A,detailed,reconstruction,of,changes,in,the,factors,and,parameters,of,soil,erosion,over,the,past,250,years,in,the,forest,zone,of,European,Russia(Moscow,region)

Anrey Zhikin ,Dri Fomihev ,Nezh Ivnov ,Tom′ˇs Dost′l ,All Yurov ,Mikhil Komissrov ,Josef Kr′s

a V.V.Dokuchaev Soil Science Institute,Pyzhevskiy Pereulok 7,Moscow,119017,Russian Federation

b Faculty of Geography,Lomonosov Moscow State University,Leninskie Gory,GSP-1,Moscow,119991,Russian Federation

c Department of Landscape Water Conservation,Faculty of Civil Engineering,Czech Technical University in Prague,Th′akurova 7,Prague,16629,Czech Republic

d Ufa Institute of Biology UFRC,Russian Academy of Sciences,Pr.Oktyabrya 69,Ufa,450054,Russian Federation

Keywords:Anthropogenic soil erosion Soil erosion history Crop rotation Magnetic tracer method WaTEM/SEDEM

A B S T R A C T Accelerated soil erosion is a major threat to soil,and there are great variations in the rate of soil erosion over time due to natural and human-induced factors.The temperate forest zone of Russia is characterized by complex stages of land-use history(i.e.active urbanization,agricultural development,land abandonment,etc.).We have for the first time estimated the rates of soil erosion by the WaTEM/SEDEM model(rainfall erosion)and by a regional model(snowmelt erosion)over the past 250 years(from 1780 to 2019)for a 100-km2 study site in the Moscow region of Russia.The calculations were made on the basis of a detailed historical reconstruction of the following factors:the location of the arable land,crop rotation,the rain erosivity factor,and the maximum snow water equivalent.The area of arable land has decreased more than 3.5-fold over the past 250 years.At the end of the 20th century,the rates of gross erosion had declined more than 5.5-fold(from 28×103 to 5×103 tˑha-1ˑyr-1)in comparison with the end of the 18th century.Changes in the boundaries of arable land and also the relief features had led to a significant intra-slope accumulation of sediments.As a result of sediment redeposition within the arable land,the variation in net soil erosion was significantly lower than the variation in gross soil erosion.The changes in arable land area and in crop composition are the factors that have to the greatest extent determined the changes in soil erosion in this territory.

Soil erosion is the gravest threat among all soil degradation processes(Montanarella et al.,2016),and it mainly occurs as a combination of natural and human-induced soil erosion processes(Poesen,2018).Fluctuations in these factors lead to spatio-temporal changes in erosion-accumulative processes.Studies of changes in soil erosion over time have evolved in recent decades,and most of these studies have shown the impacts of climate change on soil erosion(Sobol et al.,2015;Li&Fang,2016).More than 250 papers on this topic were published in 2014(Li & Fang,2016).A much smaller number of studies analyzed the impact of changes in landuse on the development of soil erosion.These studies were conducted in Hungary(Jordan et al.,2005;Szilassi et al.,2006),in Slovakia(Smetanova et al.,2017),in Austria and the Czech Republic(Dev′atýet al.,2019),in France(Foucher et al.,2014),in Greece and Macedonia(Rothacker et al.,2018),in the European Union(Panagos et al.,2015),in Ukraine(Lisetskii,1991;Kovalchuk,1997,p.440),on the border of the former Soviet Union(Wuepper et al.,2020),and in Russia(Sidorchuk,1995;Lisetskii & Pichura,2020;Litvin et al.,2017;Spatio-temporal patterns,2019;Zhidkin et al.,2016).These studies also pointed to the use of soil conservation technologies that affected changes in land-use and in land-cover.Reductions in soil losses were promoted by land abandonment,by afforestation,by the use of perennial grasses or special crop sequences,and by tillage practices(i.e.No-till,Strip-till,conservation treatment).It is very difficult to compare the results of these studies,since the investigations were carried out on various time scales(from one decade to 11,500 years),in different areas(from small catchments of 5—6 km2to the whole area of the EU and the European part of Russia)and were carried out with the use of various methods(modeling,analysis of sediment depositions,tracers,etc.)Nevertheless,each of these studies is of great importance for soil science,since the future development of soil erosion processes needs to be based on knowledge and data acquired in the course of research work carried out in the past.

The variation in rates of soil erosion caused by changes in landuse has displayed specific features for different territories/zones in the European part of Russia(Spatio-temporal patterns,2019).The most significant differences are noted for the northern parts(forest zone)and the southern parts(steppe and forest-steppe zones)of Russia.In the southern part,the most intensive land development took place in the last third of the 19th century and in the early 20th century,after which period the area of cultivated land changed slightly.In turn,the maximum plowing in the central provinces of European Russia,located in the forest zone,took place after the“abolition of serfdom”(1861—1868).In the last 150 years,there has been a significant reduction in arable land area(Spatio-temporal patterns,2019).

Unfortunately,most of the studies focused on historical changes in soil erosion have been carried out for the southern regions,in the steppe and forest-steppe zones(Sidorchuk,1995;Zhidkin et al.,2016;Litvin et al.,2017;Spatio-temporal patterns,2019;Lisetskii& Pichura,2020).Only one study(Litvin et al.,2017)has made a detailed analysis of changes in soil erosion for the forest zone in European Russia.However,the research carried out by Litvin et al.(2017)covered only a very short period of time(1980—2014).Significant political and economic transformations(the collapse of the USSR)took place during this period,which resulted in a reduction in the area of arable land(especially in the forest zone).The intensity of soil erosion on arable lands in the European territory of Russia decreased by 15%.In the forest zone it decreased by 44%,while in the steppe zone it increased by 19% between 1980 and 2014.According to Tsymbarovich et al.(2020),the decrease in the volume and in the rate of soil erosion in Russia over the past 30—40 years is connected not only with the abandonment and overgrowing of arable land,but also with certain climatic trends,such as a decrease in the maximum depth of soil freezing,in the maximum flow of spring runoff,and in the thickness of the layers of flood runoff.

The work presented here aims to provide the first study of changes in soil erosion in the temperate forest zone of Russia(on a key site in the Moscow region)over the last 250 years.The most significant tasks were:i)to make a quantitative assessment of the factors of soil erosion and of the change in these factors over the past 250 years on the study site;ii)to calculate the rates of soil erosion for different periods of time on the study site;and iii)to analyze the causes and the trends of the changes in soil erosion.

2.1.Study site description

The study site is located in the Moscow Region,40 km north of the city of Moscow,in the southern temperate broadleaf and mixed forest zone(Fig.1A),between the Uchinskoye Reservoir and its northern tributaries(Fig.1B).The total area of the study site is about 100 km2(Fig.1C).The relief is a hilly moraine plain(Fig.1C)within the southern slope of the Klinsko-Dmitrov Ridge.According to Sorokina and Kozlov(2009),forests with a predominance of coniferous species and forests of secondary succession currently occupy 60% of the territory.The remaining lands are occupied by arable land(35%),meadows and swamps(5%).The parent rocks are represented by a clay loam mantle,and the soil cover is mainly represented by Umbric Albeluvisols(Sorokina & Kozlov,2009).

This area is a typical suburban landscape,which was formerly used mainly for agriculture.The land-use in the territory has a complex history;the first mention in the annals of the area dates back to year 1501,in connection with the construction of a church in the village of Eldigino(in the central part of the study site).Agriculture in this territory therefore began about 500 years ago.The study site is representative of the Klinsko-Dmitrov Ridge area in terms of land-use history,relief,and other soil erosion factors.

2.2.Model selection

Mathematical modeling is the most widely-used method for assessing the effects on soil erosion of historical changes in landuse.Other methods are difficult to use for a long period of a few hundred years.For the study area,there was only a very limited choice of methods for assessing the soil erosion.Unfortunately,it would not have been correct to use traditional methods(soilmorphological,radio-cesium,correlative sediments,etc.)at the study site.There had been no monitoring observations of sediment runoff in this region.The method of erosion modeling was therefore chosen,and the verification of the calculated rates of erosion on key slopes was based on the magnetic tracer method.

Soil erosion in the study area is caused mainly by rainfalls and to a slight extent by snowmelt.Rainfall erosion was estimated by the WaTEM/SEDEM model(Van Oost et al.,2000;Van Rompay et al.,2001).The calculation algorithm is based on RUSLE(Renard et al.,1997):

where:A denotes the annual soil losses(tˑha-1yr-1);R is the rain erosivity factor(R-factor)(MJˑmmˑha-1ˑh-1ˑyr-1);K is the soil erodibility factor(K-factor)(tˑhaˑhˑha-1ˑMJ-1ˑmm-1);LS is the topographical slope and length factor(LS-factor);C is the covermanagement factor(C-factor);and P is the erosion control factor(P-factor).The P-factor was not taken into consideration in this study,due to its lack of relevance for the study area.

There are very few models of soil erosion caused by snowmelt,by contrast with the great number of models of soil erosion caused by rainfall(Karydas et al.,2014).We decided to use the Russian regional model of the State Hydrological Institute(Instructions for determination,1979,p.62)in the Larionov modification(Larionov,1993).The model algorithm is based on the following equation:

where:As represents the annual snowmelt soil losses(tˑha-1yr-1);f is a function of H,i.e.the slope snowmelt runoff layer(mm);K is the soil erodibility factor;L and S is the topographical slope and length factor(dimensionless).

The maps of rainfall soil erosion(Fig.5)and snowmelt soil erosion(Fig.6),and also a map of study area(Fig.1)and the period of time for which the soil was under the plow(Fig.4)were created using the ArcGIS 10.1 program.

Fig.1.Location of the Moscow region in the European part of Russia(a);location of the study site in the Moscow region(b);elevation map of the study site and the locations of the soil sampling points(c).

2.3.Data sources and parameterization of the model

2.3.1.Land-use structure

The first cartographic information about land-use on the study site appeared at the end of the 18th century.The land in the Moscow region was surveyed in the 1770s and 1780s within the framework of a general land surveying exercise.The maps of the general land survey were created using non-instrumental methods.The accuracy is therefore low,which makes the maps difficult to rectify and interpret(Goldenberg & Postnikov,1985).

The selection criteria for subsequent maps were:a large scale and a satisfactory display of the borders of the arable land,and a sufficient time period between the years in which the maps were created.The maps were downloaded in the form of scanned images with a geographic reference from the ETOMESTO(2019)website.A list of maps and an estimate of the accuracy of the referencing and digitizing land-use boundaries on the maps is presented in Table 1.

Digitization of the boundaries of the arable land was carried out on the scale of maps(Table 1).The minimum mapping unit of landcover was also assessed in accordance with the information on themap.Digitization was carried out only on the boundaries of arable land.Some maps showed treeless areas that were not divided into arable land and pasture.In this case,an area without forests and swamps was classified as arable land.This assumption is permissible,since deforestation was mainly associated with plowing in the region.The number of livestock in the Moscow region was 10—20 head per 100 people at the middle of the 19th century(Militarystatistical collection,1867).This is several times lower than in the Tula region and in more southern regions.The pastures in the Moscow region were therefore not numerous and,in general,they were confined to periodically abandoned or fallow lands.

Table 1Maps used to determine land-use boundaries and their characteristics.

The time periods were allocated according to the dates of the historical maps,and also in accordance with significant historical changes.The following periods were distinguished:1780—1830;1831—1881;1882—1927;1928—1947;1948—1964;1965—1991;1992—2003;2004—2019.Year 1830 was allocated provisionally in connection with a reduction of plowed land at about this time,and also a change in the composition of crop rotations(Fig.2).The other time lines were clearly defined;1881 is associated with the abolition of serfdom and the resulting increase in population in the area;1927 is associated with the beginning of collectivization and the 5-year plans of the USSR;1947 is associated with the end of the Second World War and the beginning of the post-war restoration of agriculture;1964 is associated with the beginning of policies introduced under the leadership of L.I.Brezhnev,with massive purchases of grain abroad and a tendency to reduce the area of arable land in the region under study;1991 is associated with the collapse of the USSR and a sharp increase in fallow lands due to the crisis;2003 is associated with the improvement of the situation following the economic crisis that came after the collapse of the USSR,and with the development of previously abandoned lands.In general,the identified periods are in good agreement with the reconstruction of the main underlying drivers of land-use transitions in European Russia from 1770 to 2010(Matasov et al.,2020).

Fig.2.The share of agricultural crops in the crop rotation and the C-factor for different periods.

2.3.2.Cover-management factor(C-factor)

The C-factor is used to reflect the effect of biomass cover and soildisturbing activities on erosion rates.The C-factor value depends on many components:the composition and the condition of agricultural crops,soil cultivation technology,the direction of arable ridges,etc.In a long historical retrospective,it is possible to evaluate only the change in the composition of field crops,and the accuracy of these estimates will vary greatly for different time periods.

For the 18th—19th centuries,information on the composition of crop rotation was obtained from expert estimates of land-use systems in non-chernozem provinces in the south of the forest zone(Gauthier,1937,p.412;Milov,2001,p.572).In the 18th—19th centuries,a system with three field crops in rotation(winter grains—spring grains—fallow)was practiced in the study area.Until the mid-19th century,the land-use included periodically abandoned cropland together with crop rotation.On periodically abandoned croplands,the frequency of the fallow period was 8—15 years after several years of plowing.In some districts,fallow land accounted for up to 30—35%of the arable land.

The composition of crop rotation in the late 19th and early 20th centuries was estimated from land censuses(Encyclopedic Dictionary,1896;Agriculture of Russia,1923).These estimates are averaged over several regions.The composition of crop rotation between 1921 and 1928 was determined by statistical materials directly related to the Moscow region(Agriculture of the USSR,1929).From 1928 to the 1980s,the crop composition was determined by the“five-year plans”in the USSR,and this information was published in statistical collections for each region.The composition and the area of field crops in recent years is available on the Internet resource created by the Ministry of Agriculture of Russia(MAR,2020).All crops cultivated during the study period were grouped(Fig.2)with regional data of C-factor values(Larionov,1993).The C-factor for periodically abandoned cropland was calculated as follows:the first year was considered to be fallow(close to black fallow,but with sparse weed vegetation)(C=0.72);the second year was the average between fallow and perennial grasses(C=0.41);the third and subsequent years were perennial grasses(C=0.1)The value of the C-factor for periodically abandoned cropland averaged about 0.2.

2.3.3.Rain and snow erosivity factor(R-factor)

The R-factor is a multi-annual average index that measures the rainfall kinetic energy and the rainfall intensity to describe the effect of rainfall on sheet and rill erosion.Pluviographic measurements in Russia are made with analog devices,and historical data are not routinely digitized.At the same time,there has been a complete database since 1947 for daily precipitation measurements for a station located 25 km to the southwest of the study site(Russia Research Institute of Hydrometeorological Information,World Data Centre(RIHMI-WDC),Roshydromet).

where:Dj are the daily amounts of erosive precipitation(precipitation with an intensity of more than 10 mm per day),α and β are the coefficients according to Fan et al.(2013),and k is the number of days.

The R-factor was calculated only for the warm season(liquid precipitation),but was converted to conventional units of measurement(MJˑmmˑha-1ˑh-1ˑyr-1)by extrapolating the same value to other months.The calculated R-factor values were compared with the new global dataset(Panagos et al.,2017).The maximum snow water equivalents were obtained at the same station as the rainfall precipitation records.

2.3.4.Soil erodibility(K-factor)

The K-factor represents both the susceptibility of soil to erosion and the rate of runoff,as measured under the standard unit plot condition.The following procedures were set up for determining the K-factor.Soil samples were taken at 37 points(Fig.1C),where the locations of the sampling points were determined by the following criteria:the maximum coverage of the study area and the most diverse land-use history.Soil samples were taken from arable horizons from a depth of 0—25 cm,using a hand sampler(JMC,Newton,IA,USA;inner diameter:4.5 cm),and then they were dried to an air-dry state and were prepared for subsequent analyses.The organic carbon content in the soil samples was determined by the Tyurin photometric method(Orlov&Grisina,1981)using a Specord M40 spectrophotometer(VEB Carl Zeiss,Jena,Germany).The particle size distribution was then analyzed using a Bluewave(Microtrac MRB,Pennsylvania,USA)laser diffraction particle size analyzer.

The soil cover on arable land is very uniform,as is confirmed by the data(Sorokina&Kozlov,2009)and by the results of the analysis of the soil properties for the 37 points.Further extensive sampling was therefore not necessary for the general analysis.

The K-factor at each point was calculated according to the formula of Renard et al.(1997):

where:M is the textural factor with M=(msilt+mvfs)×(100-mc);mc[%]is the clay fraction content(＜0.002 mm);msilt[%]is the silt fraction content(0.002—0.05 mm);mvfs[%]is the very fine sand fraction content(0.05—0.1 mm);OM[%]is the organic matter content;s is the soil structure class(s=1:very fine granular,s=2:fine granular,s=3,medium or coarse granular,s=4:blocky,platy or massive);p is the permeability class(p=1:very rapid,…,p=6:very slow).

2.3.5.Topographical slope and length factor(LS-factor)

But then, looking at him again, she fancied he still breathed, and, hastily fetching some water from the nearest fountain, she sprinkled it over his face, and, to her great delight, he began to revive

The LS-factor represents the effect of slope length and topography on erosion.It is the ratio of soil loss from the actual field to the soil loss from a 72.6-foot(22.1-m)length and 9%slope on the same soil type.

The LS-factor was calculated by WaTEM/SEDEM,using the digital elevation model(DEM),which was constructed from the elevations of a detailed topographic map with a cell size of 20×20 m.Contributing areas were computed by multiple flow algorithm.

2.4.Verification of the calculated soil erosion rates

Verification was carried out by comparing the soil erosion rates calculated by the model with data obtained using the magnetic tracer(fly ash)method.The magnetic tracer method is based on a study of spherical magnetic particles(fly ash)in the soil(Gennadiev et al.,2013;Olson et al.,2013;Zhidkin et al.,2015,2016).The properties of spherical particles,e.g.a high degree of resistance to destruction,scatter and weight close to soil fine fraction,and moving only in solid form,are well suited for quantitative assessments of soil erosion.These particles are formed during the combustion of coal,e.g.in the furnaces of steam locomotives.

At a distance of 20—30 km from the railway,the content of spherical magnetic particles is sufficient to use the magnetic tracer method.The method can therefore be used over large areas,including anywhere in the study site.A railway was built in the study area in 1860,so the period of estimated soil erosion using this method is about 150 years.To compare the results,model calculations were also carried out for a 150-year period,considering the changes in crop composition,arable land boundaries and rainfall erosivity in this period.

A detailed analysis of erosion-accumulative processes was carried out on three slopes with different morphologies and different land-use histories.Slope EK1(Fig.1C)is located near the village of Novovoronino.This territory was plowed from year 1830—1881,then was abandoned from 1881 to 1927,and was then plowed again after 1927,making a total plowing duration of～125 years.The average slope is 2.8°,the maximum slope is 4.0°,with south-and south-west exposure.Slope EK2(Fig.1C)is located near the village of Alyoshino.It was plowed from year 1830—1947,was abandoned from 1947 to 1964,and was then plowed again after 1964.The average slope is 1.9°,the maximum slope is 3.7°,and the exposure of the slope is to the south.Slope EK3(Fig.1C)is located in the central part of the study site,near the village of Eldigino.This territory was plowed throughout the studied period of 250 years,and has an average slope of 2.0°.The maximum slope is 4.5°,with a north-and north-east exposure.These three catenas are therefore representative of the entire area,which does not contain many slopes exceeding 3°.Soil samples were taken on the slopes every 70—80 m from the watershed to the foot of the slope along two parallel catenas 3—5 m apart on slopes EK1 and EK3,along one catena on slope EK2.A total of 48 soil samples were analyzed for the content of spherical magnetic particles(10 samples on the reference site,8 samples on slope EK1,20 samples on slope EK2,and 10 samples on slope EK3).

3.1.Changes in economic and natural factors in the study area

3.1.1.Changes in land-use over the last 250 years

The total area of land ever plowed between 1780 and 2019 is 7300 ha(65% of the study site).The area of arable land varied significantly within different periods(Fig.3).Currently,the area of arable land is at its lowest point(900 ha),while in the 18th century the area of arable land was at its highest point(3500 ha).The average cultivation time of the same lands is not high,i.e.,less than 80 years(Fig.4).During the entire study period of 250 years,only～4%of the arable land was constantly under the plow.In the period 1831—1881,～13% of the arable land was under the plow.

Fig.3.Changes in the amount of soil erosion,sediment deposition and the area of arable land over the last 250 years on the study site.

The land was under the plow for the longest time near towns and villages(Fig.4).However,with increasing the distance from the center of the village,the duration of land plowing decreased.Reduction of arable land took place mainly in areas prone to waterlogging,including at the foot of the slopes and on flat nearwatershed areas.The northern part of the study area has a more dissected topography and,as a result,it is less susceptible to waterlogging.The duration of plowing was therefore greater in the northern part of the study area than in the southern part,and the arable lands are confined to river valleys and gullies.

Arable land was mainly located on gentle slopes up to 2°in steepness,and in different periods the area of these lands varied between 72% and 76% of the total area of arable land.The area of arable land on slopes 2—4°in steepness varied between 21.1% and 23.8%,whereas the area of arable land on steep slopes(＞4°)ranged only between 3.1%and 4.8%.However,these areas on a steep slope made a significant contribution to the amount of soil loss.The gross erosion on steep slopes amounted to between 17% and 24% of the total amounts of washout in different periods.Steep slopes adjoin river valleys and have high connectivity,as a result of which a large proportion of the sediment washed away from the steep slopes entered directly into the river valleys.

Although there were only slight variations in the average values of the steepness of the plowed slopes,the boundaries of the arable land at the foot of the slopes varied significantly.The change in the boundaries of the arable land in the lower parts of the slopes had a great influence on the total amounts of soil loss and on the removal of sediments outside the arable land.

The C-factor remained almost unchanged for over a century from the middle of the 18th century to the end of the 19th century,when the three-crop farming system dominated in the studied region and there were no big changes in land-use.By the end of the 19th century,the system of using periodically abandoned croplands had gradually disappeared almost completely,due to the depletion of the reserve of unplowed territories suitable for agriculture.At the same time,there was an increase in the area of fallow land.This trend led to an increase in the C-factor in terms of the entire area of arable land.

Fig.4.Duration of plowing in the study area.

The introduction of row crops into the crop rotation,which occurred in the last quarter of the 19th century,could have increased the C-factor,but this was offset by a decrease in the area of fallow land.Until the last decades of the 20th century,the following trend continued:the growth of areas under row crops was offset by a decrease in fallow land and by an increase in the proportion of perennial grasses.

When the Soviet Union collapsed in the last decade of the 20th century,there was a long economic crisis in the country,which led to a shortage of fertilizers and a reduction in the range of crops.There was massive abandonment of arable land and a return to the practice of using periodically abandoned croplands.As a consequence,a large amount of fallow land,a decrease in the share of row crops,and significant areas under perennial grasses significantly reduced the C-factor to the minimum for the entire period.

Over the past two centuries there has been a reduction in arable land within the study site,and this reduction was observed for the entire southern part of the forest zone of the European territory of Russia.This reduction process has especially accelerated in the last 20 years,due to a massive transfer of agricultural land in the vicinity of Moscow into use for summer cottages.

3.1.2.Rain and snow erosivity factor changes

The results of the long-term average values of the R-factor estimated by Eq.(3)were verified,and were compared with the estimate made on the basis of pluviographs by the international community(Panagos et al.,2017).The relative error in this matching was 7%(underestimation)on average for locations selected within the 6×6°vicinity around the study site(which corresponds approximately with the territory of the administrative region).Using the chosen technique,the temporal variability ofRyrcould also be estimated.This could not have been done on the basis of time-averaged global databases(Panagos et al.,2017).

The mean R-factor for the period from 1947 until 2019 was 279 MJˑmmˑha-1ˑh-1ˑyr-1.Instrumental precipitation data for earlier historical periods were lacking,or contained large gaps.We therefore decided to use the mean R-factor value for all years before 1947.To justify this decision,we used the June—July—August(JJA)average self-calibrating Palmer Drought Severity(PDS)Index,based on the network of 697 annual treering chronologies distributed over the European Russia Drought Atlas domain for 1795—2015(Cook et al.,2020).Specifically,we used the time series for the third rotated empirical orthogonal function(Varmix 3)—the spatiotemporal pattern of wetness in the area including the study site.As shown in Table 2,the frequency of the wettest year with PDS above one standard deviation is a relatively stable characteristic,so the R-factor is considered to be relatively stable.The second part of the 19th century was wetter than the first part,but not to an extent that could greatly increase the uncertainly of the erosion model more than other factors.A nonlinear trend toward an increase from minimal values in the period between 1947 and 1964 to 30% higher values between 1964 and 1991,and a subsequent decrease,can be seen in the time-series of the R-factor for the instrumental period.The reconstruction from tree rings and the instrumental data agree onthis tendency.The temporal variability outweighs the spatial variability within the study area.Usage of the R-factor for the wet period between 1970 and 1980 that can be found in the literature can therefore lead to an overestimate of the rain erosivity.

The maximum snow water equivalent,estimated annually at the onset of snowmelt and averaged for 1947—2019,was 139 mm,which was slightly larger(144 mm)for 1947—1991,and slightly lower(136 mm)for 1991—2019.There was synoptic interannual variability within each of the two periods,but no notable climatic variability—the snow water equivalent is a very stable characteristic for the study area;it has not until now been influenced by climate change on a decadal timescale.

3.1.3.Soil erodibility on the study site

The results of soil samples analyses showed that for the study site the average soil organic carbon concentration was 1.3%,and the clay fraction was 12.7%.Additional support for the use of Eq(4)is given in the recommendations of Panagos et al.(2014),while the soil properties of the study site are appropriate for calculating the K-factor.

Within the key area,the K-factor varied slightly from 0.075 to 0.090 tˑhaˑhˑha-1ˑMJ-1ˑmm-1with an average value of 0.082 tˑhaˑhˑha-1ˑMJ-1ˑmm-1(the coefficient of variation was only 4%;Table 3).The low variation coefficient of soil erodibility allows to use the average value of the K-factor for calculating erosionaccumulative processes,without any additional computations associated with changes in time and space.

3.2.Soil erosion and sediment deposition

Average rates of soil erosion caused by rainfall varied from 5.2 to 8.2 tˑha-1ˑyr-1for the study site from 1780 to 2019.The largest rates of soil erosion(7.6—8.2 tˑha-1ˑyr-1)were from the late 18th century to the early 20th century.From the middle of the 20th century to the present time,the rate of soil erosion has decreased steadily:from 6.2 to 7.3 tˑha-1ˑyr-1in 1947—1991 to 5.2—5.3 tˑha-1ˑyr-1from 1991 until the present time(Fig.5).The rates of intra-slope sediment accumulation within arable land decreased gradually over the entire study period from 6.5 tˑha-1ˑyr-1to 2.0 tˑha-1ˑyr-1.

Gross soil erosion,determined by the rate of erosion and the area of plowing,varies significantly between 5×103tˑyr-1and 28×103tˑyr-1for the study area(Fig.3).The largest amounts of soil erosion are confined to the period of the most intensive agriculture,from the end of the 18th century to the early 20th century,while the smallest amounts of gross soil erosion are noted in the presentday period.

Net soil erosion,which is defined as the amount of sediments carried outside the arable land,was calculated on the basis of the difference between the amounts of soil erosion and the amounts of intra-slope accumulation.The amounts of net erosion from the site varied between 3.1 tˑha-1yr-1and 8.1 tˑha-1yr-1.The amounts of net erosion therefore varied much less than the volumes of gross erosion(Fig.3).

The average rate of soil erosion caused by snowmelt runoff did not exceed 1 tˑha-1ˑyr-1(Fig.6).The lowest rates of soil erosion have been recorded in recent years,which confirms the tendency towards a decrease in the values for the intensity of soil erosion caused by snowmelt runoff,as described by Litvin et al.(2017).

3.3.Verification of the calculated soil erosion rates

The rates of soil erosion obtained by the magnetic tracer(fly ash)method vary significantly in catenas where repeated measurements were made(Table 4).For example,on slope EK1,the rates obtained by the magnetic tracer method were almost twice as high for different catenas.This variation is due to the presence of a microrelief(in the form of small hollows)and local redistribution of the magnetic tracer into micro-depressions as a result of tillage erosion.Nevertheless,the average long-term rates of soil erosion calculated by the WaTEM/SEDEM model are in good agreement with the average data on soil erosion rates obtained by the magnetic tracer method(Table 4).On all slopes,the soil erosion rates calculated by WaTEM/SEDEM are within the range of variation of the results obtained by the magnetic tracer method.We can therefore assume that the model calculations quite satisfactorily reflect the long-term average annual rates of soil erosion.

4.1.Accuracy and reliability of the results

The research was carried out with a high level of detail.However,the soil erosion modelling results are obviously approximate,due to the lack of accurate historical information.Various factors with varying levels of detail were taken into account.The C-factor was considered approximately—on average for the site and on average for the time intervals(Fig.2).In addition,important information about soil-disturbing activities is missing.The details of land-use boundaries and climate change mapping are time-dependent.Accurate data on the climate in this area have been published since 1947,and maps with high accuracy of the position of the borders of arable land(2—5 m)have been available since 1931(Table 1).Kfactors and LS-factors have been taken into account best of all,sincethey have been almost unchanged over time.In addition,the soil cover of the studied area is very uniform(Table 3).

The estimates of the changes in soil erosion are therefore approximate,and should not be taken too literally.Despite all the shortcomings,however,the results are important,because they demonstrate the fundamentally different trends in soil erosion in the forest and steppe zones of Russia(see Section 4.3).Similar theoretical estimates were published in(Spatio-temporal patterns,2019).

The lack of data prevents a complete comparison of the results with the literature data.However,the data that we have obtained are in good agreement with estimates over the past four decades,presented in(Litvin et al.,2017).The rate of soil erosion in the 1980s in the Moscow region averaged 5.1 tˑha-1ˑyr-1;in the 2010s,the erosion rate decreased by 24%,i.e.,to 3.9 tˑha-1ˑyr-1(Litvin et al.,2017).In the study site,the rate of soil erosion in the 1980s was 7.3 tˑha-1ˑyr-1,and in the 2010s the rate decreased by 27% to 5.2 tˑha-1ˑyr-1.Thus,the reduction in soil erosion for the Moscow region and for the study site turned out to be very close to each other(24%for the Moscow region,and 27%for the study site).The absolute values for the study site are slightly higher than the values for the Moscow region as a whole,since the study site is located in the Klinsko-Dmitrov Ridge,which is an erosion-threatened part of the Moscow region(Fig.1B).

The rates of soil erosion calculated over a secular period of time are confirmed by the magnetic tracer method estimates(Table 4).According to the review(Alewell et al.,2019),the errors of the(R)USLE-based model calculation tended to be greater for the lower soil-loss values,with overprediction on plots with low erosion rates and underprediction on plots with high erosion rates.On the study site,the rate of soil erosion is moderate.That is probably why the calculation results are satisfactory.

4.2.The main factors of changes in the rate of soil erosion

The most variable factor was the change in the area of arable land(Table 3).The maximum variation in the area of the arable land was 3.7-fold between 1780 and 2019.The area of arable land in the study site has been decreasing steadily over the past 250 years(Fig.3).Other soil erosion factors did not have a directional trend.As a result,the strengthening of some erosion factors did not lead to an increase/decrease in soil erosion,due to the weakening of other factors.

Changes in climate and in crop composition affected the changes in the intensity of soil erosion to a lesser extent than the share of area of plowing.Nevertheless,the contribution of changes in climate and in crop composition must be taken into account,since about 10—20%of the average rate of erosion is determined by the variation of each of these factors over time.In addition,variations in climate,crop composition,and the position of the arable land relative to the topography determined the absence of a direct relationship between the area of plowing and the amount of soil erosion and sediment deposition.In some periods of time(in particular,the mid 19th century and the second half of the 20th century),the synergistic effect of these factors led to an increase in net erosion while reducing the area of arable land(Fig.3).

4.3.Erosion trends

The trend identified in the study site differs from findings reported in the literature data for the changes in soil erosion in the forest-steppe and steppe parts of European Russia.In the steppe parts,the area of arable land gradually increased and plowing was almost never abandoned.Plowing was primarily carried out on flat lands.As a result,the duration of the plowing period did not significantly affect the degree of soil erosion in the steppe zone of Russia(Golosov et al.,2021;Smirnova et al.,2020).In the study site(in the temperate forest zone),by contrast,the area of arable land varies greatly at different times.Due to the different situation in the study area and the variations of other factors,the volume of gross erosion grew 5.6-fold over the past 250 years on the studied site.However,the rate of erosion has generally decreased over the past 250 years in the temperate forest zone,despite small fluctuations(Fig.3).In particular,at the end of the 19th century and in the middle of the 20th century,an abrupt increase in the area of arable land and soil erosion is observed.This was due to the massive plowing of land in the 19th century,after the“abolition of serfdom”,and in the mid 20th century,due to the end of the Second World War.Although there were some fluctuations,the general trends in soil erosion in the forest zones and in the steppe zones are opposite.

It is important to note that the duration of plowing in the study area depended primarily on the location or,more precisely,on the proximity of settlements(villages)and roads(Fig.4).The duration of plowing largely determines the volume of erosion losses and,as a consequence,the degree of erosion of the soil cover.Thus,the soil cover is much more eroded near old villages than far away from them.That is,the social infrastructure in the forest zone of Russia had a huge impact on the spatial structure of soil erosion,on a par with,or even higher than,natural factors.

In addition,the important role of the past borders of arable land in the formation of the erosion-accumulative structure has been revealed in the study area,and also,for example,in the South-West Foreland of the West Carpathians(Smetanova et al.,2017).The changes in the configuration of the borders of the arable land was the determining factor in intra-slope sediment deposition,which is rather high(40—80%)on the studied site(Fig.3).Intra-slope sediment deposition smoothed out the variation in net erosion,which amounted to 2.6-fold over the past 250 years.

The southern part of the forest zone of Russia has a complex multi-stage land-use history,with a clear trend of decreasing rates and decreasing amounts of soil erosion over the past 250 years.Currently,the amount of gross erosion is more than 5.5 times lower than in the 19th century.The decrease in the rate and in the amount of soil erosion is primarily due to changes in the area of arable land,which has decreased 3.5—4-fold in the past 50 years compared with the end of the 18th century.Lands near the villages were under the plow for the longest time,and they were almost never abandoned even during periods of massive reduction in plow land.As a result,the infrastructure largely determined the degree of soil erosion in this area.The variation in the natural factors of soil erosion was less than 20%,but in some periods of time the synergistic effect of unidirectional changes of natural factors led to an increase in erosion while there was a reduction in the area of arable land.Changes in cropland boundaries had a large impact not only on soil erosion,but also on intra-slope sediment accumulation,which contributed to a decrease in variability in net erosion compared to the variability of gross erosion.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Acknowledgements

This research was supported by the Russian Foundation for Basic Research(RFBR)within scientific project N°18—35—20011.We also thank the anonymous reviewers for useful comments and recommendations,which have been very helpful in improving the manuscript.