Reconstruction of Seasonal Net Erosion in a Mediterranean Landscape (Alento River Basin, Southern Italy) over the Past Five Decades

: In the low Mediterranean basin, late spring and autumn rainfall events have the potential to increase discharge and transport substantial amounts of sediment soil (that is, the net soil erosion from a watershed). For the Alento River Basin (ARB), located in the low Tyrrhenian coast of Italy, we estimated changes of net erosion as dependent on the seasonality of antecedent soil moisture and its control on rainfall-runo ﬀ and erosivity. Based on rainfall and runo ﬀ erosivity sub-models, we developed a simpliﬁed model to evaluate basin-wide sediment yields on a monthly basis by upscaling point rainfall input. For the period 1951–2018, the reconstruction of a time series of monthly net erosion data indicated a decreasing trend of the sediment yield after 1991. Revegetation and land abandonment that occurred in the last decades can explain such a decrease of net erosion, which occurred even when rainfall erosivity increased. This response, obtained at the basic scale, does not exclude that rapidly developing mesoscale convective systems, typically responsible for the heaviest and most destructive rainfall events in the ARB, can a ﬀ ect small catchments, which are the most vulnerable systems to storm-driven ﬂash ﬂoods and soil erosion hazards during soil tilling in spring and at beginning of autumn.


Introduction
Environmental changes are a prominent topic for Earth and environmental sciences, but its importance increases during crucial changes and different types of climate extremes that potentially lead to crises of some kind [1,2]. Extreme climate events are often associated with land degradation [3,4]. Soil erosion, in particular, is a pervasive form of soil degradation and a matter of increasing concern because of its implications for food security with the rapidly increasing world population [5]. Modeling processes that produce geomorphological hazards require understanding of how landscape components respond to forced conditions of land use change and to the climatic regime [6,7]. This is valuable to inform the assessment of future planning [8,9], but soil erosion monitoring systems tracking downstream sediment movement may be costly, and require focused efforts to manage land and water resources [10]. Because of this cost, modeling is playing an increasingly significant role [11]. This applies to the quantification of sediment dynamics, which is key to Earth-system science as documented in geology [12], biogeochemistry [13], and human activities [14]. It is also key to advancing our quantitative understanding and predictive capabilities of regional and sub-regional sediment fluxes. In the last decade, for instance, some geomorphological studies of long-term scale have affected the  River morphology is complex in the study region. The upstream part of the basin presents a narrow alluvial valley with steep slopes. Downstream, however, the river makes a turn towards the south. Subsequently, the river assumes a braided configuration down until the reservoir of Piano della Rocca, in the commune of Prignano Cilento (40 • 20 N, 15 • 04 E). Further downstream, the river mostly takes a meandering character. The geological nature of the rocks is dominated by the "Flysch of the Cilento" (i.e., limestone and silicoclastic substrata), wherein the main river basins (Alento, Calore, Mingardo, Bussento) are established [32]. Overall, the basin area is not prone to gully erosion, as it is dominated by erosion-resistant lithologies. With the only exception of the far northern/north-eastern part of the catchment, which is characterized by the presence of limestones pertaining to the Apennine Chain, these formations are quite homogeneous in hydrogeological terms and may be merged into a single hydrogeological complex of arenaceous-marly-clayey formation, which is relatively poorly permeable. River morphology is complex in the study region. The upstream part of the basin presents a narrow alluvial valley with steep slopes. Downstream, however, the river makes a turn towards the south. Subsequently, the river assumes a braided configuration down until the reservoir of Piano della Rocca, in the commune of Prignano Cilento (40°20′ N, 15°04′ E). Further downstream, the river mostly takes a meandering character. The geological nature of the rocks is dominated by the "Flysch of the Cilento" (i.e., limestone and silicoclastic substrata), wherein the main river basins (Alento, Calore, Mingardo, Bussento) are established [32]. Overall, the basin area is not prone to gully erosion, as it is dominated by erosion-resistant lithologies. With the only exception of the far northern/northeastern part of the catchment, which is characterized by the presence of limestones pertaining to the Apennine Chain, these formations are quite homogeneous in hydrogeological terms and may be merged into a single hydrogeological complex of arenaceous-marly-clayey formation, which is relatively poorly permeable.

Data Collection
Daily rainfall data for the period 1951-2000 were collected from the rain gauge network of the Servizio Idrografico and Mareografico Nazionale (SIMN, National Hydrographic and Marine Service) [   average value was available [34]. Under this limited calibration condition, credibility of final output estimates was founded on the sub-model validation. The model was thus calibrated against long-term average net erosion and then validated for its erosivity and runoff submodels, using monthly-aggregated data, as determined in the ARB from the sub-periods 2002-2008
Monthly vegetation cover fraction was assessed with Normalized Difference Vegetation Index (NDVI) data, as derived from the GIMMS-KNMI Climate Explorer platform (http://climexp.knmi.nl), and rearranged to characterize the inter-annual evolution [16]. Olive orchards and sclerophyllous Mediterranean vegetation prevail along the coast, whereas forest landscape is dominant in the inner area, mainly represented by Quercus cerris or Fagus sylvatica woods. Not negligible is also the presence of riparian forest cover, dominated by Salix alba, Populus nigra, Populus alba, and Alnus glutinosa. Smallholder agriculture (arable land and orchards) dominates, sustained by mechanization, road infrastructure, availability of groundwater stocks, and water storage for irrigation purposes.

Net Erosion Model
Net soil water erosion is a measure of average sediment yield (soil net erosion) occurring basin-wide over time (Figure 3a), resulting from the sum of the sediment produced by all erosional sources, including overland flow, ephemeral gully, and stream channel areas [28], minus the amount of sediment deposited on such transfer zones and on the valley floodplains. The result is the amount of sediment conveyed downstream to the outlet of the basin. Four environmental factors determine the amount of water erosion and sedimentation. They are climate, soil, topography, and land-use, which operate independently and interactively. Basic characteristics and spatio-temporal features are thus taken into account in a hierarchical structure for discovering erosional phenomenon. In particular, the evolution over time of net erosion reflects the magnitude and frequency of individual storm events, which are nested within larger events occurring on different time scales [21].
(NDVI) data, as derived from the GIMMS-KNMI Climate Explorer platform (http://climexp.knmi.nl), and rearranged to characterize the inter-annual evolution [16]. Olive orchards and sclerophyllous Mediterranean vegetation prevail along the coast, whereas forest landscape is dominant in the inner area, mainly represented by Quercus cerris or Fagus sylvatica woods. Not negligible is also the presence of riparian forest cover, dominated by Salix alba, Populus nigra, Populus alba, and Alnus glutinosa. Smallholder agriculture (arable land and orchards) dominates, sustained by mechanization, road infrastructure, availability of groundwater stocks, and water storage for irrigation purposes.

Net Erosion Model
Net soil water erosion is a measure of average sediment yield (soil net erosion) occurring basinwide over time (Figure 3a), resulting from the sum of the sediment produced by all erosional sources, including overland flow, ephemeral gully, and stream channel areas [28], minus the amount of sediment deposited on such transfer zones and on the valley floodplains. The result is the amount of sediment conveyed downstream to the outlet of the basin. Four environmental factors determine the amount of water erosion and sedimentation. They are climate, soil, topography, and land-use, which operate independently and interactively. Basic characteristics and spatio-temporal features are thus taken into account in a hierarchical structure for discovering erosional phenomenon. In particular, the evolution over time of net erosion reflects the magnitude and frequency of individual storm events, which are nested within larger events occurring on different time scales [21].  Figure 3b outlines the role played in sediment transport by mesoscale rainstorms accounted at the basin scale (BGE, basin gross erosion), while also assuming that the distribution of local showers play an important role in determining torrential flows rich in sediment in the individual river catchments of the basin (CGE, catchment gross erosion).
The model structure suggests that spring-summer (May to September) precipitation is an important factor to estimate the relative contribution of individual catchments (upper tributaries river) to the sediment (CGE) moving within the basin drainage system. In contrast, winter precipitation mostly contributes to basin-wide transient response (BGE within lower tributary river).  Figure 3b outlines the role played in sediment transport by mesoscale rainstorms accounted at the basin scale (BGE, basin gross erosion), while also assuming that the distribution of local showers play an important role in determining torrential flows rich in sediment in the individual river catchments of the basin (CGE, catchment gross erosion).
The model structure suggests that spring-summer (May to September) precipitation is an important factor to estimate the relative contribution of individual catchments (upper tributaries river) to the sediment (CGE) moving within the basin drainage system. In contrast, winter precipitation mostly contributes to basin-wide transient response (BGE within lower tributary river).
Since the procedure for determining rainfall erosivity suggested by Wischmeier and Smith [36] is applicable to the computation of annual erosion, its use to estimate soil loss from single storms would imply considerable errors [37] and motivate a reinterpretation of the original formulation. Foster et al. [30] and Thornes [31] elaborated the concept of the balance between driving and resisting forces in sediment budget. We further arranged this solution to model net erosion on a monthly basis (NETAM, Mg km −2 month −1 ) as: where the term within round brackets is the modified Foster algorithm; R S is the rainfall-erosivity indicator associated with splash erosion; R Q (mm) is the runoff term, associated with transport erosion; S (m −1 ) is the mean slope of the basin; the erodibility coefficient (lithology factor) k = 0.0145 and the shape parameters (which play an adjusting role on the model inputs) n = 2 and m = 2 were arranged from Wainwright and Mulligan [11], and u was determined by calibration; exp(-ν·VCF), with ν = 0.07, is the exponential vegetation function [31], with VCF (%) being the vegetation cover fraction [16]; A and α are erosivity scale coefficients, whose values were determined by calibration. Our approximation is that hydraulically rough and vegetated surfaces reduce flow velocity and, hence, soil interril transport capacity [28]. This is reflected in the low values attributed to parameters k and ν in Equation (1). Then, as canopy cover reduces soil detachment caused by raindrop impact, it also reduces interrill sediment transport capacity by attenuating raindrop impact. Based on this understanding, the power of rainfall as prevailing storm erosivity in summer and autumn is captured by the daily rainfall term of R S , while in winter and spring, runoff is captured by the monthly rainfall terms of R Q . In the ARB, predominant water erosion derives from interactions between the detachment on hillslope areas caused by water drops falling on soil, and successive runoff towards downslope up to flow in the drainage networks. This linkage of processes occurs within a fluctuating and continuous interplay of disturbing and resistance forces. In this way, soil erosion by water mostly occurs when the detachment of particles and their subsequent transportation experience a greater driving force than the force binding particles into the vegetated slope. With all these processes, rainfall is used by nature as both a driving and a resisting factor. To better detail this, firstly the erosive influence of rainfall increases with water amount, intensity, and runoff; secondly, and opposing this influence, the protective effect of vegetation increases with precipitation amount.
To further explain the single terms of Equation (1), arranging from Diodato and Aronica [38], we obtain: where dx is the daily maximum rainfall (mm) in each j month; the scale-factor f (j m ) is as follows: The semi-parametric function f (j m ) modulates the intra-seasonal storm intensity proxy during rainfalls.
The following R Q term represents, instead, the erosivity mostly associated with runoff erosion: where p is the amount of rainfall (mm) in the current month and p j−1 (mm) is the rainfall in the previous month; w is an indicator of soil humidity, in the form of a semi-parametric function, to modulate the intra-seasonal humidity after precipitation:

Results and Discussion
Overall results of the calibrated model ) and sub-model (rainfall erosivity and runoff) validation are first presented, followed by the model-based reconstruction of net erosion data for the period 1951-2018. The long-term trend is discussed at the annual scale, before highlighting the net erosion variability at the monthly scale.

Model Calibration
For the calibration period 1951-1990, over which long-term annual mean of net erosion data was available for the ARB, the values of the coefficients u = 2, A = 1500 (which converts values of eroded soil from mm to Mg km −2 ), and α = 0.1 in Equation (1) were obtained by approximating the model output to the silting value determined experimentally from the degree of filling of the dam of Piano della Rocca (490 Mg km −2 year −1 ), covering 24% of the entire basin [34]. If the figure of 490 Mg km −2 year −1 for the period 1951-1990 is extrapolated for the whole of the basin, then the overall erosion rate calculates to 2042 Mg km −2 year −1 . The calibrated estimate was 2041 Mg km −2 year −1 for the same period.

Semi-Quantitative Validation
To ensure that the model serves its intended purpose, a semi-quantitative verification with inter-monthly variability was done, since sub-models of Equations (2) and (4) do not include any scale parameter. Figure 4 shows the performance of these sub-models. In particular, Figure 4a displays that the rainfall-erosivity component is in agreement with RUSLE-based erosivity data [35]. Figure 4b also reflects a satisfactory performance between predicted and actual runoff data [33].
Overall results of the calibrated model ) and sub-model (rainfall erosivity and runoff) validation are first presented, followed by the model-based reconstruction of net erosion data for the period 1951-2018. The long-term trend is discussed at the annual scale, before highlighting the net erosion variability at the monthly scale.

Model Calibration
For the calibration period 1951-1990, over which long-term annual mean of net erosion data was available for the ARB, the values of the coefficients u = 2, A = 1500 (which converts values of eroded soil from mm to Mg km −2 ), and α = 0.1 in Equation (1) were obtained by approximating the model output to the silting value determined experimentally from the degree of filling of the dam of Piano della Rocca (490 Mg km −2 year −1 ), covering 24% of the entire basin [34]. If the figure of 490 Mg km −2 year −1 for the period 1951-1990 is extrapolated for the whole of the basin, then the overall erosion rate calculates to 2042 Mg km −2 year −1 . The calibrated estimate was 2041 Mg km −2 year −1 for the same period.

Semi-Quantitative Validation
To ensure that the model serves its intended purpose, a semi-quantitative verification with intermonthly variability was done, since sub-models of Equations (2) and (4) do not include any scale parameter. Figure 4 shows the performance of these sub-models. In particular, Figure 4a displays that the rainfall-erosivity component is in agreement with RUSLE-based erosivity data [35]. Figure  4b also reflects a satisfactory performance between predicted and actual runoff data [33]. This indicates that, at basin scale, net erosion is not the result of the runoff amount only, but of the combination of rainfall erosivity by both raindrop impact and surface runoff. As well, vegetation covers the soil during several months, over which erosion patterns may change [39].
In Figure 5, the model appears to correctly compute the main effects and trends associated with sediment yield, represented in this case by the sand extracted every year (proxy of the net erosion) in the Alento valley [40]. We evaluated the relative performance of the NETAM, without comparing the absolute estimates. Coevolution between material extracted at the valley of ARB (histogram) and simulated net erosion (blue curve) illustrates a substantial agreement, with the only exception of around 1985 (corresponding to the beginning of the construction of the dam in 1984 [34]). This indicates that, at basin scale, net erosion is not the result of the runoff amount only, but of the combination of rainfall erosivity by both raindrop impact and surface runoff. As well, vegetation covers the soil during several months, over which erosion patterns may change [39].
In Figure 5, the model appears to correctly compute the main effects and trends associated with sediment yield, represented in this case by the sand extracted every year (proxy of the net erosion) in the Alento valley [40]. We evaluated the relative performance of the NETAM, without comparing the absolute estimates. Coevolution between material extracted at the valley of ARB (histogram) and simulated net erosion (blue curve) illustrates a substantial agreement, with the only exception of around 1985 (corresponding to the beginning of the construction of the dam in 1984 [34]).  Figure 6 shows the temporal evolution of annual sediment exports from ARB during the period 1951-2018, as calculated with Equation (1). Part of the estimated sediment was trapped by the dam built in 1994. However, no refinement was brought to the original dataset, as the sediment trapped in the reservoir is still the effective erosion that occurred from the several catchments composing the basin. Figure 6a, in particular, shows the actual evolution of net erosion that, after the first years with low erosional rate, reveals an increase from 1960 at a roughly constant trend that extended until 1990, before the change-point detected in 1991 (Figure 6c) with the Mann-Whitney-Pettitt test [41]. After this year, the sediment rate underwent a continuous irregular decrease until the end of the timeseries.   Figure 6 shows the temporal evolution of annual sediment exports from ARB during the period 1951-2018, as calculated with Equation (1). Part of the estimated sediment was trapped by the dam built in 1994. However, no refinement was brought to the original dataset, as the sediment trapped in the reservoir is still the effective erosion that occurred from the several catchments composing the basin.  Figure 6 shows the temporal evolution of annual sediment exports from ARB during the period 1951-2018, as calculated with Equation (1). Part of the estimated sediment was trapped by the dam built in 1994. However, no refinement was brought to the original dataset, as the sediment trapped in the reservoir is still the effective erosion that occurred from the several catchments composing the basin. Figure 6a, in particular, shows the actual evolution of net erosion that, after the first years with low erosional rate, reveals an increase from 1960 at a roughly constant trend that extended until 1990, before the change-point detected in 1991 (Figure 6c) with the Mann-Whitney-Pettitt test [41]. After this year, the sediment rate underwent a continuous irregular decrease until the end of the timeseries.   Figure 6a, in particular, shows the actual evolution of net erosion that, after the first years with low erosional rate, reveals an increase from 1960 at a roughly constant trend that extended until 1990, before the change-point detected in 1991 (Figure 6c) with the Mann-Whitney-Pettitt test [41]. After this year, the sediment rate underwent a continuous irregular decrease until the end of the time-series.

Annual Net Erosion Reconstruction
Over the first period, 1951-1990, the average estimated net erosion value is of 2041 Mg km −2 year −1 (± 889 Mg km −2 year −1 standard deviation), while in the last period, 1991-2018, the average value stands at 568 (±436 Mg km −2 year −1 standard deviation), with a marked decrease of 1473 Mg km −2 year −1 compared to the previous period. This decrease is also accompanied by an amplification of the interannual variation coefficient of net erosion, which passes from 0.42 for the period before the change-point, to 0.58 for the following period. The growing seasonal irregularity of the precipitation is probably the major driver of the increasing interannual variability of soil erosion. Over 1991-1998, forest cover doubled and cropland roughly halved due to decades of land abandonment and reduction of human pressure [42], and this is likely the cause of decrease in net erosion during the period 1991-2018, although rainfall-erosivity kept on rising (Figure 6b). Thus, vegetation cover exerted a great resistance to the hydrological hazard, since vegetation underwent a general increase after 1990 [16]. However, during the most extremes hydrological events (e.g., precipitation at hours or sub-hourly scales), soil erosion in small catchments could represent a large risk for soil mobilization and transport, which can contribute to nutrient and organic carbon losses.

Net Erosion Monthly Variability and Timing
The use of monthly data helps summing up consecutive hydro-geomorphological events over an appropriate time scale, with respect to hydrological timing and crop growing cycles or scheduling of tillage practices. The modeled results, obtained over 68 years, show that there is a significant variation of sediment transport at the intra-seasonal scale in both past   (Figure 7a) and recent (1991-2018) periods (Figure 7b, grey bars). Almost half of suspended solid transport occurs in autumn (43% and 54%, respectively, in the two periods) and approximately one third of the annual flux occurs in winter (39% and 29%, respectively). However, the 95th percentile (Figure 7, empty bars) is distributed differently than to mean values, with more divergence in April and September (Figure 7, red bars). This divergence represents a high risk of soil erosion in correspondence to the months with tilled soil in both the periods, although results evidence a decrease of net erosion in all months during the recent phase 1991-2018 (Figure 7b). Over the first period, 1951-1990, the average estimated net erosion value is of 2041 Mg km −2 year −1 (± 889 Mg km −2 year −1 standard deviation), while in the last period, 1991-2018, the average value stands at 568 (±436 Mg km −2 year −1 standard deviation), with a marked decrease of 1473 Mg km −2 year −1 compared to the previous period. This decrease is also accompanied by an amplification of the interannual variation coefficient of net erosion, which passes from 0.42 for the period before the change-point, to 0.58 for the following period. The growing seasonal irregularity of the precipitation is probably the major driver of the increasing interannual variability of soil erosion. Over 1991-1998, forest cover doubled and cropland roughly halved due to decades of land abandonment and reduction of human pressure [42], and this is likely the cause of decrease in net erosion during the period 1991-2018, although rainfall-erosivity kept on rising (Figure 6b). Thus, vegetation cover exerted a great resistance to the hydrological hazard, since vegetation underwent a general increase after 1990 [16]. However, during the most extremes hydrological events (e.g., precipitation at hours or sub-hourly scales), soil erosion in small catchments could represent a large risk for soil mobilization and transport, which can contribute to nutrient and organic carbon losses.

Net Erosion Monthly Variability and Timing
The use of monthly data helps summing up consecutive hydro-geomorphological events over an appropriate time scale, with respect to hydrological timing and crop growing cycles or scheduling of tillage practices. The modeled results, obtained over 68 years, show that there is a significant variation of sediment transport at the intra-seasonal scale in both past   (Figure 7a) and recent (1991-2018) periods (Figure 7b, grey bars). Almost half of suspended solid transport occurs in autumn (43% and 54%, respectively, in the two periods) and approximately one third of the annual flux occurs in winter (39% and 29%, respectively). However, the 95th percentile (Figure 7, empty bars) is distributed differently than to mean values, with more divergence in April and September ( Figure  7, red bars). This divergence represents a high risk of soil erosion in correspondence to the months with tilled soil in both the periods, although results evidence a decrease of net erosion in all months during the recent phase 1991-2018 (Figure 7b). The autumn season seems to maintain the primacy of erosion rates, in past as in recent times. Rizzi [43] documented disasters in the Alento coast in autumn and winter during past times.
In winter, rainfall and average sediment are significant, but most of the erodible particles are transported by the first floods of the preceding autumn. Spatial timeline of storminess also shows a decadal trend (Figure 8). The increasing trend has affected practically the entire basin, especially for The autumn season seems to maintain the primacy of erosion rates, in past as in recent times. Rizzi [43] documented disasters in the Alento coast in autumn and winter during past times.
In winter, rainfall and average sediment are significant, but most of the erodible particles are transported by the first floods of the preceding autumn. Spatial timeline of storminess also shows a decadal trend (Figure 8). The increasing trend has affected practically the entire basin, especially for storms of 24-h duration (Figure 8b). Then, the increased variability and amount of storms found at the Gioi Cilento station can affect the areas around the station. In particular, it is understood that an average increase of 10 mm per half a century affected the storms of 1-h duration (Figure 8a), and 10-20 mm the storms of 24-h duration (Figure 8b). The areas of the basin more interested from storm increases are those included along the transect zone around the villages of Cicerale (40 •  storms of 24-h duration (Figure 8b). Then, the increased variability and amount of storms found at the Gioi Cilento station can affect the areas around the station. In particular, it is understood that an average increase of 10 mm per half a century affected the storms of 1-h duration (Figure 8a), and 10-20 mm the storms of 24-h duration (Figure 8b). The areas of the basin more interested from storm increases are those included along the transect zone around the villages of Cicerale (40°21 N, 15°08′ E), Gioi Cilento (40°17 N, 15°13′ E), and Vallo della Lucania (40°14 N, 15°16′ E).

Conclusions
Land use change has been recognized throughout the world as an important driver of climatedriven geomorphological processes, which may also trigger changes in carbon cycling [44]. Soil erosion rates may be expected to change in response to changes in climate and vegetation for a variety of reasons, the most direct of which is the change in the erosive power of rainfall and resistance forces, respectively. However, modeling rainfall-driven soil erosion rates is difficult because of the lack of long-term data in river basins. In particular, complex models are often not adequate to reconstruct net erosion (or sediment yield) changes because they require a considerable amount of highresolution input data, not always available on long timescales. Thus, the use of parsimonious models offers an interesting possibility to reconstruct net erosion series on a monthly basis. This is what we have done with the NETAM, developed on the original Foster and Thornes algorithms, in a test site, the Alento River Basin (~400 km 2 in Southern Italy). Though the model developed for the ARB is not easily transferable for applications in other basins, it provided a peculiar and unique opportunity for modeling erosion responses to climate and land cover changes, where documented hydrological processes at basin scale also support input-data generation and interpretation of results. The ARB is a catchment with extensive natural areas. The development of agricultural and natural areas is favored by the presence of farming practices and a markedly seasonal climate. Thanks to the continuous observation of selected physical environmental variables, we were able to establish seasonal patterns of weathering processes and identify the factors that control rainfall erosivity and

Conclusions
Land use change has been recognized throughout the world as an important driver of climate-driven geomorphological processes, which may also trigger changes in carbon cycling [44]. Soil erosion rates may be expected to change in response to changes in climate and vegetation for a variety of reasons, the most direct of which is the change in the erosive power of rainfall and resistance forces, respectively. However, modeling rainfall-driven soil erosion rates is difficult because of the lack of long-term data in river basins. In particular, complex models are often not adequate to reconstruct net erosion (or sediment yield) changes because they require a considerable amount of high-resolution input data, not always available on long timescales. Thus, the use of parsimonious models offers an interesting possibility to reconstruct net erosion series on a monthly basis. This is what we have done with the NETAM, developed on the original Foster and Thornes algorithms, in a test site, the Alento River Basin (~400 km 2 in Southern Italy). Though the model developed for the ARB is not easily transferable for applications in other basins, it provided a peculiar and unique opportunity for modeling erosion responses to climate and land cover changes, where documented hydrological processes at basin scale also support input-data generation and interpretation of results. The ARB is a catchment with extensive natural areas. The development of agricultural and natural areas is favored by the presence of farming practices and a markedly seasonal climate. Thanks to the continuous observation of selected physical environmental variables, we were able to establish seasonal patterns of weathering processes and identify the factors that control rainfall erosivity and runoff and, in turn, net erosion. Cold and wet cycles in winter and wet and dry cycles in spring-autumn are the main processes involved in landscape weathering, thereby controlling slope development together with rainfall-related erosion processes. The main observed feature is the reaction of the ARB to all rainfall events. Hydrological events show high fluctuations of the suspended sediment by month-to-month, and by year-to-year, deriving from a heterogeneous temporal distribution related to seasonal variations of the hydro-climatic forcing (that is, surface erosivity and runoff) and the vegetation cover. In this way, NETAM values were obtained for the period 1951-2018 by using parsimonious erosion sub-models and land cover statistics from documented agrarian sources. We conclude that if pulses of sediment fluctuation in the ARB have always been driven mainly by natural climatic oscillations, then land abandonment and revegetation are the causes of the observed reduction of net soil erosion in the last decades. This study adds to a growing body of literature on the development of methodological frameworks and tools that could be used to outline scenarios of soil erosion and instability risks resulting from climate changes (e.g., increasing heavy rainfall events), and changes in land use and management practices in central Italy [45][46][47].
Author Contributions: N.D. conceived the study, performed the analysis, and drafted the manuscript. G.B. contributed to the analysis and wrote the final manuscript.
Funding: This is an investigator-driven study without any grant support.