4.1. C Stocks in Tree Biomass and Soil
Based on our observation of the forest structure and composition during inventories (
Table 1), the Andean PDF was rather similar to the UNF, which could explain why carbon stocks per hectare in aboveground tree biomass were not significantly different. PDF in the Chilean Andes tended to have been subjected to less human-intervention than in the Coastal mountains, coinciding with the lower demographic pressure in the surrounding area. However, DEF in both locations (especially in the Coastal mountains) have been more severely subjected to interventions over the last five decades, mostly by extensive cattle grazing and unsustainable exploitation of the wood resources, resulting in small trees and low stand density (i.e., few tree stems per hectare). This is likely why the C stock in tree biomass is so low in Andean DEF and even less in the Coastal range.
With respect to the larger C stock in Coastal UNF compared to Andean UNF (60.2 vs. 49.5 Mg·C·ha
−1 in 2013, and 77.2 vs. 63.7 Mg·C·ha
−1 in 2015), the differences can be attributed to the higher tree density (50% of all trees per hectare are
Nothofagus obliqua vs. 35% in the Andean UNF), which compensates for the larger DBH and HT of trees in the Andean UNF. In a previous investigation [
25], researchers working with a 150-year-old unmanaged second-growth
Nothofagus pumilio forest on an Andisol soil in the Chilean Patagonia reported 229 Mg·C·ha
−1 for aboveground tree biomass, which greatly exceeds the values reported in this study for
Nothofagus obliqua. One possible explanation is the totally different stand characteristics of these two
Nothofagus species. In the Patagonian study, tree density exceeded 1000 stems·ha
−1, which is more than twice the densities measured in the Andean and Coastal forests. Additionally, trees in Patagonia belonged to six diameter classes with most of the individuals pertaining to the 30–50 cm classes. In contrast, trees in Andean and Coastal UNF belonged to 13 and 8 different diameter classes, respectively, with most of the individuals pertaining to the 5–25 cm classes. Finally, trees in Patagonia were 10% and 35% taller than those in Andean and Coastal UNF, respectively.
When making comparisons of the C stocks in tree biomass between 2013 and 2015 at both locations, one may perceive how the gains were affected by the different degradation levels (i.e., the more disturbed the forest, the smaller and less vigorous are the trees, which resulted in lower increases of annual C stocks). This was clearly observed in the inventory that was performed in 2015, as compared with that of 2014, where we found the lowest tree DBH and HT increases, as a result of a severe six-month drought in the 2015 summer in south-central Chile. For example, DBH increments were up to four and two times lower in DEF and UNF at both sites, respectively, vs. those measured in 2014. Furthermore, there was no difference in height growth increment in DEF in the 2014–2015 vs. 2013–2014 periods, but the height growth increment was 25% lower in UNF when comparing the same periods. On the other hand, trees with DBH < 10 cm were the most affected, especially in DEF, and many of them were without almost any increment of growth in the last year. Tree mortality was also by far the highest in DEF, per the results of the third inventory.
In DEF of both locations, especially in the Coastal range, we observed an important variation of soil C concentrations and bulk density across depths because these soils have been highly subjected to interventions over the years. This may explain why their C stocks are slightly higher—although not significantly so—than those of PDF, with the same trend in both sites. The lower C stock variability in the Coastal PDF, compared to the C stock in DEF may explain why it is statistically lower than that of UNF.
As stated by others [
34], soil C stock calculation depends on variables such as C concentration, horizon thickness, and bulk density, which all have their own variances and errors. In temperate forests dominated by
Nothofagus pumilio in southern Chile, [
25,
35] the reported SOC stocks were 100 and 102 Mg·ha
−1, respectively, which were measured in Andisols at 0–40 cm depth, whereas others [
36] reported 132 Mg·C·ha
−1 in Andisols at 0–30 cm depth under a second-growth
N. obliqua forest. While the latter value is more similar to the results obtained in this study in UNF, the reported value of 100 Mg·C·ha
−1 reported by [
25] corresponds to a forest subjected to moderate interventions, mostly from light thinning and cattle grazing in the summer months and may therefore be compared to the Andean PDF, with a soil organic C stock in a similar range (119 Mg·C·ha
−1 in 2013).
Regarding non-volcanic soils, Stolpe et al. [
3] reported substantially lower organic C stocks in an Alfisol soil under highly degraded and typical
Acacia caven forests (Espinales) in a semi-arid region of central Chile. In their study, the carbon stock at 0–40 cm depth was 17.5 Mg·C·ha
−1 (10%–25% forest cover) and 25.2 Mg·C·ha
−1 (26%–50% cover) in the degraded and typical conditions, respectively. These results show that the proportion of forest cover is highly related to site degradation, which has a direct effect on the formation of soil organic matter. SOC stocks can vary substantially between ecosystems and depend not only on the soil type, but also on the prevalent climate, management, and vegetation [
37], which in turn will affect the quality of plant residues that fall on the ground and the microbial biomass capacity to decompose and incorporate it into the SOM [
25].
Although the Coastal DEF is substantially more degraded than the Andean one, as is evidenced by smaller trees and a deficient forest structure and composition as previously discussed, it is interesting to see that its absolute and relative soil C stocks remain slightly higher than in the degraded Andean soil, even more so considering that the Coastal soil is an Ultisol (non-volcanic soil) that is generally characterized by a lower stock of organic C as part of the SOM, compared to Andisols. That the Coastal DEF soil is more degraded than the Andean one, even though it has more total C, is reflected by lower C concentrations in the macro-, meso-, and microaggregate fractions at 20–40 cm depth in both measurement periods, and in the macro- and mesoaggregate fractions at 0–20 cm depth in 2015 (
Table 3 and
Table 4). A previous investigation [
25] which compared degraded pastures with
Nothofagus pumilio forests in southern Chile reported similar trends. Undoubtedly, all the numbers presented in
Section 3.1 show the strong effect of site disturbance and the resulting degradation over the years, as well as the factor of geographical location (and the corresponding climate in the Coastal mountains, with generally lower annual precipitations) on C storage potential in native ecosystems.
When comparing SOC stocks between 2013 and 2015 in the different forest conditions at both locations, a possible explanation for the slightly lower values in 2015, although not significant, could be due to the different sampling points where soil samples were extracted between the years. This does not happen when measuring C gains in the aboveground tree biomass because the same individuals are monitored over the years. In 2013, the higher C concentrations in the soil samples (as reflected in the LF and macroaggregates shown in
Table 3 and
Table 4) may also be attributed to soil heterogeneity and the presence of C-rich pockets in the surveyed sectors [
1,
38].
4.2. Soil Bulk Density
The more degraded the forests, as caused by higher intervention by humans and grazing by cattle, the more they are usually accompanied by high levels of soil compaction. As a result of increasing soil compaction, bulk density increases not only at the surface but also with increasing soil depth. Bulk density is also affected by soil genesis, whereby the Andean soils tend to have a lower bulk density than those of the Coastal mountains because of their volcanic origin which also favors higher contents of organic matter. The lower bulk density is also accompanied by higher amounts of C and N that are sequestered in volcanic soils, with the result that the soils are therefore somewhat protected from mineralization and leaching, and other possible losses from the system [
39].
The discrepancies that were observed in PDF could be attributed to the difficulty in finding the “most representative” of the condition in the sense that partial degradation is an intermediate state that is sometimes more similar to full degradation and sometimes more similar to non-degradation states. Also, there is very limited quantitative characterization or metrics that are known that permit a clear definition of this intermediate condition. According to [
40], the relation between SOC and the different states of forests remain uncertain. Additionally, given the small variations that may occur in forest C stocks [
41], it can be difficult to establish threshold values in order to define degradation levels.
The abnormally high bulk density at 0–5 cm depth in Andean DEF (0.96 g·cm
−3) and at 20–40 cm depth in PDF (1.22 g·cm
−3) could be attributed to the presence of horses being grazed throughout the year since plot fencing was established just a month before the period of soil sampling. Ongoing presence of cattle and sheep has also been observed in Coastal DEF and PDF. New measurements of soil bulk density should be performed only in the exclusion plots in order to test this hypothesis. With respect to UNF at both locations, the low soil bulk density at 0–40 cm depth can probably be explained by the larger amount of roots found in this condition which, according to [
42] who studied their effect on several soil parameters in temperate woodlands, contribute to lower the bulk density by action of root penetration across the soil profile.
Panichini et al. reported bulk densities of 0.54 and 0.74 g·cm
−3 at 0–19 cm and 19–41 cm depths, respectively, in temperate
Nothofagus pumilio rainforests on a Chilean Andisol, whereas the values of our study in the Andean UNF ranged from 0.78 to 0.86 g·cm
−3 at 0–20 cm and 20–40 cm depths, respectively [
35]. In northern Argentine Patagonia, Candan and Broquen [
43] reported an average bulk density of 0.72 g·cm
−3 in the upper 5 cm of the A-horizon of an Andisol under an unmanaged, almost pristine
Nothofagus sp. mixed forest, which is highly similar to the value obtained in Andean UNF at 0–5 cm depth. On the other hand, Dorner et al. [
44] found bulk densities ranging from 0.8 to 1.1 g·cm
−3 in a Chilean prairie (
Lolium multiflorum-Avena strigosa) Ultisol without major disturbances, while the bulk density values obtained in this study were 0.84, 0.98, and 1.05 g·cm
−3 in Coastal UNF, DEF, and PDF, respectively, at 0–20 cm depth.
4.3. C Concentrations in Soil Fractions
As observed in
Table 3 and
Table 4, the C concentrations within the LF (un-decomposed organic matter of plant origin) were significantly higher (
p < 0.05) in the Andean UNF than in DEF at 0–20 cm depth in the first measurement period, and higher in the Coastal UNF than in DEF at all depths in both measurement periods, although no significant differences were found between the two years (except at 20–40 cm depth in the Coastal DEF). This lack of difference is attributed to the relatively short time span, which is not sufficient for any significant changes of SOC in this fraction. This is especially true in the Andes where the level of degradation is lighter, which implies that changes may be more difficult to observe. Also, the different physicochemical properties of Andisols, such as the presence of amorphous clay (allophane or imogolite) having a high specific surface area and a pH-dependent charge, large amounts of humus, high content of water at 1500 kPa tension, high phosphate retention, and slow decomposition of incorporated organic materials may also help to provide an explanation for slow changes in SOC [
45]. On the other hand, UNF have a larger input of new plant debris (not decomposed) throughout the year than DEF, with a higher diversity and number of species, and a larger litterfall (i.e., constant input of leaves, cones, fruits, and twigs). The LF is a coarse and unstable organic matter, whereby its content in soil tends to be affected by rapid changes (i.e., influenced by the type of management performed) [
3,
37]. Additionally, according to [
46], the LF is highly sensitive to present forest management but not to historical ones. The carbon in mineral dominated aggregates, however, is part of older humic substances and therefore more likely a “historical remnant”. This helps understand why Andean and Coastal forests that have not been subjected to interventions (or managed) between 2013 and 2015 have similar C concentrations within their LF at almost all depths.
The fact that C concentrations in macroaggregates (>212 µm) were higher in Andean and Coastal UNF than DEF at 0–40 cm depth in both measurement periods shows a condition of better management over the years. The same trends were observed for meso- and microaggregates, although the differences were not always significant, showing that the Andean UNF sequester the largest amount of C in its macro-, meso-, and microaggregates. The most labile fraction (>212 µm) is an indicator of sustainable management because its content in the soil changes in the short term (various months to a few years), depending on vegetation, type of management, and equilibrium between plant residue input and subsequent decomposition in the soil [
3,
47]. However, the reasons why C concentrations in macroaggregates at 0–20 cm depth are significantly lower in 2015 than 2013 remain unclear and cannot be fully explained at this point, even more so considering that sampling plots were fenced at the beginning of the study in order to prevent animal intrusion and any other associated site disturbance, and that this kind of trend did not occur in the other soil fractions. As hypothesized in
Section 4.1, however, a specific explanation based on findings of [
1,
38] might be that the presence of C-rich pockets could have been inadvertently included in the first set of sampling points in the surveyed sectors, but absent in the second set. More research is needed to test this hypothesis.
The significantly larger C concentrations in mesoaggregates (212–53 µm) at 0–40 cm depth in UNF vs. DEF in 2015 indicate that these have the best soil quality, which could be attributed to the minimum anthropogenic intervention over the last 50 years [
25]. The better quality of organic matter found in UNF at both locations is also reflected by greater presence of N-NH
4 and available K and S [
48]. The quality of soil organic matter is essential in order to maintain productivity of the forest and assure its long-term sustainability. Soil organic matter increases also allow the sequestration of surplus atmospheric CO
2.
Our results for the LF were in the same range as those obtained for macroaggregates by [
25] who performed physical fractionation of an Andisol in a
Nothofagus pumilio forest. In their study, the LF was not measured independently like in this work, which impedes a direct comparison with organic C concentrations of the respective macroaggregates. Similar ranges of C concentrations were also found at all depths in the meso- and microaggregate fractions.
The observation that organic C concentrations were higher in micro- than in mesoaggregates, and significantly higher than in macroaggregates in all levels of degradation at both sites, is consistent with a study performed by [
43] in a
Nothofagus sp. mixed forest. These authors reported that the smallest aggregates (<50 µm) at 0–5 cm depth in an Andisol had the highest SOC concentrations comparing with the LF, the 2000–250 and the 250–50-µm size fractions. Additionally, they found that the LF (8000–2000 µm) contained the second largest amount of organic C, which is also in agreement with our study. Other works (e.g., [
49]) have confirmed that large amounts of SOM are required for the formation of bigger aggregates.
Regarding the changes of organic C concentration through the soil profiles from 0 to 40 cm depth, decreases were found in every level of degradation and location, as expected, with the highest values in the first 5 cm (
Table 3 and
Table 4). Similar but more pronounced decreases were observed for the N concentrations [
50]. As a result, the C/N ratios increased with depth, which indicated greater SOM stability with soil depth.