.
The positive correlation between particulate C and N with exchangeable-base content in relation to PC1 (Figure 7) indicates that the replenishment of nutrients in ager systems with higher biomass input occurs through the deposition of organic materials. A consequence of this fact is that if there are changes in particulate SOM input rates, nutrient supplies will be altered, potentially limiting plant development.
The most important variables for describing CP2 (y-axis of the scatter plot), which explained for 25.8% of the total variance, were: C and N stocks, relative intensity of aliphatic groups (C-alkyl) and carbohydrate-type structures (C-O-alkyl) (extracted from Hanke et al., 2023 a,b), gravimetric and volumetric moisture, effective and potential CEC, H+ and pH in water (Figure 7). Concerning this ordination axis, the sample units positioned above the origin tend to exhibit higher SOM stocks with a more aliphatic composition and a lower proportion of carbohydrates, lower active acidity, higher cation exchange and retention capacity, as well as gravimetric moisture. As observed for CP1, a similar transitional gradient among sample units was also noticed for CP2, ranging from the initial system (T), agroforests with younger ages (SAF1 and SAF3), to more advanced succession systems (SAF7 and Forest). As previously discussed, the increase in C and N stocks with the advancement of system development time leads to an increase in the soil's water hold capacity, as well as an increase in ion exchange capacity.
Due to the positioning of samples from the reference system (Forest - surface layer) with higher positive values in relation to CP1 and CP2, it is emphasized that the increase in SOM stocks can lead to a series of improvements in soil chemical and physical fertility. By applying the minimum spanning tree algorithm (set of edges connecting sample units), it was possible to observe that a progressive transition occurs through the set of similarities among samples, starting (left side of the diagram) from subsurface samples to the surface layers of the initial systems (T and SAF1), progressing to the surface layers of intermediate systems (SAF3 and SAF7), and finally ending at the reference system's surface. This transition is driven by the soil's chemical, physical, and physicochemical variables, influenced by the SOM accumulation dynamics. In summary, the adoption of production practices that result in high input of C and N can progressively beneficially alter soil fertility-related attributes.