Study site

This work was carried out on 2 land uses, a human-induced pastureland and an adjacent forest of Abies religiosa used for grazing in the Agua Blanca ejido in NT. Although the cartographic records of 1972 already show the presence of this pastureland, the National Agrarian Registry (RAN 1998; Franco-Maass et al 2006) indicates that these lands were originally A. religiosa forest that were transformed as of 1929, when the land was redistributed as result of the agrarian reform, whose objective was to give greater weight to livestock activities. In this work the time of more than 40 years was used as a reference based on existing cartographic records. NT is located in the State of Mexico between 18°51′ and 19°19′ N and 99°38′ and 100°09′ W in the Trans-Mexican Volcanic Belt that crosses the central portion of Mexico (Figure 1). Cold climate dominates the NT region; average annual temperature ranges from −2 to 5°C. Average temperature in the coldest month is less than 0°C and in the warmest month ranges from 0 to 6.5°C. The region has a summer rainfall regime and frequent snowfalls in winter, mostly in January and February. Maximum extreme temperatures may reach 21°C in summer, while minimum extreme temperatures may decrease to −10°C in winter. Average annual rainfall varies from 200 to 1800 mm (Challenger and Soberón 2008), with greater rainfall from May to October; July is the rainiest month (García 1990). The dominant soil type is Andosol, although other types such as Phaeozem, Regosol, Cambisol, and Latosol are also present (Vargas 1984). The dominant ecosystems in the PAFF are temperate forests with tree species of the genera Abies, Pinus, and Quercus (3000 to 4100 m). High mountain grassland dominated by species of the genera Festuca and Calamagrostis is found at elevations of 4100 to 4500 m (Rzedowski 1978).

FIGURE 1 

Location of study area, Protection Area for Flora and Fauna (PAFF) NT (top right). PAFF in a LANDSAT 8 image (2017) was combined with a digital elevation model and a classification of land cover (bottom right). Land cover and sampling sites: P1–P3 livestock; P4–P6 Abies religiosa forest (left). (Map by Farid Uriel Alfaro Ramírez)

Overall, the land ownership system in NT area is collective, including 53–56 agrarian units. In some sites of the Agua Blanca ejido, only 67.5% of shareholders manage areas ranging from 0.25 to 2 ha, and 37.5% grow seasonal agricultural crops: oats (Avena sativa) and beans (Vicia faba L.). After harvest, livestock graze on the stubble (Martínez-Hernández et al 2014; Mastretta-Yanes et al 2014). There are 19 homes, none of which have electricity, a refrigerator, a washing machine, a personal computer, a landline, a cell phone, or Internet connection; 5% have an automobile, 5% have television, 26% have a radio, 26% have piped water, and 37% have a toilet.

Sampling design

A combined stratified and systematic sampling based on a 2 × 2 factorial design was applied in the present study. Two land uses (extensive livestock grazing on grasslands and in forest) were considered at 2 soil depths (0–5 cm versus 5–25 cm). The forest is one of the best-conserved Abies religiosa (H.B.K.) Cham. & Schltdl, (sacred fir) forests in the area (Mastretta-Yanes et al 2014). Community members of the ejido use human-induced grasslands that run parallel to the sacred fir forests close to the area where the community lives. These human-induced grasslands and the forest are therefore exposed to the same environmental conditions, elevation, etc. For each land use, 3 sites (used as repetitions in the statistical analysis; Figure 1) with the greatest possible environmental homogeneity were selected; the 3 grassland sites were parallel to the 3 forest plots. In each site, 3 nonexperimental plots of 12 m × 12 m were systematically established. Plots were divided into 9 quadrants of 4 m × 4 m, and the corners of each plot were delimited with wooden stakes.

Sampling and physicochemical soil analysis

Soil samples were taken once from each of the 9 quadrants per plot at depths of 0–5 cm and 5–25 cm with a cylindrical bore of 5 cm in diameter, resulting in a total of 54 samples per land use and 108 samples in total. Each sample was deposited in a plastic polyethylene bag that was labeled and placed in a cooler and then transported to the laboratory of the Institute of Rural Agricultural and Livestock Science (Instituto de Ciencias Agropecuarias y Rurales, ICAR) of the Autonomous University of the State of Mexico (UAEM). Once transported to the laboratory, the samples were dried at ambient temperature for 1 week and then sieved with a 2-mm mesh to separate their components (soil, rocks, and roots). The bulk density (BD) of the soil samples was calculated using the cylinder method (Elliot et al 1999). Soil fraction less than 2 mm in size was used to determine soil moisture, SOM, SOC, and pH. Moisture relative content (%) was determined using the gravimetric method (20 g of soil dried at 105°C for 48 h; Jarrel et al 1999). SOM (%) and SOC (%) were quantified by the modified Walkey and Black method (Nelson and Summers 1996). Soil pH was measured using the potentiometric method and a soil:distilled water ratio of 1:2. Finally, soil compaction was measured in situ at depths of 0–5 and 5–25 cm with a portable penetrometer (Turf-Tec Soil Compaction Tester), which measured resistance to soil penetration (MPa).

The following measurement was performed on additional soil samples taken from 3 of the preestablished quadrants in a diagonal pattern in each plot, beginning with the upper right quadrant. Unaltered samples were collected with a PVC pipe of 15 cm in length and 8 cm in diameter; samples were kept intact for the determination of hydraulic conductivity (Ks). Water infiltration in soil was indirectly measured by determining the saturated Ks (Morikawa-Sakura and Yoshitaka 2014), which showed the soil's capacity to conduct water under saturated conditions. In total, 18 soil samples, i.e., 9 per soil use (3 subsamples per plot), were collected to evaluate infiltration. The samples were placed in a cooler and transported to the Soil Physics Laboratory of the Postgraduate College, Montecillo Campus, where hydraulic conductivity was measured in saturated soils under a constant charge in a permeameter, which is based on the Darcy equation and reported in centimeters per hour (cm h⁻¹) (Klute and Dirksen 1986).

Statistical analysis

The data obtained for SOC and SOM relative content, C stocks (mg C ha⁻¹), soil moisture (%), BD (g cm⁻³), and pH were used to perform a Kolmogorov-Smirnov normality test. When normality distribution was not fulfilled, a logarithmic transformation of the data was performed (Sokal and Rohlf 1994). One-way ANOVA was calculated considering a 2 × 2 factorial design (2 land uses × 2 soil depths), followed by a Tukey's multiple test comparison at a confidence level of 95%. Then, and given that the samples for measuring water infiltration were fewer than those for evaluating the other variables, and that all variables were not similarly evaluated at both depths, the data were also summed (mg C ha⁻¹) and averaged (BD, soil moisture, pH, SOM, and SOC) to compare all variables including infiltration in relation to land uses. Relationships between variables were independently evaluated for each soil depth and land use by determining the Pearson correlation coefficients. All statistical analyses were performed using the software SPSS Statistics 22.0.

Soil organic matter, soil organic carbon, and infiltration per land use

In the present study, the SOC and SOM relative content (%), C stocks (Mg ha⁻¹) and water infiltration were measured in livestock pasture and forest sites as supporting and regulating ecosystems services of mountain soils. Results showed that SOC and SOM (Figures 2A, 2B) differed significantly (P = 0.009 in both cases) between the 2 evaluated land uses. SOC was significantly lower in the livestock pasture (12.6 ± 0.79%) than in the forest (14.65 ± 0.58%). Contrary to SOC relative content, the stocks were not significantly different between land uses (Figure 2C). Similarly to SOC, SOM was found to be lower in the livestock pasture (21.7 ± 1.37%) than in the forest (25.7 ± 0.88%). In addition, infiltration was found to be twice as high in the forest (Figure 2D), and soil compaction was greater in the livestock pasture (1612.66 ± 52.21 MPa; Figure 3C) than in the forest (1551.38 ± 28.73 MPa). Similarly, BD was significantly higher (P = 0.013) in the livestock pasture (0.86 ± 0.25 g cm⁻³; Figure 3A) than in the forest soils (0.73 ± 0.06 g cm⁻³). Finally, the pH (P = 0.000; Fig. 3B) had a value of 6.0 (± 0.12) and 5.34 (± 0.05) in the livestock pasture and forest, respectively.

FIGURE 2 

(A) Soil organic carbon, (B) soil organic matter, (C) carbon stocks, and (D) infiltration in soil under both land uses. Different letters indicate significant differences among land use means (P ≤ 0.05).

FIGURE 3 

(A) Bulk density, (B) pH, (C) and soil compaction in soil under both land uses. Different letters indicate significant differences among land use means (P ≤ 0.05).

Soil organic carbon, soil organic matter, and infiltration by depth

Results indicated that soil depth is an important factor that influences the regulating ecosystem services in each land use. In this respect, SOC and SOM relative content (%) varied significantly according to soil depth and land use (Figure 4). For different soil depths of the same land use, significant differences (P < 0.05) were found in SOC relative content (%), with lower values at 5–25 cm (Figure 4A). There were also significant differences between the superficial layer of one land use and deeper soil of the other land use, with the same pattern of the lowest value in the deeper soil layer. SOM (Figure 4B) showed a significant difference between depths in the same land use with higher SOM in 0–5 cm soil layer than 5–25 cm. Differences between land use were recorded in the superficial soil layer with lower SOM in the livestock use.

FIGURE 4 

(A) Soil organic carbon, (B) soil organic matter, and (C) carbon stocks in 2 soil layers per land use. Different letters indicate significant differences among means (P ≤ 0.05) of the interaction between soil depths and land uses.

With respect to C stocks (Mg ha⁻¹; Figure 4C), significant differences were recorded between the 2 depths in both land uses, with the C stocks triple the amount in the deeper layer of 5–25 cm in comparison to the surface layer of 0–5 cm (ca. 60 Mg C ha⁻¹ in both land uses). Soil moisture (Figure 5A) was significantly lower in the layer of 5–25 cm (8.8 %) than in the surface layer in both land uses; meanwhile, BD (Figure 5B) was significantly greater in the layer of 0–5 cm in the livestock pasture (0.88 g cm⁻³) than in the forest. The pH values were grouped at both soil depths. Significant differences in pH (Figure 5D) were found only between land uses; the livestock pasture was less acidic.

FIGURE 5 

(A) Soil moisture, (B) bulk density, (C) pH, and (D) and soil compaction in soil under under both land uses. Different letters indicate significant differences among means (P ≤ 0.05) of the interaction between soil depths and land uses.

The relative content of SOC and SOM in addition to the C stocks, water infiltration, BD, and percentage moisture were related in different ways depending on soil depth and land use. In the livestock pasture, the Pearson correlation coefficient showed a significantly positive relationship between SOC and C stocks at both depths of 0–5 cm (r = 0.728) and 5–25 cm (r = 0.800), as well as between SOM and C stocks in the livestock soils (Table 1). The BD of livestock soil was negatively and significantly correlated with SOC and SOM relative content (%) in the soil layers of 0–5 cm (r = −0.749) and 5–25 cm (r = −0.682). Soil compaction in the livestock pasture also showed significant correlation with soil moisture in the soil layer of 5–25 cm (r = 0.697; Table 1). In the forest (Table 2), the BD was significantly related with C stocks (Mg C ha⁻¹) at both depths of 0–5 cm (r = 0.779) and 5–25 cm (r = 0.955), and water infiltration was inversely and significantly related (r = −0.755) with soil compaction.

TABLE 1 

Pearson's correlation matrix for supporting and regulating mountain soil services, as well as some physicochemical properties of livestock soils at 2 depths (0–5 and 5–25 cm).^a)

TABLE 2 

Pearson's correlation matrix for supporting and regulating mountain soil services, as well as some physicochemical properties of forest soils at 2 depths (0–5 and 5–25 cm).^a)

Effect of land use on soil organic carbon, soil organic matter, and infiltration

In the present study, SOC and SOM values were slightly lower in the livestock pasture (Figure 2A and 2B, respectively) than in the forest. These results are consistent with those reported by Cruz et al (2012) in zones near the present study area. But while Cruz et al found much higher differences in SOM content (27.1% at 10 cm of depth in forest soils A. religiosa versus 18.3% in grazed soils) and SOC (15.8% in forest versus 10.6% in grazed soils), the results of the present study showed only slight differences between forest and livestock grazing (Figure 3C). When considering SOC stocks per unit area (Mg C ha⁻¹), differences between land uses were not significant. This can partly be explained by the fact that C inputs and stabilization are very slow, and variation in C can generally be detected only after 10 to 15 years following land uses changes toward livestock (Swift 2001; Lal 2004). However, that would depend on soil type, livestock load, and vegetative cover type. Thus, both vertical and horizontal SOC distribution depend on plant functional types, i.e., grasses or forest trees, and their differences in SOM shoot/root allocation, with higher SOC in forest (50%) compared to grasslands (42%) in the upper 20 cm of soil (Jobbágy and Jackson 2000). Thus, in this study A. religiosa forest is characterized by a closed and shadowy canopy, which prevents the development of conspicuous shrubs and herbaceous strata and encourages abundant growth of lichens and mosses: these characteristics maintain lower temperatures and higher moisture levels, promoting the higher accumulation of SOM in the upper soil (Rzedowski 1978; Jobbágy and Jackson 2000; Rzedowski and Rzedowski 2001). At the same time, in the grasslands of the study site, grasses go through all stages of development and then senesce; then litter from senescent grasses is incorporated into the soil, suggesting that the livestock load in the study site is low.

Another study found a greater transference and accumulation of SOC in grazed alpine soils than in conserved alpine grasslands and even native forest soils (8.3 Mg ha⁻¹ in grazed soils versus 2.1 Mg ha⁻¹ in forest soils) (Andrade-Castañeda et al 2014). On the other hand, the significant reduction in SOC relative content (Figure 3A) and the greater BD in the livestock pasture of the present study indicate that livestock grazing simultaneously impacted both variables; however, the differences in the C stocks between both land uses are not clear because of the compensatory effect of BD, which is fundamentally related to SOC relative content (Walter et al 2016). In the surface soil layer (5 cm) of livestock soils of this study, high SOC (%; Figure 4A) and BD (Figure 5B) were recorded. According to Amésquita and Pinzón (1991), greater livestock loads lead to an increase in BD in the first 15 cm of soil. Moreover, several other studies (Amésquita and Pinzón 1991; Klumpp et al 2009; Keller and Hakansson 2010) have found a decrease in soil porosity as a result of livestock grazing and, as a consequence, a decrease in the flow of water from the surface soil layer to deeper layers. In the present study, the flow of water through infiltration was 1.6 times lower in livestock pasture soil (Figure 2D) than in forest soil. Accordingly, the regulating ecosystem service of soil C storage has not been significantly impacted after 40 years of livestock grazing, despite a corresponding significant reduction in SOM content and increased soil compaction (greater BD). This could be a result of the capacity of Andosol soils to recover from stressful events such as animal trampling, having higher resilience as confirmed in situ during fieldwork and in other studies (Martínez et al 2008; Dörner et al 2009).

Vertical variability of soil organic carbon, soil organic matter, and infiltration

Relative contents of SOC and SOM, such as C stocks at different depths (Figure 4) varied vertically between the livestock pastures and forestland. In both the livestock pasture and forest, significantly greater relative content of SOC and SOM were recorded at a depth of 0–5 cm than at 5–25 cm. However, overall, the average SOM values were lower in the livestock land use.

Contrary to the results for SOC relative content (%), C stocks were significantly lower in the surface soil layer (0–5 cm) and contained only 22% of the C stored in the first 25 cm of soil in both land uses of the ejido of Agua Blanca. The first 25 cm of soil is highly important in the conservation and storage of organic C in Andosol soils; in addition, this soil layer is significant for SOM accumulation and as an interface enabling the flow of water to greater depths. Changes in the upper layer are slow, according to Geissen et al (2009), who found that changing land use did not lead to the chemical degradation of soil, even after 15 years; while physical properties such as BD can experience drastic changes, as shown by high soil compaction at soil depths of 0–20 and 20–40 cm in permanent pastures. In line with the latter, Fernández-Rebollo et al (2015) and Bell et al (2011) reported that the trampling of sheep and goats only affects the first 10 cm of the surface soil layer. Hewins et al (2018) conclude that long-term livestock grazing may enhance SOC concentrations in shallow mineral soil and affirm that climate rather than grazing is the key modulator of soil C storage across northern grasslands.

Finally, in the present study, the evaluated variables were more stable between the soil layers in the forest, where C stocks were significantly related with BD. An inverse relationship between soil compaction and water infiltration also was found. Greater variability in evaluated variables was encountered between the soil layers of livestock soils with compaction, SOM, and SOC lower in the surface layer.

Environmental perspectives and implications

Impacts of livestock grazing on soils of the Agua Blanca ejido in NT, where extensive livestock ranching is practiced, indicate that soils can retain their capacity to store C and conduct water to deeper soil layers under certain conditions. This finding confirms that extensive livestock ranching maintains certain soil functions and ecosystem services in the long term in this region and possibly in other regions and ecosystems around the world (Piñeiro et al 2010). In zones with conditions similar to those of the present study area, economic reimbursement schemes compensate livestock ranchers for their contribution to maintaining ecosystem services and public environmental goods (Uribe et al 2011). Ultimately, the goal of such schemes is to encourage farmers to maintain or adopt sustainable grazing systems and to protect remaining forest ecosystems. In some respects, promoting extensive livestock ranching and forest conservation at the same time may seem contradictory, but the present study supports the merging of these perspectives. Nevertheless, critical and fundamental variables for guiding livestock management strategies such as SOM and BD should be further studied and identified to enable greater equilibrium between the needs of livestock ranchers and the maintenance of supporting and regulating ecosystem services of mountain soils along the elevational gradients of the high mountain systems.