Contemporary Long-Term Trends in Water Discharge, Suspended Sediment Load, and Erosion Intensity in River Basins of the North Caucasus Region, SW Russia

: For the ﬁrst time, contemporary trends in water discharge, suspended sediment load, and the intensity of overall erosion in the river basins of the North Caucasus region, as one of Russia’s most agriculturally developed geographic areas, were identiﬁed. The study was carried out using monitoring data of the Federal Service for Hydrometeorology and Environmental Monitoring of the country for 21 rivers by comparing two periods: 1963–1980 and 2008–2017. According to the study’s results, trends of an increase in the mean annual water discharge (by 2–97%) and the essential reduction in its intra-annual variability have been found in most of the studied rivers. On the contrary, the trends of reduction in annual suspended sediment load and the intensity of erosion in the river basins were identiﬁed in most of the study region. Their most essential and statistically signiﬁcant decreases (by 47–94%) were recorded within the Stavropol Upland, which several decades ago was considered one of the most erosion-dangerous territories of the entire country, as well as in some river basins of the central part of the Greater Caucasus’s northern slope (by 17–94%). The changes in climate (reducing the depth of soil freezing and meltwater runoff on the soil) and land use/cover (reduction of acreage and load (pressure) of agricultural machinery on the soil, reducing livestock on pastures, and the transfer of water from the neighboring, more full-ﬂowing rivers) are considered the leading causes of the aforementioned trends. The ﬁndings will contribute to solving some economic and environmental problems of both the region and adjacent territories and water areas. combined diagrams of temporal dynamics in March and the depth of snow cover in February, which are derivatives of the changes mentioned above. We have purposefully identiﬁed the two ﬁrst 10-day periods in March in terms of soil temperature. In Russia’s southern regions, snowmelt processes occur comparatively quickly, not dragging on as long as, for example, in the East European Plain’s forest zone for a month or more. Consequently, it is essential to show when the processes of formation and temporary changes in the snowmelt runoff and erosion processes caused by it are the most active. M.A.K.; resources, A.V.G. A.G.S.; data curation, A.V.G.; preparation, A.V.G., A.G.S., M.A.K.; writing—review A.V.G.; visualization, and supervi-sion, A.V.G.; A.V.G.; funding acquisition,


Problem Formulation
The last half-century in Eastern Europe has been characterized by noticeable climate change [1][2][3][4][5][6], which was a regional consequence of global warming, and land use/cover changes that began shortly before and especially after the collapse of the Union of Soviet Socialist Republics (the USSR) at the end of 1991 [7][8][9][10][11][12]. One of the environmental consequences of these changes were hydrological changes, which manifested, in particular, through a seasonal and long-term transformation in the water regime of rivers [13][14][15][16]. As these and other researchers note, the intra-annual variability of the rivers' water discharge has noticeably decreased over the recent decades, primarily due to a reduction in maximal water discharge during the snowmelt-induced flood season, and an increase in baseflow during low-water periods. Despite a relatively good knowledge of the various The Stavropol Upland is composed mainly of clays, sandstones, and limestones; loams (including loess-like ones) are predominant in the eastern part of the upland's surface. The prevailing modern soils in the upland are medium-humus chernozems and low-humus ones in its western part. In the Kuma-Manych Depression and the western part of the Caspian Depression, loamy castanozems and solonetzes are most widespread [36]. The mean annual precipitation amount ranges from 400 mm in the east to 600 mm in the west of the upland and less than 300-400 mm in the lower part of the river basins. This is in marked contrast to the Greater Caucasus, where the upper "floor" (starting from an altitude of about 2000 m) is located in the area of influence of the western air transport (the impact of the Atlantic Ocean and the Mediterranean Sea). At the height of about 2000 m, the ridge's northern slope has approximately 1500-2500 mm of annual precipitation in the western sector, while in the eastern sector-less than 1500 mm (https://bigenc.ru (accessed on 21 December 2020)).
May and June are the months with the most precipitation. On the contrary, January and February (and November in the east of the upland) are the months with the least precipitation. Winters are rather mild (−3 to −5 • C, on average); summers are comparatively warm (20 to 26 • C, on average). In the past, steppe vegetation prevailed in the river basins, and motley grasses and feather grasses (Stipa and others) were widespread there; broadleaf forests were often found in numerous small dry valleys. Natural landscapes have been plowed over a large area-up to 60-70% in the basins' main (central) parts [36]. Due to a combination of significant dissection of the local topography, a relatively high steepness of local slopes, and relatively less fertile soils, in the sub-basin of the upper reaches of the Kalaus River (upriver of the village of Sergiyevka (Sérgiyevka), see Figure 1), the share of cultivated land is lower there: in 1985 (i.e., during the late USSR) and 2015, it was estimated at 46% and 40%, respectively [37].
The Stavropol Upland is located in the Stavropol Krai (66,160 km 2 ; more than 2.8 million people), one of the most soil-fertile and agriculturally developed administrative regions of all Russia (see Figure 1). In 2006-2017, cultivated land occupied, on average, about 45% of the krai's total area (see below).
The Kalaus River as a reference river. The total length of the Kalaus River (see Figure 1) is 436 km; the river's weighted average gradient is from 1.6 to 0.7‰ and less in the lowest reaches. The altitude difference between the headwater and the mouth of the river is about 660 m. The banks of the river are relatively high and steep ( Figure 2).  The modern total sediment load of the Kalaus River is almost entirely (approximately 98%) represented by its suspended sediment component [36]. Until the 1980s and possibly even the 1990s, spring (snowmelt-induced) suspended sediment load in the rivers of the basin amounted to 45-70% of the total annual load, while summer and autumn-winter sediment load did not exceed 40% and 10-15%, respectively [36]. In small rivers of the Kalaus River basin, sediment load and water discharge were (and still are) regulated by a comparatively large number of ponds formed by earthen dams in small dry valleys. This is especially true for the sub-basin of the middle reaches of the river. The ponds were created mainly during the late USSR (the 1970s and the 1980s). The Kalaus River has 81 tributaries with a total length of 936 km. The main sources of water in the river are snowmelt-induced and rainfall-induced waters. The underground component is relatively small. The hydrological regime of the Kalaus River is a regime with spring (snowmelt-induced) floods (February to March) and rainfall-induced floods in the warm season (partly March to partly November). In the summer-autumn season, it is difficult to irrigate agricultural land; in the conditions of the region's relatively arid climate, this circumstance results in the need for artificial irrigation at the expense of adjacent territories (mainly the Kubán River basin, see Figure 1). This also applies to the Kuma River, which additionally receives water from the neighboring full-flowing Térek River. The location of main irrigation canals in the study area can be found in Supplementary Material 1.
Several decades ago, the Kalaus River belonged to the small group of rivers with the greatest turbidity (mostly sediment concentration) among all Russian rivers. This was owing to dissected local topography (this contributes to a high degree of sediment delivery to the river network), relatively erosion-poorly stable rocks (primarily loess-like loams overlapping bedrocks) composing the surface of the river basin, a high degree of plowing of soils in the basin, and also due to active deformations (mainly lateral) of the riverbeds of the Kalaus River's hydrographic network in loamy and clay riverbanks (see Figure 2).
The modern total sediment load of the Kalaus River is almost entirely (approximately 98%) represented by its suspended sediment component [36]. Until the 1980s and possibly even the 1990s, spring (snowmelt-induced) suspended sediment load in the rivers of the basin amounted to 45-70% of the total annual load, while summer and autumn-winter sediment load did not exceed 40% and 10-15%, respectively [36]. In small rivers of the Kalaus River basin, sediment load and water discharge were (and still are) regulated by a comparatively large number of ponds formed by earthen dams in small dry valleys. This is especially true for the sub-basin of the middle reaches of the river. The ponds were created mainly during the late USSR (the 1970s and the 1980s).

Hydrological Data
To assess contemporary trends in river water discharge (further-discharge) and suspended sediment load (further-load), as well as the overall intensity of erosion processes within the Stavropol Upland, the study used the results of long-term observations at 13 state hydrological (gauging) stations in the basins of two rivers-the Kalaus River and the Kuma River, which drain the prevailing area of the upland (see Figure 1; Table 1). The study's basic (reference) materials were the results of long-term discharge/load monitoring at four hydrological stations in the Kalaus River basin (see Figure 1; Table 1). The basin's area covered by the monitoring is 9100 km 2 , i.e., about 94% of the basin's total area.
To evaluate the regionality of discharge/load-trends identified for the Stavropol Upland, available long-term series of the studied hydrological characteristics at eight additional gauging stations (sub-basins) of neighboring rivers (the Terek River and the Belaya River) of the central part of the northern slope of the Greater Caucasus were also analyzed (see Table 1; Figure 1). Thus, the total area covered by hydrological monitoring in this study is 55,554 km 2 .
The state monitoring results were available from hydrological collections of the former USSR [38][39][40][41][42], the Russian State Water Register (http://voda.mnr.gov.ru (accessed on 14 October 2020)), and the automated information system of state monitoring for water bodies (http://gmvo.skniivh.ru (accessed on 10 October 2020)) of the Federal Agency for Water Resources of the Ministry of Natural Resources and Environment of the Russian Federation.
The data on discharge and load in the Kalaus River basin were available for the years presented in Figure 3. Unfortunately, due to lack of available data, there were interruptions in the series of observations, the most significant of which was confined to 1981-2007/2009. The discharge data collected were summarized in the long-term series on mean annual, mean monthly, maximal, and minimal (per annum) discharge (m 3 s −1 ) at the aforementioned hydrological stations. For the remaining 17 rivers studied, these data were limited only to mean annual values. According to [43], observations for the discharge of rivers of the former USSR and modern Russia were carried out according to a single standard program, several times daily (mainly by converting water level into discharge). S-river basin areas; A-mean absolute altitudes of the river basins; I-weighted average gradients of the rivers; L-lakes, reservoirs, and ponds; F-forestland area; P-plowland area; WD mean (m 3 s −1 ) and M(WD mean ) (L s −1 km −2 )-mean annual and specific mean annual water discharge, respectively (for 1963-1980 and 2008-2017); ND-no data. Note: The data on L, F, and P are for the 1970s [38].
The load data collected for the Kalaus River basin were summarized in the longterm monitoring series on mean annual and maximal (per annum) suspended sediment concentration (kg m −3 ), as well as mean annual and mean monthly load (kg s −1 ) for the aforementioned hydrological stations. For all of the remaining 17 rivers studied, these data were limited only to mean annual values. According to [44], the average error in estimating the mean annual values of the load in gauging stations in the former USSR was 8-10%. It could probably have somewhat increased in the last decades. S-river basin areas; A-mean absolute altitudes of the river basins; I-weighted average gradients of the rivers; L-lakes, reservoirs, and ponds; F-forestland area; P-plowland area; WDmean (m 3 s -1 ) and M(WDmean) (L s -1 km -2 )-mean annual and specific mean annual water discharge, respectively (for 1963-1980 and 2008-2017); ND-no data. Note: The data on L, F, and P are for the 1970s [38]. Figure 3. The years of observations (available data) for water discharge (in blue color) and suspended sediment load (in brown color) of the Kalaus River and the Ula River at the analyzed hydrological stations, the Stavropol Krai, SW Russia. The Kalaus River: 1-at Sergiyevka, 2-at Svetlograd, 3-at Vozdvizhenskoye; 4-the Ula River at Staromarievka (see Figure 1, Table 1).
The load data collected for the Kalaus River basin were summarized in the long-term monitoring series on mean annual and maximal (per annum) suspended sediment concentration (kg m -3 ), as well as mean annual and mean monthly load (kg s -1 ) for the aforementioned hydrological stations. For all of the remaining 17 rivers studied, these data were limited only to mean annual values. According to [44], the average error in estimating the mean annual values of the load in gauging stations in the former USSR was 8-10%. It could probably have somewhat increased in the last decades.

Climate Data
Long-term data on monthly/annual air temperature and precipitation for two periods studied were collected and then analyzed for eight regional meteorological stations (see Figure 1). For the stations in the city of Stavropol and the city of Budyonnovsk, reliable information on long-term changes in soil temperature (March) and snow cover depth (February) were also collected and processed. All of these data were available from the All-Russia Research Institute of Hydro-Meteorological Information-World Data Center (http://meteo.ru accessed on 30 December 2020).

Land Use/Cover Data
Information on the distribution of cultivated land area within the Stavropol Krai (where the reference Kalaus River basin and the predominant part of the Kuma River basin are located) for different time intervals (during the late USSR period (1970,1975,1980,1985,1986,1987) and the post-USSR period (annually for 1992-2017)) were collected and analyzed to semi-quantitatively estimate the probability of the anthropogenic (mainly agricultural) impact on the long-term temporal dynamics in discharge and load. The information was detailed for principal crops such as cereals, perennial, annual crops, and sunflower crops. The sources of the information are statistical collections (bulletins) of the former USSR [45,46] and also electronic statistical resources of the Russian Federation (https://fedstat.ru accessed on 30 December 2020).
To qualitatively assess the potential impact of the number of livestock (cattle, sheep, and goats; units per annum) on temporal changes of the above-mentioned hydrological variables, relevant information on the Stavropol Krai cumulatively for all types of farms for 1970, 1975, 1980, 1985, 1986, 1987, 1990, 1995, 2000, 2005, 2010, 2014, 2018, and 2019 Figure 3. The years of observations (available data) for water discharge (in blue color) and suspended sediment load (in brown color) of the Kalaus River and the Ula River at the analyzed hydrological stations, the Stavropol Krai, SW Russia. The Kalaus River: 1-at Sergiyevka, 2-at Svetlograd, 3-at Vozdvizhenskoye; 4-the Ula River at Staromarievka (see Figure 1, Table 1).

Climate Data
Long-term data on monthly/annual air temperature and precipitation for two periods studied were collected and then analyzed for eight regional meteorological stations (see Figure 1). For the stations in the city of Stavropol and the city of Budyonnovsk, reliable information on long-term changes in soil temperature (March) and snow cover depth (February) were also collected and processed. All of these data were available from the All-Russia Research Institute of Hydro-Meteorological Information-World Data Center (http://meteo.ru (accessed on 19 October 2020)).

Land Use/Cover Data
Information on the distribution of cultivated land area within the Stavropol Krai (where the reference Kalaus River basin and the predominant part of the Kuma River basin are located) for different time intervals (during the late USSR period (1970,1975,1980,1985,1986,1987) and the post-USSR period (annually for 1992-2017)) were collected and analyzed to semi-quantitatively estimate the probability of the anthropogenic (mainly agricultural) impact on the long-term temporal dynamics in discharge and load. The information was detailed for principal crops such as cereals, perennial, annual crops, and sunflower crops. The sources of the information are statistical collections (bulletins) of the former USSR [45,46] and also electronic statistical resources of the Russian Federation (https://fedstat.ru (accessed on 2 November 2020)).
Additionally, information on the annual number of the main types of farm machinery (tractors and grain combine harvesters) in the Stavropol Krai cumulatively for all types of farms for 1970-1975 and 2007-2019 (https://gks.ru (accessed on 5 November 2020)) were collected and processed to identify the dynamic of the "load" (units per 10 3 ha) of this machinery on cultivated land within the administrative region during these periods.

Hydrological Data Processing
Two periods were distinguished in the long-term monitoring series of both discharge and load/concentration because of differences in their characteristics and dynamics: 1963-1980 (used as a baseline period) and 2008-2017. The first period preceded the onset of active climate change and river discharge in the region (around 1980), while the second period, on the contrary, became the time of relatively more significant climatic and hydrological transformations in Eastern Europe [1][2][3]5,[13][14][15][16]. On the other hand, the period from the beginning of the 1960s to 1987/1988 was marked by the highest level of agricultural development in the country [26] and, accordingly, in the study region, while 2008-2017 was a time of relative stabilization in agriculture degradation that followed the USSR's collapse in 1991 (the 1990s and early 2000s) [9,12,21,47]. Consequently, distinguishing these two periods makes it possible to reveal contemporary discharge/load trends of the regional rivers studied against the background of simultaneously occurring changes in climate and land use/cover over the last six decades.
Along with the calculation of averaged long-term values, specific values of mean annual, maximal (per annum), and minimal (per annum) water discharge were also calculated for each distinguished monitoring period: M(WD)-integral specific discharge (L s −1 km −2 ) upriver of the corresponding hydrological station for the corresponding monitoring period; and ∆M(WD)-differential specific discharge (L s −1 km −2 ), which is calculated as follows in Equation (1): WD i (L s −1 )-mean or extremal discharge of a river at a hydrological station i (with a total basin area S i (km 2 ) upriver of this station); WD i+1 (L s −1 )-mean or extremal discharge of the river at the nearest hydrological station i + 1 located upriver (with a total basin area S i+1 (km 2 ) upriver of this station). When ∆M(WD) i > ∆M(WD) i+1 , a larger runoff (discharge) is formed in the sub-basin (S i − S i+1 ) compared to the sub-basin S i+1 located upriver. However, when ∆M(WD) i < ∆M(WD) i+1 , the opposite situation is noted. This allows identifying the areas (sub-basins) of the most intense water runoff formation and its yearto-year variability in a particular river sub-basin.
Along with the calculation of averaged long-term values, specific values of mean annual suspended sediment load were also calculated for each distinguished monitoring period: SSY-integral suspended sediment yield (Mg km −2 y −1 ; 1 Mg = 1 ton) upriver of the corresponding hydrological station for the corresponding monitoring period; and ∆SSY-differential suspended sediment yield (Mg km −2 y −1 ), which is calculated as follows in Equation (2): SSL i (Mg y −1 )-annual load of a river at a hydrological station i (with a total basin area S i (km 2 ) upriver of this station); SSL i+1 (Mg y −1 )-annual load of the river at the nearest hydrological station i + 1 located upriver (with a total basin area S i+1 (km 2 ) upriver of this station). When ∆SSY i > ∆SSY i+1 , a larger sediment yield is formed in the sub-basin (S i − S i+1 ) compared to the sub-basin S i+1 located upriver; however, ∆SSY i < ∆SSY i+1 , the opposite situation is noted. This allows identifying the areas (sub-basins) of the most intense sediment formation (mechanical denudation, including primarily sheet/rill/gully erosion) and its year-to-year variability in a particular river sub-basin.
Despite some methodological flaws, river sediment load/yield was/is widely used in world practice to estimate the temporal variability of erosion (more broadly-mechanical denudation) intensity in fluvial systems of various scales [21,[48][49][50][51][52][53][54][55][56]. It should always be borne in mind that sediment load/yield cannot serve as an absolute measure of all erosion products due to a complex mechanism for delivering sediment from erosional areas to riverbeds. It depends on a large set of factors such as geological structure, morphometry/morphology of slopes, the density of permanent and temporary streams, basin area, climatic characteristics, features of natural landscapes, and the nature of their anthropogenic transformation, etc. [26,[57][58][59][60][61][62]. This is especially important to consider when assessing the spatial heterogeneity of erosion (mechanical denudation) intensity. However, the importance and accuracy of sediment yield/load as an indicator of the temporal dynamics of erosion (mechanical denudation) in the basin increases noticeably when assessing the overall variability of the intensity of erosion within any one river basin for different time intervals (the temporal dynamics of sediment load/yield relative to one gauging station) [21].

Land Use/Cover Data Processing
To identify the role of agricultural activity (land cover changes) in long-term sedimentload/erosion changes, the integral (generalized) erosion resistance index-C-factor (ΣC)was calculated for the warm (rainfall) season (April to November) during two periods (1970-1991 (the late USSR) and 2006-2017) for the whole Stavropol Krai. The index is a weighted average that considers both the area occupied by one or another crop (or agrocenosis) in the corresponding period and its individual C-index. The ΣC index was calculated only for the crops listed above as follows (Equation (3)): C 1 , C 2 . . . C n -individual C-indices for different analyzed crops (natural meadow vegetation-0.016, perennials-0.02, sunflowers-0.43, cereals-0.48, and annuals-0.48), averaged over the warm period [61]. F 1 , F 2 . . . F n -the average areas occupied by these crops in the corresponding period. ΣC varies from 0 to 1. The lower ΣC, the better one or other crop (or agrocenosis) protects the soil against sheet/rill erosion processes.
The changes in the "load" (τ, units per 10 3 ha) of farm machinery on cultivated land within the Stavropol Krai between 1970-1975 and 2007-2019 were calculated as follows in Equation (4): n-the average annual number (units) of tractors or grain combine harvesters for the corresponding period; F-the average cultivated land area (ha) for the corresponding period: for tractors-the total area of cultivated land in the Stavropol Krai, for grain combine harvesters-only cultivated land area under grain crops.

Statistical Processing
To identify statistically significant differences in all averaged values of the studied hydrological and other characteristics between the aforementioned monitoring periods, the Student's t-test was used. All of the averaged values were calculated with a 95% confidence interval.

Long-Term Trends in Water Discharge
Analysis of the available data on the long-term water discharge dynamics at all three hydrological stations along the Kalaus River allowed drawing the following findings ( Figure 4).
From the 1930s-1940s to the 2000s-2010s, there has been a steady upward trend in the discharge of the Kalaus River in most of its basin. The mean annual discharge in 2008/2010-2017 increased by about two times compared to 1963-1980, and 4.4 times compared to 1930/1932-1952/1953 in the middle and lower reaches of the river (see Figure 4a). However, in the river's upper reaches, the discharge was even lower in the recent decades than in 1963-1980. The noted trend was mainly due to an increase in the baseflow (base discharge), which was expressed, for example, by a significant rise in minimal water discharge of the river, especially in its middle and lower reaches: between 1963-1980 and 2008/2010-2017, the minimal (per annum) water discharge increased there by 9.2-9.7 times, while in the upper reaches of the river, only by 2.5 times (see Figure 4c). On the other hand, maximal (per annum) water discharge in the river in 1963-1980 also had a general tendency to increase (by 47-60% compared to 1930/1932-1953), which, after the 1980s, was most likely replaced by the opposite trend (a decrease by 31-66% between 1963-1980 and 2008/2010-2017) (see Figure 4b).
Based on the temporal dynamics of differential specific water discharge (see Equation (1)), the sub-basin of the upper reaches of the Kalaus River turned out to be the territory with the greatest contribution to the formation of maximal discharge during the entire analyzed time (Figure 5b). It is noteworthy that in this part of the river basin, there have been significant reduction in the maximal discharge between 1963-1980 and 2008/2010-2017.
The differential specific minimal discharge (see Equation (1)) of the Kalaus River did not change much between the analyzed hydrological stations in 1963-1980, increasing slightly in the sub-basin of the middle reaches. Over the last decades, the situation has . Contemporary long-term changes in Kalaus River's water discharge at the analyzed hydrological stations (see Figure 1; Table 1). WD mean -mean annual water discharge (a); WD max -maximal (per annum) water discharge (b); WD min -minimal (per annum) water discharge (c); N-an indicator of intra-annual variability, or intra-annual amplitude of water discharge (N = WD max /WD min ) (d); 1-the second-degree polynomial trend line, 2-the exponential trend line; p-the probability of statistically significant difference between 1963-1980 and 2008/2010-2017 (the Student's t-test).
The aforementioned long-term discharge dynamics resulted in a significant reduction in its intra-annual variability or amplitude (N, the difference between maximal and minimal discharge per annum) in the Kalaus River over the past decade compared to 1963-1980. This reduction was substantial in the river's middle and upper reaches (see Figure 4d).
Based on the temporal dynamics of differential specific water discharge (see Equation (1)), the sub-basin of the upper reaches of the Kalaus River turned out to be the territory with the greatest contribution to the formation of maximal discharge during the entire analyzed time (Figure 5b). It is noteworthy that in this part of the river basin, there have been significant reduction in the maximal discharge between 1963-1980 and 2008/2010-2017. changed dramatically due to a sharp increase in the minimum discharge river's middle and lower reaches (see Figure 5c).  (1)) of th the analyzed hydrological stations (see Figure 1; Table 1)

Long-Term Trends in Suspended Sediment Load/Concentration
Analysis of the available data on the long-term dynamics of sedim tration at all three hydrological stations along the Kalaus River allowed lowing findings: Since the 1930s-1940s (at least), there has been a steady general upw iment load/concentration of the Kalaus River, which lasted until the begin (or maybe later). Moreover, the trend in suspended sediment load wa ( Figure 6).
Herewith, the sharpest increase in the content of suspended sedime occurred after 1968: during 1968-1980, a concentration of suspended se with a value of more than 1000 g m -3 was observed during, on average, annual time, while a concentration of more than 5000 (but less than 50,0 41%. For comparison, during 1948-1967, these values were only 14% a tively (Supplementary Material 2). Due to the period 1968-1980 (the mid laus River was considered one of Russia's most sediment-reach rivers. A (the mid-1990s?), sediment load/concentration began to reduce at all th drological stations along the Kalaus River (see Figure 6).
When analyzing the temporal dynamic of sediment load, it is impo Figure 5. Changes in differential specific water discharge (see Equation (1)) of the Kalaus River at the analyzed hydrological stations (see Figure 1; Table 1)  The differential specific minimal discharge (see Equation (1)) of the Kalaus River did not change much between the analyzed hydrological stations in 1963-1980, increasing slightly in the sub-basin of the middle reaches. Over the last decades, the situation has changed dramatically due to a sharp increase in the minimum discharge, especially in the river's middle and lower reaches (see Figure 5c).

Long-Term Trends in Suspended Sediment Load/Concentration
Analysis of the available data on the long-term dynamics of sediment load/concentration at all three hydrological stations along the Kalaus River allowed drawing the following findings: Since the 1930s-1940s (at least), there has been a steady general upward trend in sediment load/concentration of the Kalaus River, which lasted until the beginning of the 1980s (or maybe later). Moreover, the trend in suspended sediment load was more essential ( Figure 6). anthropogenic) factors. On the contrary, the relatively weak formation of sediment load in the river basin was dated to 2008-2017, despite the rather large range of discharge changes along the river during these years (see Figure 7).   Herewith, the sharpest increase in the content of suspended sediment in river waters occurred after 1968: during 1968-1980, a concentration of suspended sediment particles with a value of more than 1000 g m −3 was observed during, on average, about 75% of the annual time, while a concentration of more than 5000 (but less than 50,000) g m −3about 41%. For comparison, during 1948-1967, these values were only 14% and 4.4%, respectively (Supplementary Material 2). Due to the period 1968-1980 (the mid-1990s?), the Kalaus River was considered one of Russia's most sediment-reach rivers. After the late 1980s (the mid-1990s?), sediment load/concentration began to reduce at all three analyzed hydrological stations along the Kalaus River (see Figure 6).
When analyzing the temporal dynamic of sediment load, it is important to consider its changes relative to discharge in river basins to minimize surface water runoff's role when analyzing the causes of temporal variability in the load formation. Figure 7 shows a more or less clear differentiation of groups of years with a various mean annual load of the Kalaus River relative to the mean annual discharge of the river in these years. The years of the relatively active (enhanced) forming the load of sediment in the river's basin are comparatively compactly grouped in the time interval from 1960 to 1980. This indicates that, other things being equal, at the same mean annual water discharge, the increased mean annual load of sediment during this period could be due to external (mainly anthropogenic) factors. On the contrary, the relatively weak formation of sediment load in the river basin was dated to 2008-2017, despite the rather large range of discharge changes along the river during these years (see Figure 7). igure 6. Contemporary long-term changes in mean annual suspended sediment load (SSLmean) and suspended sediment ncentration (SSC) of the Kalaus River at the analyzed hydrological stations (see Figure 1; Table 1). SSC(1)-averaged aximal (per annum) SSC; SSC (2)   To correctly analyze the effect of changing environmental conditions within the river basin's surface on sediment load/concentration, it is important to evaluate the spatiotemporal changes in the differential suspended sediment yield [57]. The largest relative and absolute decrease in this yield between 1963-1980 and 2008/2010-2017 occurred in the sub-basin of the upper reaches of the Kalaus River (upriver of Sergiyevka-by 12.5 times); the smallest (and statistically insignificant)-in the sub-basin of the middle reaches of the river, between Sergiyevka and Svetlograd-by 1.5 times (see Figure 8e). It follows from these facts that the sub-basin of the upper reaches of the Kalaus River during 2008/2010-2017 ceased to be the predominant source of suspended sediment formation in the entire river basin compared to 1963-1980 when this sub-basin exceeded the sub-basin of the middle reaches by the intensity of river sediment formation by 3.4 times, and the sub-basin of the lower reaches-by 15.5 times. The sub-basin of the contingently lower reaches of the Kalaus River, located between Svetlograd and Vozdvizhenskoe (about 50% of the entire studied part of the river basin), reduced its role as a source of sediment in the river by more than half in 2008/2010-2017 compared to 1963-1980 (see Figure 8d,e). Thus, the main territory of river sediment formation in the Kalaus River basin in 2008/2010-2017 became the sub-basin of the river's middle reaches, which is currently more than twice active concerning sheet/rill/gully erosion and sediment formation than the sub-basin of the upper reaches of the river. The proportion of transit suspended sediment in the lower and especially in the middle reaches of the river decreased between the indicated periods ( Figure 8f). Consequently, the middle sub-basin of the Kalaus River should become the arena of the most active erosion-control measures and sediment interception measures in the present and the future. between 1963-1980 and 2008/2010-2017 occurred in the upper reaches of the river (by 6.6-12.5 times), the smallest-in its middle and lower reaches (by 3.3-6.0 times) (Figure 8a). During 1963-1980, the integral suspended sediment yield tended to decrease regularly downriver. In 2008/2010-2017, this tendency was replaced by the one in which the largest integral suspended sediment yield was observed in the middle reaches of the river, the smallest-in the upper reaches (see Figure 8b,c). Figure 8. Contemporary long-term changes in some spatiotemporal characteristics of suspended sediment load/yield/concentration of the Kalaus River at the analyzed hydrological stations (see Figure 1; Table 1). Z-the ratio between 1963-1980 and 2008-2017 (times) for: 1-mean annual suspended sediment load (SSLmean), 2-mean annual suspended sediment concentration, 3-maximal (per annum) suspended sediment concentration (based on the data from Figure 6) (a); 4-the power trend line for SSYmean-annual suspended sediment yield (Mg km -2 y -1 ) upriver of the corresponding station during the corresponding monitoring period (c) (M1-upriver of Sergiyevka, M2-upriver of Svetlograd, M3-upriver of Vozdvizhenskoye (b)); ∆SSYmean-differential annual suspended sediment yield (Mg km -2 y -1 ) (see Equation (2)) (e); Sriver basin area (S1-upriver of Sergiyevka, S2-upriver of Svetlograd, S3-upriver of Vozdvizhenskoye); ∆S-differential river basin area; ∆-relative reduction in ∆SSYmean; I-mean riverbed gradient upriver of the corresponding station; δthe ratio of the difference in SSLmean between neighboring hydrological stations (i.e., ∆SSLmean) to the river's length between these stations (∆L) (for the station at Sergiyevka-between the station and the river's headwater): δ = ∆SSLmean/∆L, kg s -1 10 km -1 (d); the probability of statistically significant difference (the Student's t-test) between: p-1963-1980 and 2008- Contemporary long-term changes in some spatiotemporal characteristics of suspended sediment load/yield/concentration of the Kalaus River at the analyzed hydrological stations (see Figure 1; Table 1). Z-the ratio between 1963-1980 and 2008-2017 (times) for: 1-mean annual suspended sediment load (SSL mean ), 2-mean annual suspended sediment concentration, 3-maximal (per annum) suspended sediment concentration (based on the data from Figure 6) (a); 4-the power trend line for SSY mean -annual suspended sediment yield (Mg km −2 y −1 ) upriver of the corresponding station during the corresponding monitoring period (c) (M 1 -upriver of Sergiyevka, M 2 -upriver of Svetlograd, M 3 -upriver of Vozdvizhenskoye (b)); ∆SSY mean -differential annual suspended sediment yield (Mg km −2 y −1 ) (see Equation (2)) (e); S-river basin area (S 1 -upriver of Sergiyevka, S 2 -upriver of Svetlograd, S 3 -upriver of Vozdvizhenskoye); ∆S-differential river basin area; ∆-relative reduction in ∆SSY mean ; I-mean riverbed gradient upriver of the corresponding station; δ-the ratio of the difference in SSL mean between neighboring hydrological stations (i.e., ∆SSL mean ) to the river's length between these stations (∆L) (for the station at Sergiyevka-between the station and the river's headwater): δ = ∆SSL mean /∆L, kg s −1 10 km −1 (d); the probability of statistically significant difference (the Student's t-test) between: p-1963-1980 and 2008-2017, p 1 -A and B, p 2 -B and C; P-the proportion of suspended sediment of "local" (5; in the corresponding sub-basin) and transit (6) origin at the studied hydrological stations (f).
There was a generally steady temporal shift tendency in maximal (per annum) mean monthly water discharge and suspended sediment load in the Kalaus River from early spring to early summer since the 1930s. In the 1930s-1950s, these maximums were observed, on average, in March, whereas in 2010-2017-in June (Figure 9). Despite several abnormal years (1950, 1953, 1965, 1971, 1973, 1977, 1978, 2012, and 2013), when discrep-ancies between these maximal mean monthly water discharge and sediment load were recorded (see Figure 9c), these changes in discharge/load were observed in the same month in the majority of cases. An explanation of the highlighted anomalies is given in Supplementary Material 3. Figure 9. Temporal shift tendencies in the month with the maximal mean monthly water discharge (a; WD m/max ) and suspended sediment load (b; SSL m/max ) of the Kalaus River at the town of Svetlograd (see Figure 1; Table 1) from the 1930s to 2017. 1-the power trend line; (c) graphical linear correlation between WD m/max and SSL m/max . a Hereinafter, the averaged month for the corresponding period.

Contemporary Hydrological Changes in the Basins of Kuma River and the Additionally Studied Rivers of the North Caucasus Region
Throughout the studied length of the Kuma River, as well as in its tributaries, there were steady and statistically significant in the overwhelming majority of cases downward trends in mean annual SSL between 1963-1980 and 2008-2017, despite the regular increase (statistically insignificant in most cases) in mean annual discharge in this river basin ( Figure 10; Table 2). Especially essential reduction in sediment load/yield occurred in the highest parts of the Kuma River basin, i.e., the lower part of the Greater Caucasus's northern slope and on the Stavropol Upland (the Tomuzlovka River basin). Downriver, these changes were less significant. Figure 10. Changes in mean annual water discharge (WDmean) and mean annual suspended sediment load (SSLmean) along the Kuma River (a western tributary of the Caspian Sea, SW Russia) between 1963-1980 and 2008-2017. ∆M(WDmean) and ∆SSYmean-differential specific water discharge and suspended sediment yield, respectively; S-river basin area; ∆S-differential river basin area; the sixth-degree polynomial trend line: 1-for WDmean, 2-for SSLmean; p-the probability of statistically significant difference for the indicated averaged values between the periods (the Student's t-test); ∆-relative changes; Z-absolute changes (times). Hydrological stations at (see Figure 1; Table 1): Bk-Bekeshevskaya, A-Aleksandriyskaya, N-Novozavedennoye, Z-Zelenokumsk, Bd-Budyonnovsk, and V-Vladimirovka.
There were unidirectional discharge trends between the two analyzed periods in the Terek River's upper reaches (see Table 2; Figure 1). However, in sediment load/yield, its decrease trends were traced in all rivers of this basin, but they were statistically insignificant (see Table 2).
No unidirectional and statistically significant discharge trends were revealed between 1963-1980 and 2008-2017 for the three studied rivers' basins in the western part of the Greater Caucasus's northern slope (see Figure 1). However, the changes in load/yield and ∆SSY mean -differential specific water discharge and suspended sediment yield, respectively; S-river basin area; ∆S-differential river basin area; the sixth-degree polynomial trend line: 1-for WD mean , 2-for SSL mean ; p-the probability of statistically significant difference for the indicated averaged values between the periods (the Student's t-test); ∆-relative changes; Z-absolute changes (times). Hydrological stations at (see Figure 1; Table 1): Bk-Bekeshevskaya, A-Aleksandriyskaya, N-Novozavedennoye, Z-Zelenokumsk, Bd-Budyonnovsk, and V-Vladimirovka. Table 2. Period-to-period changes in mean annual water discharge (WD mean , m 3 s −1 ), mean annual suspended sediment load (SSL mean , kg s −1 ), and annual suspended sediment yield (SSY mean , Mg km −2 y −1 ) of some analyzed rivers of the North Caucasus region (see Table 1; Figure 1) during 1963-2017 (∆-relative changes). Terek River basin:

River/Hydrological Station (River Basin Area) Variables
Terek River/Kotlyarevskaya (8920 km 2 ) There were unidirectional discharge trends between the two analyzed periods in the Terek River's upper reaches (see Table 2; Figure 1). However, in sediment load/yield, its decrease trends were traced in all rivers of this basin, but they were statistically insignificant (see Table 2).
No unidirectional and statistically significant discharge trends were revealed between 1963-1980 and 2008-2017 for the three studied rivers' basins in the western part of the Greater Caucasus's northern slope (see Figure 1). However, the changes in load/yield were expressed in upward trends in all of these basins, but they were also statistically insignificant (see Table 2). Figure 11 cartographically summarizes all of the above information on the distribution of contemporary trends in mean annual water discharge and suspended sediment load in all 21 studied river basins of the region, taking into account their statistical significance. The general prevailing tendency to increase in the last more than half a century is clearly visible in water discharge. In contrast, the suspended sediment load showed a predominant tendency to decrease in the analyzed rivers. Moreover, the following relatively clear pattern was observed: all studied rivers flowing east of the submeridian upper reaches of the Kuban River tended to decrease the sediment load in the specified period, while rivers flowing westward, but within the northern slope of the Greater Caucasus, tended to increase suspended sediment (a statistically insignificant increase).

Discussion
Considering the study region's physical geographic conditions, the leading causes of the above-mentioned long-term changes in discharge and load of the Kalaus River and Kuma River basins could be, in our opinion, changes in climate and land use/cover. We exclude any significant role of the processes associated directly with the Earth's internal  Table 1); trend changes: b-a statistically significant (p ≤ 0.05) increase, c-a statistically insignificant (p > 0.05) increase, d-a statistically significant decrease, e-a statistically insignificant decrease.

Discussion
Considering the study region's physical geographic conditions, the leading causes of the above-mentioned long-term changes in discharge and load of the Kalaus River and Kuma River basins could be, in our opinion, changes in climate and land use/cover. We exclude any significant role of the processes associated directly with the Earth's internal energy (mainly tectonic movements) in the indicated dynamics.

The Impact of Climate and Climate-Induced Changes
The collected and processed data on climate change in the region under study, based on long-term air temperature/precipitation observations at two meteorological stations closest to the Kalaus River and Kuma River basins, are summarized in Table 3. Table 3. Changes in monthly/annual air temperature and precipitation at meteorological stations in Stavropol and Budyonnovsk (the Stavropol Krai, SW Russia, see Figure 1) between 1963-1980 and 2008-2017.  Following these data, it was possible to reveal the influence of modern climate change on discharge and load/yield of the rivers (on the example of the Kalaus River) during the main seasons of their formation-the spring snowmelt period (snowmelt-induced flood, mostly March) and the warm (rainy) season (partly March to November).

Changes in Surface Water/Sediment Runoff Formation during Snow Melting
In the relatively cool/cold months in the region (November to March), only in March, there was a statistically significant rise in air temperature (see Table 3). This trend was accompanied by a statistically significant increase in rainfall this month, sometimes in the snowfall form.
A distinctive feature of both this region and Russia s vast expanses outside it is that the snow, falling during the winter months, accumulates there in the form of snow cover, except for evaporation losses. With the advent of spring, the cover melts, forming a surface runoff and carrying out the soil erosion destruction primarily where there are favorable environmental conditions for this (mainly within arable land). Figure 12 demonstrates the combined diagrams of temporal dynamics of temperature on the soil (to a depth of not more than 10 cm from the surface) in March and the depth of snow cover in February, which are derivatives of the changes mentioned above. We have purposefully identified the two first 10-day periods in March in terms of soil temperature. In Russia's southern regions, snowmelt processes occur comparatively quickly, not dragging on as long as, for example, in the East European Plain's forest zone for a month or more. Consequently, it is essential to show when the processes of formation and temporary changes in the snowmelt runoff and erosion processes caused by it are the most active.
A distinctive feature of both this region and Russia´s vast expanses outside it is that the snow, falling during the winter months, accumulates there in the form of snow cover, except for evaporation losses. With the advent of spring, the cover melts, forming a surface runoff and carrying out the soil erosion destruction primarily where there are favorable environmental conditions for this (mainly within arable land). Figure 12 demonstrates the combined diagrams of temporal dynamics of temperature on the soil (to a depth of not more than 10 cm from the surface) in March and the depth of snow cover in February, which are derivatives of the changes mentioned above. We have purposefully identified the two first 10-day periods in March in terms of soil temperature. In Russia's southern regions, snowmelt processes occur comparatively quickly, not dragging on as long as, for example, in the East European Plain's forest zone for a month or more. Consequently, it is essential to show when the processes of formation and temporary changes in the snowmelt runoff and erosion processes caused by it are the most active. Figure 12. Contemporary long-term changes in March temperature of the topsoil (≤10 cm deep) (a) and snow cover depth (SD) (b) in February, according to data from the Stavropol's meteorological station (see Figure 1). 1-averagely for the first 10-day period of March; 2-averagely for the second 10-day period of March; R 2 -the coefficient of approximation for the sixth-degree polynomial trend line (the linear trend for SD/T correlation); γ-the proportion of years with temperatures below 0°C in the corresponding period; λ-the proportion of years with a snow cover depth in February of more than 10 cm in the corresponding period; Cv-the coefficient of variation; r-the coefficient of linear correlation. On the SD diagram (c), qualitative assessment of the possibility of water runoff formation caused by snow melting on the soil (based mainly on combined temporal Figure 12. Contemporary long-term changes in March temperature of the topsoil (≤10 cm deep) (a) and snow cover depth (SD) (b) in February, according to data from the Stavropol's meteorological station (see Figure 1). 1-averagely for the first 10-day period of March; 2-averagely for the second 10-day period of March; R 2 -the coefficient of approximation for the sixth-degree polynomial trend line (the linear trend for SD/T correlation); γ-the proportion of years with temperatures below 0 • C in the corresponding period; λ-the proportion of years with a snow cover depth in February of more than 10 cm in the corresponding period; C v -the coefficient of variation; r-the coefficient of linear correlation. On the SD diagram (c), qualitative assessment of the possibility of water runoff formation caused by snow melting on the soil (based mainly on combined temporal dynamics of T(1) and SD during 1977-2018) is: 3-not assessed owing to lack of data on T(1) until 1977; 4-low and very low; 5-high and very high.
Since the late 1980s, according to data from the Stavropol's meteorological station, a steady tendency towards an increase in the temperature of the topsoil was found, especially in the first half of March (with a sufficiently high statistical significance), when snow melting is usually observed in the upland (see Figure 12a). After 1997, the topsoil's negative temperatures at the beginning of March became more an exception than the rule that dominated until the mid-1980s (including, most likely, the 1950s-1970s). Due primarily to an increase in air temperatures in March, this trend was also partially determined by snow cover depth changes. This was especially well-manifested when comparing two periods: 1977-1997 and 1998-2018 (see Figure 12b). The snow cover's warming effect on the soil was comparatively effective, primarily with snow depths more than 10 cm (the direct and relatively close relationship between snow depths and soil surface temperatures). An inverse relationship was observed between these characteristics (see Figure 12c) with snow depths of less than 10 cm. It indicates, most likely, the essential role of the thermal air balance in forming the thermal balance of the topsoil in the absence of a relatively deep snow cover. The smaller this depth (<5 cm), the stronger the influence of the thermal balance of air (direct solar radiation is also possible) on temperatures of the topsoil, which determine the depth of its freezing before and during the snow-melting time.
In the context of our study, the most important is the resultant of all of these changesa change in the conditions of formation of water runoff and sediment load/yield in the studied rivers of Ciscaucasia. According to the Barabanov's rule [63,64], formulated based on long-term observations at experimental runoff plots in the steppe and forest-steppe zones of European Russia, at a certain minimum value of one of the three limiting factors (water reserves in the snow, soil freezing depth, and soil moisture), no surface water runoff is formed regardless of the indicators of other factors. If the soil is frozen to ≤50 cm deep (the farther south, the less this critical level), then there is no surface runoff regardless of its moisture and water reserves in the snow. The bulk of the meltwater is filtered into the soil and then into underlying rocks, replenishing groundwater volumes. The effect of the depth of soil freezing on surface runoff is also confirmed by modeling and field experiments [65][66][67][68].
Unfortunately, we did not have data on the depth of soil freezing over the last several decades in the region. Still, there is a direct and close correlation (ceteris paribus) between this value and the topsoil's temperature [63,64,[69][70][71]. Hence, we had reason to restore (at least at a qualitative level) the probability of surface runoff and its long-term dynamic over the past few decades. As shown in Figure 12b, the active formation of snowmeltinduced surface water runoff could potentially be in 1977-1997: of 21 years of this period, 13 years (62%) could well have this runoff during the first half of March. This runoff could carry out active erosion work and, therefore, sediment formation.
It can also be assumed with high probability that even before the 1980s, the conditions for the formation of snowmelt runoff and snowmelt-induced sheet/rill/gully erosion in the Kalaus River basin and most of the Kuma River basin differed little or were even more favorable than in the 1980s-1990s. We do not have direct data on soil temperature to verify our findings. Nonetheless, the fact that winters (especially in January and February) in the 1960s-1980 were more severe than since the 2000s (see Table 3) is an essential indirect argument in favor of these findings, taking into account the Barabanov's rule noted. Moreover, a smaller amount of precipitation (the depth of snow cover), especially in January (see Table 3), could also contribute to more significant soil freezing in the 1960s-1980.
The following period, 1998-2018, was contrasted concerning the period 1977-1997. Snowmelt runoff and erosion processes in the Kalaus River basin, taking part in the river's sediment load formation, could be observed only during three years (14%) in 1998-2018. In most years, there was an active transformation of spring surface meltwater into soil/subsoil (mainly groundwater) runoff during this period. The consequence of this was a previously marked reduction in maximal water discharge (in the river's upper reaches, see Figures 4a and 5b), maximal and mean annual concentration of suspended sediment, and the load (annual values) of suspended sediment of the Kalaus River and the Kuma River (see Figures 6, 8 and 10) in the last decades.
As for climate monitoring data on the eastern part of the Stavropol Upland, the climatic trends noted above were also stored there in the last decades, according to the Budyonnovsk's meteorological station. The following temporal dynamics were found there: in temperatures on the surface of the soil (the second 10-day period of March): 1977-1997: 0.96 ± 1.61 • C, 1998-2018: 3.81 ± 1.48 • C (0.01 < p < 0.05); in depths of the snow: 1938-1973: 7.1 ± 3.3 cm, 1977-1997: 4.6 ± 2.2 cm, 1998-2018: 3.6 ± 3.4 cm (for all periods p > 0.05). These data on Budyonnovsk can also be considered quite representative for the Kuma River s entire basin (see Figure 1), at least for its larger, plain part. Table 4 summarizes the inter-period dynamics of precipitation with different depths in all seasons since 1966 for two meteorological stations in Stavropol and Budyonnovsk. Despite a general decrease in the number of rains (and snowfalls) with a small precipitation depth (≤10 mm) in the last three decades compared with 1966-1980, there was an increase in precipitation with a depth of >20-30 mm. These precipitation depths can create surface water runoff that can erode the soil and form sediment. An increase in heavy rainfall (in the warm season) is a general trend that was observed almost throughout the southern half of the East European Plain and the Ciscaucasia with global warming. It was proved that the frequency and intensity of heavy rains in Russia as a whole had risen sharply over the past half-century, increasing by 1-2% every ten years [72]. It is likely that the noted dynamic could result in a slight increase in the intensity of soil erosion predominantly on cultivated land of the Stavropol Upland in the warm season and therefore increase the load of suspended sediment in the Kalaus River and Kuma River networks. However, it should be kept in mind that all heavy rains (with a depth of more than 30 mm) occur in the region during the warm season (see Table 4) when cultivated land is comparatively well protected by either crops or stover, which also protects the soil against both water erosion and wind erosion. Therefore, not every heavy rain and not everywhere (depending on sowing crops) could cause proportionally enhanced soil erosion. The temporal shift in the month with the largest discharge and load in the Kalaus River from the beginning of spring (March) to the beginning of summer (June) during the last 80 years (especially since the 1970s), noted earlier in Figure 9, can be explained, on the one hand, by the weakening of surface water runoff caused by snow melting due to warming at the end of the cold season. On the other hand, it could be owing to a relative increase in rainfall-induced runoff in the warm season.

River Water Transfer
The increase in minimal and mean annual water discharge of the Kalaus River and the Kuma River (the middle and lower reaches of the rivers), noted above in Figures 4 and 10, was due to the transfer of water from the neighboring more full-flowing rivers of the region (the Kuban River and Terek River) through the system of irrigation canals, which began to function since the 1960s-1970s (see Supplementary Material 1). For example, the Pravoyegorlykskiy (Right Yegorlyk) Canal, as the principal object of the Yegorlyk irrigation system, which has been operating since the mid-1970s, has an average water discharge into the Kalaus River of about 4-5 m 3 s −1 . This roughly corresponds to the magnitude of the increase in the minimum discharge of the Kalaus River between the periods studied. Here we would like to pay attention to the previously noted decrease in the dependence of sediment load on temporal discharge changes for 2008-2017 in the middle and lower reaches (see Figure 7b,c) of the river compared to its upper reaches (see Figure 7a). The main reason for this, in our opinion, is the aforementioned transfer of water from the basins of neighboring rivers. This water entering the Kalaus River through canals is not dynamically related to sediments coming from the river catchment due to soil/gully erosion, except for those that have formed as a result of riverbed deformations.
Unfortunately, we do not have quantitative data on the volumes of water discharged from the canals directly into the studied rivers, since significant volumes of this water were/are spent on irrigation before reaching the river network. This issue requires further study.

Cultivated Land Area Changes Impact
After the collapse of the Soviet Union in 1991/1992, agricultural degradation took place in Russia and neighboring countries of Eastern Europe, manifested, among other things, by a significant reduction in the total cultivated land area [10][11][12]17,21,73]. This degradation also affected, albeit with less intensity, the Stavropol Krai. The total area of cultivated land in this administrative region of Russia decreased by 20% from 1970-1991 to 2006-2017 ( Figure 13).
The reduction in the area of cultivated land was accompanied by a change in the structure of crops: a slight increase in the area of grain crops (cereals) and a significant reduction in areas of perennial-clover, alfalfa, etc. (by 79%) and annual (by 80%) crops. Regarding perennial crops' anti-erosion properties, it should be noted that they sufficiently reliably protect the soil against erosion due to their dense root system and significant projective cover. For instance, the washing-out of soil material under perennial grasses was on average about 50 times less than on cultivated land, according to the earlier studies by I.A. Kuznik in 1961 [74] in the forest-steppe of the Trans-Volga region of European Russia. The reduction in the area of cultivated land was accompanied by a change in the structure of crops: a slight increase in the area of grain crops (cereals) and a significant reduction in areas of perennial-clover, alfalfa, etc. (by 79%) and annual (by 80%) crops. Regarding perennial crops' anti-erosion properties, it should be noted that they sufficiently reliably protect the soil against erosion due to their dense root system and significant projective cover. For instance, the washing-out of soil material under perennial grasses was on average about 50 times less than on cultivated land, according to the earlier studies by I.A. Kuznik in 1961 [74] in the forest-steppe of the Trans-Volga region of European Russia.
The sown area under sunflower increased significantly (almost by 54%, p < 0.001) between 1970-1987 and 2010-2015 in the krai, from 170.1 ± 28.0 ×10 3 ha to 262.4 ± 15.3 ×10 3 ha [46]. However, it is important to clarify that the main sunflower sown area is currently located chiefly in the western part of the Stavropol Krai [47], west of the main studied river basins of the Stavropol Upland.
This temporal dynamic gives a clear idea of the general change in the Stavropol Krai's sowing fund because all of these crops together occupied (and still occupy) the predominant area in the region (in 1977-1991: 80.2%; in 2006-2017: 86.3%). In general terms, it was also characteristic of the Kalaus River basin, which occupies about 15% of the administrative region's total area. However, in some parts of this basin, agricultural degradation could be varied. Due to the noted reduction in cultivated land area, abandoned land area that is now overgrown with natural meadow-steppe vegetation, which reliably protects the soil against all types of erosion, has increased.
The integral (generalized) erosion resistance index-C-factor (see Equation (3))-decreased insignificantly between the indicated periods (-4.8%), from 0.42 to 0.40, despite a noticeable decrease in the total area of cultivated land. This is explained by the fact that on cultivated land remained after the reduction, the C-factor, on the contrary, slightly increased (9.5%), from 0.42 to 0.46, due primarily to the expansion of cereals and sunflower crops, characterized by an average level of erosion resistance. From the data obtained, it  [46]. However, it is important to clarify that the main sunflower sown area is currently located chiefly in the western part of the Stavropol Krai [47], west of the main studied river basins of the Stavropol Upland.
This temporal dynamic gives a clear idea of the general change in the Stavropol Krai's sowing fund because all of these crops together occupied (and still occupy) the predominant area in the region (in 1977-1991: 80.2%; in 2006-2017: 86.3%). In general terms, it was also characteristic of the Kalaus River basin, which occupies about 15% of the administrative region's total area. However, in some parts of this basin, agricultural degradation could be varied. Due to the noted reduction in cultivated land area, abandoned land area that is now overgrown with natural meadow-steppe vegetation, which reliably protects the soil against all types of erosion, has increased.
The integral (generalized) erosion resistance index-C-factor (see Equation (3))decreased insignificantly between the indicated periods (−4.8%), from 0.42 to 0.40, despite a noticeable decrease in the total area of cultivated land. This is explained by the fact that on cultivated land remained after the reduction, the C-factor, on the contrary, slightly increased (9.5%), from 0.42 to 0.46, due primarily to the expansion of cereals and sunflower crops, characterized by an average level of erosion resistance. From the data obtained, it might be concluded that the dynamic of crop areas, expressed through changes in the C-factor, can be one of the causes of the previously noted sediment load changes in the Kalaus River and the Kuma River. However, its overall effect was most likely not decisive. It is also likely that this influence was much more essential in some sub-basins and small catchments of the Kalaus River and Kuma River basins.
Despite a slight increase in erosion potential on the remaining cultivated land area in the recent decades (growth of the C-factor), it could be partially compensated by reducing the "load" (pressure) of agricultural machinery on plowed soils. The number of the machinery has been steadily (and very significantly) declining in Russia over the past decades after the USSR collapse [75]. However, this decrease in the Stavropol Krai (Table 5) was not so sharp compared to the rest country. Apparently, this circumstance reduced the topsoil's compaction, thereby improving its filtration properties and, consequently, increased its erosion resistance relative to the increased heavy rains, as was mentioned earlier. Information on the mechanism of the impact of agricultural machinery on chernozems can be found, for example, in [76].  It should be noted that in the neighboring agriculturally developed administrative region of Russia, in the Krasnodar Krai, located between the Stavropol Upland and the Black Sea and the Sea of Azov (see Figure 1), a similar tendency in plowland area changes was also observed: this area reduced by almost 15% from 1970-1987 and 1996-2017. At the same time, the intensity of sheet/rill/gully erosion and load/yield rates could also increase, even compared to the late USSR, near large settlements (cities) of the region studied, where active urbanization processes have taken place in the recent decades. For instance, this case was observed near the city of Stavropol in the Ula River basin. This river's mean annual load increased by 1.7 times between 1963-1980 and 2008-2017 (Supplementary Material 4). The influence of urbanization and suburban economy on the long-term dynamics of discharge and load/yield requires a separate thorough study.

Livestock Changes Impact
There was a significant reduction in the number of livestock (cattle, goats, and sheep) ( Figure 14) within old pastures that existed since the time of the late USSR, and new ones formed after abandoning cultivated land in the last decades. A significant decrease in livestock numbers on pastures after the USSR collapse was also noted almost everywhere in the Greater Caucasus's northern slope [75]. Most likely, this resulted in a proportional decrease in the impact on soils of the region through a decrease in the topsoil compaction during trampling [77][78][79] and an increase in the biomass of meadow grassy associations. These associations increase the erosion resistance of the soil. The indicated temporal dynamics were generally correlated well with the dynamics of sediment load changes in the Kalaus River (see Figures 6 and 8) and the Kuma River (see Figure 10).
Under these conditions, it was quite natural, not only the reducing of the intensity of erosion and associated sediment load/yield in various parts of the fluvial network (as confirmed by the temporal dynamics of sediment load of the rivers, noted earlier) but the gradual stabilization of erosion landforms created previously (most likely in the 1950s-1990s); for example, ephemeral and typical gullies (see Figure 14). Under these conditions, it was quite natural, not only the reducing of the intensity of erosion and associated sediment load/yield in various parts of the fluvial network (as confirmed by the temporal dynamics of sediment load of the rivers, noted earlier) but the gradual stabilization of erosion landforms created previously (most likely in the 1950s-1990s); for example, ephemeral and typical gullies (see Figure 14).

Neighboring River Basins of the North Caucasus Region
Based on the results from Table 2, we assume that similar trends in load/yield have occurred in the recent decades in the central part of the Ciscaucasia-to the east of the dividing line between the basins of the Kuma River and Kuban River (see Figures 1 and  11). These trends were more statistically significant for plain river basins (the Kuma River basin) than for piedmont/low-mountain ones (see Figure 11).
In the western part of the Greater Caucasus's northern slope with wetter (more forested) landscapes (see Table 1; Figure 1), the opposite tendency (but statistically insignificant) toward an increase in sediment load/yield in local rivers has been observed in the last decades (see Table 2). A commensurate rise in mechanical denudation rates in the western part of the Greater Caucasus was also indicated by [80] based on sedimentation rates in two small lakes in this region from the 1960s to the present. Therefore, the aforementioned (by the example of the Kalaus River) combined effect of climate and land use/cover changes on reducing sediment load manifested most effectively in the lowland (plain) part of the Ciscaucasia, possibly including the lowland part of the Kuban River basin (we did not have reliable data on river sediment load). This reduction in sediment load also took place in the central part of the Greater Caucasus's northern slope, where the overall (natural-anthropogenic) rates of mechanical denudation are generally high [81]. It was in these parts of the Ciscaucasia and the northern slope of

Neighboring River Basins of the North Caucasus Region
Based on the results from Table 2, we assume that similar trends in load/yield have occurred in the recent decades in the central part of the Ciscaucasia-to the east of the dividing line between the basins of the Kuma River and Kuban River (see Figures 1 and 11). These trends were more statistically significant for plain river basins (the Kuma River basin) than for piedmont/low-mountain ones (see Figure 11).
In the western part of the Greater Caucasus's northern slope with wetter (more forested) landscapes (see Table 1; Figure 1), the opposite tendency (but statistically insignificant) toward an increase in sediment load/yield in local rivers has been observed in the last decades (see Table 2). A commensurate rise in mechanical denudation rates in the western part of the Greater Caucasus was also indicated by [80] based on sedimentation rates in two small lakes in this region from the 1960s to the present. Therefore, the aforementioned (by the example of the Kalaus River) combined effect of climate and land use/cover changes on reducing sediment load manifested most effectively in the lowland (plain) part of the Ciscaucasia, possibly including the lowland part of the Kuban River basin (we did not have reliable data on river sediment load). This reduction in sediment load also took place in the central part of the Greater Caucasus's northern slope, where the overall (natural-anthropogenic) rates of mechanical denudation are generally high [81]. It was in these parts of the Ciscaucasia and the northern slope of the Greater Caucasus (its central part) that the decrease in river sediment load/yield occurred primarily due to a general reduction in the rate of sheet/rill/gully erosion in interfluves of the river basins studied. In particular, this could result from a decrease in mudflows' contribution to river sediment load formation. They were (and still are) considered the most active precisely in the center and the east of the Greater Caucasus [82].
As for the western part of the northern slope of the Greater Caucasus and its foot (approximately west of the dividing line between the basins of the Kuma River and Kuban River), climate warming could have an essential influence on the long-term upward trend of sediment load in local river basins where the rates of natural-anthropogenic mechanical denudation are relatively generally low [81]. This is largely due to the good preservation of forests in large areas in these basins (see Figure 1, Table 1), preventing mechanical denudation. In conditions of statistically significant regional warming and statistically insignificant increase in precipitation (especially in the cold season) (Supplementary Material 5), the influence of the forestland (its relative area) on long-term dynamics could be decisive. In the well-forested landscapes, this can intensify lateral channel erosion (natural riverbed deformations) due to increased discharge and therefore increase its products in riverine waters. According to our preliminary calculations, the relationship between the forestland area in river basins and the growth rates of load/yield in the recent decades was very close for this part of the study region: the linear correlation coefficient was 0.98, according to the three studied rivers (the Belaya, Urup, and Fars rivers).

Comparison with Neighboring Regions of Europe
Finally, mention should be made that the above-mentioned hydrological (a reduction in spring snowmelt-induced discharge) and sheet/rill/gully erosion intensity changes (a decrease in annual river load/yield) in the basins of the main studied plain rivers of the Ciscaucasia (the Kalaus River and the Kuma River), in river basins of the central part of the northern slope of the Greater Caucasus (the upper Terek River basin) and possibly the lowland part of the Kuban River basin, were similar in their common features to those previously described in many regions of the neighboring East European Plain [13][14][15][16][17][19][20][21]23,25,[27][28][29]71,73,83,84], in the Baltic countries (Estonia, Latvia, and Lithuania) [85], as well as in Eastern Scandinavia [86], Poland, Belarus, and in the north of Ukraine [87]. That is, it was a sizeable regional phenomenon for Eastern and Central Europe.
The revealed modern tendencies towards a decrease in sediment load in most of the North Caucasus region's rivers correspond well with the general global tendency to decrease in the large rivers' sediment load [88].

Conclusions
Based on the work made, we have come to the following key conclusions: (i) From the 1960s to the last decades, there was a steady (but not always statistically significant) tendency to increase mean annual water discharge in the Ciscaucasia's rivers studied. The tendency was accompanied by a substantial reduction in discharge's intraannual variability (N) in the rivers, especially in their upper reaches. These changes were chiefly caused by an increase in the discharge (mainly baseflow) in the warm season due to water transfer through irrigation canal systems from the neighboring river basins, characterized by much larger runoff (the Kuban River basin and the Terek River basin). The baseflow increase could be partly due to the redistribution of the spring snowmelt-induced surface runoff to the underground runoff under regional/global warming. The intra-annual variability reduction in the discharge was partly due to a decrease in maximal water discharge, which for the last at least 90 years have a maximal average level in the 1930s-1970s. The most significant reduction in water discharge in the last decades was recorded in subbasins of the upper reaches of the rivers that drain the Stavropol Upland (e.g., the upper reaches of the Kalaus River). As for the studied rivers' discharge of the Greater Caucasus's northern slope, multidirectional and statistically differently significant trends in its changes were revealed there over the past decades.
(ii) There has been a predominant tendency to reduce suspended sediment load in river basins over the past more than half a century in the North Caucasus region. The most significant decreases were noted within the Stavropol Upland between 1963-1980 and 2008-2010: in the Kalaus River basin-from 70 to 92%, in the Kuma River basin-from 47 to 94%. In other words, such an essential decrease in sediment load/yield took place exactly where there were some of the sediment-richest rivers in all of Russia flowed in the middle of the 20th century. A moderate decrease in sediment load/yield was found in some river basins of the central part of the Greater Caucasus's northern slope (17 to 94%). Statistically insignificant tendencies of an increase in suspended sediment load/yield between the periods under consideration were revealed in the river basins in the west of the Greater Caucasus. The greatest suspended sediment reduction rates were found in the sub-basins of the Ciscaucasia's studied rivers' upper reaches. These rivers' sub-basins are characterized by the highest average altitude, a high degree of relief dissection, a relatively more humid climate, and comparatively more severe winters. All of the noted main changes in river load/yield were directly related to a decrease in the intensity of mechanical denudation processes (mainly sheet/rill/gully erosion processes) within these river basins due to joint changes in climate (reducing the depth of soil freezing and reducing the runoff of meltwater on it) and land use/cover (reducing cultivated land area and agricultural machinery "load" (pressure) on it, reducing livestock on pastures, etc.).