Tessala mountains are part of the Tell Atlas of Algeria situated on a hundred kilometers between Oran to the North and the plain of Sidi Bel Abbes to the South. They give reliefs from 500 to 1,000 m of altitude, where the mount Tessala culminates to 1,061 m. The landscape draws a rugged morphology with slopes exceeding 50% in some parts, accentuated by a very marked ravine. These slopes are dissected by a large number of wadis (an Arabic term traditionally referring to a valley) and tributaries that carry fine and stony materials downstream in the plain of Sidi Bel Abbés (Pouquet, 1952). The main lithological units are represented by marls, limestone marls, and clay marl; they are met mainly in the North and North-West part (Insid, 2019). The sandstone and limestone marl of the Eocene are found mainly in the South-East. The gray marls are located in the southern part of the piedmonts. Tessala Mount has semiarid climate characteristics, characterized by a clear dry and wet season, the average annual temperature is ∼21°C, and the rainfall is low and irregular (356 mm) and occurs mainly in autumn and winter. The period from late April to mid-October is generally dry with high summer temperature and the aridity index is 12.7 (Belahcene, 2019). The vegetation cover of the forest area of Tessala mountains is composed of oak coppice (99 ha); matorral (310 ha); forest (209 ha); dense garrigue and sparse garrigue (382 ha) (Saidi, 2017).

The sampling campaign involved 7 sites; each site is characterized by a distinctive vegetation formation (Supplementary Table S2 and Supplementary Figure S2). Soil samples were collected randomly from each location in May 2017 at soil depth ranging from 0 to 30 cm. Given the morphology of the terrain and the nature of the harsh rugged relief of the study area, a series of six transects with a total length of approximately 8.5 km was selected; these six series represent 7 vegetation coves including matorral (M); sparse garrigue (GC); forest (F); dense garrigue (GD); olive grove (OL); oak coppice (TC); and herbaceous vegetation (VH). A random sampling but still dependent on the relief and the distribution of plant formations was followed for each virtual line. In total 32 composite soil samples (each composite is composed of 3 samples) were collected; this gave one sample per 265 m. Soil samples were classified according to the World Reference Base for Soil Resources (WRB) (WRB, 2007); see Supplementary Table S2. The soil samples were air-dried and sieved using a 2 mm sieve for physical and chemical analyses. The mineral fractions were measured by the “Robinson” method (Gee and Or, 2002). The principle is to remove any cement such as carbonates, oxides, and organic substances by oxidation with hydrogen peroxide, the dispersion is done in sodium hexametaphosphate, the samples were pipetted at different times and different depths, following different sedimentation intervals, and the duration and the depth were calculated with Stokes’s law. All the granulometric data were expressed in the percentage of fine earth (<2 mm). The pH was measured using a pH meter (Hanna Instruments, Netherlands). For the soil samples suspensions with a ratio of 1: 2.5 (m/v) of the fine earth and water, as shown by Thomas (1996), the salinity was evaluated by the electrical conductivity (EC) in millisiemens per centimeter (mS cm⁻¹) using a conductivity meter (Hanna Instruments, The Netherlands); a suspension of fine earth and water with a ratio 1:5 (m/v) was considered for this analysis (Rhoades, 1996). The bulk density (BD) in g cm⁻³ was determined by the direct sampling method using a 5.5 cm diameter and 5 cm height cylinder (Blake and Hartge, 1986). The samples taken from the cylinder were dried then passed through a sieve with a diameter of 2 mm in order to measure the content of coarse particles (expressed in %). The OC content was determined by the modified Walkley and Black method, based on the oxidation of OC by potassium dichromate (K₂Cr₂O₇) in sulfuric acid (Gillman et al., 1986). The soil inorganic carbon (SIC, CaCO₃ content) was determined for total carbonates by Bernard calcimeter method (Sparks et al., 1996) using HCl acid. The CO₂ released by the reaction was measured by a gas burette (Bernard calcimeter). For active carbonates, the method described by Drouineau (1942) was used, the calcium associated with oxalates gives insoluble calcium oxalate, and the excess of ammonium oxalate is then determined by a solution of potassium permanganate in a sulfuric solution. The physical and chemical characteristics of the soil are reported in Supplementary Table S1.

The C stocks were calculated for 0–30 cm depth, with the C stocks for each sample calculated using the equation recommended (FAO, 2017; FAO, 2019) as follows: SOC stocks  ( t C ha −1 ) =  0 . 1 * C * BD * T * ( 1 - CP ) (1) where SOC stock is a soil organic carbon stock (t C ha⁻¹), C is the carbon content (g C kg⁻¹ soil), BD is the bulk density (g cm⁻³), T is the thickness of the soil horizon (cm), and CP is the gravel content (g g⁻¹ soil).

The percentage of fine particles was used to calculate the stable carbon potential in soils following the equation below according to Hassink (1997): C saturation ( % ) = 4 . 09   +  0 . 37 ( % clay  + % fine silt ) (2) The stable C value can be compared to the measured soil C and from that carbon saturation deficit can be calculated (Angers et al., 2011): C saturation deficit = C sat - C actual (3) The critical thresholds of carbon represent a minimum level of organic C below which the soil structure is weakened and the risks of its degradation become stronger because of the low structural stability. This value was calculated after Autfray et al. (2009) as follows: C critical ( % ) = [ 0 . 32   ( % Clay  + % fine Silt ) +  0 . 87 ] / 10 (4)

An ANOVA test was performed to determine the significance of differences between measured averages. A Welch test was realized to verify if there is a difference between soil SOC stocks values under the 7 studied vegetation covers. A principal component analysis (PCA) was performed to determine if there is a correlation between variables followed by parallel hierarchical cluster analysis (HCA) (Supplementary Figure S1) to verify the information provided by the PCA. The HCA was applied to the variables and not to the individuals (represented here by the samples for the 7 sites); the aim is to group similar variables and above all to highlight the variables that have close proportions of SOC, that is, those which have more influence on the accumulation of SOC. Finally, two bivariate correlation tests with two quantitative variables were performed, one between the SOC stocks and the coarse silt and the other between SOC stocks and altitude. All data collected were processed by SPSS Statistics software version 24.0 (IBM Corp, 2016).

The average of the total OC contained in the first 30 cm of the soil varies according to the sampled sites; the highest values were attained in the topsoil under matorral and sparse garrigue vegetation covers, with rates of 3.1% and 3%. The lowest percentages were recorded under forest and herbaceous vegetation covers, with 0.8% and 1.1%. The contents recorded under dense garrigue, olive groves, and oak coppice were in the same range between 2% and 2.6% (Supplementary Table S1). The maximum value of SOC stocks averages (121 t ha⁻¹) stored in Tessala Mount soils at 30 cm depth was noted under dense matorral of green and kermes oak followed by sparse garrigue with a value of 112 t ha⁻¹, while the soils under olive groves contain 77 tz ha⁻¹ and soils under dense garrigue and oak coppice have an average of 73 t ha⁻¹. Finally, the soils under herbaceous vegetation have an average of 48 t ha⁻¹. The minimum SOC stocks average value is recorded under the clear forest of Aleppo pine and eucalyptus with 38 t ha⁻¹ (Table 1).

The actual OC measured for the first 30 cm of Tessala Mount soils records an average of 22 g OC kg⁻¹; it varies between 31.7 g OC kg⁻¹ under dense matorral of oak and 8.6 g OC kg⁻¹ under the clear forest of Aleppo pine and eucalyptus (Table 2). Our results are in the range of previous findings for woodland topsoil in the mountain forest of Belezma in Algeria, where the content of SOC stocks varied from 36 g OC kg⁻¹ under holm oak to 53 g OC kg⁻¹ under cedar (Bensid et al., 2015). Ben Hassine et al. (2012) recorded a SOC content average equal to 0.8% (8 g OC kg⁻¹) in Tunisia under almost similar climatic conditions. Boulmane et al. (2010) indicated that more than 80% of SOC stocks can be stored in the top 30 cm of soils of the Mediterranean oak forest of the Moroccan Middle Atlas. In their study, 63 t ha⁻¹ of SOC stocks was recorded for the dense oak grove located at 1,670 m altitude and 47 t ha⁻¹ for the clear oak located at 1,450 m altitude. Bounouara et al. (2017) recorded 100 t ha⁻¹ and 168 t ha⁻¹ of SOC stocks in Skikda region, northern east Algeria, at an altitude varying between 13 m and 140 m.

The SOC stocks at Tessala Mount vary from one site to another; this variability could be explained by the diversification in terms of vegetation, exposure, and altitude. Several studies emphasized that SOC stocks can vary regionally following environmental factors such as precipitation, temperature, elevation, slope, soil texture, and land use (Yimer et al., 2006; Liu et al., 2011; Gutiérrez-Girón et al., 2015). The type of vegetation plays a crucial role in determining SOC spatial distribution as the trees species composition can control the amount and composition of organic matter inputs especially in the upper soil layers (Díaz-Pinés et al., 2011). Aboveground vegetation properties such as stand age, leaf area index, aboveground biomass, mean tree height, tree diameter, stand density, tree basal area, and litter depth are all vegetation properties that have a potential effect on SOC stocks (Li et al., 2010).

Other studies indicated that plant formations have a profound effect on soil carbon stocks (Arasa-Gisbert et al., 2018; Yao et al., 2019; Koga et al., 2020). Boulmane et al. (2010) found that the density of a plant stand may have no significant effect on carbon sequestration potential in biomass, but it certainly has a significant effect on carbon sequestration in soils and more precisely on the layer 0–15 cm (between 52.7 t ha⁻¹ and 33.2 t ha⁻¹). However, a Canadian study conducted by Lagacé Banville (2009) under different climatic conditions indicates a significant correlation between plant formation and SOC stocks distribution of hardwood and closed canopy conifers with 13.4 and 12.2 kg C m⁻² (134 and 122 t ha⁻¹), respectively; the highest intakes are in sites with the highest plant density.

A bivariate correlation test, between two quantitative variables, the altitude and OC, was performed; the results showed a significant (p<0.05) positive correlation between the altitude and OC (Figure 1). The soils under dense matorral of holm and kermes oak have an average SOC stocks value of 121 t ha⁻¹ at an average altitude of 955 m, whereas soils under Aleppo pine and eucalyptus forest recorded the lowest value with only 38 t ha⁻¹ at an average altitude of 790 m (Figure 2). This difference may be explained by the influence of temperature and humidity on the decomposition rates of soil organic matter and thus the SOC accumulation. Geomorphological terrain attributes, such as slope and elevation, play an important role in controlling carbon stock variations (Chaplot et al., 2001; Moghiseh et al., 2013). The elevation is considered as one of the various important environmental factors controlling soil OC content in arid and semiarid rangelands (Shedayi et al., 2016; Sharafatmandrad, 2019). Besides, low temperatures at higher elevation may suppress the decomposition rates of soil organic matter (Yimer et al., 2006; Wang et al., 2013; Gutiérrez-Girón et al., 2015). Water is the most important key that can influence SOC storage, as microbial activity increases in well-aerated and moist soils which in turn increases the rate of organic matter decomposition, especially under warm to hot climates (Beare et al., 1994; Ogle et al., 2005). Bounouara (2018) found that, for surface horizons (0–30 cm), the stocks of OC decreased from upstream to downstream, while for depth horizons (>30 cm), the opposite can be observed; in his study the carbon stocks have gradually increased from mountain soils to the lowlands. In a similar study on natural forests in northern Iran, Moghiseh et al. (2013) have shown that slope aspect and site altitude have a great effect on SOC storage; higher SOC storage was observed at a higher altitude compared to low altitude sites.

The maximum average of SIC stocks (123 t ha⁻¹) stored in Tessala Mount soils at 0–30 cm was noted under a clear forest of Aleppo pine and eucalyptus, followed by herbaceous vegetation with a value of 87 t ha⁻¹, while the soils under olive groves contain 84 t ha⁻¹ and soils under dense garrigue and oak coppice have, respectively, an average of 21 t ha-1 and 19 t ha⁻¹. The minimum SIC stocks average value was recorded under sparse garrigue and dense matorral of green and kermes oak with 12 t ha⁻¹ for each (Table 1). Our results indicated that SIC has a negative correlation with SOC stocks (p < 0.05); soil with higher values of SIC exhibited less SOC accumulation and vice versa (Figure 3). For example, at SIC of 3% a maximum SOC stocks value was recorded (121 t ha⁻¹), while the soil with 22% of the total carbonates content only recorded 38 t ha⁻¹ of SOC stocks (Figure 2). The results came in contradiction to previous studies findings, which indicate that SIC has a positive effect on soil organic matter stabilization and soil structure (Álvaro-Fuentes et al., 2009; Shi et al., 2017; Martí-Roura et al., 2019; Quijano et al., 2020). Exchangeable Ca correlates positively with the concentration of SOC and its resistance to oxidation due to the positive effect on the aggregation and structural stability of the soil and by the indirect influence on the accumulation and occlusion of OC (Rowley et al., 2018). However, the exact mechanisms behind this relationship and the actual consequences on occluded SOC stocks remain unclear as many factors such as soil texture, pH, organic inputs, and distribution may interfere with these mechanisms. Martí-Roura et al. (2019) found that microbial carbon use efficiency was more closely associated with land use than with carbonate content.

A bivariate correlation test, between coarse silt and SOC stocks, was performed to show the relationship between the coarse silt and SOC accumulation; the results showed a significant marginal correlation of coarse silt with SOC stocks (p < 0.05) (Figure 4). Increasing in coarse silt contents would tend to relatively increase SOC stocks. For example, under dense matorral of holm and kermes oak, an average value of 121 t ha⁻¹ for SOC stocks under 41.3% of coarse silt content was noted, whereas under Aleppo pine and eucalyptus forests the lowest value of SOC stocks was recorded with only 38 t ha⁻¹ and an average percentage of coarse silt of 14.6% (Supplementary Table S1). Soil texture has a distinct influence on organic carbon accumulation because of the physical protection of SOM induced by the aggregation of fine particles, such as clay and fine silt (Feller et al., 1991; Jobbágy and Jackson, 2001). Previous studies showed that forests soils with high contents of silt ranging from fine to coarse aggregates are essential for OC stabilization (Balesdent et al., 1998). Coarse silts and sands fractions ranging from 20 to 2000 µm have the potential to associate plant debris at various stages of decomposition (Feller et al., 1991; Schulten et al., 1993). Forest soils contain particularly large amounts of SOM associated with the coarser soil fraction (Martí-Roura et al., 2019).

The principal components selected are the F1 axis, which provides the most important statistical information with an inertia rate of 37.05% and the F2 axis with 15.69%. These two axes restore 52.74% of the initial information and highlight two groups confirmed by parallel hierarchical cluster analysis (HCA) (Figure S1). The choice of these groups was made based on the principal component analysis (PCA) (Figure 5) and HCA.

Group 1: SOC, C, altitude, coarse silt, pH; Group 2: coarse particles, fine silt, clay, bulk density, sand, total and active CaCO₃, SIC.

Based on the information provided by the PCA, it was noted that elevation, OC, pH, and coarse silt correlate with stocks; PCA shows as variable; OC, altitude, pH, and coarse silt are factors that would influence SOC stocks.

On the basis of the information provided by the sampled sites and the measured edaphic variables, the result of the analysis confirms that the variables of group 1 are correlated with axis 1 which essentially indicates the geographical and physicochemical characteristics of the sites studied represented by northern exposure, high elevation, high CO, and high coarse silt ratio; all of these factors positively contributed to the increase in SOC stock.

On the contrary, the variables of group 2 are not correlated with axis 1; this can be explained by the geographical and physicochemical characteristics of these sites, with a southern exposure, a low altitude, and a low CO content; all these factors are not conducive to the storage of CO in soils. It has been noted that, in addition to the physical and chemical properties of topsoil sites, other geographical factors appear to influence OC storage, such as exposure and altitude. The northern slopes in Tessala area benefit from milder climatic conditions, with lower temperatures compared to the southern slopes. In addition, the sea breeze in summer on the northern slopes, with low evaporation rates and significant rainfall, has created a favorable environment for the accumulation of the organic matter and thus influenced the OC storage (Saidi, 2017). Also, the high altitude of the northern slopes is associated with cooler local climate conditions compared to the southern exposure which has low altitude and thus warmer climate. Soil carbon budget tends to increase when moving from a warmer climate to a cooler climate; this is attributed to the accelerated decomposition of organic matter in warmer and dry climates compared to cold and wet climates (Schimel et al., 1994; Bachelet et al., 2001).

The calculation of soil carbon saturation levels has made it possible to estimate the soil carbon storage capacity. Table 2 (values are presented in g kg⁻¹) shows that the maximum soil storage capacity is noted under olive grove with 23.1% of carbon, in the second position the oak coppice with 21%, then under dense garrigue with 16.7%, and under herbaceous vegetation and sparse garrigue with 13.8% and 13.7%, respectively. Lastly, forest and scrubland recorded 12% and 11%, respectively. The critical threshold of carbon varies between 0.7% and 1.7%. The average rate of SOC concentration in the study area is 22 g C kg⁻¹, and the average critical threshold is 11 g kg⁻¹ (Table 2). The average saturation rate is 160 g C kg⁻¹; this means that it is possible to increase the average carbon content to 138 g C kg⁻¹. In Tunisia, under nearly similar pedoclimatic conditions to our study area (Ben Hassine et al., 2012), a critical threshold was recorded that varies between 2.7% and 1.79% (27 g C kg⁻¹ and 17.9 g C kg⁻¹). The average deficit of Tessala Mount topsoils is 137 g C kg⁻¹. For agricultural topsoils Angers et al. (2011) note an average deficit of 8.1 g C kg⁻¹, with a maximum of up to 500 g C kg⁻¹. The lowest deficit recorded of 78 t C ha⁻¹ was for the soils under dense matorral located at a high altitude with a North exposure. As described in this study smaller deficits occurred at higher altitudes, probably due to the combined effect of the cooler temperature and the presence of meadows. Table 2 shows that the current carbon content (C_actual) of Tessala Mount soils is very low and closer to the critical threshold (C_{critical threshold}) than its maximum storage capacity represented by saturation carbon (C_sat). The carbon deficit (C_{.sat. deficit}) in soils of the study area is very alarming because it is very close to saturation levels.