Deforestation effects on soil erosion rates and soil physicochemical properties in Iran: a case study of using fallout radionuclides in a Chernobyl contaminated area

Deforestation for farming and grazing purposes has become a global challenge. To study the impact of deforestation on soil erosion rates and soil physicochemical properties, Zarivar Lake watershed, Kurdestan Province, Iran, was selected. Converting the steep hillslopes naturally under oak forest to rainfed vineyards has been one of the most common land-use changes in the area. We used Cs and Pbex radionuclides and quantified the Chernobyl-derived Cs 20 fallout with Pu. The soil samples were collected from two adjacent and similar hillslopes, one of which is under natural forest, while the other is under rainfed vineyard. Using Cs/239+240Pu rates and a simple unmixing of the Cs sources indicated that 50.2±10.0% of Cs was Chernobyl-derived. The mean reference inventory values of Cs, Pbex, and Pu were estimated to be at 6152±1266, 6079±1511, and 135±31 Bq m, respectively. At the forested hillslope, net soil erosion rates based on Cs, and Pbex, techniques were estimated to be at 5.0 and 5.9 Mg ha yr, respectively, resulting 25 in Sediment Delivery Ratios (SDRs) of 96 and 70%. However, at the vineyard hillslope, the net soil redistribution rates were at 25.9 and 32.5 Mg ha yr for Cs and Pbex, respectively, resulting in respective SDRs of around 95 and 92%. Both Cs and Pbex indicated that as a result of deforestation, soil erosion has increased by approximately five times. Percolation Stabilities (PS) in forest and vineyard topsoil are about 309 and 160 gr H2O 600s classified as rapid and moderate PSs, respectively. Rapid PS in forest soil implies high aggregate stability, whereas moderate PS in vineyard soils 30 indicates that they are generally weakly-structured. All in all, the results of the present study revealed that deforestation and converting natural vegetation to cropland prompted soil loss and deteriorated physicochemical properties of the soil.

deforestation and converting forests to cultivated lands have increased drastically. In fact, over the past 50 years, the land-use changes in Iran have been more rapid than ever before and are expected to 40 accelerate in the future (Emadodin et al., 2012). The area of natural forests in Iran has decreased from 19 million ha in the 1950s to 12.4 million ha in the 1990s (DEI, 2003). During the past 50 years, the area of Iran's cultivated land has expanded by more than five times, increasing from 2.6 million ha (Wilber, 1948) to 18.5 million ha (DEI, 2003). Simultaneously, more than 2.5 million ha of Iran's agricultural land has converted to urban areas during the last decade, threatening food security. 45 Consequently, deforestation and cultivation of marginal lands, including fragile upland ecosystems, have become more prevalent in the country. There is general agreement on the fact that this change in land-use on steep slopes leads to an increase in soil erosion (Collins et al., 2001;Nunes et al., 2011;Alatorre et al., 2012;Rudiarto and Doppler, 2013;Zhang et al., 2017;Nabiollahi et al., 2018) and a decrease in soil quality and health (Nabiollahi et al., 2018). Many studies have shown that cultivation 50 results in loss of organic matter (OM) (Vagen et al., 2006;Ozgoz et al., 2013;Karamesouti et al., 2015;Schweizer et al., 2017) which in turn influences the chemical, physical, and biological properties of the soil (Batlle-Bayer et al., 2010;Vinhal-Freitas et al., 2017;Zhang et al., 2017). In general, considerable impacts of deforestation on soils are biological degradation, soil compaction and physical degradation, and accelerated erosion (Lal, 1989). Additionally, land-use change is considered to be one of the main 55 causes of carbon emissions after fossil fuel consumption (Watson et al., 2000).
To study the impact of deforestation on soil physicochemical properties and soil erosion rates on steep slopes, Zarivar Lake watershed, Kurdestan Province, northern-west of Iran, was selected. The watershed is located in the Zagros forests in the Zagros mountain ranges, northern-west and west of the country.
Due to the dominance of oak species, these forests are known as western oak forests. Converting the 60 steep hillslopes naturally under oak forest to rainfed vineyards has been one of the most common landuse changes in the area. The converted farmland is prone to erosion due to steep topography and the inappropriate land management practices (e.g., leaving the soil surface bare, and intensive tillage) in the past and the present. So far, only limited studies sought to quantify soil loss and changes in soil properties in the Zarivar watershed (e.g. Ebrahimi-Mohammadi et al., 2012). (Meusburger et al., 2016). 137 Cs (half-life 30.2 years), is an anthropogenic isotope, produced by atmospheric testing of thermonuclear weapons during the 1950s and 1960s (i.e. global fallout) or the nuclear incidents like the one which took place in Chernobyl NPP in April-May 1986. On the contrary, 70 210 Pb (half-life of 22.3 years) is a geogenic radionuclide, originated from the decay of gaseous 222 Rn, a daughter of 226 Ra (Mabit et al., 2008). A proportion of 222 Rn produced from the lithogenic sources escapes and produces 210 Pb which is deposited on the soil surface and is called "excess" or "unsupported" 210 Pb ( 210 Pbex), because it is not in equilibrium with its parent 226 Ra, therefore, 210 Pbex can be used as a soil redistribution tracer. 75 Generally, the global pattern of global-derived 137 Cs fallout indicates that inputs are distributed between 160 and 3200 Bqm -2 depending on latitude (UNSCEAR, 1969;Garcia Agudo, 1998). To estimate soil redistribution rates using 137 Cs measurements, additional 137 Cs fallout inputs which were originated from the Chernobyl incident must be taken into account. If 137 Cs inventories are to be converted into annual soil erosion rates in the contaminated areas, the proportion of 137 Cs Chernobyl fallout should be 80 determined . To quantify the relative proportion of Chernobyl fallout to the total 137 Cs inventory, plutonium (Pu) radioisotopes have recently been suggested (Alewell et al., 2017) and applied successfully (Alewell et al., 2014;Meusburger et al., 2016;2018; weapon manufacturing, nuclear fuel reprocessing and nuclear power plant accidents (Ketterer and Szechenyi, 2008;Alewell et al., 2017). Globally, it was observed that the Pu 1950s-60s fallout is very similar to that of 137 Cs distribution. However, in contrast to 137 Cs, Pu is found in the non-volatile fraction of nuclear fuel debris released from reactor accidents like the one in Chernobyl accident.
Hence, specific regions of Russia, Ukraine, Belarus, Poland, the Baltic countries, and Scandinavia were 90 the geographic zones over which Pu Chernobyl fallout was deposited (Mietelski and Was, 1995). Therefore, Chernobyl Pu is unlikely to be found in other areas; as a result, it cannot be seen as a significant contributor to the total Pu activity deposited in Iran. Therefore, the relative proportion of Pu isotopes can reveal whether the area received additional 137 Cs deposition from the Chernobyl accident or not. Pu radioisotopes have been applied successfully in Germany (Schimmack et al., 2001;2002), Switzerland (Alewell et al., 2014;Meusburger et al., 2016;2018), Australia (Everett et al., 2008;Tims et al., 2010;Hoo et al., 2011;Smith et al., 2012;Lal et al., 2013), and China (Xu et al., 2013).
With this study, we aim to assess the impact of deforestation on physical degradation of soils in a case study in Iran. Conversion to farmland alters soil properties, and these changes may further enhance soil erosion. An important factor in soil resistance to erosion is aggregate stability. Percolation Stability (PS) 100 is a suitable method to assess aggregate stability (Mbagwu and Auerswald, 1999), with the advantage of less effort compared to the wet sieving method. Through rapid wetting, PS calculates the resistance of aggregates against slaking, thus triggering numerous subprocesses of erosion (Auerswald, 1995). Also, crusts contribute to run off and soil erosion (Lal, 1989;Valentin and Bresson, 1997;Fageria et al., 2010). The crust formation is associated with low organic matter (OM) content and a high percentage of 105 silt (Lal, 1989;Valentin and Bresson, 1997). As monitoring changes in permeability and soil aggregate stability are often time-consuming and costly; various indices were developed for predicting soil crusting using more available data such as soil texture and organic matter content (Udom and Kamalu, 2016). We compared two indices, namely, Sealing Index (SI) (Valentin and Bresson, 1997) and Crusting Index (CI) (FAO, 1978). 110 The present study was undertaken to quantify the impact of deforestation on and soil loss and soil physicochemical properties in a representative region in Northwest Iran. The objectives of this study were (i) to estimate the proportion of 137 Cs Chernobyl input using 239+240 Pu in the study site, (ii) to quantify soil redistribution rates using 137 Cs and 210 Pbex in two adjacent hillslopes under different landuses, and (iii) to evaluate the effects of deforestation on soil loss and soil physicochemical properties.
Meteorological Organization). The average annual, minimum and maximum air temperatures are 12.8, -4.6 and 35.4°C, respectively (IMA, 2008). 125 Zarivar Lake is closed by mountains covered by deciduous oak forests of various species such as Quercus spp. (oaks), Pistacia mutica (wild pistachio), Crategus spp. and Pyrus spp. (Ghazanfari et al., 2004;Haidari et al., 2012). The forest is an open forest type. The ground flora normally presents a high degree of cover (up to 100%) which is exploited for grazing. According to the study of pollen in the lake cores, forests began to grow in the region around 5000 years ago (Van Zeist and Bottema, 1977). 130 The parent materials in ridges are metamorphic rocks of complexly folded black shale and minor metamorphosed limestone and sandstones (Geological survey and mineral exploration of Iran).
However, parent materials in the lowlands consist of Quaternary sediments, namely, lacustrine deposits which include sand, marl, and loam (Geological survey and mineral exploration of Iran). According to soil taxonomy, the soils of the watershed are Typic Haploxeralfs and Typic Haploxerepts (Soil as well as scattered orchards of walnut. Also, rainfed crops such as vineyard and cereal crops, namely, 140 wheat, barley, and lentil, are cultivated in the area. The vineyards are extensively cultivated on sloping lands. Generally, different soil erosion types, including sheet, rill, stream bank, gully are found in the watershed. However, conservation measures like terraces and check dams have been constructed in some parts of the watershed during the last decade.

Soil sampling design
FRNs as soil erosion tracers are normally used by comparing their inventories in sites affected by soil redistribution processes with their baseline inventory at a reference site (Mabit et al., 2008(Mabit et al., , 2012. A crucial step that should be taken when using FRN techniques is to find an undisturbed reference site (Mabit et al. 2008(Mabit et al. , 2012. Here an undisturbed, flat site, having a cover of perennial grass, in clearance from the study sites (Fig. 2). Eleven bulk samples were collected to determine the initial inventory of FRNs at the reference site. For collecting samples, manual soil column cylinder augers (inner diameter of 9 cm) were designed and built for stony soils. Its cutting edge was made up of titanium so that it could be able to penetrate the stony soils easily. A sectioned core was collected at 2 cm increments 155 using a scraper plate sampling device designed by Campbell et al. (1988).
At the two studied hillslopes, a multi-transect sampling approach was adopted. In each site, a total of 33 bulk soil cores have been gathered along the three parallel transects in the main slope direction (Fig. 1).
Soil sampling was performed up to 30 and 40 cm depth in forest and vineyard hillslopes, respectively to ascertain that all FRNs in the soil profile were included. A sectioned core at 5 cm increments was also 160 collected in an erosional site in each hillslope.

Soil sample pre-treatment and Radioisotopic analysis
Soil samples were air-dried, disaggregated, sieved to <2 mm, grounded and then homogenized. Soil samples were analyzed for 137 Cs and 210 Pb and 226 Ra by gamma spectrometry using N-type HPGe detector. Gamma spectrometry measurements were performed in the Iranian Nuclear Medicine 165 Research Institute.
Before analyzing 210 Pb, sub-soil samples were sealed for one month to achieve equilibrium between 226 Ra and its daughter 214 Pb. 137 Cs, total 210 Pb and 214 Pb activities were determined from the net peak areas of gamma rays at 661.6, 46.5 and 352 keV, respectively. The counting times ranged from 12 to 24 h with a precision of 5% to a maximum of 20% at the 95% level of confidence. 170 To determine the proportion of Cs Chernobyl fallout in the samples, 39 bulk and increment samples were selected, which included reference bulk samples, reference increment samples, a transect in forest and increment samples of an erosional site. According to the method described by Ketterer et al. (2012), the samples were prepared for the analysis of Plutonium isotopes. The measurement of Plutonium isotopes ( 239+240 Pu) was conducted using a Thermo X Series II quadrupole ICP-MS instrument at the 175 University of Cadiz, Spain. The ICP-MS instrument was equipped with a high-efficiency dissolving sample introduction system. A detection limit of 0.1 Bq kg -1 for 239+240 Pu was obtained for samples of nominal 1 g of dry-ashed material; for 239+240 Pu activities> 1 Bq kg-1 . The measurement error was 1-3%. 239 Pu and 240 Pu masses existing in the sample (which were determined by isotope dilution https://doi.org/10.5194/soil-2021-2 Preprint. Discussion started: 25 January 2021 c Author(s) 2021. CC BY 4.0 License. calculation) were presented as the summed 239+240 Pu activity so that it could correspond with alpha 180 spectrometric determinations of Pu activity. The 240 Pu/ 239 Pu atom ratios were determined through the same analytical run. Data quality was evaluated through the analysis of preparation blanks (soils or rocks devoid of Pu), duplicates, and control samples with known 239+240 Pu activities.

Soil sampling and soil physicochemical properties
For this study soil samples were collected from two adjacent similar hillslopes, one of which is under natural forest while the other hillslope is under rainfed vineyard, which was deforested in 1981. The 190 hillslopes were selected having similar slope gradient, length, and exposures ( Fig. 2). Disturbed and undisturbed soil samples (for the latter we used a 5 cm diameter and 5 cm high steal ring) were collected along transects at six and seven points in the vineyard and forested hillslope, respectively.
Both undisturbed and disturbed samples were gathered from two depths of 0-20 ad 20-40 cm. The disturbed samples were air-dried, disaggregated and sieved to <2 mm. Soil pH (McLean, 1982) and 195 Electric Conductivity (ECe) (Rhoades, 1982) were determined in the saturated paste and saturated paste extract, respectively. Soil texture, or Particle Size Distribution (PSD) (wet sieving and pipette method, Gee and Bauder, 1986), and bulk density (BD) (core method; Blake and Hartge, 1986) were measured.
Porosity (ƒ) of the soils was calculated using soil bulk density (Danielson and Suterland, 1986). The soil moisture at field capacity (FC) and the wilting point was (PWP) determined by a pressure plate 200 apparatus (Klute, 1986) at 33 and 1500 kPa, respectively. The Available Water Capacity (AWC) was calculated by subtracting FC and PWP. Percolation stability was measured in a laboratory test with airdried 1 to 2 mm aggregates (Siegrist et al., 1998)  Total Organic Carbon (TOC) (Walkley-Black method; Nelson and Sommers, 1982), Carbonate Calcium Equivalent (CCE) (Jackson, 1958) and Cation Exchange Capacity (CEC) (determined using Rhoades, 1982 method) were also measured using standard methods. TN was determined by Kjeldahl method (Bremner, 1996). The soil C:N ratio was calculated by dividing the SOC concentration by the TN 210 concentration. The soil organic carbon stock (SOCs) was computed with the equation proposed by Hiederer and Köchy (2011) as well. Soil permeability was measured using double-ring infiltrometers, and a soil profile was also studied in each hillslope (Scholten, 1997).
Additionally, soil erodibility (K) factor was estimated using Wischmeier and Smith (1978). The equation takes into account the OM content, aggregate stability, infiltration rate, and particle size 215 distribution. Crusting hinders germination of feeble seedlings and reduces infiltration, which on steep slopes, can lead to runoff and soil erosion. To assess soil's susceptibility to crusting, soil physical quality indices were also computed, which included Sealing Index (SI) (Valentin and Bresson, 1997) and Crusting Index (CI) (FAO, 1978) using equations 1, and 2, respectively.
(1) = × 100 + Value of the sealing index, if less than 5%, is considered high crusting or sealing risk, whereas a value 220 more than 9% indicates a low sealing risk and 7% is regarded as the threshold value (Udom and Kamalu, 2016).
The values over 1.6 for CI represent that soils are prone to intense crusting, whereas values below 1.2 are implying low crusting risk. 225 Basically, we performed a statistical comparison between topsoils and subsoils properties of the two land-use types. This was done by one-way analysis of variance (ANOVA). All statistical analyses were performed using SPSS software ver. 22 (IBM, Armonk, NY, USA).
The 137 Cs value obtained in our study site was substantially higher than those reported for reference inventories in Iran so far (Table 1). It was hypothesized that the study site received Chernobyl-derived 137 Cs fallout. The CHELSA (Climatologies at high resolution for the earth's land surface areas) dataset 240 (Karger et al., 2017) confirmed that the study site received around 150, and 100 mm rainfall in April and May 1986, respectively (the time when the radioactive Chernobyl cloud travelled across Iran; Fig.   5). Also, depth profiles of 137 Cs in proximate (< 1 km distance) lake cores showed two distinct peaks (supplementary Fig. 1), which points out to the influence of 137 Cs Chernobyl fallout.
At our study site, the average atomic ratio of 240 Pu/ 239 Pu was 0.184 ± 0.020 ranging from 0.121 to 0.262 245 (Fig. 6). In addition, to determine the proportion of the 137 Cs Chernobyl input at our study site, the mean activity ratio of 137 Cs/ 239+240 Pu from global originated fallout was employed using the value reported by Hodge et al. (1996) at 38.4 (as of 1 July 1994) after values being decay-corrected. The Chernobyl 137 Cs was derived from subtraction of reference site inventories from those calculated using the abovementioned ratio. The 137 Cs Chernobyl input in the study site was estimated to be at 50.2±10.0 %. 250

Fallout radionuclides inventories and soil redistribution rates at slope transects
For 137 Cs and 210 Pbex, 239+240 Pu the MODERN and the DMM were applied at the forested site, while at the vineyard site, the MBM without tillage component was used ( Table 2). The mean and standard deviation of 239+240 Pu inventory was at 130.9±54.5 Bq m -2 at the forested hillslope, 137 Cs inventories were at 5389.9±2227.9 and 4646.6±1921.8 Bq m -2 , respectively for forested and vineyard hillslopes and hillslopes (Fig. 4). The majority of sampling points on the steep forested slope and cultivated field showed a lower inventory compared to those of the reference site for each FRN, implying the dominance of erosive processes. At specific sampling points, 210 Pbex was below the detection limit.
Both 137 Cs and 210 Pbex showed moderately lower average inventories in vineyard hillslope. 260 At forested hillslope, net soil erosion rate estimated by MODERN for 210 Pbex (at 5.9 Mg ha -1 yr -1 ) was slightly more than that of 137 Cs (at 5.0 Mg ha -1 yr -1 ) and 239+240 Pu (at 1.7 Mg ha -1 yr -1 ), yielding respective SDRs of 96, 70 and 66% (Table 2). By contrast, the ones estimated by DMM were considerably different for these two FRNs, standing at 2 and 5.1 Mg ha -1 yr -1 for 137 Cs and 210 Pbex, respectively. At the vineyard hillslope, the net soil redistribution rates estimated by MBM II were at 265 25.9 and 32.5 Mg ha -1 yr -1 for 137 Cs and 210 Pbex respectively, which was about five times more than those estimated by MODERN at the forested hillslopes, resulting in respective SDRs of around 95 and 92% (Table 2). In general, SDRs calculated in both hillslopes using different models were profoundly high, indicating that most of the eroded soil has been removed from the hillslopes.
The correlation between inventories of 210 Pbex and 239+240 Pu with 137 Cs along a transect in forested 270 hillslope was examined (Fig. 7). A stronger positive correlation was observed between 137 Cs and 239+240 Pu (ca. r 2 =0.82) than between 137 Cs and 210 Pbex (ca. r 2 =0.56). Also, the soil redistribution rates of three FRNs in a forested transect were estimated using the MODERN and DMM (Fig. 8). DMM estimated soil losses for 210 Pbex were higher than those of 137 Cs in most sampling points; yet MODERN estimated soil losses and gains for 239+240 Pu were by far the highest along the transect (Fig. 8). 275 The sectioned profiles of 137 Cs, 210 Pbex and 239+240 Pu in an erosional site in the second transect (midtransect) of forested hillslope were compared (Fig. 9 a, b, and c). Total inventories of 137 Cs, 210 Pbex and 239+240 Pu were 5095, 4670 and 19 Bq m -2 , respectively, indicating inventory changes of -17, -23 and -86%. While inventory changes of 137 Cs and 210 Pbex were comparatively similar, that of the 239+240 Pu was notably high (Fig. 9 a, b, and c). Depth distribution of 137 Cs and 210 Pbex at an erosional vineyard site was

Change of physicochemical properties of soils due to land-use change and soil erosion
Deforestation and land-use change from natural vegetation to agriculture have significantly increased ECe, pH, BD, and the K factor, whereas they have significantly decreased OM, TN, C:N, CEC, AWC, 285 ƒ, and PS in surface soil (Table 3). Likewise, there was a significant rise in pH, BD, and K in the subsurface soil of vineyard compared to the subsoil of forest while clay, OM, TN, C:N, CEC, AWC, ƒ, and PS experienced a notable decline (Table 3).
OM and TN in both depths were reduced in the vineyard compared to the forest; however, this reduction was greater in the surface layer (Table 3). According to Table 3, the average OM content of 290 the topsoil of the forest is twice as much as the dry farming vineyard i.e. 2.8 and 1.4%, respectively, which correspond to a carbon stock of 28.8 and 19.8 Mg ha -1 , respectively. Similarly, the OM content in forest and vineyard subsoil were 1.7 and 1.2%, respectively, indicating a carbon stock of 19.5 and 15.2 Mg ha -1 , respectively. The C:N ratio dropped with depth in both soils, most probably as a result of lower OM turnover (Table 3). Moreover, bulk density of the vineyard topsoil was significantly greater than 295 that of forest (1.09 kg m -3 and 1.37 kg m -3 in the forest and vineyard topsoil, respectively) (Table 3).
Similarly, following the deforestation, the porosity of the surface and subsurface soils under vineyard declined by around 11 and 10%, respectively. Mbagwu and Auerswald (1999) classified PS into three classes, namely, rapid (>250 gr H2O 600s -1 ), moderate (250-150 gr H2O 600s -1 ), and slow (<150 gr H2O 600s -1 ). PS in forest and vineyard topsoil 300 were about 309 and 160 gr H2O 600s -1 which classified as rapid and moderate PSs, respectively. Rapid PS in forest soil implies high aggregate stability reflecting the higher OM content, whereas moderate PS in vineyard soils indicates that they are generally weakly-structured. Furthermore, two indices were calculated to evaluate the surface crusting risk. The sealing index (SI) in all soil samples was less than 5% indicating a high risk of sealing. Crusting Index (CI) in both topsoil and subsoil of the forest showed 305 a low sealing risk (<1.2), while it was considered to have a moderate sealing risk in surface and subsurface soils of the vineyard (1.2-1.6). K-values in forest surface and subsurface soils were 0.044 and 0.054, respectively, whereas respective values for surface and subsurface vineyard soils were 0.051 and 0.067 (Table 3). The depth profiles at the reference site of three FRNs showed an exponential decrease with depth (Fig.   3). 137 Cs is mostly associated with the fine mineral fraction (Lujaniene et al., 2002;Qiao et al., 2012) while 239+240 Pu in soils exhibits strong binding to organic matter and sesquioxides. As such, both FRNs 315 show a slight migration into the upper subsurface horizons. 210 Pbex is absorbed by both, organic matter and clay-sized mineral particles, and seem to show less migration into subsurface horizons. Evidently, 210 Pbex does have constant fallout while the other two FRNs don't (Mabit et al., 2008), which can be another reason for its less immigrations into subsurface horizons.
The mean reference inventory value of 137 Cs was substantially higher than those reported for reference 320 inventories in Iran so far (Table 1). However, it should be noted that our study site has a 2-3 times higher MAP as compared to the study sites investigated so far. FRN fallout is strongly related to MAP and is the main cause for high global 137 Cs fallout (Meusburger et al., 2020). Owing to substantial amounts of rainfall after Chernobyl incident (CHELSA dataset; Karger et al., 2017), and the presence of two peaks in depth profiles of 137 Cs in proximate lake cores (supplementary Fig. S1), a hypothesis was 325 formed suggesting that the study site had been contaminated by Chernobyl-derived 137 Cs fallout.
Although other inventories of the 137 Cs in different parts of Iran have not shown unusually high inventories (e.g. Afshar et al., 2010;Ayoubi et al., 2012;Rahimi et al., 2013;Khodadadi et al., 2018), moderately high inventories around 4000 Bq m -2 observed in North of Iran (Gharibreza et al., 2020) were relatively compatible with our findings in the northwest of Iran (Table 1). 330 The 240 Pu/ 239 Pu atom ratio of global fallout in mid-latitudes of Northern Hemisphere is 0.180 ± 0.014 (Kelley et al., 1999) while the ratio of Chernobyl fallout is between 0.37 to 0.41 (Muramatsu et al., 2000;Ketterer et al., 2004). Thus, our measured 240 Pu/ 239 Pu ratios (with average of ca 0.184 ± 0.020) are in accordance with the integrated atmospheric fallout value of 0.18 ± 0.014 (Kelley et al., 1999), implying that the source of Pu is a result of the global atmospheric fallout (Fig. 6). The proportion of the 335  to 80% Alewell et al., 2014), the contribution found in our site was notably less.
The mean reference inventory value of 210 Pbex stood at 6079±1511 Bq m -2 which was moderately higher 340 than the inventory of 210 Pbex at 5825±297 Bq m -2 in central Iran with mean annual precipitation of 330 mm (Khodadadi et al., 2018). The annual deposition flux of 210 Pbex was estimated to be 205 Bq m -2 yr -1 for the study area, the amount of which is within the range of the reported global annual deposition fluxes of 210 Pbex, i.e. from 23 to 367 Bq m -2 .
No data has yet been published on 239+240 Pu for Iran. The estimated reference inventory of 135 ±31 Bq 345 m -2 is considerably higher than the values reported for reference sites in Europe, China and Australia (Table 4). This deviation of 239+240 Pu inventory in our reference site from those values might be attributed to high initial bomb-derived deposition in the study site in 1953-1964, and the differences in the locations and the climates. However, measurements from other parts of the country are required to confirm the hypothesis of a high initial bomb-derived fallout. 350 The coefficient of variations (CV) for all three FRNs were around 20% at the reference site (Fig. 4).
Indeed, there is an inevitable spatial variability of FRNs inside any reference site; however, due to interception of fallout by the tree canopies and their trunks in the forest, a forested reference site, in particular, may represent relatively higher CVs (Wallbrink and Murray, 1996). Such high CVs have also been reported in the literature; Nagle et al. (2000), for instance, reported a CV of 28% for forest and 355 coffee reference sites in tropical mountains. Sutherland (1991) also found CVs ranging from 17.9 to 31.8 in reference sites of Saskatchewan, Canada. However, the 210 Pbex reference inventory values represented a higher spatial variability with a CV of 25%, which was in agreement with Meusburger et al. (2016) who reported a relatively higher CV for 210 Pbex (ca. 28%) than 137 Cs (ca. 18%) and 239+240 Pu (ca. 15%). Teramage et al. (2015) presented a CV of 28% for 210 Pbex at a coniferous forest in Japan. 360 As the key assumption of FRNs method is the homogeneity of the initial fallout in the reference site, the heterogeneity can obscure or even compromise the usage of this method (Haugen, 1992;Golosov et al., 2008;Mabit et al., 2013). As highlighted by Sutherland (1996) and Mabit et al. (2012), the allowable CV was estimated to be 20% for FRNs at 90% level of confidence, with which the selected reference site here agreed, that is to say, our CV was approximately 20% as well. Moreover, 20% CV of 239+240 Pu 365 confirmed that surplus 137 Cs Chernobyl fallout distributed uniformly in the study area. Unusually high https://doi.org/10.5194/soil-2021-2 Preprint. Discussion started: 25 January 2021 c Author(s) 2021. CC BY 4.0 License. heterogeneity of 137 Cs reference inventories was, however, found in the Swiss alpine areas (Alewell et al., 2014) which was essentially because of partial snow coverage at the end of April-May, 1986 and that most likely caused heterogeneous redistribution during the snowmelt process. However, unlike Swiss Alps, the possibility of snow coverage in the so-called time period in the study area was unlikely. At the vineyard hillslope, the net soil erosion rates estimated by MBM II were at 25.9 and 32.5 Mg ha -1 yr -1 for 137 Cs and 210 Pbex, respectively. Both FRNs estimated that as a result of deforestation, soil 380 erosion has increased by approximately 5 times. Gharibreza et al. (2013) who used 137 Cs method showed that following deforestation in a watershed in Malaysia, soil loss magnitude surged by 7-13 times, which was compatible with our findings.
It should be noted that different FRNs inherently account for diverse time spans (Meusburger et al., 2016). Relatively high 137 Cs Chernobyl input at the site under investigation (~50%) reveals that 137 Cs is 385 more likely to be an indicator of a medium-term soil erosion process, i.e. mainly from 1986 onwards.
However, the time frame captured by 239+240 Pu has been recorded from mid-1960s onward. Moreover, although 210 Pbex inventories are influenced by the last 100 years of fallout deposition, due to its ongoing fallout through time, it is more sensitive to the two last decades' erosive events causing soil redistribution (Porto et al., 2013(Porto et al., , 2014. At forested hillslope, the net soil erosion rate estimated by 390 239+240 Pu was far less than that of 137 Cs ( A trend analysis of average annual precipitation of two adjacent synoptic stations, i.e. Sanandaj and Marivan was done using the Mann-Kendall test and Sen's slope estimates (Salmi, 2002). Mean annual 395 precipitation of both stations has shown a clear downward trend from 1959 onwards. Additionally, Balling et al. (2016) showed an increasing trend for extreme precipitation events over the period 1951-2007. Khodadadi et al. (2018 also reported a higher amount of soil loss derived by 10 Pbex compared to that of 137 Cs. As pointed out by Porto et al. (2013Porto et al. ( , 2014, 210 Pbex inventories are more likely to reflect soil redistribution caused by erosive events over the last 15 to 20 years. As found in similar studies, 400 however, measurement uncertainties of 210 Pbex may affect the estimated soil loss magnitudes to some extent. The soil redistribution rates of three FRNs in a forested transect were estimated using both the MODERN model and DDM (Fig. 8). Results obtained for 137 Cs by the MODERN represented a wider range of soil redistribution magnitude, while the DMM outputs tended to level off the extreme soil 405 redistribution rates, which were in accordance with Meusburger et al. (2018).
Inventory changes in sectioned profiles of 137 Cs, 210 Pbex and 239+240 Pu in an erosional site at forested hillslope were relatively different (Fig. 9 a, b, and c). In other words, while inventory changes of 137 Cs and 210 Pbex (-17, and -23 respectively) were quite identical, the inventory change of the 239+240 Pu (-86%), was notably high (Fig. 9 a, b, and c). Supposedly, changes in microtopography of the hillslope 410 over time might have brought about such discrepancies in FRNs inventories. Depth distribution of 137 Cs and 210 Pbex at an erosional vineyard site confirmed that an erosion process has occurred on the site ( Fig.   9 d and e). Yet, 210 Pbex showed a higher rate of soil erosion. Such an incongruity was also found by Wakiyama et al. (2010) when applying 137 Cs and 210 Pbex in Japan. This can suggest that erosion has increased at the sampling point in response to the climate variabilities over the last decades. On the 415 whole, changes in land-use and land management (Gaspar et al. 2013;Zhang et al. 2006), microtopography and the rainfall characteristics during the different timescales might have a significant effect on the estimated soil erosion rates using different FRNs.

Change of physicochemical properties of soils due to land-use change and soil erosion
Soil organic matter content (OM), a key quality index of soil (Pathak et al., 2004), influences the 420 chemical, physical, and biological properties of soil, therefore directly impacting microbial and plant https://doi.org/10.5194/soil-2021-2 Preprint. Discussion started: 25 January 2021 c Author(s) 2021. CC BY 4.0 License.
growth. OM and TN in both depths were reduced in the vineyard compared to the forest (Table 3). Puget and Lal, 2005;Khormali et al., 2009;Hernanz et al., 2009;and Nabiollahi et al., 2018 also reported a considerable reduction in SOM after the land-use change, which were congruent with our findings. This reduction can partly be due to a lower litter turnover and biomass exiting from the 425 hillslope. In addition, tillage operations which are done to eliminate weed growth and conserve moisture are probably going to increase organic-matter oxidation (i.e. mineralisation) by breaking up the soil aggregates. In addition, factors such as low vegetation cover, less shading, increased soil temperature, tillage practices, and subsequent susceptibility of soils to erosion after deforestation are key factors prompting the loss of OM (Carter et al., 1998;Gregorich et al., 1998;Six and et al., 2000). Also, the 430 breaking down of the larger aggregates to smaller ones following deforestation increases the loss of OM (Nardi et al., 1996). Forest soils had higher C:N ratio in both depths. This trend was comparable to those reported by Puget and Lal (2005) and Franzluebbers et al. (2000). In fact, the ratio mostly corresponds to the amount of plant residues entering the soil organic matter pool (Diekow et al., 2005;Puget and Lal, 2005). The C:N ratio dropped with depth in both soils, most probably as a result of lower OM 435 turnover (Table 3).
Bulk density of the vineyard topsoil was significantly greater than that of forest (Table 3). In fact, as stated above, deforestation leads to the reduction of OM; consequently, aggregate stability is reduced, which in turn gives rise to soil aggregate rupture, resulted from slaking. In addition, soil compaction, which is the result of tillage operation, does increase subsequently (Celik, 2005;Puget and Lal, 2005). 440 This will lead to a reduction of ƒ and increase of the BD (Celik, 2005;Bahrami et al., 2010). Similarly, available water content (AWC) of forest topsoil was higher by approximately 5%. The vineyard subsoil retained as much AWC as forest subsoil.
PS in forest and vineyard topsoil were classified as rapid and moderate PSs, respectively. Rapid PS in forest soil implies high aggregate stability reflecting the higher OM content, whereas moderate PS in 445 vineyard soils indicates that they are generally weakly-structured. It is also believed that fungal hyphae play an important role in aggregate stability under forest (Ternan et al., 1996). Mbagwu and Auerswald (1999)  vineyard was roughly as much as that of the subsoil sample forest, implying that it was affected by the land-use change history. From among all properties shown in table 3, the highest coefficient of variability occurred in PS, which was notably higher in vineyard surface and subsurface soils.
Microaggregates resulting from aggregate breakdown clog large pores, thus decreasing infiltration rate and increasing runoff generation (Auerswald, 1995). The sealing index (SI) in all soil samples was less 455 than 5% indicating a high risk of sealing which can be due to a high percentage of silt (Valentin and Bresson, 1997;Lal, 1989). However, SI was significantly higher in both topsoil and subsoil of forest owing to higher OM content. Udom and Kamalu (2016) reported that SI varied from 2.40 to 3.10 in topsoil and from 2.17 to 4.76 in the subsoil of tropical soil susceptible to seasonal flooding, meaning that the risk of sealing was considerably high in their investigated sites. In contrast, Crusting Index (CI) 460 in both topsoil and subsoil of the forest showed a low sealing risk, while it was considered to have a moderate sealing risk in surface and subsurface soils of the vineyard. In other words, although SI indicated a high risk of sealing for all studied soils, CI the suggested that risk of crusting in both surface and subsurface soils of the vineyard were more than those of forest soils. This implies that the CI can be an advantageous index over SI as it can not only take the particle size classes of silt (i.e. fine and course 465 silt) into consideration but also better differentiate between the soils with different OM.
The K factor depends on particle size distribution, permeability, organic matter content, and structure.
K-values in forest surface and subsurface soils were lower than those for vineyard soils. The reported K factor in China ranged from 0.007 to 0.043 (Wang et al., 2013), which is similar to the K-values in the USA varying from 0.004 to 0.063 (Wischmeier and Smith, 1978;Romkens et al., 1997). Comparing 470 these values with our site, we can conclude that there are high values of the soil erodibility in the study area. In other words, the soils of both land-uses are extremely prone to soil erosion owing to the high amount of silt. Following the land-use change, K factor value in the vineyard has increased up to 14% in topsoil and around 20% in the subsoil, which was in agreement with Boix-Fayos (2001); Khormali et al., (2009) andEvrendilek et al. (2004). Based on the study of Sanchez-Maranon et al. (2002), following 475 the land-use change, the K-value has risen up to around 59% in southern Spain. Also, Celik (2005) studied the impact of cultivated land on soil properties in the south of Turkey and reported that cultivation practices have increased the amount of K-value by 2.4. Even though significant differences in clay content in forested and deforested topsoil sites were not observed, the clay content in forest topsoil was slightly less than that of cropland. On the other hand, the 480 clay content of forest subsoil was considerably higher, reflecting that clay lessivation might have occurred in the forest soils. Besides, subsoils characterized by higher clay content may be drawn to the surface by regular deep ploughing, resulting in the minor increase of clay content in the cropland topsoil. All in all, the results of the present study revealed that deforestation and converting natural vegetation to cropland prompted soil degradation, deteriorating physicochemical properties of the soil. 485

Conclusion
Forests and forest soils store carbon, and therefore play a decisive role in mitigating climate change impacts. In Iran, over the past decades, a rapidly growing population has induced an increasing demand for food, so one of the most rapid land-use change, i.e. conversion of land under natural vegetation into arable lands, has been occurring. The present study was undertaken to quantify the impact of 490 deforestation on soil loss with fallout radionuclides and soil physicochemical properties.
The deposition of the 137 Cs fallout from different sources (Chernobyl fallout vs. global fallout) can be determined via 239+240 Pu isotopes if Pu origins exclusively from global fallout. The 240 Pu/ 239 Pu atom ratios in the reference samples confirmed the global fallout origin of Pu. From the 137 Cs/ 239+240 Pu ratio, it was evident that half of the 137 Cs found in the site was Chernobyl-derived. The mean reference 495 inventory of 137 Cs at 6152±1266 Bq m -2 was relatively higher than previous reports in different parts of Iran, which might be due to the location of the site in Iran (the site was closer to the Chernobyl site than other measured sites in Iran), with higher mean annual precipitation, and relatively high rainfall after Chernobyl incident. The mean reference inventories of 210 Pbex estimated at 6079±1511 Bq m -2 was in accord with the reported value in the central part of Iran. The mean reference inventory of 239+240 Pu at 500 135 ±31 Bq m -2 was higher than values reported in other parts of the globe, which might be attributed to high initial bomb-derived deposition in the study site during 1953-1964, and the location and climate diversity. Nevertheless, measurements from other parts of the country are required to confirm the hypothesis of a high initial bomb-derived fallout.
Both 137 Cs and 210 Pbex radionuclides indicated that deforestation has increased annual soil loss by about 505 five times. At the forested site, net soil losses for 137 Cs and 210 Pbex were around 5.0 in and 5.9 Mg ha -1 https://doi.org/10.5194/soil-2021-2 Preprint. Discussion started: 25 January 2021 c Author(s) 2021. CC BY 4.0 License. yr -1 (using MODERN model), respectively, and at the vineyard hillslope, they were about 25.9 and 32.5 Mg ha -1 yr -1 (using DMM), respectively. Notable is that the values obtained by both techniques in each land-use were consistent with each other; however, moderately higher net soil erosion estimated by 210 Pbex at both sites could be a reflection of climate variabilities over the last decades. 510 Converting forest to vineyard resulted in a significant deterioration in soil quality as was indicated by a significant decline in OM, TN, C:N, CEC, AWC, ƒ, and PS, and a significant increase in ECe, pH, BD, and the K factor in both topsoil and subsoil. As a result of deforestation, OM content, which is the most important soil quality indicator, has declined significantly at the vineyard hillslope, leading to a carbon stock loss of about 10.1 and 4.2 Mg ha -1 in topsoil and subsoil, respectively. Furthermore, the land-use 515 change weakened the aggregate stability significantly as the PS approximately decreased by half in the vineyard. It also made the vineyard soils susceptible to erosion as K factor surged by 13 and 19% in topsoil and subsoil, respectively. 239+240 Pu proved to be a valuable tool to quantify the relative contribution of Chernobyl-derived 137 Cs in contaminated areas, which is a prerequisite for applying conversion models with the 137 Cs technique. 520 Obviously, without this knowledge, the results of the 137 Cs method in those areas would not be accurate. Moreover, as found in similar studies, high measurement uncertainties of 210 Pbex are shown to restrain its application to a great extent. In general, having the highest standard deviation among all measured physicochemical properties, PS is identified to be a powerful tool to study soil aggregate stability and soil quality with low costs. All in all, the results of the present study revealed that deforestation and 525 converting natural vegetation to cropland prompted soil degradation and erosion, deteriorating physicochemical properties of the soil.   **Different capital letters indicate significant differences at the p-level<0.05 in subsoil properties of two land-uses.