Soil organic carbon along an altitudinal gradient in the Despeñaperros Natural Park , southern Spain

Soil organic carbon (SOC) is extremely important in the global carbon (C) cycle as C sequestration in non-disturbed soil ecosystems can be a C sink and mitigate greenhouse-gas-driven climate change. Soil organic carbon changes in space and time are relevant to understand the soil system and its role in the C cycle. This is why the influence of topographic position on SOC should be studied. Seven topographic positions from a toposequence between 607 and 1168 m were analyzed in the Despeñaperros Natural Park (Jaén, SW Spain). Depending on soil depth, one to three control sections (0–25, 25–50 and 75 cm) were sampled at each site. The SOC content in studied soils was below 30 g kg and strongly decreases with depth. These results were related to the gravel content and to the bulk density. The SOC content from the topsoil (0–25 cm) varied largely through the altitudinal gradient ranging between 27.3 and 39.9 g kg. The SOC stock (SOCS) varied between 53.8 and 158.0 Mg ha in the studied area, which had been clearly conditioned by the topographic position. Therefore, results suggest that elevation should be included in SOCS models and estimations at local and regional scales.


Introduction
Soils are an important carbon reservoir (Barua and Haque, 2013;Yan-Gui et al., 2013).In fact, the primary terrestrial pool of organic carbon (OC) is soil, which accounts for more than 71 % of the Earth's terrestrial OC pool (Lal, 2010).In addition, soils have the ability to store C for a long time (over the last 5000 years) (Brevik and Homburg, 2004).Soils play a crucial role in the overall C cycle, and small changes in the soil organic carbon stock (SOCS) could significantly affect atmospheric carbon dioxide (CO 2 ) concentrations, and through that global climate change.Within the C cycle, soils can be a source of greenhouse gases through CO 2 and methane (CH 4 ) emissions, or can be a sink for atmospheric CO 2 through C sequestration in soil organic matter (OM) (Breuning-Madsen et al., 2009;Brevik, 2012).
Climate, soil use and soil management affect OC variability, particularly in soils under Mediterranean type of climate, characterized by low OC content, weak structure and readily degradable soils (Hernanz et al., 2002).In temperate climates, recent studies show differences in C sequestration rates in soils depending on use and management (Muñoz-Rojas et al., 2012a, b), climate and mineralogical composition (Wang et al., 2010), texture, slope and elevation (Hontoria et al., 2004), and tillage intensity and no-till duration (Umakant et al., 2010).Soil conservation strategies are being seen as a strategy to increase soil OM content (Barbera et al., 2012;Batjes et al., 2014;Jaiarree et al., 2014;Srinivasarao et al., 2014;Fialho and Zinn, 2014).
Several studies have been carried out to estimate differences in soil organic carbon (SOC) dynamics in relation to soil properties, land uses and climate (Eshetu et al., 2004;Lemenih and Itanna, 2004;Muñoz-Rojas et al., 2013).Although the impact of topographic position on soil properties on SOC content is widely recognized (Venterea et al., 2003;Fu et al., 2004;Brevik, 2013), relatively few studies have been conducted to examine the role of topographic position (Fernández-Romero et al., 2014;Lozano-García et al., 2014).
The spatial variation of soil properties may also be significantly influenced by aspect (which may induce microclimate variations), physiography, parent material, and Published by Copernicus Publications on behalf of the European Geosciences Union.L. Parras-Alcántara et al.: Soil organic carbon along an altitudinal gradient vegetation (López-Vicente et al., 2009;Brevik, 2013;Ashley et al., 2014;Bakhshandeh et al., 2014;Dingil et al., 2014;Gebrelibanos et al., 2014;Kirkpatrick et al., 2014).Ovales and Collins (1986) evaluated soil variability due to pedogenic processes across landscapes in contrasting climatic environments and concluded that topographic position and variations in soil properties were significantly related.McKenzie and Austin (1993) and Gessler et al. (2000) found that variations of some soil properties could be related to the slope steepness, length, curvature and the relative location within a toposequence.Both studies suggest that the assessment of the hillslope sequence helps to understand variations of soil properties in order to establish relationships among specific topographic positions and soil properties.Asadi et al. (2012) found that the integrated effect of topography and land use determined soil properties.Topography is a relevant factor controlling soil erosion processes through the redistribution of soil particles and soil OM (Cerdà and García Fayos, 1997;Ziadat and Taimeh, 2013).
The topographic factor has been traditionally included in the study of the spatial distribution of soil properties (Fernández-Calviño et al., 2013;Haregeweyn et al., 2013;Ozgoz et al., 2013;Wang and Shao, 2013).Over time, many researchers have quantified the relationships between topographic parameters and soil properties such as soil OM and physical properties such as particle size distribution, bulk density and depth to specific horizon boundaries (McKenzie and Austin, 1993;Gessler et al., 1995Gessler et al., , 2000;;Pachepsky et al., 2001;Ziadat, 2005).Soil OM content has been negatively correlated with the topographic gradient (Ruhe and Walker, 1968) and slope gradient (Nizeyimana and Bicki, 1992).However, quantitative relationships between soil topography and soil physical-chemical properties are not well established for a wide range of environments (Hattar et al., 2010).
Research along altitudinal gradients has shed light on the effects of climate on soil properties.Ruiz-Sinoga et al. (2012) found a strong relationship between soil OM and elevation, which was due to reduced decomposition rates with lower temperatures.High erosion rates have been found under dry climates and low altitudes in Israel (Cerdà, 1998a, b), which support the idea of high OM losses due to soil erosion in dry areas.
In line with this, in Mediterranean natural areas there is no information about the soil variability; also little data is available related to the control topography exerts on soil properties (Lozano-García and Parras-Alcántara, 2014).Therefore, the aims of this study are (i) to quantify SOC contents and their vertical distribution in a natural forest area, (ii) to assess the SOCS differences in soils along an altitudinal gradient and (iii) to assess their relationship with soil depth in a Mediterranean natural area.

Study site
The Despeñaperros Natural Park (76.8 km 2 ) is one of the best-preserved landscapes in southern Europe.It is located within the eastern Sierra Morena (province of Jaén, southeastern Spain), at coordinates 38  1).Well-preserved Mediterranean woodlands and scrublands occupy the study area and deer hunting habitat is the main land use.

Soil sampling and analytical methods
Seven sites were selected along a topographic gradient in a south-facing slope in the Despeñaperros Natural Park (Table 1).Soil samples were collected at each site following a random sampling design according to FAO (2006).Each selected point was sampled using soil control sections (SCSs) at different depths (S1: 0-25, S2: 25-50 and S3: 50-75 cm).Soil control sections were used for a uniform comparison between studied soils.Four replicates of each soil sample were analyzed in laboratory (17 sampling points × 1, 2 or 3 SCS × 4 replicates).
Soil samples were air-dried at constant room temperature (25 • C) and sieved (2 mm) to discard coarse particles.The analytical methods used in this study are described in Table 2.
Statistical analysis was performed using SPSS Inc. ( 2004).The physical and chemical soil properties were analyzed statistically for each SCS of different soil groups (PH, CM, RG and LP), including the average and standard deviation (SD).The statistical significance of the differences in each variable between each sampling point and soil type was tested using the Anderson-Darling test at each control section for each soil type.Differences with p < 0.05 were considered statistically significant.Nelson and Sommers (1982) Parameters calculated from study data SOC stock (Mg ha −1 ) Total SOC stock (Mg ha −1 ) a 3 cm diameter, 10 cm length and 70.65 cm 3 volume.b Prior to determining the particle size distribution, samples were treated with H 2 O 2 (6 %) to remove organic matter (OM).Particles larger than 2 mm were determined by wet sieving and smaller particles were classified according to USDA standards ( 2004).c Where SOC is the organic carbon content (g Kg −1 ), d the thickness of the soil layer (cm), δ2 mm is the fractional percentage (%) of soil mineral particles > 2 mm in size in the soil, and BD the soil bulk density (Mg m −3 ).

Soil properties
The soils are stony soils, acidic, with low base concentrations, oligotrophic and with slightly unsaturated complex change and located in areas of variable slopes ranging between 5 and 38 %.Phaeozems are the most developed soils in the study area.They are deep, dark, and well humidified with high biological activity and high vegetation density on gentle slopes and shady side foothills.Cambisols are developed and deep soils; however, Leptosols are the least developed and shallowest soils.Phaeozems are the most pedogenically developed soils in the study area.They are found on gentle slopes (< 3 %), usually in shaded areas on Ordovician sandstones.The gravel content is variable, ranging between 7 and 31 % (total weight).Texturally they are sandy soils at the surface and silty-clay-loam or silty-clay soils at depth, with a horizon sequence A0/A1/AB/Bt/C1.These soils show Luvic (lv) characteristics (Luvic Phaeozems (lv-PH)) and are > 1 m in depth with pH along the profile ranging from 6.3 to 5.6 at depth and about 4.3 % OM content (Tables 1 and 3).Cambisols are less developed soils than Luvic Phaeozems; however, these soils are more developed and deeper than Regosols and Leptosols.They appear in areas of variable slope (3-38 %) and are > 1 m in depth characterized by a cambic horizon (Bw) on Ordovician quartzite (Table 1) with approximately 20 % gravel content.At the surface they are sandy soils (< 60 % sand content) with high clay content in the Bw horizon and increasing clay content with depth (Table 3).The horizon sequences were A0/A1/AB/BW/BC/C1 or A0/A1/AB/BW.These soils are characterized by low OM content at depth.Gallardo et al. (2000) showed that the low OM content could be explained by the semiarid Mediterranean conditions.In addition, Parras-Alcántara et al. (2013a) found there is less OM and fewer mineral aggregates in sandy soils, thus favoring high levels of OM transformation.Because of this, Hontoria et al. (2004) suggested that physical variables determine soil development in the driest areas of Spain to a greater degree than management or climatic variables.The Cambisols topsoil has humic (hu) characteristics, with > 5 % OM content (Table 3) due to plant debris accumulation in the A0 horizon.This OM is poorly structured and partially decomposed, thereby reducing the amount and increasing the OM evolution degree with depth.In line with this, Bech et al. (1983) reported that the free OM concentration in the surface horizon was higher than 90 %, while humic and fulvic acid concentrations were less than 2 % in soils with Quercus ilex spp.ballota vegetation.Free OM was reduced and humidification increased up to 30 % in deeper layers.
Regosols can be found in steeply sloping areas (> 8 %) characterized by high water erosion and subject to rejuvenation processes.We found Eutric (eu), Dystric (dy) and Umbric (um) Regosols (Table 1) on sandstone and quartzite parent materials with > 25 % gravel content in surface layers that in some cases disappeared in depth.These soils are sandy loam in surface layers and silty clay in deep layers, with different horizon sequences (A0/A1/AB/BC/C1, A0/A1/AC/C1 and A1/AC/C1).Eutric Regosols are deeper soils (> 80 cm) that are loamy with high gravel content (25.1-32.2%) at the surface decreasing with deep, acidic pH (5.9) and high OM content (6.7 %) at the surface.The Dystric Regosols are stony soils that are shallow (< 40 cm), loamy at the surface and sandy at depth with high gravel content (> 40 %) at the surface, acidic pH (6.2) and high OM content (7.3 %) in the surface horizon (Table 3).The Umbric Regosols are also stony soils that are deep soils (> 70 cm), loamy with high gravel content (40 %) in the surface decreasing to 11 % at depth, have acidic pH (5.6) and high OM content (6.5 %) (Table 3).
Leptosols are the least developed soils of the study area.Lithic (li), Mollic (mo) and Eutric (eu) Leptosols were identified (Table 1) formed in sandstone, quartzite and slate on variable slopes (1.5-46 %).Horizon sequences A1/AC/C1, A1/AC, and AC/C1 and A1 were found.The gravel content was variable (> 40 % in the topographically elevated areas and decreasing with depth) with high sand content (> 50 %) in the surface layers.One characteristic of these soils is that the clay content increased with depth, reaching up to 30 %.According to Recio et al. (1986), the physicalchemical properties of the soils in the study area are due to lithology, while their low edaphic development is conditioned by age.According to Nerger et al. (2007), the alteration and pedogenesis processes taking place in these soils  3).

Distribution of soil organic carbon
Generally, soils in the study area are characterized by > 3 % OC content, making them part of the 45 % of the mineral soils of Europe that have between 2 and 6 % OC content (Rusco et al., 2001).Soil OM content decreased with depth at all topographic positions (A, B, C and D positions) (Table 4).However, this property cannot be observed in the lowest topographic positions (E, F and G positions) due to the low edaphic development (Umbric Leptosols, Lithic Leptosols and Mollic Leptosols) as only one SCS exists (S1: 0-25 cm) (Tables 1 and 4).
The soils in this study are characterized by high sand content at the surface (S1) varying between 59.2 and 34.2 % for C and F positions respectively, and reduced sand content with depth in all studied soils (Table 3), affecting OM development.Clay content reaches 45 % in C: S3.In addition, the mineral medium may play an important role in soil humidification processes, so we can explain low soil OM concentrations with depth due in part to soil texture, because soil OM tends to decrease with depth in virtually all soils, regardless of textural changes.Clays over sands induce a decrease in soil OM with depth; also, the aggregate formation between OM and the mineral fraction is reduced favoring high OM levels in sandy soils at depth (González and Candás, 2004).Furthermore, Gallardo et al. (2000) argued that the relatively low concentrations of OM in depth could be explained by the climate (Mediterranean semiarid).Similar results have been found by Corral-Fernández et al. (2013), Parras-Alcántara et al. (2014) and Lozano-García and Parras-Alcántara (2013a) in the Pedroches Valley, near the study area.
Another key issue is that the clay fraction increased with depth in the B and C positions (reaching a clay content of as high as 45 % (C: S3)) and its relation with soil OM at depth (S2: 25-50 cm), which was characterized by high OM contents as compared to S3 (B: 2.0/0.6 %; C: 1.8/0.06%) (Tables 3 and 4).Burke et al. (1989) and Leifeld et al. (2005) have identified high OM levels in soils with high clay content in depth indicating clay stabilization mechanisms in the L. Parras-Alcántara et al.: Soil organic carbon along an altitudinal gradient soil.This effect can be observed in the B and C topographic positions, where an increase in clay content was observed at depth as compared to the upper horizons (B: S1-17.2 %/S2-22.1 %; C: S1-16.1 %/S2-35.7 %).This OM increase may be due to carbon translocation mechanisms (dissolved organic carbon), soil biological activity and/or the root depth effect (Sherstha et al., 2004).
Soil OM appears to be concentrated in the first 25 cm (S1), where the mineralization and immobilization C processes should be slightly active.In the surface layer (S1), OM was variable along the toposequence studied ranging between 39.9 and 27.3 g kg −1 at the B and F positions, respectively (Table 4).In this regard, it is important to point out that the S1 layer can reach over 60 % of the total soil organic carbon (T-SOC) values documented, corresponding to 60, 64.4 and 63 % for the B, C and D positions respectively as compared to the rest of the soil profile (S2 or S2+S3).Batjes (1996) states that for the 0 to 100 cm depth approximately 50 % of soil organic carbon (SOC) appears in the first 30 cm of the soil.Jobbágy and Jackson (2000) showed that 50 % of SOC is concentrated in the first 20 cm in forest soils to 1 m depth.Civeira et al. (2012) showed that SOC in the upper 30 cm of soils in Argentina is much higher than in the 30-100 cm interval.Data provided by these authors and the results obtained in this study may be comparable because in this study we used a 75 cm depth and the mentioned authors used a 1m depth.Furthermore, Jobbágy and Jackson (2000) indicated that changes in SOC were conditioned by vegetation type (which determines the vertical distribution of roots) and to a lesser extent the effect of climate and clay content.Despite this, climatic conditions can be a determining factor in the SOC concentrations for surface horizons, whereas clay content may be the most important element in deeper horizons; also, clay contributes to stabilize OM by protecting physically microbial activity and reducing C outputs.This effect is important under homogeneous climate conditions (as those in the study area).At the regional-global scale, the precipitation contributes to maximize SOC and temperature accelerates mineralization process decreasing the SOC (Post et al., 1982).
Results of T-SOC analysis in the studied area did not show up great along the toposequence.T-SOC depended on the degree of development of the soil that appeared at each topographical position.The T-SOC was highest at the B (66.5 g kg −1 ), D (58.1 g kg −1 ) and C (52.3 g kg −1 ) positions, corresponding to Cambisols-Regosols-Leptosols, Regosols, and Phaeozems-Cambisols-Regosols respectively.Leptosols showed the lowest T-SOC content with 27.3 g kg −1 , 31.9 g kg −1 , 32.7 g kg −1 and 38.1 g kg −1 at the F, G, E and A topographic positions, respectively.Similarly, > 60 % of SOC was concentrated in the S1 layer of deeper soils (B, C and D).
Precipitation and temperature varied through the studied toposequence, where precipitation increases and temperature decreases with increasing elevation (Parras-Alcántara et al., 2004).T-SOC content was not affected by climatic variations, but depended on the soil development in each landscape position.Reduced T-SOC contents were observed at the lowest topographic positions, where soils were shallower.This is in agreement with Power and Schlesinger (2002), who concluded that topographic position affects T-SOC, due to low OM decomposition rates under low temperatures.

Soil organic carbon stocks
Soil organic carbon stocks in the study area showed a reduction with depth in all topographic positions (Table 4).This SOCS reduction along the profile is linked to OM reduction with depth, which also depends on the gravel content and the bulk density (Table 3).
Higher values of SOCS (up to 91.1 Mg ha −1 ) were found in the upper SCS at elevated topographic positions (highest value at the B position).The lowest SOCS values were found at the G position (53.8Mg ha −1 ), the lowest site in the toposequence.This trend of decreasing SOCS with decreasing elevation is constant except at the A and E positions.Both are poorly developed soils with high OM content in the surface horizon.
In the S1 SCS, between 53.8 and 58.0 % of the SOCS were found at the D and B topographic positions.This constituted 63.0 and 60.0 % of T-SOC in these topographic positions.This shows that the gravel content and bulk density affects the SOCS in the surface horizons of the toposequence studied, and, therefore, SOCS decreases when SOC increases.In the most developed soil, similar SOC and SOCS concentrations (B: 60 % -SOC; 58 % -SOCS) were observed in the S1 layer, conditioned by bulk density and gravel content.In addition, SOCS decreased in depth conditioned by reduction of gravel content and increasing bulk density.This is not in agreement with Tsui et al. (2013) and Minasny et al. (2006), who suggested a negative relation between bulk density and depth as a consequence of high OM content at the surface, linked to low clay concentrations (Li et al., 2010).In this sense, we observed that high SOCS depended on the SOC concentration and the clay content.However, the SOC concentration affected the SOCS to a lesser degree so that in S2 (25-50 cm) we found > 10 % of SOCS related to SOC (C position).
In contrast, low SOCS can be found in S3 except at the B topographic position (19.1 Mg ha −1 ).This situation could be due to the fact that pedological horizons were generally different than the SCS divisions (S1: 0-25 cm; S2: 25-50 and S3: 50-75 cm) (Hiederer, 2009); in other words, the SCS divisions often led to the mixing of two or more soil horizons (depending on thickness horizon) in any given SCS division.
In all studied soils, the clay content increased with depth.This clay content increase is associated with higher values of SOC (B: S2 and C: S2).In line with this, we can explain high SOCS concentrations in clayey soils caused by clay stabilization mechanisms on SOC; this effect can observed at the A  4), showing a correlation between S1 and S2, due to carbon translocation processes as dissolved organic carbon, bioturbation and/or deep rooting (Sherstha et al., 2004).

Soil organic carbon stocks along the altitudinal gradient
The SOCS results along the toposequence were also studied.
It is important to point out that total SOCS (T-SOCS) was influenced by topographical position in the toposequence analyzed.T-SOCS increased exponentially with elevation from G (607 m a.s.l.) to B site (1009 m a.s.l.), with the exception of the highest topographic position, A (1168 m a.s.l.), with an exponential regression relationship (Fig. 1).Similar results were found by Ganuza and Almendros (2003), Leifeld et al. (2005) (Balesdent et al., 2000;Paustian et al., 2000).Cerdà (2000) indicated that this effect could occur regardless of climatic conditions.As can be seen in Table 4, T-SOCS decrease was not homogeneous.In some cases, rapid changes were found, while in other situations gradual changes were noted.Abrupt changes in T-SOCS occurred between the B/C and D/E topographic positions, showing T-SOCS differences of 38 Mg ha −1 and 44 Mg ha −1 respectively.Gradual changes in T-SOCS occurred between the C/D, E/F and F/G topographic positions with variations of 3 Mg ha −1 , 13 Mg ha −1 and 6 Mg ha −1 respectively.Some authors have concluded that the SOCS reduction can be explained by soil physical properties -mainly texture (Corral-Fernández et al., 2013;Parras-Alcántara et al., 2013b).The studied soils are sandy at the surface, with clay increasing with depth, except in E, F and G sites (soils that have S2 and/or S3 SCS); therefore, OM stabilizing mechanisms are produced, reducing the aggregate formation between SOC and mineral fraction at depth.As a result, the SOCS content is lower with sandy soils (Nieto et al., 2013).González and Candás (2004) and Parras-Alcántara et al. (2013a) obtained similar results: the first in sandy-loamy soils and the second in Mediterranean clayey soils.In addition, low SOC levels are conditioned by the climatic characteristics of southern Europe (Gallardo et al., 2000).

Conclusions
Soils found in the Despeñaperros Natural Park include Phaeozems, Cambisols, Regosols and Leptosols.Phaeozems are the deepest and most developed soils, and Leptosols are the least developed and shallowest soils.These soils are characterized by low OM content with depth due to the semiarid Mediterranean conditions and the high sand content.The studied soils are characterized by organic residue accumulation in the surface horizons.
The SOC content decreased with depth at all topographic positions and the clay fraction increased with depth.The mineral medium played an important role in soil humidification processes.In addition, the SOC in the S2 layers is characterized by high SOC values with respect to the S3 layers indicating clay stabilization mechanisms in the soil.We can explain this increase due to carbon translocation mechanisms (dissolved organic carbon), soil biological activity and/or the root depth effect.
With respect to T-SOC content, there is not a large difference between T-SOC along the toposequence.The T-SOC of these soils depends on the degree of development of the soils found at each topographic position.We can observe a T-SOC reduction at the lowest topographic positions for less developed soils and a T-SOC increase at the L. Parras-Alcántara et al.: Soil organic carbon along an altitudinal gradient highest topographic positions in the more developed soils.SOCS in the study zone shows a reduction with depth in all topographic positions.This SOCS reduction along the profile is linked to OM and gravel content reduction and an increase in bulk density with depth.The T-SOCS increased with altitude, due to the higher turnover of organic material (plants) and the lower decomposition rate due to lower temperatures.

Figure 1 .
Figure 1.Exponential regression model for T-SOCS versus altitudinal gradient.T-SOCS: total soil organic carbon stock

Table 1 .
Soil groups of the study area at each of the seven topographic positions with properties.The key refers to the reference soil groups of the IUSS Working Group WRB (2006) with lists of qualifiers.

Table 2 .
Methods used in field measurements, laboratory analysis and to make calculations from study data.

Table 3 .
Properties of the soils evaluated (average ±SD * ) in the Despeñaperros Natural Park.a.s.l.: meters above sea level; SCS: soil control section; BD: bulk density; OM: organic matter.* Standard deviation.Numbers followed by different lower-case letters within the same column have significant differences (P < 0.05) at different depths, considering the same topographic position.Numbers followed by different capital letters within the same column have significant differences (P < 0.05) considering the same SCS at different topographic positions. m

Table 4 .
Soil organic carbon (SOC) content and soil organic carbon stock (SOCS) (average ±SD * ) in the Despeñaperros Natural Park.
m a.s.l.: meters above sea level; SCS: soil control section; SOC: soil organic carbon; T-SOC: total SOC; SOCS: soil organic carbon stock; T-SOCS: total SOCS.* Standard deviation.Numbers followed by different lower-case letters within the same column have significant differences (P < 0.05) at different depths, considering the same topographic position.Numbers followed by different capital letters within the same column have significant differences (P < 0.05) considering the same SCS at different topographic positions.usually occur on low slopes.The Lithic Leptosols are the least developed soils at this study site, with thicknesses ranging between 10 and 15 cm in areas of steep slope.In flat areas, their low development is due to their extreme youth.