Introduction

The Mountains, which encompass over 25% of the Earth’s land area, support a population of more than 12% of the world’s inhabitants. Significantly, over 50% of the world’s human population either directly or indirectly benefited from the goods and services provided by mountain ecosystems1. Soil erosion is the major form of land degradation, affecting hilly and mountainous landscapes globally, with serious implications for productivity, non-point source pollution, and carbon dynamics. Forest cover dominates hilly and mountainous regions, providing stability and conservation benefits by mitigating soil erosion across diverse climates and ecosystems. Forest resists erosion through their canopy, litter cover, and extensive root networks, which enhance soil porosity, stability and carbon storage2. Despite these benefits, reliable information on soil erosion rate and soil carbon stock (SOCS) across different forest types in the north-western Indian Himalayan region remain limited. Processes of soil erosion, including detachment, breakdown, transport, and deposition, impact the soil organic carbon (SOC) pool, leading to its depletion and increased greenhouse gas emissions. However, the overall contribution of erosion-induced carbon loss remains poorly understood. Globally, erosion may displace 4.0 to 6.0 Pg C year−1, with ≈ 20% of this being emitted due to mineralization, underscoring the need for conservation measures to mitigate these emissions and enhance carbon sequestration in soils and biota3.

The Himalayas are identified as one of the global soil erosion hotspots4, despite having a significant portion of their land covered by forest. Based on the State of Forest Report5, approximately 41% of the land area of the Indian Himalayan region (IHR) is covered with forests. Out of this, 16.9% is classified as very dense forest cover, 45.4% as moderate forest cover, and the remaining 37.7% falls as open forest. The Himalayan forests’ diversity and distinctiveness have made a substantial contribution to the abundance of biodiversity elements at all levels, which has led to the region being recognised as one of the 36 Global Biodiversity Hotspots. They are extremely diverse and exhibit significant differences in terms of structure, phenology, and function when compared to both tropical and temperate forests. Additionally, they also differ in terms of ecological processes6. The vegetation in the Himalayan region varies from tropical dry deciduous at lower altitudes to alpine meadows above the regions where trees can grow7. The biomass productivity of the virgin forests in this region, ranging from 17.0 to 21.0 t ha−1 year−1, is comparable to that of highly productive forests worldwide, which typically range from 15.0 to 30.0 t ha−1 year−1among the major forest types in the region8. The forest vegetation regulates hydrological regimes, including streamflow and atmospheric moisture, at both local and regional scales, which in turn influences a range of ecosystem services. The provision of water for human use and the role of forests in soil development and maintaining soil fertility are critical ecosystem services9. Therefore, it is essential to understand the extent of soil erosion in forested areas, including different forest types, to develop effective erosion mitigation strategies and ensure the conservation of these natural ecosystems.

Despite the strong link between forests, water, and soils, our understanding of forest hydrology in the Indian Himalayan Region (IHR) is limited10. Uncertainty persists regarding which forest types (broadleaf or conifer) and tree forms (evergreen or deciduous) are most effective in supporting soil-water conservation while maintaining essential ecosystem functions. Soil erosion studies are crucial for designing effective afforestation and conservation programs in the IHR. The lack of reliable long-term climate data and comprehensive weather station networks at the desired scale hinders the application of models and conventional methods in mountainous regions2. Although, Fallout radionuclides (FRNs), such as cesium-137 (137Cs), offer a valuable alternative for investigating retrospective soil erosion, providing several advantages over traditional erosion and sedimentation measurement methods11. 137Cs, a man-made isotope with a 30.2-year half-life, originated from atmospheric nuclear weapon testing in the 1950s and 1960s and is widely used as a soil erosion tracer. It binds strongly to clay particles and moves via surface runoff, enabling measurement of erosion or deposition rates by comparing radionuclide levels to a stable reference site with no erosion12. 137Cs measurements offer estimates of soil redistribution over a span of around 60 years, since the highest levels of 137Cs fallout occurred in 1963. The scientific community has widely examined and argued the advantages and disadvantages of the 137Cs technique13,14. In response to this discourse, constructive suggestions to enhance this method have been put forward15. Mabit et al.13 suggested, and this was further affirmed by Zhang et al.15, that it is essential to employ a robust sampling design when using this method to ensure reliable estimates of soil erosion rates. Recently, Kumar et al.16 argued that, for the hilly and mountainous regions of the Himalayas, FRNs are most suitable than traditional measurement and modelling methods.

Soil carbon is a critical component of soil that significantly influences ecosystem and agro-ecosystem functions, impacting fertility, water availability, and other soil properties17. Soil organic matter (SOM), derived from decomposing plant and animal materials, acts as both a carbon sink and source, responding to climate, land-use changes, and atmospheric CO₂ levels. Effective forestry practices are essential for enhancing carbon sequestration potential in soils18. Carbon capture and storage is a crucial method for mitigating CO2emissions. Soils serve as the primary storage sites in the carbon cycle on land. Approximately 40% of the world’s SOCS is found in forest ecosystems19. The Himalayan regions in India have extensive forest vegetation covering one-fifth of the country. These areas also have one-third of the country’s SOCS, which are highly valued for their unique conservation importance and the abundance of biodiversity20. Soils have a significant influence on the global carbon budget and can greatly contribute to the release of carbon under the scenario of climate change21. Hence, it is imperative to enhance soil conservation methods to prevent the depletion of carbon stocks, particularly in mountain forests. The management of extensive regions of Himalayan forests at lower altitudes can be considered significant carbon sinks, thereby reducing atmospheric CO2 levels. Additionally, these forests can enhance soil health by increasing several parameters related to soil quality. The carbon stock of Oak (Quercus leucotrichophora) forest ranges from 64 to 72 t ha−1, while Pine (Pinus roxburghii) forest has a carbon stock ranging from 49.6 to 60.0 t ha−122. Kumar et al.23 and David Raj et al.24 identified that soil erosion as a one of the significant cause of carbon loss in the Himalayas. Hence, measurement and monitoring of soil erosion and associated carbon loss is necessary for the sustainable management of Himalayan region.

In the north-western Indian Himalayas, soil erosion studies remain limited, particularly with respect to quantitative measurements using advanced radioisotope techniques such as Fallout Radionuclide-137Cs. To date, no studies have applied the 137Cs method to assess soil erosion across various forest cover types in this region25,26. Since soil carbon loss in forest ecosystems is predominantly driven by erosion processes, the study provides a reliable approach by measuring soil erosion across different forest/tree types—pine (Pinus), deodar (Cedar), oak (Quercus), rhododendron, mixed forest, degraded forest and apple orchard in the Tehri Dam catchment of the north-western Himalayas using the 137Cs method. Additionally, this study integrates soil organic carbon stock (SOCS) measurements (0–15 cm depth) to examine the relationship between SOCS and soil erosion rates, offering insights into the resilience of forest ecosystems in mitigating soil erosion and carbon loss. Hence, objectives of the study were to (i) quantify soil erosion rates in different forest types in the north-western Himalayas using the FRN-137Cs technique, (ii) assess the SOCS across these forest types, and (iii) estimate soil erosion-associated carbon loss in terms of net displaced soil carbon and the emissions resulting from eroded carbon across different forest/tree types.

Methods

Study area

The Tehri Dam catchment falls in Tehri-Garhwal and Uttarkashi districts of Uttarakhand, India, in the north-western Indian Himalayas, between 78°9’15” E to 79°24’55” E and 30°20’20” N to 31°27’30” N (Fig. 1). According to the Köppen climate classification system, the area comes under the Cwb and Dwb categories27. In the Tehri-Garhwal district, the mean maximum and minimum temperatures are 29.4 °C and 14.7 °C, respectively, with an average annual precipitation of 1216 mm. In the Uttarkashi district, the mean maximum and minimum temperatures are 23.6 °C and 13.1 °C, respectively, with an average annual precipitation of 1289 mm28. Most of the catchment area is covered by forest (with species such as Cedrus deodara, Pinus roxburghii, Quercus leucotrichophora, Rhododendron arboreum, Myrica esculenta), along with degraded forest, agricultural land, barrenland, and snow/glaciers.

The soil types can be broadly classified into three groups. The first group includes relatively shallow, highly drained, fine-loamy, moderately erosive, and slightly stony classified as Dystric Eutrudepts soils. The second type consists of loamy soils, classified as Lithic Udorthents, which are found on steep slopes with sandy loam and loamy surfaces. These soils are very shallow, highly drained, severely eroded, and highly stony. The third major soil type belongs to Typic Udorthents, which occurs on moderately sloping land surface and are characterized as moderately deep, loamy texture, excessive drainage, moderate erosion, and slight stoniness29. The soils are slightly acidic in nature with pH values ranged from 3.1 to 8.2, with a mean value of 5.4 ± 1 and 5.6 ± 0.9 for forest and degraded forest, respectively. The electrical conductivity (EC) values ranged from 2 to 844 µS cm⁻¹, with a mean value of 89.3 ± 98.6 µS cm⁻¹ for forest and for 100.8 ± 118.7 µS cm⁻¹ degraded forest. The majority of soils were sandy loam (45%), followed by loam (38%), silt loam (7%), loamy sand (4%), sandy clay loam (3%), clay loam (3%), silty clay loam (1%), and sand (0.17%). The sand content was observed 60.9 ± 12.1% in degraded forest and 55.6 ± 10.9% in forest. The silt content was observed 32.2 ± 10.4% in forest and 28.1 ± 9.5% in degraded forest. The clay content was observed 12.2 ± 8.3% in forest and 11 ± 5.5% in degraded forest. The total organic carbon (TOC) content ranged from 0.99 to 13.0%, with an average of 4.6%30.

Fig. 1
figure 1

Location of the Tehri dam catchment and soil sampling sites.

Full size image

Soil sampling

Grid and transect-based sampling along hillslopes are often sufficient for smaller areas31. However, in large catchments, the substantial variability of 137Cs necessitates meticulous consideration of causes of variation. To achieve this, the catchment was divided into soil-landscape units using the iso-sector approach32. The study employed purposive stratified random sampling based on strata such as forest/tree types and hillslope positions (upper, middle, and lower). This approach resulted in the generation of 18 unique soil-landscape units33. Based on this, soil samples were collected from 145 sites during 2022–2023 (Fig. 1). For the sampling, soil core samples were collected using a stainless-steel cylindrical core upto a depth of 35–50 cm from small clearings between trees/plants in both forested and degraded areas, where rainfall is less obstructed by the canopy. This approach was adapted to minimize the effect of canopy interception on the deposition of caesium34,35. Additionally, the reference site samples (45–50 cm) were also collected near to the sampled sites. In most of the cases the reference site samples were up to 0–10 km limit of the sampled site and some were collected from the published literature24,36,37. In each sampling sites three individual cores were selected within a 1 m−2 to reduce uncertainty and obtaining representative soil samples. The exponential shape factor of 137Cs distribution for the conversion models was calculated from depth-incremental sampling conducted in the same catchment, and which was available in the published literature24,36,37,38. The detailed sampling strategy can be found in David Raj and Kumar39.

Soil analysis

Gamma spectrometry analysis

The 137Cs activity concentrations in soil samples were analysed using a Gamma spectrometer equipped with a high-purity germanium detector (BEGE-5030, Canberra Industries, USA) at Mangalore University. Each soil sample weighed 250–350 g (< 2 mm) and counted over 60,000 s, with activity concentration derived from the 661.6 keV gamma emission. Efficiency and calibration details for the Gamma Spectrometer were previously provided by Karunakara et al.40. The 137Cs inventory (Bq m⁻²) for the entire soil layer was then calculated using 137Cs concentrations (Bq kg⁻¹), the mass of the < 2 mm fraction, and the cross-sectional area (m−2) of sampling tool41.

Soil carbon and bulk density analysis

For soil organic carbon stock measurement, same site 0–15 cm soil sample also been collected. After soil sample collection, the samples underwent air drying, disaggregation, and sieving through a < 2 mm sieve. To estimate soil organic carbon (Walkley-Black method)42 and bulk density (Keen’s cup), the samples were analyzed following standard procedures in Central Analytical Laboratory (CAL), Indian Institute of Remote Sensing.

Quantifying soil erosion rate

Typically, in undisturbed and stable soils, the 137Cs concentration decreases exponentially with depth, which can be modelled using a specific function43,44. However, it relies on numerous simplifying assumptions and does not consider for the time-varying nature of 137Cs fallout or the gradual changes in its depth distribution after atmospheric deposition. Although, the profile shape model is straightforward and easy to apply.

$$\:Y=\:\frac{10}{\left(t-1963\right)P}ln\left(1-\frac{X}{100}\right){h}_{o}$$
(1)

Y represents the annual soil loss (t ha−1 year−1), t denotes the year of sample collection, and X indicates the percentage loss of 137Cs in the total inventory relative to the local 137Cs reference value. Where, h0 is the coefficient describing profile shape (kg m-2).

Quantifying soil organic carbon stock

To determine soil organic carbon stocks (SOCS) in t ha−1, the equation provided by FAO45 was used:

$$\:SOC\:stock={OC}_{i}\times\:{BD}_{i}\times\:(1-{gG}_{i})\times\:{T}_{i}\times\:0.1$$
(2)

OCi represents the organic carbon content (mg C per gram) in the fine earth fraction (particles < 2 mm) within depth increment i. BDi is the soil’s bulk density (g per cm³) for this depth increment, while gGi refers to the mass fraction of coarse mineral fragments within the sample. Consequently, (1 - gGi) indicates the proportion of fine earth (g of fine particles per gram of total soil) in the depth increment i. Ti is the thickness of this depth layer in centimeters. A factor of 0.1 converts values from mg C cm⁻² to t C ha⁻¹. This study focuses exclusively on SOC stock in the surface layer (0–15 cm).

Estimation of soil erosion-associated carbon loss

Soil erosion-induced carbon loss is calculated primarily through two mechanisms: the displacement of carbon due to soil erosion and carbon emissions from eroded soils, which result from the mineralization of the displaced carbon. The C-displaced is influenced by erosion rate, SOC concentration, and Carbon Enrichment Ratio (CER) values. The CER, defined as the ratio of the organic carbon content in eroded sediment to that of the source soil46. Mandal et al.47 provided CER values for various soil erosion classes in Uttarakhand state, based on the surface soil and sediment concentrations from experimental plots. Thus, the C-erosion rate was determined using the following formula:

$$\:C\:Erosion\:\left(t\:{ha}^{-1}\:{year}^{-1}\right)=\frac{Soil\:loss\:\left(t\:{ha}^{-1}\:{year}^{-1}\right)\times\:SOC\:\left(\%\right)\times\:CER}{100}$$
(3)

A detailed literature analysis, encompassing 12 studies from various global agro-climatic zones, established a mean estimate of 30% for the fraction of eroded carbon that becomes oxidized and released into the atmosphere47. This value serves as a key input for calculating carbon emissions resulting from soil erosion. Thus, the amount of carbon emitted during soil erosion can be determined by applying the following equation:

$$\:C\:Emission\:\left(t\:{ha}^{-1}\:{year}^{-1}\right)=C\:Erosion\:\left(t\:{ha}^{-1}\:{year}^{-1}\right)\times\:0.3$$
(4)

Statistical analysis

Correlation analysis was used to detect the association between soil erosion, soil carbon stock, and associated carbon loss. Additionally, one-way ANOVA and Tukey’s post hoc test were employed to determine significant differences in soil erosion and soil carbon stock across different forest/tree types. All analyses were conducted using R programming software.

Results and discussions

Measurement and modelling of 137Cs reference and sampled inventory

The measured reference inventory ranges from 1408.6 to 2246.7 Bq m−2 with a mean value of 1834.9 ± 232.9 Bq m−2 (supplementary file Table 1). The coefficient of variation ranged from 0.83 to 21.6% with an average of 14.5%. Previous investigations have reported standard deviations up to 30% considered satisfactory for reference sites48. In sampled sites 137Cs activity ranged from 223.3 to 1885.0 Bq m−2 with an average value of 1155.01 ± 341.5 Bq m−2. Mariappan et al.41reported a modeled reference inventory of 1685 Bq m⁻² in the foothills, while Prokop and Poreba49 measured 1220 Bq m−2 in the north-east Himalayas, both at lower latitudes and higher longitudes than this study. Mandal et al.50 found 137Cs levels in Dehradun between 944 and 1170 Bq m−2. Tagami et al.51 examined 137Cs distribution across South Asia, revealing a geometric mean of 1576 Bq m−2 in the 30–40° latitude band, with a range from 860 to 3731 Bq m−2. In Srinagar, Jammu and Kashmir (34.083° N), 137Cs was recorded at 2685 Bq m−2 while in New Delhi (28.583° N), it was 1580 Bq m−2(decay-corrected to 2022/2023)51,52. Models were also used to verify the reliability of measured 137Cs reference values49. Recent studies found that in Tehri dam catchment area 137Cs concentration at reference locations varied from 1409 to 2172.8 Bq m−224,36,37,38.

As the catchment is extensive and highly rugged, it was not feasible to use each reference point for every sample at every location. Due to this limited data, it was necessary to substantiate the dependability of the measured reference inventory by comparing it with global reference inventory models. Thus, the study compared the measured 137Cs with different global fallout methods and found that the Pálsson53 method (based on rainfall) significantly correlates with the measured values, with a low RMSE value. Several models were employed to estimate 137Cs fallout using monthly scale rainfall data from the India Meteorological Department (IMD), resulting in a reference inventory between 1504 and 2006 Bq m−2 (supplementary Fig. 1). All methods produced inventory estimates that were fairly consistent in magnitude with the measured reference inventories at the reference site. Thus, 21 IMD grid station rainfall data were used to generate a modeled reference inventory of 137Cs for the Tehri dam catchment. To improve accuracy close to the measured 137Cs values, bias correction was applied. Several studies have utilized global fallout models for soil erosion assessment when 137Cs measurements are not available41. Therefore, the both modelled (bias corrected) and measured global fallout of 137Cs was utilized for soil erosion assessment.

Estimation of FRN-137Cs -based soil erosion rate

The majority area of the Tehri Dam catchment is covered with forest, with tree types including pine, oak, deodar, mixed forest, rhododendron, apple and degraded forest. The soil erosion rate at different forest type has been estimated (Table 1; Fig. 2). The mixed forest and oak were estimated lowest mean soil erosion rate of 3.7 ± 2.2 and 4.2 ± 3.7 t ha−1 year−1 correspondingly. The rhododendron had an estimated value of 5.5 ± 0.9 t ha−1 year−1. Deodar forest had an estimated soil erosion rate of 6.0 ± 3.9 t ha−1 year−1. The pine forest had a soil erosion rate of 7.0 ± 3.6 t ha−1 year−1. The Apple orchard had the value of 10.2 ± 2.3 t ha−1 year−1. Whereas the degraded forest has the highest soil erosion rate of 15.5 ± 6.4 t ha−1 year−1.

Table 1 Descriptive statistics of soil erosion rate under different forest/tree types of north-western Himalayas.
Full size table

A study conducted by Kalambukkattu et al.54 using RUSLE model reported that average potential soil erosion rate under forest was < 10 t ha−1 year−1, and wasteland/barrenland has more than 100 t ha−1 year−1. They reported that average annual soil erosion rate was estimated as 27.5 t ha−1 year−1. Another study conducted by David Raj et al.55 in Shiwalik Himalayas stated an average of 23.4 t ha−1 year−1 with scrub land/degraded forest (35.6 t ha−1 year−1) and forest having 13.6 t ha−1 year−1. Sooryamol et al.56 conducted a study in lesser Himalayas found a high average soil erosion rate of 29.3 t ha−1 year−1 whereas, they observed that forest having lowest soil erosion and degraded forest/scrubland having higher soil erosion rate. A global study conducted by Borrelli et al.57 using RUSLE model states that bare soil and agriculture have the highest soil erosion whereas forest cover has lowest erosion. Xiong et al.58 conducted a study using runoff plot indicated that bare lands exhibited highest average soil loss values, ranging from 10.6 to 109.2 t ha−1 year−1. Forest lands experienced the lowest soil loss, ranging from 0.2 to 0.6 t ha−1 year−1. However, agricultural land had an average soil erosion rate of 22.1 ± 47.0 t ha−1 year−1, and forest land had an average rate of 0.4 ± 0.7 t ha−1 year−1. Borrelli et al.25conducted a systemic review and states that according to different modelling studies, bare soil has the highest soil loss, followed by agriculture, grassland and lowest in forest land. The distribution of soil erosion rates across the various land use/cover types aligns with the findings reported by Montgomery59 and Borrelli et al.4 based on field measurements.

The ANOVA analysis (Tukey’s post-hoc test) revealed that the significant difference in soil erosion rate between several forest types. Degraded forests exhibit significant differences in soil erosion when compared to mixed (p = 0.0000), oak (p = 0.0000), deodar (p = 1e-7), pine (p = 0.0000), and rhododendron (p = 0.0001). Similarly, apple orchard showed significant differences when compared to oak (p = 0.046), and mixed forest (p = 0. 0179). Whereas pine forest showed significant difference from mixed forest (p= 0.008). These results indicated that degraded forests showed consistently significant differences in soil erosion as compared to several other forest types. Moreover, apple orchards also differ significantly from oak, and mixed forests in terms of soil erosion. Land use/land cover (LULC) particularly forests are crucial in mitigating soil erosion. They diminish the direct impact of raindrops on the soil surface, enriching organic matter content, enhancing water infiltration rates, slowing down runoff velocity, and curbing sediment transport across the soil surface60,61. Kothyari et al.62 reported a peak soil erosion rate of 5.47 t ha−1 year−1 from runoff plots (pine forest) in the Kumaon Himalayas, which receive 179.33 mm of rainfall. Soil erosion from slash-and-burn cultivation was 20 times greater than that from natural forests in mountainous areas. Erosion rates in northern Brazil’s deforested areas were 115 Mg ha−1 for recently cleared forest soils, 8.6 Mg ha−1 for grass-covered soils, and barely 1.2 Mg ha−1for shrub and tree-covered soils63. According to Mandal and Sharda64 soil depth of north-western Himalayas varies from 25 to 150 cm and soil loss tolerance limit (SLTL) is generally 7 − 5 to 10 t ha−1 year−1. The forest land observed lowest mean soil erosion rate of 5.9 ± 3.6 t ha−1 year−1 which is lower than the SLTL. Although, the degraded forest showed soil erosion rate of 15.5 ± 6.4 t ha−1 year−1, which 1.5 to 2-fold higher than the SLTL.

Fig. 2
figure 2

(Source: author, generated using R 4.3.2; https://www.r-project.org/).

Forest/tree type-based soil erosion rate using rain-cloud plot.

Full size image

Figures 3 and 4shows the different type of forest/trees present in the Tehri dam catchment area. Canopy cover, sapling density, litter depth, and woody debris are key ecological factors affecting soil erosion. The canopy reduces raindrop size and erosive force, while saplings, litter, and woody debris shield the soil, limiting detachment and adding surface roughness that slows the soil movement to downslope. However, their impact on runoff was less significant than that of rainfall65. In mixed forest different species and multi-story vegetation is present (Fig. 3d). They provide greater soil erosion control benefits compared to monocultures. Enhancing tree species diversity can be an effective strategy for soil erosion management60. In a subtropical forest experiment, Song et al.66found that tree diversity was found to reduce soil erosion by influencing both the tree canopy and the development of biological soil crusts. Typically, forests with multiple species exhibit more complex canopy layers and a greater variety of functional traits (such as leaf area), which can enhance rainfall interception by the canopy and decrease the kinetic energy of raindrops, thereby lowering the risk of soil erosion67. Canopy cover was crucial in reducing runoff and soil loss, while the litter cover beneath plants was essential for controlling erosion during intense rainfall68. Thus, the broad leaved oak and mixed forest witnessed lower soil erosion compared to needle leaved pine and deodar forest (Figs. 3 and 4). Wakiyama et al.69 reported that broad-leaved species have a comparatively lesser soil erosion rate than needle-leaved species, with understory vegetation significantly reducing the soil erosion rate. Figure 4a-d illustrates different types of pine forest stands observed in the Tehri Dam catchment. Moreover, pine forests show allelopathic inhibition of understory growth (Fig. 4a, b, c, d)70. Additionally, fires in pine forests disrupt the successional process by impeding the establishment of late successional species71, which reduces the multi-storied protection of the forest floor from other tree and shrub species. The density of canopy cover in a forest also plays a major role in determining the soil erosion rate. Neto et al.72 reported that sparse vegetation experiences higher soil erosion compared to dense vegetation.

Fig. 3
figure 3

Different tree species found in Tehri dam catchment area – (a) needle leaved deodar, (b) broad leaved oak, (c) broad leaved rhododendron, (d) broad leaved mixed forest, (e) broad leaved apple tree, and (f) scrub/degraded forest.

Full size image
Fig. 4
figure 4

Different types of pine forest (a) open forest, (b) moderately dense forest and (c) dense pine forest, and (d) sparse understory growth in pine forest.

Full size image

The Table 2 reveals the field observation characteristics of different forest/tree types. The forests with more closed canopies, abundant understory growth, and rich litter layers tend to have darker, more fertile soils. However, even in these cases, gully erosion remains a persistent issue, likely driven by external factors like topography, slope steepness, and rainfall intensity, which can overwhelm the natural erosion control provided by vegetation (Fig. 5a). Forest types with open canopies, such as pine and degraded forests, show significantly poorer soil conditions and higher susceptibility to severe erosion. The absence of understory vegetation and litter exacerbate soil erosion, as there is low organic matter to stabilize the soil (Fig. 5b). Overall, the table highlights the critical role of vegetation structure in maintaining soil health and reducing erosion, but it also points to the limitations of vegetation alone in mitigating severe erosion, especially in challenging landscapes. Mixed forests, with broad-leaved trees, demonstrate the healthiest ecosystem in terms of canopy closure, understory abundance, and litter accumulation. The large root system, substantial floor litter, and dense canopy layers all contribute to the low rates of soil erosion. Raindrops are captured and absorbed by the thick surface layer, which also encourages water infiltration. It minimizes soil erosion by reducing runoff water velocity73. The Degraded forest exhibited the worst conditions, with both broad- and needle- leaved trees and an open canopy, leading to poor understory growth and an absence of litter. The light-coloured soil indicates significant depletion of organic matter, and the erosion is severe, with highly dissected gullies and exposed rocks (Fig. 3f). This suggests that the ecosystem has been heavily disturbed, possibly due to deforestation or overgrazing, leading to accelerated soil degradation and due to erosion. The absence of vegetation and very steep slopes enhance gully erosion, and in extreme cases, lead to deep incisions in the landscape, causing severe erosion and concentrated water flow (Fig. 5c & d). In such cases, effective soil conservation in these areas likely requires both maintaining vegetative cover and implementing additional erosion control measures like terracing or gully plugging. In absence of these measures, it will affect soil quality and health, eventually lowering soil resilience against erosion. These underscore the vital importance of maintaining forest diversity and canopy cover in preventing soil erosion, enhancing carbon sequestration, and promoting long-term ecosystem resilience, which are crucial for combating climate change and preserving the environmental balance in mountainous landscapes.

Table 2 Characteristics of different forest/tree types observed in the field.
Full size table
Fig. 5
figure 5

Erosion features found in forested landscape: (a) gully erosion in pine forest, caused by concentrated runoff water flow, (b) exposed roots in oak forest, due to surface erosion, (c) large gullies in steep sloped deodar forest, formed by intense runoff and (d) less vegetated and highly dissected landscape where sparse vegetation accelerates erosion and fragments the terrain.

Full size image

Estimation of soil carbon stock

The Fig. 6 illustrates the variation in SOCS (0–15 cm) across different forest and tree types. The highest SOCS was observed in mixed forests, with a mean of 73.71 ± 32.2 t ha−1 and a relatively large standard deviation (SD) indicating significant variability in the SOCS within this forest type. Oak forests follow with a mean SOCS of 59.89 ± 24.4 t ha−1. Rhododendron forest showed a mean SOCS of 56 ± 18.7 t ha−1. Deodar forests exhibited a mean SOCS of 48.76 ± 19.4 t ha−1, while pine forests have an almost similar mean SOCS of 48.5 ± 21.7 t ha−1 but slightly higher variability. Degraded forests, with a mean SOCS of 46.33 ± 19.9 t ha−1, showed a noticeable reduction in SOCS compared to the other natural forests. The apple orchards present the lowest SOCS, with a mean of 23.41 ± 4.3 t ha−1, demonstrating the impact of agricultural land use on carbon storage. The ANOVA analysis (Tukey’s post-hoc test) revealed that the significant difference in SOCS between different forest types. Specifically, oak-apple (p = 0.0409) and mixed-apple (p = 0.0005) showed significant differences, with both oak and mixed vegetation types having higher SOCS compared to apple. Additionally, significant differences areobserved between mixed-degraded (p = 0.0182) and mixed-pine (p = 0.0005). These results suggest that natural forests, especially mixed, oak, deodar and rhododendron, store more SOC than degraded lands and orchards. The variability in mixed forests may result from differences in composition or management. Figure 7 illustrates how varying understory vegetation and litter layers across forest types such as the dense multi-story canopy in mixed forests and the thick litter layers in oak, rhododendron, and deodar forests contribute to differing SOCS. Denser litter and canopy layers in mixed and oak forests are linked to higher SOCS, while the sparser ground cover in deodar forests corresponds to comparatively lower SOC accumulation. In contrast, degraded and orchards showed lower SOC, highlighting the negative impact of land degradation on carbon storage. This underscores the importance of forest type and management in enhancing soil carbon storage for ecosystem health and climate mitigation.

Fig. 6
figure 6

(Source: author, generated using R 4.3.2; https://www.r-project.org/).

Soil organic carbon stock under different forest/tree types.

Full size image

In Giri catchment under chir pine forests, Negi and Gupta74 observed that SOC density (up to 30 cm depth) was varied between 29.71 and 89.03 t ha−1 with an average value of 57.33 t ha−1. According to Sharma et al.75 SOCS varied between 40.3 t ha−1 on SW aspect (Pinus roxburghii) and 177.5 t ha−1 on NE aspect (moist Cedrus deodara) in 30 cm. The SOCS up to 100 cm ranged between 56.7 and 337.8 t ha−1, with a mean of 168.1 ± 93.0 t ha−1in broadleaved and conifer forests76. SOCS in oak forests (0–90 cm depth) was found to be 171.79 t ha−1, over twice that of pine forests (73.67 t ha−1)77. These values are lower than those reported for Central Himalayan oak forests (166.8–440.1 t ha−1) by Rana et al.78 but align with pine forest estimates (62 t ha−1) from Jina79 and Jina et al.80. Lower SOC values have also been noted in this region for oak (96.4 t ha−1) and pine (61.1 t ha−1) forests81. SOCS was observed to decrease with altitude in oak (185.6 to 160.8 t ha−1) and pine (141.6 to 124.8 t ha−1) forests20. Forests (including oak and pine) showed the SOCS of 168.8 ± 74.4 t ha−1 in the top 1-m of soils82. Total carbon stock in soils of different forest zones and altitudes in Bhutan varies from (0–10 cm) 20.4 to 78.5 t ha−183. SOCS of deodar (0–30 cm) is 89.7 to 105.3 t ha−184. The higher SOCS in mixed and forests likely results from dense canopies and higher litter input, promoting carbon storage. Dense vegetation in oak sites supports greater soil organic carbon accumulation than in coniferous sites. Conversely, pine forests have lower SOC due to wider tree spacing (Fig. 4a), which reduces litter input and overall carbon storage in these soils85. At similar depths and forest types, our SOCS values were comparable to existing studies, but with increasing depth, the stock tends to be higher than in our findings. Biomass and carbon stocks in forest vegetation are governed by geographical location, tree/plant species, stand age, rainfall, climate, and vegetation cover77,85,86, which together affect forests’ capacity to conserve soil and water, sequester carbon, reduce greenhouse gas emissions, provision of resources, support of biodiversity, and enhancement of environmental quality73.

Fig. 7
figure 7

Different types of understory vegetation cover and litter: (a) mixed dense forest featuring a multi-story canopy and a thick layer of litter covering the forest floor, (b) oak forest with a dense litter layer and scattered branches on the ground, (c) rhododendron forest with a dense accumulation of litter on the forest floor, and (d) deodar forest with sparse grass cover but a thick layer of litter covering the ground.

Full size image

Soil erosion induced carbon displacement and carbon emission

Fig. 8
figure 8

(Source: author, generated using R 4.3.2; https://www.r-project.org/).

Soil erosion associated carbon loss under different forest types.

Full size image

Soil erosion, sediment transport, and deposition processes on slopes can lead to SOC loss. These processes redistribute SOC across the landscape, increase oxidation, and create both SOC sources and sinks87. Fig. 8a compares the net carbon displaced across various forest and tree types. Degraded forests have the highest mean carbon displacement (1.42 ± 0.7 t ha-1 year-1). Rhododendron and pine forests follow, with mean values of 0.98 ± 0.5 and 0.8 ± 0.5 t ha-1 year-1, respectively. Deodar, mixed, and apple orchards exhibited lower carbon displacement, with means ranging from 0.67 to 0.79 t ha-1 year-1. Oak forests showed the lowest net carbon displaced (0.58 ± 0.3 t ha-1 year-1) indicating minimal carbon loss. This highlighted that degraded forests experience the highest net carbon loss, while more stable forest ecosystems like oak, deodar and mixed show less carbon displacement. The pairwise comparison showed that degraded forests had significantly higher net carbon displacement compared to other forest types, with notable differences observed between degraded and oak (p = 0.0002), degraded and deodar (p = 0.0055), degraded and mixed (p = 0.0026), and degraded and pine (p = 0.0012). However, no statistically significant differences were found between other forest types, as all other p-values exceeded 0.05, indicating that carbon displacement among these types (such as deodar-oak, pine-oak, and apple-rhododendron) do not differ substantially. The Fig. 8b illustrates the erosion induced carbon emission for different forest or tree types. The degraded forest showed the highest carbon emission mean at 0.43 ± 0.2 t ha-1 year-1. The rhododendron forest had a mean of 0.3 ± 0.2 t ha-1 year-1. Pine and apple orchards both had similar means of 0.24 t ha-1 year-1. Mixed forests had a slightly lower mean of 0.22 ± 0.1 t ha-1 year-1. The deodar forest showed a mean of 0.20 ± 0.1 t ha-1 year-1, and the oak forest had the lowest mean at 0.17 ± 0.1 t ha-1 year-1. The pairwise comparisons of carbon emissions between various forest or tree types, showing the difference in means same trends as the net carbon displaced. Kumar et al.23 reported that the carbon erosion in the watershed of lesser Himalayas ranged between 0.31 and 0.66 t ha⁻¹ year⁻¹. Forested areas exhibited the lowest SOC loss at 0.31 t ha⁻¹ yr⁻¹, primarily due to the protective cover provided by vegetation and the continuous addition of organic matter through litter fall. Additionally, minimal soil erosion in these areas contributed to the reduced carbon loss. In contrast, scrubland/degraded forest experienced the highest carbon erosion, reaching 0.66 t ha⁻¹ year⁻¹, attributed to sparse vegetation cover and highly erodible soils, which lead to greater soil and carbon loss. Average carbon displaced across the watershed was 0.45 t ha⁻¹ year⁻¹ with 2200 mm of rainfall. This is comparable to rates reported by Boix-Fayos et al.87 in a Mediterranean catchment (0.20 t ha⁻¹ year⁻¹ with 583 mm rainfall), Sitaula et al.88 in the Hindu Kush Himalayas (0.26 t ha⁻¹ yearr⁻¹), and Mandal et al.50 in Doon Valley (0.38 t ha⁻¹ year⁻¹ with 1625 mm rainfall).

The correlation plot (Fig. 9a) shows a moderate negative relationship between soil erosion rate and SOCS, with a Pearson correlation coefficient (r) of −0.30, indicating that as soil erosion increases, SOCS tends to decrease. The p-value of 0.00036 suggests that this relationship is statistically significant. The negative correlation implies that erosion is a key factor in reducing organic carbon content in soils, as erosion typically removes topsoil, which is rich in organic matter. Figure 9b illustrates the relationship between soil organic carbon (SOC) displacement (t ha−1 year−1) and soil erosion rate (t ha−1 year−1), revealing a positive correlation (R = 0.52, p-value = 2e⁻¹¹). This indicates that as the soil erosion rate increases, the amount of SOC displaced also increases. The Fig. 9c shows a slightly weaker, yet still positive correlation (R = 0.50, p-value = 1.7e⁻¹⁰) between soil organic carbon percentage and carbon displacement. It suggests that higher SOC percentages correlate with greater displacement, though the relationship is somewhat weaker than that of erosion rate. Both p-values indicate robust statistical significance. These positive correlations highlight the critical role of erosion in the loss of soil carbon, which can affect soil fertility and carbon cycling in ecosystems. Eroded SOC is closely linked to soil erosion, although this relationship is influenced by the severity of erosion, the concentration of SOC in the topsoil, and the Carbon Enrichment Ratio (CER). Mandal et al.47 observed that the correlation between eroded soil and carbon loss associated with erosion is notably strong (r = 0.82), suggesting that as erosion increases, the corresponding carbon loss also rises in direct proportion.

Accelerated soil erosion leads to preferential removal of soil organic carbon (SOC), as it is concentrated near the soil surface and is lighter than mineral particles89. Komissarov and Ogura90 also found that the presence of silt and clay fractions enriches the 137Cs and SOC concentrations. During erosion, SOC is detached, transported, redistributed, and eventually deposited in depressional sites where it may become buried with sediments, although much of it is also mineralized to CO2 and CH43,89. This redistribution and exposure to varying environmental conditions, such as temperature and moisture, accelerate SOC decomposition and greenhouse gas (GHG) emissions, including CO2, CH4, and N2O91,92. Soil erosion thus depletes SOC stocks, especially in agroecosystems and degraded forests, affecting the global carbon cycle and contributing to climate change93. Additionally, the breakdown of soil aggregates during erosion releases SOC that would otherwise be protected from microbial decomposition94. This depletion of SOC in eroded soils compared to un-eroded soils highlights the significant role of erosion in soil degradation. The SOC pool, which holds twice the carbon of the atmospheric pool, is thus a critical factor in climate regulation. Detecting SOC loss during erosion can be achieved using isotopic tracers like 137Cs95, and studies showed a strong negative correlation between erosion and SOC content96.

Fig. 9
figure 9

Relationship between (a) soil erosion rate and soil organic carbon stock; (b) soil carbon displaced with soil erosion rate (t ha−1 year−1) and (c) soil organic carbon (%) (95% CI).

Full size image

Deforestation is a significant threat to forests, especially tropical forests, which are disappearing rapidly due to natural and human causes73,97. Soil erosion exacerbates disturbances in tropical forests, reducing resilience and increasing vulnerability to degradation97. Implementing effective land management practices to control soil erosion offers substantial climate benefits by enhancing carbon sequestration and stabilizing forest ecosystems. In the Tehri Dam catchment, scrub land/degraded forest covers a land are of 18,664 ha (approximately 2.5% of the land area). Through forest regeneration and erosion control, degraded lands can sequester approximately 0.50 t C ha−1 year−147, leading to an estimated 9.33 kilotons of carbon sequestered annually. Recognizing and addressing carbon storage dynamics through conservation measures is crucial for reducing emissions, enhancing soil and vegetation sequestration, and contributing to climate change mitigation in this region. Land-use changes, including afforestation and the management can significantly impact the regional carbon sequestration rate by facilitating the absorption of carbon dioxide (CO2) into plant biomass. Additionally, establishing mixed-species forests enhances forest stability and helps to mitigate rapid decomposition of soil organic matter, promoting a healthier ecosystem as well98.

Limitation and future scope of the study

The present study provides in-situ (137Cs) soil erosion measurements across different forest/tree types in the north-western Himalayan region, where no prior attempts have been made. The sampling design effectively captures variations in soil erosion rates under different land cover types. However, we did not account for variations in soil erosion rates across different slope classes. To reliably estimate the influence of slope on soil erosion, a larger number of samples would be required. Hence, there is significant scope for future research to explore the influence of topography (slope and hillslope position) on soil erosion and its associated carbon loss. Additionally, soil types vary across the catchment, but this aspect was not considered in the current study. Future research could include these variations to provide a more comprehensive understanding. Moreover, the relationships between various soil properties, soil erosion rates, and 137Cs concentrations remain poorly understood in this region and necessitates further investigation. In this study, we used a uniform particle size correction factor due to the unavailability of detailed particle size data. Future studies could incorporate this factor to improve the accuracy of soil erosion estimates. To further reduce uncertainties in reference location data, multiple temporal samples could be collected as suggested by Arata et al.99. As most of the sampled slopes in this study are steep (> 50%), traditional measurement methods like runoff plots are not feasible. Thus, fallout radionuclides (FRNs) are particularly suitable for hilly and mountainous regions of the Himalayas16. While the current study used a single radionuclide tracer (137Cs), future research could use multiple tracers to increase the reliability of soil erosion estimates69. Additionally, detailed spatial mapping could be achieved by integrating 137Cs-derived soil erosion rates with machine learning techniques to enhance predictive accuracy and spatial representation.

This study utilized in-situ measurements for soil erosion and carbon content; however, values for carbon displacement and emissions from eroded soils were obtained from literature due to the unavailability of in-situ data for these parameters in the specific study area. While these literature-derived values were chosen from studies in similar climatological zones to approximate local conditions, they may not fully capture the local variations in carbon dynamics. Direct, site-specific measurements of carbon displacement and emissions would provide a better understanding of these processes in the Himalayan context. Despite these limitations, our findings offer valuable insights into land use/land cover patterns associated with higher carbon emissions and displacement, informing carbon management strategies in erosion-prone, data-sparse regions of the Himalayas.

Conclusions

In the hilly and mountainous regions, soil erosion is a significant environmental concern. It leads to substantial losses in soil organic carbon content from the topsoil. Most of these regions are covered with natural vegetation, which helps to conserve soil and mitigate the adverse effects of erosion. However, forested areas have been less studied, as they contribute less to overall soil erosion, and there is a lack of data on soil erosion rates and induced carbon loss in the Himalayas. To address this gap, we conducted a study to measure the soil erosion rates under different forest and tree types in the Tehri Dam catchment of the north-western Himalayan region. Our findings indicated that mixed forest exhibited the lowest soil erosion rates and highest soil organic carbon stock, while degraded forests showed significantly higher rates of soil erosion and lower soil organic carbon stock.  Correspondingly, the highest erosion-associated carbon loss was observed in degraded forests, whereas the lowest loss was found in oak, deodar, and mixed forests. This study highlights that soil erosion is a major factor contributing to the reduction of carbon content in forest soils, as it not only displaces soil but also leads to the oxidation of carbon from eroded materials. The regeneration of forest cover in degraded areas has the potential to enhance carbon sequestration, serving as a nature-based solution to mitigate the adverse effects of climate change. The findings of this study underscore the urgent need for targeted conservation efforts in degraded forest areas to restore soil health and enhance carbon storage. By prioritizing forest regeneration initiatives, policymakers and land managers can improve ecosystem resilience, enhance biodiversity, and contribute to climate change mitigation. Furthermore, understanding the relationship between forest types, soil erosion, and carbon loss can guide sustainable land management practices that promote environmental sustainability and carbon neutrality in the Himalayan region.