The erosivity of rainfall in the watershed of Mellegue varies between 20 and 60 MJ.mm/ha.h.year in 2002 while it fluctuates between 55 and 91 MJ.mm/ha.h.year in 2020 (Fig. 4). The average values deduced from the R factor in 2020 (75 MJ.mm/ha.h.year) are higher than those of 2002 (44 MJ.mm/ha.h.year). The general pattern of rainfall aggressiveness within the basin shows an increasing gradient from North to South. Consequently, the values of R show that the basin is subjected to a high pluvial aggressiveness which reflects a significant erosive power of precipitation on the basin. Regardless of the year, the spatial distribution of R values shows that 30% of the area of the basin (generally located in the north) is exposed to intense erosive power from rainfall.
Vegetation cover factor (C)
The C factor reflects the cover and degree of crop production. For 2002 and 2020, maps of the spatial distribution of C index show that low vegetation density classes are mostly vulnerable to erosion (Fig. 7).
The results obtained from the C factor map in 2002 confirm that 77% of the area of the basin shows a very low vegetation rate (C factor > 0.5) and only 23% of the area is well protected with a C factor < 0.5. This protected area was reduced in 2020 where it occupies only 13% of the total area of the basin (C < 0.5). The low vegetation cover reflects open forests, cultivated land, degraded rangelands and asylvatic areas which are generally very sensitive to erosion. Values below 0.5 are related to dense vegetation such as forests, matorrals and arboriculture while values between 0.5 and 0.9 relate to moderately covered vegetation such as open matorrals and sparse forests. The extreme values of the C factor which tend towards 1 relate to bare soils and harvested field crops.
The analysis of the spatial distribution of the C factor in 2020, compared to 2002, confirms that sensitive areas to water erosion (with a C factor > 0.5) have been well developed, especially downstream of the catchment area. The increase of low vegetation cover areas (C factor > 0.5) has been intensified by the conversion to land that favours water erosion such as cultivated land and steppes. The spatial distribution of C factor confirms that the study area experienced a degradation of dense vegetation following the socio-political events occurred in Tunisia on 2011, manifested in the form of anthropogenic actions and anarchic land conversion that have severely impacted forests and vegetations of the region [53].
Anti-erosion practices factor (P)
Cultivation techniques, commonly used, such as cultivation following the contour lines direction, ridging, terracing or alternating strips and ridging are effective practices for soil conservation against erosion. The values of P factor depend on erosion control or the agricultural practice carried out and also on the slope.
In this study, P values were concluded on the basis of slope. Low and medium values correspond to areas with low to moderate slope. The spatial distribution of the P factor within the catchment area shows that values between 0.55 and 0.6 (areas with low slope) occupy 76% of the catchment area while values between 0. 6 and 0.8 (zones with moderate slope) represent 11.9% of the area. The steeply sloping areas (P between 0.8 and 0.95) define 7.1% of the basin area. Finally, the values which tend towards the external higher limit (equal to 1) and which correspond to land without anti-erosion practices constitute 5% of the study area (Fig. 8).
Soil loss assessment
The estimated soil losses were obtained by combining the factors of the RUSLE model, which are soil erodibility (K), climatic aggressiveness (R), vegetation cover (C), topographic factor (LS) and anti-erosion practices (P). The combination of these factors in a GIS environment provides accurate maps of soil losses which we can observe the spatial distribution of this phenomenon throughout the entire catchment area.
Soil losses in the Mellegue watershed were estimated at 25,584 t/year for 2002 with an average of 1.58 t/ha/year and a standard deviation of 52 t/ha/year, against a total loss estimated at 53,822 t/year in 2020 with an average of 1.78 t/ha/year and a standard deviation of 66 t/ha/year. This states the importance and great variability of the water erosion phenomenon, which has intensified considerably in 2020 comparing to 2002. Soil losses have been grouped into 5 classes (Figs. 9 and 10, Tables 4 and 5). The first-class concerns areas where soil loss is less than 7 t/ha/year. It represents 94.9% of the area of the basin in 2002 and 91.7% in 2020. This class mainly dominates where it is seen throughout the basin. The second-class concerns areas where soil loss is between 7 and 15 t/ha/year. It occupies 2.1% in 2002 and 3.8% in 2020 of the basin total area. The third class represents areas of soil loss estimated between 15 and 30 t/ha/year. This class constitutes 0.9% (90 km²) in 2002 and 1.7% (181 km²) of total area of the basin in 2020. The fourth class contains areas whose soil loss is between 30 and 60 t/ha /year. The occupations of this class are very low since it represents only 0.3% (31 km²) of the total area of the basin in 2002 against 0.8% (85 km²) in 2020. The last class concerns losses in upper soils at 60 t/ha/year. It bears witness to the mountainous areas as well as the areas with friable substrate located on either side of the basin. This class represents 0.3% (33 km²) of the total area in 2002 against 0.5% (57 km²) in 2020.
Thus, in-depth analysis show that the areas with a high erosion rate (greater than 30 t/ha/year) evolved from 64 km² (i.e., 0.6% of the total area) in 2002 to 142 km² (i.e., 1 .3% of the total area) in 2020. This figure reflects a non-compensable soil loss mainly by the effect of pedogenesis. The results show a reduction in areas with a low risk of erosion (between 0 and 7 t/ha/year) against an increase in all the other classes. This confirms the degradation of soils and the intensification of the phenomenon of erosion in the watershed of Mellegue.
These results obtained, either for 2002 or 2020, are very close to the other studies which were established under similar conditions, applied only on upstream of the watershed of Mellegue where low soil loss areas represent 80% of the area of the basin [36]. In addition, the reported results are close to several other studies whose methodology is based on RUSLE modelling where climatic and spatial characteristics are similar to the Mellegue catchment [54, 55].
Minimal soil loss occupies most of the watershed. It is the result of the action of land with a low slope, generally less than 3%. This is also explained by the low values of the LS factor and above all by the protective effect of vegetation in sloping areas. On the other hand, values of soil losses greater than 30 t/ha/year (64 km² in 2002 and 142 km² in 2020) are mainly located in areas with steep slopes where the soil is fragile and composed of marl and clay. This area is characterized by high to very high erosion intensity. The variation in soil losses shows that it is greater downstream than upstream of the basin. The phenomenon intensified further in 2020. Estimated soil losses in the Algerian part of the basin increased from 14,265 t/year (i.e., 56% of the total loss) in 2002 to 30,603 t/year (i.e., 57% of total loss) in 2020 (Table 5). By analogy, the sediments loss from the Tunisian part of the basin have also evolved, starting from 11,319 t/year in 2002 to 23,219 t/year in 2020. Although the Tunisian area belonging to the basin constitutes 40% of the area, it seems that the estimated soil losses of this part are close to the sediment losses from the Algerian territory of the basin. However, the spatial-temporal comparative tools of GIS, carried out on the maps of 2002 and 2020, show the appearance of 13,954 new areas at risk of water erosion in 2020 compared to 2002. Almost 60% of these areas are observed in Algerian territory while the rest is in Tunisia.
Sediments extracted from upstream are likely to settle downstream than being transported out of the basin. In addition, the local sedimentation that took place in the flow axes and the superficial depressions will contribute to the filling of the watercourses and drainage problems may appear [56]. This phenomenon promotes slope instability, gully erosion and landslides. Anthropogenic action influences the intensity of water erosion. The intensification of agricultural land use observed within the watershed during the period from 2002 to 2020 has contributed to accelerating water erosion. Several causes are involved, such as the choice of cropping systems and the excessive exploitation of land as well as the application of unsuitable agricultural techniques, namely: too frequent ploughing and following the slope direction, overgrazing and poor management of irrigation [57].
Many causes can lead to the intensification of soil loss. The area has witnessed urban extension taking place during the period (2002–2020) which took place rapidly at the expense of several classes of land use, in particular forest cover [30]. The artificial soils of the urban environment will prevent the infiltration of rainwater and therefore the volumes of water to be evacuated will be enormous. This process increases soil erosion, which generates different patterns of degradation. The reduction of forest space in the region has increased the process of erosion causing floods and mudslides. The outlet of the watershed forming part of the Jendouba region constitutes the favourable environment for water erosion [58]. In addition, the region has been marked by repetitive historic floods causing material and human damage [59]. Maps dating from 2002 and 2020 show a high rate of mobilized sediment in this region exceeding 1000 t/ha/year.
The integration of the RUSLE model into GIS environment ensured efficient management of the large volume of data relating to the various factors of water erosion. It also possible to generate a synthetic map of the possible erosion rates, expressed in tonnes of sediment lost per hectare per year. This map offers to visualize spatially the distribution of the vulnerability to erosion through the entire area of the Mellegue watershed. However, it should be pointed out that the universal equation of the RUSLE model only makes it possible to evaluate soil losses caused by sheet erosion. The model was based on data applied to small areas, which poses problems of uncertainty when used on a large scale and under different conditions than those where the model was initially adopted. Thus, despite the fact that the reliability of the results obtained is subject to discussion, they can nevertheless guide decision-makers in planning the necessary measures to combat erosion in areas where the risk of erosion is preponderant.
Finally, it should be noted that the application of the empirical RUSLE model under conditions different from those in which it was developed exposes it to numerous errors and criticisms which will lead to a suspect and irrelevant estimate of soil losses. Moreover, the model is considered applicable if all these factors are non-zero and do not highlight the deposits. However, the uncertainty is always tolerable, the results will be more reliable if the field measurements as well as the laboratory analyses are rigorously carried out [35].
Table 4
Comparative soil loss rates in the Mellegue watershed between 2002 and 2020
Class (t/ ha /an) | Erosion rate | 2002 | 2020 |
Tunisia | Algeria | Sum | Tunisia | Algeria | Sum |
Area (Km²) | Area (%) | Area (Km²) | Area (%) | Area (Km²) | Area (%) | Area (Km²) | Area (%) | Area (Km²) | Area (%) | Area (Km²) | Area (%) |
Missing data | - | 70 | 1.6% | 87 | 1.4% | 157 | 1.5% | 72 | 1.6% | 80 | 1.3% | 152 | 1.4% |
0 — 7 | Very weak | 4173 | 94.9% | 5837 | 95.0% | 10010 | 94.9% | 4036 | 91.7% | 5633 | 91.7% | 9669 | 91.7% |
7 — 15 | Weak | 96 | 2.2% | 127 | 2.1% | 223 | 2.1% | 162 | 3.7% | 239 | 3.9% | 400 | 3.8% |
15 — 30 | Moderate | 32 | 0.7% | 58 | 0.9% | 90 | 0.9% | 74 | 1.7% | 108 | 1.8% | 181 | 1.7% |
30 — 60 | High | 15 | 0.3% | 16 | 0.3% | 31 | 0.3% | 34 | 0.8% | 51 | 0.8% | 85 | 0.8% |
> 60 | Very high | 12 | 0.3% | 21 | 0.3% | 33 | 0.3% | 22 | 0.5% | 35 | 0.6% | 57 | 0.5% |
Total | - | 4399 | 100 | 6145 | 100 | 10545 | 100 | 4399 | 100 | 6145 | 100 | 10545 | 100 |
Table 5
Comparative average soil losses in the Mellegue watershed between 2002 and 2018
| 2002 | 2020 |
| Tunisia | Algeria | Total | Tunisia | Algeria | Total |
Number of risk areas | 6775 | 9487 | 16262 | 12583 | 17633 | 30216 |
Total soil loss (t /an) | 11319 | 14 265 | 25584 | 23219 | 30603 | 53822 |
Average (t/ha/an) | 1.67 | 1.50 | 1.58 | 1.85 | 1.74 | 1.78 |
Standard deviation (t/ha/an) | 50 | 15.8 | 52 | 59 | 65.25 | 66 |