Agroforestry Systems Enhance Soil Moisture Retention and Aquifer Recharge in a Semi-Arid Mexican Valley

Abstract

Agroforestry systems (AFSs) have been recognized for their ecological potential, yet quantitative assessments of their hydrological functions in semi-arid regions remain limited. This study evaluates soil moisture retention and potential aquifer recharge in two agroforestry systems compared to a traditional rainfed maize system in the semi-desert region of Celaya, Mexico, where aquifer depletion is a growing concern. Field measurements during the 2022 rainy season included precipitation, soil moisture at multiple depths, and soil physical properties across seven vegetation covers. The results show significantly higher moisture content, improved uniformity, and enhanced recharge potential under tree species such as Bursera graveolens and Lysiloma divaricatum. These effects are attributed to vegetation cover, organic matter input, and reduced evaporation. This study provides empirical evidence supporting the integration of AFSs into regional water management strategies, offering a nature-based solution for aquifer recovery and climate adaptation in arid landscapes.

1. Introduction

Sustainable management of agroecosystems has become essential to mitigate climate change and biodiversity loss, particularly in semi-arid regions where water scarcity is intensifying. Agroforestry systems (AFSs), which integrate trees with crops or livestock, are widely acknowledged for their multifunctional benefits, enhancing soil quality, providing ecosystem services, and regulating hydrological cycles [1]. Research has shown that agroforestry can improve infiltration rates, reduce runoff, and increase aquifer recharge compared to conventional agriculture [2,3,4,5,6,7]. Under forest cover, soil porosity and moisture retention have been found to be improved in tropical and temperate zones [8,9,10].

However, these findings are mainly based on global case studies, and their relevance to arid and semi-arid environments remains limited, since ecosystems tend to respond differently due to the complex interactions among rainfall patterns, vegetation structure, and soil characteristics [1,11,12]. Mexico’s agroforestry systems are traditionally practiced by indigenous and rural communities and represent significant ecological and cultural reserves [13]. Nevertheless, hydrological research on Mexican AFS has been fragmented and often limited to specific soil parameters or crop performance. In other instances, agroforestry systems have not been directly assessed. Instead, infiltration rates have been inferred from hydrological balance models applied at the watershed scale. These watersheds typically exhibit a range of soil types, vegetation covers, and agricultural production systems, which adds complexity to isolating the effects of agroforestry practices [14,15,16,17]. In terms of soil moisture dynamics and aquifer recharge, there is a critical lack of long-term, integrative studies assessing how tree species composition, canopy structure, and spatial arrangements affect soil moisture dynamics. Murray-Nuez et al. [18] and Luna-Robles et al. [19] have addressed soil properties and infiltration at local scales, but a national synthesis by vegetation type or climate zone has not been attempted. Groundwater depletion is particularly severe in the semi-arid regions of Central Mexico [20,21], yet policy frameworks for recharge management remain uninformed by agroecological research.

Agroforestry configurations such as arboreta and intercropped systems have not been systematically compared with conventional monoculture plots under uniform soil, climate, and species conditions. The mechanisms by which specific species or canopy traits influence water retention and recharge remain poorly understood. In regions where aquifer levels are declining at an alarming rate, this limits the translation of agroforestry into water management policy.

This study aims to quantify soil moisture retention and potential aquifer recharge in the Celaya Valley under semi-arid conditions by evaluating two agroforestry systems and a conventional rainfed maize system. Agroforestry configurations differ in terms of vegetation composition and spatial arrangement. The hypothesis is that systems with greater canopy cover, such as those dominated by Bursera graveolens and Lysiloma divaricatum, will exhibit better soil moisture retention and greater groundwater recharge. The findings are intended to support the development and application of nature-based solutions for sustainable water management in arid and semi-arid Mexican agricultural landscapes.

2. Materials and Methods

2.1. Experimental Site

This study was conducted in the Experimental Field of the Forest, Agriculture, and Livestock Research Center (INIFAP), Bajío branch, located in Celaya, Guanajuato, central Mexico (20°34′ N, 100°48′ W), at an elevation of 1767 m above sea level. The climate is classified as semi-arid (Cwa), with annual precipitation ranging from 400 to 600 mm and average annual temperatures between 15 and 25 °C [22]. The soil type corresponds to vertisols, with high clay content, which affects infiltration and water retention capacities [23].

This area lies within the Celaya Valley aquifer zone, which is considered severely depleted. According to CONAGUA [20], this aquifer exhibited a mean annual drawdown of 1.77 m in 2013 and a cumulative overextraction between 386.58 and 440.78 Mm³ [21]. Groundwater constitutes the primary water source for irrigation and urban use in the region, making this site relevant for studying recharge potential under agroforestry systems.

2.2. Experimental Design

The experimental layout consisted of two agroforestry systems and a conventional control plot. The first agroforestry system (Ar) was a dense arboretum established with trees planted at 3 × 3 m spacing. The species composition included Prosopis laevigata (mesquite), Bursera graveolens (copal), and Leucaena esculenta (huaje). The second system (AIC) was an intercropped agroforestry configuration, where trees were spaced at 5 × 3 m and interspersed with maize (Zea mays). The AIC system included Prosopis laevigata, Lysiloma divaricatum (palo prieto), and Hesperalbizia occidentalis (palo blanco) as the main tree components. The control plot was a conventional rainfed maize field with typical regional management practices and a planting density of approximately 100,000 plants per hectare. The experimental setup is visually summarized in Figure 1.

The selection of tree species in each system was based on their adaptability to semi-arid environments, multifunctionality, and ecological relevance [24,25,26,27,28,29,30,31].

2.3. Measurement of Tree Species

On 15 September 2022, tree canopy diameter, height, and canopy were measured using a tape measure and a free mobile application called HabitApp [32]. Measurements were taken from five trees of each species at the same time of day (12:00 pm), from all selected trees, to avoid bias. The selected trees were those whose soil moisture was measured under their canopy, as well as five other trees near the measurement site.

2.4. Measurement of Soil and Climate Variables

Measurements at the experimental site were taken between 3 August and 12 October 2022.

2.4.1. Soil Characterization

Soil samples were collected from both agroforestry systems and the control plot to characterize their physical and chemical properties. At each plot, soil pits were excavated to a depth of one meter, allowing for the identification and morphological description of the A and B horizons. From each horizon, composite samples were obtained using a 5 cm diameter auger, combining material from at least four different locations within each plot to reduce spatial variability. In the laboratory, texture was determined using a Bouyoucos hydrometer method [33], bulk density was measured through the core sampling method [34], and true density was calculated using a pycnometer. Organic carbon content was estimated using the Walkley–Black method [35], which quantifies the oxidizable fraction of organic matter. These methods were chosen due to their proven reliability in soil hydrology research [36,37], particularly under arid and semi-arid conditions, where infiltration and water retention are strongly influenced by soil texture and organic matter content [38,39,40].

2.4.2. Soil Moisture Content

Soil moisture content was measured on five dates: 3 August, 17 August, 15 September, 28 September, and 12 October 2022, designated as dates 1, 2, 3, 4, and 5, respectively. Volumetric moisture content was measured using a FIEDSCOUT 300 ^® Time Domain Reflectrometer (Spectrum Technologies, Inc. Aurora, IL, USA) with an accuracy of ±3% VWC (volumetric water content) at the intersection points of two rectangular, perpendicular grids (Figure 2a,c); 150 cm long and 30 cm deep in the dense agroforestry system (Figure 2b) and 500 cm long and 30 cm deep in the AIC agroforestry system (Figure 2d). Each mesh has 18 measuring points, arranged in three rows and six columns, separated by 10 cm and 30 cm, respectively. Due to the limitations of the measuring equipment and because it has been observed that the major lateral roots of species similar to those in the study are located within the top 30 cm of soil [41], moisture measurements were taken at depths of 10, 20, and 30 cm. Three measurements were taken for each agroforestry system, one for each tree species; in the dense system (Ar), measurements were taken under the cover of four adjacent trees of the same species (Figure 2a), while in the agroforestry system intercropped with crops (AIC), they were taken under the cover of two adjacent trees of the same species (Figure 2c). Moisture was measured in the control system in the center of the plot using two rectangular and perpendicular meshes; 120 cm long horizontally by 30 cm thick, with measuring points every 30 cm horizontally and every 10 cm deep.

To relate soil moisture behavior and potential aquifer recharge to precipitation, precipitation was recorded during the rainy season (summer) using a Davis Vantage pro-2^® (Davis Instruments Corp., Hayward, CA, USA), weather station, model 6162, located less than two kilometers from the studied systems. The station has a discharge dipstick every 0.25 mm, with an accuracy of ±4% for rainfall intensities up to 50 mm h⁻¹ and ±5% for rainfall intensities greater than 50 mm h⁻¹.

2.5. Soil Moisture Distribution Uniformity

Christiansen’s uniformity coefficient was calculated to quantify soil moisture uniformity in the agroforestry systems and the control system:

$$C{U}_C=\left[1-{\displaystyle \frac{{\sum }_{i=1}^n\left|{\theta }_i-\theta \right|}{n\theta }}\right]\times 100$$

(1)

where $C{U}_C$ is the Christensen uniformity coefficient, $\theta$ is the average soil moisture, ${\theta }_i$ is the average soil moisture of the i-th measurement site, and $n$ is the number of measurement sites [42].

2.6. Potential Aquifer Recharge Depth

The potential aquifer recharge depth was calculated using the following equation:

$$LRPA=\left({\theta }_{obs}-{\theta }_{cc}\right)\times 1cm | {\theta }_{obs}\ge {\theta }_{cc}$$

(2)

where $LRPA$ is the potential aquifer recharge depth (cm), ${\theta }_{obs}$ is the average observed soil moisture, and ${\theta }_{cc}$ is the soil moisture at field capacity. ${\theta }_{obs}$ was obtained by averaging the soil moisture under each vegetation cover, and ${\theta }_{cc}$ was obtained from the texture and organic matter content [43].

2.7. Statistical Analysis

Soil moisture content was assessed at three depths (10, 20, and 30 cm) under seven vegetation covers (six from the agroforestry systems and the control), for five dates, averaging the 12 replicates at each depth and vegetation cover. To evaluate the effect of vegetation cover and soil depth on soil moisture content, a completely randomized design with an asymmetric two-factor combinatorial arrangement and Tukey multiple comparison with a p-value < 0.05 was used. Factor 1 was soil depth with three depth levels, and Factor 2 was vegetation cover with seven cover levels. Calculations were performed using Minitab 18 Statistical Sofware [44].

3. Results and Discussion

3.1. Characteristics of Tree Species

The tree species included in each agroforestry system presented distinct structural attributes that likely influenced their respective hydrological performances (

Table 1. Means and standard deviations of the dendrometric measurements of the study species.

Species	High (m)	Crown Diameter (m)	Cover (%)
Prosopis leaviagata (Ar)	5.00 ± 1.00	5.23 ± 0.72	62.40 ± 5.87
Bursera graveolens (Ar)	4.34 ± 0.50	4.04 ± 1.68	73.80 ± 4.12
Leucaena esculenta (Ar)	3.90 ± 0.65	3.83 ± 1.21	75.20 ± 3.84
Prosopis leaviagata (AIC)	4.30 ± 0.44	5.70 ± 0.58	58.05 ± 3.69
Lysiloma divaricatum (AIC)	4.70 ± 1.15	6.25 ± 1.04	71.05 ± 8.22
Hesperalbizia occidentalis (AIC)	6.00 ± 0.61	7.51 ± 1.06	71.65 ± 10.77

). In the Ar system, Bursera graveolens and Leucaena esculenta exhibited high canopy coverage values (73.80% and 75.20%, respectively), despite having moderate crown diameters and heights (4.04 m and 3.83 m in diameter; 4.34 m and 3.90 m in height). These values suggest a compact, closed canopy that favors shading and reduces evaporation, but may also limit uniform rainwater infiltration due to higher interception [45,46].

In contrast, the AIC system included species such as Hesperalbizia occidentalis and Lysiloma divaricatum, which reached greater heights (6.00 m and 4.70 m) and have broader crown diameters (7.51 m and 6.25 m). These traits, combined with intermediate canopy coverage (~71%), suggest that the AIC system may allow more rain to reach the soil surface while still reducing evapotranspiration through partial shading [47]. Notably, Prosopis laevigata presented lower canopy coverage in the AIC system (58.05%) than in the Ar system (62.40%), reinforcing the hypothesis that the spacing and vertical structure in AIC favor infiltration and reduce canopy interception losses [48].

These morphological differences between the systems help explain the higher soil moisture and recharge values observed in the AIC plots. Therefore, beyond the presence of vegetation, it is the structural configuration and species-specific architecture that modulate hydrological function, supporting the central hypothesis of this study [49].

3.2. Soil Characteristics

Figure 3 shows the soil profiles of the three production systems, which illustrate the structural differences influenced by management practices. In the soils under agroforestry systems, a better aggregate structure was observed, characterized by angular aggregates in the A horizon and columnar structures in the B horizon. In contrast, the control system exhibited a more fractured structure, associated with long-term cultivation [50]. According to USDA classification [43], the soil texture in both A and B horizons of the three systems was clay, with values of up to 70.55% clay in the A horizon (

Table 2. Texture and thickness of soil horizons of the studied agroforestry systems.

Horizon	Thickness, Particle Content, and Texture	System
		Dense Agroforestry (Ar)	Intercropped Agroforestry (AIC)	Control
A	Thickness (cm)	14.00	4.00	Close to zero
	Clay (%)	70.55	66.95	71.95
	Silt (%)	17.50	20.00	22.20
	Sand (%)	11.95	13.05	5.55
	Texture	Clay	Clay	Clay
B	Thickness (cm)	26.00	28.00	30.00
	Clay (%)	44.45	64.45	56.96
	Silt (%)	15.00	20.00	22.50
	Sand (%)	40.55	15.55	20.55
	Texture	Clay	Clay	Clay

). These findings are consistent with the classification of these soils as vertisols by González et al. [23].

The organic carbon content and bulk density varied between systems and horizons. The Ar system exhibited the highest organic carbon content in the A horizon (2.61%), followed by the AIC system (1.30%) and the control (1.04%). In the B horizon, values were lower across systems: 0.52% in Ar, 0.78% in AIC, and 0.26% in the control. Correspondingly, total porosity in the A horizon was 33.23% for Ar, 32.68% for AIC, and 28.90% for the control. In the B horizon, values were 30.01% in both Ar and control, and 32.01% in AIC.

A moderate-to-strong positive correlation ($r=0.68$) was found between organic carbon content and total soil porosity across both A and B horizons. This relationship supports the hypothesis that increased organic matter contributes to improved soil structural properties, particularly in semi-arid environments where aggregation and pore formation are strongly influenced by biological activity and organic inputs [51,52,53]. The higher porosity observed in the Ar and AIC systems may be attributed to the continuous deposition of organic matter via litterfall and root turnover, which improves aggregation and creates biopores [38,39,40]. In the control system, the relatively lower organic carbon but comparable porosity may reflect long-term tillage effects or structural alterations due to continuous cropping [39].

The decomposition of organic matter also leads to the formation of angular aggregates, particularly in the A horizon, enhancing porosity and water retention capacity. This effect is further supported by field observations and comparative studies, such as those conducted in the Tarim River Basin, China, where higher vegetation cover was linked to greater soil carbon and aggregate stability [38]. Moreover, the differences in porosity and organic carbon are directly reflected in soil moisture dynamics, as will be discussed in the following section.

These findings reinforce previous work emphasizing the central role of soil organic matter in water regulation under agroforestry systems in semi-arid regions [51,52,53]. Nevertheless, the magnitude of the correlation and the variability among systems suggest that these results, while significant, should not be overgeneralized to all semi-arid agroecosystems without considering differences in soil mineralogy, climate variability, and land-use history.

In addition to the structural indicators described, the hydraulic behavior of the soils was also evaluated (

Table 3. Hydraulic parameters of the soil horizons of the studied agroforestry systems.

Horizon	Parameter	System
		Dense Agroforestry (Ar)	Intercropped Agroforestry (AIC)	Control
A	${\theta }_s \left({\mathrm{cm}}^3{ \mathrm{cm}}^{-3}\right)$	0.539	0.536	0.553
	${\theta }_{cc}\left({\mathrm{cm}}^3{\mathrm{cm}}^{-3}\right)$	0.453	0.453	0.449
	${\theta }_{pmp}\left({\mathrm{cm}}^3{\mathrm{cm}}^{-3}\right)$	0.347	0.348	0.345
	$D_a \left({\mathrm{g} \mathrm{cm}}^{-3}\right)$	1.220	1.230	1.190
	$K_s \left({\mathrm{mm} \mathrm{h}}^{-1}\right)$	1.460	1.320	2.490
B	${\theta }_s \left({\mathrm{cm}}^3{ \mathrm{cm}}^{-3}\right)$	0.462	0.529	0.518
	${\theta }_{cc}\left({\mathrm{cm}}^3{\mathrm{cm}}^{-3}\right)$	0.388	0.455	0.458
	${\theta }_{pmp}\left({\mathrm{cm}}^3{\mathrm{cm}}^{-3}\right)$	0.269	0.349	0.350
	$D_a \left({\mathrm{g} \mathrm{cm}}^{-3}\right)$	1.430	1.250	1.280
	$K_s \left({\mathrm{mm} \mathrm{h}}^{-1}\right)$	0.990	0.960	0.730

). The moisture content at saturation (${\theta }_s$), at field capacity (${\theta }_{cc}$), and at the permanent wilting point (${\theta }_{pmp}$), the apparent density ($D_a$), and the saturated hydraulic conductivity ($K_s$) were estimated according to the pedotransfer functions proposed by Saxton and Rawls [42]. In the A horizon, the two agroforestry systems and the control displayed similar ${\theta }_s$, ${\theta }_{cc}$, and ${\theta }_{pmp}$ values due to their shared clayey texture. However, $K_s$ differed notably: the control system exhibited the highest value ($~2.5 {\mathrm{cm} \mathrm{h}}^{-1}$), while the agroforestry systems showed a lower $K_s$ ($~1.3 {\mathrm{cm} \mathrm{h}}^{-1}$), possibly due to structural differences such as cracking under conventional tillage [50].

In the B horizon, the dense agroforestry system showed the highest $K_s$ ($~1.0 {\mathrm{cm} \mathrm{h}}^{-1}$), despite its higher bulk density ($~1.43 {\mathrm{g} \mathrm{cm}}^{-3}$) and slightly lower porosity, while the control system had the lowest $K_s$ ($~0.7 {\mathrm{cm} \mathrm{h}}^{-1}$). This suggests that, even under compacted subsurface conditions, agroforestry practices enhance soil hydraulic behavior, likely due to improved aggregation and biopore networks.

Although all soils share the same textural class, the observed variations in $K_s$ and $D_a$ highlight structural differences induced by organic matter inputs and vegetation cover. The agroforestry systems, especially the dense one, maintained higher levels of organic carbon, which contribute to greater aggregate stability and infiltration. These effects are not accounted for by the pedotransfer functions used, which omit the role of structure and vegetation. Thus, while model-based estimates are useful, they likely underestimate the functional improvements introduced by agroforestry. The benefits are directly observable in soil moisture data, which will be analyzed in the next section.

3.3. Behavior of Soil Moisture Content

Figure 4 shows the behavior of the average daily moisture in the soil under different vegetation covers, for the two agroforestry systems and the control system, during the rainy season from 3 August to 12 October 2022. In general, in the control soil the moisture content was lower than in the soils under the different vegetation covers of the two agroforestry systems. Also, higher moisture contents were observed in the soil of the dense Ar agroforestry system than in the soil of the AIC agroforestry system. The soils reacted with an increase in moisture to the rain event of 15 August, the largest of the period analyzed; in all cases they exceeded the moisture value corresponding to field capacity (${\theta }_{cc}$), with a longer period in the soils to the two agroforestry systems than in the control soil, but also the moisture content in the soils of the agroforestry systems exceeded the value corresponding to the saturation value (${\theta }_s$) obtained from Saxton and Rawls [43].

As seen in Figure 4, the behavior of moisture in the soil is not consistent with the presence of rain on some dates. The above is explained because in some cases moisture measurement days (3 and 17 August, 15 and 28 September, and 12 October) did not come immediately after a rain event, so its immediate effect is not shown. In this way, on 3 August, an increase in moisture appears because there was precipitation that day and on previous days, then the increase was sustained until reaching a maximum value in the measurement taken on 17 August due to the precipitation on that day and the previous days (15 and 16 August). The moisture subsequently decreased because it was measured until 15 September, and the effect of the precipitation from 5 to 8 September was not detected, since a considerable period of time occurred between the dates of moisture measurement. Then, the moisture decreased, because until the next moisture measurement date (28 September) there was no precipitation. Finally, an increase in moisture was recorded on 12 October due to the rainfall on 5 and 6 October.

In the dense Ar system, Bursera graveolens cover had the highest moisture content, followed by Leucaena esculenta and Prosopis leaviagata. In the AIC system, the soil under Prosopis leaviagata cover was also the one with the lowest moisture. The moisture content of the soil under Bursera graveolens cover was higher than the moisture under Prosopis leaviagata cover because Bursera graveolens cover is the second densest after Leucaena esculenta cover, and also the trees are short, which favors the reduction of evapotranspiration [54]. In contrast, the moisture content under Prosopis leaviagata cover was the lowest of all covers, largely because its cover is lower, allowing more solar radiation to pass through, resulting in greater evapotranspiration and greater soil moisture loss [55]. In the AIC system, being more open and having greater exposure to the sun, the trend is that the moisture content in the soil was lower compared to the denser Ar system, in which more moisture is retained, which could be advantageous or disadvantageous depending on the intended application of the system [56,57].

3.4. Uniformity of Moisture Distribution in the Soil

Figure 5 shows the average uniformity coefficient of the distribution of the soil moisture for all vegetation cover in the three production systems between 4 August 2022 and 12 October 2022, demonstrating significant statistical differences in the three systems studied. The dense agroforestry moisture (Ar) under Prosopis leaviagata cover and under Bursera graveolens cover did not present statistical differences with the control soil, but they did present differences with the moisture under the Leucaena esculenta cover in Ar system and with the moisture under the cover of the three species in the AIC system. The moisture under Leucaena esculenta cover in the Ar system did not present significant statistical differences with Prosopis leaviagata and Herperalbizia occidentalis cover in the AIC system, but did with Lysiloma divaticatum in AIC system. In the Ar system, Bursera graveolens cover presented the greatest uniformity of soil moisture distribution (79.56%), followed by Prosopis leavagata (76.97%) and Leucaena esculenta (74.60%, which has some correspondence with its dendrometric measurements (Table 1); Bursera graveolens, being a species with a larger leaf size than the other species in the Ar system, exhibits greater buffering of soil moisture [27]. In the intercropped agroforestry system (AIC), Lysiloma divaricatum presented the greatest uniformity of moisture distribution at 81.99%, followed by Herperalbizia occidentalis at 74.29% and the Prosopis leaviagata at 72.75%; these data show some correlation with the vegetation cover and morphology results presented in Table 1 [24,29]. The highest values were found for Lysiloma divaricatum cover in the AIC system and Bursera graveolens cover in the Ar system, which can be associated with the fact that they are species that present the greatest cover and crown diameter, thereby reducing evaporation from the surface layer of the soil. In the control system, the distribution uniformity was 77.98%. This result, which is greater than in the soils under some species cover in the two agroforestry systems, may be due to the fact that the corn crop had high cover during the study period, preventing evaporation of the surface layer of the soil, thereby promoting greater vertical moisture uniformity. In general, the average moisture distribution uniformity in the soil was similar in the three production systems: 76.88% in the Ar system, 76.34% in the AIC system, and 77.08% in the control system, with a standard deviation of 0.38%.

The results in Figure 4 and Figure 5 show that the species of Bursera graveolens in the Ar system and Lysiloma divaricatum in the AIC system are the ones that generate more water retention in the soil and with a more uniform distribution than the rest of the tree species and the control.

3.5. Potential Recharge to the Aquifer

There is a significant difference for potential recharge on the dates of greatest precipitation, which were 15 August 2022 and 7 September 2022. The greatest potential recharge of water to the aquifer occurred under Bursera graveolens cover (25.98 cm cm⁻¹), followed by Leucaena esculenta cover (19.77 cm cm⁻¹) and Prosopis leaviagata cover in the Ar system (14.06 cm cm⁻¹). In the intercropped system (AIC), the highest potential water recharge to the aquifer occurred under Lysiloma divaricatum and Hesperalbizia occidentalis cover (with 23.88 and 23.47 cm cm⁻¹, respectively), followed by Prosopis leaviagata cover (21.45 cm cm⁻¹). Bursera graveolens, Lysiloma divaricatum, and Herperalbizia occidentalis cover are those that generated the greatest potential recharge to the aquifer, and the least recharge occurred in the control plot that does not have tree cover. In general, the AIC system reached the highest average potential recharge to the aquifer with 22.93 cm cm⁻¹, followed by the Ar system with 19.93 cm cm⁻¹, and the control system with 9.64 cm cm⁻¹ (

Table 4. Potential recharge to the aquifer (cm of water per cm of soil thickness) of the soil in the studied systems. Tukey means from highest to lowest: a, b.

Cover	$\mathbf{Recharge} (\mathbf{c}\mathbf{m} \mathbf{c}{\mathbf{m}}^{-1})$
	15 August 2022	7 September 2022
Prosopis leaviagata Ar b	14.06	0.00
Bursera graveolens Ar a	25.98	6.03
Leucaena esculenta Ar a	19.77	0.00
Prosopis leaviagata AIC a	21.45	0.00
Lysiloma divaricatum AIC a	23.88	0.00
Hesperalbizia occidentalis AIC a	23.47	0.00
Control b	9.64	0.00

).

3.6. Relationship Between Moisture Behavior and Vegetation Cover

The statistical analysis indicated that there is no dependence between the main factors of soil depth and vegetation cover that explains the soil moisture content, for all measurement dates. The phenomenon is complex, since there is no linear relationship between the density of the vegetation cover and the moisture content, except in some of the species studied. Such behavior was observed by Yang et al. [58], as they found an inverse relationship in the moisture content in the soil and the vegetation cover and discovered that there was a higher moisture content in the dry season that in the wet season. The moisture content was different both for the three depths, as well as for the seven vegetation covers evaluated (p-value < 0.05). For the first date, the highest moisture content was under the Lysiloma divaricatum and Herperalbizia occidentalis cover in the AIC system and at a soil depth at 10 cm for all the systems studied, possibly due to some precipitation event that occurred before data collection (Figure 6a). For the second date, the highest moisture content was found under Bursera graveolens cover in the Ar systems, while there were no differences in the three levels of soil depth (Figure 6b). This is because soil moisture content becomes homogenized throughout the soil profile after a rainfall event [50]. For the third date, the highest moisture retention was achieved under the Ar system (Figure 6c), since the agroforestry arrangement is denser and prevents greater evapotranspiration than what can occur in the AIC system and the control [59]. For the fourth date, the highest moisture retention was reached under Bursera graveolens and Lysiloma divaricatum covers in the AIC and Ar systems, at a soil depth of 30 cm (Figure 6d). For the fifth and last date, the highest soil moisture content was reached under Bursera graveolens and Lysiloma divaricatum covers, in the Ar and AIC systems respectively, and at a depth of 30 cm (Figure 6e). This is because weeks after the precipitation event the water moves to a greater depth [60]. In general, in two of the plant species with the highest plant cover (Bursera graveolens in the Ar system and Lysiloma divaricatum in the AIC system), the highest moisture contents were present, and in one of the species with the lowest plant cover (Prospis leaviagata), the lowest moisture values were observed. In sand dunes in China they observed an inverse behavior, as they observed an inverse correlation between the degree of vegetation cover and moisture content [58]. In our study, in general, higher moisture was observed at greater depths, but a reverse situation was observed in sand dunes in China [58].

4. Conclusions

Agroforestry systems have been shown to enhance soil structural properties and water regulation functions under semi-arid agricultural conditions. A positive correlation exists between soil organic carbon and porosity, and improved hydraulic behavior is observed in agroforestry-managed soils, indicating that integrating woody vegetation into cropping systems can promote conditions favorable for aquifer recharge. As compared to conventional rainfed maize fields, dense agroforestry and an intercropped system improved soil water retention and infiltration.

This research has important practical implications: agroforestry can help alleviate aquifer depletion in regions that are experiencing overexploitation of groundwater. In addition to improving soil porosity and water storage capacity, agroforestry practices contribute to broader hydrological sustainability at the landscape level. Evidence indicates that agroforestry can complement existing groundwater management policies in semi-arid regions.

In terms of policy, it is recommended that agroforestry practices be incorporated into local and regional water conservation programs, with incentives for farmers to adopt diversified vegetation structures. Additionally, land-use planning strategies should integrate agroecological models that emphasize ecosystem services in addition to crop yields.

Future research should involve more frequent measurements of soil moisture at greater depths, as well as measuring water retention by the canopy and evapotranspiration. This would aim to refine the understanding of the relationship between precipitation characteristics, vegetation, and soil, enabling the development of mathematical models for predictive purposes.