Varying particle size selectivity of soil erosion along a cultivated catena

Judit Alexandra Szabó; Boglárka Keller; Csaba Centeri; István Gábor Hatvani; József Kovács; Zoltán Szalai; Gergely Jakab

doi:10.1515/geo-2022-0585

Article Open Access

Varying particle size selectivity of soil erosion along a cultivated catena

Judit Alexandra Szabó , Boglárka Keller , Csaba Centeri , István Gábor Hatvani , József Kovács , Zoltán Szalai and Gergely Jakab

Published/Copyright: December 22, 2023

Published by

Become an author with De Gruyter Brill

Submit Manuscript Author Information

From the journal Open Geosciences Volume 15 Issue 1

Abstract

Sheet erosion is a complex multi-factor-dependent process with high spatial heterogeneity on hillslopes. Although the individual factors have been well studied, their aggregated effect on size-selective erosional processes is highly uncertain. Therefore, this study concentrates on the aggregate size distribution and effective particle size distribution (PSD) of the aggregates in the soil loss, collected from different simulated hillslope positions and surface conditions. These simulated hillslope positions combine moisture content from the extremely dry to the saturated with related slope positions of 2, 5, and 12% steepness and different surface roughness (tilled and crusted surfaces) modelled in a laboratory rainfall simulator. Using hierarchical cluster analysis, the PSD of the aggregates was separated into three groups based on the differences in the 59–116 µm range of the PSD histograms, namely, macro-aggregates, 50–250 µm sized micro-aggregates, and <50 µm sized fractions were classified into distinct groups, although some micro-aggregate samples were classified into the macro-aggregate group. PSDs from the 50–250 µm aggregate size fraction were clustered into a group of macro-aggregates if the PSD changed with time (during the rainfall event), notably on rough surfaces. The role of the specified size range in the classification is believed to be due to the parallel presence of aggregates and single particles in this range. As aggregates have a lower density than mineral particles, they tend to be enriched in soil loss under low-energy runoff conditions. Moreover, all samples in the <50 µm fraction clustered into the macro-aggregate group were eroded from the smooth/crusted surface, probably due to the presence of larger particles. The results indicate that the combined effect of erosional factors is not apparent, and the impact of the crust and extreme moisture content on the selectivity and size distribution of the sediment requires further investigation.

Keywords: selective erosion; particle size distribution; rainfall simulation; aggregate size distribution

1 Introduction

Sheet erosion is a potential hazard for agricultural fields, being the root cause of organic matter (OM) loss [1,2] and non-point source of nutrient pollution [3]. As a consequence of the three main processes of erosion – aggregate detachment, sediment transport, and sedimentation – fine-grained material, OM, and soil organic carbon (SOC) are enriched in the soil loss [4,5]. Sheet erosion is a size-selective process [6,7,8]. The study of this selectivity is critical because the 63–250 µm sized aggregates play an essential role in primary silt- and clay-bounded OM mobilization [1]. It has been stated that the coarse-size aggregate fractions have the largest SOC concentrations [9]. Furthermore, the clay size fraction of the sediment adsorbs the most nutrients and chemicals applied by agrotechnics [10], hence knowledge of the size distribution of the eroded material and the study of the selective process is crucial for soil protection planning.

The basis of selectivity, or preferential mobilization, is the hierarchical structure of soil particles. Primary soil particles form micro- (<250 µm) and macro-aggregates (>250 µm) [11]. The aggregates can break down if the raindrop energy is equal to or greater than the critical energy level of the aggregates. The continuously produced pre-detached or bounded material is transported if the power of the flow reaches a critical value [12]. Consequently, as part of a complex process, the aggregate and particle size distribution (PSD) of the eroded material could be influenced by various combinations of the same factors that control splash erosion, runoff, and soil loss.

Although the effects of the major influencing factors, such as soil type, initial soil moisture content, vegetation, slope steepness, rainfall intensity, surface roughness, and soil crust [13,14,15,16] on soil loss are generally well studied [17], their role in selectivity and selectivity itself need further investigation [1,18,19]. An increasing number of studies [20,21,22] have identified slope steepness and surface properties as the main factors influencing particle size selectivity. In addition, fine particle selectivity in eroded sediment is also related to the rainfall intensity [23].

It was determined that the loss of particles with size 2–8 μm was significantly influenced by slope steepness, whereas those with >8 and <2 μm were influenced more by raindrop impact on silty loam soil [22]. Surface roughness has a crucial effect on sediment PSD, and the clay fraction is transported in an aggregated form. In contrast, the silt fraction is eroded mainly as primary/single particles [20]. Based on a study conducted on upper, mid, and lower slope positions, the role of surface properties of soil (including texture, vegetation cover, slope steepness, SOC content, soil OM composition, and initial moisture content) on selectivity has also been reported [21].

In erosion processes, slope gradient is a crucial property [12]; the amount of runoff and soil loss is proportional to the slope gradient [13,14,24]. Thus, the geomorphological position representing changing slope steepness could determine the occurrence and magnitude of erosional processes. At the field scale, it was concluded that the aggregates containing high molecular weight SOM were deposited at a higher geomorphologic position, such as the upper or mid-slope. At the same time, the less polymerized SOM components were enriched at the low slope position [25], indicating that different types of SOM were transported to different distances. This selective SOM sedimentation was presumed to be related to the aggregate and particle size of the sediment.

Single soil particles, especially the smallest ones, form aggregates and the small aggregates may create macro-aggregates. Some aggregates are broken due to raindrops and transportation, whereas others are sedimented in an aggregated form. Thus, soil loss contains both aggregates and single particles. PSD reflects the distribution after the complete destruction of all aggregates. Changes in the aggregate size distribution (ASD) and PSD in the soil loss compared to those in the original soil are good indicators of the selectivity processes. For example, Ding and Huang [20] and Martinez-Mena et al. [26] used the effective and ultimate PSD ratios per fraction to study the occurrence of selectivity in different fractions. However, the basics of ASD and PSD changes along a catena owing to soil erosion and redistribution have not yet been studied in detail.

The aim of this study is to determine the following:

What is the dominant aggregate size fraction in runoff under different surface conditions?
What is the dominant particle size fraction in runoff under different surface conditions?
What is the difference between the PSDs of the micro- and macro-aggregates?
What effect does the evolved crust/surface roughness have on aggregate fractions’ sediment dynamics and PSDs?

We characterized erosional selectivity by comparing the effective PSD histograms of the aggregate size classes. The catena scale was adapted to represent the inhomogeneity of erosion in the different morphological positions of a slope. It was also used to model soil heterogeneity and typical surface conditions. Because the influencing factors of soil erosion are not independent, moisture content was combined with surface roughness and linked to slope steepness to model the morphological positions of a catena and examine the differences in selectivity during the erosion process.

2 Materials and methods

2.1 Study site and soil characteristics

The study area was a crop field under conventional tillage with autumn mouldboard ploughing. The studied soil and its location represent loess-derived, highly erodible hilly croplands dominating parts of Europe and Asia. The sampled soil is located 154–170 m above sea level in the Gödöllő Hills near Ceglédbercel, Hungary (47.249846°N; 19.678744°E). The soil was collected from the upper third of the cultivated catena from the cultivated layer using a circle of 2 m in diameter. The climate is humid continental, with an average annual temperature of 10.2°C and an average annual precipitation of 530 mm. The soil was classified as eroded eutric calcaric Cambisol loamic [27] with a silty-clay loam texture that developed on loess parent material and had 15.3% sand (>20 µm), 74.7% silt (20–8 µm), and 10% clay content (<8 µm) by laser diffraction. The ASD of the in situ tilled layer determined by wet sieving was 4.81% <50 µm, 24.63% 50–250 µm, 32.00% 250–1,000 µm, and 38.57% >1,000 µm [28]. The soil pH was 7.64 (H₂O), CaCO₃ content was 5.1%, and SOC content was 13.640 g kg⁻¹ [4]. After seedbed preparation, the soil was collected directly from the tilled horizon (approximately 0–15 cm) and placed into the sample flume (Figure 1a).

Figure 1

Methodological components. (a) seedbed condition soil with preserved clods in the flume; (b) front view of the simulator; (c) soil sample collection: sieve series; and (d) schematics of sample collection; soil loss is separated using a series of sieves to obtain aggregate size classes.

2.2 Rainfall simulation

The plot-scale lab simulator (Figure 1b) is located at Eötvös Loránd University, Budapest, Hungary [28,29,30,31]. The kinetic energy of the simulated 80 mm h⁻¹ rainfall using the 1/2HH-40WSQ nozzle was 17 J m⁻² mm⁻¹ at 20 kPa [32] and the size of the soil sample flume was 0.5 m × 1.0 m × 0.2 m (0.1 m³).

2.3 Experimental design

Experiments were designed along the spatial gradient of a representative theoretical catena applying idealized values (Figure 2). Slope steepness, moisture content, and surface roughness were used to model the natural soil surface characteristics of the morphological positions representing the shoulder, backslope, and toeslope (Figure 2; Table 1).

Figure 2

Slope gradient and surface state combinations along an idealized catena (photos: Judit Alexandra Szabó).

Table 1

Coding of the experiments

Morphological position	Slope gradient (%)	Surface status	Surface description	Experiment code
Toeslope	2	Dry	Crusted and sealed surface	2D
Toeslope	2	Wet	Saturated soil “muddy”	2W
Backslope	5	Tilled	Tilled surface with field capacity moisture content	5T
Backslope	5	Crusted	Crusted surface on saturated soil	5C
Shoulder	12	Tilled	Tilled surface with field capacity moisture content	12T
Shoulder	12	Crusted	Crusted surface on saturated soil	12C

2.3.1 Surface state description

Six simulated hillslope positions and surface conditions, hereafter “surface states,” were selected for the study represented by one simulation each (Table 1) with the combinations of moisture contents of extremely dry and saturated, and slope steepness of 2, 5, and 12%. The combination of conditions, was necessary to distinguish between the detachments of particles of various sizes under different surface conditions and steepness. For example, on backslope (5% slope steepness) and shoulder (12% slope steepness) positions, recently tilled and crusted surfaces are the most frequent surface character, while at the bottom of the catena on the toeslope position (2% slope steepness), sedimentation and even inland inundation can occur. Therefore, extreme moisture content surfaces (wet and dry) were simulated on this slope position.

All the measurements were performed on the same soil under various conditions. To create natural soil surface characteristics, the flume was once filled with soil collected in the field. Before the first experiment, the soil was compacted with three successive rainfall events. After compaction, the upper 10 cm of the soil was manually “tilled” by hoeing. This surface preparation was used before the tilled (T) and 2W experiments. Heavy rainfalls erode the surface of the soil, which is called sheet erosion. The aggregates are destroyed during this process, and the soil particles are transported and sedimented by the runoff. But beneath the soil surface at >1 cm depth, precipitation does not affect the soil structure [33]. Thus, simplifying the process of heavy rainfalls creates crusts on the surface that tillage can destroy. One of the most important reasons for tillage is soil structure formation. In theory, ploughing creates the same soil aggregation each year. Therefore, tillage does remove the formal effects of rainfalls and provides a standard and comparable soil condition to investigate.

The tilled experiments (5T and 12T) represent cultivated, rough surfaces with clods at two different slope angles. Their initial moisture content was below field capacity (15–18%). Crusted surface experiments (5C and 12C) were repeated for tilled surface experiments per slope angle. After rainfall simulation on a tilled surface, the soil crust evolved. The surface was untouched, and the second rainfall simulation was conducted on the crusted surface one day after the tilled experiment. In this case, the surface was smoother than before, crusted, and the moisture content was increased to the field capacity (24–27%). Following these four experiments, the flume was flooded for a month, with continuous modelling of water coverage/pluvial flood (2W). The slope for the 2W experiment was set to 2% because water can accumulate on surfaces with lower slope angles, where the runoff is not a dominant process. After this toeslope simulation, the soil was left for 3 months to dry, and the surface remained untouched after 2W. During these months, a dry and crusted surface evolved (2D) before the sixth simulation.

2.3.2 Sampling strategy

The sampling, measurements, and data processing consisted of four main steps in each experiment.

Rainfall simulation and soil loss collection;
Soil loss separation by aggregate sizes and weight measurement of the aggregate fractions;
PSD measurement of the aggregate size fractions with laser diffraction (Horiba LA- 950);
Hierarchical cluster analysis (HCA) of the PSD curves.

2.4 Aggregate size separation and soil loss calculation

During the simulation, using a sieve series, the runoff was immediately separated into four aggregate size classes according to their size: A (>1,000 µm), B (250–1,000 µm), C (50–250 µm), and D (<50 µm) (Figure 1c and d) to directly measure the ASD of soil loss. In each simulation, three temporal sections (I, II, and III) were collected separately [28,29]. Three litres of runoff were collected from each temporal section. Altogether, 72 runoff samples were collected (six simulations × three temporal sections × four aggregate size fractions). All the samples were dried at 60°C for 24 h and weighed to the nearest 0.01 g. Because of the different durations of the precipitation to determine the amount of soil loss per hour, normalization was needed. Soil loss values were calculated using the equation of equilibrium runoff (mm h⁻¹) per simulation and sediment concentration values [29] (Table S1). The ASD of the in situ soil was determined using wet sieving with the same sieve series as a control.

2.5 PSD measurements

All samples were analysed using a Horiba LA-950 V2 particle sizer. A size range of 0.03 µm to 3 mm was detected with two light sources, the red (650 nm) and blue LED (405 nm) monochromatic laser beams, to measure the smaller and larger particles, respectively. The continuous size distribution is displayed as a histogram with 93 classes on a logarithmic scale.

The samples were dispersed in sodium pyrophosphate (Hungarian Standard MSZ-08 0205-78 1978) and sonicated for 15 min (12 W) as a pre-treatment method. This pre-treatment method was more of the nature of physical destruction than binding agent removal because we aimed to study effective PSD; the presence of stable micro-aggregates within the macro-aggregates, which are resistant to physical forces, had to stay together, as they occur in nature.

The Mie theory was used to calculate the PSD, and two different refractive indices (RIs) were set to exclude the effect of RI on the PSD during the evaluation. In the first evaluation round, the RI of quartz was selected because it is the dominant mineral in the soil. Therefore, according to the Horiba RI library, 1.45 as real RI and 0.2 as imaginary RI values were set. However, in soils, the mineral composition, including the content of organic-mineral complexes, varies. Therefore, the second round RIs was set based on the RI calculator of the software with 1.51 as real RI and 0.1 as imaginary RI. Both values were close to the RI of quartz suggested in the literature [34,35].

The particle size classes were separated according to the criteria of the United States Department of Agriculture (USDA) [36] into seven categories of sand, silt, and clay. The upper size boundary of the clay was set to 8 µm owing to its better comparability to that obtained via the gravimetric method [37]. The aggregate and particle size classes overlapped (Table 2).

Table 2

Particle size categories used in the study and their overlapping

Particle size range	USDA category	Aggregate size class
>2,000 µm	Gravel	A (>1,000 µm)
1,000–2,000 µm	Very coarse sand	A (>1,000 µm)
500–1,000 µm	Coarse sand	B (250–1,000 µm)
250–500 µm	Medium sand	B (250–1,000 µm)
100–250 µm	Fine sand	C (50–250 µm)
50–100 µm	Very fine sand	C (50–250 µm)
2–50 µm	Silt	D (<50 µm)
<2µm	Clay	D (<50 µm)

2.6 Statistical analysis

HCA was applied to classify the Z-score transformed particle size ranges using Ward’s method [38] with squared Euclidean distance, as in numerous earth science studies conducted previously [39,40] using the factoextra package [41] in R [42]. To verify the HCA results, that is, how well the groups were defined, linear discriminant analysis (LDA) [43,44] and the indices incorporated into the Nbclust package [45] were used as in numerous other studies (e.g. [46,47]). LDA uses a discriminant function which splits the linear combination of the predictor (independent) variables into sets by a linear plane (in the case of LDA), providing the percentage of correctly classified observations and, consequently, the best discrimination between the groups. Wilks’ λ quotient was estimated to define which size class played the most significant role in determining the formation of cluster groups [48]. The value of λ is the ratio of the within-group sum of squares to the total sum of squares, ranging between 0 and 1. If λ = 1, the parameter does not affect the formation of the cluster groups. In contrast, the smaller the ratio, the more the parameter affects the formation of the groups [49]. The analyses were performed in R [42] and in IBM SPSS Statistics 22.0.

3 Results and discussion

The amount of soil loss and mean sediment concentrations were expected to be functions of the geomorphological position (slope steepness) and soil conditions (Table S1). However, the present study focused on soil loss composition.

3.1 ASD of soil loss

The ratio of macro-aggregates >1,000 µm differed between the untreated bulk soil (control) and all soil losses from different geomorphological positions and surface states (Figure 3a). The ratio of this fraction to soil loss was decreased to almost 0 from 40% in the untreated bulk soil. This substantial decrease was probably the result of the mechanical breakdown of the macro-aggregates by raindrop impact [50] on the weak soil structure. In contrast, for example, in the case of a well-structured Mollisol, the ratio of >1,000 µm aggregates in soil loss could be 8% at a 5% slope eroded by 50 mm h⁻¹ rainfall intensity [51]. Another possible explanation for the low ratio of the >1,000 µm fraction is that the runoff energy was insufficient to transport the largest aggregates.

$Figure 3 Percentage of aggregate size fractions: (a) >1,000 µm; (b) 250–1,000 µm; (c) 50–250 µm; and (d) <50 µm in the eroded material compared to that in the untreated bulk soil (control sample). The boxes show the interquartile range and the black horizontal line in the middle of each box is the median value. The two upright lines represent the data within the 1.5 interquartile range [39].$

Figure 3

Percentage of aggregate size fractions: (a) >1,000 µm; (b) 250–1,000 µm; (c) 50–250 µm; and (d) <50 µm in the eroded material compared to that in the untreated bulk soil (control sample). The boxes show the interquartile range and the black horizontal line in the middle of each box is the median value. The two upright lines represent the data within the 1.5 interquartile range [39].

The ratio of the 250–1,000 µm sized macro-aggregates in soil loss varied with slope steepness (Figure 3b). At 5% slope, the ratio of the 250–1,000 µm sized aggregates was more or less the same as in the control sample. The values showed high variability, which could be related to the varying energy of runoff triggered by the various spatial concentrations of flow on rough surfaces [15]. In the runoff from the tilled surface (5T), the ratio of these macro-aggregates was slightly higher than that of the crusted surface (5C). It would be difficult to say that the size of the aggregates in the runoff was originally 250–1,000 µm or that they were just the result of broken >1,000 µm aggregates, as reported by Zuo et al. [52]. In the case of a 12% slope (shoulder position), the highest ratio of 250–1,000 µm sized macro-aggregates was measured in the soil loss. On steeper slopes, the runoff has higher energy and can transport larger amounts and heavier sediments [53]. Moreover, as in the case of 5% slope, the surface state affects the ratio of this size class. Runoff on steep slopes can increase the runoff’s macro-aggregate transport capacity, such as the shoulder position. As their ratio was up to 40%, we can state that smaller macro-aggregates might play a significant role in soil redistribution, even on moderate slopes. In the runoff from the crusted surface, we measured a smaller ratio of 250–1,000 µm sized aggregates, which can be explained by the effect of surface micro topology [29] and/or the lack of available aggregates for erosion on the surface as a consequence of splash erosion and aggregate breakdown [50]. At 2% slope steepness, the lowest ratio (<10%) of 250–1,000 µm size aggregates was measured in the runoff. On a 2% gentle slope, the runoff has lower energy than on 5 or 12% slopes, which could be one of the reasons for the low macro-aggregate ratio [53,54]. Additionally, the two experiments related to two extreme moisture content statuses were where the amount of available macro-aggregates was reduced by strong slaking [50] and crust formation during the drying period. In addition, the toeslope position is a sedimentation area where the transported soil can accumulate, affecting in situ ASD [55]. The size composition of runoff from the toeslope may be double-selected and poor in macro-aggregates.

The ratio of the 50–250 µm sized micro-aggregates in soil loss was higher than that in control in the case of 5 and 12% slope experiments, and similar in 2% slope experiments. This size group shows high variability in almost all experiments, which means that this size group is favourable in all types of runoff, and the ratio of this group is strongly dependent on the other group sizes.

The ratio of <50 µm sized group with micro-aggregates and individual particles increased in the runoff in all experiments, but the change volume was the largest in 2W and 2D experiments, from 4 to 80%. In addition, the runoff of the wet experiment (2W) had the highest variability in the ratio of the <50 µm group. Comparing the wet and dry experiments on 2% slope, the former resulted in high variability, independent of particle size. Therefore, the surface states in this case strongly defined the transportable soil material, and stable aggregates could be delivered from the wet surface. In contrast, the amount of available aggregates on the dry surface covered by a thick crust was limited. In the other experiments, the ratio of the <50 µm sized group was higher on a crusted surface per slope angle and higher on 5% slope than on 12% slope. The slope angle affects groups of larger particles more [53]. However, the surface state difference was the most relevant in this group. Crust formation drastically decreased the amount of detachable aggregates, e.g. in the 2D experiment.

Generally, the mobilized soil was enriched in fine aggregates relative to the in situ source soil. The fine fraction ordinarily enhances soil loss, as reported by Martinez-Mena et al. [26], Leguédois and Bissonnais [56], and Armstrong et al. [57]. In the 50–250 µm fraction (Figure 3c), there was only a slight enrichment in the transported material, except in the toeslope position, where this fraction was decreased. At the same time, the ratio of the smallest size class (Figure 3d) showed a strong increase in all cases, which underlines the observation of Lu et al. [51], who detected maximum of 30.2 times higher <250 µm aggregates in the soil loss of a Mollisol, which was 96.8% of the total aggregate loss. In the present study, the ratio of <250 µm aggregates in soil loss reached 90% in the toeslope geomorphologic position. Beyond the variations in runoff energy, the differences in aggregate stability at different slope positions [58] may contribute to changes in the ASD of soil loss.

3.2 Transportation based on aggregate class sizes

The PSD of the aggregate size classes (>1,000 µm (A); 250–1,000 µm (B); 50–250 µm (C), and <50 µm (D)) were compared in two different ways (Figure 4). In addition to the traditional classification of the ratios of clay, silt, and sand according to USDA size classes (Figure 4b, d, and f), HCA was applied using detailed PSD data (Figure 4a, c, and e). The HCA separated the samples into three groups at 50% of the maximum linkage distance (Figure S1; Table S2). The classification results were validated 100% using LDA (Table S3) and 15 out of the 26 calculated indices (Table S4), out of which three were proposed as the best number of clusters. The classification was driven by the 59–116 µm size class being the most influential according to the Wilks’ λ quotient (Table S5). Two groups can be considered as “homogeneous” in terms of empirical group element characteristics, but not statistically [59], and contain PSDs of the same aggregate size class. In contrast, the third group has a dominant aggregate size class, but some samples from the other aggregate size classes were also classified in the dominant class (Table S2, Figure S1). RI changes did not affect the result of classification; thus, either the predefined RI of quartz or software-calculated RI was suitable to measure the PSD of this soil sample. Here we present the classification based on the RI of quartz.

$Figure 4 PSDs of the tree groups identified by HCA (Section 2.6) driven by the 59–116 µm size class being the most influential in the classification. The group of “single particles” contains only PSDs from <50 µm fraction independent of the slope and surface conditions. (a) PSD histograms of the group; (b) USDA classification. The group of “micro-aggregates” contains only PSD from 50 to 250 µm fraction. (c) PSD histograms of the group; (d) USDA classification. The heterogeneous “macro-aggregates” group includes all samples of macro-aggregates (>1,000 µm (A); 250–1,000 µm (B) size classes) independent of the treatment and the remaining samples from the micro-aggregate fractions. (e) PSD histograms of the group; dotted line: samples from the 50–250 µm fraction; dashed line samples from the <50 µm fraction; and (f) USDA classification.$

Figure 4

PSDs of the tree groups identified by HCA (Section 2.6) driven by the 59–116 µm size class being the most influential in the classification. The group of “single particles” contains only PSDs from <50 µm fraction independent of the slope and surface conditions. (a) PSD histograms of the group; (b) USDA classification. The group of “micro-aggregates” contains only PSD from 50 to 250 µm fraction. (c) PSD histograms of the group; (d) USDA classification. The heterogeneous “macro-aggregates” group includes all samples of macro-aggregates (>1,000 µm (A); 250–1,000 µm (B) size classes) independent of the treatment and the remaining samples from the micro-aggregate fractions. (e) PSD histograms of the group; dotted line: samples from the 50–250 µm fraction; dashed line samples from the <50 µm fraction; and (f) USDA classification.

The “single particles” group contains only PSD from <50 µm fraction independent of slope and surface conditions. Of the 18 samples (<50 µm fraction) measured, 13 were classified into this group. PSD curves can be described with one peak at 10 µm (4–7%), and the median was the smallest among the groups, approximately 6–10 µm. From a classical viewpoint, the silt fractions dominate in the <50 µm sized soil loss.

The “micro-aggregates” group contains only PSD from 50 to 250 µm fraction (Figure 4c and d). Of the 18 measured samplheterogeneous group containedes (50–250 µm fraction), 12 were classified into this group. The PSD curves can be described by an elongated peak between 1 and 30 µm and a craggy peak around 90 µm (Figure 4c). The former peak is related to the single particles as was in the “single particles” group. The latter is about the range that rules the classification most, with a higher proportion of 7%. The median values were 20–40 µm. Two peaks PSD had already been measured in the <250 µm fraction and in the case of both the control and eroded samples of loam, with a major peak near 63 µm and a minor peak in the coarse clay size fraction [35]. The bimodal PSD curve for the soil loss samples indicates the complex nature of transportation. It leads to different travel distances of particles of various sizes [6], suggesting the joint delivery of single particles and aggregates. Moreover, the bimodal distribution is also attributed to transport mechanism differences between fine and coarser sediments [7,60].

According to the USDA classification system, this group contains the least silt fraction. At the same time, the very fine sand and fine sand fraction together could make up 50% of the soil loss (12C experiment) (Figure 4d). The size range of the micro-aggregates overlapped with that of very fine sand and fine sand [61], and the parallel presence of single quartz particles and stable micro-aggregates was also observed in these samples [29]. This means that in this group, mostly clay and silt fractions were transported as micro-aggregates, with the occurrence of larger individual quartz particles.

The heterogeneous group contained all samples of macro-aggregates (>1,000 µm (A); 250–1,000 µm (B) size classes) independent of the treatment. Smaller-sized samples, which were not classified into the former two groups, were placed in this group (Figure 4e and f). The PSDs classified in this group were bimodal, with one peak located at approximately 10 µm and the other near 50 µm, with approximately 4% peak height. This latter peak indicates smaller particles compared to the peak found in the “micro-aggregate” group. This difference in size may indicate the various origins of these particles. The abscissa is in logarithmic scale; hence, the peak at 50 µm suggests a wider range of particle sizes than 10 µm. The median value of the histograms in this mixed group was approximately 10–15 µm (Figure 4e), in the silt range, which dominated the samples.

Some PSDs of 50–250 µm aggregates were also classified into this group (Figure 4e, dotted line). In most of these samples, the silt and clay ratio together was higher than that of the samples in the 50–250 µm group. These samples were from tilled or dry surfaces and were mainly collected at the beginning of the runoff process, with high variability with time among the replications [29]. The second exception was related to some samples of the <50 µm fraction (Figure 4e, dashed line), which were not classified alongside the other <50 µm fraction samples, and were collected from the crusted surfaces.

The results suggest that the aggregate size classes of soil loss are related to their PSD, where the observed peaks differ in size. Micro-aggregates contain larger particles (∼90 µm) than macro-aggregates with lower modus (50 µm). Therefore, stable “runoff-resistant” micro-aggregates are delivered separately. The less stable, smaller-sized micro-aggregates and the very fine sand-sized single particles tend to build up macro-aggregates. As the formation rules of macro-aggregates are much less studied than those of micro-aggregates and no consensus has been reached, secondary aggregate formation may vary considerably owing to different environmental conditions [62,63]. However, the differences between aggregates and single particles are much better understood. Theoretically, aggregates and individual mineral particles differ in density [64], which may affect delivery. Consequently, due to their lower density, micro-aggregates are expected to be more detached and transported than individual particles of the same size. This effect is definite at lower runoff energy (less steep slope), where micro-aggregates might overrepresent the overlapping range in soil loss (Figure 4f).

The limitations of the presented findings lie in the fact that these are the results of the study on a single-tilled topsoil; therefore, generalization is impossible. Moreover, the results measured at topographical hotspots of the idealized catena were not interpreted using real topography. Accordingly, investigations should be extended to understand field-scale trends.

4 Conclusion

The present study used a partially new approach to understand the size-selective processes of sheet erosion, as used in the experimental setup, to simulate different and frequent situations occurring in the field. This approach allowed us to expand the opportunities of a laboratory rainfall simulator spatially. The ASD of soil loss was significantly affected by slope steepness, initial moisture content, or surface roughness, whereas the PSD was found to be a function of the size of the aggregates. The most effective size range was 59–116 µm, which may reflect micro-aggregates and individual particles. As stable micro-aggregates often act as shelters of labile OM or available nutrients, their enrichment in soil loss may be a function of the aggregate ratio in soil loss. The results indicate the complex effects of erosion factors, which are not obvious and require more detailed studies.

Acknowledgements

The authors would like to thank Paul Thatcher for his work on our English version. Judit Alexandra Szabó wishes to dedicate the article to Zorka Szabó.

Funding information: The present research was supported by the NKFIH via reference number K-143005.
Conflict of interest: Authors state no conflict of interest.

References

[1] Nadeu E, de Vente J, Martínez-Mena M, Boix-Fayos C. Exploring particle size distribution and organic carbon pools mobilized by different erosion processes at the catchment scale. J Soil Sed. 2011;11(4):667–78.10.1007/s11368-011-0348-1Search in Google Scholar

[2] Schiettecatte W, Gabriels D, Cornelis WM, Hofman G. Enrichment of organic carbon in sediment transport by interrill and rill erosion processes. Soil Sci Soc Am J. 2008;72(1):50–5.10.2136/sssaj2007.0201Search in Google Scholar

[3] Di Stefano C, Ferro V, Palazzolo E, Panno M. SW—soil and water: Sediment delivery processes and agricultural non-point pollution in a Sicilian Basin. J Agric Eng Res. 2000;77(1):103–12.10.1006/jaer.2000.0567Search in Google Scholar

[4] Jakab G, Szabó J, Szalai Z, Mészáros E, Madarász B, Centeri CS, et al. Changes in organic carbon concentration and organic matter compound of erosion-delivered soil aggregates. J Soil Sed. 2016;75(2):144.10.1007/s12665-015-5052-9Search in Google Scholar

[5] Mahmoodabadi M, Sajjadi SA. Effects of rain intensity, slope gradient and particle size distribution on the relative contributions of splash and wash loads to rain-induced erosion. Geomorphology. 2016;253:159–67.10.1016/j.geomorph.2015.10.010Search in Google Scholar

[6] Asadi H, Ghadiri H, Rose CW, Rouhipour H. Interrill soil erosion processes and their interaction on low slopes. Earth Surf Process Landf. 2007;32(5):711–24.10.1002/esp.1426Search in Google Scholar

[7] Asadi H, Moussavi A, Ghadiri H, Rose CW. Flow-driven soil erosion processes and the size selectivity of sediment. J Hydrol. 2011;406:73–81. 10.1016/j.jhydrol.2011.06.010.Search in Google Scholar

[8] Slattery MC, Burt TP. Particle size characteristics of suspended sediment in hillslope runoff and stream flow. Earth Surf Process Landf. 1997;22:705–19.10.1002/(SICI)1096-9837(199708)22:8<705::AID-ESP739>3.0.CO;2-6Search in Google Scholar

[9] Ouyang W, Xu X, Hao Z, Gao X. Effects of soil moisture content on upland nitrogen loss. J Hydrol. 2017;546:71–80.10.1016/j.jhydrol.2016.12.053Search in Google Scholar

[10] Young RA, Onstad CA. Predicting particle-size composition of eroded soil. Trans ASAE. 1976;19(6):1071–5.10.13031/2013.36178Search in Google Scholar

[11] Tisdall JM, Oades J. Organic matter and water‐stable aggregates in soils. Eur J Soil Sci. 1982;33(2):141–63.10.1111/j.1365-2389.1982.tb01755.xSearch in Google Scholar

[12] Kinnell PIA. Raindrop‐impact‐induced erosion processes and prediction: a review. Hydrol Process. 2005;19(14):2815–44.10.1002/hyp.5788Search in Google Scholar

[13] Assouline S, Ben-Hur M. Effects of rainfall intensity and slope gradient on the dynamics of interrill erosion during soil surface sealing. Catena. 2006;66(3):211–20.10.1016/j.catena.2006.02.005Search in Google Scholar

[14] Defersha MB, Melesse AM. Effect of rainfall intensity, slope and antecedent moisture content on sediment concentration and sediment enrichment ratio. Catena. 2012;90:47–52.10.1016/j.catena.2011.11.002Search in Google Scholar

[15] Gómez JA, Nearing MA. Runoff and sediment losses from rough and smooth soil surfaces in a laboratory experiment. Catena. 2005;59(3):253–66.10.1016/j.catena.2004.09.008Search in Google Scholar

[16] Kinnell PIA. The effect of slope length on sediment concentrations associated with side-slope erosion. Soil Sci Soc Am J. 2000;64(3):1004–8.10.2136/sssaj2000.6431004xSearch in Google Scholar

[17] Poesen J. Soil erosion hazard and mitigation in the Euro-Mediterranean region: do we need more research? Hung Geogr Bull. 2015;64(4):293–9.10.15201/hungeobull.64.4.3Search in Google Scholar

[18] Suleymanov A, Gabbasova I, Suleymanov R, Abakumov E, Polyakov V, Liebelt P. Mapping soil organic carbon under erosion processes using remote sensing. Hung Geogr Bull. 2021;70(1):49–64. 10.15201/hungeobull.70.1.4.Search in Google Scholar

[19] Sun L, Wang C. Effects of the extreme rainfall events on particle size selectivity on the simulated Gobi surface using indoor rainfall experiments. Catena. 2023;220(Part B):106716.10.1016/j.catena.2022.106716Search in Google Scholar

[20] Ding W, Huang C. Effects of soil surface roughness on interrill erosion processes and sediment particle size distribution. Geomorphology. 2017;295:801–10.10.1016/j.geomorph.2017.08.033Search in Google Scholar

[21] Koiter AJ, Owens PN, Petticrew EL, Lobb DA. The role of soil surface properties on the particle size and carbon selectivity of interrill erosion in agricultural landscapes. Catena. 2017;153:194–206.10.1016/j.catena.2017.01.024Search in Google Scholar

[22] Kiani-Harchegani M, Hamidreza S, Singh VP, Asadi H. Effect of rainfall intensity and slope on sediment particle size distribution during erosion using partial eta squared. Catena. 2019;176:65–72. 10.1016/j.catena.2019.01.006.Search in Google Scholar

[23] Wu X, Wei Y, Cai C, Yuan Z, Liao Y, Li D. Effects of erosion-induced land degradation on effective sediment size characteristics in sheet erosion. Catena. 2020;195:104843.10.1016/j.catena.2020.104843Search in Google Scholar

[24] Chaplot V, Le Bissonnais Y. Field measurements of interrill erosion under different slopes and plot sizes. Earth Surf Process Landf. 2000;25(2):145–53.10.1002/(SICI)1096-9837(200002)25:2<145::AID-ESP51>3.3.CO;2-VSearch in Google Scholar

[25] Szalai Z, Szabo J, Kovacs J, Meszaros E, Albert G, Centeri CS, et al. Redistribution of soil organic carbon triggered by erosion at field scale under subhumid climate, Hungary. Pedosphere. 2016;26(5):652–65.10.1016/S1002-0160(15)60074-1Search in Google Scholar

[26] Martinez-Mena M, Rogel JA, Albaladejo J, Castillo VM. Influence of vegetal cover on sediment particle size distribution in natural rainfall conditions in a semiarid environment. Catena. 2000;38(3):175–90.10.1016/S0341-8162(99)00073-9Search in Google Scholar

[27] WRB. 2015 World reference base for soil resources 2014, update 2015 International soil classification system for naming soils and creating legends for soil maps. World Soil Resources Reports 106. Rome: FAO; 2015. p. 203.Search in Google Scholar

[28] Szabó JA, Jakab GI, Szabó B. Spatial and temporal heterogeneity of runoff and soil loss dynamics under simulated rainfall. Hung Geogr Bull. 2015;64(1):25–34.10.15201/hungeobull.64.1.3Search in Google Scholar

[29] Szabó JA. The effect of topography and tillage on aggregate and particle size distribution of soil loss. PhD thesis. Hungary: Eötvös Loránd University, 2019.Search in Google Scholar

[30] Szabó JA, Centeri C, Keller B, Hatvani IG, Szalai Z, Dobos E, et al. The use of various rainfall simulators in the determination of the driving forces of changes in sediment concentration and clay enrichment. Water. 2020;12(10):1–17.10.3390/w12102856Search in Google Scholar

[31] Szabó JA, Király CS, Karlik M, Tóth A, Szalai Z, Jakab G. Rare earth oxide tracking coupled with 3D soil surface modelling: An opportunity to study small-scale soil redistribution. J Soils Sediments. 2020;20:2405–17.10.1007/s11368-020-02582-7Search in Google Scholar

[32] Strauss P, Pitty J, Pfeffer M, Mentler A. Rainfall simulation for outdoor experiments. In: Jamet P, Cornejo J, editors. Current Research Methods to Assess the Environmental Fate of Pesticides. Idaho Falls, ID, USA: INRA Editions; 2000. p. 329–33.Search in Google Scholar

[33] Assouline S. Rainfall-induced soil surface sealing: A critical review of observations, conceptual models, and solutions. Vadose Zone J. 2004;3:570–91. 10.2136/vzj2004.0570.Search in Google Scholar

[34] Eshel G, Levy GJ, Mingelgrin U, Singer MJ. Critical evaluation of the use of laser diffraction for particle-size distribution analysis. Soil Sci Soc Am J. 2004;68(3):736–43.10.2136/sssaj2004.7360Search in Google Scholar

[35] Warrington DN, Mamedov AI, Bhardwaj AK, Levy GJ. Primary particle size distribution of eroded material affected by degree of aggregate slaking and seal development. Eur J Soil Sci. 2009;260:84–93.10.1111/j.1365-2389.2008.01090.xSearch in Google Scholar

[36] Ditzler C, Scheffe K, Monger HC, editors. Soil survey manual USDA Handbook 18. Washington, D.C.: Government Printing Office; 2017. p. 639.Search in Google Scholar

[37] Konert M, Vandenberghe J. Comparison of laser grain size analysis with pipette and sieve analysis: a solution for the underestimation of the clay fraction. Sedimentology. 1997;44:523–35.10.1046/j.1365-3091.1997.d01-38.xSearch in Google Scholar

[38] Ward JH. Hierarchical grouping to optimize an objective function. J Am Stat Assoc. 1963;58(301):236–44.10.1080/01621459.1963.10500845Search in Google Scholar

[39] Kovács J, Tanos P, Korponai J, Székely IK, Gondár K, Gondár-Sőregi K, et al. Analysis of water quality data for scientists. In: Voudouris K, Voutsa D, editors. Water Quality Monitoring and Assessment. Rijeka: Tech Open Access Publisher; 2012. p. 65–90.10.5772/32173Search in Google Scholar

[40] Young FJ, Hammer RD. Defining geographic soil bodies by landscape position, soil taxonomy, and cluster analysis. Soil Sci Soc Am J. 2000;64:989–98. 10.2136/sssaj2000.643989x.Search in Google Scholar

[41] Kassambara A, Mundt F. Factoextra: Extract and visualize the results of multivariate data analyses. R Package Version 1.0.7; 2020. https://CRAN.R-project.org/package=factoextra.Search in Google Scholar

[42] Core Team R. R: A language and environment for statistical computing. Vienna, Austria: R Foundation for Statistical Computing; 2022. https://www.R-project.org/.Search in Google Scholar

[43] Duda RO, Hart PE, Stork DG. Pattern Classification. Hoboken, USA: Wiley InterScience; 2000.Search in Google Scholar

[44] McLachlan G. Discriminant Analysis and Statistical Pattern Recognition. Hoboken, New Jersey: Wiley & Sons; 2005. p. 552.Search in Google Scholar

[45] Charrad M, Ghazzali N, Boiteau V, Niknafs A. NbClust: An R package for determining the relevant number of clusters in a data set. J Stat Softw. 2014;61(6):1–36. 10.18637/jss.v061.i06.Search in Google Scholar

[46] Lucas C, Wong P, Klein J, Castro TBR, Silva J, Sundaram M, et al. Longitudinal analyses reveal immunological misfiring in severe COVID-19. Nature. 2020;584:463–9. 10.1038/s41586-020-2588-y.Search in Google Scholar PubMed PubMed Central

[47] Chambers JM, Wyborn C, Ryan ME, Reid RS, Riechers M, Serban A, et al. Six modes of co-production for sustainability. Nat Sustain. 2021;4:983–96. 10.1038/s41893-021-00755-x.Search in Google Scholar

[48] Afifi A, Clark VA, May S, Raton B. Computer-Aided Multivariate Analysis. 4th edn. USA: Chapman & Hall/CRC; 2004. p. 489.Search in Google Scholar

[49] Kovács J, Nagy M, Czauner B, Kovács ISZ, Borsodi AK, Hatvani IG. Delimiting sub-areas in water bodies using multivariate data analysis on the example of Lake Balaton (W Hungary). J Environ Manage. 2012;110:151–8.10.1016/j.jenvman.2012.06.002Search in Google Scholar PubMed

[50] Le Bissonnais Y. Aggregate stability and assessment of soil crustability and erodibility: I. Theory and methodology. Eur J Soil Sci. 1996;47:425–37. 10.1111/j.1365-2389.1996.tb01843.x.Search in Google Scholar

[51] Lu J, Zheng F, Li G, Bian F, An J. The effects of raindrop impact and runoff detachment on hillslope soil erosion and soil aggregate loss in the Mollisol region of Northeast China. Soil Till Res. 2016;161:79–85.10.1016/j.still.2016.04.002Search in Google Scholar

[52] Zuo FL, Li XY, Yang XF, Wang Y, Ma YJ, Huang YH, et al. Soil particle-size distribution and aggregate stability of new reconstructed purple soil affected by soil erosion in overland flow. J Soils Sediments. 2020;20:272–83. 10.1007/s11368-019-02408-1.Search in Google Scholar

[53] Shi P, Arter P, Liu X, Keller M, Schulin R. Soil aggregate stability and size-selective sediment transport with surface runoff as affected by organic residue amendment. Sci Total Environ. 2017;607:95–102. 10.1016/j.scitotenv.2017.07.008.Search in Google Scholar PubMed

[54] Shi ZH, Fang NF, Wu FZ, Wang L, Yue BJ, Wu GL. Soil erosion processes and sediment sorting associated with transport mechanisms on steep slopes. J Hydrol. 2012;454–455:123–30. 10.1016/j.jhydrol.2012.06.004.Search in Google Scholar

[55] Jasik O, Kodesova R, Kubis A, Stehlikova I, Drábek O, Kapicka A. Soil aggregate stability within morphologically diverse areas. Catena. 2015;127:287–99. 10.1016/j.catena.2015.01.010.Search in Google Scholar

[56] Leguédois S, Bissonnais YL. Size fractions resulting from an aggregate stability test, interrill detachment and transport. Earth Surf Process Landf. 2004;29(9):1117–29.10.1002/esp.1106Search in Google Scholar

[57] Armstrong A, Quinton JN, Heng BCP, Chandler JH. Variability of interrill erosion at low slopes. Earth Surf Process Landf. 2011;36(1):97–106.10.1002/esp.2024Search in Google Scholar

[58] Pavlu L, Kodesova R, Vasát R, Fér M, Klement A, Nikodem A, et al. Estimation of the stability of topsoil aggregates in areas affected by water erosion using selected soil and terrain properties. Soil Till Res. 2022;219:105348. 10.1016/j.still.2022.105348.Search in Google Scholar

[59] Kovács J, Kovács S, Magyar N, Tanos P, Hatvani IG, Anda A. Classification into homogeneous groups using combined cluster and discriminant analysis. Environ Model Softw. 2014;57:52–9.10.1016/j.envsoft.2014.01.010Search in Google Scholar

[60] Hao H, Wang J, Guo Z, Hua L. Water erosion processes and dynamic changes of sediment size distribution under the combined effects of rainfall and overland flow. Catena. 2019;173:494–504.10.1016/j.catena.2018.10.029Search in Google Scholar

[61] Han Z, Li J, Li Y, Gu X, Chen X, Wei C. Assessment of the size selectivity of eroded sediment in a partially saturated sandy loam soil using scouring experiments. Catena. 2021;201:105234. 10.1016/j.catena.2021.105234.Search in Google Scholar

[62] Six J, Paustian K, Elliott ET, Combrink C. Soil structure and organic matter I. distribution of aggregate-size classes and aggregate-associated carbon. Soil Sci Soc Am J. 2000;64:681–9. 10.2136/sssaj2000.642681x.Search in Google Scholar

[63] Denef K, Six J. Clay mineralogy determines the importance of biological versus abiotic processes for macroaggregate formation and stabilization. Eur J Soil Sci. 2005;56:469–79. 10.1111/j.1365-2389.2004.00682.x.Search in Google Scholar

[64] Zimmermann M, Leifeld J, Schmidt MWI, Smith P, Fuhrer J. Measured soil organic matter fractions can be related to pools in the RothC model. Eur J Soil Sci. 2007;58:658–67. 10.1111/j.1365-2389.2006.00855.x.Search in Google Scholar

Received: 2023-06-08

Revised: 2023-10-12

Accepted: 2023-11-16

Published Online: 2023-12-22

This work is licensed under the Creative Commons Attribution 4.0 International License.

Supplementary material

Articles in the same Issue

https://doi.org/10.1515/geo-2022-0585

Keywords for this article

selective erosion; particle size distribution; rainfall simulation; aggregate size distribution

Creative Commons

BY 4.0

Varying particle size selectivity of soil erosion along a cultivated catena

Article

Abstract

1 Introduction

2 Materials and methods

2.1 Study site and soil characteristics

2.2 Rainfall simulation

2.3 Experimental design

2.3.1 Surface state description

2.3.2 Sampling strategy

2.4 Aggregate size separation and soil loss calculation

2.5 PSD measurements

2.6 Statistical analysis

3 Results and discussion

3.1 ASD of soil loss

3.2 Transportation based on aggregate class sizes

4 Conclusion

Acknowledgements

References

Supplementary Material

Articles in the same Issue

Articles in the same Issue

Articles in the same Issue