Seasonal and Land Use Effects on Soil Respiration and Its Controlling Factors in Arid Lands from Northeastern Mexico

Abstract

The extreme conditions in arid ecosystems make these environments sensitive to environmental changes. Particularly, land use and seasonal changes are determinants of their soil carbon dynamics. The effect of those elements on soil respiration (R_S) is still poorly known in several arid regions of the world. This study investigates the seasonal effect on the R_S and its controlling factors throughout different land use systems in northeastern Mexico. R_S and 34 biotic and abiotic variables were measured across agricultural crops, natural shrublands, livestock farms, walnut orchards, and industrially influenced soils during the dry and wet seasons. Six variables (soil water content, soil organic matter, soil temperature, silt, and pH) were found as drivers of R_S on both local and regional scales. Seasonal and land use had a transversal effect on R_S and its controlling factors. R_S dynamics were primarily modulated by soil water content, with the wet season and managed lands showing increased sensitivity to climatic and anthropogenic changes. These results indicate that land management strategies are critical for carbon cycling, particularly in water-limited regions like northeastern Mexico, where land use changes are occurring at an accelerated pace.

1. Introduction

Soils are the largest carbon reservoir on earth, containing three times more carbon than vegetation or the atmosphere [1,2]. They can act as a source or sink of the atmospheric CO₂ flux [3], with arid lands regulating the trend and interannual variability of the sinking behavior [4,5,6]. Arid lands cover roughly 47.2% of the earth’s surface, enhancing their importance in the global carbon storage processes [7]. At a large scale, the carbon dynamic of these ecosystems is mainly regulated by abiotic factors such as atmospheric temperature, precipitation, and shortwave radiation [4]. At lower scales, biotic elements such as soil organic matter degradation and microbial [8,9,10] and vegetation activity [11,12,13,14] also play a key role.

Soil respiration (R_S) represents the below-ground CO₂ biologically generated by the activity of its organisms both autotrophs (CO₂ fixers) and heterotrophs (not CO₂ fixers) [15]. The low precipitation and temperature extremes in arid ecosystems make them highly sensitive to natural and anthropogenic changes, often reflected in highly variable R_S rates [11,16,17,18,19]. The R_S dynamic in arid environments has been related to factors such as photosynthesis, root expansion, soil moisture, and temperature, which tend to be determinants [10,20,21,22]. Seasonal variation is another critical factor that modulates the CO₂ exchange in arid ecosystems [4,6]. Particularly, it has been shown that irregular rainfall patterns with shorter wet periods and longer drought intervals can modulate R_S seasonality and its controlling factors. For example, Huxman, et al. [23] showed that precipitation pulse size and frequency regulate the carbon balance by determining the temporal activity of microbial respiration and photosynthetic activity, resulting in arid ecosystems acting as carbon sources following rainfall. Similarly, it has been shown that precipitation legacies control the carbon uptake and ecosystem respiration of the following seasons, making the gross ecosystem productivity more sensitive to precipitation than respiration [24]. In general, microbial respiration increases following very small or moderate precipitation events, while photosynthetic activity (related to root respiration) is associated with relatively larger events or a series of small ones [23].

Among the anthropogenic activities that modify the soil dynamics, land use change is a driving element in altering the soil carbon balance and, subsequently, the R_S activity of natural ecosystems [15]. Even though studies carried out in recent years, the effects of land use changes on R_S patterns in arid zones still are not completely known, and different outcomes have been recorded. Specifically, when agricultural lands are compared with natural areas of the same region. For example, higher respiration values in a tropical dry forest than in a pasture area were obtained in Brazil [25]. Similarly, Zhang et al. [26] showed that woodland had higher R_S rates in a land use conversion sequence of the same arid area in China. Conversely, other studies have shown higher R_S rates in agricultural crops than in natural environments [20,27,28]. In addition, only a few studies have compared the seasonal changes throughout different land use systems, most of them in China [26,29,30,31,32]. In semi-arid regions of Africa, R_S rates in savanna ecosystems under grazing pressure are lower than in undisturbed areas [33]. At the same time, intensive cultivation further reduces R_S by disrupting soil structure and depleting carbon stocks [34]. Similarly, in Australian arid lands, degraded areas exhibit reduced R_S compared to native shrublands, highlighting the influence of land management on carbon dynamics [35]. These examples underscore the need for studies to address the critical knowledge gaps in understanding seasonal patterns and the controlling factors of R_S across diverse land uses of arid ecosystems at regional scales.

In Mexico, extensive land use changes have taken place in northern arid zones. Many areas of the Chihuahuan desert biogeographic region (the largest North American desert) have experienced a strong impact from agriculture and industry [36], which modify the soil carbon dynamics at different magnitudes. Therefore, measuring the R_S throughout specific temporal and spatial scales is necessary to understand its dynamics best in the region. This study investigates the seasonal effect on R_S and its controlling factors throughout different land use systems in Mexico southeast of the Chihuahuan Desert. Therefore, we hypothesized that the soil water availability, modulated by the seasonal changes, will primarily affect R_S dynamics in the different land uses, showing higher R_S rates during the wet season and in land uses with agricultural practices.

2. Materials and Methods

2.1. Study Area

This study was carried out within the Chihuahuan desert biogeographic region, at the Saltillo, Ramos Arizpe, General Cepeda, and Arteaga municipalities of Coahuila de Zaragoza state in northern Mexico (Figure 1). The dominant climate in the region is BWhw, corresponding to very arid, semi-warm, with a mean annual temperature between 18° and 22 °C [37]. The main edaphological categories are Calcisol, Kastanozem, Leptosol, Lithosol, Regosol, and Xerosol [38]. The dominant vegetation types are microphile and rosetophile desert scrub, with species such as Fouquieria splendens, Larrea tridentata, Yucca carnerosana, Yucca filifera, Dasylirion cedrosanum, Echinocactus spp., Echinocereus spp., Mammillaria spp., and Opuntia spp. [39]. According to the precipitation registers in the region [37], two main seasons were considered: (1) dry season, comprising February, April, and December 2019 (cumulative rainfall 21.2 mm); and (2) wet season, comprising June, August, and October 2019 (cumulative rainfall 149 mm) [40]. Hereafter, these periods will be referred to the as dry season and wet season, respectively.

2.2. Sampling Design

Forty-five sampling sites were used with five land uses: agricultural crops (AC), livestock farming (LF), walnut orchard (WO), industrial influence (II), and natural shrubland (NS). The general characteristics of the land use systems sampled are shown in

Table 1. General characteristics of land use systems studied in arid lands from Chihuahua de Zaragoza, México.

Land Use	Altitude (m.a.s.l.)	Climatic Variables and Soil Group	General Characteristics
Agricultural crops	1300–1700	MAT: 17.5–18.4 °C MAP: 268–493 mm SG: Calcisol, Leptosol, and Regosol	Intensity gradient from fallow fields to fields with three annual crops. Main crops: corn (Zea mayz), sorghum (Sorghum spp.), and oatmeal (Avena sativa). Irrigation systems: seasonal and flood.
Industrial influence	1240–1860	MAT: 17.5–18.5 °C MAP: 268–485 mm SG: Calcisol, Leptosol, and Regosol	Induced grassland (Aristida adscensionis, Bouteloua gracilis) with residual vegetation of microphile and rosetophile desert scrub (Opuntia spp. Larrea tridentata, Cylindropuntia spp.). Placed roughly 100 m from industries with emissions. No management.
Livestock farming	1369–1780	MAT: 16.7–18.5 °C MAP: 374–493 mm SG: Calcisol, Kastanozem, Leptosol, and Regosol	Intensity gradient from semi-confined conditions, semi-confined with grown sorghum (50 head/ha), from intensive exploitation (700 to 1500 head/ha). Cattle with occasional sheep.
Natural shrublands	1240–1850	MAT: 16.7–18.5 °C MAP: 268–493 mm SG: Calcisol, Kastanozem, Leptosol, and Regosol	Microphile and rosetophile desert scrub of variable high (50 to 180 cm). No evident land use changes. Species such as Fouquieria splendens, Larrea tridentata, Yucca spp., Dasylirion cedrosanum, and several Cactecaea species, such as Echinocactus spp., Echinocereus spp., Mammillaria spp., Opuntia spp., etc.
Walnut orchards	1369–1780	MAT: 17.5–18.5 °C MAP: 268–493 mm SG: Calcisol, Leptosol, and Regosol	Walnuts trees (Juglans spp. and Carya spp.) with a productivity gradient from commercial to local exploitation. Irrigation systems: seasonal, drip, and sprinkler.

Mean annual temperature and precipitation [37]; soil group [38].

. Nine sampling sites were established at each land use, and five spaced (30 m) sampling points in a row per site (replicates) were taken (n = 225). A total of 35 parameters corresponding to the soil respiration (R_S), environmental variables (air temperature, relative humidity, photosynthetically active radiation, soil temperature, and soil water content), and abiotic (bulk density, sand, silt, clay, total C and N, electrical conductivity, pH, and soil nutrients) and biotic (microbial biomass carbon and soil organic matter) soil properties were taken at each sampling point. Soil nutrient measurements correspond to 19 chemical compounds (Al₂O₃, CaO, Cl, Cr₂O₃, CuO, Fe₂O₃, K₂O, MgO, MnO, Na₂O, NiO, P₂O₅, PbO, SO₃, SiO₂, SrO, TiO₂, ZnO, and ZrO₂).

Sampling was carried out from February to December 2019 during the maximum R_S activity hours (12:00 to 16:00) [40,41] with the sites measured randomly. The R_S and environmental variables were measured every other month (February, April, June, August, October, and December), whereas the electrical conductivity (EC), pH, microbial biomass carbon (MBC), and soil organic matter (SOM) were registered only twice: during the dry (February) and wet (August) seasons. Bulk density (BD), sand, silt clay, total C and N, and soil nutrients were measured only once at the first sampling event (February).

2.3. Measurement of Soil Respiration, Environmental Variables, and Soil Properties

R_S was measured with a portable dynamic closed chamber (SRC-1; PP System, Amesbury, MA, USA) connected to an infrared gas analyzer (EGM-5; PP Systems, Amesbury, MA). The chamber was placed on an external PVC collar (5 cm depth × 10 cm diameter) temporarily inserted to a depth of three centimeters into the soil 24 h before the R_S measurements to minimize the impact of this insertion [42]. Regarding the environmental variables, the soil temperature (T_soil) and soil water content (SWC) were recorded at 8 cm depth by using a Hydra Probe II (Stevens Water Monitoring Systems Inc., Portland, OR, USA). The air temperature (T_air) and relative humidity (RH) were recorded with a WatchDog (mod. 1450) microstation, and the photosynthetically active radiation (PAR) was recorded with an MQ-200 sensor (Apogee Instruments, Inc., Logan, UT, USA).

pH and EC were determined in a 1:2.5 (w:v) aqueous suspension (Orion Star A211 and Orion Star A212, respectively; both from Thermo Fisher Scientific, Waltham, MA, USA). The MBC was obtained by the fumigation extraction method [43]. The SOM was evaluated by the loss on ignition method (at 400 °C for 4 h). The BD was analyzed by the volumetric method using a graduated cylinder. The soil texture (sand, silt, and clay content) was determined by granulometry with a Horiba LA 950 V2 Particle Size Analyzer. Total C and N were measured on air-dried soil samples using a C:N elemental analyzer (Flash Smart, Thermo Fisher Scientific, Waltham, MA, USA). Total concentrations of soil nutrients were determined by X-ray fluorescence (Bruker S4 Pioneer, Billerica, MA, USA).

2.4. Data Analysis

R_S data measured with EMG-5 (g CO₂ m⁻² h⁻¹) were converted to micromoles of carbon dioxide per square meter per second (µmol CO₂ m⁻² s⁻¹) divided by the conversion factor 0.1584 [44]. Prior to the data analyses, outliers were validated (by using Tukey’s method), and a random forest algorithm imputation was applied for missing data (<8% of the total data in all the cases) [45,46]. Seasonal and land use effects were analyzed by computing a one- or two-way analysis of variance (ANOVA). Tukey’s tests were carried out to test differences among land uses. Normality and homoscedasticity of variance were tested. The natural logarithm transformation was implemented to meet the assumptions when necessary. Nutrient content data were summarized to be used in the structural equation models (see below) by computing a principal component analysis (PCA, two first components). PCA results are shown in the Supplementary Materials (Figure S1 and Table S1). ANOVAs and PCA were computed in R software version 4.2.1 [47].

Structural equation models (SEMs) were used to identify the factors controlling R_S, incorporating all the previously analyzed data and various hypothetical causal relationships among variables. Additionally, SEMs were used to examine significant differences in the R_S drivers across seasons and between the studied land uses. An initial model considering that environmental (T_air, T_soil, HR, SWC, and PAR), soil abiotic (soil texture, pH, EC, and nutrients), and biotic (SOM, MBC) had an influence over R_S was constructed. Models considered a complete set of hypotheses based on the literature [48,49,50] and previous experience [40,41,51]. From the general model, a multigroup SEM was used to test whether the studied factors were linked by the same causal structure in each season (dry or wet) and to identify the paths that did not behave similarly in the two seasons. For this analysis, the same hypothetical model was used for each season group separately. Since a lack of fit was detected in the fully constrained multigroup model, a series of nested models, where equality constraints were removed one at a time, were developed to detect which one would significantly improve the model [52,53]. To test for differences in parameter values between the two seasons, χ² and AIC statistics were used.

The degree of fit between observed and expected covariance structures was assessed in the multigroup model by the root mean square error of approximation statistic (RMSEA), where values < 0.08 indicate a good fit, between 0.08 and 0.10 provide a moderate fit, and >0.10 suggest a poor fit [54]. The models’ fit to the data was additionally evaluated by Bentler and Bonett’s normed-fit index (NFI), in which values ranging between 0 and 1 and those >0.9 indicate an acceptable fit [55]. The maximum likelihood algorithm was used to estimate standardized path coefficients [56]. All SEM analyses were performed by using SPSS^® AMOS 26.0 software.

3. Results

3.1. Soil Respiration, Environmental Variables, and Soil Properties

Both season and land use significantly (p < 0.001) affected all sampled variables at each season, as shown by the ANOVA results (

Table 2. Seasonal and land use variation for the soil respiration, and the environmental, physicochemical, and biological soil properties measured in arid lands from Coahuila de Zaragoza, Mexico. AC = agricultural crops, II = industrial influence, LF = livestock farming, NS = natural shrublands, WOs = walnut orchards.

	AC	II	LF	NS	WOs	Average
Dry season
RS	1.67 ± 0.1 aB	0.36 ± 0.03 bB	1.74 ± 0.1 aB	0.58 ± 0.06 bB	1.73 ± 0.08 aB	1.22 ± 0.04 B
PAR	988.36 ± 45.6 bB	1220.44 ± 44.92 aB	966.23 ± 44.92 bB	1222.53 ± 47.24 aA	983.36 ± 53.37 bA	1076.18 ± 21.6 B
Tair	31.82 ± 0.37 bB	33.22 ± 0.31 aB	30.03 ± 0.3 cB	31.88 ± 0.33 bB	32.35 ± 0.33 abB	31.86 ± 0.15 AB
RH	26.07 ± 0.59 abB	22.58 ± 0.44 cB	27 ± 0.44 aB	24.77 ± 0.55 bB	24.56 ± 0.57 bcB	24.99 ± 0.24 B
Tsoil	28.96 ± 0.35 bcB	31.87 ± 0.32 aB	28.69 ± 0.32 cB	30.09 ± 0.33 bB	29.15 ± 0.29 bcB	29.75 ± 0.15 B
SWC	12.08 ± 0.71 aA	4.06 ± 0.29 cB	9.03 ± 0.58 bB	2.95 ± 0.23 cB	11.07 ± 0.59 aB	7.84 ± 0.27 B
EC	1268.03 ± 74.79 aA	721.83 ± 32.87 bA	1059.83 ± 59.05 aA	814.6 ± 49.19 bA	1139.12 ± 53.91 aA	1000.68 ± 28.16 A
pH	8.04 ± 0.05 aA	8.21 ± 0.05 aA	7.8 ± 0.05 bA	8.16 ± 0.04 aA	8.11 ± 0.05 aA	8.07 ± 0.02 A
MBC	38.49 ± 4.29 abB	20.36 ± 1.13 cB	51.31 ± 5.02 aB	28.98 ± 3.07 bcB	42.34 ± 3.62 abB	36.3 ± 1.78 B
SOM	6.15 ± 0.32 aA	3.61 ± 0.24 bA	6.91 ± 0.66 aA	3.7 ± 0.21 bA	6.78 ± 0.29 aA	5.43 ± 0.16 A
Wet season
RS	2.56 ± 0.11 aA	0.81 ± 0.06 bA	2.53 ± 0.11 aA	1.11 ± 0.09 bA	2.62 ± 0.11 aA	1.93 ± 0.05 A
PAR	1426.62 ± 61.14 aA	1434.36 ± 48.93 aA	1355.37 ± 54.82 aA	1374.08 ± 60.95 aA	637.62 ± 56.55 bB	1245.61 ± 27.9 A
Tair	35.34 ± 0.34 aA	34.62 ± 0.32 aA	34.74 ± 0.27 aA	35.61 ± 0.3 aA	35.26 ± 0.29 aA	35.11 ± 0.14 A
RH	37.68 ± 0.68 aA	34.28 ± 0.6 bA	37.13 ± 0.57 aA	34.32 ± 0.59 bA	38.24 ± 0.63 aA	36.33 ± 0.28 A
Tsoil	33.78 ± 0.35 abA	34.68 ± 0.28 aA	33.13 ± 0.32 bA	34.94 ± 0.28 aA	32.61 ± 0.31 bA	33.83 ± 0.14 A
SWC	13.4 ± 0.6 aA	6.63 ± 0.34 bA	13.49 ± 0.7 aA	6.43 ± 0.44 bA	13.08 ± 0.6 aA	10.61 ± 0.28 A
EC	482.76 ± 49.54 aB	452.72 ± 57.44 aB	458.92 ± 80.66 aB	351.69 ± 52.6 aB	495.19 ± 68.14 aB	448.26 ± 28 B
pH	7.56 ± 0.07 bB	7.9 ± 0.04 aB	7.6 ± 0.06 bB	7.86 ± 0.04 aB	7.73 ± 0.05 abB	7.73 ± 0.02 B
MBC	75.93 ± 5.53 aA	49.05 ± 3 bA	78.37 ± 5.45 aA	57.58 ± 6.47 bA	77.35 ± 2.79 aA	67.66 ± 2.13 A
SOM	4.27 ± 0.32 bB	3.21 ± 0.25 bA	6.15 ± 0.35 aA	3.36 ± 0.22 bA	6.42 ± 0.32 aA	4.68 ± 0.16 B
One sampling event variables
BD	1.11 ± 0.012 b	1.19 ± 0.015 a	1 ± 0.02 c	1.18 ± 0.015 a	1.04 ± 0.011 c	1.11 ± 0.01
Sand	26.09 ± 1.48 b	33.41 ± 1.77 b	45 ± 2.58 a	44.56 ± 2.18 a	32.76 ± 1.77 b	36.36 ± 1.01
Silt	68.59 ± 1.3 a	63.11 ± 1.6 a	52.55 ± 2.21 b	53.42 ± 2.08 b	63.15 ± 1.43 a	60.16 ± 0.88
Clay	5.21 ± 0.36 a	3.47 ± 0.29 bc	2.25 ± 0.4 cd	1.63 ± 0.17 d	3.6 ± 0.36 b	3.23 ± 0.17
CT	5.1 ± 0.28 c	5.43 ± 0.29 c	7.63 ± 0.45 a	5.68 ± 0.38 bc	6.88 ± 0.38 ab	6.15 ± 0.17
NT	0.27 ± 0.02 b	0.19 ± 0.01 c	0.47 ± 0.03 a	0.21 ± 0.01 bc	0.4 ± 0.02 a	0.31 ± 0.01

R_S = soil respiration (μmol CO₂ m⁻² s⁻¹), PAR = photosynthetically active radiation (mmol m⁻² s⁻¹), T_air = air temperature (°C), RH = relative humidity (%), T_soil = soil temperature (°C), SWC = soil water content (%), EC = electrical conductivity (mS cm⁻¹), MBC = microbial biomass carbon (mg C glucosa kg⁻¹), SOM = soil organic matter (%), BD = bulk density (g cm ⁻³), CT = total carbon (%), and NT = total nitrogen (%). Means and standard errors with different letters denote significant differences in decreasing order (α = 0.05). Lowercase letters indicate comparisons among land uses, and capital letters are comparisons between seasons.

and

Table 3. Effect of season (S), land use (LU), and the interaction (S × LU) in the respiration, environmental, physicochemical, and biological soil properties of arid lands from Coahuila de Zaragoza, México.

		Factor
Variable		S	LU	S × LU
	df	1	4	4
RS (μmol CO2 m−2 s−1)	F	163.42 ***	161.18 ***	2.74 *
PAR (mmol m−2 s−1)	F	26.34 ***	31.48 ***	17.81 ***
Tair (°C)	F	262.91 ***	7.86 ***	7.43 ***
RH (%)	F	989.03 ***	15.71 ***	3.98 **
Tsoil (°C)	F	416.24 ***	22.58 ***	4.08 **
SWC (%)	F	67.32 ***	105.99 ***	2.68 *
	df	1	4	4
EC (mS cm−1)	F	217 ***	10.15 ***	5.45 ***
pH	F	109.55 ***	16.96 ***	2.12
MBC (mg C glucosa kg −1)	F	147.92 ***	18.91 ***	0.63
SOM (%)	F	16.43 ***	57.24 ***	2.59 *
	df	–	4	–
BD (g cm −3)	F	–	30.44 ***	–
Sand (%)	F	–	16.92 ***	–
Silt (%)	F	–	15.46 ***	–
Clay (%)	F	–	17.85 ***	–
TC (%)	F	–	8.81 ***	–
TN (%)	F	–	44.73 ***	–

R_S = soil respiration, PAR = photosynthetically active radiation, T_air = air temperature, RH = relative humidity, T_soil = soil temperature, SWC = soil water content, EC = electrical conductivity, MBC = microbial biomass carbon, SOM = soil organic matter, BD = bulk density, TC = total carbon, and TN = total nitrogen. *** p < 0.001, ** p < 0.01, * p < 0.05.

). Nevertheless, an interaction effect of land use and season (p < 0.05) was observed, except for pH and MBC (Table 3). Regarding environmental variables analyzed, Figure 2 and Table 2 show the highest and lowest R_S values were recorded during the wet season in walnut orchards (WOs) and the dry season in sites under the industrial influence (II), respectively. The highest PAR value was recorded during the wet season in II sites, whereas the lowest was observed during the wet season in walnut orchard (WO) areas. The highest and lowest T_air values were observed during the wet season in natural shrublands (NS) and during the dry season in livestock farming (LF) sites, respectively. T_soil had the highest value throughout the wet season in NS and the lowest throughout the dry season in LF sites. The highest and lowest RH values were shown throughout the wet and dry seasons in WO and II sites, respectively. The SWC showed the highest and lowest values during the wet and dry seasons in LF sites and NS, respectively. The highest EC was observed during the dry season in agricultural crops (AC), whereas the lowest was registered during the wet season in NS. The most basic pH was measured throughout the dry season in soils from II, whereas the lowest values were detected during the wet season in AC. The MBC was highest during the wet season in LF sites and it was lowest during the dry season in II sites. Finally, SOM was highest throughout the dry season in LF sites and lowest during the wet season in II sites.

3.2. R_S Controlling Factors

SEMs showed the causal-effect relation of biotic and abiotic variables controlling R_S. At the regional scale (i.e., overall data, Figure 3), this analysis exhibits the robust control of SOM content influencing positively or negatively the abiotic (SWC, soil temperature, pH, silt content) and biotic (microbial biomass) soil factors, which results in a direct and indirect influence over the R_S. The SWC was the main controlling factor of the R_S variability regardless of the season. Even if total effects are considered, the SWC exerts a stronger influence over R_S than SOM (

Table 4. Standardized direct, indirect, and total effects of the main controlling factors for soil respiration (R_S) during the dry and wet season in arid lands from Coahuila de Zaragoza, Mexico (See Figure 3).

	Effect	SOM	SWC	Tsoil	pH	MBC	Silt
Dry season	Direct	0.2	0.526	0.059	---	0.135	−0.155
	Indirect	0.253	−0.004	0.011	−0.042	---	0.157
	Total	0.453	0.522	0.07	−0.042	0.135	0.003
Wet season	Direct	0.228	0.502	−0.056	---	0.037	−0.001
	Indirect	0.2	−0.001	0.003	−0.012	0	0.069
	Total	0.428	0.501	−0.053	−0.012	0.037	0.067

SOM = soil organic matter, SWC = soil water content, T_soil = soil temperature, and MBC = microbial biomass carbon.

). Furthermore, multigroup SEM also showed how the controlling factors of R_S variability significantly (p < 0.05) changed in importance (e.g., SWC and SOM), no longer had an influence (i.e., silt and MB), or even shifted the influence over the R_S (i.e., T_soil). Specifically, the SWC showed a stronger influence over the R_S during the dry season than the wet one (standardized regression weight (SRW) of 0.53 and 0.50, respectively). On the contrary, SOM showed a slightly stronger influence over the R_S during the wet season than the dry one (SRW of 0.23 and 0.20, respectively). Both silt percentage and MBC only influenced the R_S during the dry season. Furthermore, T_soil positively influenced the R_S during the dry season and negatively during the wet (SRW of 0.06 and −0.06, respectively) (Figure 3).

SEM by land use (Figure 4 and Figure 5), in addition to the consistent and strong influence of SWC over R_S (except for II during dry season), exhibit the positive control of T_soil over R_S under the dry season (i.e., AC, LF, WO; Figure 4a,c,d) and are negative during the wet one (LF, II; Figure 5c,e). Other controls were less consistent across land use and season. Silt content negatively influenced the R_S for LF (both seasons), WOs (dry season; Figure 4d), and positively in NS (wet season; Figure 5b). SOM positively influenced II R_S regardless of the season and negatively influenced it in WOs. The MBC influenced the R_S positively on LF and II (dry season; Figure 4c,e) and negatively influenced it in WOs (wet season; Figure 5d). pH only had an effect over R_S during the dry season on NS and LF (Figure 4b,c).

4. Discussion

4.1. R_S Seasonal Dynamic

The annual mean R_S rate recorded in this study (1.57 μmol CO₂ m⁻² s⁻¹; Table 2) exhibits noticeable distinctions compared to other arid zones with varying land uses in China (2.27 μmol CO₂ m⁻² s⁻¹ growing season [20,26]) and Brazil (2.03 μmol CO₂ m⁻² s⁻¹, [25]). Still, it is consistent with that previously recorded in natural shrublands from northwestern Mexico (1.4. μmol CO₂ m⁻² s⁻¹) [57]. Inherent variations in the controlling factors of land CO₂ exchange at continental scales (atmospheric temperature, precipitation, and shortwave radiation) partially account for the above-mentioned similarities and differences in the R_S annual rates [4,6]. According to those spatial variations, those studies also showed a significant seasonal change, which is mirrored in an increase in the R_S during the wet season, matching our results. This R_S seasonal pattern is common in other natural environments, such as the dry tropical forest [58], tropical cloud forest [59], pine forest [9], pine-oak forest [60], and arid meadows [61]. Nevertheless, some other environments have reached a negative response of R_S to precipitation (e.g., coastal wetlands [62] and other semi-humid or humid ecosystems [63]). Thus, data about seasonal and precipitation effects on the R_S from the different habitats around the world are still necessary to understand the soil C dynamic best.

The magnitude of the seasonal change obtained in the studied region (145% to 191% higher during the wet season throughout the different land uses; Table 2) is smaller when compared to other arid environments in Mexico (295% in chaparral vegetation [57]), India (229% to 256% through fragmented tropical thorn forests; [10]), Brazil (1089% and 1220% in dry tropical forest and degraded pasture, respectively [25]), and Africa (421 to 598% through three different land uses; [34]). In addition to the continental variations in environmental variables that drive R_S at large scales (mentioned above), the C dynamic is also influenced by variations in the SOM added to the soil by plants and its specific modifications through the mineral matrix [64]. Therefore, factors such as water availability, vegetation type [40,57,65,66], microclimatic variations [67], and high spatial heterogeneity in the vegetation composition [14,68] can help to explain the differences regarding the magnitude of the R_S seasonal change between sites at lower scales. Thus, R_S seasonal change recorded in our overall results may be attributable to a higher sensitivity of the autotrophic respiration to drought when microorganisms are in dormant states [69] but also to the capacity of microorganisms to rapidly resuscitate from that state after being stimulated by the increase of the SWC during the wet season [9,40,70]. This is supported by the direct effect of the SOM over MBC, and R_S (Figure 3 and Table 4), which also suggests an important contribution of the heterotrophic component [71,72,73].

4.2. Land Use Effects on R_S

Despite the strong seasonal effect on R_S, as expected, land use also influenced the recorded R_S rates, aligning with findings documented in other arid zones [20,29,74]. Our results show R_S rates observed across the land uses with inherent management practices (managed lands: agricultural crops, livestock farming, and walnut orchards) were 250% to 372% higher (Table 2) than those from areas without specific management practices (natural shrublands and sites under industrial influence). Despite this, the recorded R_S for agricultural crops (annual mean: 2.11 μmol CO₂ m⁻² s⁻¹) falls within the global cropland annual range (0.47–4.16 μmol CO₂ m⁻² s⁻¹ [75]). This pattern of higher R_S for agricultural lands compared to native vegetation aligns with observations in various studies [20,27,29,32], underscoring the negative impact of agricultural practices on the soil’s carbon dioxide stocks of arid ecosystems. Nevertheless, it has also been recorded that implementing sustainable management practices or cultivating specific crops, such as some native ones, can mitigate the adverse effects on R_S [76,77], potentially even acting as a C sink [78]. To propose more sustainable practices, further studies are necessary to investigate the potential effect of those practices on the C dynamics of agricultural lands in the region.

Prima facie results revealed heightened SOM, SWC, and MBC values in agricultural lands compared to sites without management practices (Figure 2 and Table 2). Irrigation and fertilization practices, along with cattle excrement commonly found in agricultural lands (Table 1), directly lower pH and EC while increasing SOM, carbon, and water availability (Table 2; Figure 4 and Figure 5). These changes create favorable conditions that enhance microbial activity, boosting R_S [35,79,80]. Since the agricultural sites in the region often continue their management practices even during the dry, the consistent stimulation of R_S controlling factors may also explain the R_S seasonal variation when comparing sites with and without agronomic management. Specifically, during the dry season, R_S values were 288% to 480% higher in sites with agronomic management, contrasting with the wet season where the difference ranged from 228% to 323% (Table 2). Although the influence of specific management practices, such as fertilization, was not analyzed in this study, it is possible that they strongly influence the R_S rates obtained. This has been shown in other studies that record an increase from 18% to 48% after nitrogen (N) or phosphorous (P) fertilization [79,80]. Similarly, tillage and fallowing practices tend to increase CO₂ emissions from 0.2 to 3 times in crops regarding bare soils [81,82]. The influence of these non-sustainable management practices (that increase the loss of soil CO₂) on the regional R_S dynamic requires a more detailed investigation.

Regarding sites without management practices, SOM and water content showed a higher influence on R_S with positive effects regardless of seasonal changes (Figure 4 and Figure 5). Changes in the native vegetation concerning the characteristic structure of sites with management practices can influence the R_S by modifying microhabitat conditions [83], such as microclimate, detritus amount and quality, and autotrophic respiration [66,84]. Our data suggest that the lowest R_S values recorded in land uses without management practices can be explained by a lower ecological dynamic derived from simpler structural components or fewer consumables for heterotrophic respiration. Those sites displayed the lowest vegetation density (except when compared with livestock farming sites, personal observation), which directly influences variables that limit microorganism activity (lower EC, SWC, and SOM, along with higher pH), resulting in a reduced carbon flux (lower MBC and R_S values) (Table 2 and Figure 2). A similar response of the R_S and microbial activity due to changes in structural components of ecosystems (mainly related to the vegetative component) has been recorded in other arid environments [11,12].

4.3. Dynamics of R_S Controlling Factors Across Land Uses and Seasons

This study identified six variables as the R_S primary controlling factors: SWC, SOM, MBC, T_soil, Silt, and pH. Among these, the SWC consistently emerged as the most influential factor, irrespective of the season or land use (Table 4, Figure 3, Figure 4 and Figure 5). The dominance of SWC contrasts with studies that emphasize T_soil as the primary R_S driver [22,31,72,85,86]. In general, the contribution of T_soil or SWC to R_S depends on the type of ecosystem and is strongly influenced by factors such as water and soil nutrient availability, microbial communities, and autotrophic organisms. For example, in arid ecosystems with varying land uses, soil temperature tends to be closely correlated with R_S in agricultural sites, particularly during the wet season [33,77,82]. However, our findings mainly revealed a weak direct effect of T_soil on R_S across the different land uses and seasons. Similarly, the influence of T_soil on MBC was found to be weak, with significant effects observed only during the dry season in LF and WO, and during the wet season in areas affected by industrial activities (Figure 4 and Figure 5). In areas with native vegetation of arid ecosystems, it has been shown that soil temperature acts as a limiting factor for microbial activity, negatively affecting R_S [25,87]. The lower influence of T_soil on R_S in sites without management practices in our results may be explained by higher temperatures coupled with the limited availability of SOM for microbial activity. Conversely, in managed sites, factors such as constant irrigation, higher vegetation cover (indirectly showed by the PAR values in crop sites), and humidity generated by cow excrements may contribute to relatively lower temperatures and a weaker T_soil influence on the R_S [28]. These findings emphasize the importance of considering the interactions among the R_S and its controlling factors in further studies in the region.

Despite the dominant role of the SWC, total weights (Table 4) indicated that SOM is the second most important controlling factor. High drought stress in the region diminishes plants’ activity, leading to low SOM levels that restrict microbial activity [88,89]. These patterns were particularly observed in soils with industrial influence during the dry season, where SOM served as the primary constraint of microbial biomass carbon (Figure 4b). In general, our results show that SOM can either positively or negatively influence both abiotic (SWC, T_soil, pH, silt content) and biotic (microbial biomass) controlling factors (Figure 3 and Figure 4), thereby shaping the spatial and seasonal dynamic of R_S. The quality of SOM, in terms of its fraction proportions (recalcitrant, extra recalcitrant, and labile), may explain the differing influence of SOM on the R_S. Labile SOM requires less energy for microbial degradation, which supports a higher abundance of microorganisms. Conversely, more recalcitrant SOM is less bioavailable, thereby reducing its contribution to R_S. This agrees with previous studies that highlight the significance of SOM in soil dynamics [73,90]. Drought stress is the main factor positively associated with R_S and the dehydrogenase activity during the decomposition of organic matter. In contrast, phenol oxidase activity reflects the breakdown of recalcitrant carbon in heterotrophic respiration, a key process in lignin decomposition. The activity of phenol oxidase is primarily determined by seasonal changes in soil moisture [91,92]. During dry periods, phenol oxidase its activity is restricted due to reduced substrate availability, microbial proliferation, and nutrient content, resulting in lower R_S, matching with our results. In contrast, wet periods show increased phenol oxidase activity, consistent with the idea that soil water availability and seasonal rain events are critical drivers of carbon flux in arid ecosystems [4,10,11,23,24,57].

Although the effects of pollutants were not directly analyzed in this study, the presence of heavy metals was recorded in some of the sites influenced by industrial activity. Among the nine industrialized sampled sites, two contained Cr₂O₃ (0.057%, maximum recorded concentration) and one contained PbO (0.078%, maximum recorded concentration). The negative effects of Cr and Pb pollution on microbial activity (arylsulphatase, alkaline phosphatase, and urease) and R_S rates have been demonstrated in roadside soils of intensive traffic regions [93]. Our results yielded sites under industrial influence exhibited higher PAR and lower vegetation density (personal observation), as well as reduced MBC, and R_S values compared to natural shrublands (non-significant differences; Table 2). These findings support the assumption that exposure to industrial volatile compounds and suspended particles negatively impacts R_S dynamics. However, the influence of industrial activity on regional R_S values warrants further investigation to understand its long-term impacts on C dynamics.

The consistently high negative effect of sites with management practices on pH, as shown in our results (Table 2 and Figure 2), aligns with findings from other studies [94,95]. Low pH values in these land uses were negatively correlated to MBC, regardless of season, directly influencing the R_S (Figure 4 and Figure 5). The high SOM content in such sites may contribute to an increase in the H⁺ release during its degradation, thereby lowering pH and negatively affecting MBC [96]. Although soil acidification is a global challenge for carbon cycling and sustainable soil management, the relationship between these remains insufficiently understood. The dynamics of soil pH are complex due to its variable responses to several factors (e.g., climate, topography, parent material, soil biology, human disturbances, sampling design, and measurement errors) [97]. Nevertheless, our results confirm the negative impact of land use changes, particularly those involving agronomic management, on soil pH and their subsequent effects on the R_S dynamics.

Silt content also influences R_S differently depending on land use. In livestock farming and walnut orchards, silt has a negative effect on R_S, whereas in natural shrublands, it exhibits a positive impact. In livestock farming, animal trampling can compact the soil, especially those with more silt, reducing porosity and limiting aeration, decreasing oxygen availability for microorganisms and R_S [95,98]. Conversely, in the natural shrublands, silt particles may promote microaggregate formation, protecting organic matter and enhancing the respiratory activity of specific microorganisms even when their total biomass is limited [99,100].

This research provides new insights into the critical drivers of R_S across various land uses and seasons, advancing our understanding of carbon dynamics in arid environments. By identifying SWC, SOM, and MBC as the main controlling factors, this study offers valuable indicators for monitoring soil carbon dynamics under both natural and managed conditions. Future research should continue exploring how changing precipitation patterns, intensified land use, and industrial pollutants might interact to affect soil carbon fluxes, particularly under climate change scenarios. This knowledge will be essential for designing targeted interventions aimed at enhancing carbon sequestration in arid ecosystems, thereby contributing to global climate change mitigation efforts.

5. Conclusions

The findings of this study underscore the complex interplay between seasonal variations, land use practices, and soil characteristics in regulating soil respiration (R_S) in arid ecosystems. The central role of soil water content (SWC) as the primary R_S driver highlights the vulnerability of arid soils to changing rainfall patterns. Increased R_S rates during the wet season and in managed lands point to elevated sensitivity to hydric inputs, which are likely intensified by agricultural and livestock practices. These results suggest that land management strategies have substantial implications for C cycling, especially in water-limited regions like northeastern Mexico, where land use change is accelerating.

The high R_S rates observed in agricultural lands, driven by elevated soil organic matter (SOM) and microbial biomass carbon (MBC), highlight a dual role for land management in either exacerbating or mitigating soil carbon release. While intensive agricultural practices can amplify R_S, potentially increasing CO₂ emissions, practices aimed at conserving soil moisture and organic matter could help stabilize C storage. Our study also reveals that unmanaged or industrially influenced lands display markedly lower R_S, potentially due to suppressed vegetation growth and microbial activity from pollution exposure. Our results confirm the fragility of CO₂ dynamics to changes in land use in arid regions, highlighting the importance of resuming sustainable actions that help mitigate the increase in greenhouse gas emissions. This finding points to the need for further research into the ecological impacts of industrial activities and the varying management practices of agricultural sites on soil carbon dynamics and microbial communities in the region.