Assessing the Input and Output Flows and Nutrients Balance Analysis at Catchment Level in Northern Ethiopia

Soil fertility depletion in smallholder farms is one of the fundamental biophysical causes for declining use per capita food production in Ethiopia. In the present study, resource flow analysis was made at catchment scale in northern Ethiopia, analyzing nutrient balances. JMP 5, a computer program for statistics and used in applications such as design experiment and scientific research was used to process and to analyze the data of different resources flow into and out of the watershed. Nutrient balances for N, P and K from four nutrient fluxes entering and four nutrient fluxes leaving the watershed were calculated. Some of the fluxes (e.g. wet deposition and gaseous losses) were estimated using transfer functions. At the catchment level, full nutrient balance results indicate a depletion rate of -42.5 to 13.1 kg N ha -1 year -1 (for the rich) and -32.7 kg N ha -1 year -1 , -11.8 kg K ha -1 year -1 (for the medium) in the upper landscape, -49 kg N ha -1 year -1 , -14.3 kg K ha -1 year -1 ( for the rich) and -28.5 kg N ha -1 year -1 , -11.8 kg K ha -1 year -1 (for the medium) in the middle landscape, -57 kg N ha -1 year -1 , -16.1 kg K ha -1 year -1 (for the rich) and -33.5 kg N ha -1 year -1 , -12.1 kg K ha -1 year -1 (for the medium) in the lower landscapes. Less negative value of nutrient balances of the poor socio-economic groups across the three landscapes shows N depletion in the poor socio-economic groups with -17.7kg N ha -1 year -1 and -5.59 kg K ha -1 year -

Soil fertility depletion in smallholder farms is one of the fundamental biophysical causes for declining use per capita food production in Ethiopia.In the present study, resource flow analysis was made at catchment scale in northern Ethiopia, analyzing nutrient balances.JMP 5, a computer program for statistics and used in applications such as design experiment and scientific research was used to process and to analyze the data of different resources flow into and out of the watershed.Nutrient balances for N, P and K from four nutrient fluxes entering and four nutrient fluxes leaving the watershed were calculated.Some of the fluxes (e.g.wet deposition and gaseous losses) were estimated using transfer functions.At the catchment level, full nutrient balance results indicate a depletion rate of -42.5 to 13.1 kg N ha -1 year -1 (for the rich) and -32.7 kg N ha -1 year -1 , -11.8 kg K ha -1 year -1 (for the medium) in the upper landscape, -49 kg N ha -1 year -1 , -14.3 kg K ha -1 year -1 ( for the rich) and -28.5 kg N ha -1 year -1 , -11.8 kg K ha -1 year -1 (for the medium) in the middle landscape, -57 kg N ha -1 year -1 , -16.1 kg K ha -1 year -1 (for the rich) and -33.5 kg N ha -1 year -1 , -12.1 kg K ha -1 year -1 (for the medium) in the lower landscapes.Less negative value of nutrient balances of the poor socio-economic groups across the three landscapes shows N depletion in the poor socio-economic groups with -17.7kgN ha -1 year -1 and -5.59 kg K ha -1 year -1 , -21 kg N ha -1 year -1 and -12.8 kg K ha -1 year -1 and -19.7 kg N ha -1 year

INTRODUCTION
Agriculture is the main economic activity of Ethiopia.It is dominated by smallholder farming (MoFED, 2002;CSA, 2009).The contribution of smallholder agriculture to the country is very high.It accounts for about 45% of the GDP, 85% of the exports and 80% of the total employment (EPA, 1997).However, agricultural productivity is *Corresponding author.E-mail: k.gebremedhin@yahoo.com.Tel: +251(0) 344444899.Fax: +251(0) 344444898.
continuously challenged by land degradation, which is manifested in various ways.For example, footpaths develop into gullies, soils become thin and stony, topsoil is gone due to accelerated soil erosion etc (Stocking and Murnaghan, 2001).Over 50% of the highlands in general and cropped areas of Ethiopia are in an advanced stage of land degradation (Elias, 2002).This is because of the continuous cultivation at least since the 13th century (Tewolde Berhan, 2006).Therefore, soil organic matter (SOM) content and nutrients are generally lower, where land degradation is more severe (Elias, 2002) it also leads to poor soil structure consequently to water erosion.Therefore, soils in many areas of the country especially in Tigray become shallow and stony (Stocking and Murnaghan, 2001).
The decline in soil physical, chemical and biological properties is revealed in many parts of the densely populated highlands of Ethiopia.For example, in Tigray nitrogen and phosphorus are highly deficient (Mitiku et al., 2003).Nitrogen in the cultivated surface soils was between 0.07-0.13% in Melbe area of Tigray (Tegene, 1996).Moreover, the soil depth in many areas of Ethiopia is less than 20 to 30 cm, this means that it is reaching the lower limits of productivity of the arable land and has lost much of its capacity to retain moisture; with consequent decline in agricultural yield (Stocking and Murnaghan, 2001;Elias, 2002;World Bank, 2007).The annual grain production loss estimate due to burning of dung as fuel than using for soil fertility improvement is estimated at 50,000 tons per annum, while the loss due to accelerated erosion is around 40,000 t by 1990.This will be accelerated into 170,000 t in 2010, if not controlled (EPA, 1997).
Soil nutrient depletion is rarely directly linked to food shortage, as it is a gradual process unlike natural disasters (e.g.drought, flooding and, etc.).Nevertheless, several studies have revealed that lack of plant nutrients is one of the principal causes for low agricultural productivity and food insecurity in Africa (Smaling, 1993;Sanchez, 2002;Haileselassie, 2005).Agriculture in Ethiopia is no exception: More soil nutrients are exported compared to natural and anthropogenic inputs (Elias et al., 1998;Okumu, 2000).Stoorvogel and Smaling (1990) predicted for Ethiopia that the national nutrient balances would be on average: -47 kg N ha -1 , -15 kg P 2 O 5 ha -1 and -38 kg K 2 O ha -1 for the year 2000.This prediction was twice as high as the average value for sub Saharan Africa and indicates the severity of nutrient depletion in Ethiopia.
Previous efforts to quantify nutrient balances were at sub-continental and national scales without addressing the large spatial variability of nutrient balances within a region and catchment level.Hence, there is a need for assessment of resource flow analysis at catchment level (scale) in different landscapes and under different socioeconomic conditions, as a basis for the design of technically feasible, ecologically non-degrading, and economically viable nutrient management strategies.The study described in this paper responds to that need of reporting on assessment of macro-nutrient balances at catchment scale in May-leba, Northern Ethiopia.
The purpose of this study is to investigate the potential of resources flow analysis to tackle the increasingly severe soil fertility deterioration in May-leba catchment, Northern Ethiopia.Based on primary, available secondary data and orders-of-magnitude estimations, the resources (nutrient) flows through the catchment system are analyzed, traced and quantified, and the key material sources identified.Scenario calculations allow then to determine most effective mitigation measures which are either favorable for nutrient conservation or which are responsible for high nutrient losses under current management practices.This kind of information is critical for decision makers to plan and implement integrated nutrient management strategies at a regional and catchment level.

Location
The May-Leba catchment lies (longitude: 13°14'35" and latitude: 39°15'53") is located at 35 km West of Mekelle, along the road from Mekelle to Abi-Adi, about 10 km east of Hagere-Selam town (Figure 1) .It is part of the woreda of Dogua Tembien.The lower edge of the study area is the May-Leba dam at 2290 m.a.s.l.(Van de Wauw et al., 2008).

Soils and climate
The major soils of May-leba catchment are Leptosol, Cambisol, Vertisol, Regosol and Fluvisol (Gebresamuel et al., 2010).They contain 1.1-1.6 percent organic carbon, 0.09-0.16percent total nitrogen and 39.1-57.9cmol (+) kg -1 soil Cation Exchange Capacity ( Van de Wauw et al., 2008).The climate is characterized by an ustic (intermediate between the aridic and the udic) moisture regime implying dryness with a distinct rainy season in July and August.The average rainfall is 716mm.The site is located in the semi-arid agro-ecological zone.Rainfall is distributed between the short rainfall season (March to April) and the main rainy season (June to September).The average monthly temperature varies between 12 and 19 o C (Van de Wauw et al., 2008).

Land use
The dominant land use in the May-Leba catchment is cropland (about 50%).The rest is used for grazing (30%), housing (15%) and a small part (5%) is closed for grazing and cropping ( Van de Wauw et al., 2008).The major crops grown in the study area are barley (Hordeum vulgare), wheat (Triticum sp.), teff (Eragrostis tef) and grass pea (Lathyrus sativus) where the total area of the study area is 1733.6haand 50% of this is cropland therefore, the relative share of the total cultivated crop land among these crops is 797.456ha(Gebresamuel et al., 2010).

Quantifying of resources flow
The N, P and K balances were calculated from a combination of four input and four output flows using the nutrient/material balance model developed by Stoorvogel and Smaling (1990).The input flows monitored were mineral fertilizer, manure, compost, ash and atmospheric deposition.The major output flows quantified were harvested products (like straw, crop grain), crop residues, hay and gaseous losses.Quantification of the N, P and K in the input and output flows was achieved through the combination of different methods: field measurement, use of empirical quantitative relations (i.e.transfer functions) and assumptions based on secondary data from a variety of sources.The type of data required and the method of quantification for each of the input and output flows are summarized in Tables 1 and 2.

Quantifying Inputs
For N, P, K and OC primary data were obtained on type and amount of mineral fertilizer, organic inputs (manure and compost), ash whereas atmospheric deposition was found by transfer function.Field observation was made on type and amount of resources input and output, type of cereal crop grown and their grain and residue yield were recorded.Routine field measurement of the transfer of resources in and out of each land use type was monitored by maintaining a book keeping of material flows.The quantity of nutrient inputs via wet deposition (IN4) depends on the amount of precipitation and proximity to a city (Mengel and Kirkby, 1996).The method of Stoorvogel and Smaling (1990) and Smaling (1993) were used to estimate wet deposition as the square root of average annual rainfall (in mm) using coefficients of 0.014 for N, 0.053 for P and 0.11 for K (Stoorvogel and Smaling, 1990;Smaling, 1993).
Where: IN4 is expressed in kilograms per hectare per year; and rainfall in millimeters per year.

Quantifying outputs
Removal of harvested products (OUT1), crop residues (OUT2) and hay (OUT3) are usually the major pathways of nutrient losses from agricultural soils.The amount varies considerably depending on crop type, soil type, agronomic practices and plant nutrient uptake (Brady and Weil, 2002).Reliable information on crop nutrient composition is, therefore, essential for quantifying this important nutrient export.Biomasses, grain, residue yields of crops were determined by harvesting from a 12 m 2 areas at ground level from randomly selected fields.The harvested area was randomly demarcated in each field and the plot was harvested manually at crop maturity on dates preferred by farmers.Samples were air dried Table 1.Type of data required and method of quantification of the four input processes employed in the calculation of N, P and K balances.
Table 2. Type of data required and method of quantification of the four output processes employed in the calculation of N, P and K balances.The quantity of crop residues (OUT2) and hay (OUT3) were calculated from residues/products ratio given in FAO (1986).It is assumed that in most parts of Ethiopia 85% of total residues are removed from arable lands being used either as energy sources or for animal feed.

Output process Code and nutrients
OUT2N = R * N * F = (kg N/ha ), OUT2P= R * P * F = (kg N/ha) and OUT2K= R * K * F = (kg N/ha) R = amount of residues (kilograms/ha); N = N content of crops residues (kg N/ kilograms harvested product); P = P content of crops residues (kg P/ kilograms harvested product); K = K content of crops residues (kg K/ kilograms harvested product), and F = removal factor of crop residues Gaseous losses (OUT4) can be an important pathway of N fluxes in many agricultural production systems.Stoorvogel and Smaling (1990) linked a limited number of reliable data on denitrification of sub Saharan African soils using multiple regression analysis.Some other studies estimate gaseous losses as function of soil N, fertilizer N, precipitation and soil clay contents (Van den Bosch et al., 1998;Smaling, 1993).More recent studies (FAO, 2003) includes volatilization losses in OUT4 estimation as given by the equation below in which OC is organic C content (in percent), R is mean annual precipitation (mm), Nf is mineral N fertilizer and organic fertilizer N (kg N ha -1 ).
OUT4 = (0.025 + 0. 000855R + 0.01725Nf + 0.117OC) + 0.113Nf Nutrient export in the harvested products and residues was derived from the yield data for each crop.Sub-samples of harvested products and residues of wheat, barley, teff and grass pea were taken for determination of nutrient content at the Tigray agricultural Research Institute (TARI) laboratory.

Chemical analysis of the resources flow
The resources flow samples were processed for laboratory test including proper registration, air-drying, grinding, sieving through 2 mm sieve and storage (Sahlemedhin and Taye, 2000).The analysis was conducted on ground and sieved (< 2 mm) samples.The material samples were used for chemical property analysis.The following resources flow parameters have been analyzed: organic carbon, total nitrogen, available phosphorus, and available potassium.
The percentage of organic carbon was determined by a modified Walkley-Black procedure (Smith and Weldon, 1940).The organic matter was obtained by wet oxidation technique, that is, calculation of 1.724 × percentage of carbon (Black, 1965).It is based on the assumption that organic matter contains 58% organic carbon.
The wet digestion of the Kjeldahl procedure (Bremmer and Mulvaney, 1982) was used to test for total nitrogen while Olsen's method was employed to determine available phosphorus (Olsen and Sommers, 1982;Anderson and Ingram, 1993).Available K was extracted by ammonium acetate extraction method (Sahlemedhin and Taye, 2000).

Chemical analysis of plant biomass
The N, P and K contents of the plants of different crop resources were analyzed in the soil laboratory of the Tigray Agriculture Research Institute (TARI).The preparation of the plant material was through drying the green material at a maximum of 60 to 70°C, grinded to pass through a 0.15 mm mesh and 10 g is taken for analysis (Anderson and Ingram, 1993).The concentration of the total nitrogen in plant was determined by the Kjeldahl method.The organic nitrogen is oxidized into ammonium by acid hydrolysis with H2SO4 together with the reagent potassium sulfate to raise temperature and to hasten the rate of decomposition, copper sulfate and selenium powder were used as catalyst.The plant analysis procedure for Phosphorous and Potassium concentration was done following ashing method.About 1.0 g of ground plant sample was dried out in aluminum dish over night at 105°C in an oven.The ash was dissolved in concentrated HCl and diluted with de-ionized water.After addition of color reagent (molybdatevandate-solution), the phosphorous concentration was measured by spectrophotometer and K by flame photometry.

Data analysis
The types of analysis that were carried out were the analysis of variance (ANOVA) to evaluate how resources flow (treatments) varies with landscapes.Regression and correlation analyses among the wealth ranking, landscape and resources flow was conducted.

Major source of nutrients flow and their contribution to soil fertility
The major sources of N, P, K and OC for the homestead fields was animal manure along with homestead refuse, leftover feed of animals from around the houses.Household refuse and animal manure from around the hut were used on homestead fields to improve the physical properties and nutrient retention of the soils.In contrast, mineral fertilizer (DAP and Urea) were the main sources of nutrient addition to the distance fields.

Nutrient inputs to the catchment
Farmers use specific soil fertility management strategies for different parts of their farms.They grow mainly permanent and vegetable crops near the homestead and others (cereals, pulses and oil crops) on more distant fields.Applications of manure and inorganic fertilizers reflect this differentiation (Tilahun et al., 2001).Currently, DAP and Urea are the only inorganic fertilizers applied by smallholders.Potassium application from inorganic fertilizer (IN1) is not reported in Ethiopia (NFIA, 2001).Fertilizer trials conducted on major cereal crops (for N, P and K) also indicated that cereal crops were not responsive to K.But the importance of long term K fertilization must not be overlooked, as leaching, erosion and depletion via crop harvest may deplete K stocks (Haileselassie, 2005).For all the Socio-economic groups, inflows into the catchment are in the form of chemical fertilizer, organic fertilizer like compost and manure and ash.Inflows of the inorganic fertilizers urea and diammonium phosphate (DAP) were small i.e. differences in the magnitude of IN1 across the socio-economic groups may have similar underlying causes these are like less awareness about using inputs by the farmers, approach of the agricultural extension situations (Table 3).
Nitrogen and phosphorus inputs from inorganic fertilizer were (3.29 kg N ha -1 year -1 , 0.56 kg P ha -1 year -1 ) and (23.01 kg N ha -1 year -1 , 4.18 kg P ha -1 year -1 ), (10.5 kg N ha -1 year -1 , 1.93 kg P ha -1 year -1 ) and (28.39 kg N ha , 0.28 kg P ha -1 year -1 ) higher in the resource endowed households than the medium and poor households in the upper, middle and lower landscape respectively.Hence, the chemical fertilizer inflows of N and P were higher for the rich farmers than for the medium and the poor in all the landscapes.The quantity of fertilizer used by different groups of farmers and the area of application are presented in Table 3.The data revealed a highly significant difference (p<0.01) for N among households in the quantity of fertilizer used across all socio-economic groups in all the landscapes.Richer farmers purchase significantly larger amounts of Mean values along the columns with different letters indicate significant difference at P<0.05 level of confidence.
fertilizer compared to the poor farmers in all the landscapes.This is because there is no relatively financial constraint in the richer farmers in all the landscapes (p<0.01) and also there are a sizeable number of livestock that could secure the rich farmers than the poor from the debts of the chemical fertilizer.
There is no significant difference across the landscapes in the quantity of fertilizers used.This finding in agreement with the findings by Elias (2002) who observed no statistically significant difference among socio-economic groups in the rate of fertilizer application The national scale nutrient inputs from urea and DAP were estimated at 12.24 kg N ha -1 year -1 and 12.87 kg P ha -1 year -1 (Haileselassie et al., 2005).Demeke et al. (1998) reported a similar trend of inorganic fertilizer applications across all regional states of Ethiopia.In addition to soil fertility, intensity of crop cultivation, cropping pattern, livestock management system (open grazing, confined management) and differences in other competitive uses of manure (e.g.household energy) were the major causes of variation in the application of IN2 across the different socio-economic groups (Table 4).
The rich endowed households had remarkably higher amount of IN2 compared to the other socio-economic groups in all the landscapes.This can be accounted for homestead farms that require regular application of manure.The rich famers also have better livestock management practices that contributed access to dung and increased IN2 in this socio-economic group.Compared to the medium and poor farmers the rich farmers apply a high amount of nutrient inputs from organic fertilizer (IN 2 ).Nitrogen, Phosphorus and Potassium inputs from organic fertilizer were higher in the rich endowed households that is, (4.8 kg N ha -1 year -1 , 0.1 kg P ha -1 year -1 and 1 kg K ha -1 year -1 ) higher than the medium, (6.7 kg N ha -1 year -1 , 0.12 kg P ha -1 year -1 and 2.73 kg K ha -1 year -1 ) and the poor in the upper landscape and (7.1 kg N ha -1 year -1 , 0.1 kg P ha -1 year -1 and 2.31 kg K ha -1 year -1 ) higher than the medium, (10.7 kg N ha -1 year -1 , 0.2 kg P ha -1 year -1 and 7.21 kg K ha -1 year -1 ) and the poor in the middle landscape and (3.7 kg N ha -1 year -1 , 0.1 kg P ha -1 year -1 and 1.85 kg K ha -1 year -1 ) higher than the medium, (4.4 kg N ha -1 year -1 , 0.12 kg P ha -1 year -1 and 4.2 kg K ha -1 year -1 ) and the poor in the lower Mean values along the columns with the same letter are not significantly different at P<0.05 confidence interval.
landscape.Hence, the organic inflows of N, P and K were higher for the rich farmers than for medium and poor in all of landscapes.Thus, the data revealed a significant difference (p<0.01 for N, p<0.05 for P and p<0.01 for K) in the amount of organic inputs (manure and compost) moved (used) to fields for soil fertility maintenance among households of different socio-economic groups (Table 4).Richer farmers used significantly (p<0.01)larger volumes of manure compared to poor farmers in the three landscapes.This is because there are a sizeable number of livestock in the rich endowed households than the poor farmers.
As would be expected, the statistical correlation test showed a strong and positive correlation (r=0.992,p<0.01) between the number of livestock held per household and the amount of manure for soil fertility.This is in agreement with the findings of Elias (2002).The variation in the amount of manure used among households is largely explained by differences in the herd size.Nonetheless, in societies (Burkina Faso and Northern Benin) where animal dung is used for household energy supply (fuel), a significantly negative relationship would occur (Millennium Ecosystem Assessment, 2005).
No significant variation was observed in the amount of chemical fertilizer, compost, manure and ash transferred to the fields for soil fertility maintenance across landscape positions (Table 5).The total national estimate of IN2 application rates for all cropping systems was 29 kg N ha -1 year -1 , 7.2 kg P ha -1 year -1 and 34.3 kg K ha -1 year -1 (Haileselassie et al., 2005).Ash (IN3) and atmospheric deposition (IN4) were the other nutrient inputs considered in this study.Globally, N deposition rates are estimated to be 5 kg N ha -1 year -1 for less densely populated and non-industrial countries and range from 20 kg N ha -1 year -1 to 50 kg N ha -1 year -1 in countries of Western Europe and parts of China (Sheldrick et al., 2003).Mengel and Kirkby (1996) also discussed that atmospheric deposition for N can be as high as 60 kg ha -1 year -1 depending on proximity to a city and amount of precipitation.Atmospheric deposition estimates showed no clear difference between socio-economic groups.

Nutrient outputs from the catchment
Removals in crops (OUT1) and (OUT2) were the major causes of export of nutrients from the soil followed by hay (OUT3) and gaseous (OUT4) losses in most of the socioeconomic groups (Table 6).
Differences in the landscapes and corresponding primary productivities were the main cause of variation (OUT1) among the socio-economic groups (Table 6).Notably, socio-economic groups with a high IN1 level had also a high OUT1 (e.g. the rich in the three landscapes).This imply for nutrient management is assessing the effects of future land use change on nutrients; and soil management practice influence the soil nutrients related soil processes, such as erosion, oxidation, mineralization, and leaching, etc. and consequently modify the processes of transport and re-distribution of nutrients.At the catchment level, the losses of N, P and K via harvested crops (OUT1) varied from (28.69 to 82.64 kg ha -1 year -1 , 0.12 to 0.32 kg ha -1 year -1 and 9.37 to 18.02 kg ha -1 year -1 ) across all the landscapes.As was the case for products harvested, there were clear differences in OUT2 across socio-economic groups (Table 6).This was caused by differences in yields and cropping pattern this indicates that the yearly sequence and spatial arrangement of crops and fallow on a given area where selection of crops and their varieties is to be made depending on the soil and rain fail situation in the rained areas are Teff-Barley/Wheat-Teff; Barley-Teff-Wheat-Grass pea/Bean; Barley/Wheat -Bean-Wheat/Barley.The losses of N, P and K via crop residues (OUT2) varied from (5.1 to 13.3 kg ha -1 year -1 , 0.611 to 1.21 kg ha -1 year -1 and 6.29 to 13.55 kg ha -1 year -1 ) along with the landscapes.High crop residue losses were observed in the rich farmers compared to the medium and the poor, because the rich farmers have a sizeable number of livestock so that they do not leave the crop residues in their farms for soil fertility maintenance.This finding is in agreement to the study at a national scale where export through residues (OUT2) was estimated at 9.8 kg N ha -1 year -1 , 1.5 kg P ha -1 year -1 , but contrasts for K which is 18.5 kg ha -1 year -1 (Haileselassie et al., 2005) The results suggest that denitrification (OUT4) may be the least important cause of N loss.Indeed, averages are of the magnitude of 4.73 kg ha -1 year -1 , 4.17 kg ha -1 year -1 and 4 kg ha -1 year -1 for the rich, medium and the poor in the study area, values which are close to the result of Van den Bosch et al. (1998).Denitrification losses predicted by Stoorvogel and Smaling (1990) were higher than the present study's estimate (7 kg ha -1 year -1 ). Rich farmers of the study areas had medium level of attention.Of course there are very limited numbers of farmers were found to have a high attention of integrated soil fertility and nutrient management efficiency for sustainable crop production.

Partial nutrient balance
The partial nutrient balance of the study considered only the most important inputs and outputs.The inputs include mineral fertilizer (IN1), organic inputs (IN2) while the output harvest products or grain (OUT1) and residues removed (OUT2).The partial nutrient balances are calculated as the difference between sum of inputs (IN1 and IN2) and sum of outputs (OUT1 and OUT2).Based on the partial input-output nutrient balance the study area shows negative balance for Nitrogen and Potassium regardless of the wealth status and landscape.Marked variations were observed among the socioeconomic groups (Figure 2).The rich with a higher rate of fertilizer or dung input (in the three landscapes) have a higher negative balance even though harvests were also higher.It is also observed apparently that strong positive balances of P in all the socio-economic groups exist in all the three landscape positions.Higher inorganic and organic nutrient inputs and relatively strong negative partial nutrient balances in the rich households might be due to intensively cultivated farms, however nutrientsaving techniques, such as soil and water conservation practices, were practiced more frequently by the rich group.In all the socio-economic groups and the landscapes, the level of inputs (IN1 and IN2) was clearly lower than the outputs (OUT1 and OUT2), resulting in a negative partial balance for all nutrients except phosphorus (P).The losses of N and K (-29.91 kg N ha -1 year -1 , -14.420 kg K ha -1 year -1 ), (-23.20 kg N ha -1 year -1 , -11.65 kg K ha -1 year -1 ) and (-15.77kg N ha ) in the upper, (-35.48 kg N ha -1 year -1 , -13.67 kg K ha -1 year -1 ), (-18.99 kg N ha -1 year -1 , -11.34 kg K ha -1 year -1 ) and (-12.65 kg N ha -1 year -1 , -12.63 kg K ha -1 year -1 ) in the middle and (-34.61kg N ha -1 year -1 , -14.35 kg K ha -1 year -1 ), (-21.90 kg N ha -1 year -1 , -12.11 kg K ha -1 year -1 ) and (-24.49kg N ha -1 year -1 , -9.90 kg K ha -1 year -1 ) in the lower landscapes of the rich, medium and poor socio-economic groups respectively.Higher negative balance was greater in lower, middle and upper landscapes respectively.The average nutrient depletion from the catchment of the study area were -33.32 kg N ha -1 year -1 , -14.15 kg K ha -1 year -1 for the rich, -21.36 kg N ha -1 year -1 , -11.7 kg K ha -1 year -1 for the medium and -17.64 kg N ha -1 year -1 , -10.31 kg K ha -1 year -1 for the poor; this relation shows nutrient depletion on the partial nutrient balance for different agro-ecological conditions of the central highlands of Ethiopia where similar to the study area was also reported by Dechert et al. (2005) as -33 kg N ha -1 year -1 , -39 kg K ha -1 year -1 for the rich; -29 kg N ha -1 year -1 , -39 kg K ha -1 year -1 for the medium and -16 kg N ha -1 year -1 , -35 kg K ha -1 year -1 for the poor socio economic groups in the entire catchment.
At the national scale, the partial balances were positive for all nutrients considered.These values were comparable with the mean partial balance of 26 farms in the Kisii, Kakamega and Embu districts in Kenya where the agroecological zones are similar (De Jager et al., 1998) In many studies partial nutrient balance at land use level are negative for the Teff based farming system in the central high lands of Ethiopia -28 kg N ha -1 year -1 , -87 kg K ha -1 year -1 for Barley; -21 kg N ha -1 year -1 , -23 kg K Kg/ha/year for Faba bean and -11 kg N ha -1 year -1 , -51 kg K ha -1 year -1 for Maize (Dechert et al., 2005;Van Dung et al., 2008) However, the study by Haileselassie (2005) reported that the partial nutrient balance is showing positive for the Tigray Region (+10 N, +6 P, +10 K kg ha -1 year -1 ) and (+10 N, +11 P, +7 K kg ha -1 year -1 ) for Ethiopia at national level.So far there is no detailed study conducted in the study area except the plot level study in the region by Hengsdijk et al. (2005) who reported that a negative balance -27 N ha -1 .year -1 . This report shows a lower estimate as compared to the national level.The lower estimates reported in this study and by others might be due to the severe level of degradation of the region (Hagos et al., 2002;Mitiku et al., 2003;Tewolde Berhan, 2006), as compared to the other well-endowed areas of Ethiopia (Elias et al., 1998), undermining the soils' poor in Nitrogen.
On the other hand the study by Abegaz (2005) in Teghane Atsbi, Tigray Region, reported nutrient depletion between -56.5 to (-115) kg N ha -1 year -1 , 0 to (-5.8) kg P ha -1 year -1 and -34.6 to (-112) kg K ha -1 year -1 . The high nutrient depletion in the country in general and Tigray Region in particular are because of limited applications of organic fertilizers like crop residues and manure, and the socio-economic problems in the mineral fertilizer (Abegaz, 2005).
The results of the partial nutrient balance showed that the nutrient removals by the crop harvest and crop residues were the greatest contributors to the N and K negative balances except for the phosphorous in the three landscapes and in all the socio-economic groups, particularly in the case of rich farmers in the middle landscape.These differences are associated with the differences in crop production, that is, higher crop production by rich farm group.

Full nutrient balance
The full nutrient balance results also demonstrate a large variability in the socio-economic groups (Figure 3).Variations in the high rates of nutrient mining across the socio-economic groups were mainly explained by land holding, family size, herds (with cattle dung mainly used for fuel) and use of external inputs.The balances of N and K for the rich, medium and poor farmers in the lower, middle and upper landscapes were all negative.Whereas the balances of P showed positive values for all the socio-economic groups and in all landscape positions.
The full nutrient balances results were similar across the three landscapes.Higher negative nutrient balances were found highest in the rich farmers followed by the medium and the poor as the result of chemical and physical soil fertility of the cultivated fields of the rich is generally higher and they implement more frequently nutrient-saving techniques, such as soil and water conservation practices and apply more external and internal inputs.The N and K depletion in rich and medium socio-economic groups with high intensive management of their farms -42.5 kg N ha -1 year -1 , -13.1 kg K ha -1 year -1 (for the rich) and -32.7 kg N ha -1 year -1 , -11.8 kg K ha -1 year -1 (for the medium) in the upper landscape, -49 kg N ha -1 year -1 , -14.3 kg K ha -1 year -1 ( for the rich) and -28.5 Kg/ha /year , -11.8 kg K ha -1 year -1 (for the medium) in the middle landscape, -57 kg N ha -1 year -1 , -16.1 kg K ha -1 year -1 (for the rich) and -33.5 kg N ha -1 year -1 , -12.1 kg K ha -1 year -1 (for the medium) in the lower landscapes.Comparably nutrient imbalances were low in the case of the poor socio-economic groups across the three landscapes.N depletion in the poor socio-economic groups which apply less intensive farm management were (-17.7 kg N ha -1 year -1 , and -5.59 kg K ha -1 year -1 ), (-21 kg N ha -1 year -1 , and -12.8 kg K ha -1 year -1) and (-19.7 kg N ha -1 year -1 , and -7.52 kg K ha -1 year -1 ) in the upper, middle and lower landscapes respectively (Figure 3).This was largely due to, lower rainfall, lower biomass production and lower soil fertility.N and K balances for all the socio-economic groups across the three landscapes were negative (Figure 3).In all the socio-economic groups and the landscapes, positive full balances for P was obtained as 7.132 kg P ha -1 year -1 , 8 kg P ha -1 year -1 and 7.13 kg P ha -1 year -1 for the rich, 6.86 kg P ha -1 year -1 , 6.71 kg P ha -1 year -1 and 6.32 kg P ha -1 year -1 for the medium and 3.48 kg P ha -1 year -1 , 3.97 kg P ha -1 year -1 and 6.77 kg P ha -1 year -1 for the poor in the upper, middle and lower landscapes respectively.P depletion rates generally were not as high as N and K rates.This can be explained by the low quantity of P in plant and soil systems, and because its ability to fix in the soil.Similar trends of P depletion were reported by Sheldrick et al. (2003).This is mainly the impact of higher IN1.As was the case for N and K, farms of the poor households with less nutrient management showed a lower degree of P depletion, while the farms of rich with high farm management show comparatively high amounts of P depletion.
Many nutrient balance studies results in Ethiopia reported negative values.The present study reveal lower nutrient depletion than reported in previous studies where the full nutrient balance is negative for Tigray Region with the depletion rate of -41 N, -1 P, -36 K kg ha -1 year -1 ; the national level depletion rate for N, P and K was calculated at -122 kg ha -1 year -1 , -13 kg ha -1 year -1 and -82 kg ha -1 year -1 , respectively.It may be because other factors, like inputs through sedimentation and outputs like leaching, erosion are not calculated (Dechert et al., 2005;Haileselassie, 2005Haileselassie, , 2007)).Moreover farmers in the study areas are introducing different extension approaches like using the chemical fertilizers, organic inputs to replenish their farm lands so that the depletion rate of the macro nutrients are less than the national depletion rate.

CONCLUSION AND RECOMMENDATION
From this study we can conclude that there is higher nutrient mining rate, particularly for N in all the landscapes primarily due to large removals of nutrients in harvested output and residue as a result of poor nutrient management by these farmders, leading to low nutrient use efficiency, especially N. The use and management of specific source of soil nutrients like application of mineral fertilizer and organic inputs and nutrient depletion rates differ significantly among socio-economic groups.The highest rates were recorded for the rich farm group, followed by the medium and poor farm groups.These differences are associated with the differences in crop production as the result of chemical and physical soil fertility of the cultivated fields of the rich is generally higher and these rich farmers implement more frequently nutrient-saving techniques, such as soil and water conservation practices and apply more external and internal inputs.On the other hand, the low rates of depletion in the poor farm group are partly associated with low crop production, due to low indigenous soil fertility, low inputs and poor crop management.
Under the prevailing natural and socio-economic conditions, farmers' current strategies (practices) are inadequate to cope with the decline in soil fertility entirely at farm (catchment) level in all the landscapes.Fertilizer use is seen as the only way to maintain or improve soil fertility, and soil management is particularly lacking in the return of organic matter to the soil.However, despite all the extension efforts, fertilizer consumption in the study area as well as in the region level is extremely low.Therefore, there is a need for targeted agronomic interventions, improvements in nutrient use efficiency from different inputs, awareness creation through integrated nutrient management to mitigate nutrient losses and there by improve the sustainability of crop production at farm (catchment) level.
Finally this study recommends that farmers should have a tendency to give more weight for the application of organic and inorganic inputs, proper management of crop residues and sustaining soil conservation measures.In addition, further research about the quantification and analysis of resources flow through the catchment scale and their effect on soils fertility management, landscape positions and farm management practices is required.This can be done by conducting researches on improving the quality and quantity of organic inputs enhancing the macro nutrients at catchment level and further long-term and short-term research on socio-economic conditions and effects on soil fertility, especially on the relationship between social/wealth status and the use of organic and inorganic inputs.

Figure 1 .
Figure 1.Location map of the May-Leba catchment.
content in applied manure Field measurement Nutrient content in applied fertilizer Field measurement N-losses Transfer functions and weighed using a hand held scale or sensitive balance to give values for total biomass, grain, and residue yields and hay.OUT1N = Y * N = kg N/ha, OUT1P= Y * P = kg P/ha and OUT1K= Y * K = kg K/ha Y = yield (kilograms/ha), N = N content of crops (kg N/ kilograms harvested product); P = P content of crops (kg P/ kilograms harvested product), and K = K content of crops (kg K/ kilograms harvested product).

Figure 2 .
Figure 2. Partial nutrient balance (Kg/ha/year) for different resources inflow and outflow across the socio-economic groups in the landscape positions in the catchment, northern Ethiopia.

Figure 3 .
Figure 3. Full nutrient balance (Kg/ha/year) for different resources inflow and outflow across the socio-economic groups in the landscape positions in the catchment, northern Ethiopia.

Table 3 .
Resources inflow (kg/ha/yr) across the socio-economic groups in the three landscapes in May-leba catchment, Northern Ethiopia.

Table 4 .
The different resources inputs (kg/ha/year) into the catchment across the socio-economic groups (wealth status) in May-leba catchment, Northern Ethiopia. a

Table 5 .
The different resource inputs (kg/ha/year) into the catchment across the Landscapes in May-leba catchment, Northern Ethiopia. a

Table 6 .
Resources outflow (kg/ha/yr) across the socio-economic groups in the three landscapes in May-leba catchment, Northern Ethiopia.