FTIR-derived soil degradation indices and stochastic modelling of organic matter–sediment dynamics in a Mediterranean watershed: A Northern Apennines case study

Manuel La Licata; Odunayo D Adeniyi; Ruth H Ellerbrock; Nisha Bhattarai; Alberto Bosino; Natalie Papke; Jörg Schaller; Michael Maerker

doi:10.1371/journal.pone.0330252

. 2025 Aug 21;20(8):e0330252. doi: 10.1371/journal.pone.0330252

FTIR-derived soil degradation indices and stochastic modelling of organic matter–sediment dynamics in a Mediterranean watershed: A Northern Apennines case study

Manuel La Licata ^1,^2,^3,^*, Odunayo D Adeniyi ^2,⁴, Ruth H Ellerbrock ⁵, Nisha Bhattarai ⁵, Alberto Bosino ⁶, Natalie Papke ¹, Jörg Schaller ⁵, Michael Maerker ^1,^2,³

Editor: Przemysław Mroczek⁷

PMCID: PMC12370054 PMID: 40839698

Abstract

In this study we explored the relationships between Soil Organic Matter (SOM) properties, serving as potential indicators of soil degradation and erosion, and environmental, geomorphic, and hydrological characteristics in an agricultural-forested Mediterranean watershed. SOM composition of fluvial sediments sampled across the watershed was analysed using FTIR spectroscopy to calculate FTIR-based proxies for the relative hydrophobicity of SOM, Cation Exchange Capacity (CEC), and organic-matter-cation associations. To investigate geospatial relationships between SOM composition influencing erosion susceptibility and the factors driving its variability at the watershed scale, such as terrain characteristics, soil properties, lithological, and LULC data, we used a Random Forest modelling approach. Our findings indicate that the size and configuration of the contributing areas associated with the sampling points played a crucial role in interpreting the relationships between SOM composition and environmental factors. Oak, hornbeam, and chestnut forests influence hydrophobic organic matter accumulation, making soils more prone to water erosion, where clay content potentially intensifies erosion susceptibility under particular climatic conditions. Moreover, SOM chemical components were spatially linked to sediment dynamics and organic matter connectivity across the watershed, with topographic features such as elevation and channel network base level being key factors. Also, CEC was found to be a potential indicator of soil erosion in geomorphologically active areas. Lastly, carbonate-rich soils appeared to positively influence organic matter-cation associations, potentially enhancing aggregate stability and reducing erosion susceptibility. This study provides significant new insights into the complex relationships between SOM composition, environmental predictors, and soil erosion in Mediterranean watersheds, supporting novel research hypotheses and perspectives from both a scientific and applicative point of view.

1 Introduction

Soil is a vital resource that sustains life by facilitating the exchange of mass, energy, and biodiversity [1]. It plays an essential role in supporting ecosystems by processing inputs and generating outputs that influence groundwater, vegetation, atmosphere, and surface waters [2,3]. However, soils are essentially non-renewable on human timescales, making them vulnerable to degradation due to complex interactions occurring across various spatial and temporal scales [4]. Soil degradation, exacerbated by human activities and climate changes, is a significant threat to social-ecological landscapes, particularly through soil erosion, which undermines land productivity, ecosystem services, and socio-economic stability [5–7]. Some regions such as the Mediterranean are more vulnerable to land degradation triggered by geomorphic activity, particularly due to pronounced seasonal climatic variability, steep slopes, and a long history of human pressure [8–10]. Nonetheless, proper management practices offer the potential for soil restoration and recovery, highlighting the importance of understanding the relationships between soil functions, degradation, and resilience [11], particularly in landscapes with complex nature-human interactions [12].

Soil Organic Matter (SOM) is a key factor influencing soil response to external forcings, playing a critical role in maintaining soil quality and health [13]. Moreover, it influences a range of physical, chemical, and biological properties [14,15], such as aggregate stability, bulk density, water-holding capacity, erodibility, and compaction, as well as nutrient levels, Cation Exchange Capacity (CEC), and pH [16,17]. Many SOM-related properties are indeed used as indicators of soil degradation susceptibility [4], particularly to water erosion [18,19].

SOM plays a crucial role in influencing soil erodibility [20,21], which is essential in regions vulnerable to erosion like the Mediterranean. However, the relationship between SOM and erosion processes is complex and multifaceted, as SOM’s interactions with soil minerals and water infiltration behaviour can either enhance or reduce erosion. Thus, SOM chemical composition is critical in determining soil structure, as well as its response to water infiltration and overland flow [19].

For instance, hydrophobic organic compounds typically reduce water infiltration capacity with the consequence increase in surface runoff [22], thereby intensifying soil erosion in certain conditions [23]—such as after prolonged dry periods followed by heavy rainfall or in areas with limited vegetation cover and variable soil moisture, high slope gradients, and affected by wildfires [e.g., 21,24–28]. These conditions are typical of Mediterranean-type ecosystems [29]. However, the impact of hydrophobicity on water-induced erodibility depends on the balance between hydrophobic and hydrophilic components [30,31] and varies with environmental conditions and soil characteristics [32–35]. Additional SOM-related properties, such as CEC and organic matter-cation associations [36,37], can also serve as indicators of land degradation processes. High levels of these properties improve soil aggregate stability, increasing resistance to disintegration under erosive forces [19]. Anyway, the effect of SOM to these properties is complex and depends on several factors, e.g., chemical SOM characteristics, pH, soil texture, clay minerals, and pedogenic oxides [38–40].

However, the watershed-scale relationships between SOM chemical characteristics, environmental features, and landscape connectivity, along with their impact on organic matter depletion, redistribution, and sediment dynamics, remain relatively underexplored. Addressing these relationships is essential for understanding the role of erosion-sedimentation patterns in carbon cycling and sequestration [41]. Recent studies underscore the need to investigate the chemical properties and spatial distribution of transported organic matter, since it varies significantly due to factors such as land use, altitude, clay content, pH, human disturbance, and hydrological connectivity [42,43]. Additionally, studies focusing on the effects of climatic events, such as rainstorms, on Dissolved Organic Matter (DOM) composition and related export dynamics highlight the implications for water quality protection and pollutant contamination [44,45]. Such interactions in geomorphologically dynamic and human-altered Mediterranean environments remain poorly studied. This emphasizes the need for interdisciplinary studies investigating the interplay between SOM characteristics and environmental factors for understanding watershed system functioning in soil degradation-prone environments [46].

This study aims to propose a novel methodological framework to investigate the geospatial relationships between SOM properties, serving as potential indicators of soil degradation and erosion, and the environmental, geomorphological, and hydrological characteristics in a Mediterranean watershed. The upper Val d’Arda (N-Apennines, Italy) was selected as the study area, representing an intensely altered agricultural-forested temperate Mediterranean watershed with a highly dynamic geomorphological setting [47].

According to Johnston & Aochi [48] and Margenot et al. [49], Fourier-transform infrared (FTIR) spectroscopy can be employed to analyse SOM composition, focusing on specific functional groups and their interactions with cations. This technique is particularly suitable for its rapidity, cost-effectiveness, non-destructive nature, and minimal sample requirements [49]. FTIR spectroscopy enables the analysis of SOM organic constituents by examining absorption bands, such as hydrophobic alkyl (CH, CH₂, CH₃) groups, hydrophilic carbonyl and carboxyl (C = O) groups, hydroxyl (OH) groups, and polysaccharide (C–O–C) groups [50]. By relating the intensities of these absorption bands in ratios (i.e., FTIR indices), it is possible to derive useful information on SOM composition, including relative hydrophobic group content [indicating the potential wettability of SOM; [51,52] and CEC [39,50]. Moreover, interactions between organic matter and polyvalent cations can be assessed by analysing an absorption band linked to both relative carbonate content and the COO^- functional group [53–55].

Thus, to assess the relationships between SOM composition influencing erosion susceptibility and the environmental factors driving its variability at the watershed scale, FTIR indices derived from fluvial sediment samples collected across the watershed were integrated with terrain, soil properties, lithological, and land use data using a Random Forest (RF) machine learning model [56,57]. This integration aimed to uncover new insights into erosion processes, sediment dynamics, and organic matter connectivity at watershed scale. This study, which is novel in its holistic and interdisciplinary approach, not only builds on existing research by linking SOM composition with soil erosion, but also offers a new methodological approach for understanding these processes in complex Mediterranean landscapes. The findings contribute to the broader field of soil conservation by emphasizing the critical role of SOM characterization in assessing soil erosion and organic matter dynamics, offering significant implications for socio-ecological landscapes, where soil erosion and degradation assessments are critical challenges.

2 Study area

The upper Val d’Arda is an agricultural-forested watershed located in the western Emilia-Romagna region of the Northern Apennines (Emilian Apennines), Italy (Fig 1a). The watershed spans approximately 14 km from southwest to northeast and covers around 88 km², with elevation ranging from 285 to 1356 m a.s.l (Fig 1b). In the upper part of the study area, the watershed is drained by two main channels, which converge into the Arda river before flowing into the Mignano reservoir: the Lubiana and Arda creeks (Fig 1b).

Fig 1 — (b) The upper Val d’Arda-Mignano watershed. The main elevation peaks, elevation distribution, hydrographic elements, and other relevant toponyms are represented on the map. The location and distribution of sampling sites (fluvial sediments) are also reported. Base Map: Digital Terrain Model 5x5, Ed. 2014 (© Archivio Cartografico, Regione Emilia-Romagna; https://metasfera.regione.emilia-romagna.it/ricerca_metadato?uuid=r_emiro:2016-08-08T155835) (c) Explanatory excerpt showing the location of some sampling sites along the main channels and stream tributaries (d, e, f, g, h, i). Base Map: Technical Regional Map 1:5,000 (© Archivio Cartografico, Regione Emilia-Romagna; https://metasfera.regione.emilia-romagna.it/ricerca_metadato?uuid=r_emiro:2017-11-24T113154).

The study area represents a typical Mediterranean temperate watershed, characterized by decreasing mean annual temperature and increasing annual precipitation with altitude, and rainfall peaks in April, October, and November exceeding 100 mm. Summers are hot with minimal water deficit [47]. As with many Mediterranean watersheds, the study area consists in a complex palimpsest landscape, with features inherited from I) the past morphoclimatic evolution, II) the activities of past cultures and the remnants of geomorphic processes that partially determine the management of the land, and III) the actual geomorphic setting [9,58]. At the same time, it exhibits highly diversified environmental conditions, with substantial variability in lithological, geomorphological, and Land Use/Land Cover (LULC) characteristics, making it particularly suitable for geospatial watershed-scale modelling. Furthermore, its accessibility as an anthropized watershed provides good opportunities for intensive sampling campaigns and field-based research.

Particularly, the study area features geological formations dating from the Upper Cretaceous to the Paleocene-Eocene periods [59]. According to Martini & Zanzucchi [60], these formations can be grouped based on similar lithological characteristics: I) Silicified Calcilutites and Silty Clays, II) Varicoloured Clays and Shales, III) Carbonate Turbidites, IV) Ophiolitic/Sedimentary Breccias and Olistoliths, V) Arenaceous-Pelitic Turbidites. Landscape evolution is shaped by the interaction of lithological characteristics and local structural features [61]. The watershed alternates morphological traits typical of hilly and mountainous settings, thereby influencing both geomorphic processes and vegetation patterns [47].

The geological setting significantly drives the complex geomorphology of the study area, along with peculiar morpho-structural characteristics. The prevalence of ‘weak rocks’ with a prevalent clayey component makes the watershed highly prone to intense erosion and large mass movements [47,62]. Landslides are the dominant land degradation process, exhibiting considerable variability in magnitude and frequency. In the study area, landslide activity is largely dominated by periodic reactivation of pre-existing large-scale landslide bodies, primarily triggered by intense or prolonged rainfall events. Fluvial undercutting at the toe of landslide deposits also plays a critical role in destabilizing valley slopes. Also, streambank and upland water erosion contribute to severe land degradation, facilitating sediment delivery to the channel network and depositing substantial sediment volumes in the Mignano reservoir [63]. Erosion processes like gullying and subsurface erosion are constrained by lithological features, particularly associated with claystones.

Forest vegetation dominated by oaks, hornbeams, and chestnuts is the most representative LULC type in the watershed, shaping the hilly landscape at lower elevations, while beech forests prevail in mountainous areas at higher altitudes. Water erosion processes such as rill-interrill erosion and gullying appear to be less prevalent in forested environments. This is likely due to a combination of the protective effect of the canopy against rainfall and the inherent difficulty in identifying lower-magnitude erosion processes, especially considering that these areas are generally less accessible for detailed field investigations. Coniferous coppice woods are also common, especially because they are employed to restore degraded and eroded steep slopes. Rainfed arable lands represent the predominant agricultural landscape in the watershed but are significantly affected by rill-interrill erosion, which is strongly influenced by seasonal variations due to changing land cover [64].

The soils of the study area are affected by biochemical alteration and decarbonation, with Calcaric Cambisols being dominant on various parent materials. Calcaric Regosols are found on steep, erosive slopes with scarce vegetation, while Eutric Cambisols and Dystric Cambisols are present on more stable slopes [65,66].

3 Materials and methods

In order to accomplish the objectives of this study, we adopted an interdisciplinary methodological workflow as shown in Fig 2: 1) to assess and visualize geospatial data and prepare the sampling design, we delineated the watershed boundaries and the channel network through GIS-based hydrological modelling from a Digital Terrain Model (DTM); 2) we conducted a sampling campaign to collect fluvial sediments across the watershed from the main channels and their tributaries; 3) samples collected were analysed through Fourier Transform Infrared (FTIR) spectroscopy to characterize SOM composition and calculate four different FTIR proxies (i.e., indices) as potential indicators of soil erosion susceptibility. A dataset of ‘target variables’ was then compiled using values of the calculated indices; 4) several environmental variables including terrain features, soil properties, lithological and LULC datasets were collected in a GIS environment; 5) for each sampling point, we extracted data of the environmental variables from the respective upstream contributing areas through three experimental designs, accounting for different hydrological and geomorphological settings. Then, three datasets of ‘feature variables’ were compiled using values extracted from the variables used; 6) the datasets were finally analysed through a RF machine learning approach to assess the non-linear and geospatial relationships between target and feature variables.

Access to the study area and sampling sites did not require any permission from local authorities, as the watershed is an open-access area, and no private property was violated in any way.

In the present study we used the Projected Coordinate System WGS84 – UTM 32N (EPSG: 32632).

3.1 Watershed and channel network delineation

The selected watershed was delineated using the System for Automated Geoscientific Analyses (SAGA) [version 8.1.1; [67] based on a 5 m DTM of the Emilia-Romagna Region [RER; 68]. A Gaussian Filter of radius 3 was applied for filtering errors and artifacts. The DTM was pre-processed for hydrological modelling using the Deepen Drainage Routes tool, with a threshold of 10 m [69]. Then, we derived channel network and watershed boundary by computing Fow Directions [70] and Flow Accumulation [71,72] (Fig 2). A Strahler order of 5 was used as threshold for the computation of the main channel network [73].

3.2 Sampling of fluvial sediments

To accomplish a representative characterization of soil and associated organic matter eroded from hillslopes in the upper Val d’Arda, we collected a total of 73 fluvial sediment samples (i.e., VAC 01–73; Fig 1b; Fig 2). We systematically collected samples along the main channels (i.e., Lubiana creek and Arda river) and main tributaries, particularly before and after their intersections (Fig 1c). The sampling design was aimed at capturing a representative cross-section of channel bed sediments, assumed to reflect the material transported from upstream contributing areas to the respective downstream sampling points [74; i.e., each sampling site is considered as the outlet of its upstream area]. Hence, our approach ensured comprehensive coverage of the watershed contributing areas.

The sampling campaign was carried out at the end of May 2023, right after an intense hydro-meteorological event that impacted the Emilia-Romagna Region [75,76]. Hence, freshly deposited river sediments were collected shortly after the heavy rainfall event. These sediments were identified by several characteristics indicative of recent deposition, including: I) a light greyish colour; II) a very poorly weathered appearance, unlike older, oxidized deposits; III) a well-sorted grain size distribution ranging from clay to coarse sand; IV) the presence of thin and horizontal layering as well as cross-bedding, ripple marks and micro-dunes; and V) the absence of vegetation or significant biological activity above the surface (e.g., Fig 1d-i). This uniformity in deposition conditions among the samples is critical for maintaining high organic content and preserving biochemical properties [77]. Moreover, samples were collected at locations associated with reduced flow velocity, including small point bars, riffle-pool sequences, and areas downstream of physical obstacles. When sediments are deposited under similar hydrodynamic conditions, they generally exhibit more consistent physical and biochemical properties, which is crucial for comparative analysis [78,79]. Based on these considerations, we assume that the differences observed in the sampled material (Fig 1) can be attributed to a set of environmental variables that influenced the chemical composition of SOM from its origin in the upstream areas to its deposition at the sampling sites.

Sampling along tributaries was carried out at locations close to the intersection with the main stream. In the latter case, particular attention was paid to select locations that remain unaffected by water rise from the main stream during flood events, thus maintaining the integrity of the samples. Four replicates were collected at each sampling site cross-section and later mixed in the field to obtain a representative composite sample [80]. The samples were then air-dried, disaggregated, homogenized, and sieved to ≤ 2 mm [81].

3.3 Fourier transform infrared (FTIR) spectroscopy analysis

We analysed SOM functional groups for the 73 fluvial sediment samples through FTIR spectroscopy, using the potassium bromide (KBr) transmission technique, following Ellerbrock et al. [82] and Ellerbrock & Gerke [50,83] (Fig 2). The laboratory procedure, along with equipment used, software, and specific settings, is reported in S1 Appendix (Supporting Information). In the end, we obtained the absorption spectra of sediment samples in a range of wavenumbers (WN) between 3900 and 400 cm^-1. We further processed and interpreted spectra data using a common procedure, as reported in Ellerbrock et al. [84].

The FTIR spectra exhibit bands that represent absorption of infrared light at frequencies (i.e., wavenumbers, WN; cm^-1) specific to the type and excitation behaviour of the sample [49]. Here, we focused on specific WN regions of the spectra to characterize the following functional groups: I) C–H (band A), II) C = O (band B), III) Carbonate/ν_sCOO^- (band C), IV) C-O-C/Si-O-Si (band D) (Fig 3). The C–H functional groups (i.e., hydrophobic alkyl groups) were analysed in the WN region 3020−2800 cm^-1 [85] (Fig 3). Within this range, the bands at ~ 2920 cm^-1 and ~ 2860 cm^-1 are related to asymmetric (A₁) and symmetric (A₂) stretching vibrations of methyl and methylene groups. The absorption intensities of bands A₁ and A₂ were measured as the vertical distance (i.e., height) from a ‘baseline’ plotted between tangential points on absorption minima to the A₁ and A₂ maxima, following Ellerbrock et al. [86]. Then, the A₁ and A₂ bands were combined into a single band A (Fig 3). The C = O functional groups typically cause bands in the following WN regions: I) 1740−1698 cm^-1 (typical for carboxyl groups in ketones, carboxylic acids, or amides) and II) 1640–1600 cm^-1 (asymmetric carboxylate stretching, i.e., ν_asCOO^- band) [87] (Fig 3). The C = O band intensities were measured as the height from the total baseline of the spectra to the respective band maxima [84]. Hence, the band intensities within the before mentioned WN regions are denoted as Bands B₁ and B₂, respectively. Then, the B₁ and B₂ bands were combined into a single band B (Fig 3). Likewise, the C = O groups cause a third band in the WN region 1450−1382 cm^-1, which accounts for the symmetric vibration of the carboxylate groups [ν_sCOO^- band; 88,89]. However, in FTIR analysis of carbonate-rich soils the ν_sCOO^- band is strongly overlapped by the carbonate band [90], which shows a maximum intensity at around 1420 cm^-1 [53]. Therefore, according to Alvarez-Puebla et al. [54], the ν_sCOO^-/carbonate band (i.e., band C; Fig 3) intensity may be related to the formation of organic matter-cation associations due to a potentially high Ca²⁺ cation content. Finally, the C-O-C functional groups, indicating the presence of polysaccharides or ether groups often associated with hydrophilic components of SOM [84], were analysed in the WN region 1120−1000 cm^-1 [55; band D] (Fig 3). However, in soil/sediment samples, the band D may be overlapped by an intense Si-O-Si band due to the presence of silicates and aluminosilicates.

We calculated, for each sample, four indices as the ratios of the band peaks (i.e., A/B, A/D, B/D, and C/D; Table 1; Fig 2). Then, a dataset of ‘target variables’ was compiled using values of the calculated indices (Fig 2). Afterwards, a Spearman’s rank correlation analysis was performed to assess the rank correlation between the indices. Moreover, the index values were classified into 10 classes according to the Jenks natural breaks method [91] and displayed in a GIS environment to identify and visualize spatial patterns.

Table 1. Indices for SOM characterization calculated as FTIR band ratios, used in the present study.

Index	Bands ratio	Meaning	References
$\frac{A}{B}$	$\frac{(A_{1} + A_{2})}{(B_{1} + B_{2})}$	Relative hydrophobicity of SOM. A higher ratio indicates a higher content of hydrophobic groups, suggesting that the sediment has a reduced wettability.	[51] [52]
$\frac{A}{D}$	$\frac{(A_{1} + A_{2})}{D}$	Relative hydrophobicity of SOM. A higher ratio indicates a higher content of hydrophobic groups, suggesting that the sediment has a reduced wettability.	[84]
$\frac{B}{D}$	$\frac{(B_{1} + B_{2})}{D}$	Relative Cation Exchange Capacity (CEC) of SOM. A higher ratio is often correlated with a higher CEC because carbonyl and carboxyl groups can participate in cation exchange processes.	[39] [50]
$\frac{C}{D}$	$\frac{C}{D}$	For SOM, it can be used as a relative measure of organic matter-cation associations. A higher ratio suggests a greater degree of interaction between SOM and cations.	[54] [55]
$\frac{C}{D}$	$\frac{C}{D}$	Depending on the mineral composition, it is a measure of the relative carbonate content in the sample.	[53]

Open in a new tab

3.4 Preparation of environmental variable datasets

In this study, we used several watershed-scale raster datasets as environmental variables, including terrain (morphometric) features and soil properties, as well as vector data on lithology and LULC (Fig 2).

In particular, we performed a DTM-based terrain analysis using the SAGA GIS software to characterize the main land-surface features which potentially influence hydrological processes, soil erosion and, in general, sediment dynamics [92,93] (Fig 2). Table 2 summaries the terrain variables used in this study, along with a description of their geomorphological significance. As they were calculated from the DTM [68], the final resolution of the outputs is 5 m. Detailed information on the methods used for calculating these variables is included in S1 Appendix (Supporting Information).

Table 2. Terrain variables used in the present study, along with their geomorphological significance. Sources: [92,94–102].

Terrain variable	Method/ Reference	Unit	Geomorphological significance
Elevation (DTM)	[68]	Meters	Climate, vegetation, potential energy.
Flow Accumulation	[72]	Meters²	Contributing catchment area, steady-state runoff rate.
Slope	[103]	Degrees	Precipitation, overland and subsurface flow velocity and runoff rate, soil water content.
Aspect	[103]	Degrees	Flow direction, solar insolation, evapotranspiration, influence on vegetation distribution.
Profile curvature	[103]	1/ meters	Flow acceleration and deceleration, soil erosion and deposition rates.
Tangential curvature	[103]	1/ meters	Local flow convergence and divergence.
General curvature	[103]	1/ meters	General measure of terrain convexity, identifying convex surface as interfluves and peaks and concave surface as valleys and cavities.
Total curvature	[103]	1/ meters	Comprehensive measure that sums up all components of surface curvature, influencing water flow and erosion patterns.
Downslope Distance Gradient	[104]	Meters	Rate of elevation change over a horizontal distance, providing a measure of slope steepness and flow direction for understanding flow paths, erosion potential, and landscape stability.
SAGA Wetness Index	[92]	Dimensionless	Spatial distributions and extent of areas of high soil moisture and saturation excess overland flow as a function of upslope contributing area, soil transmissivity, and slope.
Terrain Ruggedness Index	[105]	Meters	Heterogeneity or ruggedness of the terrain (terrain complexity).
Stream Power Index	[94]	Meters	Potential erosive power of flowing water (based on the assumption that discharge is proportional to the specific catchment area).
Channel Network Base Level	[102]	Meters	Erosion potential and sediment transport/deposition efficiency in relation to the landscape topography.
Vertical Distance to Channel Network	[102]	Meters	Identification of fluvial terrace systems, palaeo-surfaces, planation surfaces, and erosion and deposition zones.

Open in a new tab

Moreover, the RER provided the regional raster datasets of soil physical and chemical properties (https://datacatalog.regione.emilia-romagna.it/catalogCTA/) (Fig 2). The datasets, which are related to the topsoil (0–30 cm), were elaborated at 1:50,000 scale by means of a Digital Soil Mapping approach, with a pixel resolution of 100 m. In the present study, we considered the following soil properties:

Soil Organic Carbon (SOC) stock (expressed in Mg ha^-1) [106];
Soil Organic Carbon (SOC) content (expressed in %) [107];
Clay content (particles with diameter ≤ 2 µm, in %) [108];
Silt content (particles with diameter > 2 µm and ≤ 50 µm, in %) [108];
Sand content (particles with diameter > 50 µm and ≤ 2 mm, in %) [108];
Skeleton content (particles with diameter > 2 mm, in %) [108];
Soil pH [109].

All the datasets are provided along with the related pixel-based cartographic accuracy (i.e., low, medium, or high accuracy), defined as the standard deviation of the estimated value in each pixel (i.e., the lower the standard deviation, the more accurate the estimate). For the study area, all the datasets have, on average, a medium accuracy. Further details concerning the full methodology (i.e., spatial prediction, validation, regionalization, and accuracy evaluation) can be found in Ungaro et al. [106–109]. Afterwards, all the datasets were resampled at 5 m resolution using the Nearest Neighbour method. Due to the nature of the data, the initial resolution and processing methods of these datasets may potentially introduce errors and uncertainties related to data representation and spatialization. Nevertheless, resampling and alignment with outputs derived from the DTM ensures consistency in spatial resolution, which is essential for accurate and homogeneous pixel value sampling across all datasets and for reliable statistical analysis.

Additionally, the RER also provided the regional shapefile datasets of land use and lithological data (Fig 2). In particular, the following datasets were used:

Geological units, 1:10,000 scale [110];
Land use, 1:10,000 scale [111].

Data were managed using the Esri^© ArcMap software (version 10.3.1; ArcGIS). Then, geological units were merged based on lithological characteristics according to Servizio Geologico d’Italia [59] and Martini & Zanzucchi [60], obtaining 5 lithological groups as shown in Table 3. Also, land use classes were grouped to obtain 10 LULC types as shown in Table 4 (rivers and water bodies were excluded).

Table 3. Lithological groups of the upper Val d’Arda-Mignano watershed. Shapefile data derived from the Geological Units database of the Emilia-Romagna Region [1:10,000 scale; 110]. The description of the lithological groups was adapted from the Geological Sheet 198 – Bardi [1:50,000 scale; 59].

Label	Lithological group	Description
LI_1	Silicified Calcilutites and Silty Clays	Silicified calcilutites in medium to thick beds with intercalations of marly limestones. Silty clays with frequent intercalations of thin-bedded fine turbiditic sandstones (e.g., Palombini Shales).
LI_2	Varicoloured Clays and Shales	Varicoloured clays and shales with intercalations of fine turbiditic sandstones and thin layers of calcilutites (e.g., Cassio Varicoloured Clays).
LI_3	Carbonate Turbidites	Turbidites composed of calcareous marls, marly limestones, and marls with an arenitic or calcarenitic base. Alternation of sandstones, shales, calcareous marls, and marls (e.g., Monte Cassio Flysch).
LI_4	Ophiolitic/Sedimentary Breccias and Olistoliths	Matrix-supported and grain-supported breccias, containing sedimentary and ophiolitic elements. Olistoliths made up of calcilutites, sandstones, serpentinites, basalts, granites, and jaspers (e.g., Ophiolitic Sandstones of the Pietra Parcellara Complex).
LI_5	Arenaceous-Pelitic Turbidites	Alternations of arenaceous-pelitic rocks in thin to medium beds, with rare intercalations of marls and thin to medium bedded laminated limestones (e.g., Scabiazza Sandstones).

Open in a new tab

Table 4. LULC types of the upper Val d’Arda-Mignano watershed. Shapefile data derived from Land use database of the Emilia Romagna Region [1:10,000 scale; 111].

Label	LULC type
LU_1	Coniferous forests
LU_2	Beech forests
LU_3	Oak, hornbeam and chestnut forests
LU_4	Riparian vegetation
LU_5	Cultivated fields
LU_6	Anthropic areas
LU_7	Bare surfaces/ rocky outcrops
LU_8	Sparse vegetation in evolution
LU_9	Meadows
LU_10	Bushes and shrubs

Open in a new tab

3.5 Variables extraction based on different experimental designs

We extracted data related to the variables reported in the previous section from the contributing areas (CAs) of the sampling points, in a GIS environment (Fig 2). In this study, we consider the CA of each sample as the region where soil erosion and downstream sediment transport occur, with the sampling site representing the respective outlet. To determine the best strategy for extracting data from CAs, we set up three different experimental designs, representative for different hydrological and geomorphological settings (Fig 4):

Fig 4 — The coloured areas delimited in the three experimental designs were used to extract values from the feature variables. This methodology was replicated for each of the 73 sediment samples (Fig 1b). Base Maps: Hillshade Map derived from DTM 5x5, Ed. 2014 (© Archivio Cartografico, Regione Emilia-Romagna; https://metasfera.regione.emilia-romagna.it/ricerca_metadato?uuid=r_emiro:2016-08-08T155835) and Technical Regional Map 1:5,000 (© Archivio Cartografico, Regione Emilia-Romagna; https://metasfera.regione.emilia-romagna.it/ricerca_metadato?uuid=r_emiro:2017-11-24T113154). Maps produced using the Esri© ArcMap software (version 10.3.1; ArcGIS).

Experiment 1. We used the entire hydrological areas upstream of each sampling point (hereinafter called ‘Upstream areas Exp_1’; Fig 4a). In particular, CAs are calculated with the Deterministic 8 method following O’Callaghan & Mark [112].
Experiment 2. CAs were narrowed to sediment sources and sinks (hereinafter called ‘Sources and sinks Exp_2’; Fig 4b), without discriminating between the type of landform/process, the activity status, the connectivity to the drainage system, nor between the erosional or depositional dynamic. We used the entire shapefile mask (i.e., “dissolved”) of the ‘Inventory Map’ (IM) of sediment sources and sinks of the upper Val d’Arda published in La Licata et al. [62,113]. The latter was subsequently clipped within single CAs.
Experiment 3. CAs were further narrowed to the geomorphological areas with the potential to produce sediment and deliver it to the main channels (hereinafter called ‘Hazard areas Exp_3’; Fig 4c). We employed the ‘Hazard Map of sediment Production and Delivery’ (HPD) of the upper Val d’Arda published in La Licata et al. [62], only focusing on the areas characterized by a medium, high, or very high hazard for sediment sourcing and delivery. These hazard-prone areas mostly include landslides highly connected to the drainage system and overlaid by additional processes like rill-interrill erosion, gullying, stream incision, and bank erosion, as well as steep rock walls affected by rockfalls and debris flows. The HPD was generated as output of the HOTSED model, covering geospatial information of sediment sources/sinks and related dynamics with the assessment of structural and functional properties of sediment connectivity [62,114]. Also in this case, the HPD map was clipped within single CAs.

Finally, we extracted for each experiment data of the environmental variables, as follows:

Mean value of the raster in the CA: Elevation, Slope, Aspect, Profile Curvature, Tangential Curvature, General Curvature, Total Curvature, SAGA Wetness Index, Terrain Ruggedness Index, Stream Power Index, Channel Network Base Level, Vertical Distance to Channel Network, SOC stock, SOC %, Clay %, Silt %, Sand %, Skeleton %, and pH.
Absolute value of the raster at the sampling point (outlet of the CA): Flow Accumulation and Downslope Distance Gradient.
Proportion of each class in the CA for lithological groups and LULC types: Silicified Calcilutites and Silty Clays, Varicoloured Clays and Shales, Carbonate Turbidites, Ophiolitic/Sedimentary Breccias and Olistoliths, Arenaceous-Pelitic Turbidites, Coniferous forests, Beech forests, Oak, hornbeam and chestnut forests, Riparian vegetation, Cultivated fields, Anthropic areas, Bare surface/Rocky outcrops, Sparse vegetation in evolution, Meadows, and Bushes and Shrubs.

In addition, we calculated the Altitude Difference as the difference between the mean elevation of the CA and the elevation at the sampling point. Moreover, the Area, Length, and Width of each CA were calculated as geometric parameters. Specifically, Length and Width were determined as the longest and shortest sides of a minimum bounding rectangle that encloses the CA, respectively.

In the end, we obtained three different datasets related to the three experimental designs, each containing 40 ‘feature variables’.

3.6 Determination of non-linear and geospatial relationships using Random Forest algorithm

The datasets were analysed through a Random Forest (RF) model to determine the non-linear and geospatial relationships between target and feature variables (Fig 2). Initially, feature selection was performed using the Spearman’s correlation filter from the ‘FSelector’ package in R [115]. We selected the top 10 most significant features for each target variable, helping to streamline the data and focus on the most influential variables. The RF algorithm, which is an ensemble machine learning method [56], was then used to model the relationships between the selected features and target variables. RF operates by constructing multiple decision trees during training and outputting the mean prediction of the individual trees. It excels in handling complex and non-linear relationships by averaging the results from numerous trees to improve accuracy and control overfitting [57]. The RF prediction is performed using the rf function in the ‘caret’ package in R [116,117].

To evaluate model performance, we implemented Leave-One-Out Cross-Validation (LOOCV) on the entire dataset as it is beneficial for small datasets, detects overfitting, and provides error estimates with comparatively good bias and variance properties [118]. In LOOCV, each data point is used once as a test set while the remaining data points form the training set. This technique ensures that every observation is tested exactly once, providing a comprehensive evaluation of model performance. Within each training dataset, we performed a Grid Search to optimize the ‘mtry’ parameter of the RF model. The ‘mtry’ parameter, which controls the number of variables randomly sampled as candidates at each split in the decision trees, was tuned between 2 and 10 using 10-fold cross-validation. The optimal ‘mtry’ value was used to develop the final model, which was tested on the single data point left out in each iteration of LOOCV.

The performance of each RF model was assessed using R-square (R²), Lin’s Concordance Correlation Coefficient (CCC), Mean Squared Error (MSE), Root Mean Squared Error (RMSE), Bias, and Nash-Sutcliffe Efficiency (NSE), as commonly defined in literature [e.g., 57,119–121]. These metrics were performed with the help of goof function from the ‘ithir’ package [122] and ‘hydroGOF’ package in R [123].

For each model developed, post-hoc interpretability was conducted using the ‘iml’ package in R [124]. We first assessed the importance of each feature using the variable permutation method. This technique evaluates the effect of permuting (shuffling) each feature on the model’s performance, thereby indicating the importance of each feature in predicting the target variable. Additionally, we employed Accumulated Local Effects (ALE) plots to explore interactions between features and target variables. ALE plots provide insights into how the predicted values change as a feature varies, accounting for interactions with other features. The ALE plot helps in understanding the local effect of each feature while controlling for the effects of other features, thus revealing nuanced relationships and interactions [125].

4 Results

4.1 FTIR indices for SOM characterization (target variables)

The FTIR analysis carried out on collected samples (i.e., VAC 01–73; Fig 1b) provided the intensity of the band maxima related to selected SOM functional groups (Fig 3): I) hydrophobic alkyl groups (Bands A₁ and A₂), II) hydrophilic carboxyl and carboxylate groups (Bands B₁ and B₂), III) carbonate content/symmetric vibration of COO^- (Band C), and IV) polysaccharides hydrophilic groups (C-O-C; Band D) (Fig 3).

Over the 73 analysed samples, Band A₁ has a mean value of 0.005, a median of 0.004, and a standard deviation (STD) of 0.003. Band A₂ has a mean value of 0.01, a median of 0.01, and a STD of 0.04. Band B₁ has a mean value of 0.11, a median of 0.11, and a STD of 0.03. Band B₂ has a mean value of 0.20, a median of 0.19, a STD of 0.05. Band C has a mean value of 0.86, a median of 0.77, and a STD of 0.51. Band D has a mean value of 1.32, a median of 1.24, and a STD of 0.42.

Furthermore, 4 indices were calculated as FTIR band ratios (Table 1) to characterize the SOM in the samples, in terms of relative hydrophobicity (i.e., A/B and A/D), relative CEC (i.e., B/D), and relative intensity of organic matter-cation interactions or relative carbonate content (i.e., C/D) (Table 5). A/B and A/D show relatively low values with little variation, indicating that these indices are consistent across the samples. In particular, A/B index has a mean value of 0.04, a median of 0.04, and a STD of 0.02. Similarly, A/D index has a mean value of 0.01, a median of 0.01, and a STD of 0.01. Instead, B/D index shows more variability, but its values still remain within a moderate range. B/D index has a mean value of 0.25, a median of 0.24, and a STD of 0.09. Differently, C/D index exhibits the most significant variation, with a wide range indicating that the relative carbonate content, and consequently the interaction between SOM and calcium, varies considerably among the samples. C/D index has a mean value of 0.71, a median of 0.56, and a STD of 0.47.

Table 5. FTIR Indices calculated for the 73 fluvial sediment samples in the upper Val d’Arda-Mignano watershed (cfr. Table 1).

Sample	A/B	A/D	B/D	C/D	Sample	A/B	A/D	B/D	C/D
VAC01	0.067	0.021	0.306	0.994	VAC38	0.028	0.006	0.215	0.377
VAC02	0.075	0.052	0.689	2.012	VAC39	0.010	0.002	0.176	0.275
VAC03	0.026	0.006	0.220	0.601	VAC40	0.021	0.003	0.155	0.334
VAC04	0.056	0.017	0.296	0.708	VAC41	0.034	0.008	0.244	0.996
VAC05	0.039	0.013	0.322	1.428	VAC42	0.043	0.010	0.239	1.070
VAC06	0.060	0.013	0.224	0.764	VAC43	0.037	0.009	0.244	0.695
VAC07	0.074	0.034	0.465	1.655	VAC44	0.044	0.013	0.285	1.136
VAC08	0.037	0.012	0.325	1.435	VAC45	0.032	0.008	0.236	0.722
VAC09	0.054	0.014	0.257	1.222	VAC46	0.045	0.012	0.272	0.724
VAC10	0.059	0.021	0.360	0.433	VAC47	0.046	0.007	0.162	0.331
VAC11	0.044	0.015	0.335	0.830	VAC48	0.011	0.002	0.202	0.175
VAC12	0.037	0.012	0.321	1.005	VAC49	0.016	0.003	0.179	0.214
VAC13	0.067	0.023	0.348	1.878	VAC50	0.041	0.010	0.237	0.488
VAC14	0.040	0.009	0.234	0.764	VAC51	0.014	0.003	0.183	0.295
VAC15	0.031	0.007	0.234	0.532	VAC52	0.027	0.006	0.219	0.495
VAC16	0.025	0.006	0.253	0.384	VAC53	0.021	0.005	0.237	0.502
VAC17	0.047	0.006	0.125	0.174	VAC54	0.033	0.007	0.204	0.584
VAC18	0.016	0.004	0.235	0.267	VAC55	0.030	0.011	0.367	0.822
VAC19	0.005	0.001	0.219	0.233	VAC56	0.031	0.008	0.277	0.910
VAC20	0.012	0.002	0.134	0.166	VAC57	0.037	0.012	0.318	1.215
VAC21	0.045	0.009	0.209	0.377	VAC58	0.025	0.005	0.184	0.389
VAC22	0.054	0.011	0.206	0.542	VAC59	0.042	0.012	0.300	1.179
VAC23	0.039	0.007	0.174	0.559	VAC60	0.015	0.003	0.189	0.310
VAC24	0.012	0.002	0.131	0.176	VAC61	0.047	0.008	0.166	0.271
VAC25	0.027	0.006	0.208	0.550	VAC62	0.033	0.005	0.147	0.116
VAC26	0.029	0.004	0.139	0.206	VAC63	0.019	0.004	0.219	0.269
VAC27	0.011	0.002	0.165	0.201	VAC64	0.052	0.013	0.241	1.107
VAC28	0.013	0.002	0.148	0.195	VAC65	0.036	0.012	0.321	1.400
VAC29	0.013	0.002	0.175	0.219	VAC66	0.046	0.013	0.272	0.788
VAC30	0.056	0.016	0.276	0.888	VAC67	0.020	0.005	0.225	0.446
VAC31	0.033	0.007	0.208	0.480	VAC68	0.040	0.007	0.182	0.537
VAC32	0.052	0.016	0.312	1.286	VAC69	0.049	0.011	0.213	0.531
VAC33	0.051	0.019	0.369	2.070	VAC70	0.040	0.010	0.242	0.641
VAC34	0.017	0.004	0.241	0.385	VAC71	0.041	0.014	0.336	0.871
VAC35	0.048	0.016	0.345	0.846	VAC72	0.093	0.022	0.241	1.341
VAC36	0.049	0.012	0.246	0.961	VAC73	0.067	0.017	0.251	1.300
VAC37	0.024	0.005	0.222	0.342	–	–	–	–	–

Open in a new tab

Table 6 summaries the Spearman correlation coefficients among the calculated indices. Indices A/B and A/D are strongly correlated (i.e., 0.92), indicating that SOM hydrophobicity is well characterized in this study. The high correlation between A/D and C/D indices (i.e., 0.86) implies that hydrophobicity and relative carbonate content in the samples may be closely linked. However, even though the high correlation between B/D and C/D indices, their slightly lower coefficient value (i.e., 0.83) suggests that the two indices are capturing slightly different aspects of SOM-cation dynamics in the study area. The moderate correlation between A/B and C/D indices (i.e., 0.68), as well as the lower correlation between A/B and B/D indices (i.e., 0.55), suggests that the spatial distribution and accumulation of hydrophobic vs. hydrophilic SOM components might be driven by different landscape processes and dynamics (e.g., water fluxes, sediment erosion, transport, and deposition) and/or environmental conditions.

Table 6. Spearman’s rank correlation coefficients among the calculated FTIR indices (cfr. Table 1).

	A/B	A/D	B/D	C/D
A/B	1.00	0.92	0.55	0.68
A/D	0.92	1.00	0.81	0.86
B/D	0.55	0.81	1.00	0.83
C/D	0.68	0.86	0.83	1.00

Open in a new tab

Considering the spatial distribution of collected samples (Fig 1b), A/B index varies significantly across the landscape, with higher values (i.e., indicating more hydrophobic SOM) concentrated in specific areas (Fig 5a). In the uppermost areas of the watershed, A/B values are generally lower, indicating a lower ratio of hydrophobic to hydrophilic functional groups (Fig 5a). In correspondence of the middle sections of the watershed there is a noticeable increase in A/B, whereas in the lower watershed, particularly near the confluence of main channels and larger tributaries, the index values are more variable. Main channels, particularly those flowing through the middle part of the watershed, tend to exhibit higher A/B values, while the tributaries generally show lower A/B values, especially in the upper and middle parts of the watershed (Fig 5a). Similarly, in the uppermost areas of the watershed, the A/D index is generally low (Fig 5b), suggesting a lower ratio of hydrophobic alkyl groups (C-H) to hydrophilic polysaccharides (C-O-C) and silicate (Si-O-Si) groups, while in the middle parts of the watershed the index values show some increase, although they remain relatively low overall (Fig 5b). In the lower watershed, near the confluence of main channels and tributaries, A/D values remain low to moderate, with fewer areas showing high ratios (Fig 5b). In general, both the main channels and tributaries display low A/D values, with only small increases in specific areas, suggesting minimal differences in hydrophobicity between these watercourse types. The B/D index exhibits a similar trend of A/B across the landscape, with a clear increase from the upper to the lower watershed (Fig 5c). The main channels show moderate to high B/D values (i.e., highlighting a greater proportion of carbonyl groups), particularly in the middle and lower watershed, while the tributaries generally exhibit lower values, especially in the upper reaches (Fig 5c). Finally, the C/D index shows a noticeable increase from the upper to the lower watershed, particularly in the main channels (Fig 5d). Main channels have consistently higher C/D values compared to tributaries (Fig 5d).

Fig 5 — Values are classified into 10 classes according to the Jenks natural breaks method, for a better visual representation. Base Map: Hillshade Map derived from DTM 5x5, Ed. 2014 (© Archivio Cartografico, Regione Emilia-Romagna; https://metasfera.regione.emilia-romagna.it/ricerca_metadato?uuid=r_emiro:2016-08-08T155835). Maps produced using the Esri© ArcMap software (version 10.3.1; ArcGIS).

4.2 Datasets selected as feature variables

The GIS-based terrain analysis provided 14 terrain variables that have been used as feature variables (Fig 6). Moreover, the raster datasets for 7 soil properties (i.e., SOC stock, SOC %, Clay %, Silt %, Sand %, Skeleton %, and pH) have been added to the list of feature variables used in the modelling procedure, as well as the lithology and land use shapefile data (Fig 7). Basics statistics including min-max range, mean, and standard deviation for each raster variable are reported in S2 Appendix (Supporting Information).

The geological units have been subdivided into 5 groups based on lithological characteristics as reported in Table 3 and Fig 7h. In particular, the Silicified Calcilutites and Silty Clays group is well extended in the study area, especially in the right-side of the watershed, covering a total of 19.4 km². Varicoloured Clays and Shales group extends only for 4.9 km² and is limited to the left-side of the watershed. Carbonate Turbidites group extends for 46.4 km² and it is widespread in the whole watershed. Ophiolitic/Sedimentary Breccias and Olistoliths group extends for 5.2 km² and is limited to the upper parts of the watershed. Finally, Arenaceous-Pelitic Turbidites group is extended for 12.7 km².

Land use classes have been subdivided into 10 unified LULC types as reported in Table 4 and Fig 7i. The LULC types with the related extent in the study area are: I) Coniferous forests, 3.8 km², II) Beech forests, 13.8 km², III) Oak, hornbeam and chestnut forests, 42.2 km², IV) Riparian vegetation, 0.7 km², V) Cultivated fields, 15.7 km², VI) Anthropic areas, 3.7 km², VII) Bare surfaces/Rocky outcrops, 0.4 km², VIII) Sparse vegetation in evolution, 0.9 km², IX) Meadows, 2.6 km², X) Bushes and shrubs, 4.0 km².

Overall, these data represent a substantial set of environmental variables and watershed characteristics that potentially influence the geospatial variability of organic matter characterization indices used as potential indicators of erosion susceptibility.

4.3 Data extraction from feature variables across the experiments

The data extraction procedure across the three experiments have provided three different datasets, all of which encompass 73 rows (i.e., samples), 4 fields containing the four FTIR indices (i.e., target variables; Table 5) and 40 fields containing data extracted from the feature variables (Figs 6 and 7). The results of the data extraction from the feature variables, for the three experiments, are reported within the related datasets at the following link: https://doi.org/10.5281/zenodo.15097417 (they will not be part of the discussion).

4.4 Random Forest models’ performance evaluation

We employed three different modelling strategies (i.e., Experiments 1–3; Fig 4) for relating different hydrological/geomorphic settings to FTIR-derived indices used as indicators of soil erosion susceptibility (Table 1).

Experiment 1 provided the best performance for the A/B index, in terms of model accuracy and agreement between predicted and observed values (R² = 0.39, CCC = 0.59, NSE = 0.38), followed by Experiment 3 (R² = 0.35, CCC = 0.53, NSE = 0.35). Experiment 2 showed slightly lower predictive power (R² = 0.33, CCC = 0.53, NSE = 0.31) (Table 7). For the A/D index, Experiment 1 similarly exhibited the best model performance in terms of predictive power (R² = 0.45, CCC = 0.62, NSE = 0.45), followed by Experiment 2 (R² = 0.37, CCC = 0.54, NSE = 0.37). Experiment 3 showed the lowest performance (R² = 0.32, CCC = 0.48, NSE = 0.32) (Table 7). For the B/D index, Experiment 3 performed better (R² = 0.35, CCC = 0.50, NSE = 0.35) than Experiment 1 (R² = 0.32, CCC = 0.50, NSE = 0.32) and Experiment 2 (R² = 0.33, CCC = 0.47, NSE = 0.33) (Table 7). For the C/D index, Experiment 1 showed the best performance (R² = 0.39, CCC = 0.57, NSE = 0.39), while Experiment 2 and Experiment 3 produced lower but comparable performance (R² = 0.38, CCC = 0.56, NSE = 0.38) (Table 7).

Table 7. Models’ performance metrics for the three experiments (cfr. Fig 4).

FTIR Index	Experiment	R-squared	CCC	MSE	RMSE	Bias	NSE
$\frac{A}{B}$	1	0.39	0.59	0.00	0.01	0.00	0.38
	2	0.33	0.53	0.00	0.01	0.00	0.31
	3	0.35	0.53	0.00	0.01	0.00	0.35
$\frac{A}{D}$	1	0.45	0.62	0.00	0.01	0.00	0.45
	2	0.37	0.54	0.00	0.01	0.00	0.37
	3	0.32	0.48	0.00	0.01	0.00	0.32
$\frac{B}{D}$	1	0.32	0.50	0.00	0.07	0.00	0.32
	2	0.33	0.47	0.00	0.07	0.00	0.33
	3	0.35	0.50	0.00	0.07	0.00	0.35
$\frac{C}{D}$	1	0.39	0.57	0.13	0.36	0.01	0.39
	2	0.38	0.56	0.13	0.37	0.00	0.38
	3	0.37	0.56	0.14	0.37	0.00	0.37

Open in a new tab

Considering MSE, RMSE, and Bias, for the A/B index, the models consistently exhibited similar performance across all three experiments (Table 7). Notably, the MSE remained at 0.00, the RMSE was 0.01, the Bias was also 0.00. This consistency implies that the models displayed uniform predictive accuracy and minimal error for this index across the three experimental settings. Similarly, for the A/D index, the models also achieved an MSE of 0.00, an RMSE of 0.01, and a Bias of 0.00, indicating comparable model performance (Table 7). However, for the B/D index, the models showed a higher RMSE of 0.07 across all three experiments, with a Bias of 0.00 and an MSE of 0.00 (Table 7), suggesting that while the error was slightly elevated, the model predictions remained unbiased. For the C/D index, Experiment 1 yielded a slightly higher MSE of 0.13, an RMSE of 0.36, and a Bias of 0.01 (Table 7), indicating a minor degradation in performance compared to other indices. In contrast, Experiments 2 and 3 for the same index produced MSE values of 0.13 and 0.14, RMSE of 0.37, and Bias of 0.00 (Table 7). This suggests that the model performed similarly in Experiments 2 and 3 but slightly better in Experiment 1 based on the MSE and RMSE metrics.

Hence, Experiment 1 performed generally better than the other two experimental designs, demonstrating that the entire upstream hydrological contributing areas (Fig 4a) provide the best way for investigating the interplay between SOM composition and selected feature variables in this landscape setting (Table 7).

4.5 Feature importance and variable interactions

In this section we report the results for the best performing experiment for each FTIR index (Table 7): A/B, Experiment 1; A/D, Experiment 1; B/D, Experiment 3; C/D, Experiment 1. Fig 8 shows permutation based on feature importance, while ALE plots are provided in Figs 9–12. Additionally, the feature importance permutation plots and ALE plots for all the experiments are included in Supporting Information (S1 and S2 Figs).

In general, the results provided valuable insights into the effectiveness of various environmental factors in predicting geospatial variability of SOM composition in the upper Val d’Arda-Mignano watershed.

The most important variable influencing A/B index is Oak, hornbeam, and chestnut forests, followed by Vertical Distance to Channel Network (Fig 8a), emphasizing the significant role of vegetation type and proximity to the valley bottom (i.e., main channels) in shaping SOM hydrophobic and hydrophilic characteristics. In particular, ALE plots reveal a noticeable increase in A/B as the presence of oak, hornbeam and chestnut forests approaches 0.30 of proportion in contributing areas, indicating that this forest type has a strong positive influence on SOM hydrophobicity (Fig 9), while over 0.50 the effect of this variable tends to level off. Similarly, ALE plots show a steady increase in A/B with increasing vertical distance from the channel network, especially between 200 and 300 m (Fig 9). Carbonate Turbidites and Aspect show moderate influence in the prediction of A/B index (Fig 8a), with ALE plots suggesting that the relationship between mineral and SOM composition, as well as for slope orientation, has a relatively minor and inconsistent influence compared to other features (Fig 9). Channel Network Base Level, Bare surfaces/Rocky outcrops, Beech forests, Sand %, Clay %, and Elevation show relatively lower importance (Fig 8a).

Vertical Distance to Channel Network is the most significant variable also in predicting A/D index, followed by Oak, hornbeam and chestnut forests, pointing out the critical role of topography and forest type in influencing hydrophobicity (Fig 8b). In particular, ALE plots show a steady increase in A/D as the vertical distance from the channel network increases above 250 m, whereas also in this case the positive effect of oak, hornbeam and chestnut forests emerges over 0.30 of proportion in the contributing areas (Fig 10). Clay %, Channel Network Base Level, and Carbonate Turbidites also appear to have a relevant role in influencing A/D prediction by the RF model (Fig 8b). Particularly, there’s a significant increase in A/D at high clay percentages (> 30%), as well as at high proportion of carbonate turbidites (> 0.90), highlighting the significant effects of soil texture and lithological characteristics (Fig 10). Conversely, there is a slight decrease in A/D as the channel network base level increases above 600 m (Fig 10).

Concerning B/D index, Elevation and Channel Network Base Level emerge as the most influential variables (Fig 8c), emphasizing that elevation gradient and proximity to erosional base level strongly influence the spatial distribution of the index. Particularly, Fig 11 shows a sharp decline in index values around 500 m of elevation, followed by a moderate decrease until a threshold is reached (around 800 m). Above 800 m, the effect on the index tends to level off and becomes more stable, exhibiting a slightly negative effect (similarly to the channel network base level; Fig 11). SOC % is also an important variable (Fig 8c), showing a strong positive effect below 1.75%, followed by a sharp decrease as the percentage of SOC increases (Fig 11). Conversely, Oak, hornbeam, and chestnut forests, which also has a valuable effect on B/D index (Fig 8c), exhibits a negative effect below 0.25 of proportion in the contributing areas, followed by a moderate increase between 0.25 and 0.50 (Fig 11). Other variables such as SOC stock, Beech forests, Downslope Distance Gradient, and Aspect are progressively decreasing in importance (Fig 8c).

Channel Network Base Level has the highest importance for C/D index (Fig 8d), where ALE plots highlight a strong positive effect of the proximity to erosional base level on carbonate content below 500 m, as well as a substantial negative effect above 600 m (Fig 12). Anthropic areas is the second most important variable (Fig 8d), exhibiting a significant increase as the proportion of the variable increases (Fig 12), suggesting that human activities have a significant effect on carbonate content, as well as on SOM composition and its interaction with cations like calcium. Also in this case Oak, hornbeam, and chestnut forests is a critical variable, followed by Altitude difference, Vertical Distance to Channel Network, Downslope Distance Gradient, and Aspect (Fig 8d). Notably, also in this case Elevation exhibits a threshold of 800 m, which separates the positive effect of elevation at lower altitudes and its negative effect at higher ones (Fig 12).

5 Discussion

5.1 FTIR-derived indices as indicators of soil degradation and erosion processes

The results of the RF model performance evaluation suggest that considering broader areas widely influenced by surface hydrological and geomorphic processes (Experiment 1), rather than focusing solely on sediment-related landforms and associated processes (Experiments 2 and 3), offers a more valuable framework for modelling indicators of soil erosion susceptibility in the upper Val d’Arda (Fig 4; Table 7). This is particularly evident for the two proxies representing the relative hydrophobicity of SOM and for the organic matter-cation associations (Table 1). It is important to note that Experiment 2 limits the contributing areas (CAs) of the sampling points to regions specifically characterized by mapped geomorphic features, including areas impacted by rill-interrill erosion, various types of landslides, badlands, gullies, and fluvial erosion [62,113], whereas Experiment 3 focuses exclusively on hazard-prone areas [62]. However, fluvial sediment collected during the sampling campaign appear to not reflect very well these processes (Fig 1b). Thus, we hypothesize that this evidence might be linked to the type of erosion process captured in the sampling. Sheet erosion was likely the dominant erosion process during the heavy rainfalls occurred in the upper Val d’Arda in May 2023, pouring out sediment and organic matter from upland areas into the drainage system. Although the watershed experienced severe sediment supply into the rivers, the hydro-meteorological event occurred in May 2023 likely impacted less on geomorphic activity than many other watersheds in the same region [i.e., Emilia-Romagna; see 126–128]. Indeed, no significant landslide (re)activations, hydro-geological hazards, or floods were observed by the Authors or reported by local authorities in the watershed [see also 75,76]. Hence, considering that diffuse sheet erosion is often difficult to detect in the field or from aerial images, especially under tree canopy, areas affected by this process could not have been adequately covered in the maps used for Experiments 2 and 3 [62] (Figs 4b, c). This might explain why this process is better represented in CAs of Experiment 1 (Fig 4a), particularly in forested areas of the watershed (Fig 7i).

Conversely, relative CEC is better predicted in hazard-prone hotspots affected by land degradation, characterized by relatively high geomorphic potential and sediment connectivity [62] (Experiment 3; Fig 4c; Table 7). Hence, our results suggest that the B/D index could be a valuable proxy for modelling soil degradation by geomorphic activity in most vulnerable areas of the watershed. Such degraded areas are probably characterized by a reduced CEC, due to the degradation of topsoil rich in SOM and organo-mineral fine fractions caused by erosion and associated processes [129,130]. Other studies previously underscored a link between reduced CEC levels, low SOM content, and increasing erosion rates in Mediterranean environments, highlighting the critical role of organic matter for soil structural stability and erosion prevention [131,132]. From a certain point of view, this evidence imply that diffuse erosion processes might have affected areas already impacted by other geomorphic processes, as more susceptible to further degradation. However, this evidence appears to contrast with the results obtained from the other three indices (Table 7), as it suggests that hazard-prone hotspots played a more important role in sediment production and export during the study period [62]. As a possible explanation, we hypothesize that the organic matter analysed in the individual samples, whose signal is reflected in the calculation of the various indices (Tables 1 and 5), may originate from diverse areas within the respective CA. Soluble Dissolved Organic Matter (DOM) might have a different origin influenced by various factors and may have followed distinct pathways compared to organic matter in particulate form and/or associated with sediments, while the latter could have been transported differentially based on the characteristics of soil particles [133]. Moreover, Berhe [134] observed that the decomposition rates of organic matter differ considerably along dynamic toposequences and across soil depth gradients, where the effect of transfer of topsoil and associated SOM from eroding to depositional landforms, as well as of its burial by subsequent erosion events, depends on local environmental conditions. This could have implications for the role of CEC proxy as potential indicator of land degradation as, generally, the greater the degree of decomposition of the SOM, the higher the CEC of the SOM [135]. Nadeu et al. [136] already demonstrated how different land use and geomorphic settings can affect sediment sourcing and SOC dynamics in Mediterranean watersheds. Nevertheless, this study was not aimed at conducting a detailed molecular chemical analysis of organic matter to determine its origin within the watershed, nor at performing a pedological analysis to investigate soil properties in relation to degradation dynamics. Thus, the approach employed in this study is unable to offer more detailed insights into the observed relationships. However, these preliminary findings serve as a crucial starting point for further in-depth studies.

5.2 Geospatial relationships between target and feature variables

Oak, hornbeam, and chestnut forests was a recurrent critical variable for all the indices (Fig 8). This relationship might partially explain why all the investigated FTIR indices display an increasing trend in their values from upper to lower parts of the watershed (Fig 5). That is, in the Northern Apennines this forest type is typically distributed at lower-middle elevations (Fig 7i; i.e., hilly vegetation belt), whereas it is replaced by beech forests in upper areas (Fig 7i; i.e., montane vegetation belt) [137]. However, the effect of oak, hornbeam, and chestnut forests seems to be more pronounced for the relative hydrophobicity of SOM and organic matter-cation associations, as the proportion of this forest type increases, values of the related indices also significantly rise (Figs 9, 10, and 12).

On the one hand, this may imply that the organic matter inputs from this forest type contribute to increasing hydrophobic compounds in SOM relative to hydrophilic ones, by providing hydrocarbons, triglycerides, lipids, or waxes [22,138]. Forests generally provide important ecosystem services protecting soil from surficial water erosion [e.g., stabilization of soil, increasing organic matter content, and protection from rain drops erosive force; [139]. However, under the same conditions, hydrophobicity may stabilize or destabilize soil aggregates depending on soil texture and SOM content [19,23], thus, leading to a differential susceptibility to erosion. In particular, our results indicate that high clay content (i.e., Clay% > 30–32%) significantly improves the accumulation of hydrophobic SOM relative to polysaccharides (i.e., A/D index; Fig 8b), whereas the influence of this variable is negligible at lower concentrations (Fig 10). This may imply that once a critical amount of clay is present in the soil, it begins to have a more pronounced effect on the retention of hydrophobic compounds into organo-mineral complexes, although they produce low energy bonds that may be reversible [140,141]. Anyway, in many Mediterranean watersheds such as the upper Val d’Arda, clay-rich soils can experience intensified shrinking-swelling phenomena which enhance soil cracks under dry conditions [47,142,143]. Bosino et al. [144] highlighted that after long dry periods, these cracks may allow huge water infiltration into the upper topsoil horizon, leading to a subsequent closure of the fissures in the surface crust. During moist conditions the clay rich substrate enhance rapid runoff processes due to reduced infiltration rates, especially in shallow soil profiles with minimum storage capacities [144]. However, structural weaknesses in high-hydrophobic soils can create preferential flow paths [145], leading to uneven wetting, aggregate breakdown, and increased erosion rates under heavy rainstorms. Nevertheless, other studies pointed out that high organic matter content can stabilize clays reducing swelling and crack formation [146]. Therefore, our findings might have relevant implications for Mediterranean watersheds that are similar to the upper Val d’Arda, where clay content in soils ranges from 7 to 47% (Fig 7c), with a mean value of ~ 30% (cfr. S2 Appendix).

On the other hand, in the case of C/D index (Figs 8d and 12), the degree of decomposition or humification of the litter produced by oak, hornbeam, and chestnut trees, as well as the relative carbonate content in the samples, might provide specific compounds responsible for improving associations between organic matter and cations [39,135] (Table 1), thus promoting soil aggregate stability [19].

SOM hydrophobic characteristics appear to be largely affected also by Vertical Distance to Channel Network (VDCN) (Fig 8a, b). Following Berhe & Kleber [133] and Maerker et al. [102], the redistribution of SOM throughout the watershed driven by water fluxes, as affected by the topographic setting and landform position in a toposequence, could lead to a differential stabilization of the organic matter, depending on whether it is transferred through eroding areas or accumulated in depositional environments [147,148]. In this case, the relative amount of hydrophobic vs. hydrophilic components measured in our fluvial sediment samples might reflect the organic matter connectivity between the respective contributing areas and the fluvial system [149]. That is, hydrophilic compounds, being more soluble in water, are more easily transferred as DOM in areas closer to main channels (in vertical distance), where runoff volume is typically higher and flooding is more frequent and intense. Therefore, the retention of DOM within riparian vegetation or floodplain wetlands [149] might favour the accumulation of hydrophilic components such as C = O and C-O-C groups, as suggested by the negative effect of VDCN on the A/B and A/D indices below 200 m (Figs 9 and 10). This allows hydrophobic compounds to be retained in poorly connected areas, farther away from the valley bottom (in vertical distance) (Figs 9 and 10), thereby rendering soils more susceptible to water erosion by affecting water infiltration capacity and surface runoff.

Similarly, the spatial patterns of CEC and organic matter-cation associations are highly affected by topography, as reflected by the Channel Network Base Level (CNBL) (Fig 8c, d). Indeed, Figs 11 and 12 show that CNBL has a particular effect on these indices, where their values decrease as the erosional base level increases (see also Figs 5 and 6 m). Therefore, our results suggest that areas with a lower CNBL (closer to the reservoir) tend to accumulate more carboxylic acids, as well as more carbonates relative to silicates [84,87]. Carboxylic acids are likely more stable and exhibit stronger interactions with cations, as C = O functional groups are characteristic of more decomposed and reactive organic matter [e.g., humified materials and lignin degradation products; 50]. This suggests that as organic matter moves through the watershed, it alternates between phases of storage and remobilization depending on hydrological conditions, undergoing successive stages of decomposition that enhance its chemical reactivity [87], particularly when buried [134]. Therefore, long retention times under certain hydrological conditions might promote the formation and accumulation of carboxyl groups derived from decomposition processes, thus promoting the formation of stable organo-mineral complexes. This implication is supported also by the influence of the Downslope Distance Gradient (DDC) (Fig 8c, d), as areas with higher DDC values, such as in the lower watershed (see Fig 6i), appear to have a positive effect on the indices (Figs 11 and 12). Anyway, as with many other watersheds in the Northern Apennines, in the study area the past and current riverbed evolution is a function of fluvial morphodynamics, as well as lithological and structural characteristics of the landscape [61]. La Licata et al. [47] emphasised that in the lower part of the upper Val d’Arda, where gentle open slopes composed of pelitic and chaotic rock formations prevail and the valley bottom is wider, slope and fluvial sediments can be easily stored in depositional storages. According to Doetterl et al. [150], these findings underscore the importance of adopting a landscape-based modelling approach for assessing the complexity of SOM dynamics and the relations with their shaping factors, informing useful predictions of organic matter cycling and soil carbon sequestration in degraded watersheds.

In this regard, particular attention should be paid to B/D index (Fig 8c). Even though the influence of Elevation is more or less relevant in all experiments (Fig 4; cfr. Feature Importance plots in S1 Fig), the peculiar effect revealed by Experiment 3 is likely related to the distribution of geomorphologically highly active areas (e.g., landslide deposits connected to the drainage system and affected by additional processes like surficial soil erosion and/or stream incision, and steep rock walls affected by rockfalls and localized debris flows) with respect to the elevation gradient (Fig 11; Fig 4c) [see 62]. In other words, highly active sediment sources and eroding steep slopes, mostly distributed at medium-high elevations, likely retain less SOM rich in C = O groups because the export of organic matter is higher and the accumulation of decomposed compounds is limited [134]. Nonetheless, the Experiment 3 could have biased the effect of SOC% on the B/D index, as in the other experiments SOC content seems to not to play an important role in shaping CEC. Indeed, generally high SOC content improves CEC [16], with a consequent effect on soil aggregate stability [19]. However, Fig 11 shows a significative positive effect of SOC content only for low percentage (around 1.75%), while the effect of increasing SOC content appears to be negligible. This evidence may indicate that in geomorphologically active degraded areas, a minimum threshold of SOC is needed to contribute to CEC. However, the continued disturbance and removal of organic matter in these areas may be overriding the beneficial effects of increasing SOC content (Fig 11).

Human-altered landscapes (i.e., Anthropic areas; Fig 7i; Table 4), including urban network, mining areas, construction sites, excavations, manufacturing facilities, infrastructures, farming and livestock settlements, and agricultural areas (other than cultivated fields), seem to promote a greater degree of organic matter-cation interactions (i.e., C/D index; Fig 8d). This could be due to soil amendments (e.g., fertilizers) or certain land management practices that enhance nutrient retention and cation exchange processes. Moreover, carbonate-rich soils (i.e., Carbonate turbidites; Fig 7h) appear to have a positive effect on C/D index (Fig 12), likely due to the presence of calcium and other cations, which improve soil stability. This finding is explained by the widespread distribution of carbonate-rich lithologies in the contributing areas of the analysed samples in Experiment 1 (Fig 4a). Particularly, it is worth noting that samples with absorption spectra showing a high-intensity band C (i.e., higher or similar compared to band D; Fig 3) have a mean proportion of Carbonate Turbidites of 0.55 (C/D > 1) and 0.41 (C/D close to 1), which is quite higher than other lithologies (Table 3; see available Datasets). Thus, these results underscore the relation between C/D index and the relative carbonate content in the samples (Table 1) [53,55].

5.3 Limitations of the methodological approach and improvements

Despite these promising findings, certain limitations need to be acknowledged.

In general, the significance of these findings relies on the assumption that the FTIR indices used in this study can serve as proxies for soil hydrophobicity, CEC, and organic matter–cation interactions (Table 1) within this environmental context, as supported by evidence drawn from the literature [39,50–55,84]. The obtained results represent an initial preliminary attempt towards a better understanding of such relationships. However, these results should be interpreted with caution, e.g., as we discussed for their relevance in terms of erosion processes. That is, an empirical evaluation linking FTIR indices to actual erosion measurements—necessary to establish these indices as direct indicators—is currently lacking. Nonetheless, additional studies are needed to explore more detailed relationships between these potential FTIR-based indicators and the chemical and pedological characteristics of sediment source areas.

The relatively higher errors associated with C/D, which are likely caused by interference problems (e.g., signals from SOM and soil mineral components, or effects of land use vs. soil genesis), suggest that the model’s capability to predict SOM chemical characteristics with high variability (cfr. STD = 0.47) should be interpreted cautiously (Table 7). This underscores the need for further refinements in modelling approaches to better capture the complexity SOM-environment interactions, especially in areas with diverse vegetation and soil types. Considering the potential overlapping in absorbance bands between organic and mineral fractions, especially in the case of band C (Carbonate/ν_sCOO^-) and band D (C-O-C/Si-O-Si) (Fig 3), further improvements should disentangle the effect of organic vs. mineral compounds in the spatial distribution of the indices. On the other hand, if the organic component is not separated from the mineral fraction, accounting for the particle size effect and effective organic matter content on organic matter–sediment interactions and the detectability of SOM characteristics could reveal additional insights into the variability of the measured band intensities [49,151]. This might be of particular interest whether the sediment’s particle size distribution and/or organic constituents allow for hypotheses regarding the origin of the sediment within the watershed or enable inferences about the conditions and the selectivity of the mechanism responsible for its erosion and transport [78,79,152,153]. Anyway, recent advancements in FTIR analysis [e.g., 154,155] suggest that characterizing silicate and carbonate components could provide additional insights into the role of these compounds on aggregate stability and hydro-mechanical properties of soils [156]. Additionally, the reliance on FTIR indices alone may limit the range of SOM characterization, as these indices provide only a partial view on the broader ranging organic matter dynamics in the watershed (cfr. Table 7).

Furthermore, the RF model used has potential limitations, including the lack of tuning for other hyperparameters such as the number of trees (ntree) or the minimum node size, which could significantly impact model performance. Since the rf function in the ‘caret’ package only allows tuning of the ‘mtry’ parameter [116,117], the optimization process may not fully exploit the potential of the RF algorithm. Also, RF may struggle with extrapolation or interpretability in certain cases, and comparing its performance with other machine learning models like gradient boosting machines or support vector machines could provide insights into whether a different model better captures the relationships in the data. Despite the reliability afforded by the LOOCV and ensemble structure of RF model, residual uncertainties may persist due to potential imbalances in the datasets, the underrepresentation of specific geomorphic features where geomorphological mapping is limited by a dense forest cover, or variability in SOM composition that is not captured in the sampling. These factors may limit the model’s transferability to unsampled areas having diverse environmental conditions.

Future research should focus on expanding the spatial and temporal scales of investigation to include different watersheds with varying climatic and geomorphological conditions. This would allow for a more comprehensive understanding of how SOM composition interacts with environmental factors across different contexts. Moreover, integrating additional variables such climatic data and land management practices, as well as incorporating dynamic modelling approaches, could provide further insights into the relationships between soil erosion susceptibility and SOM dynamics in Mediterranean environments.

6 Conclusion

With this interdisciplinary work, we provide significant new evidence on the relationships between SOM-related properties, serving as potential indicators of soil degradation and erosion, and environmental, geomorphic, and hydrological characteristics in a temperate agricultural-forested area of the Northern Apennines chosen as representative of Mediterranean environments. Particularly, our approach explored SOM-related indices as proxies for relative hydrophobicity, CEC, and the degree of organic matter-cation interactions, identifying the most influential environmental predictors driving their spatial variability at watershed scale and evaluating their effect on the observed values. Moreover, the significance of our approach exceeded the initial objectives, providing valuable new insights into the complex landscape-based relationships between environmental factors, geomorphology, connectivity, and SOM dynamics. The key findings of this study can be summarized as follows:

Relative hydrophobicity of SOM and organic matter-cation interactions are better predicted in areas influenced by broader hydrological and low-magnitude geomorphic processes during the study period, as reflected by the entire contributing areas of the sampling points (Experiment 1), likely reflecting diffuse erosion processes like sheet erosion. This highlights the importance of considering low-intensity geomorphic regimes, beyond more intensive processes like gullying and landsliding, for assessing and managing sediment supply to river systems during hydrometeorological events, particularly in forested areas;
CEC was identified as a proxy for detecting and modelling soil degradation caused by geomorphic activity in the most vulnerable areas of the watershed, as reflected by the hazard-prone contributing areas used in Experiment 3. This evidence partially contrasts with the other indices, suggesting that diffuse erosion may preferentially affect already degraded areas. This underscores the complexity of SOM dynamics in environments with similar characteristics and supports new research directions exploring the differential origins and transport pathways of SOM based on its chemical properties and associations with mineral particles on hillslope and sub-watershed scales;
Areas dominated by oak, hornbeam, and chestnut forests, on carbonate-rich bedrocks, and located farther from the valley bottom, tend to have more hydrophobic soils, making them potentially more prone to water erosion. Additionally, high clay content in soil appears to favour the retention of hydrophobic compounds, potentially intensifying shrinking-swelling phenomena, soil cracking, and localized water erosion. These findings carry significant implications for Mediterranean watersheds having similar vegetation, as well as for regions with seasonally contrasting climates and lithologies dominated by weak, stratified, clay-rich rocks;
Topographic features such as elevation and channel network base level were found to significantly influence the spatial patterns of CEC and organic matter-cation interactions. This suggests that topography-driven erosion potential and landscape connectivity play critical roles in the redistribution and retention of organic matter across landscape compartments. These insights suggest that such landscape-based modelling could be a valuable approach for predicting organic matter cycling and carbon sequestration in complex, dynamic Mediterranean watersheds with similar landscape configuration;
Organic matter-cation associations appear to be strongly influenced by relative carbonate content, likely due to higher concentrations of calcium and other cations, as well as specific land management practices. This has important implications for soil erosion control strategies, as stronger SOM-cation interactions enhance soil aggregate stability, reducing susceptibility to erosion.

While providing significant findings that support new research hypotheses and perspectives, at this stage, our interdisciplinary approach is able to explain only a moderate portion of the variability in the SOM indices, indicating the need for further improvements to enhance model performance. Future studies could focus on implementing broader spatial and temporal analyses, integrating climatic and land management variables. Additionally, more refined geostatistical methods could be applied to spatially model and regionalize soil erosion susceptibilities by integrating the most influential predictive features.

Supporting information

S1 Appendix. Additional information regarding laboratory procedures and methodology.

(PDF)

pone.0330252.s001.pdf^{(85.6KB, pdf)}

S2 Appendix. Basic statistics for the raster variables selected as feature variables.

(PDF)

pone.0330252.s002.pdf^{(55.7KB, pdf)}

S1 Fig. Permutation-based variable importance for all the experiments.

(PDF)

pone.0330252.s003.pdf^{(426.2KB, pdf)}

S2 Fig. Accumulated Local Effect (ALE) plots for all the experiments.

(PDF)

pone.0330252.s004.pdf^{(1.3MB, pdf)}

Acknowledgments

We thank the staff of Consorzio di Bonifica di Piacenza for the valuable support during the field surveys. We were also supported by the Belmont ABRESO project.

Abbreviations

ALE: Accumulated Local Effects
CA: Contributing Area
CARG: Italian Official Geological Cartography
CCC: Concordance Correlation Coefficient
CEC: Cation Exchange Capacity
CNBL: Channel Network Base Level
DDC: Downslope Distance Gradient
DOM: Dissolved Organic Matter
DTM: Digital Terrain Model
FTIR: Fourier Transform Infrared
GIS: Geographic Information System
HOTSED: Hotspots of sediment sources and related dynamics
HPD: Hazard Map of sediment Production and Delivery
IM: Inventory Map
LOOCV: Leave-One-Out Cross-Validation
LULC: Land Use/Land Cover
MSE: Mean Squared Error
NSE: Nash-Sutcliffe Efficiency
RER: Emilia-Romagna Region
RF: Random Forest
RMSE: Root Mean Squared Error
SOC: Soil Organic Carbon
SOM: Soil Organic Matter
STD: Standard Deviation
VDCN: Vertical Distance to Channel Network
WN: Wavenumber.

Data Availability

The datasets related to the three experiments, along with the related raster and shapefile data, contributing areas, and a shapefile containing the location of the sampling points, are available at the following link: https://doi.org/10.5281/zenodo.15097417. The code of the Random Forest model implemented in this study is available at the following link: https://github.com/Odunayo3/SoilDegradationModeling.git. Additional information and outputs are available along with the Supporting Information files.

Funding Statement

This research was conducted with the financial support of the Earth and Environmental Sciences PhD-PON program (Research & Innovation, 2014e2020, Education and research for recovery - REACT-EU, DOT1322534-4) of University of Pavia, Department of Earth and Environmental Sciences. This research was also funded by the PRIN 2022 project by the Italian Ministry of University and Research, entitled: ”Full cOveRage, Multi-scAle and multi-sensor geomorphological map: a practical tool for TerrItOrial plaNning – FORMATION” (2022C2XPK7_004). This research was also supported by the Polish National Agency for Academic Exchange - Urgency Grants program project contract no. BPN/GIN/2024/1/00008. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

References

1.FAO, ITPS. Status of the World’s Soil Resources (SWSR) – Main Report. Rome: Food and Agriculture Organization of the United Nations, Intergovernmental Technical Panel on Soils. 2015. [Google Scholar]
2.Giardino JR, Houser C. Introduction to the Critical Zone. Developments in Earth Surface Processes. Elsevier. 2015. 1–13. doi: 10.1016/b978-0-444-63369-9.00001-x [DOI] [Google Scholar]
3.Banwart SA, Bernasconi SM, Blum WEH, de Souza DM, Chabaux F, Duffy C, et al. Soil Functions in Earth’s Critical Zone. Advances in Agronomy. Elsevier. 2017. 1–27. doi: 10.1016/bs.agron.2016.11.001 [DOI] [Google Scholar]
4.Lal R. Restoring Soil Quality to Mitigate Soil Degradation. Sustainability. 2015;7(5):5875–95. doi: 10.3390/su7055875 [DOI] [Google Scholar]
5.Montgomery DR. Soil erosion and agricultural sustainability. Proc Natl Acad Sci U S A. 2007;104(33):13268–72. doi: 10.1073/pnas.0611508104 [DOI] [PMC free article] [PubMed] [Google Scholar]
6.Borrelli P, Robinson DA, Fleischer LR, Lugato E, Ballabio C, Alewell C, et al. An assessment of the global impact of 21st century land use change on soil erosion. Nat Commun. 2017;8(1):2013. doi: 10.1038/s41467-017-02142-7 [DOI] [PMC free article] [PubMed] [Google Scholar]
7.FAO. Outcome document of the Global Symposium on Soil Erosion. Rome: Food and Agriculture Organization of the United Nations. 2019. [Google Scholar]
8.Thornes JB. The evolving context of Mediterranean desertification. In: Geeson NA, Brandt CJ, Thornes JB. Mediterranean desertification: A mosaic of processes and responses. Chichester: John Wiley. 2002. 5–11. [Google Scholar]
9.García-Ruiz JM, Nadal-Romero E, Lana-Renault N, Beguería S. Erosion in Mediterranean landscapes: Changes and future challenges. Geomorphology. 2013;198:20–36. doi: 10.1016/j.geomorph.2013.05.023 [DOI] [Google Scholar]
10.Borrelli P, Panagos P, Alewell C, Ballabio C, de Oliveira Fagundes H, Haregeweyn N, et al. Policy implications of multiple concurrent soil erosion processes in European farmland. Nat Sustain. 2022;6(1):103–12. doi: 10.1038/s41893-022-00988-4 [DOI] [Google Scholar]
11.Lal R. Degradation and resilience of soils. Phil Trans R Soc Lond B. 1997;352(1356):997–1010. doi: 10.1098/rstb.1997.0078 [DOI] [Google Scholar]
12.Viñals E, Maneja R, Rufí-Salís M, Martí M, Puy N. Reviewing social-ecological resilience for agroforestry systems under climate change conditions. Sci Total Environ. 2023;869:161763. doi: 10.1016/j.scitotenv.2023.161763 [DOI] [PubMed] [Google Scholar]
13.Karlen DL, Andrews SS, Doran JW. Soil quality: Current concepts and applications. Advances in Agronomy. Elsevier. 2001. 1–40. doi: 10.1016/s0065-2113(01)74029-1 [DOI] [Google Scholar]
14.Reeves DW. The role of soil organic matter in maintaining soil quality in continuous cropping systems. Soil and Tillage Research. 1997;43(1–2):131–67. doi: 10.1016/s0167-1987(97)00038-x [DOI] [Google Scholar]
15.Wander M. Soil organic matter fractions and their relevance to soil function. In: Magdoff F, Weil RR. Soil organic matter in sustainable agriculture. Boca Raton: CRC Press. 2024. 67–102. [Google Scholar]
16.Murphy BW. Impact of soil organic matter on soil properties—a review with emphasis on Australian soils. Soil Res. 2015;53(6):605. doi: 10.1071/sr14246 [DOI] [Google Scholar]
17.Olsson L, Barbosa H, Bhadwal S, Cowie A, Delusca K, Flores-Renteria D, et al. Land degradation. In: Shukla PR, Skea J, Calvo Buendia E, Masson-Delmotte V, Pörtner H-O, Roberts DC, et al. Climate change and land: An IPCC special report on climate change, desertification, land degradation, sustainable land management, food security, and greenhouse gas fluxes in terrestrial ecosystems. Cambridge: Cambridge University Press; 2019. doi: 10.1017/9781009157988.006 [DOI] [Google Scholar]
18.Conforti M, Buttafuoco G, Leone AP, Aucelli PPC, Robustelli G, Scarciglia F. Studying the relationship between water-induced soil erosion and soil organic matter using Vis–NIR spectroscopy and geomorphological analysis: A case study in southern Italy. CATENA. 2013;110:44–58. doi: 10.1016/j.catena.2013.06.013 [DOI] [Google Scholar]
19.Obalum SE, Chibuike GU, Peth S, Ouyang Y. Soil organic matter as sole indicator of soil degradation. Environ Monit Assess. 2017;189(4):176. doi: 10.1007/s10661-017-5881-y [DOI] [PubMed] [Google Scholar]
20.Panagos P, Meusburger K, Ballabio C, Borrelli P, Alewell C. Soil erodibility in Europe: a high-resolution dataset based on LUCAS. Sci Total Environ. 2014;479–480:189–200. doi: 10.1016/j.scitotenv.2014.02.010 [DOI] [PubMed] [Google Scholar]
21.Bettoni M, Maerker M, Sacchi R, Bosino A, Conedera M, Simoncelli L, et al. What makes soil landscape robust? Landscape sensitivity towards land use changes in a Swiss southern Alpine valley. Sci Total Environ. 2023;858(Pt 2):159779. doi: 10.1016/j.scitotenv.2022.159779 [DOI] [PubMed] [Google Scholar]
22.Capriel P. Hydrophobicity of organic matter in arable soils: influence of management. European J Soil Science. 1997;48(3):457–62. doi: 10.1111/j.1365-2389.1997.tb00211.x [DOI] [Google Scholar]
23.Jordán A, Zavala LM, Mataix-Solera J, Doerr SH. Soil water repellency: Origin, assessment and geomorphological consequences. CATENA. 2013;108:1–5. doi: 10.1016/j.catena.2013.05.005 [DOI] [Google Scholar]
24.Dekker LW, Ritsema CJ. Wetting patterns and moisture variability in water repellent Dutch soils. Journal of Hydrology. 2000;231–232:148–64. doi: 10.1016/s0022-1694(00)00191-8 [DOI] [Google Scholar]
25.Doerr SH, Shakesby RA, Walsh RPD. Soil water repellency: its causes, characteristics and hydro-geomorphological significance. Earth-Science Reviews. 2000;51(1–4):33–65. doi: 10.1016/s0012-8252(00)00011-8 [DOI] [Google Scholar]
26.Shakesby RA, Doerr SH, Walsh RPD. The erosional impact of soil hydrophobicity: current problems and future research directions. Journal of Hydrology. 2000;231–232:178–91. doi: 10.1016/s0022-1694(00)00193-1 [DOI] [Google Scholar]
27.Fox DM, Darboux F, Carrega P. Effects of fire-induced water repellency on soil aggregate stability, splash erosion, and saturated hydraulic conductivity for different size fractions. Hydrol Process. 2007;21:2377–84. doi: 10.1002/hyp.675 [DOI] [Google Scholar]
28.Bettoni M, Maerker M, Bosino A, Conedera M, Simoncelli L, Vogel S. Land use effects on surface runoff and soil erosion in a southern Alpine valley. Geoderma. 2023;435:116505. doi: 10.1016/j.geoderma.2023.116505 [DOI] [PubMed] [Google Scholar]
29.Cerdà A, Lavee H, Romero‐Díaz A, Hooke J, Montanarella L. Preface. Land Degrad Dev. 2010;21(2):71–4. doi: 10.1002/ldr.968 [DOI] [Google Scholar]
30.Piccolo A, Mbagwu JSC. Role of Hydrophobic Components of Soil Organic Matter in Soil Aggregate Stability. Soil Science Soc of Amer J. 1999;63(6):1801–10. doi: 10.2136/sssaj1999.6361801x [DOI] [Google Scholar]
31.Mataix-Solera J, Doerr SH. Hydrophobicity and aggregate stability in calcareous topsoils from fire-affected pine forests in southeastern Spain. Geoderma. 2004;118(1–2):77–88. doi: 10.1016/s0016-7061(03)00185-x [DOI] [Google Scholar]
32.Barthès B, Roose E. Aggregate stability as an indicator of soil susceptibility to runoff and erosion; validation at several levels. CATENA. 2002;47(2):133–49. doi: 10.1016/s0341-8162(01)00180-1 [DOI] [Google Scholar]
33.Mirbabaei SM, Shahrestani MS, Zolfaghari A, Abkenar KT. Relationship between soil water repellency and some of soil properties in northern Iran. CATENA. 2013;108:26–34. doi: 10.1016/j.catena.2013.02.013 [DOI] [Google Scholar]
34.Nciizah AD, Wakindiki IIC. Physical indicators of soil erosion, aggregate stability and erodibility. Archives of Agronomy and Soil Science. 2015;61(6):827–42. doi: 10.1080/03650340.2014.956660 [DOI] [Google Scholar]
35.FAO. Soil erosion: the greatest challenge to sustainable soil management. Rome: Food and Agriculture Organization of the United Nations; 2019. [Google Scholar]
36.Celi L, Schnitzer M, Nègre M. Analysis of carboxyl groups in soil humic acids by a wet chemical method, fourier-transform infrared spectrophotometry, and solution-state carbon-13 nuclear magnetic resonance. a comparative study. Soil Science. 1997;162(3):189–97. doi: 10.1097/00010694-199703000-00004 [DOI] [Google Scholar]
37.Tipping E. Cation Binding by Humic Substances. Cambridge University Press. 2002. doi: 10.1017/cbo9780511535598 [DOI] [Google Scholar]
38.McKenzie N, Jacquier D, Isbell R, Brown K. Australian soils and landscapes—An illustrated compendium. Melbourne: CSIRO Publishing. 2004. [Google Scholar]
39.Kaiser M, Ellerbrock RH, Gerke HH. Cation Exchange Capacity and Composition of Soluble Soil Organic Matter Fractions. Soil Science Soc of Amer J. 2008;72(5):1278–85. doi: 10.2136/sssaj2007.0340 [DOI] [Google Scholar]
40.Solly EF, Weber V, Zimmermann S, Walthert L, Hagedorn F, Schmidt MWI. A Critical Evaluation of the Relationship Between the Effective Cation Exchange Capacity and Soil Organic Carbon Content in Swiss Forest Soils. Front For Glob Change. 2020;3:98. doi: 10.3389/ffgc.2020.00098 [DOI] [Google Scholar]
41.Zheng H, Miao C, Huntingford C, Tarolli P, Li D, Panagos P, et al. The Impacts of Erosion on the Carbon Cycle. Reviews of Geophysics. 2025;63(1):e2023RG000829. doi: 10.1029/2023rg000829 [DOI] [Google Scholar]
42.Shang P, Lu Y, Du Y, Jaffé R, Findlay RH, Wynn A. Climatic and watershed controls of dissolved organic matter variation in streams across a gradient of agricultural land use. Sci Total Environ. 2018;612:1442–53. doi: 10.1016/j.scitotenv.2017.08.322 [DOI] [PubMed] [Google Scholar]
43.Tang J, Wang W, Yang L, Cao C, Li X. Variation in quantity and chemical composition of soil dissolved organic matter in a peri-urban critical zone observatory watershed in Eastern China. Sci Total Environ. 2019;688:622–31. doi: 10.1016/j.scitotenv.2019.06.270 [DOI] [PubMed] [Google Scholar]
44.Zhao C, Wang C-C, Li J-Q, Wang C-Y, Wang P, Pei Z-J. Dissolved organic matter in urban stormwater runoff at three typical regions in Beijing: chemical composition, structural characterization and source identification. RSC Adv. 2015;5(90):73490–500. doi: 10.1039/c5ra14993b [DOI] [Google Scholar]
45.Zhou Y, Liu M, Zhou L, Jang K-S, Xu H, Shi K, et al. Rainstorm events shift the molecular composition and export of dissolved organic matter in a large drinking water reservoir in China: High frequency buoys and field observations. Water Res. 2020;187:116471. doi: 10.1016/j.watres.2020.116471 [DOI] [PubMed] [Google Scholar]
46.Ferreira CSS, Seifollahi-Aghmiuni S, Destouni G, Ghajarnia N, Kalantari Z. Soil degradation in the European Mediterranean region: Processes, status and consequences. Sci Total Environ. 2022;805:150106. doi: 10.1016/j.scitotenv.2021.150106 [DOI] [PubMed] [Google Scholar]
47.La Licata M, Bosino A, Bettoni M, Maerker M. Assessing landscape features and geomorphic processes influencing sediment dynamics in a geomorphologically highly active Mediterranean agroecosystem: The upper Val d’Arda case study (Northern Apennines, Italy). Geomorphology. 2023;433:108724. doi: 10.1016/j.geomorph.2023.108724 [DOI] [Google Scholar]
48.Johnston CT, Aochi YO. Fourier transform infrared and Raman spectroscopy. In: Sparks DL, Page AL, Helmke PA, Loeppert RH. Methods of soil analysis, Part 3: Chemical methods. Madison: Soil Science Society of America, American Society of Agronomy. 1996. 269–321. [Google Scholar]
49.Margenot AJ, Parikh SJ, Calderón FJ. Fourier‐transform infrared spectroscopy for soil organic matter analysis. Soil Science Soc of Amer J. 2023;87(6):1503–28. doi: 10.1002/saj2.20583 [DOI] [Google Scholar]
50.Ellerbrock RH, Gerke HH. Explaining soil organic matter composition based on associations between OM and polyvalent cations. J Plant Nutr Soil Sci. 2018;181(5):721–36. doi: 10.1002/jpln.201800093 [DOI] [Google Scholar]
51.Ellerbrock RH, Gerke HH, Bachmann J, Goebel M-O. Composition of Organic Matter Fractions for Explaining Wettability of Three Forest Soils. Soil Science Society of America Journal. 2005;69(1):57. doi: 10.2136/sssaj2005.0057 [DOI] [Google Scholar]
52.Leue M, Gerke HH, Ellerbrock RH. Millimetre‐scale distribution of organic matter composition at intact biopore and crack surfaces. European J Soil Science. 2013;64(6):757–69. doi: 10.1111/ejss.12098 [DOI] [Google Scholar]
53.Reig FB, Adelantado JVG, Moya Moreno MCM. FTIR quantitative analysis of calcium carbonate (calcite) and silica (quartz) mixtures using the constant ratio method. Application to geological samples. Talanta. 2002;58(4):811–21. doi: 10.1016/s0039-9140(02)00372-7 [DOI] [PubMed] [Google Scholar]
54.Alvarez-Puebla RA, Garrido JJ, Valenzuela-Calahorro C, G. Goulet PaulJ. Retention and induced aggregation of Co(II) on a humic substance: sorption isotherms, infrared absorption, and molecular modeling. Surface Science. 2005;575(1–2):136–46. doi: 10.1016/j.susc.2004.11.027 [DOI] [Google Scholar]
55.Ellerbrock RH, Gerke HH. FTIR spectral band shifts explained by OM–cation interactions. J Plant Nutr Soil Sci. 2021;184(3):388–97. doi: 10.1002/jpln.202100056 [DOI] [Google Scholar]
56.Breiman L. Random Forests. Machine Learning. 2001;45(1):5–32. doi: 10.1023/a:1010933404324 [DOI] [Google Scholar]
57.Adeniyi OD, Brenning A, Bernini A, Brenna S, Maerker M. Digital Mapping of Soil Properties Using Ensemble Machine Learning Approaches in an Agricultural Lowland Area of Lombardy, Italy. Land. 2023;12(2):494. doi: 10.3390/land12020494 [DOI] [Google Scholar]
58.Simoni A, Ponza A, Picotti V, Berti M, Dinelli E. Earthflow sediment production and Holocene sediment record in a large Apennine catchment. Geomorphology. 2013;188:42–53. doi: 10.1016/j.geomorph.2012.12.006 [DOI] [Google Scholar]
59.Servizio Geologico d’Italia. Carta geologica d’Italia alla scala 1:50.000, foglio 198 - Bardi. Roma: Istituto Poligrafico e Zecca dello Stato. 1999. [Google Scholar]
60.Martini A, Zanzucchi G. Note illustrative della Carta Geologica d’Italia alla scala 1:50.000, Foglio 198 - Bardi. Roma: Servizio Geologico d’Italia. 2000. [Google Scholar]
61.Pellegrini L, Vercesi PL. Landscapes and Landforms Driven by Geological Structures in the Northwestern Apennines. World Geomorphological Landscapes. Springer International Publishing. 2017. 203–13. doi: 10.1007/978-3-319-26194-2_17 [DOI] [Google Scholar]
62.La Licata M, Bosino A, Sadeghi SH, De Amicis M, Mandarino A, Terret A, et al. HOTSED: A new integrated model for assessing potential hotspots of sediment sources and related sediment dynamics at watershed scale. International Soil and Water Conservation Research. 2025;13(1):80–101. doi: 10.1016/j.iswcr.2024.06.002 [DOI] [Google Scholar]
63.Patro ER, De Michele C, Granata G, Biagini C. Assessment of current reservoir sedimentation rate and storage capacity loss: An Italian overview. J Environ Manage. 2022;320:115826. doi: 10.1016/j.jenvman.2022.115826 [DOI] [PubMed] [Google Scholar]
64.Staffilani F, Bonapasta D, Marucci FE. Carta dell’erosione idrica attuale della Regione Emilia-Romagna. Servizio Geologico, Sismico e Suoli, Regione Emilia-Romagna. 2019. [Google Scholar]
65.Regione Emilia-Romagna. I suoli dell’Emilia-Romagna. Note illustrative. Servizio Cartografico – Ufficio Pedologico. 1994. [Google Scholar]
66.IUSS Working Group WRB. World Reference Base for Soil Resources 2014, update 2015: International soil classification system for naming soils and creating legends for soil maps. Rome: FAO. 2015. [Google Scholar]
67.Conrad O, Bechtel B, Bock M, Dietrich H, Fischer E, Gerlitz L, et al. System for Automated Geoscientific Analyses (SAGA) v. 2.1.4. Geosci Model Dev. 2015;8(7):1991–2007. doi: 10.5194/gmd-8-1991-2015 [DOI] [Google Scholar]
68.Regione Emilia-Romagna. Modello Digitale del Terreno 5x5, Edizione 2014. Archivio Cartografico. 2019. https://www.geoportale.regione.emilia-romagna.it/catalogo/dati-cartografici/altimetria/layer-2 [Google Scholar]
69.Lidberg W, Nilsson M, Lundmark T, Ågren AM. Evaluating preprocessing methods of digital elevation models for hydrological modelling. Hydrological Processes. 2017;31(26):4660–8. doi: 10.1002/hyp.11385 [DOI] [Google Scholar]
70.Wang L, Liu H. An efficient method for identifying and filling surface depressions in digital elevation models for hydrologic analysis and modelling. International Journal of Geographical Information Science. 2006;20(2):193–213. doi: 10.1080/13658810500433453 [DOI] [Google Scholar]
71.Freeman TG. Calculating catchment area with divergent flow based on a regular grid. Computers & Geosciences. 1991;17(3):413–22. doi: 10.1016/0098-3004(91)90048-i [DOI] [Google Scholar]
72.Quinn P, Beven K, Chevallier P, Planchon O. The prediction of hillslope flow paths for distributed hydrological modelling using digital terrain models. Hydrological Processes. 1991;5(1):59–79. doi: 10.1002/hyp.3360050106 [DOI] [Google Scholar]
73.Quantitative analysis of watershed geomorphology. Eos Trans AGU. 1957;38(6):913–20. doi: 10.1029/tr038i006p00913 [DOI] [Google Scholar]
74.Collins AL, Pulley S, Foster IDL, Gellis A, Porto P, Horowitz AJ. Sediment source fingerprinting as an aid to catchment management: A review of the current state of knowledge and a methodological decision-tree for end-users. J Environ Manage. 2017;194:86–108. doi: 10.1016/j.jenvman.2016.09.075 [DOI] [PubMed] [Google Scholar]
75.Valente M, Zanellati M, Facci G, Zanna N, Petrone E, Moretti E, et al. Health System Response to the 2023 Floods in Emilia-Romagna, Italy: A Field Report. Prehosp Disaster Med. 2023;38(6):813–7. doi: 10.1017/S1049023X23006404 [DOI] [PMC free article] [PubMed] [Google Scholar]
76.Arrighi C, Domeneghetti A. Brief communication: On the environmental impacts of the 2023 floods in Emilia-Romagna (Italy). Nat Hazards Earth Syst Sci. 2024;24(2):673–9. doi: 10.5194/nhess-24-673-2024 [DOI] [Google Scholar]
77.Collins AL, Blackwell M, Boeckx P, Chivers C-A, Emelko M, Evrard O, et al. Sediment source fingerprinting: benchmarking recent outputs, remaining challenges and emerging themes. J Soils Sediments. 2020;20(12):4160–93. doi: 10.1007/s11368-020-02755-4 [DOI] [PMC free article] [PubMed] [Google Scholar]
78.Koiter AJ, Owens PN, Petticrew EL, Lobb DA. The behavioural characteristics of sediment properties and their implications for sediment fingerprinting as an approach for identifying sediment sources in river basins. Earth-Science Reviews. 2013;125:24–42. doi: 10.1016/j.earscirev.2013.05.009 [DOI] [Google Scholar]
79.Owens PN, Blake WH, Gaspar L, Gateuille D, Koiter AJ, Lobb DA, et al. Fingerprinting and tracing the sources of soils and sediments: Earth and ocean science, geoarchaeological, forensic, and human health applications. Earth-Science Reviews. 2016;162:1–23. doi: 10.1016/j.earscirev.2016.08.012 [DOI] [Google Scholar]
80.Lizaga I, Gaspar L, Blake WH, Latorre B, Navas A. Fingerprinting changes of source apportionments from mixed land uses in stream sediments before and after an exceptional rainstorm event. Geomorphology. 2019;341:216–29. doi: 10.1016/j.geomorph.2019.05.015 [DOI] [Google Scholar]
81.Lizaga I, Quijano L, Gaspar L, Ramos MC, Navas A. Linking land use changes to variation in soil properties in a Mediterranean mountain agroecosystem. CATENA. 2019;172:516–27. doi: 10.1016/j.catena.2018.09.019 [DOI] [Google Scholar]
82.Ellerbrock RH, Höhn A, Rogasik J. Functional analysis of soil organic matter as affected by long‐term manurial treatment. European J Soil Science. 1999;50(1):65–71. doi: 10.1046/j.1365-2389.1999.00206.x [DOI] [Google Scholar]
83.Ellerbrock RH, Gerke HH. Characterization of Organic Matter Composition of Soil and Flow Path Surfaces Based on Physicochemical Principles—A Review. Advances in Agronomy. Elsevier. 2013. 117–77. doi: 10.1016/b978-0-12-407685-3.00003-7 [DOI] [Google Scholar]
84.Ellerbrock RH, Niessner D, Deumlich D, Puppe D, Gerke HH. Infrared-based analysis of organic matter composition in liquid and solid runoff fractions collected during a single erosion event. Soil and Tillage Research. 2024;235:105901. doi: 10.1016/j.still.2023.105901 [DOI] [Google Scholar]
85.Capriel P, Beck T, Borchert H, Gronholz J, Zachmann G. Hydrophobicity of the organic matter in arable soils. Soil Biology and Biochemistry. 1995;27(11):1453–8. doi: 10.1016/0038-0717(95)00068-p [DOI] [Google Scholar]
86.Ellerbrock RH, Gerke HH, Böhm C. In Situ DRIFT Characterization of Organic Matter Composition on Soil Structural Surfaces. Soil Science Soc of Amer J. 2009;73(2):531–40. doi: 10.2136/sssaj2008.0103 [DOI] [Google Scholar]
87.Ellerbrock RH, Gerke HH, Deumlich D. Soil organic matter composition along a slope in an erosion-affected arable landscape in North East Germany. Soil and Tillage Research. 2016;156:209–18. doi: 10.1016/j.still.2015.08.014 [DOI] [Google Scholar]
88.Berthomieu C, Hienerwadel R. Fourier transform infrared (FTIR) spectroscopy. Photosynth Res. 2009;101(2–3):157–70. doi: 10.1007/s11120-009-9439-x [DOI] [PubMed] [Google Scholar]
89.Dudev T, Lim C. The effect of metal binding on the characteristic infrared band intensities of ligands of biological interest. Journal of Molecular Structure. 2012;1009:83–8. doi: 10.1016/j.molstruc.2011.09.007 [DOI] [Google Scholar]
90.Tinti A, Tugnoli V, Bonora S, Francioso O. Recent applications of vibrational mid-Infrared (IR) spectroscopy for studying soil components: a review. j cent eur agric. 2015;16(1):1–22. doi: 10.5513/jcea01/16.1.1535 [DOI] [Google Scholar]
91.Jenks GF. The data model concept in statistical mapping. International Yearbook of Cartography. 1967;7:186–90. [Google Scholar]
92.Boehner J, Selige T. Spatial prediction of soil attributes using terrain analysis and climate regionalisation. In: Boehner J, McCloy KR, Strobl J. SAGA - Analysis and modelling applications. Goettingen: Department of Geography, University of Goettingen. 2006. 13–28. [Google Scholar]
93.Olaya V, Conrad O. Chapter 12 Geomorphometry in SAGA. Developments in Soil Science. Elsevier. 2009. 293–308. doi: 10.1016/s0166-2481(08)00012-3 [DOI] [Google Scholar]
94.Moore ID, Grayson RB, Ladson AR. Digital terrain modelling: A review of hydrological, geomorphological, and biological applications. Hydrological Processes. 1991;5(1):3–30. doi: 10.1002/hyp.3360050103 [DOI] [Google Scholar]
95.Wilson JP, Gallant JC. Primary topographic attributes. Terrain analysis: Principles and applications. New York: John Wiley & Sons. 2000. 51–85. [Google Scholar]
96.Gruber S, Peckham S. Chapter 7 Land-Surface Parameters and Objects in Hydrology. Developments in Soil Science. Elsevier. 2009. 171–94. doi: 10.1016/s0166-2481(08)00007-x [DOI] [Google Scholar]
97.Olaya V. Chapter 6 Basic Land-Surface Parameters. Developments in Soil Science. Elsevier. 2009. 141–69. doi: 10.1016/s0166-2481(08)00006-8 [DOI] [Google Scholar]
98.Blaga L. Aspects regarding the significance of the curvature types and values in the studies of geomorphometry assisted by GIS. Analele Universităţii din Oradea, Seria Geografie. 2012;22(2):327–37. [Google Scholar]
99.Wilson JP. Digital terrain modeling. Geomorphology. 2012;137(1):107–21. doi: 10.1016/j.geomorph.2011.03.012 [DOI] [Google Scholar]
100.Wilson JP, Bishop MP. 3.7 Geomorphometry. Treatise on Geomorphology. Elsevier. 2013. 162–86. doi: 10.1016/b978-0-12-374739-6.00049-x [DOI] [Google Scholar]
101.Florinsky IV. An illustrated introduction to general geomorphometry. Progress in Physical Geography: Earth and Environment. 2017;41(6):723–52. doi: 10.1177/0309133317733667 [DOI] [Google Scholar]
102.Maerker M, Schillaci C, Kropáček J. Morphometric terrain analysis to explore present day geohazards and paleolandscape forms and features in the surroundings of the Melka Kunture prehistoric site, Upper Awash Valley, Central Ethiopia. AUC GEOGRAPHICA. 2018;53(1):10–9. doi: 10.14712/23361980.2018.2 [DOI] [Google Scholar]
103.Zevenbergen LW, Thorne CR. Quantitative analysis of land surface topography. Earth Surf Processes Landf. 1987;12(1):47–56. doi: 10.1002/esp.3290120107 [DOI] [Google Scholar]
104.Hjerdt KN, McDonnell JJ, Seibert J, Rodhe A. A new topographic index to quantify downslope controls on local drainage. Water Resources Research. 2004;40(5):W05602. doi: 10.1029/2004wr003130 [DOI] [Google Scholar]
105.Riley SJ, De Gloria SD, Elliot R. A terrain ruggedness that quantifies topographic heterogeneity. Inter-mountain Journal of Science. 1999;5(1–4):23–7. [Google Scholar]
106.Ungaro F, Tarocco P, Aprea A. Carta del carbonio organico immagazzinato nei suoli della regione Emilia-Romagna, strato 0-30 cm, scala 1:50,000. Terza ed. Bologna: Regione Emilia-Romagna, Direzione Generale cura del territorio e dell’ambiente, Servizio Geologico, Sismico e dei Suoli. 2023. [Google Scholar]
107.Ungaro F, Tarocco P, Aprea A. Carta del contenuto percentuale di carbonio organico nei suoli della regione Emilia-Romagna, strato 0-30 cm, 1:50,000. Terza ed. Bologna: Regione Emilia-Romagna, Direzione Generale cura del territorio e dell’ambiente, Servizio Geologico, Sismico e dei Suoli. 2023. [Google Scholar]
108.Ungaro F, Tarocco P, Aprea A. Carta della tessitura dei suoli della Regione Emilia-Romagna, strato 0-30 cm, 1:50,000. Seconda edizione. Note illustrative. Bologna: Regione Emilia-Romagna, Direzione Generale cura del territorio e dell’ambiente, Servizio Geologico, Sismico e dei Suoli. 2023. [Google Scholar]
109.Ungaro F, Tarocco P, Aprea A. Carta del pH dei suoli della Regione Emilia-Romagna, strato 0-30 cm, 1:50,000. Seconda edizione. Note illustrative. Bologna: Regione Emilia-Romagna, Direzione Generale cura del territorio e dell’ambiente, Servizio Geologico, Sismico e dei Suoli. 2023. [Google Scholar]
110.SGSS - Servizio Geologico, Sismico e dei Suoli. Banca dati geologica, 1:10.000 – Unità geologiche – 10k. Geoportale Regione Emilia-Romagna. 2006. https://geoportale.regione.emilia-romagna.it/catalogo/dati-cartografici/informazioni-geoscientifiche/geologia/banca-dati-geologica-1-10.000/layer-11 [Google Scholar]
111.Regione E-R. Coperture vettoriali uso del suolo di dettaglio 2017 – Edizione 2020. Settore innovazione digitale, dati, tecnologia e polo archivistico. 2020. https://geoportale.regione.emilia-romagna.it/catalogo/dati-cartografici/pianificazione-e-catasto/uso-del-suolo/layer-9 [Google Scholar]
112.O’Callaghan JF, Mark DM. The extraction of drainage networks from digital elevation data. Computer Vision, Graphics, and Image Processing. 1984;28(3):323–44. doi: 10.1016/s0734-189x(84)80011-0 [DOI] [Google Scholar]
113.La Licata M, Bosino A, De Amicis M, Terret A, Maerker M. A GIS-based multiscale mapping framework to assess and visualize complex and polygenetic geomorphic systems – A novel ‘Geomorphic Entities’ approach. Journal of Maps. 2025;21(1):2437256. doi: 10.1080/17445647.2024.2437256 [DOI] [Google Scholar]
114.La Licata M, Bosino A, Sadeghi SH, Maerker M. Assessing hotspots of sediment sources and related sediment dynamics through the integration of geomorphological data, sediment connectivity and sediment transport modelling – The HOTSED model. EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024. 2024; EGU24-4101. doi: 10.5194/egusphere-egu24-4101 [DOI] [Google Scholar]
115.Romanski P, Kotthoff L, Schratz P. Package ‘FSelector’: Selecting Attributes. R package version 0.34. 2023. doi: 10.32614/CRAN.package.FSelector [DOI] [Google Scholar]
116.Kuhn M. Building Predictive Models inRUsing thecaretPackage. J Stat Soft. 2008;28(5). doi: 10.18637/jss.v028.i05 [DOI] [Google Scholar]
117.R Development Core Team. R: A language and environment for statistical computing. Vienna: R Foundation for Statistical Computing. 2016. [Google Scholar]
118.Arlot S, Celisse A. A survey of cross-validation procedures for model selection. Statist Surv. 2010;4:40–79. doi: 10.1214/09-ss054 [DOI] [Google Scholar]
119.Gupta HV, Kling H, Yilmaz KK, Martinez GF. Decomposition of the mean squared error and NSE performance criteria: Implications for improving hydrological modelling. Journal of Hydrology. 2009;377(1–2):80–91. doi: 10.1016/j.jhydrol.2009.08.003 [DOI] [Google Scholar]
120.Hastie T, Tibshirani R, Friedman J. The Elements of Statistical Learning. Springer New York. 2009. doi: 10.1007/978-0-387-84858-7 [DOI] [Google Scholar]
121.Singh AK, Kumar P, Ali R, Al-Ansari N, Vishwakarma DK, Kushwaha KS, et al. An Integrated Statistical-Machine Learning Approach for Runoff Prediction. Sustainability. 2022;14(13):8209. doi: 10.3390/su14138209 [DOI] [Google Scholar]
122.Bellon-Maurel V, Fernandez-Ahumada E, Palagos B, Roger J-M, McBratney A. Critical review of chemometric indicators commonly used for assessing the quality of the prediction of soil attributes by NIR spectroscopy. TrAC Trends in Analytical Chemistry. 2010;29(9):1073–81. doi: 10.1016/j.trac.2010.05.006 [DOI] [Google Scholar]
123.Zambrano-Bigiarini M. Package ‘hydroGOF’: Goodness-of-fit functions for comparison of simulated and observed hydrological time series. 2024. doi: 10.5281/zenodo.839854 [DOI] [Google Scholar]
124.Molnar C. iml: An R package for Interpretable Machine Learning. JOSS. 2018;3(26):786. doi: 10.21105/joss.00786 [DOI] [Google Scholar]
125.Molnar C. Interpretable machine learning: A guide for making black box models explainable. 2nd ed. 2024. [Google Scholar]
126.Berti M, Pizziolo M, Scaroni M, Generali M, Critelli V, Mulas M, et al. RER2023: the landslide inventory dataset of the May 2023 Emilia-Romagna event. Copernicus GmbH. 2024. doi: 10.5194/essd-2024-407 [DOI] [Google Scholar]
127.Ferrario MF, Livio F. Rapid Mapping of Landslides Induced by Heavy Rainfall in the Emilia-Romagna (Italy) Region in May 2023. Remote Sensing. 2023;16(1):122. doi: 10.3390/rs16010122 [DOI] [Google Scholar]
128.Cremonini L, Randi P, Fazzini M, Nardino M, Rossi F, Georgiadis T. Causes and Impacts of Flood Events in Emilia-Romagna (Italy) in May 2023. Land. 2024;13(11):1800. doi: 10.3390/land13111800 [DOI] [Google Scholar]
129.Krull ES, Skjemstad JO, Baldock JA. Functions of soil organic matter and the effect on soil properties. Canberra: CSIRO Land & Water. 2004. [Google Scholar]
130.Lal R. Soil erosion and carbon dynamics. Soil and Tillage Research. 2005;81(2):137–42. doi: 10.1016/j.still.2004.09.002 [DOI] [Google Scholar]
131.Pardini G, Gispert M, Dunjó G. Runoff erosion and nutrient depletion in five Mediterranean soils of NE Spain under different land use. Sci Total Environ. 2003;309(1–3):213–24. doi: 10.1016/S0048-9697(03)00007-X [DOI] [PubMed] [Google Scholar]
132.Pardini G, Gispert M, Emran M, Doni S. Rainfall/runoff/erosion relationships and soil properties survey in abandoned shallow soils of NE Spain. J Soils Sediments. 2017;17(2):499–514. doi: 10.1007/s11368-016-1532-0 [DOI] [Google Scholar]
133.Berhe AA, Kleber M. Erosion, deposition, and the persistence of soil organic matter: mechanistic considerations and problems with terminology. Earth Surf Processes Landf. 2013;38(8):908–12. doi: 10.1002/esp.3408 [DOI] [Google Scholar]
134.Berhe AA. Decomposition of organic substrates at eroding vs. depositional landform positions. Plant Soil. 2012;350(1–2):261–80. doi: 10.1007/s11104-011-0902-z [DOI] [Google Scholar]
135.Stevenson FJ. Humus chemistry – Genesis, composition and reactions. 2nd ed. New York: John Wiley and Sons. 1994. [Google Scholar]
136.Nadeu E, Berhe AA, de Vente J, Boix-Fayos C. Erosion, deposition and replacement of soil organic carbon in Mediterranean catchments: a geomorphological, isotopic and land use change approach. Biogeosciences. 2012;9(3):1099–111. doi: 10.5194/bg-9-1099-2012 [DOI] [Google Scholar]
137.Ubaldi D, Puppi G, Zanotti AL. Cartografia fitoclimatica dell’Emilia-Romagna in scala 1:500.000. Bologna: Regione Emilia-Romagna. 1996. [Google Scholar]
138.Hardie MA, Cotching WE, Doyle RB, Lisson S. Influence of climate, water content and leaching on seasonal variations in potential water repellence. Hydrological Processes. 2012;26(13):2041–8. doi: 10.1002/hyp.8312 [DOI] [Google Scholar]
139.Blanco-Canqui H, Lal R. Soil Erosion Under Forests. Principles of Soil Conservation and Management. Springer Netherlands. 2010. 321–44. doi: 10.1007/978-1-4020-8709-7_12 [DOI] [Google Scholar]
140.Lützow M v., Kögel‐Knabner I, Ekschmitt K, Matzner E, Guggenberger G, Marschner B, et al. Stabilization of organic matter in temperate soils: mechanisms and their relevance under different soil conditions – a review. European J Soil Science. 2006;57(4):426–45. doi: 10.1111/j.1365-2389.2006.00809.x [DOI] [Google Scholar]
141.Chilom G, Rice JA. Organo‐Clay Complexes in Soils and Sediments. Biophysico‐Chemical Processes Involving Natural Nonliving Organic Matter in Environmental Systems. Wiley. 2009. 111–45. doi: 10.1002/9780470494950.ch3 [DOI] [Google Scholar]
142.Kasanin-Grubin M. Badlands and environmental change. Geogr Helv. 2008;63(1):15–25. doi: 10.5194/gh-63-15-2008 [DOI] [Google Scholar]
143.Bosino A, Pellegrini L, Omran A, Bordoni M, Meisina C, Maerker M. Litho-structure of the Oltrepo Pavese, Northern Apennines (Italy). Journal of Maps. 2019;15(2):382–92. doi: 10.1080/17445647.2019.1604438 [DOI] [Google Scholar]
144.Bosino A, Szatten DA, Omran A, Crema S, Crozi M, Becker R, et al. Assessment of suspended sediment dynamics in a small ungauged badland catchment in the Northern Apennines (Italy) using an in-situ laser diffraction method. CATENA. 2022;209:105796. doi: 10.1016/j.catena.2021.105796 [DOI] [Google Scholar]
145.Leue M, Ellerbrock RH, Gerke HH. DRIFT Mapping of Organic Matter Composition at Intact Soil Aggregate Surfaces. Vadose Zone Journal. 2010;9(2):317–24. doi: 10.2136/vzj2009.0101 [DOI] [Google Scholar]
146.Del Prete M, Bentivenga M, Amato M, Basso F, Tacconi P. Badland erosion processes and their interactions with vegetation: A case study from Pisticci, Basilicata, Southern Italy. Geogr Fis Dinam Quat. 1997;20:147–55. [Google Scholar]
147.Berhe AA, Harden JW, Torn MS, Harte J. Linking soil organic matter dynamics and erosion‐induced terrestrial carbon sequestration at different landform positions. J Geophys Res. 2008;113(G4). doi: 10.1029/2008jg000751 [DOI] [Google Scholar]
148.Berhe AA, Harden JW, Torn MS, Kleber M, Burton SD, Harte J. Persistence of soil organic matter in eroding versus depositional landform positions. J Geophys Res. 2012;117(G2). doi: 10.1029/2011jg001790 [DOI] [Google Scholar]
149.Wohl E. Connectivity in rivers. Progress in Physical Geography: Earth and Environment. 2017;41(3):345–62. doi: 10.1177/0309133317714972 [DOI] [Google Scholar]
150.Doetterl S, Berhe AA, Heckman K, Lawrence C, Schnecker J, Vargas R, et al. A landscape-scale view of soil organic matter dynamics. Nat Rev Earth Environ. 2025;6(1):67–81. doi: 10.1038/s43017-024-00621-2 [DOI] [Google Scholar]
151.Ellerbrock RH, Kaiser M. Stability and composition of different soluble soil organic matter fractions–evidence from δ¹³C and FTIR signatures. Geoderma. 2005;128(1–2):28–37. doi: 10.1016/j.geoderma.2004.12.025 [DOI] [Google Scholar]
152.Dawson LA, Hillier S. Measurement of soil characteristics for forensic applications. Surface & Interface Analysis. 2010;42(5):363–77. doi: 10.1002/sia.3315 [DOI] [Google Scholar]
153.Sadeghi SH, Singh VP, Kiani-Harchegani M, Asadi H. Analysis of sediment rating loops and particle size distributions to characterize sediment source at mid-sized plot scale. CATENA. 2018;167:221–7. doi: 10.1016/j.catena.2018.05.002 [DOI] [Google Scholar]
154.Ellerbrock R, Stein M, Schaller J. Comparing amorphous silica, short-range-ordered silicates and silicic acid species by FTIR. Sci Rep. 2022;12(1):11708. doi: 10.1038/s41598-022-15882-4 [DOI] [PMC free article] [PubMed] [Google Scholar]
155.Ellerbrock RH, Stein M, Schaller J. Comparing silicon mineral species of different crystallinity using Fourier transform infrared spectroscopy. Front Environ Chem. 2024;5:1462678. doi: 10.3389/fenvc.2024.1462678 [DOI] [Google Scholar]
156.Barbosa LAP, Stein M, Gerke HH, Schaller J. Synergistic effects of organic carbon and silica in preserving structural stability of drying soils. Sci Rep. 2024;14(1):8330. doi: 10.1038/s41598-024-58916-9 [DOI] [PMC free article] [PubMed] [Google Scholar]

PLoS One. 2025 Aug 21;20(8):e0330252. doi: 10.1371/journal.pone.0330252.r001

Author response to Decision Letter 0

12 Feb 2025

Attachment

Submitted filename: PONE-D-25-04924 Response_to_Editor.docx

pone.0330252.s005.docx^{(47KB, docx)}

PLoS One. doi: 10.1371/journal.pone.0330252.r002

Decision Letter 0

Przemysław Mroczek

15 May 2025

PONE-D-25-04924

Integration of FTIR-based soil degradation indices with stochastic modelling to assess spatial patterns of organic matter-sediment dynamics in a Mediterranean watershed

PLOS ONE

Dear Dr. La Licata,

Thank you for submitting your manuscript to PLOS ONE. After careful consideration, we feel that it has merit but does not fully meet PLOS ONE’s publication criteria as it currently stands. Therefore, we invite you to submit a revised version of the manuscript that addresses the points raised during the review process.

Manuscript presents a valuable and methodologically sound case study integrating FTIR-based soil characterisation with spatial modelling. Reviewer 1 recommends acceptance, while Reviewer 2 requests minor revisions regarding title precision, figure quality, and clarification of erosion processes. In addition to these points, I require the authors to address several editorial concerns:

(1) avoid overgeneralisation of results beyond the studied catchment;

(2) moderate the interpretation of FTIR indices as direct erosion proxies, clearly stating their indirect nature;

(3) discuss the potential sampling bias due to post-event sediment collection;

(4) critically reflect on model uncertainties and limitations; and

(5) better integrate lithological and geomorphological controls into the analysis.

These revisions are necessary to meet PLOS ONE’s standards of scientific transparency and accurate interpretation.

Please submit your revised manuscript by Jun 29 2025 11:59PM. If you will need more time than this to complete your revisions, please reply to this message or contact the journal office at plosone@plos.org . When you're ready to submit your revision, log on to https://www.editorialmanager.com/pone/ and select the 'Submissions Needing Revision' folder to locate your manuscript file.

Please include the following items when submitting your revised manuscript:

A rebuttal letter that responds to each point raised by the academic editor and reviewer(s). You should upload this letter as a separate file labeled 'Response to Reviewers'.
A marked-up copy of your manuscript that highlights changes made to the original version. You should upload this as a separate file labeled 'Revised Manuscript with Track Changes'.
An unmarked version of your revised paper without tracked changes. You should upload this as a separate file labeled 'Manuscript'.

If you would like to make changes to your financial disclosure, please include your updated statement in your cover letter. Guidelines for resubmitting your figure files are available below the reviewer comments at the end of this letter.

If applicable, we recommend that you deposit your laboratory protocols in protocols.io to enhance the reproducibility of your results. Protocols.io assigns your protocol its own identifier (DOI) so that it can be cited independently in the future. For instructions see: https://journals.plos.org/plosone/s/submission-guidelines#loc-laboratory-protocols . Additionally, PLOS ONE offers an option for publishing peer-reviewed Lab Protocol articles, which describe protocols hosted on protocols.io. Read more information on sharing protocols at https://plos.org/protocols?utm_medium=editorial-email&utm_source=authorletters&utm_campaign=protocols .

We look forward to receiving your revised manuscript.

Kind regards,

Przemysław Mroczek, Dr. hab.

Academic Editor

PLOS ONE

Journal Requirements:

When submitting your revision, we need you to address these additional requirements.

1.Please ensure that your manuscript meets PLOS ONE's style requirements, including those for file naming. The PLOS ONE style templates can be found at

https://journals.plos.org/plosone/s/file?id=wjVg/PLOSOne_formatting_sample_main_body.pdf and

https://journals.plos.org/plosone/s/file?id=ba62/PLOSOne_formatting_sample_title_authors_affiliations.pdf

2. Please note that PLOS ONE has specific guidelines on code sharing for submissions in which author-generated code underpins the findings in the manuscript. In these cases, we expect all author-generated code to be made available without restrictions upon publication of the work. Please review our guidelines at https://journals.plos.org/plosone/s/materials-and-software-sharing#loc-sharing-code and ensure that your code is shared in a way that follows best practice and facilitates reproducibility and reuse.

3. In your Methods section, please provide additional information regarding the permits you obtained for the work. Please ensure you have included the full name of the authority that approved the field site access and, if no permits were required, a brief statement explaining why.

4. Thank you for stating the following financial disclosure:

“This research was conducted with the financial support of the Earth and Environmental Sciences PhD-PON program (Research & Innovation, 2014e2020, Education and research for recovery - REACT-EU, DOT1322534-4) of University of Pavia, Department of Earth and Environmental Sciences. This research was also funded by the PRIN 2022 project by the Italian Ministry of University and Research, entitled: ”Full cOveRage, Multi-scAle and multi-sensor geomorphological map: a practical tool for TerrItOrial plaNning – FORMATION” (2022C2XPK7_004). This research was also supported by the Polish National Agency for Academic Exchange - Urgency Grants program project contract no. BPN/GIN/2024/1/00008.”

Please state what role the funders took in the study. If the funders had no role, please state: "The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript."

If this statement is not correct you must amend it as needed.

Please include this amended Role of Funder statement in your cover letter; we will change the online submission form on your behalf.

5. When completing the data availability statement of the submission form, you indicated that you will make your data available on acceptance. We strongly recommend all authors decide on a data sharing plan before acceptance, as the process can be lengthy and hold up publication timelines. Please note that, though access restrictions are acceptable now, your entire data will need to be made freely accessible if your manuscript is accepted for publication. This policy applies to all data except where public deposition would breach compliance with the protocol approved by your research ethics board. If you are unable to adhere to our open data policy, please kindly revise your statement to explain your reasoning and we will seek the editor's input on an exemption. Please be assured that, once you have provided your new statement, the assessment of your exemption will not hold up the peer review process.

6. We note that Figures 1 and 4-7 in your submission contain map/satellite images which may be copyrighted. All PLOS content is published under the Creative Commons Attribution License (CC BY 4.0), which means that the manuscript, images, and Supporting Information files will be freely available online, and any third party is permitted to access, download, copy, distribute, and use these materials in any way, even commercially, with proper attribution. For these reasons, we cannot publish previously copyrighted maps or satellite images created using proprietary data, such as Google software (Google Maps, Street View, and Earth). For more information, see our copyright guidelines: http://journals.plos.org/plosone/s/licenses-and-copyright.

We require you to either (1) present written permission from the copyright holder to publish these figures specifically under the CC BY 4.0 license, or (2) remove the figures from your submission:

1. You may seek permission from the original copyright holder of Figures 1 and 4-7 to publish the content specifically under the CC BY 4.0 license.

We recommend that you contact the original copyright holder with the Content Permission Form (http://journals.plos.org/plosone/s/file?id=7c09/content-permission-form.pdf) and the following text:

“I request permission for the open-access journal PLOS ONE to publish XXX under the Creative Commons Attribution License (CCAL) CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). Please be aware that this license allows unrestricted use and distribution, even commercially, by third parties. Please reply and provide explicit written permission to publish XXX under a CC BY license and complete the attached form.”

Please upload the completed Content Permission Form or other proof of granted permissions as an "Other" file with your submission.

In the figure caption of the copyrighted figure, please include the following text: “Reprinted from [ref] under a CC BY license, with permission from [name of publisher], original copyright [original copyright year].”

2. If you are unable to obtain permission from the original copyright holder to publish these figures under the CC BY 4.0 license or if the copyright holder’s requirements are incompatible with the CC BY 4.0 license, please either i) remove the figure or ii) supply a replacement figure that complies with the CC BY 4.0 license. Please check copyright information on all replacement figures and update the figure caption with source information. If applicable, please specify in the figure caption text when a figure is similar but not identical to the original image and is therefore for illustrative purposes only.

The following resources for replacing copyrighted map figures may be helpful:

USGS National Map Viewer (public domain): http://viewer.nationalmap.gov/viewer/

The Gateway to Astronaut Photography of Earth (public domain): http://eol.jsc.nasa.gov/sseop/clickmap/

Maps at the CIA (public domain): https://www.cia.gov/library/publications/the-world-factbook/index.html and https://www.cia.gov/library/publications/cia-maps-publications/index.html

NASA Earth Observatory (public domain): http://earthobservatory.nasa.gov/

Landsat: http://landsat.visibleearth.nasa.gov/

USGS EROS (Earth Resources Observatory and Science (EROS) Center) (public domain): http://eros.usgs.gov/#

Natural Earth (public domain): http://www.naturalearthdata.com/

7. Please review your reference list to ensure that it is complete and correct. If you have cited papers that have been retracted, please include the rationale for doing so in the manuscript text, or remove these references and replace them with relevant current references. Any changes to the reference list should be mentioned in the rebuttal letter that accompanies your revised manuscript. If you need to cite a retracted article, indicate the article’s retracted status in the References list and also include a citation and full reference for the retraction notice.

Additional Editor Comments:

Dear Authors,

Thank you for submitting your manuscript entitled “Integration of FTIR-based soil degradation indices with stochastic modelling to assess spatial patterns of organic matter-sediment dynamics in a Mediterranean watershed” to PLOS ONE. I appreciate the considerable effort invested in this interdisciplinary study, which integrates FTIR spectroscopy with geomorphological and statistical modelling approaches to investigate organic matter and sediment dynamics in a Mediterranean catchment.

The manuscript has been reviewed by two external referees. Reviewer 1 provided a very positive evaluation, stating that your research is technically sound, the data are sufficient and appropriately analysed, and the manuscript is clearly written. They commended the scientific relevance of your study and acknowledged the improvements made since the initial submission. Reviewer 2 also recognised the value of your work but raised several points requiring attention. Specifically, they suggested that the manuscript’s title should be more precise, explicitly reflecting the study area rather than implying a broader regional scope. They noted deficiencies in figure quality, particularly Figure 1, and recommended improvements in visual presentation, including the addition of geological and land use maps, as well as photographs of typical lithologies. Furthermore, they pointed out inconsistencies in figure referencing within the text and encouraged a more nuanced discussion of erosion processes, especially regarding the role of hydrophobicity in soil stability.

In addition to the reviewers' comments, I would like to draw your attention to several further issues that require careful consideration. While your integration of FTIR-derived indices with Random Forest modelling is methodologically sound, the interpretation of these proxies as direct indicators of erosion susceptibility appears overstated. The manuscript lacks empirical validation linking the FTIR indices to actual erosion measurements, sediment yields, or field-observed processes. This disconnect should be addressed with a more cautious interpretation of your findings, clearly acknowledging the proxy-based and indirect nature of your assessments.

Moreover, your conclusions frequently extrapolate results from a relatively small catchment in the Northern Apennines to “Mediterranean watersheds” in general. Such generalisations are not sufficiently substantiated and should be carefully delimited to the specific physiographic and climatic context of your study area. A more balanced formulation that frames your findings as case-study insights rather than general regional conclusions is recommended.

Another aspect requiring attention concerns the sampling strategy. Given that samples were collected following a significant rainfall event, it would be pertinent to discuss how this may have influenced sediment composition and whether such conditions are representative of the typical sediment dynamics in the catchment. This context is essential for interpreting your results.

Additionally, while your Random Forest models are well described, there is little discussion of model uncertainties, potential overfitting, or limitations. Providing a critical reflection on the robustness of your models and discussing the possible influence of dataset characteristics would enhance the scientific transparency of your work.

Finally, the integration of lithological controls and the role of mass movements in sediment connectivity and SOM dynamics are not fully explored. Although mentioned in the text, these factors are not quantitatively assessed nor sufficiently linked to your FTIR-based analyses. A more detailed examination of how different lithologies and geomorphological processes affect your observed patterns would significantly strengthen the study.

I therefore invite you to revise your manuscript by addressing both the reviewers’ comments and the editorial concerns outlined above. I believe these improvements will enhance the clarity, scientific rigour, and overall impact of your work.

Yours sincerely,

Przemysław Mroczek

Academic Editor

PLOS ONE

[Note: HTML markup is below. Please do not edit.]

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #1: Yes

Reviewer #2: Yes

**********

2. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #1: Yes

Reviewer #2: Yes

**********

3. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #1: Yes

Reviewer #2: Yes

**********

4. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #1: Yes

Reviewer #2: Yes

**********

5. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #1: I believe that authors put sufficient data in order to explain their approach and to enable reproducibility. This is a a valuable contribution for the study of landscape evolution in Mediterranean watersheds. I also understand that the MS went through one round of review with different reviewers and authors made significant changes.

Reviewer #2: The title of the article suggests research in the Mediterranean region, whose coastline is very long, extensive, and is based on many countries, regions, etc. Meanwhile, when reading the text, we read that the authors are basically studying the catchment area of one river flowing into the Adriatic Sea, which concerns the region of Emilia-Romagna, northern Italy, the Arda River. This means that the title should be changed, precisely defining the area.

Fig. 1. The quality of these images is low, it needs to be improved. The hatching also raises concerns because the maximum height of 1300m is marked as if it were a height of over 5000m.

Line 134 and others: I suggest writing Roman numerals with a capital letter.

The authors described the catchment area of the studied river, also showing the lithology of the bedrock. They indicate in the text that mass movements occur there. Maybe it would be worth including a geological and soil map in this place? It would be good to explain why mass movements occur, whether it is the effect of soil instability resulting from heavy rains, the arrangement of rocks in the subsoil, or improper use of soils. This information has an impact on the rest of the text, especially since the authors themselves indicate that mass movements destroy soils (line 149). It is also worth considering how fields are ploughed, which can facilitate or hinder surface runoff of water on slopes.

Table 3 is interesting, but the authors could include photographs of typical examples of rocks and a map of agricultural use of the research area.

After Fig. 5, the authors describe Figure 7 (line 439) in the text, so it should be moved so that the order of citation of graphics in the text is maintained. There is no description of Figure 6 in this paragraph. This needs to be supplemented. There is only one reference in line 473, but it is too little for the information contained there.

Line 571, requires consideration, the hydrophobicity of matter alone does not have to stop erosion, which can be caused by, for example, soil instability and mechanical washing out. In addition, chemical compounds can appear in polluted waters that cause emulsification of such compounds.

**********

6. PLOS authors have the option to publish the peer review history of their article (what does this mean? ). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy .

Reviewer #1: Yes: Bülent Arıkan

Reviewer #2: Yes: Miłosz Huber

**********

[NOTE: If reviewer comments were submitted as an attachment file, they will be attached to this email and accessible via the submission site. Please log into your account, locate the manuscript record, and check for the action link "View Attachments". If this link does not appear, there are no attachment files.]

While revising your submission, please upload your figure files to the Preflight Analysis and Conversion Engine (PACE) digital diagnostic tool, https://pacev2.apexcovantage.com/ . PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Registration is free. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email PLOS at figures@plos.org . Please note that Supporting Information files do not need this step.

PLoS One. 2025 Aug 21;20(8):e0330252. doi: 10.1371/journal.pone.0330252.r003

Author response to Decision Letter 1

6 Jun 2025

Dear Editor,

Dear Reviewers,

We would like to express our gratitude for your valuable comments and suggestions, which have significantly contributed to improve the manuscript. We have tracked the changes throughout the manuscript and provided a clean copy without tracked changes, as requested by the journal. We have addressed your requests and comments throughout the manuscript. Please find below our responses. The specific references to the modified lines are referred to the “clean copy”.

Reviewers’ comments:

Reviewer_1: The title of the article suggests research in the Mediterranean region, whose coastline is very long, extensive, and is based on many countries, regions, etc. Meanwhile, when reading the text, we read that the authors are basically studying the catchment area of one river flowing into the Adriatic Sea, which concerns the region of Emilia-Romagna, northern Italy, the Arda River. This means that the title should be changed, precisely defining the area.

Corresponding author (CA): Thank you so much for the valuable comment. We agree with you and modified the title accordingly. Now the title sounds: “Integration of FTIR-based soil degradation indices with stochastic modelling to assess spatial patterns of organic matter-sediment dynamics in a Mediterranean watershed – A case study from the Northern Apennines”. We believe that this title better reflect the regional specificity of the study watershed, while keeping the attention of the reader to the Mediterranean. Hence, we believe that the reader should be clearly informed from the title that we are dealing with Mediterranean characteristics, in spite of we are studying only one watershed in the Apennines. The title is 183 characters without spaces or 211 with spaces, thereby being consistent with journal requirements (i.e., max 250 characters).

Reviewer_2: Fig. 1. The quality of these images is low, it needs to be improved. The hatching also raises concerns because the maximum height of 1300m is marked as if it were a height of over 5000m.

CA: We greatly appreciate your comment and your attention to the quality of the figures. It is possible that the version of the figure displayed in the submission document does not show the overall map and the photographs below with optimal clarity. However, we have provided all images at a resolution of 500 dpi, and we are confident that the final publication will ensure high-quality rendering. If by “image quality” you referred to something else, we kindly ask you to clarify this point…

Figure 1 provides all the necessary elements to properly outline the study area in its geographical context, offering both a regional overview and a basic characterization of the topography of the area. This is intended to help the reader better understand the distribution of sampling sites across the landscape. A zoomed-in panel further illustrates the relative position of the samples along the valley floor, accompanied by representative photographs (in high resolution). Therefore, we believe that this figure offers a useful and essential tool for contextualizing the sampling design within the studied watershed. The legend clearly shows the elevation range of the study area, leaving no particular ambiguity about the topographic configuration or actual elevation of the watershed. The DTM was classified using graduated colours, based on a scale bar that is commonly used for elevation data, and semi-transparently overlaid on a hillshade map to improve the visualization of landscape morphology (an approach widely used for this type of map). Thus, we understand that the original colour scale, from light-blue (valley floors) to white (main peaks), may have been somewhat misleading at first glance, as it may have suggested higher alpine-like peaks. For this reason, we have adjusted the scale bar by removing the white colour, making the visualization more appropriate for the actual elevation range of the study area. Please check the new Figure 1 provided along with the revised submission. Anyway, please consider that Fig. 1a has been changed to accomplish journal requirements (see comment below; cfr. comment ‘Journal_6’).

Reviewer_3: Line 134 and others: I suggest writing Roman numerals with a capital letter.

CA: Thank you for the comment. Roman numerals have been rewritten using capital letters throughout the text.

Reviewer_4: The authors described the catchment area of the studied river, also showing the lithology of the bedrock. They indicate in the text that mass movements occur there. Maybe it would be worth including a geological and soil map in this place? It would be good to explain why mass movements occur, whether it is the effect of soil instability resulting from heavy rains, the arrangement of rocks in the subsoil, or improper use of soils. This information has an impact on the rest of the text, especially since the authors themselves indicate that mass movements destroy soils (line 149). It is also worth considering how fields are ploughed, which can facilitate or hinder surface runoff of water on slopes.

CA: Thank you for your suggestion and consideration. In our opinion, including geological and soil maps in the Study Area chapter would go beyond the scope of this work, considering that this study does not aim to explore relationships between erosion and landslide processes and specific variables such as all the outcropping geological formations or soil types. In this study, geological formations are grouped based on lithological characteristics and relevant/available soil properties are taken separately as single variables. Moreover, for the whole study area, the available soil map is at a scale of 1:250,000, while more detailed information (1:50,000) is only available for limited parts of the watershed. Therefore, the integration of pedological data in this form would likely not be particularly useful. The core of the study lies in the relationship between the so-called ‘target variables’ (FTIR indices) and the ‘feature variables’, which are analyzed within a machine learning framework rather than from a qualitative perspective. All lithological and pedological features that are relevant and available in an appropriate format (e.g. raster, shapefile) have already been incorporated into the model, independently from the classifications found in soil maps. Thus, we believe that integrating geological or soil maps would not contribute substantially to the analysis, since these variables are already considered in the model and are presented as part of the results, particularly in Figure 7, which already shows the spatial distribution of lithological, pedological, and land use data.

However, we agree that adding some further contextual information in the Study Area chapter could help clarify key aspects of the area's geomorphology. Therefore, we have revised the part referring to landslides as follows: “Landslides are the dominant land degradation process, exhibiting considerable variability in magnitude and frequency. In the study area, landslide activity is largely dominated by periodic reactivation of pre-existing large-scale landslide bodies, primarily triggered by intense or prolonged rainfall events. Fluvial undercutting at the toe of landslide deposits also plays a critical role in destabilizing valley slopes” (lines 153-157).

Regarding the suggestion to provide information on how the fields are ploughed, we unfortunately do not have detailed data on this. Most of the cultivated areas in the study area are classified as rainfed arable lands, with very few hectares of vineyards and orchards (they are not enough to justify a separate class as they are very small and sporadic polygons in the overall land use map). Please note that the variable ‘cultivated fields’ is among the feature variables integrated into the model, but the results did not reveal any significant relationship between this variable and the FTIR indices. Please also note that in the Study area chapter, we explicitly stated that: " Rainfed arable lands represent the predominant agricultural landscape in the watershed..." (line 168).

Reviewer_5: Table 3 is interesting, but the authors could include photographs of typical examples of rocks and a map of agricultural use of the research area.

CA: Thank you for your appreciation. In our opinion, adding photographs of representative rock types could be somewhat interesting, but it would not substantially add value to the research, as the characteristics of the lithological groups are already clearly described in Table 3. Moreover, this information is relevant to our study because it relates to the parent material of the soils. Additionally, the characteristics of sediments may be affected by the bedrock from which they come from, rather than to rock outcrops or lithological material themselves. Therefore, including pictures of rocks or outcrops may not be entirely relevant in this context. Regarding agricultural land use in the study area, a classification comprising 10 land use classes (including agricultural areas) has already been integrated into the model. A map showing the classified land use is also presented as part of the results in Figure 7.

Reviewer_6: After Fig. 5, the authors describe Figure 7 (line 439) in the text, so it should be moved so that the order of citation of graphics in the text is maintained. There is no description of Figure 6 in this paragraph. This needs to be supplemented. There is only one reference in line 473, but it is too little for the information contained there.

CA: Thank you for your comment. We understand the point you raised. However, Figure 6 was already cited at line 451 (previously line 436 in the former submitted version), before the citation of Figure 7 at line 454 (previously line 439). Therefore, the consistency in the figure numbering and citation order is maintained. Also, we would like to clarify that the textual information regarding basic statistics of the raster-based feature variables (Figures 6 and 7) is not reported in the manuscript text but only in the Supporting Information, as explicitly stated in the manuscript (lines 454-455). This was a specific request from the Editor during the initial revision round, before the manuscript was sent out for peer review. Indeed, the Editor specifically asked us to move all numerical and technical data (such as these basic statistics) to the Supporting Information, in order to avoid overloading the main text with non-essential content and to help readers focus on the most relevant information for interpretation and discussion. On the other hand, the subsequent information referring to Figure 7 (lines 456-467) mainly concerns the shapefile-based variables (i.e., lithology and land use), from which proportions were extracted. In this case, the text does not only present numerical data that could be moved to the Supporting Information, but also provides qualitative and descriptive insights into the distribution of these variables across the study area. These elements are indeed important to support the discussion (you will see that Figs. 7h and 7i have been extensively cited in the discussion).

Reviewer_7: Line 571, requires consideration, the hydrophobicity of matter alone does not have to stop erosion, which can be caused by, for example, soil instability and mechanical washing out. In addition, chemical compounds can appear in polluted waters that cause emulsification of such compounds.

CA: Thank you for your comment. However, in the discussion, we avoid any reference to a "protective" or "preventive" role of hydrophobicity against soil erosion. On the contrary, we stated the pedological significance of hydrophobicity on soil erosion in the Introduction, before outlining the objectives of the present work. We are fully aware that erosion can occur in areas affected by soil instability, which is the reason why we decided to focus on three different hydrological/geomorphological settings that take these aspects into account. All of this is addressed in the Discussion (see paragraphs below).

--- --- --- --- --- --- --- --- --- --- --- --- --- --- --- --- --- --- --- --- --- --- --- --- --- --- --- --- --- ---

Editor’s additional comments:

Editor_1: In addition to the reviewers' comments, I would like to draw your attention to several further issues that require careful consideration. While your integration of FTIR-derived indices with Random Forest modelling is methodologically sound, the interpretation of these proxies as direct indicators of erosion susceptibility appears overstated. The manuscript lacks empirical validation linking the FTIR indices to actual erosion measurements, sediment yields, or field-observed processes. This disconnect should be addressed with a more cautious interpretation of your findings, clearly acknowledging the proxy-based and indirect nature of your assessments.

CA: Thank you so much for the comment. We agree with your point of view. We have added a short paragraph in the section ‘5.3 Limitations of the methodological approach and improvements’: “In general, the significance of these findings relies on the assumption that the FTIR indices used in this study can serve as proxies for soil hydrophobicity, CEC, and organic matter–cation interactions (Table 1) within this environmental context, as supported by evidence drawn from the literature [39,50-55,84]. The obtained results represent an initial preliminary attempt towards a better understanding of such relationships. However, these results should be interpreted with caution, e.g., as we discussed for their relevance in terms of erosion processes. That is, an empirical evaluation linking FTIR indices to actual erosion measurements—necessary to establish these indices as direct indicators—is currently lacking. Nonetheless, additional studies are needed to explore more detailed relationships between these potential FTIR-based indicators and the chemical and pedological characteristics of sediment source areas” (Lines 737-745). Moreover, we would like to stress the fact that in the manuscript we are frequently referring to these indices as “potential indicators”, thereby implicitly reflecting their nature in relation to the methodology and the approach employed in the study. In any case, in the manuscript we never refer to definitive evidence or unquestionable results, but we consistently use modal verbs (e.g., could, might, may, etc…). In this regard, we have slightly refined some sentences (please check the Tracked Changes Copy).

Editor_2: Moreover, your conclusions frequently extrapolate results from a relatively small catchment in the Northern Apennines to “Mediterranean watersheds” in general. Such generalisations are not sufficiently substantiated and should be carefully delimited to the specific physiographic and climatic context of your study area. A more balanced formulation that frames your findings as case-study insights rather than general regional conclusions is recommended.

CA: Thank you for the valuable comment. We agree that, in the Conclusion, we should refer to these extrapolations more cautiously. Therefore, we have made some corrections:

- Lines 785-786: we have clearly specified that we selected a “temperate agricultural-forested area of the Northern Apennines chosen as representative of Mediterranean environments”;

- Lines 802-803: “This underscores the complexity of SOM dynamics in environments with similar characteristics and supports…”;

- Lines 810-812: “These findings carry significant implications for Mediterranean watersheds having similar vegetation, as well as for regions with seasonally contrasting climates and lithologies dominated by weak, stratified, clay-rich rocks”;

- Lines 816-818: “These insights suggest that such landscape-based modelling could be a valuable approach for predicting organic matter cycling and carbon sequestration in complex, dynamic Mediterranean watersheds with similar landscape configuration”.

Editor_3: Another aspect requiring attention concerns the sampling strategy. Given that samples were collected following a significant rainfall event, it would be pertinent to discuss how this may have influenced sediment composition and whether such cond

Attachment

Submitted filename: Response to Reviewers.docx

pone.0330252.s006.docx^{(48.9KB, docx)}

PLoS One. doi: 10.1371/journal.pone.0330252.r004

Decision Letter 1

Przemysław Mroczek

18 Jun 2025

PONE-D-25-04924R1Integration of FTIR-based soil degradation indices with stochastic modelling to assess spatial patterns of organic matter-sediment dynamics in a Mediterranean watershed – A case study from the Northern ApenninesPLOS ONE

Dear Dr. La Licata,

The manuscript now fully meets PLOS ONE’s methodological and data requirements and addresses all major reviewer and editor concerns. A minor revision is requested solely to streamline the title and refine the keywords to avoid unnecessary repetition and ensure better indexing. No further substantive changes are required, and I recommend acceptance once these final editorial points have been implemented.

Please submit your revised manuscript by Aug 02 2025 11:59PM. If you will need more time than this to complete your revisions, please reply to this message or contact the journal office at plosone@plos.org . When you're ready to submit your revision, log on to https://www.editorialmanager.com/pone/ and select the 'Submissions Needing Revision' folder to locate your manuscript file.

Please include the following items when submitting your revised manuscript:

A rebuttal letter that responds to each point raised by the academic editor and reviewer(s). You should upload this letter as a separate file labeled 'Response to Reviewers'.
A marked-up copy of your manuscript that highlights changes made to the original version. You should upload this as a separate file labeled 'Revised Manuscript with Track Changes'.
An unmarked version of your revised paper without tracked changes. You should upload this as a separate file labeled 'Manuscript'.

We look forward to receiving your revised manuscript.

Kind regards,

Przemysław Mroczek, Dr. hab.

Academic Editor

PLOS ONE

Journal Requirements:

Please review your reference list to ensure that it is complete and correct. If you have cited papers that have been retracted, please include the rationale for doing so in the manuscript text, or remove these references and replace them with relevant current references. Any changes to the reference list should be mentioned in the rebuttal letter that accompanies your revised manuscript. If you need to cite a retracted article, indicate the article’s retracted status in the References list and also include a citation and full reference for the retraction notice.

Additional Editor Comments :

Dear Dr La Licata and Co-authors,

Thank you very much for submitting the revised version of your manuscript entitled “Integration of FTIR-based soil degradation indices with stochastic modelling to assess spatial patterns of organic matter–sediment dynamics in a Mediterranean watershed – A case study from the Northern Apennines” to PLOS ONE. I sincerely appreciate the significant effort you have devoted to addressing the reviewers’ and editorial comments. The manuscript has improved considerably in terms of methodological transparency, a more balanced interpretation of proxy-based findings, and an overall clearer presentation of the research context.

Before I can proceed with a final decision, I would kindly ask you to address two small but important editorial points to ensure that your manuscript meets the journal’s best practice for clarity and discoverability. Firstly, while your current title correctly reflects the study’s scope and regional focus, it is rather long and includes redundant wording that makes it unnecessarily complex. PLOS ONE encourages concise titles that convey the essential elements without superfluous phrases. I therefore strongly recommend shortening it to the following form: “FTIR-derived soil degradation indices and stochastic modelling of organic matter–sediment dynamics in a Mediterranean catchment: a Northern Apennines case study.” This version preserves all key terms but removes the overly extended structure and the phrase “to assess spatial patterns”, which is implicit in your methods and discussion.

Secondly, I encourage you to refine your set of keywords to avoid repeating phrases already present in the title. For example, “Organic matter indices”, “Mediterranean watershed”, and “Geomorphological modelling” duplicate core title terms and thus reduce the added value of the keywords for search engines and indexing. A more effective set could include terms that highlight important aspects of your study not directly stated in the title. I suggest the following as an example: Soil erosion risk, FTIR spectroscopy, Soil organic matter composition, Random Forest modelling, Watershed connectivity, and Sediment transport processes. This selection emphasises the main processes, analytical approach, and spatial system interactions central to your manuscript.

Once you have adjusted the title and keywords accordingly, I will be pleased to proceed with the final acceptance. No further substantive revisions are needed. Thank you again for your thoughtful and thorough response to the peer review process and for choosing PLOS ONE for the publication of your work. I look forward to receiving your final version soon.

With best regards,

Przemysław Mroczek, Dr hab.

Academic Editor, PLOS ONE

[Note: HTML markup is below. Please do not edit.]

PLoS One. 2025 Aug 21;20(8):e0330252. doi: 10.1371/journal.pone.0330252.r005

Author response to Decision Letter 2

18 Jul 2025

Dear Editor,

We would like to thank you very much for all the work you have done with us in these months, during several revision processes, which has led to a result that we are very satisfied. Please find below our responses.

Editor’ comments:

Editor_1: Firstly, while your current title correctly reflects the study’s scope and regional focus, it is rather long and includes redundant wording that makes it unnecessarily complex. PLOS ONE encourages concise titles that convey the essential elements without superfluous phrases. I therefore strongly recommend shortening it to the following form: “FTIR-derived soil degradation indices and stochastic modelling of organic matter–sediment dynamics in a Mediterranean catchment: a Northern Apennines case study.” This version preserves all key terms but removes the overly extended structure and the phrase “to assess spatial patterns”, which is implicit in your methods and discussion.

Corresponding author (CA): Thank you very much for the comment and the suggestion. We have decided to adopt the title you proposed, with just one word changed. We chose to keep the term “watershed” instead of “catchment” for consistency with the terminology used in the text. The title now reads as follows: "FTIR-derived soil degradation indices and stochastic modelling of organic matter–sediment dynamics in a Mediterranean watershed: a Northern Apennines case study".

Editor_2: Secondly, I encourage you to refine your set of keywords to avoid repeating phrases already present in the title. For example, “Organic matter indices”, “Mediterranean watershed”, and “Geomorphological modelling” duplicate core title terms and thus reduce the added value of the keywords for search engines and indexing. A more effective set could include terms that highlight important aspects of your study not directly stated in the title. I suggest the following as an example: Soil erosion risk, FTIR spectroscopy, Soil organic matter composition, Random Forest modelling, Watershed connectivity, and Sediment transport processes. This selection emphasises the main processes, analytical approach, and spatial system interactions central to your manuscript.

CA: Thank you very much for your comment and suggestion. Also in this case, we decided to use essentially the keywords you proposed, with only minor adjustments. Now the keywords are: “Soil erosion susceptibility, FTIR spectroscopy, Soil organic matter composition, Random Forest modelling, Landscape connectivity, Sediment transport processes”. We chose to avoid the term “risk” because in applied geomorphology this term has a specific meaning that does not apply to the scope of our work. Therefore, we would prefer to avoid potential misunderstandings among specialized readers. Moreover, compared to your proposal, we decided to keep “landscape connectivity” instead of “watershed connectivity,” as this term (and its related definition) is more relevant and consistent with the literature cited in the manuscript.

Attachment

Submitted filename: Response_to_Reviewers_auresp_2.docx

pone.0330252.s007.docx^{(30.8KB, docx)}

PLoS One. doi: 10.1371/journal.pone.0330252.r006

Decision Letter 2

Przemysław Mroczek

30 Jul 2025

FTIR-derived soil degradation indices and stochastic modelling of organic matter–sediment dynamics in a Mediterranean watershed: a Northern Apennines case study

PONE-D-25-04924R2

Dear Dr. La Licata,

We’re pleased to inform you that your manuscript has been judged scientifically suitable for publication and will be formally accepted for publication once it meets all outstanding technical requirements.

Within one week, you’ll receive an e-mail detailing the required amendments. When these have been addressed, you’ll receive a formal acceptance letter and your manuscript will be scheduled for publication.

An invoice will be generated when your article is formally accepted. Please note, if your institution has a publishing partnership with PLOS and your article meets the relevant criteria, all or part of your publication costs will be covered. Please make sure your user information is up-to-date by logging into Editorial Manager at Editorial Manager® and clicking the ‘Update My Information' link at the top of the page. For questions related to billing, please contact billing support .

If your institution or institutions have a press office, please notify them about your upcoming paper to help maximize its impact. If they’ll be preparing press materials, please inform our press team as soon as possible -- no later than 48 hours after receiving the formal acceptance. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information, please contact onepress@plos.org.

Kind regards,

Przemysław Mroczek, Dr. hab.

Academic Editor

PLOS ONE

Additional Editor Comments (optional):

Thank you for your thorough revision and for addressing the editorial suggestions with care and clarity. The changes made to the title, keyword selection, and data availability statement are appropriate and fully in line with the previous guidance. I appreciate the effort to enhance the clarity and coherence of the manuscript.

However, I note that the revised version labelled as "tracked changes" does not actually contain any visible change markings. While the clean version appears to incorporate the necessary revisions, the absence of a tracked version prevents a straightforward verification of specific edits. I recommend ensuring that such a file is included in future submissions where applicable.

I would like to thank you and your co-authors for your constructive engagement throughout the editorial process. I believe the manuscript is now in a form suitable for publication.

Reviewers' comments:

PLoS One. doi: 10.1371/journal.pone.0330252.r007

Acceptance letter

Przemysław Mroczek

PONE-D-25-04924R2

PLOS ONE

Dear Dr. La Licata,

I'm pleased to inform you that your manuscript has been deemed suitable for publication in PLOS ONE. Congratulations! Your manuscript is now being handed over to our production team.

At this stage, our production department will prepare your paper for publication. This includes ensuring the following:

* All references, tables, and figures are properly cited

* All relevant supporting information is included in the manuscript submission,

* There are no issues that prevent the paper from being properly typeset

You will receive further instructions from the production team, including instructions on how to review your proof when it is ready. Please keep in mind that we are working through a large volume of accepted articles, so please give us a few days to review your paper and let you know the next and final steps.

Lastly, if your institution or institutions have a press office, please let them know about your upcoming paper now to help maximize its impact. If they'll be preparing press materials, please inform our press team within the next 48 hours. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information, please contact onepress@plos.org.

You will receive an invoice from PLOS for your publication fee after your manuscript has reached the completed accept phase. If you receive an email requesting payment before acceptance or for any other service, this may be a phishing scheme. Learn how to identify phishing emails and protect your accounts at https://explore.plos.org/phishing.

If we can help with anything else, please email us at customercare@plos.org.

Thank you for submitting your work to PLOS ONE and supporting open access.

Kind regards,

PLOS ONE Editorial Office Staff

on behalf of

Dr. hab. Przemysław Mroczek

Academic Editor

PLOS ONE

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

S1 Appendix. Additional information regarding laboratory procedures and methodology.

(PDF)

pone.0330252.s001.pdf^{(85.6KB, pdf)}

S2 Appendix. Basic statistics for the raster variables selected as feature variables.

(PDF)

pone.0330252.s002.pdf^{(55.7KB, pdf)}

S1 Fig. Permutation-based variable importance for all the experiments.

(PDF)

pone.0330252.s003.pdf^{(426.2KB, pdf)}

S2 Fig. Accumulated Local Effect (ALE) plots for all the experiments.

(PDF)

pone.0330252.s004.pdf^{(1.3MB, pdf)}

Attachment

Submitted filename: PONE-D-25-04924 Response_to_Editor.docx

pone.0330252.s005.docx^{(47KB, docx)}

Attachment

Submitted filename: Response to Reviewers.docx

pone.0330252.s006.docx^{(48.9KB, docx)}

Attachment

Submitted filename: Response_to_Reviewers_auresp_2.docx

pone.0330252.s007.docx^{(30.8KB, docx)}

Data Availability Statement

[pone.0330252.ref001] 1.FAO, ITPS. Status of the World’s Soil Resources (SWSR) – Main Report. Rome: Food and Agriculture Organization of the United Nations, Intergovernmental Technical Panel on Soils. 2015. [Google Scholar]

[pone.0330252.ref002] 2.Giardino JR, Houser C. Introduction to the Critical Zone. Developments in Earth Surface Processes. Elsevier. 2015. 1–13. doi: 10.1016/b978-0-444-63369-9.00001-x [DOI] [Google Scholar]

[pone.0330252.ref003] 3.Banwart SA, Bernasconi SM, Blum WEH, de Souza DM, Chabaux F, Duffy C, et al. Soil Functions in Earth’s Critical Zone. Advances in Agronomy. Elsevier. 2017. 1–27. doi: 10.1016/bs.agron.2016.11.001 [DOI] [Google Scholar]

[pone.0330252.ref004] 4.Lal R. Restoring Soil Quality to Mitigate Soil Degradation. Sustainability. 2015;7(5):5875–95. doi: 10.3390/su7055875 [DOI] [Google Scholar]

[pone.0330252.ref005] 5.Montgomery DR. Soil erosion and agricultural sustainability. Proc Natl Acad Sci U S A. 2007;104(33):13268–72. doi: 10.1073/pnas.0611508104 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0330252.ref006] 6.Borrelli P, Robinson DA, Fleischer LR, Lugato E, Ballabio C, Alewell C, et al. An assessment of the global impact of 21st century land use change on soil erosion. Nat Commun. 2017;8(1):2013. doi: 10.1038/s41467-017-02142-7 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0330252.ref007] 7.FAO. Outcome document of the Global Symposium on Soil Erosion. Rome: Food and Agriculture Organization of the United Nations. 2019. [Google Scholar]

[pone.0330252.ref008] 8.Thornes JB. The evolving context of Mediterranean desertification. In: Geeson NA, Brandt CJ, Thornes JB. Mediterranean desertification: A mosaic of processes and responses. Chichester: John Wiley. 2002. 5–11. [Google Scholar]

[pone.0330252.ref009] 9.García-Ruiz JM, Nadal-Romero E, Lana-Renault N, Beguería S. Erosion in Mediterranean landscapes: Changes and future challenges. Geomorphology. 2013;198:20–36. doi: 10.1016/j.geomorph.2013.05.023 [DOI] [Google Scholar]

[pone.0330252.ref010] 10.Borrelli P, Panagos P, Alewell C, Ballabio C, de Oliveira Fagundes H, Haregeweyn N, et al. Policy implications of multiple concurrent soil erosion processes in European farmland. Nat Sustain. 2022;6(1):103–12. doi: 10.1038/s41893-022-00988-4 [DOI] [Google Scholar]

[pone.0330252.ref011] 11.Lal R. Degradation and resilience of soils. Phil Trans R Soc Lond B. 1997;352(1356):997–1010. doi: 10.1098/rstb.1997.0078 [DOI] [Google Scholar]

[pone.0330252.ref012] 12.Viñals E, Maneja R, Rufí-Salís M, Martí M, Puy N. Reviewing social-ecological resilience for agroforestry systems under climate change conditions. Sci Total Environ. 2023;869:161763. doi: 10.1016/j.scitotenv.2023.161763 [DOI] [PubMed] [Google Scholar]

[pone.0330252.ref013] 13.Karlen DL, Andrews SS, Doran JW. Soil quality: Current concepts and applications. Advances in Agronomy. Elsevier. 2001. 1–40. doi: 10.1016/s0065-2113(01)74029-1 [DOI] [Google Scholar]

[pone.0330252.ref014] 14.Reeves DW. The role of soil organic matter in maintaining soil quality in continuous cropping systems. Soil and Tillage Research. 1997;43(1–2):131–67. doi: 10.1016/s0167-1987(97)00038-x [DOI] [Google Scholar]

[pone.0330252.ref015] 15.Wander M. Soil organic matter fractions and their relevance to soil function. In: Magdoff F, Weil RR. Soil organic matter in sustainable agriculture. Boca Raton: CRC Press. 2024. 67–102. [Google Scholar]

[pone.0330252.ref016] 16.Murphy BW. Impact of soil organic matter on soil properties—a review with emphasis on Australian soils. Soil Res. 2015;53(6):605. doi: 10.1071/sr14246 [DOI] [Google Scholar]

[pone.0330252.ref017] 17.Olsson L, Barbosa H, Bhadwal S, Cowie A, Delusca K, Flores-Renteria D, et al. Land degradation. In: Shukla PR, Skea J, Calvo Buendia E, Masson-Delmotte V, Pörtner H-O, Roberts DC, et al. Climate change and land: An IPCC special report on climate change, desertification, land degradation, sustainable land management, food security, and greenhouse gas fluxes in terrestrial ecosystems. Cambridge: Cambridge University Press; 2019. doi: 10.1017/9781009157988.006 [DOI] [Google Scholar]

[pone.0330252.ref018] 18.Conforti M, Buttafuoco G, Leone AP, Aucelli PPC, Robustelli G, Scarciglia F. Studying the relationship between water-induced soil erosion and soil organic matter using Vis–NIR spectroscopy and geomorphological analysis: A case study in southern Italy. CATENA. 2013;110:44–58. doi: 10.1016/j.catena.2013.06.013 [DOI] [Google Scholar]

[pone.0330252.ref019] 19.Obalum SE, Chibuike GU, Peth S, Ouyang Y. Soil organic matter as sole indicator of soil degradation. Environ Monit Assess. 2017;189(4):176. doi: 10.1007/s10661-017-5881-y [DOI] [PubMed] [Google Scholar]

[pone.0330252.ref020] 20.Panagos P, Meusburger K, Ballabio C, Borrelli P, Alewell C. Soil erodibility in Europe: a high-resolution dataset based on LUCAS. Sci Total Environ. 2014;479–480:189–200. doi: 10.1016/j.scitotenv.2014.02.010 [DOI] [PubMed] [Google Scholar]

[pone.0330252.ref021] 21.Bettoni M, Maerker M, Sacchi R, Bosino A, Conedera M, Simoncelli L, et al. What makes soil landscape robust? Landscape sensitivity towards land use changes in a Swiss southern Alpine valley. Sci Total Environ. 2023;858(Pt 2):159779. doi: 10.1016/j.scitotenv.2022.159779 [DOI] [PubMed] [Google Scholar]

[pone.0330252.ref022] 22.Capriel P. Hydrophobicity of organic matter in arable soils: influence of management. European J Soil Science. 1997;48(3):457–62. doi: 10.1111/j.1365-2389.1997.tb00211.x [DOI] [Google Scholar]

[pone.0330252.ref023] 23.Jordán A, Zavala LM, Mataix-Solera J, Doerr SH. Soil water repellency: Origin, assessment and geomorphological consequences. CATENA. 2013;108:1–5. doi: 10.1016/j.catena.2013.05.005 [DOI] [Google Scholar]

[pone.0330252.ref024] 24.Dekker LW, Ritsema CJ. Wetting patterns and moisture variability in water repellent Dutch soils. Journal of Hydrology. 2000;231–232:148–64. doi: 10.1016/s0022-1694(00)00191-8 [DOI] [Google Scholar]

[pone.0330252.ref025] 25.Doerr SH, Shakesby RA, Walsh RPD. Soil water repellency: its causes, characteristics and hydro-geomorphological significance. Earth-Science Reviews. 2000;51(1–4):33–65. doi: 10.1016/s0012-8252(00)00011-8 [DOI] [Google Scholar]

[pone.0330252.ref026] 26.Shakesby RA, Doerr SH, Walsh RPD. The erosional impact of soil hydrophobicity: current problems and future research directions. Journal of Hydrology. 2000;231–232:178–91. doi: 10.1016/s0022-1694(00)00193-1 [DOI] [Google Scholar]

[pone.0330252.ref027] 27.Fox DM, Darboux F, Carrega P. Effects of fire-induced water repellency on soil aggregate stability, splash erosion, and saturated hydraulic conductivity for different size fractions. Hydrol Process. 2007;21:2377–84. doi: 10.1002/hyp.675 [DOI] [Google Scholar]

[pone.0330252.ref028] 28.Bettoni M, Maerker M, Bosino A, Conedera M, Simoncelli L, Vogel S. Land use effects on surface runoff and soil erosion in a southern Alpine valley. Geoderma. 2023;435:116505. doi: 10.1016/j.geoderma.2023.116505 [DOI] [PubMed] [Google Scholar]

[pone.0330252.ref029] 29.Cerdà A, Lavee H, Romero‐Díaz A, Hooke J, Montanarella L. Preface. Land Degrad Dev. 2010;21(2):71–4. doi: 10.1002/ldr.968 [DOI] [Google Scholar]

[pone.0330252.ref030] 30.Piccolo A, Mbagwu JSC. Role of Hydrophobic Components of Soil Organic Matter in Soil Aggregate Stability. Soil Science Soc of Amer J. 1999;63(6):1801–10. doi: 10.2136/sssaj1999.6361801x [DOI] [Google Scholar]

[pone.0330252.ref031] 31.Mataix-Solera J, Doerr SH. Hydrophobicity and aggregate stability in calcareous topsoils from fire-affected pine forests in southeastern Spain. Geoderma. 2004;118(1–2):77–88. doi: 10.1016/s0016-7061(03)00185-x [DOI] [Google Scholar]

[pone.0330252.ref032] 32.Barthès B, Roose E. Aggregate stability as an indicator of soil susceptibility to runoff and erosion; validation at several levels. CATENA. 2002;47(2):133–49. doi: 10.1016/s0341-8162(01)00180-1 [DOI] [Google Scholar]

[pone.0330252.ref033] 33.Mirbabaei SM, Shahrestani MS, Zolfaghari A, Abkenar KT. Relationship between soil water repellency and some of soil properties in northern Iran. CATENA. 2013;108:26–34. doi: 10.1016/j.catena.2013.02.013 [DOI] [Google Scholar]

[pone.0330252.ref034] 34.Nciizah AD, Wakindiki IIC. Physical indicators of soil erosion, aggregate stability and erodibility. Archives of Agronomy and Soil Science. 2015;61(6):827–42. doi: 10.1080/03650340.2014.956660 [DOI] [Google Scholar]

[pone.0330252.ref035] 35.FAO. Soil erosion: the greatest challenge to sustainable soil management. Rome: Food and Agriculture Organization of the United Nations; 2019. [Google Scholar]

[pone.0330252.ref036] 36.Celi L, Schnitzer M, Nègre M. Analysis of carboxyl groups in soil humic acids by a wet chemical method, fourier-transform infrared spectrophotometry, and solution-state carbon-13 nuclear magnetic resonance. a comparative study. Soil Science. 1997;162(3):189–97. doi: 10.1097/00010694-199703000-00004 [DOI] [Google Scholar]

[pone.0330252.ref037] 37.Tipping E. Cation Binding by Humic Substances. Cambridge University Press. 2002. doi: 10.1017/cbo9780511535598 [DOI] [Google Scholar]

[pone.0330252.ref038] 38.McKenzie N, Jacquier D, Isbell R, Brown K. Australian soils and landscapes—An illustrated compendium. Melbourne: CSIRO Publishing. 2004. [Google Scholar]

[pone.0330252.ref039] 39.Kaiser M, Ellerbrock RH, Gerke HH. Cation Exchange Capacity and Composition of Soluble Soil Organic Matter Fractions. Soil Science Soc of Amer J. 2008;72(5):1278–85. doi: 10.2136/sssaj2007.0340 [DOI] [Google Scholar]

[pone.0330252.ref040] 40.Solly EF, Weber V, Zimmermann S, Walthert L, Hagedorn F, Schmidt MWI. A Critical Evaluation of the Relationship Between the Effective Cation Exchange Capacity and Soil Organic Carbon Content in Swiss Forest Soils. Front For Glob Change. 2020;3:98. doi: 10.3389/ffgc.2020.00098 [DOI] [Google Scholar]

[pone.0330252.ref041] 41.Zheng H, Miao C, Huntingford C, Tarolli P, Li D, Panagos P, et al. The Impacts of Erosion on the Carbon Cycle. Reviews of Geophysics. 2025;63(1):e2023RG000829. doi: 10.1029/2023rg000829 [DOI] [Google Scholar]

[pone.0330252.ref042] 42.Shang P, Lu Y, Du Y, Jaffé R, Findlay RH, Wynn A. Climatic and watershed controls of dissolved organic matter variation in streams across a gradient of agricultural land use. Sci Total Environ. 2018;612:1442–53. doi: 10.1016/j.scitotenv.2017.08.322 [DOI] [PubMed] [Google Scholar]

[pone.0330252.ref043] 43.Tang J, Wang W, Yang L, Cao C, Li X. Variation in quantity and chemical composition of soil dissolved organic matter in a peri-urban critical zone observatory watershed in Eastern China. Sci Total Environ. 2019;688:622–31. doi: 10.1016/j.scitotenv.2019.06.270 [DOI] [PubMed] [Google Scholar]

[pone.0330252.ref044] 44.Zhao C, Wang C-C, Li J-Q, Wang C-Y, Wang P, Pei Z-J. Dissolved organic matter in urban stormwater runoff at three typical regions in Beijing: chemical composition, structural characterization and source identification. RSC Adv. 2015;5(90):73490–500. doi: 10.1039/c5ra14993b [DOI] [Google Scholar]

[pone.0330252.ref045] 45.Zhou Y, Liu M, Zhou L, Jang K-S, Xu H, Shi K, et al. Rainstorm events shift the molecular composition and export of dissolved organic matter in a large drinking water reservoir in China: High frequency buoys and field observations. Water Res. 2020;187:116471. doi: 10.1016/j.watres.2020.116471 [DOI] [PubMed] [Google Scholar]

[pone.0330252.ref046] 46.Ferreira CSS, Seifollahi-Aghmiuni S, Destouni G, Ghajarnia N, Kalantari Z. Soil degradation in the European Mediterranean region: Processes, status and consequences. Sci Total Environ. 2022;805:150106. doi: 10.1016/j.scitotenv.2021.150106 [DOI] [PubMed] [Google Scholar]

[pone.0330252.ref047] 47.La Licata M, Bosino A, Bettoni M, Maerker M. Assessing landscape features and geomorphic processes influencing sediment dynamics in a geomorphologically highly active Mediterranean agroecosystem: The upper Val d’Arda case study (Northern Apennines, Italy). Geomorphology. 2023;433:108724. doi: 10.1016/j.geomorph.2023.108724 [DOI] [Google Scholar]

[pone.0330252.ref048] 48.Johnston CT, Aochi YO. Fourier transform infrared and Raman spectroscopy. In: Sparks DL, Page AL, Helmke PA, Loeppert RH. Methods of soil analysis, Part 3: Chemical methods. Madison: Soil Science Society of America, American Society of Agronomy. 1996. 269–321. [Google Scholar]

[pone.0330252.ref049] 49.Margenot AJ, Parikh SJ, Calderón FJ. Fourier‐transform infrared spectroscopy for soil organic matter analysis. Soil Science Soc of Amer J. 2023;87(6):1503–28. doi: 10.1002/saj2.20583 [DOI] [Google Scholar]

[pone.0330252.ref050] 50.Ellerbrock RH, Gerke HH. Explaining soil organic matter composition based on associations between OM and polyvalent cations. J Plant Nutr Soil Sci. 2018;181(5):721–36. doi: 10.1002/jpln.201800093 [DOI] [Google Scholar]

[pone.0330252.ref051] 51.Ellerbrock RH, Gerke HH, Bachmann J, Goebel M-O. Composition of Organic Matter Fractions for Explaining Wettability of Three Forest Soils. Soil Science Society of America Journal. 2005;69(1):57. doi: 10.2136/sssaj2005.0057 [DOI] [Google Scholar]

[pone.0330252.ref052] 52.Leue M, Gerke HH, Ellerbrock RH. Millimetre‐scale distribution of organic matter composition at intact biopore and crack surfaces. European J Soil Science. 2013;64(6):757–69. doi: 10.1111/ejss.12098 [DOI] [Google Scholar]

[pone.0330252.ref053] 53.Reig FB, Adelantado JVG, Moya Moreno MCM. FTIR quantitative analysis of calcium carbonate (calcite) and silica (quartz) mixtures using the constant ratio method. Application to geological samples. Talanta. 2002;58(4):811–21. doi: 10.1016/s0039-9140(02)00372-7 [DOI] [PubMed] [Google Scholar]

[pone.0330252.ref054] 54.Alvarez-Puebla RA, Garrido JJ, Valenzuela-Calahorro C, G. Goulet PaulJ. Retention and induced aggregation of Co(II) on a humic substance: sorption isotherms, infrared absorption, and molecular modeling. Surface Science. 2005;575(1–2):136–46. doi: 10.1016/j.susc.2004.11.027 [DOI] [Google Scholar]

[pone.0330252.ref055] 55.Ellerbrock RH, Gerke HH. FTIR spectral band shifts explained by OM–cation interactions. J Plant Nutr Soil Sci. 2021;184(3):388–97. doi: 10.1002/jpln.202100056 [DOI] [Google Scholar]

[pone.0330252.ref056] 56.Breiman L. Random Forests. Machine Learning. 2001;45(1):5–32. doi: 10.1023/a:1010933404324 [DOI] [Google Scholar]

[pone.0330252.ref057] 57.Adeniyi OD, Brenning A, Bernini A, Brenna S, Maerker M. Digital Mapping of Soil Properties Using Ensemble Machine Learning Approaches in an Agricultural Lowland Area of Lombardy, Italy. Land. 2023;12(2):494. doi: 10.3390/land12020494 [DOI] [Google Scholar]

[pone.0330252.ref058] 58.Simoni A, Ponza A, Picotti V, Berti M, Dinelli E. Earthflow sediment production and Holocene sediment record in a large Apennine catchment. Geomorphology. 2013;188:42–53. doi: 10.1016/j.geomorph.2012.12.006 [DOI] [Google Scholar]

[pone.0330252.ref059] 59.Servizio Geologico d’Italia. Carta geologica d’Italia alla scala 1:50.000, foglio 198 - Bardi. Roma: Istituto Poligrafico e Zecca dello Stato. 1999. [Google Scholar]

[pone.0330252.ref060] 60.Martini A, Zanzucchi G. Note illustrative della Carta Geologica d’Italia alla scala 1:50.000, Foglio 198 - Bardi. Roma: Servizio Geologico d’Italia. 2000. [Google Scholar]

[pone.0330252.ref061] 61.Pellegrini L, Vercesi PL. Landscapes and Landforms Driven by Geological Structures in the Northwestern Apennines. World Geomorphological Landscapes. Springer International Publishing. 2017. 203–13. doi: 10.1007/978-3-319-26194-2_17 [DOI] [Google Scholar]

[pone.0330252.ref062] 62.La Licata M, Bosino A, Sadeghi SH, De Amicis M, Mandarino A, Terret A, et al. HOTSED: A new integrated model for assessing potential hotspots of sediment sources and related sediment dynamics at watershed scale. International Soil and Water Conservation Research. 2025;13(1):80–101. doi: 10.1016/j.iswcr.2024.06.002 [DOI] [Google Scholar]

[pone.0330252.ref063] 63.Patro ER, De Michele C, Granata G, Biagini C. Assessment of current reservoir sedimentation rate and storage capacity loss: An Italian overview. J Environ Manage. 2022;320:115826. doi: 10.1016/j.jenvman.2022.115826 [DOI] [PubMed] [Google Scholar]

[pone.0330252.ref064] 64.Staffilani F, Bonapasta D, Marucci FE. Carta dell’erosione idrica attuale della Regione Emilia-Romagna. Servizio Geologico, Sismico e Suoli, Regione Emilia-Romagna. 2019. [Google Scholar]

[pone.0330252.ref065] 65.Regione Emilia-Romagna. I suoli dell’Emilia-Romagna. Note illustrative. Servizio Cartografico – Ufficio Pedologico. 1994. [Google Scholar]

[pone.0330252.ref066] 66.IUSS Working Group WRB. World Reference Base for Soil Resources 2014, update 2015: International soil classification system for naming soils and creating legends for soil maps. Rome: FAO. 2015. [Google Scholar]

[pone.0330252.ref067] 67.Conrad O, Bechtel B, Bock M, Dietrich H, Fischer E, Gerlitz L, et al. System for Automated Geoscientific Analyses (SAGA) v. 2.1.4. Geosci Model Dev. 2015;8(7):1991–2007. doi: 10.5194/gmd-8-1991-2015 [DOI] [Google Scholar]

[pone.0330252.ref068] 68.Regione Emilia-Romagna. Modello Digitale del Terreno 5x5, Edizione 2014. Archivio Cartografico. 2019. https://www.geoportale.regione.emilia-romagna.it/catalogo/dati-cartografici/altimetria/layer-2 [Google Scholar]

[pone.0330252.ref069] 69.Lidberg W, Nilsson M, Lundmark T, Ågren AM. Evaluating preprocessing methods of digital elevation models for hydrological modelling. Hydrological Processes. 2017;31(26):4660–8. doi: 10.1002/hyp.11385 [DOI] [Google Scholar]

[pone.0330252.ref070] 70.Wang L, Liu H. An efficient method for identifying and filling surface depressions in digital elevation models for hydrologic analysis and modelling. International Journal of Geographical Information Science. 2006;20(2):193–213. doi: 10.1080/13658810500433453 [DOI] [Google Scholar]

[pone.0330252.ref071] 71.Freeman TG. Calculating catchment area with divergent flow based on a regular grid. Computers & Geosciences. 1991;17(3):413–22. doi: 10.1016/0098-3004(91)90048-i [DOI] [Google Scholar]

[pone.0330252.ref072] 72.Quinn P, Beven K, Chevallier P, Planchon O. The prediction of hillslope flow paths for distributed hydrological modelling using digital terrain models. Hydrological Processes. 1991;5(1):59–79. doi: 10.1002/hyp.3360050106 [DOI] [Google Scholar]

[pone.0330252.ref073] 73.Quantitative analysis of watershed geomorphology. Eos Trans AGU. 1957;38(6):913–20. doi: 10.1029/tr038i006p00913 [DOI] [Google Scholar]

[pone.0330252.ref074] 74.Collins AL, Pulley S, Foster IDL, Gellis A, Porto P, Horowitz AJ. Sediment source fingerprinting as an aid to catchment management: A review of the current state of knowledge and a methodological decision-tree for end-users. J Environ Manage. 2017;194:86–108. doi: 10.1016/j.jenvman.2016.09.075 [DOI] [PubMed] [Google Scholar]

[pone.0330252.ref075] 75.Valente M, Zanellati M, Facci G, Zanna N, Petrone E, Moretti E, et al. Health System Response to the 2023 Floods in Emilia-Romagna, Italy: A Field Report. Prehosp Disaster Med. 2023;38(6):813–7. doi: 10.1017/S1049023X23006404 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0330252.ref076] 76.Arrighi C, Domeneghetti A. Brief communication: On the environmental impacts of the 2023 floods in Emilia-Romagna (Italy). Nat Hazards Earth Syst Sci. 2024;24(2):673–9. doi: 10.5194/nhess-24-673-2024 [DOI] [Google Scholar]

[pone.0330252.ref077] 77.Collins AL, Blackwell M, Boeckx P, Chivers C-A, Emelko M, Evrard O, et al. Sediment source fingerprinting: benchmarking recent outputs, remaining challenges and emerging themes. J Soils Sediments. 2020;20(12):4160–93. doi: 10.1007/s11368-020-02755-4 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0330252.ref078] 78.Koiter AJ, Owens PN, Petticrew EL, Lobb DA. The behavioural characteristics of sediment properties and their implications for sediment fingerprinting as an approach for identifying sediment sources in river basins. Earth-Science Reviews. 2013;125:24–42. doi: 10.1016/j.earscirev.2013.05.009 [DOI] [Google Scholar]

[pone.0330252.ref079] 79.Owens PN, Blake WH, Gaspar L, Gateuille D, Koiter AJ, Lobb DA, et al. Fingerprinting and tracing the sources of soils and sediments: Earth and ocean science, geoarchaeological, forensic, and human health applications. Earth-Science Reviews. 2016;162:1–23. doi: 10.1016/j.earscirev.2016.08.012 [DOI] [Google Scholar]

[pone.0330252.ref080] 80.Lizaga I, Gaspar L, Blake WH, Latorre B, Navas A. Fingerprinting changes of source apportionments from mixed land uses in stream sediments before and after an exceptional rainstorm event. Geomorphology. 2019;341:216–29. doi: 10.1016/j.geomorph.2019.05.015 [DOI] [Google Scholar]

[pone.0330252.ref081] 81.Lizaga I, Quijano L, Gaspar L, Ramos MC, Navas A. Linking land use changes to variation in soil properties in a Mediterranean mountain agroecosystem. CATENA. 2019;172:516–27. doi: 10.1016/j.catena.2018.09.019 [DOI] [Google Scholar]

[pone.0330252.ref082] 82.Ellerbrock RH, Höhn A, Rogasik J. Functional analysis of soil organic matter as affected by long‐term manurial treatment. European J Soil Science. 1999;50(1):65–71. doi: 10.1046/j.1365-2389.1999.00206.x [DOI] [Google Scholar]

[pone.0330252.ref083] 83.Ellerbrock RH, Gerke HH. Characterization of Organic Matter Composition of Soil and Flow Path Surfaces Based on Physicochemical Principles—A Review. Advances in Agronomy. Elsevier. 2013. 117–77. doi: 10.1016/b978-0-12-407685-3.00003-7 [DOI] [Google Scholar]

[pone.0330252.ref084] 84.Ellerbrock RH, Niessner D, Deumlich D, Puppe D, Gerke HH. Infrared-based analysis of organic matter composition in liquid and solid runoff fractions collected during a single erosion event. Soil and Tillage Research. 2024;235:105901. doi: 10.1016/j.still.2023.105901 [DOI] [Google Scholar]

[pone.0330252.ref085] 85.Capriel P, Beck T, Borchert H, Gronholz J, Zachmann G. Hydrophobicity of the organic matter in arable soils. Soil Biology and Biochemistry. 1995;27(11):1453–8. doi: 10.1016/0038-0717(95)00068-p [DOI] [Google Scholar]

[pone.0330252.ref086] 86.Ellerbrock RH, Gerke HH, Böhm C. In Situ DRIFT Characterization of Organic Matter Composition on Soil Structural Surfaces. Soil Science Soc of Amer J. 2009;73(2):531–40. doi: 10.2136/sssaj2008.0103 [DOI] [Google Scholar]

[pone.0330252.ref087] 87.Ellerbrock RH, Gerke HH, Deumlich D. Soil organic matter composition along a slope in an erosion-affected arable landscape in North East Germany. Soil and Tillage Research. 2016;156:209–18. doi: 10.1016/j.still.2015.08.014 [DOI] [Google Scholar]

[pone.0330252.ref088] 88.Berthomieu C, Hienerwadel R. Fourier transform infrared (FTIR) spectroscopy. Photosynth Res. 2009;101(2–3):157–70. doi: 10.1007/s11120-009-9439-x [DOI] [PubMed] [Google Scholar]

[pone.0330252.ref089] 89.Dudev T, Lim C. The effect of metal binding on the characteristic infrared band intensities of ligands of biological interest. Journal of Molecular Structure. 2012;1009:83–8. doi: 10.1016/j.molstruc.2011.09.007 [DOI] [Google Scholar]

[pone.0330252.ref090] 90.Tinti A, Tugnoli V, Bonora S, Francioso O. Recent applications of vibrational mid-Infrared (IR) spectroscopy for studying soil components: a review. j cent eur agric. 2015;16(1):1–22. doi: 10.5513/jcea01/16.1.1535 [DOI] [Google Scholar]

[pone.0330252.ref091] 91.Jenks GF. The data model concept in statistical mapping. International Yearbook of Cartography. 1967;7:186–90. [Google Scholar]

[pone.0330252.ref092] 92.Boehner J, Selige T. Spatial prediction of soil attributes using terrain analysis and climate regionalisation. In: Boehner J, McCloy KR, Strobl J. SAGA - Analysis and modelling applications. Goettingen: Department of Geography, University of Goettingen. 2006. 13–28. [Google Scholar]

[pone.0330252.ref093] 93.Olaya V, Conrad O. Chapter 12 Geomorphometry in SAGA. Developments in Soil Science. Elsevier. 2009. 293–308. doi: 10.1016/s0166-2481(08)00012-3 [DOI] [Google Scholar]

[pone.0330252.ref094] 94.Moore ID, Grayson RB, Ladson AR. Digital terrain modelling: A review of hydrological, geomorphological, and biological applications. Hydrological Processes. 1991;5(1):3–30. doi: 10.1002/hyp.3360050103 [DOI] [Google Scholar]

[pone.0330252.ref095] 95.Wilson JP, Gallant JC. Primary topographic attributes. Terrain analysis: Principles and applications. New York: John Wiley & Sons. 2000. 51–85. [Google Scholar]

[pone.0330252.ref096] 96.Gruber S, Peckham S. Chapter 7 Land-Surface Parameters and Objects in Hydrology. Developments in Soil Science. Elsevier. 2009. 171–94. doi: 10.1016/s0166-2481(08)00007-x [DOI] [Google Scholar]

[pone.0330252.ref097] 97.Olaya V. Chapter 6 Basic Land-Surface Parameters. Developments in Soil Science. Elsevier. 2009. 141–69. doi: 10.1016/s0166-2481(08)00006-8 [DOI] [Google Scholar]

[pone.0330252.ref098] 98.Blaga L. Aspects regarding the significance of the curvature types and values in the studies of geomorphometry assisted by GIS. Analele Universităţii din Oradea, Seria Geografie. 2012;22(2):327–37. [Google Scholar]

[pone.0330252.ref099] 99.Wilson JP. Digital terrain modeling. Geomorphology. 2012;137(1):107–21. doi: 10.1016/j.geomorph.2011.03.012 [DOI] [Google Scholar]

[pone.0330252.ref100] 100.Wilson JP, Bishop MP. 3.7 Geomorphometry. Treatise on Geomorphology. Elsevier. 2013. 162–86. doi: 10.1016/b978-0-12-374739-6.00049-x [DOI] [Google Scholar]

[pone.0330252.ref101] 101.Florinsky IV. An illustrated introduction to general geomorphometry. Progress in Physical Geography: Earth and Environment. 2017;41(6):723–52. doi: 10.1177/0309133317733667 [DOI] [Google Scholar]

[pone.0330252.ref102] 102.Maerker M, Schillaci C, Kropáček J. Morphometric terrain analysis to explore present day geohazards and paleolandscape forms and features in the surroundings of the Melka Kunture prehistoric site, Upper Awash Valley, Central Ethiopia. AUC GEOGRAPHICA. 2018;53(1):10–9. doi: 10.14712/23361980.2018.2 [DOI] [Google Scholar]

[pone.0330252.ref103] 103.Zevenbergen LW, Thorne CR. Quantitative analysis of land surface topography. Earth Surf Processes Landf. 1987;12(1):47–56. doi: 10.1002/esp.3290120107 [DOI] [Google Scholar]

[pone.0330252.ref104] 104.Hjerdt KN, McDonnell JJ, Seibert J, Rodhe A. A new topographic index to quantify downslope controls on local drainage. Water Resources Research. 2004;40(5):W05602. doi: 10.1029/2004wr003130 [DOI] [Google Scholar]

[pone.0330252.ref105] 105.Riley SJ, De Gloria SD, Elliot R. A terrain ruggedness that quantifies topographic heterogeneity. Inter-mountain Journal of Science. 1999;5(1–4):23–7. [Google Scholar]

[pone.0330252.ref106] 106.Ungaro F, Tarocco P, Aprea A. Carta del carbonio organico immagazzinato nei suoli della regione Emilia-Romagna, strato 0-30 cm, scala 1:50,000. Terza ed. Bologna: Regione Emilia-Romagna, Direzione Generale cura del territorio e dell’ambiente, Servizio Geologico, Sismico e dei Suoli. 2023. [Google Scholar]

[pone.0330252.ref107] 107.Ungaro F, Tarocco P, Aprea A. Carta del contenuto percentuale di carbonio organico nei suoli della regione Emilia-Romagna, strato 0-30 cm, 1:50,000. Terza ed. Bologna: Regione Emilia-Romagna, Direzione Generale cura del territorio e dell’ambiente, Servizio Geologico, Sismico e dei Suoli. 2023. [Google Scholar]

[pone.0330252.ref108] 108.Ungaro F, Tarocco P, Aprea A. Carta della tessitura dei suoli della Regione Emilia-Romagna, strato 0-30 cm, 1:50,000. Seconda edizione. Note illustrative. Bologna: Regione Emilia-Romagna, Direzione Generale cura del territorio e dell’ambiente, Servizio Geologico, Sismico e dei Suoli. 2023. [Google Scholar]

[pone.0330252.ref109] 109.Ungaro F, Tarocco P, Aprea A. Carta del pH dei suoli della Regione Emilia-Romagna, strato 0-30 cm, 1:50,000. Seconda edizione. Note illustrative. Bologna: Regione Emilia-Romagna, Direzione Generale cura del territorio e dell’ambiente, Servizio Geologico, Sismico e dei Suoli. 2023. [Google Scholar]

[pone.0330252.ref110] 110.SGSS - Servizio Geologico, Sismico e dei Suoli. Banca dati geologica, 1:10.000 – Unità geologiche – 10k. Geoportale Regione Emilia-Romagna. 2006. https://geoportale.regione.emilia-romagna.it/catalogo/dati-cartografici/informazioni-geoscientifiche/geologia/banca-dati-geologica-1-10.000/layer-11 [Google Scholar]

[pone.0330252.ref111] 111.Regione E-R. Coperture vettoriali uso del suolo di dettaglio 2017 – Edizione 2020. Settore innovazione digitale, dati, tecnologia e polo archivistico. 2020. https://geoportale.regione.emilia-romagna.it/catalogo/dati-cartografici/pianificazione-e-catasto/uso-del-suolo/layer-9 [Google Scholar]

[pone.0330252.ref112] 112.O’Callaghan JF, Mark DM. The extraction of drainage networks from digital elevation data. Computer Vision, Graphics, and Image Processing. 1984;28(3):323–44. doi: 10.1016/s0734-189x(84)80011-0 [DOI] [Google Scholar]

[pone.0330252.ref113] 113.La Licata M, Bosino A, De Amicis M, Terret A, Maerker M. A GIS-based multiscale mapping framework to assess and visualize complex and polygenetic geomorphic systems – A novel ‘Geomorphic Entities’ approach. Journal of Maps. 2025;21(1):2437256. doi: 10.1080/17445647.2024.2437256 [DOI] [Google Scholar]

[pone.0330252.ref114] 114.La Licata M, Bosino A, Sadeghi SH, Maerker M. Assessing hotspots of sediment sources and related sediment dynamics through the integration of geomorphological data, sediment connectivity and sediment transport modelling – The HOTSED model. EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024. 2024; EGU24-4101. doi: 10.5194/egusphere-egu24-4101 [DOI] [Google Scholar]

[pone.0330252.ref115] 115.Romanski P, Kotthoff L, Schratz P. Package ‘FSelector’: Selecting Attributes. R package version 0.34. 2023. doi: 10.32614/CRAN.package.FSelector [DOI] [Google Scholar]

[pone.0330252.ref116] 116.Kuhn M. Building Predictive Models inRUsing thecaretPackage. J Stat Soft. 2008;28(5). doi: 10.18637/jss.v028.i05 [DOI] [Google Scholar]

[pone.0330252.ref117] 117.R Development Core Team. R: A language and environment for statistical computing. Vienna: R Foundation for Statistical Computing. 2016. [Google Scholar]

[pone.0330252.ref118] 118.Arlot S, Celisse A. A survey of cross-validation procedures for model selection. Statist Surv. 2010;4:40–79. doi: 10.1214/09-ss054 [DOI] [Google Scholar]

[pone.0330252.ref119] 119.Gupta HV, Kling H, Yilmaz KK, Martinez GF. Decomposition of the mean squared error and NSE performance criteria: Implications for improving hydrological modelling. Journal of Hydrology. 2009;377(1–2):80–91. doi: 10.1016/j.jhydrol.2009.08.003 [DOI] [Google Scholar]

[pone.0330252.ref120] 120.Hastie T, Tibshirani R, Friedman J. The Elements of Statistical Learning. Springer New York. 2009. doi: 10.1007/978-0-387-84858-7 [DOI] [Google Scholar]

[pone.0330252.ref121] 121.Singh AK, Kumar P, Ali R, Al-Ansari N, Vishwakarma DK, Kushwaha KS, et al. An Integrated Statistical-Machine Learning Approach for Runoff Prediction. Sustainability. 2022;14(13):8209. doi: 10.3390/su14138209 [DOI] [Google Scholar]

[pone.0330252.ref122] 122.Bellon-Maurel V, Fernandez-Ahumada E, Palagos B, Roger J-M, McBratney A. Critical review of chemometric indicators commonly used for assessing the quality of the prediction of soil attributes by NIR spectroscopy. TrAC Trends in Analytical Chemistry. 2010;29(9):1073–81. doi: 10.1016/j.trac.2010.05.006 [DOI] [Google Scholar]

[pone.0330252.ref123] 123.Zambrano-Bigiarini M. Package ‘hydroGOF’: Goodness-of-fit functions for comparison of simulated and observed hydrological time series. 2024. doi: 10.5281/zenodo.839854 [DOI] [Google Scholar]

[pone.0330252.ref124] 124.Molnar C. iml: An R package for Interpretable Machine Learning. JOSS. 2018;3(26):786. doi: 10.21105/joss.00786 [DOI] [Google Scholar]

[pone.0330252.ref125] 125.Molnar C. Interpretable machine learning: A guide for making black box models explainable. 2nd ed. 2024. [Google Scholar]

[pone.0330252.ref126] 126.Berti M, Pizziolo M, Scaroni M, Generali M, Critelli V, Mulas M, et al. RER2023: the landslide inventory dataset of the May 2023 Emilia-Romagna event. Copernicus GmbH. 2024. doi: 10.5194/essd-2024-407 [DOI] [Google Scholar]

[pone.0330252.ref127] 127.Ferrario MF, Livio F. Rapid Mapping of Landslides Induced by Heavy Rainfall in the Emilia-Romagna (Italy) Region in May 2023. Remote Sensing. 2023;16(1):122. doi: 10.3390/rs16010122 [DOI] [Google Scholar]

[pone.0330252.ref128] 128.Cremonini L, Randi P, Fazzini M, Nardino M, Rossi F, Georgiadis T. Causes and Impacts of Flood Events in Emilia-Romagna (Italy) in May 2023. Land. 2024;13(11):1800. doi: 10.3390/land13111800 [DOI] [Google Scholar]

[pone.0330252.ref129] 129.Krull ES, Skjemstad JO, Baldock JA. Functions of soil organic matter and the effect on soil properties. Canberra: CSIRO Land & Water. 2004. [Google Scholar]

[pone.0330252.ref130] 130.Lal R. Soil erosion and carbon dynamics. Soil and Tillage Research. 2005;81(2):137–42. doi: 10.1016/j.still.2004.09.002 [DOI] [Google Scholar]

[pone.0330252.ref131] 131.Pardini G, Gispert M, Dunjó G. Runoff erosion and nutrient depletion in five Mediterranean soils of NE Spain under different land use. Sci Total Environ. 2003;309(1–3):213–24. doi: 10.1016/S0048-9697(03)00007-X [DOI] [PubMed] [Google Scholar]

[pone.0330252.ref132] 132.Pardini G, Gispert M, Emran M, Doni S. Rainfall/runoff/erosion relationships and soil properties survey in abandoned shallow soils of NE Spain. J Soils Sediments. 2017;17(2):499–514. doi: 10.1007/s11368-016-1532-0 [DOI] [Google Scholar]

[pone.0330252.ref133] 133.Berhe AA, Kleber M. Erosion, deposition, and the persistence of soil organic matter: mechanistic considerations and problems with terminology. Earth Surf Processes Landf. 2013;38(8):908–12. doi: 10.1002/esp.3408 [DOI] [Google Scholar]

[pone.0330252.ref134] 134.Berhe AA. Decomposition of organic substrates at eroding vs. depositional landform positions. Plant Soil. 2012;350(1–2):261–80. doi: 10.1007/s11104-011-0902-z [DOI] [Google Scholar]

[pone.0330252.ref135] 135.Stevenson FJ. Humus chemistry – Genesis, composition and reactions. 2nd ed. New York: John Wiley and Sons. 1994. [Google Scholar]

[pone.0330252.ref136] 136.Nadeu E, Berhe AA, de Vente J, Boix-Fayos C. Erosion, deposition and replacement of soil organic carbon in Mediterranean catchments: a geomorphological, isotopic and land use change approach. Biogeosciences. 2012;9(3):1099–111. doi: 10.5194/bg-9-1099-2012 [DOI] [Google Scholar]

[pone.0330252.ref137] 137.Ubaldi D, Puppi G, Zanotti AL. Cartografia fitoclimatica dell’Emilia-Romagna in scala 1:500.000. Bologna: Regione Emilia-Romagna. 1996. [Google Scholar]

[pone.0330252.ref138] 138.Hardie MA, Cotching WE, Doyle RB, Lisson S. Influence of climate, water content and leaching on seasonal variations in potential water repellence. Hydrological Processes. 2012;26(13):2041–8. doi: 10.1002/hyp.8312 [DOI] [Google Scholar]

[pone.0330252.ref139] 139.Blanco-Canqui H, Lal R. Soil Erosion Under Forests. Principles of Soil Conservation and Management. Springer Netherlands. 2010. 321–44. doi: 10.1007/978-1-4020-8709-7_12 [DOI] [Google Scholar]

[pone.0330252.ref140] 140.Lützow M v., Kögel‐Knabner I, Ekschmitt K, Matzner E, Guggenberger G, Marschner B, et al. Stabilization of organic matter in temperate soils: mechanisms and their relevance under different soil conditions – a review. European J Soil Science. 2006;57(4):426–45. doi: 10.1111/j.1365-2389.2006.00809.x [DOI] [Google Scholar]

[pone.0330252.ref141] 141.Chilom G, Rice JA. Organo‐Clay Complexes in Soils and Sediments. Biophysico‐Chemical Processes Involving Natural Nonliving Organic Matter in Environmental Systems. Wiley. 2009. 111–45. doi: 10.1002/9780470494950.ch3 [DOI] [Google Scholar]

[pone.0330252.ref142] 142.Kasanin-Grubin M. Badlands and environmental change. Geogr Helv. 2008;63(1):15–25. doi: 10.5194/gh-63-15-2008 [DOI] [Google Scholar]

[pone.0330252.ref143] 143.Bosino A, Pellegrini L, Omran A, Bordoni M, Meisina C, Maerker M. Litho-structure of the Oltrepo Pavese, Northern Apennines (Italy). Journal of Maps. 2019;15(2):382–92. doi: 10.1080/17445647.2019.1604438 [DOI] [Google Scholar]

[pone.0330252.ref144] 144.Bosino A, Szatten DA, Omran A, Crema S, Crozi M, Becker R, et al. Assessment of suspended sediment dynamics in a small ungauged badland catchment in the Northern Apennines (Italy) using an in-situ laser diffraction method. CATENA. 2022;209:105796. doi: 10.1016/j.catena.2021.105796 [DOI] [Google Scholar]

[pone.0330252.ref145] 145.Leue M, Ellerbrock RH, Gerke HH. DRIFT Mapping of Organic Matter Composition at Intact Soil Aggregate Surfaces. Vadose Zone Journal. 2010;9(2):317–24. doi: 10.2136/vzj2009.0101 [DOI] [Google Scholar]

[pone.0330252.ref146] 146.Del Prete M, Bentivenga M, Amato M, Basso F, Tacconi P. Badland erosion processes and their interactions with vegetation: A case study from Pisticci, Basilicata, Southern Italy. Geogr Fis Dinam Quat. 1997;20:147–55. [Google Scholar]

[pone.0330252.ref147] 147.Berhe AA, Harden JW, Torn MS, Harte J. Linking soil organic matter dynamics and erosion‐induced terrestrial carbon sequestration at different landform positions. J Geophys Res. 2008;113(G4). doi: 10.1029/2008jg000751 [DOI] [Google Scholar]

[pone.0330252.ref148] 148.Berhe AA, Harden JW, Torn MS, Kleber M, Burton SD, Harte J. Persistence of soil organic matter in eroding versus depositional landform positions. J Geophys Res. 2012;117(G2). doi: 10.1029/2011jg001790 [DOI] [Google Scholar]

[pone.0330252.ref149] 149.Wohl E. Connectivity in rivers. Progress in Physical Geography: Earth and Environment. 2017;41(3):345–62. doi: 10.1177/0309133317714972 [DOI] [Google Scholar]

[pone.0330252.ref150] 150.Doetterl S, Berhe AA, Heckman K, Lawrence C, Schnecker J, Vargas R, et al. A landscape-scale view of soil organic matter dynamics. Nat Rev Earth Environ. 2025;6(1):67–81. doi: 10.1038/s43017-024-00621-2 [DOI] [Google Scholar]

[pone.0330252.ref151] 151.Ellerbrock RH, Kaiser M. Stability and composition of different soluble soil organic matter fractions–evidence from δ¹³C and FTIR signatures. Geoderma. 2005;128(1–2):28–37. doi: 10.1016/j.geoderma.2004.12.025 [DOI] [Google Scholar]

[pone.0330252.ref152] 152.Dawson LA, Hillier S. Measurement of soil characteristics for forensic applications. Surface & Interface Analysis. 2010;42(5):363–77. doi: 10.1002/sia.3315 [DOI] [Google Scholar]

[pone.0330252.ref153] 153.Sadeghi SH, Singh VP, Kiani-Harchegani M, Asadi H. Analysis of sediment rating loops and particle size distributions to characterize sediment source at mid-sized plot scale. CATENA. 2018;167:221–7. doi: 10.1016/j.catena.2018.05.002 [DOI] [Google Scholar]

[pone.0330252.ref154] 154.Ellerbrock R, Stein M, Schaller J. Comparing amorphous silica, short-range-ordered silicates and silicic acid species by FTIR. Sci Rep. 2022;12(1):11708. doi: 10.1038/s41598-022-15882-4 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0330252.ref155] 155.Ellerbrock RH, Stein M, Schaller J. Comparing silicon mineral species of different crystallinity using Fourier transform infrared spectroscopy. Front Environ Chem. 2024;5:1462678. doi: 10.3389/fenvc.2024.1462678 [DOI] [Google Scholar]

[pone.0330252.ref156] 156.Barbosa LAP, Stein M, Gerke HH, Schaller J. Synergistic effects of organic carbon and silica in preserving structural stability of drying soils. Sci Rep. 2024;14(1):8330. doi: 10.1038/s41598-024-58916-9 [DOI] [PMC free article] [PubMed] [Google Scholar]

PERMALINK

FTIR-derived soil degradation indices and stochastic modelling of organic matter–sediment dynamics in a Mediterranean watershed: A Northern Apennines case study

Manuel La Licata

Odunayo D Adeniyi

Ruth H Ellerbrock

Nisha Bhattarai

Alberto Bosino

Natalie Papke

Jörg Schaller

Michael Maerker

Roles

Abstract

1 Introduction

2 Study area

Fig 1. (a) Geographic outline of the study area in the Emilia-Romagna Region, Northern Italy.

3 Materials and methods

Fig 2. Schematic methodological workflow used in the present study.

3.1 Watershed and channel network delineation

3.2 Sampling of fluvial sediments

3.3 Fourier transform infrared (FTIR) spectroscopy analysis

Fig 3. Identification of the bands A, B, C, and D, within specific wavenumber regions, on a smoothed, baseline-corrected spectrum.

Table 1. Indices for SOM characterization calculated as FTIR band ratios, used in the present study.

3.4 Preparation of environmental variable datasets

Table 2. Terrain variables used in the present study, along with their geomorphological significance. Sources: [92,94–102].

Table 4. LULC types of the upper Val d’Arda-Mignano watershed. Shapefile data derived from Land use database of the Emilia Romagna Region [1:10,000 scale; 111].

3.5 Variables extraction based on different experimental designs

Fig 4. The three experimental designs employed in the present study: (a) ‘Upstream areas Exp_1’, (b) ‘Sources and sinks Exp_2’, (c) ‘Hazard areas Exp_3’.

3.6 Determination of non-linear and geospatial relationships using Random Forest algorithm

4 Results

4.1 FTIR indices for SOM characterization (target variables)

Table 5. FTIR Indices calculated for the 73 fluvial sediment samples in the upper Val d’Arda-Mignano watershed (cfr. Table 1).

Table 6. Spearman’s rank correlation coefficients among the calculated FTIR indices (cfr. Table 1).

Fig 5. Spatial distribution of FTIR indices (Table 1) within the upper Val d’Arda watershed (Fig 1b): (a) A/B, (b) A/D, (c) B/D, (d) C/D.

4.2 Datasets selected as feature variables

Fig 7. Soil properties of the upper Val d’Arda-Mignano watershed: (a) SOC stock, (b) SOC %, (c) Clay %, (d) Silt %, (e) Sand %, (f) Skeleton %, (g) pH, (h) lithological groups, and (i) LULC types.

4.3 Data extraction from feature variables across the experiments

4.4 Random Forest models’ performance evaluation

Table 7. Models’ performance metrics for the three experiments (cfr. Fig 4).

4.5 Feature importance and variable interactions

Fig 8. Permutation-based variable importance for the selected variables in the best-performing experiment for each FTIR index: (a) A/B, Experiment 1; (b) A/D, Experiment 1; (c) B/D, Experiment 3; (d) C/D, Experiment 1.

Fig 9. Accumulated Local Effect (ALE) plots using the RF model for assessing A/B index, Experiment 1.

Fig 12. Accumulated Local Effect (ALE) plots using the RF model for assessing C/D index, Experiment 1.

Fig 10. Accumulated Local Effect (ALE) plots using the RF model for assessing A/D index, Experiment 1.

Fig 11. Accumulated Local Effect (ALE) plots using the RF model for assessing B/D index, Experiment 3.

5 Discussion

5.1 FTIR-derived indices as indicators of soil degradation and erosion processes

5.2 Geospatial relationships between target and feature variables

5.3 Limitations of the methodological approach and improvements

6 Conclusion

Supporting information

Acknowledgments

Abbreviations

Data Availability

Funding Statement

References

Author response to Decision Letter 0

Decision Letter 0

Przemysław Mroczek

Roles

Author response to Decision Letter 1

Decision Letter 1

Przemysław Mroczek

Roles

Author response to Decision Letter 2

Decision Letter 2

Przemysław Mroczek

Roles

Acceptance letter

Przemysław Mroczek

Roles

Associated Data

Supplementary Materials

Data Availability Statement

ACTIONS

PERMALINK

RESOURCES

Similar articles

Cited by other articles

Links to NCBI Databases