Comparative assessment of soil and nutrient losses from three land uses in the central highlands of Ethiopia

Land use/land cover change drive changes in several ecosystem processes over short and long terms. In Ethiopia, the main land use/land cover change involves conversion of natural ecosystem into cultivated land. However, a recent change also involves conversion of cultivated and gra zing land into Eucalyptus woodlots. This study was conducted to analyse the effects of such land use/land cover change on soil and nutrient losses. Three land use/ land cove r types (cultivated land, grassland and Eucalyptus woodlot) were selected for a comparative assessment. A total of twelve runoff plots, each with 43.3 m 2 area and with four replications, were installed. Rainfall depth, runoff volume and sediment samples (500 ml) were collected from each plots every morning and evening for 91 days (from 4 th July to 2 nd October, 2015) in the main rainy season. The sediment samples for ten consecutive days were stored in separate containers and composite sediment concentration samples were weighed after being filtered and oven dried for 24 hours at 105 o c. From the samples taken at the end of the rainy season, separate composite a sample before filtration of one litre was analysed in the laboratory for nutrient losses. The effect of land use/land cover on soil and nutrients losses was statistically tested using analysis of variance. The study found that soil loss significantly differed between the land use /land cover types. Soil loss from cultivated land (16.8 ton/ha) was significantly higher than loss from grassland (7 ton/ha) and Eucalyptus stand (8.1 ton/ha). The soil and nutrient losses were positively correlated with runoff volume. There was higher nutrient (N and P) loss from cultivated land than grassland and Eucalyptus. From the results, it can be concluded that soil and nutrients losses are above tolerable limit, and perennial land covers including Eucalyptus stand reduce soil and nutrient losses significantly. This re-affirms the multi-purpose nature of Eucalyptus not only for socioeconomic benefit but also for soil erosion control when planted in appropriate locations.

Eucalyptus has been acclaimed to have social, economic and ecological benefits (Lemenih, 2010;Kebebew and Ayele, 2010;Bekele, 2015).Farmers are continuing planting Eucalyptus converting their farm plots due mainly its positive economic benefits (Mekonnen et al., 2007;Adimassu et al., 2010).However, the uncontrolled expansion of Eucalyptus also is raising controversies with respect to the alleged ecological effect of the species (Jagger and Pender, 2003;Chanie et al., 2013;Jaleta et al., 2016b).Eucalyptus is considered to consume higher water and nutrients and it has allopathic effect on undergrowth vegetation (Nigatu and Michelsen, 1993;Fikreyesus et al., 2011;Chanie et al., 2013).It is also considered as less desirable species for soil-water conservation (Chanie et al., 2013), which has led to banning of its plantings, once in Tigray region, which was latter lifted (Jagger and Pender, 2003).
Soil and nutrient losses from the highlands of Ethiopia is a major environmental challenge owing to high rate of soil erosion (FAO, 1998;Tekele and Hedlund, 2000;Amsalu et al., 2007).The current expansion of Eucalyptus is also feared to intensify the problem.Therefore, it is essential that studies are conducted to assess the effect of Eucalyptus on soil erosion and nutrient loss to help informed decision making for policy makers as well as plantation developers.The objective of this study was therefore to quantify soil and nutrient losses from under Eucalyptus stand in comparison with two other LULC types namely cultivated land and grassland, at Meja River watershed in Central Ethiopia.

Description of the study area
Meja River w atershed, w here this study w as done, found at Jeldu district, in West Shew a, central Ethiopia.Tiki and Sochoa sub catchment of the w atershed w ere selected to install the runoff plots for the study.The catchments are located 114 km w est of the capital, Addis Ababa, Ethiopia.The study sites is located w ithin 9° 02' to 9° 15' N and 38° 05' to 38° 12' E, altitude ranging from 2400 to 3200 m above sea level (Figure 1).The mean annual temperature ranges from17 to 25°C.The rainfall is bi-modal w ith the short rainy season from February to May and long rainy season from June to September.The mean annual rainfall is 1400 mm.The site is characterized by a mixed crop-livestock system.Wheat (Triticum vulgare), potato (Solanum tuberosum) and Enset (Ensete ventricosum) are the crops mostly grow n.The w atershed is know n for Eucalyptus expansion in the central Ethiopia.Eucalyptus w oodlots are abundant in the Meja River w atershed replacing cultivated land and marginal grazing lands.Eucalyptus globulus is the main species of Eucalyptus grow ing in the study area.The soil is characterized as Pellic Vertisol.

Experim ental design
The runoff plot (43.3 m 2 ) w as constructed in three LULC types w ith four replications in tw o adjacent sub catchments.The LULC types are cultivated land, grassland and Eucalyptus w oodlot.The cultivated land covered by w heat and the species of Eucalyptus grow ing is four years old Eucalyptus globules.The grassland w as grazed highly before the instalment of the runoff plots.The details of each runoff plots characteristics w ere given in Table 1.Tw o rain gauges w ere installed to record daily rainfall depth in the tw o sub catchments.The three LULC types at each site w ere adjacent and share other biophysical conditions.The runoff and the rainfall depth w ere registered tw ice daily: at every morning (6:00 am) and evening (6:00 pm).

Data collection and data analysis
Runoff sample of 500 ml w as collected every morning and evening from each runoff plots during the study period (from 4 th July to 2 nd October, 2015).Samples of each plot w ere separately stored for 10 consecutive days in a plastic container, w hich w ere then composited per plot at end of the 10 th day.The composite samples w ere filtered by Whatman (0.42 µm) filter paper.The filtered material w as oven dried by 105°C for 24 h, w eighed and converted to dry mass in milligram per hectare (mg/ha) to yield sediment load.Soil loss (ton/ha) w as then calculated from the volume of runoff and sediment load (Adimassu et al., 2014).
From each runoff plots, at the end of the rainy season, composite runoff sample of one litre w as taken for nutrient analysis in the laboratory.Thoroughly stirred samples w ere kept for 5 h in the laboratory for sedimentation.Then the topmost w ater in each container w as collected for runoff related nutrient analysis.Total dissolved solid, electric conductivity, pH, ammonia (NH3)-N, Nitrate (NO3)-N and Phosphate (PO4)-P w ere analysed at Ethiopian Water Works Design and Supervision Enterprise Water laboratory.The total runoff related nutrient losses of each plot w ere calculated by multiplying the average concentration of each nutrient in the runoff w ith the total runoff volume.Dissolved Ammonia w as analysed by Phenate method using spectrophotometer, modele Eleco SL-160 double beam ultraviolet (Patnaik, 2010).Dissolved Nitrate and Phosphate w as analysed by spectrophotometer, modele Eleco SL-160 double beam ultraviolet (Patnaik, 2010).
Data analysis w as done using Genstat 15 th edition statistical softw are.Analysis of variance w as used to test the effect of LULC changes on soil loss and nutrient loss statistically significant at p < 0.05.Least significant difference (LSD) w as tested to compare the mean values at p < 0.05.The runoff and soil loss mean values regressed among themselves to see the runoff-soil loss relationships.

RESULTS AND DISCUSSION
The runoff volume collected within the study period differed significantly between the LULC types.The mean runoff harvested from the cultivated land is 30% higher than that of grassland and 17% higher than that of Eucalyptus stand ( (154.8 mm) than the grassland (147.8 mm).
Similarly, there was significant difference between in soil loss among the land uses/land cover types.The soil loss from cultivated land was significantly higher than the other two LULC types (Table 3).There was no significant difference in soil loss between the grassland and Eucalyptus field.The soil loss from the cultivated land was greater than the maximum tolerable amount of soil loss which is 10 ton/ha per year in Ethiopia (Hurni and Messerli, 1981;Hurni, 1985).The soil loss from cultivated land recorded in this study matches with the recorded amount from different part of the country (Haile et al., 2006;Girmay et al., 2009;Adimassu et al., 2014).The volume of soil loss has shown the reducing trend as the crop grown fully in cultivated land and resulted comparatively less difference in the final weeks of the experiment on the study area.
Total dissolved solids (TDS), pH, Electric conductivity (Ec), dissolved Ammonia (NH 3 ), dissolved Nitrate, dissolved Phosphate were checked for the significance difference at p < 0.05.The Nitrogen and Phosphorus level in the runoff was analysed indirectly in the form of dissolved Ammonia, Nitrate and Phosphate respectively.The two above listed nutrients and others were essential to show the land quality of the area.There is no significant difference in nutrient loss among the three land uses in the study site (Table 4).One reason for this might be the sample collection time effect, which was collected at the end of the rainy season, when the soil loss amount was very low due to crop cover increment.
The nutrient analysis from the runoff has indicated that essential nutrients are washed away from the land through runoff.The amount of dissolved ammonia and nitrate lost in runoff was higher in cultivated land than the other two LULC types (Table 5).Total dissolved solids and electric conductivity were higher in grassland than the other two land use.The amount of phosphate lost from Eucalyptus field was lower than cultivated land and grassland.
There is significant correlation between land use and runoff at p < 0.01.There is also significant correlation at p < 0.05 between LULC types and soil loss (Table 6).There is positive correlation between soil loss and runoff in the study area (Figure 2).
Similar to this study Adimassu et al. (2014) has observed that the soil loss from cultivated land reduces  overtime due to the effect of direct rainfall on bare land reduced as the crops grown and cover the land.In different part of Ethiopia soil loss studies have found highly variable annual soil loss from 3.4 ton/ha per year (Walle et al., 2006) to 56.7 ton/ha per year (Gebreegziabher et al., 2008) in non-conserved cultivated land.Others found within the range given above from cultivated land (Haile et al., 2006;Nyssen et al., 2008;Girmay et al., 2009;Adimassu et al., 2014).Further exceptional soil loss was also registered up to 212 ton/ha per year in Andit Tid, Ethiopia (Haile et al., 2006).The soil loss from cultivated land has reached up to 300 ton per hectare per year in Ethiopia (Hurni, 1993;Herweg and Stillhardt;1999;SCRP, 2000).The extent of previous soil erosion, rainfall characteristic, plot size and other variables could be mentioned as causes for variation soil loss in the country.
The annual sediment yield in cultivated land is higher than grassland, plantation, and conservation area in Tigray, North Ethiopia (Girmay et al., 2009).Oppositely, Nyssen et al. (2008) has found less mean soil loss from arable land (9.9 ton/ha/year) than rangeland (17.4 ton/ha/year) in Tigray, Ethiopia.Adimassu et al. (2014) has found greater soil loss from un-conserved cultivated land comparing with cultivated land conserved by soil bund.According to Hurni (1993) the average amount of soil loss from cultivated land was 40 ton per hectare per year in Ethiopia, which is higher than this study, Girmay et al. (2009) and Adimassu et al. (2014).Defersha and Mesele (2012) have also found higher soil loss from cultivated land than other LULC types.The soil loss study, from sub catchments in the same experiment sites with this study, has found lower sediment loss (2.4 ton/ha) and this is due to some sediment might be deposited on its way before reaching the outlet.It could be the reason of reduction of the value of soil loss (Erkossa et al., 2015).Similar to this study, Girmay et al. (2009) has found no significant difference in soil loss among grassland, exclosure and plantation sites.Grassland has lower soil loss than cultivated land and bare land (Defersha and Mesele, 2012).Similar to the study by Girmay et al. (2009), this study confirms that soil loss from the cultivated land and other degraded land can be significantly reduced when lands are converted into plantation and especially exclosure.
There was no significant difference on nutrient loss from runoff associated P and K observed on cultivated land under different land management practices (Adimassu et al., 2014).The nutrient loss, specially the N and P losses, from the catchment were strongly related to the soil loss.This emphases that where there is high soil loss, there is also high nutrient loss; for instance N (9.7 kg/ha) and P (4.7kg/ha) losses were observed from catchments (Erkossa et al., 2015).Similar to the above, this study has found that higher nutrient loss where there is high soil loss in cultivated land.On contrary to the above, Girmay et al. (2009) has stated that nutrient loss was not only dependent on the sediment loss, but the soil condition of the land where nutrients losses can also influence.Soil and nutrient losses significantly reduced by application of land management practices such as soil bund on cultivated land (Adimassu et al., 2014).According to Haileslassie et al. (2006) nutrient loss rates vary between the land use types and the land management practices.

CONCLUSION AND RECOMMENDATIONS
This study has investigated that the effects of LULC change specially Eucalyptus expansion on soil and nutrient losses from runoff (N and P).The study has found that soil loss from each LULC types significantly varied.Cultivated land has contributed the major soil loss in the catchments.The soil loss from cultivated land is higher than the tolerable average soil loss limit given in the country.There was no significant effect on soil loss by changing from grassland to Eucalyptus plantation or vise verse.In other words, planting Eucalyptus on degraded land could reduce soil loss from the land.There was no significant difference in nutrient loss from runoff at each LULC types.However, the mean amount of soil nutrients loss from cultivated land is higher than other LULC types.Even, greater than the value given by Food and Agriculture Organization (FAO) for nutrient loss from cultivated land.The study has found that the negative correlation between LULC change and soil and nutrient losses in the study area.Similarly, it has found that strong correlation between the volume of soil and nutrient losses from the plots.In summary, cultivated land in the site needs land management practices such as physical and biological soil conservation measures to reduce the soil and nutrient losses beyond the tolerable limit.This study, along with other studies, which have been done in different part of the country on the effect of plantation on erosion, has proved that reduction of soil loss in plantation sites.Eucalyptus plantation or woodlot can reduce the soil and nutrient losses through soil erosion.This implies that apart from the socioeconomic benefits, Eucalyptus can also be used as biological soil erosion conservation measures on selected sites.
Further studies of Eucalyptus need to be done in diversified agro ecological zones by considering its effect on soil and nutrient losses from runoff with spacing, long term erosion controlling effect, species and topographic effect.To this extent, this study recommends the policy makers and the experts to consider Eucalyptus as one of erosion conservation option where there is land degradation reached the climax and where there is limitation of conservation trees in the area.

Figure 1 .
Figure 1.Map of the study area.

Figure 2 .
Figure 2. Runoff and soil loss relationship for the study period.

Table 1 .
Biophysical conditions of the runoff plots.

Table 2 .
Total runoff collected from each LULC types w ithin study period.

Table 3 .
Means of soil loss w ithin study period at each LULC types.

Table 4 .
Means of some nutrient loss from runoff at each LULC types.
ns nsno significant difference at p < 0.05 least significant difference.

Table 5 .
Means value of runoff, ammonia, nitrate and phosphate per hectare.

Table 6 .
Correlation betw een LULC types, runoff and soil loss.
** Correlation is significant at 0.01; *Correlation is significant at 0.05.