Introduction

Hydraulic conductivity, a critical soil physical property, significantly affects water movement through soil pores and plays a central role in regulating soil–water interactions, nutrient transport, and ecosystem functioning1,2. Among its forms, quasi-saturated hydraulic conductivity (Kqsat) is particularly relevant for understanding near-saturated flow conditions, which are important for modeling infiltration, drainage, and soil erosion—especially in fragile ecosystems like the Himalayan region3,4.

Accurate estimation of Kqsat under varying land use systems is essential, particularly in ecologically sensitive regions such as the Northwest Himalayas. This region, characterized by diverse topography, vegetation cover, and land management practices, includes agricultural fields, barren lands, and forest ecosystems dominated by Deodar (Cedrus deodara) and Oak (Quercus spp.). Each land use type influences soil structure and composition differently, thereby modifying water retention and transmission properties. Agricultural practices in the region often involve terracing, hand tillage, and application of organic manure5,6,7. These activities initially improve surface porosity but may later cause compaction due to repeated disturbance. In contrast, barren lands, often a result of deforestation or prolonged fallow conditions, lack protective vegetative cover, leading to poor aggregation, crust formation, and reduced infiltration8.

Among forested systems, vegetation type and rooting depth are important determinants of hydraulic behavior. Deodar trees, with deep taproots, improve vertical permeability and subsoil structure9. Oak trees have fibrous root systems that stabilize surface layers and enhance lateral water movement. Plant species influence hydrological behavior through these distinct root morphologies. Fibrous roots (e.g., in grasses and shallow-rooted crops) increase near-surface aggregation, while taproots facilitate deep water infiltration and reduced surface runoff10. Despite advances in measuring soil hydraulic properties, quantifying Kqsat remains difficult due to species-specific root architectures and complex plant-soil interactions. Root systems modify soil porosity, stability, and hydrophobicity in non-uniform and often unpredictable ways. This variability poses challenges in generalizing findings across heterogeneous land use systems11.

Several studies have demonstrated that vegetation restoration improves soil hydrological and structural conditions. Recent work has emphasized the role of vegetation in enhancing soil aggregate stability, bulk density reduction, and porosity. Zhang et al.12 reported increased Ksat in forest-restored hillslopes, Li et al.13 highlighted improved moisture retention due to root-soil interactions, and Thakur et al.14 documented enhanced infiltration under tree canopy cover. These studies collectively underscore the functional role of vegetation in regulating water movement through soils.

Saturated hydraulic conductivity (Ks), field capacity (FC), and saturated water-holding capacity (SWHC) are widely accepted as diagnostic indicators of soil quality and erosion potential15. However, most existing studies focus predominantly on surface soils (0–5 cm) and neglect deeper profiles, where root activity and water transport also play critical roles16. Subsoil hydraulic properties are essential for understanding plant water uptake and deep drainage processes17. Given the spatial variability and measurement challenges, there is growing interest in using multivariate statistical tools and predictive modeling to estimate Kqsat. Constrained ordination techniques such as redundancy analysis (RDA) and variance partitioning analysis (VPA) allow for the identification of key explanatory variables, while regression-based models enable extrapolation across landscapes18. These tools offer efficient, scalable alternatives to conventional field measurements19.

To our knowledge, this is the first study to integrate constrained ordination and predictive modelling for assessing quasi-saturated hydraulic conductivity across diverse land use systems in the Northwest Himalayas. The analysis aims to identify dominant physico-chemical drivers of Kqsat and provide scalable models for sustainable soil and water management. Assumptions made in this study include: (i) soil samples are representative of each land use type; (ii) vegetation cover and soil properties remained stable during sampling; (iii) point-scale measurements can be extrapolated to landscape scale under similar conditions; and (iv) transient weather effects did not significantly alter hydraulic measurements during the study period.

Materials and methods

Experimental site

This research was conducted within the Semalta watershed, which is situated between Longitude 77°53′00″ E and Latitude 30°38′00″ N in the North-West Mid Himalayan region of India. The region comprises typical gully formations and loess hilly terrain. The climate of the region is characterized as a humid sub-tropical climate, with cold winters and hot summers. The watershed has an annual average temperature of 12.5 °C, with mean monthly maximum and minimum temperatures recorded at 29.2 °C and 13.9 °C, respectively. May is the hottest month, while January is the coldest. The area receives an average annual rainfall of 1615 mm, with approximately 81% occurring during the monsoon season (June–September). April and May are the driest months of the monsoon season. According to Singh et al.15, the soil at the experimental site is classified as fine mixed hyperthermia Typic Udorthents.

Experimental design and soil sampling

Soil samples were obtained from four different land use systems, which comprised of barren land, cultivated land, oak forest, and deodar land. The cultivated area primarily utilized finger millet-fallow, pulse-fallow, and finger millet-lentil cropping systems, as well as inter/mixed cropping. No chemical fertilizers were used in this area, and the soil was prepared through hand sowing with farm yard manure (FYM). The oak forest was over 90 years old and consisted primarily of pine (Pinus roxburghii) and oak (Quercus spp.) trees, with no recorded history of disturbances. The grass-dominated rangelands were characterized by a diverse range of grass species. The barren and/or fallow land, which had been unutilized for over 15 years, was considered as private property. The forest land was owned by the government, while the community owned and utilized the grass land and barren land. In the forest land use systems, the dominant species were Cedrus deodara (Deodar) and Quercus leucotrichophora (Banj Oak), each exerting distinct influences on soil hydraulic properties. Cedrus deodara exhibits a deep taproot system that enhances vertical water movement and subsoil porosity, facilitating higher quasi-saturated conductivity in deeper layers. In contrast, Quercus leucotrichophora has a fibrous root system concentrated in the upper soil horizon, which increases surface aggregation and improves water retention near the root zone. The contrasting rooting depths and architectures of these species result in spatial variation in Kqsat, with Deodar forests promoting greater infiltration at depth and Oak forests enhancing moisture retention in surface soils. These species-specific effects are critical in explaining the observed differences in hydraulic behavior across forest plots. The distinguishing characteristics of each land use system, including dominant vegetation, slope, elevation, and disturbance history, are summarized in Table 1.

Table 1 Summary of land use systems in the watershed.
Full size table

Soil sampling and analysis

A stratified random sampling approach was adopted to capture spatial variability across four land use systems—Deodar forest, Oak forest, agricultural land, and barren land20. Ten samples were collected in each land use to analyze the soil parameters. Soil cores and excavated samples were collected from beneath the disk after each set of infiltration sites to determine the bulk density (BD) and soil aggregate mean weight diameter in the topsoil (0–15 cm) and subsoil (15–30 cm). The bulk density of soil was determined using a stainless-steel core method, which had a 5 cm internal diameter and a 5 cm height. The collected soil samples from each location of the experimental sites were oven-dried at 105 °C for 24 h to obtain the dry soil weight. The recorded dry soil weight was divided by the core volume to obtain the bulk density of soil21. In the laboratory, soil moisture was measured by mass loss from wet soil samples during oven drying (105 °C) to constant weight. Field capacity values were derived using a method developed by Qiu et al.22.

Soil aggregate analysis was conducted through the use of wet sieve methods on a Yoder apparatus, as per the methods outlined by Yoder23. The soil samples were dried, with the inclusion of any loose plant roots, leaves, and gravel or pebbles being carefully removed. The soil samples were then passed through a 10 mm sieve and retained on an 8 mm sieve. Next, 100 g of dry soil samples were uniformly spread across a set of sieves with diameters ranging from 0.125 to 4.0 mm. The samples were allowed to sit on the wet sieves for 25 min to undergo capillary re-wetting, after which a nest of sieves was submerged in a water drum and shaken for 30 min using an electric motor and pulley system. The water-stable aggregates were then collected separately in different size ranges, from 0.0–0.125 mm to 4.0–8.0 mm, and were oven-dried at 65–70 °C until they reached a constant weight. The mean weight diameter (MWD) was calculated using the formula provided by Kemper and Rosenau24.

Soil samples were obtained from the vicinity of each infiltration site at depths of 0–15 cm and 15–30 cm for the purpose of determining soil particle size distribution, soil organic carbon (SOC), and soil pH. The samples were air-dried and gently ground to pass through a 2 mm sieve for further analysis. The method used to determine particle size distribution was described by Piper25. The USDA system was utilized to determine the textural class of the soil. SOC was determined using the Walkley and Black method26, and soil pH was determined using a digital pH meter.

Quasi-saturated hydrologic conductivity measurement

The mini disk infiltrometer (MDI) has a practical length of 32.7 cm and requires 135 mL of water for field measurements of unsaturated hydraulic conductivity (K) when access to water is limited. This device consists of two chambers, with the top chamber regulating the suction applied to the porous sintered stainless steel disk, while the lower chamber contains the infiltrating water volume. The suction ranges from 0.5 to 7 cm, as per the method developed by Zhang27. The cumulative infiltration is used to determine K, which is then plotted on a graph with the x-axis representing the square root of time and the y-axis representing the cumulative infiltration. The data points are then fitted with a function28.

$${\text{I }} = {\text{ C}}_{{1}} {\text{t}}^{{{1}/{2}}} + {\text{ C}}_{{2}} {\text{t}}$$
(1)

where I is the cumulative infiltration (cm), C1 is the soil sorptivity (cm/s1/2), C2 is parameter (cm/s) related to K and t is time (s). The hydraulic conductivity of the soil (K (cm/s)) is computed using Eq. (2)

$${\text{K }} = {\text{C}}_{{2}} /{\text{A}}_{{2}}$$
(2)

where C2 is the slope of the curve of the cumulative infiltration versus square root of time and A2 is a value calculated using Eq. (3).

$$A_{2 } = \frac{{11.65\left( {n^{0.1} - 1} \right){\text{exp}}\left[ {c\left( {n - 1.9} \right) \propto h_{0} } \right]}}{{\left( { \propto r_{0} } \right)^{0.91} }}$$
(3)

(c = 2.92 if n ≥ 1.9; c = 7.5 if n < 1.9).

where n and α are van Genuchten parameters determined for one of the 12 soil texture classes29, r0 is the disk radius, h0 is the suction at the disk surface and c is a constant.

Method developed by Kutilek and Nielsen30 was applied to estimate Ks (cm/s) from MDI measurements based on Eq. (4).

$${\text{I }} = {\text{ C}}_{{1}} {\text{t}}^{{{1}/{2}}} + {\text{ mK}}_{{{\text{st}}}}$$
(4)

Where C1 (cm/s1/2) is sorptivity and m = 0.667.

Constrained ordination-based approach and predictive modelling to study quasi-saturated hydraulic conductivity under different land use systems.

Statistical analysis

Two-way analysis of variance (ANOVA) was conducted to evaluate the effects of various land-uses and soil depths on soil hydrologic and physico-chemical properties. Tukey’s honest significant difference (HSD) test was applied for post hoc analysis of average values of soil properties across different land-use types at the same or varying soil depths. Redundancy analysis (RDA) was employed to determine the relationships between explanatory variables, such as AMWD, EC, pH, BD, TP, SOC, Sand, Silt, and Clay, and study variables, including Field Capacity, Soil Moisture, and Quasi-saturated hydraulic conductivity. Pearson’s correlation coefficient analysis was used to estimate pairwise correlations. Log-transformation was performed on the physico-chemical variables to reduce heteroscedasticity in RDA. Standardization of explanatory variables was conducted to transform different scales into a 0 to 1 range. The Monte Carlo permutation test was used for the relative significance of each explanatory variable. Variance partitioning analysis (VPA) was carried out to assess the proportional contribution of explanatory variables to the change in soil hydrological parameters. Multiple linear regression (MLR) was applied to generate pseudo-pretransfer functions (PTFs). The final MLR models were developed through iterative selection of independent variables via stepwise regression. The multicollinearity diagnostic test was performed before MLR development using the variance inflation factor (VIF). All statistical analyses were executed using RStudio (R version 4.2.1). The R package “vegan” was utilized for RDA and VPA.

Results and discussion

Soil physio-chemical properties

Aggregate mean weight diameter (AMWD)

The investigation of soil aggregate stability, as measured by AMWD, revealed considerable variations across various depths and land utilization practices (Table 2). In the 0–15 cm depth range, agricultural soils demonstrated a higher AMWD (1.14 ± 0.04 mm, p < 0.05) when compared to barren land (0.69 ± 0.02 mm), deodar (1.49 ± 0.07 mm), and oak (1.56 ± 0.04 mm). This pattern persisted in the 15–30 cm depth range, with agricultural soils exhibiting a higher AMWD (0.95 ± 0.04 mm, p < 0.05) than barren land (0.71 ± 0.03 mm), deodar (1.44 ± 0.07 mm), and oak (1.42 ± 0.04 mm). The observed differences in AMWD indicate variations in soil structure and aggregation influenced by both depth and land use. Deodar and oak lands consistently displayed superior aggregate stability, emphasizing the positive impact of tree cover on soil structure. This difference may be attributed largely to the characteristics of the root systems associated with these species. Forest soils with greater root biomass, length, and density—particularly those with fibrous root systems—showed higher AMWD values, indicating improved soil aggregate stability. Fibrous roots, such as those of oak, are concentrated in the upper soil horizon and play a crucial role in promoting surface aggregation and enhancing water retention near the root zone. These roots physically bind soil particles and release organic exudates that foster aggregate formation and stabilization, resulting in a generally positive relationship between root development and MWD31,32. This relationship supports previous findings that robust root networks contribute to improved soil structure and resilience33. The dense network of fine roots in oak not only increases aggregate stability but also influences soil hydrology by improving the soil’s capacity to retain moisture in the upper layers. This, in turn, supports greater water availability for plants and may slow down water movement, promoting infiltration and reducing surface runoff34. Such effects highlight the importance of root architecture in modulating both soil physical properties and hydrological processes35. The effects of land use and soil depth on soil hydrologic and physicochemical properties are pronounced in montane ecosystems. Different land uses and plant species influence soil quality through distinct patterns of organic matter input, root distribution, and disturbance regimes36. Assessing multiple soil depths is essential, as root activity, organic matter content, and microbial processes typically decline with depth, affecting both surface and subsoil contributions to overall soil health37. In this study, the upper soil layers (0–15 cm) under forest cover consistently exhibited higher organic carbon, aggregate stability, and water retention compared to deeper layers and to soils under agriculture or barren conditions.

Table 2 Mean ± standard error of soil physio-chemical properties under different land use and soil depth, including ANOVA and post hoc test results.
Full size table

Soil texture and composition

Texture analysis indicated notable distinctions in the percentages of sand, silt, and clay across various land uses (Table 2). Deodar and oak lands exhibited increased sand content, while barren land displayed elevated clay percentages at both 0–15 cm and 15–30 cm depths. Agricultural lands showed intermediate values, suggesting a harmonious soil composition. These results suggest that land use practices contribute to disparities in soil texture, which may impact water retention, drainage, and nutrient availability38. The higher clay content in barren land may indicate increased susceptibility to compaction and reduced water infiltration39.

Chemical characteristics

The evaluation of soil chemical properties exposed distinct patterns in soil organic carbon (SOC) content and electrical conductivity (EC) across various land uses (Table 2). Consistently, deodar and oak lands displayed higher SOC values as compared to agricultural and barren lands at both depths (p < 0.05). Moreover, EC values were generally greater in deodar and oak lands. The elevated levels of SOC in tree-covered lands suggest the role of vegetation in the accumulation of organic matter. The higher EC values may indicate differences in nutrient availability or ion concentrations, reflecting the impact of vegetation types on soil chemistry. These findings align with the presence of continuous litterfall and reduced disturbance in forested systems, which promote organic matter buildup and microbial activity40. In contrast, frequent cultivation and residue removal in agricultural lands likely reduce SOC content and contribute to lower EC values due to nutrient depletion. Barren lands, lacking vegetation cover, exhibited the poorest chemical quality, further underscoring the role of plant inputs in maintaining soil fertility. The depth-wise decline in SOC across all land uses reflects reduced organic input and biological activity in subsoil layers6,7. Moreover, tree species play a complex and influential role in soil organic carbon (SOC) accumulation and overall soil quality. Species such as deodar and oak contribute to higher SOC through continuous litterfall, root turnover, and rhizodeposition, which increase organic matter inputs and stimulate microbial activity41. Elevated SOC enhances soil aggregation, nutrient cycling, and moisture retention, all of which are critical for maintaining high soil quality. The intricate interactions between plant species, SOC, and soil quality underscore the importance of diverse and well-managed forest ecosystems for sustaining soil health42.

pH and bulk density

The pH values displayed a consistent pattern across various land uses and depths, indicating a predominantly neutral to slightly acidic soil environment (Table 2). However, it is important to note that the barren land exhibited higher bulk density (BD) values at both depths (p < 0.05), which suggests soil compaction. Conversely, the Deodar and oak lands displayed lower BD values, indicating a more porous soil structure. The uniformity in pH values suggests that land use practices in the studied area have not significantly influenced soil acidity. The observed differences in bulk density emphasize the impact of land use on soil compaction, with barren land demonstrating an increased density compared to other land uses.

Soil hydrological properties

The investigation into soil hydrological properties, including field capacity, hydraulic conductivity, soil moisture, and total porosity, provides valuable insights into the water-holding capacity and drainage characteristics across different depths and land uses (Table 3).

Table 3 Mean ± standard error of soil hydrological properties under different land use and soil depth, including ANOVA and post hoc test results.
Full size table

Field capacity

In the 0–15 cm depth range, significant differences in field capacity were observed among land uses (Table 3). Deodar lands exhibited the highest field capacity (33.07 ± 1.46 cm), followed by oak lands (30.89 ± 1.06 cm) and agricultural lands (29.85 ± 1.08 cm). Barren land had the lowest field capacity (24.07 ± 0.77 cm). A similar trend was observed in the 15–30 cm depth range. These findings suggest that afforested lands, particularly deodar and oak, tend to retain higher water content at field capacity compared to agricultural and barren lands. This enhanced water retention in forested soils can be attributed to higher organic matter content, better aggregation, and improved pore continuity1,2. The relatively lower field capacity in barren land highlights the adverse effects of vegetation absence on soil structure and water-holding potential43.

Quasi-saturated hydraulic conductivity

Significant differences in hydraulic conductivity (K values) were evident across land uses and depths (Table 3). Deodar and oak lands consistently demonstrated higher K values compared to agriculture and barren land. In the 0–15 cm depth range, deodar lands had the highest K values (6.16 ± 0.33 cm, p < 0.01), while barren land had the lowest (1.24 ± 0.04 cm, p < 0.01). A similar pattern was observed in the 15–30 cm depth range. These results indicate that forested lands exhibit enhanced water permeability compared to agricultural and barren lands. The higher K values in forested soils reflect the influence of deep-rooted vegetation in maintaining pore connectivity and reducing compaction11. Root channels and biological activity in deodar and oak stands likely facilitate preferential flow paths, enhancing quasi-saturated conductivity44. In contrast, soil degradation and surface sealing in barren lands severely limit water transmission, increasing the risk of runoff and erosion15.

Soil moisture

The assessment of soil moisture content revealed significant variations across land uses and depths (Table 3). In both depth ranges, deodar lands consistently maintained higher soil moisture levels compared to other land uses. Agricultural lands exhibited intermediate soil moisture levels, while barren lands had the lowest soil moisture content. These findings suggest that afforested lands, particularly those with deodar and oak, contribute to enhanced soil water retention. This improved moisture retention in forested areas may be attributed to higher organic matter content and better canopy cover, which reduce evaporation losses and enhance infiltration45. In agricultural lands, periodic tillage and crop removal can disrupt soil structure, moderately reducing moisture retention. Barren lands, lacking both vegetation and organic inputs, exhibit poor water-holding capacity, making them more vulnerable to drought and surface runoff1,46.

Total porosity

Total porosity exhibited variations across land uses and depths, reflecting the overall pore space within the soil matrix (Table 3). Deodar and oak lands displayed higher total porosity compared to agriculture and barren land in both depth ranges. In the 0–15 cm depth range, deodar lands had the highest total porosity (0.49 ± 0.01, p < 0.01), while barren land had the lowest (0.41 ± 0.00, p < 0.01). A similar pattern was observed in the 15–30 cm depth range. These results emphasize the influence of land use on soil structure and porosity, with afforested lands exhibiting improved water-holding capacity. The greater porosity in forested soils likely results from enhanced root activity, organic matter accumulation, and minimal disturbance, which promote the development of stable macro- and micropores47. In contrast, compaction and limited biological inputs in barren and agricultural lands restrict pore development, reducing overall porosity. This structural difference plays a crucial role in regulating infiltration, aeration, and moisture dynamics across land use systems.

Relationships between soil hydrological properties and soil physio-chemical properties

The constrained ordination method RDA was used to identify the possible multivariate relationship between the soil hydrological and physicochemical parameters. Nine physio-chemical properties of soil (i.e., AMWD, EC, pH, BD, TP, SOC, Sand, Silt, and Clay) considered as explanatory variables and five soil hydraulic properties of soil (i.e., FC, SM and Kqs.5, Kqs2, Kqs5) was considered as response variables, and Fig. 1 displays the ordination diagram. Soil hydrological properties are displayed as red markers, whereas soil physicochemical values are represented by blue arrows. The percentages of total variance explained by each axis are shown by the values on the RDA axes. The length of the arrows or distance of the marker from the origin indicated how significant the association was between the hydrological and physicochemical parameters of the soil; the longer the arrow or larger distance of marker, the greater the association. The included angle of the arrows and RDA axes determined the importance of the correlation; the higher the correlation, the smaller the included angle. The positive and negative correlations between the RDA axes and soil attributes were analyzed depending on which quadrant the arrows are positioned in. Strong relationships were observed between soil hydrological properties (such as field capacity, hydraulic conductivity, and moisture content) and physicochemical properties (including aggregate stability, SOC, and bulk density). Higher aggregate stability and SOC were associated with increased field capacity and hydraulic conductivity, demonstrating the importance of organic matter and soil structure in enhancing water retention and movement7. In contrast, higher bulk density was negatively correlated with hydrological properties, indicating theadverse effect of compaction on soil water dynamics15.

Fig. 1
figure 1

Ordination biplot depicting the findings of RDA of soil physicochemical properties and hydrological properties. BD: bulk density, AMWD: aggregate mean weight diameter, TP: total porosity, SOC: soil organic carbon, FC: field capacity, pH: potential of hydrogen, EC: electrical conductivity, Kqs.5: psudo-saturated hydraulic conductivity at 0.5 cm., Kqs2: psudo-saturated hydraulic conductivity at 2 cm, Kqs 5: psudo-saturated hydraulic conductivity at 0.5 cm.

Full size image

Figure 2 illustrates scree plots of Redundancy Analysis (RDA) for Oak, Barrenland, Agriculture, and Deodar land types. Each plot shows inertia values, indicating the variance explained by each RDA axis. In all four plots, RDA1 has the highest inertia, approximately 2.5–3.5, indicating it accounts for the most variance. RDA2 also explains notable variance, with inertia values between 1.0 and 2.0. The remaining axes (RDA3, RDA4, RDA5) have minimal inertia, close to zero, contributing little to the overall variance. This pattern is consistent across all land types, indicating that the first two RDA axes are crucial in capturing primary ecological gradients and variation within these land use categories. The plots highlight RDA’s efficiency in summarizing ecological data, with a sharp inertia drop after the second axis, underscoring the first two axes’ dominance in the analysis.

Fig. 2
figure 2

Scree plot of RDA.

Full size image

The scree plot analysis of RDA evident that two redundant analysis axes (RD1 and RD2) have significant inertia gain (> 1), hence RD1 and RD2 were adequate to explain variance of the original response variables (Fig. 2).

The ordination diagram Fig. 1 revealed, RD1 and RD2 explained approximately 100% of the variability of the response variables (hydrological variables) in agriculture (99.7%), barren land (99.2%), oak (98.4%) and deodar (99.5%). The first two RDA (RD1 and RD2) axis explained variance of the response variables 76.7% and 22.95% in agriculture, 62.32% and 36.92% in barren land, 65.03% and 33.38% in oak and 74.57% and 24.94% in deodar, respectively. In oak land use, RDA1 mainly presented the changes of AMWD, Sand, Clay, SOC, and rest of the explanatory variables (TP, Silt, BD and pH) changes were presented by RD2. For barren land, RDA explain mostly variance of BD, Sand, Clay and TP, whereas RD2 explained AMWD, pH, Silt and SOC. AMWD, EC, BD, Silt, Clay and TP were RD1 were explained RD1 whereas, RD2 explain pH and sand in agriculture. Again, for deodar RD1 majorly captured the variance AMWD, EC, BD, Silt, clay, TP and rest of the explanatory variables were reelected by RD2. The Monte Carlo permutation test findings showed that, AMWD significantly affected the hydrological parameters of the soil for various land-use types (p-value 0.05). BD has significant effect on all the land uses except deodar. pH has significant effect on agriculture and EC has significant effect on agriculture and Deodar.TP has effect on barren land and agriculture. SOC only has effect on oak land use whereas clay has effect on barren land only. Further, pairwise relationships were depicted in Fig. 3. It revealed that AMWD, pH, EC, TP have significant positive correlation with SM, FC, Kqs.5, Kqs2, and Kqs5. BD has significant negative correlation with SM, FC, Kqs.5, Kqs2, and Kqs5. SOC have positive correlation with Kqs.5, Kqs2, Kqs5 and negative correlation with SM, FC. Clay has a moderate positive correlation with all the hydrological parameters except field capacity.

Fig. 3
figure 3

Pearson correlation analysis (‘ × ’ indicate no correlation at p value = 0.05).

Full size image

Figure 3 presents the Pearson correlation analysis, with ' × ' marks signifying no significant correlation (p = 0.05). Combined with the Variance Partitioning Analysis (VPA) in Figs. 4a-c, the results underscore the relative contributions of soil physicochemical parameters to variations in soil hydrological properties. The explanatory variables are grouped into soil structure (AMWD, BD, TP), chemical composition (SOC, EC, pH), and texture (Sand, Silt, Clay), highlighting their distinct impacts on hydrological properties and enabling a more comprehensive understanding of each group’s contributions to soil behavior under various conditions.

Fig. 4
figure 4figure 4

Variance partitioning analysis (VFA) to measure the effects of three subset of soil physio-chemical variables viz., on (a) Field capacity, soil moisture, Kqs.5, Kqs2, and Kqs5 (b). Kqs.5, Kqs2, and Kqs5, (c) Field capacity. The values range from the 0 to 1 equivalent to contribution ranges 0–100%.

Full size image

The results of VPA was depicted in Fig. 4a–c, reveals relative contributions of various soil physicochemical parameters to variations in soil hydrological properties. The explanatory variables were divided into three subsets: AMWD, BD and TP (subset 1), SOC, EC and pH (subset 2), and Sand, Silt, and Clay (subset 3).

Subset 1 (AMWD, BD and TP), subset 2 (SOC, EC and pH), and subset 3 (Sand, Silt, and Clay) described 84.0%, 74.0%, and 15.0% of the observed variation of the five hydrological parameters, respectively, whereas unexplained variation only 8% (p-value = 0.05). Subset 1, subset2 and subset 3, alone explained 4%, 8% and 1% of the total variation, respectively. Hence SOC, EC and pH have highest sole contribution (without considering the compound contribution) (Fig. 4a). VPA further applied to measure proportionate contributions of several soil physicochemical characteristics to variations in hydraulic conductivity at 0.5 cm, 2 cm and 5 cm and field capacity, respectively (Fig. 4b–c). The results showed that AMWD, BD and TP contributed the most (94%), followed by SOC, EC and pH (73%) and Sand, Silt, and Clay (29%) in hydraulic conductivity at 0.5 cm, 2 cm and 5 cm (p-value = 0.05) (Fig. 4b). AMWD, BD and TP alone explained 21% of the total variation, which was higher than SOC, EC and pH (2%) and Sand, Silt, and Clay (1%). The unexplained variation only 4% lesser than Fig. 4a. Figure 4c revealed the contribution of physio-chemical parameters to field capacity, where AMWD, BD and TP (subset 1), SOC, EC and pH (subset 2), and Sand, Silt, and Clay (subset 3). Subset 1 (AMWD, BD and TP), subset 2 (SOC, EC and pH), and subset 3 (Sand, Silt, and Clay) described 98.0%, 88.0%, and 21.0% of the observed variation of the five hydrological parameters, respectively, whereas unexplained variation only 10% (p-value = 0.05). Without considering compound contribution, subset 1, subset2 and subset 3, alone explained 0%, 9% and 2% of the total variation, respectively.

Figure 4 utilizes Variance Partitioning Analysis (VPA) to evaluate the individual and collective impacts of three subsets of soil physio-chemical variables on key hydrological parameters. Panel (a) showcases the cumulative influence of these variables on field capacity and soil moisture, as well as pseudo-saturated hydraulic conductivity at different depths, including Kqs.5, Kqs2, and Kqs5. Panel (b) delves deeper into the specific contributions of Kqs.5, Kqs2, and Kqs5 to these parameters, highlighting their distinct roles in soil water dynamics. In contrast, panel (c) concentrates on the isolated effect of field capacity, revealing its primary function in retaining moisture within the soil matrix. This comprehensive analysis, ranging from 0 to 1 in variance contribution, highlights the crucial role of soil physio-chemical properties in governing water movement and retention, providing valuable insights for precision agriculture and hydrological modeling.

Multiple linear regression equations for hydrological parameters

All the soil samples were used to develop pseudo pedo transfer functions for 0.5cm, 2cm and 5 cm using MLR. Moreover, MLR applied to develop equations for field capacity and soil moisture (Fig. 5a–e). A stepwise variable selection method was used to retain physicho-chemical variables in the model, that have significant contributions to the model (at p-value 0.05). Prior to finalize the regression models, multicollinearity test was done using VIF and retain the variables with less than 10 VIF values, showing that multicollinearity does not exist. VIF values of the selected variables were included in Table 4, depicted less than 10 VIF values in Table 4. For Physicho-chemical variables choose as regressor variables in the final models were (a) AMWD, EC, pH and SOC for psudo-hydraulic conductivity at 0.5 cm (Kqs.5), (b) pH, Sand and SOC for psudo-hydraulic conductivity at 2 cm (Kqs2), (c) AMWD, Silt and SOC for psudo-hydraulic conductivity at 5 cm (Kqs5), (d) EC, Sand, Silt and SOC for field capacity, and ), (d) pH, Sand, and SOC for soil moisture. The final regression models were good fit as adjusted R2 were more than 0.9 except soil moisture where adjusted R2 was 0.85 still high value (Fig. 5).

Fig. 5
figure 5figure 5

Final regression equations for hydrological parameters viz., (a) pseudo-saturated hydraulic conductivity at 0.5 cm (Kqs.5), (b) pseudo-saturated hydraulic conductivity at 2 cm (Kqs2) (c) pseudo-saturated hydraulic conductivity at 5 cm (Kqs5) (d) field capacity (%) (e) soil moisture (%). (The equations include physicho-chemical variables only with significant contribution to the model at p value 0.05).

Full size image
Table 4 VIF test for multicollinearity test of fitted regression models.
Full size table

Figure 5 presents the final regression equations linking hydrological parameters, notably pseudo-saturated hydraulic conductivity at 0.5 cm (Kqs.5) and 2 cm. These equations elucidate the variations in these parameters under different conditions, offering a quantitative framework for understanding soil permeability dynamics. Integrating these insights allows researchers to improve models predicting water movement through soil, thereby enhancing water resource management in agricultural and environmental contexts. These equations advance fundamental soil hydrology knowledge and inform sustainable water management strategies.

The findings of this study, although site-specific, have broader implications for similar montane environments characterized by steep slopes, variable vegetation cover, and shallow soils. The identified predictors of Kqsat—such as aggregate stability, bulk density, and soil organic carbon—are process-based variables that remain relevant across regions with comparable soil-forming factors and land use pressures. Therefore, the regression models and ordination-based interpretations developed here can be adapted for use in other parts of the Northwest Himalayas and analogous eco-regions, provided that local calibration is applied. This enhances the potential for upscaling the approach for regional land and water management planning under changing land use scenarios.

Unlike previous studies that focus solely on direct measurements or empirical models1,15,47, this study uniquely integrates constrained ordination (RDA/VPA) with predictive modelling to identify key drivers and estimate Kqsat across multiple land use systems in a Himalayan watershed—an approach not previously applied in this context.

The multiple linear regression equations developed in this study demonstrated high predictive accuracy (adjusted R2 > 0.90), indicating their potential applicability in estimating Kqsat from easily measurable soil variables. However, these equations are site-specific and should be applied cautiously beyond the studied region. Their reliability depends on the similarity of environmental conditions, particularly in terms of slope, soil texture, and vegetation structure48,49. Moreover, the models assume a stable relationship between predictors (e.g., bulk density, SOC, aggregate stability) and Kqsat, which may not hold under extreme climatic events, seasonal variability, or land use transitions. Therefore, for broader application, these equations should be locally calibrated and validated using independent datasets before extrapolation.

Conclusions

This study provides a comprehensive analysis of the intricate relationships between soil physicochemical properties and hydrological characteristics across diverse land uses and soil depths. The findings reveal that land management practices significantly influence soil aggregation, texture, chemical composition, and water dynamics. Agricultural lands demonstrated higher aggregate stability at shallow depths, while tree-covered areas (deodar and oak) exhibited superior soil structure, organic carbon content, and enhanced water retention capacity. In contrast, barren lands showed increased bulk density and reduced water-holding capacity, indicating vulnerability to compaction and degradation.

Advanced statistical techniques, including Redundancy Analysis (RDA), scree plot analysis, and Monte Carlo permutation tests, elucidated the dominant role of soil structure (AMWD, BD, TP) and chemical properties (SOC, EC, pH) in shaping hydrological behavior. Variance Partitioning Analysis (VPA) further quantified these contributions, with soil structural parameters explaining 84% of variability in soil moisture and hydraulic conductivity, while chemical properties accounted for 74%. Texture (sand, silt, clay) had a minor yet discernible influence (15%). Additionally, Multiple Linear Regression (MLR) models, with high predictive accuracy (adjusted R2 > 0.90), established robust pseudo-pretransfer functions for key hydrological parameters, offering practical tools for water management.