Nutrient budget analysis under smallholder farming systems and implications on agricultural sustainability in degraded environments of semiarid central Tanzania

Soil nutrient budget is one of the critical concerns in most farming systems, especially in degraded environments. This paper reveals a net nutrient removal from the agricultural system that resulted in negative budgets, particularly for N. Large variations existed between villages, with N budget ranging from -9.6 to -252.4 kg/ha/year, with a mean budget of -91.9 kg/ha/year. Various factors seem to have contributed to the negative budgets, including soil erosion, crop harvest and residue removals. The latter two contributed significantly less to nutrient losses compared to soil erosion. The main input sources for nutrients into the system were found to be manure, crop residues, atmospheric deposition and to a limited extent inorganic fertilisers. Manure application added into the soil about 13, 1.5, and 0.04 kg/ha/year of N, P and K respectively. Its use was, however, limited by the inadequate amounts available. Crop residues added about 3.9, 1.9 and 11 kg/ha/year of N, P and K respectively. The use of crop residues for nutrient recycling and soil maintenance is however, constrained by its competitive use for animal feed. Generally, this farming system seems to be characterised by negative nutrient budgets, limiting agricultural productivity and sustainability.


INTRODUCTION
Farming in the tropical environments is largely on upland soils under rainfed agriculture.These soils easily degrade when subjected to over-exploitation, for example through cultivation (Kang, 1993).In the past the traditional bush and fallow system with short cropping cycles and long fallow periods allowed for regeneration of adequate soil productivity.The system was of high biological stability and sustained agricultural production for many generations.In recent years various socio-economic factors have caused the fallow periods to be shortened or abolished as observed in many parts of the tropics (Kang, 1993;Mohamed, 1989;Adesina, 1994;Scholes et al., 1994).The shortened fallow periods have resulted in increased soil degradation in farmlands, more weed infestation, declining crop yields, and reduced production of both food and cash crops.Increasing and sustaining food and cash crop production on these inherently erodible and low fertility tropical soils is an elusive goal.
One of the reasons for low agricultural production is the gradual process of soil nutrient depletion in many of the sub-Saharan land-use systems (Smaling, 1993).Among the processes responsible for this nutrient depletion are soil erosion, crop harvesting, residue removals and leaching, phenomena that have received less public attention compared to aspects like erratic rainfall, crop failures, and civil wars which are directly related to famine and starvation (Smaling, 1993).The supply of nutrients dictates the productivity of land and dwindling soil fertility has become one of the major constraints to crop production, especially in sub-Saharan Africa (Smaling, 1993;CWFS, 1985).In most land-use systems, the amount of nutrients leaving the soil in crops, by leaching and erosion, is grossly exceeding the nutrient input to the system by natural processes such as atmospheric deposition and biological nitrogen fixation, or artificially by organic manure and mineral fertilisers (Smaling, 1993;CWFS, 1985).This has resulted into reduced land productivity and agricultural sustainability (Lynam and Herdt, 1989;Spenser and Swift, 1992;Rockström, 1997).
This paper analyses nutrient budgets in the Irangi Hills (Figure 1) located in Kondoa District, which forms an area commonly known as the Kondoa Eroded Area (KEA).It presents findings from three villages namely, Baura, Bolisa and Mafai.The main objective of the study was to determine the nutrient budget for the crop/livestock and soil management systems in the area.Specifically, the objectives were to determine input sources and pathways for nutrient loss, and to compute the difference between nutrient input and output into and from the agricultural system.The KEA is a badly degraded area in the hills where comprehensive land-rehabilitation activities have been tried out since the 1930s, including the eviction of grazing animals implemented in 1979.The studied villages are characterised by gullies as deep as 20 m.
The climate of the study area is semiarid with an average annual precipitation of 600 mm.The dry season normally lasts from June through October and the rainy season from November to May.The rainfall distribution pattern is influenced by elevation (Ngana, 1992).Mafai village, which is much higher in altitude (ranging from 1620 to 2010 m), compared to the other two villages (ranging from 1200 to 1545 m) is sub-humid, while Baura and Bolisa are semiarid.The long-term rainfall pattern for the study area shows that rainfall is highly variable between years and is also very erratic.Temperature varies from 16 to 23°C, the coolest period being in June and July.
that are removed from the soil through crop harvest and residue removals, and that returned back into the soil by the residues retained in the farm.Farmyard manure was also analysed with the aim of establishing the amounts of same nutrients that it contributes to the soil when applied.Soil samples were collected from soil profiles dug to 1 to 1.5 m depth from representative locations for major soil groups identified by the farmers during transect walks.Soil profiles were described to at least 80 cm depth, according to the internationally accepted guidelines for soil-profile description (FAO, 1990).Visual observation showed no distinct horizons; as such samples were taken at specified depth ranges as described by (Hodgson, 1985) and (FAO, 1990), at 20 cm intervals, starting with the upper 0 to 20 cm.The soil samples were then analysed for particle-size distribution, pH, organic carbon, total nitrogen, extractable phosphorus, exchangeable bases (Ca 2+ , Mg 2+ , K + and Na + ), and cation exchange capacity (CEC), in order to obtain information important for estimating nutrient availability and soil productivity.This determination essentially aimed at estimating the inherent soil fertility status and, subsequently used the findings in computing the quantities of nutrients (N, P, and K) lost from the soil through soil erosion and other pathways.

Household interviews
Interviews were also conducted to 10% (147 households) of the village households so as to solicit information on crop yield and fertiliser application whose data were later used in computing the nutrient budget.

Inputs
Nutrient input to the agricultural system included inorganic fertilisers, farm yard manure, incorporation of crop residues and atmospheric deposition, as described in the following sections.

Inorganic fertilisers (Input 1)
Information on this input based on data obtained from household interviews conducted to the farmers on the type and quantities of fertilisers they used in the preceding growing seasons.The major inorganic fertilisers used by the respondent households were Calcium Ammonium Nitrate (CAN), Nitrogen-Phosphorus-Potassium (NPK), Sulphate of Ammonia (SA), UREA and Triple Supper Phosphate (TSP).However, the amounts of fertilisers used by respective households varied considerable.

Manure (Input 2)
The quantities of manure used by the farmers based on responses regarding the number of livestock units that are kept by the interviewed farmers.One livestock unit (250 kg live weight) requires a total of about 3000 kg of dry matter per year, of which about 40% is given out as manure (Mero et al., 1988).This gives an average of 1200 kg of manure being produced by one livestock unit in a year.

Crop residues (Input 3)
This refers to that portion of crop plants which is retained in the field after harvesting the main crop and later incorporated in the soil.Nutrients returned back into the soil through this pathway was determined by using the percentage nutrient content in the stover and the proportion of the total residue that is retained in the field.Responses from the survey in Mafai village indicated that of the

Outputs
Various pathways can be discerned regarding nutrient removal from the soil.The major pathways analysed during this study were nutrient removals (mainly N, P and K) from harvested crops plants, residue removals and loss through soil erosion.The determination of these nutrient output pathways from the farming system is described below.The landscape characteristics used to calculate soil and nutrient losses through erosion summarised in Table 1.

Harvested products (Output 1)
Nutrient output in harvested products was determined from crop harvest data reported by interviewed households.Crop yield data was then multiplied by the nutrient content in the product (e.g.grain, tubers and roots) so as to obtain the total amount of nutrients (N, P and K) that are removed from the soil through crop harvest.

Residue removals (Output 2)
The quantity of residue produced per unit area was determined based on the yield of harvested products and the respective harvest index for each crop.The term harvest index is used here referring to the proportion of the total crop biomass that is harvested.What remains is considered as residue.The total residue removal was derived from the interviews on the proportion taken away as animal feed, burnt or for other purposes.The amount of residue removed was multiplied by nutrient content in the residue to estimate the amounts of N, P and K taken away from the farming system.

Soil erosion (Output 3)
Nutrient loss in eroded soil material was determined by quantifying soil loss under the different crops grown in the area and later multiplying by the nutrient content in the soil samples.Quantities of soil loss by erosion were determined using the Universal Soil Loss Equation (USLE), that is, A = R * K * L * S * C * P.Where A = predicted soil loss in tonnes/hectare/year; R = rainfall factor (half the mean annual rainfall (Roose, 1976;Babu et al.,1978;FAO, 1979;Singh et al., 1981); K = soil erodibility factor; L = slope length factor; S = slope steepness factor; C = crop cover and management factor (shortly, crop cover); P = land management (erosion control) practice factor (Young, 1991;El-Swaify and Fownes, 1992).The values of L and S in the equation above were computed based on Smaling et al. (1993), while P was computed based on Wenner (1981) as follows.
L = (d/22.13) 0.5 S = (0.43 + 0.30 * s + 0.043 * s 2 )/6.613P = 0.2 + 0.03 * s Where: d = slope length in meters; s = slope gradient in percent.Inherent soil fertility was then used to translate the soil loss into N, P and K losses.Soil erosion is a selective process, taking away nutrients and fine soil particles first.The eroded material is thus richer in nutrients than soil it originates from (Stocking, 1984;Gachene, 1987;Mnkeni, 1992).The amount by which the eroded material is richer in nutrients than the soil, from which it is taken, is referred to as the 'enrichment factor'.The calculated nutrient loss in the above procedure was then multiplied by an enrichment factor of 2.
It should be noted that although leaching and denitrification are among the contributors to soil nutrient loss, they were not considered in this study because of the aridity of the climate in the study area.Due to low precipitation and high evapotranspiration of the area, no effective leaching or denitrification takes place (Young, 1991;FAO, 1978).As such, their role in nutrient loss from the surface soil layers in this area is considerably small.

Landscape and soil characteristics
Most of the soils in the study area are severely eroded exposing the red argilic layer.A few exceptions exist in some sites in the conserved forest areas (e.g. in Mafai).Soils of this area are classified as ferric acrisols (Tosi et al., 1982), ferralic cambisols (Mowo et al., 1993) and luvisols (Payton et al., 1992).According to FAO/UNESCO (1974), the three soil types belong to the low erodibility class.The mean texture of the sampled soil profiles is medium, thus qualifying for the soil erodibility factor (K) of 0.15 in the USLE equation.Chemical soil analysis based on the top 20 cm soil layer because root systems of most annual crops grown in the area occupy this plough layer.Table 2 presents characteristics of this soil layer.
The organic carbon content ranged from low (0.31 and 0.54% in Baura and Bolisa) to high (2.62% in Mafai).The average for the three villages was classified as low (Table 2).The texture of the soils was sandy clay in Mafai, loamy sand in Baura and sandy loam in Bolisa.Based on the National Soil Service (NSS, 1994) classification, the soils in this area are strongly acid.However, the recorded pH values are higher than the critical pH value of 5 for a maize/sorghum, cowpea and cassava based cropping system of the area (Aune and Lal, 1995) and four for general plant and microbial growth (Logan, 1990).These soils also had low N, P, CEC and low to medium values of other exchangeable cations.These low figures suggest that soil fertility may be exhausted by continuous cultivation without adequate replacement.

Inorganic fertilisers (Input 1)
The inorganic fertilisers used in the studied villages were N-P-K (20:10:10), UREA (46% N), Sulphate of ammonia (SA-21% N), Calcium ammonium nitrate (CAN-26% N) and triple super phosphate (TSP-46% P 2 O 5 ).Among the respondents who reported to have used inorganic fertilisers were 8.9% in Mafai village, while in Baura and Bolisa the percentages were 2.05 and 7.53% respectively.The remaining households did not use any fertiliser during the study period, and some had never ever used inorganic fertilisers before.The total amounts of N, P, and K nutrients applied through the use of these fertilisers are presented in Table 3. Considering the total area cultivated by respondents in these villages, it has been found that on average, the application of inorganic fertilisers on a hectare basis is about 2.60, 0.65 and 0.65 kg of nitrogen, phosphorus and potassium respectively (Figure 2).

Manure (Input 2)
Laboratory analysis for nutrients in farmyard manure indicated that the manure samples contained 1.56% N, 0.18% P and 0.005% K (on a dry matter basis).The N content of the manure samples compares well with the 1.3% reported by Smaling et al. (1993).However, the values for the other two nutrients, P and K are smaller.Smaling et al. (1993) used values as high as 0.5% for P and 1.6% for K.The availability and rate of manure application are influenced by the number of households keeping livestock and number of livestock units.Among the interviewed households only 40% are keeping livestock at present.The number of livestock units reported by these households was 42 in Mafai, 92 in Baura and 180 in Bolisa.These figures correspond to a total of 54.4, 110.4,and 216 tonnes of manure per year in Mafai, Baura and Bolisa respectively.Of all respondents, 67.32% had access to farmyard manure.It can be observed that the percentage of households using manure is larger than that of households keeping livestock.This disparity is due to the fact that the nonlivestock keepers obtain some manure from relatives and/or neighbours who keep livestock.Among those using manure 37.4% are residents of Bolisa, whereas Mafai and Baura had 16.32 and 13.6% respectively.The average manure application rate for all respondent households have been found to be 1.5, 3.2, and 2.9 tonnes/household per year for Mafai, Baura and Bolisa, corresponding to an application rates are 0.44, 1.15 and 0.94 tonnes/ha/year respectively.The average application rate for the three villages is 0.84 tonnes/ha/year.Table 4 shows the application rates for various nutrients.

Crop residue additions (Input 3)
Determination of N, P and K nutrients added to the soil through crop residues was based on the percentages of crop residue retained in the field after harvesting the main crop and the nutrient concentration in the residues.The harvest indices (HI), area under each crop and amount of harvested product are presented in Table 5.For P and K the analysis included the percent of the total residue that was burned, because they are not removed from the arable land by burning (Smaling et al., 1993).The nutrient contents in the crop residues used in this computation are presented in Table 6. Figure 3 shows the total residue yield that was used in nutrient calculation for both inputs and outputs, whereas the quantities of residue-nutrients added into the soil by each crop are given in Table 7.
The largest N-input per hectare came from pigeon peas and sunflower stover which contributed 8.6 kg/ha each.Bean's residues ranked second in the quantity of N added per unit area, providing 3.4 kg/ha, followed by finger millet (3.3 kg/ha).Other crops with significant residue-N input were maize, cowpea, and sorghum (Table 7).The average N additions from all crop residue are 5.9 kg/ha in Mafai, 3.2 kg/ha in Baura and 2.7 kg/ha in Bolisa.Significant P additions per hectare were contributed by sunflower (8.6 kg) followed by cassava (2.2 kg) (Table 7).Lowest values are found in Irish and sweet potato because of the low inherent quantities of this nutrient in the residue.Pigeon peas residue provided the largest share of K additions from residues, amounting to 20.1 kg/ha, followed by maize (16.2 kg) and sunflower (15 kg/ha).The average K addition per hectare ranged between 8.3 and 13.9 kg with considerable variations between villages.Some of the crops such as beans, cowpeas and groundnuts had high nutrient concentrations in their residue, giving a high nutrient input per hectare.
However, because of the small farm areas they occupy, their contribution to total residue input has been relatively small (Table 6).

Atmospheric deposition (Input 4)
The mean annual rainfall for Kondoa district for the past 63 years  has been 643 mm (Mbegu, 1988;  HI=harvest index (Agata, 1982;Cock, 1985;Victorio et al., 1986).Ngana, 1990).Using the theoretical expressions for atmospheric depositions described above, the following results were obtained.The amount of N deposited in the study area is 3.55 kg/ha/year (that is, 0.14 * 25.36) and that of P is 3.119 kg/ha/year (that is, 0.123 * 25.36).The amount of K deposited from atmospheric sources is 2.333  kg/ha/year (that is, 0.092 * 25.36).Similar findings for N deposition are reported for Northern Nigeria (Wilkes et al., 1984).

Nutrient removals in harvested products (Output 1)
Crop harvesting entails removal of nutrients from the arable land in the harvested products.The total quantity of the nutrients removed is a function of the area cultivated, yield per unit area and the nutrient content in the product.Grain is the major product harvested.Table 5 shows the land area that was grown with the different crops and yield of harvested products involved in the analysis, while Table 6 presented the nutrient content in the harvested products.The calculated overall average nutrient removals through harvested products from the respondents' crop fields were 14.0, 1.9 and 7.2 kg/ha for N, P and K respectively.At the village level, the nutrient removals were 14.7 kg N/ha in Mafai, 12.1 kg N/ha in Baura and 15.1 kg N/ha in Bolisa.Removals of P amounted to 2, 1.6 and 2.1 kg/ha respectively.In the same order, the values for K removed per hectare were 10.7, 5.2 and 5.7 kg (Table 8).
It can be seen from the largest N removal (at a rate of 24.2 kg/ha), followed by beans (22.2 kg/ha), sugarcane (21.4 kg/ha) and groundnut (18.6 kg/ha).Lowest removals were observed under sorghum (6.4 kg/ha).This may be explained at least in part by the relatively low content of N in its (sorghum) grain compared to other cereals involved in the analysis (Table 5).The largest P removals were observed under sugarcane and groundnut that extracted from the soil 3.2 and 3.0 kg/ha respectively.Other crops with large removals were pearl millet (2.5 kg/ha) and pigeon peas (2.2 kg/ha).Lowest quantities of P were removed by finger millet grain at a rate of 1.1 kg/ha.Sugarcane featured highest in terms of extracting K from the soil (Table 8).However, because of the small area planted with this crop its total contribution to the overall nutrient loss through crop harvest was relatively small.Other crops removing large amounts of K from the arable land are cassava, Irish potato and Pigeon peas, taking away 15.3, 11.0 and 9.8 kg/ha respectively.The remaining crops had smaller rates of K nutrient extraction from the soil.The overall removals by these crops were however large because of the large land area they occupy, especially for cereals.

Crop residue removals (Output 2)
The removal of nutrients from the arable land in form of crop residues was calculated in a similar way as calculations for harvested products (Output 1).The calculation based on the proportion of the residue that is taken away from the arable land and the nutrient content in this residue.The average crop residue yield used in the analysis is presented in Figure 3.The proportion of crop residue considered here is that taken away as animal feed and for other purposes, which was 25.1% in Mafai, 25% in Baura and 44.4% in Bolisa.For nitrogen, the proportion of residue burned was also taken into consideration since N is lost by burning.The latter constituted 19.5, 48 and 26.2% of total residue in Mafai, Baura and Bolisa respectively.P and K are not taken away from the system by burning (Smaling et al., 1993).Quantities of nutrient removed through this pathway are given in Table 9.The average residue nutrient removal from the agricultural system was established to be 5.7, 0.4 and 4.3 kg K/ha.Baura and Bolisa villages featured high in the removal of nutrients (especially N) in crop residues (Table 9).The explanation to the observed trends is the larger proportion of the total residue that is taken away from the arable land by burning, for animal feed and other purposes.The two types of residue removals in Baura and Bolisa village accounted for 73% and nearly 71% of the residue produced in the farm respectively.Regarding individual crops, sunflower, pigeon peas, cowpea and bean residues had the highest N removals (18.1 and 14.1 kg/ha respectively).Other crops removed relatively small quantities of N (Table 9).Cassava removed largest quantities of P (1.2 kg/ha), followed by finger millet (0.6 kg/ha), maize and Pigeon peas (0.5 kg/ha each).Other crop residue removals accounted for limited quantities, ranging from 0.1 to 0.4 kg P/ha.As for K, largest removals in crop residues were observed in pigeon peas whose residue removed 8.7 kg/ha.Maize was the next highest ranking crop in terms of removing K (7.9 kg/ha) from the agricultural land.

Nutrient loss through soil erosion (Output 3)
The quantities of soil loss under each crop in respective villages were obtained by multiplying the factors in the USLE for each crop so as to get respective quantities of eroded soil material (Figure 4).Let us look at an example of maize crop in Mafai village and how soil loss from erosion was determined.The factors considered here are rainfall factor (R = 200), soil erodibility factor under ferric acrisols (K = 0.15), slope length factor on 100m slope (L = 2.126), slope gradient factor, with slope gradient of 20% (S = 4.11), and land management (erosion control) practice factor (P = 0.8).Multiplication of these factors gives a soil loss 83.8 tonnes/ha/year.The calculated average soil losses from cropland owned by the respondent households are 62.4,16.5 and 11.2 tonnes/ha/year in Mafai, Baura and Bolisa respectively.The average soil loss on a hectare basis was highest under sunflower, maize, sorghum, finger millet and pigeon pea crop covers (Figure 4).The erosion loss was used in the determination of nutrient losses through erosion presented in Table 10.Soil erosion contributed greatest to the N loss.The calculated figures are in the magnitude of 253.0 kg/ha/year in Mafai, 17.8 kg/ha/year in Baura and 15.2 kg/ha/year in Bolisa.

Nutrient budgets
The term nutrient budget is used here to refer to the difference between total quantities of input to and outputs from the agricultural system.In all the three villages there were negative nitrogen budgets to the magnitudes of -252.4,-9.58 and -13.53 kg/ha/year in Mafai, Baura and Bolisa respectively (Table 11).Corresponding budgets for P were 4.16, 5.67 and 4.02 kg/ha/yr, while those for K were 0.07, 8.75 and -0.33 kg/ha/yr in the same order.The mean difference between the four inputs and three output factors for the three villages has indicated values of -91.85 kg N, 4.62 kg P, and 2.82 kg K/ha/yr.Among the four input factors, application of inorganic fertilisers has shown to play a small role as nutrient input source in the study area, contributing for only 2.6 kg N, 0.65 kg P and 0.65 kg K/ha/yr (Table 11), equivalent to 11.2, 9.1 and 4.5% of total N, P and K input to the system.Manure ranked highest as N input source, third as P source and fourth as K source, providing about 56.5, 20.8 and 0.3% of input N, P and K, in the same order.Incorporation of crop residues into the soil was the major contributor to K input and second as P and N source.Residues added into the system accounted for nearly 17.0, 26.3 and 79.0% of input N, P & K respectively.Atmospheric deposition contributed significant quantities of all the three elements, N, P, and K, ranking highest as input source for P, third N source and second K source, providing about 15.3, 43.8 and 16.2% of input N, P and K respectively (Table 11).Removal of nutrients in eroded soil material has been found to be the major pathway for N losses.This was followed by nutrient removal in harvested products.Residue removals had the lowest contribution to the overall N losses (Tables 10 and 11).The mean values for the three output factors in the same order are 95.33,13.97 and 5.73 kg/ha respectively.As for P and K budget, largest removals appeared to be in form of  et al. (1993), Roose (1986) and Lewis (1987).
harvested products, withdrawing 2.0, 1.6 and 2.1 kg P/ha/yr, and 10.7, 5.2 and 5.7 kg K/ha/yr in Mafai, Baura and Bolisa respectively.The overall average P and K removals in harvested products were 1.90 and 7.2 kg/ha/yr, in that order (Table 11).

Landscape and soil features
The various landscape and soil characteristics presented in Table 2 reveal that in of the study villages the soil are generally of low organic carbon content.Such soils are characterised by low nutrient and water holding capacity which makes these soils of limited productivity compared to the sandy clay and loamy textures.The low N, P, CEC and low to medium values of other exchangeable cations that these soils had suggests that soil fertility may be exhausted by continuous cultivation without adequate nutrient replacement.

Major sources of soil nutrient inputs
The major sources of nutrient supply to the agricultural system were found to be inorganic fertilisers (Input 1), manure (Input 2), crop residue additions (Input 3) and atmospheric deposition (Input 4).Although no fertiliser recommendations exist for study area, the whole of Dodoma (Mowo et al., 1993), the quantities of inorganic fertiliser used by respondent farmers fall far below the recommended rates for other places in Tanzania.In areas with similar soil types and agro-climatic conditions such as Shinyanga, the recommended rates ranges from 12-80 kg N/ha and 10-27 kg P/ha (Tosi et al., 1982).Limited access to fertilisers and high prices were among the reasons put forward by farmers for the low fertiliser application rates.The low rates of soil nutrient replacement from inorganic fertilisers may lead to decreased crop productivity if other nutrient sources are also limiting.
Manure is another important sources of soil fertility I the study area.However, the average application rate for the three study villages was very low, ranging from 0.44 to 0.94 tonnes/ha/yr) with an average of about 0.84 tonnes/ha/year.The observed manure application rate falls far below the 5 tonnes per hectare recommended for a sorghum/millet based farming system of Dodoma region, and 7-10 tonnes/ha necessary to meet the requirements of most crops (Agata, 1982).
This situation indicates the inadequate supply of nutrients required to boost modest crop production.This nutrient application rates from manure does not compare favourably with rates applied elsewhere.In Kisii district is Kenya, for example, nutrient application rates from manure are as high as 24 kg N, 5 kg P, and 25 kg K/ha/year (Smaling and Fresco, 1993).The obvious explanation is low manure application rates in the study area attributed to the fact that only few households keep livestock.
Crop residue addition is another important source of the major soil nutrients, N, P and K.However, it contribution to the soil nutrients pool is influenced by the proportion of the plant material that is retained in the field as crop residue after harvesting the main crop and the quality (nutrient concentration) of the residues.The proportion of the crop removed from the field is referred Note: sugarcane and sunflower residues are burned, thus only N gets lost from the system.P and K are not removed (Smaling et al. 1993).
to as harvest indices (HI), as described by Agata (1982), Cock (1985) and Victorio et al. (1986).Harvest index depends on the genetic traits of the cultivar and the conditions of the environment, such as photoperiod, air temperature, solar radiation, water supply, and level nitrogen fertilisation (Mazurczyk et al., 2009).Findings presented in Table 5 revealed that the largest N-input per hectare came from pigeon peas and sunflower stover which contributed 8.6 kg/ha each.Bean's residues ranked second in the quantity of N added per unit area, providing for 3.4 kg/ha, followed by finger millet (3.3 kg/ha).These observations demonstrate that integration of leguminous crop plants in the farming system may partly help alleviate the soil fertility problems.Increased growing of these crops may enhance soil N-status taking advantage of their N-fixing properties.Pigeon peas residue also provided the largest share of K additions from residues, amounting to 20.1 kg/ha/year, followed by maize (16.2 kg) and sunflower (15 kg/ha).Although there were considerable variations between the villages regarding the nutrient input from crop residue additions, this source of soil nutrient appears to be particularly important given that many farmers cannot afford inorganic fertiliser, and have limited access to manure.Soil fertility improvement in the field would thus depend on careful management of crop residues.Atmospheric deposition (Input 4) is another source of nutrient input to the soil.However, since this input is dependent on the amount of rainfall, its contribution to the Kondoa environment is small due to the semiarid conditions.Similar findings were reported for Northern Nigeria (Wilkes et al., 1984).

Mechanisms for soil nutrient removals (outputs)
Three major outlets contribute to the net removals of soil nutrients from the agroecosystems, namely nutrient removals in harvested products (Output 1), crop residue removals (Output 2), and nutrient loss through soil erosion (Output 3).Crop harvesting entails removal of nutrients from the arable land in form of harvested products.The total quantity of the nutrients removed is a function of the area cultivated, yield per unit area and the nutrient content in the product.Grain is the major product harvested.Crops such as sunflower, beans, sugarcane, groundnut were found to be among the crops that had the largest N removal, at a rate of 24.2, 22.2, 21.4 and 18.6 kg/ha, respectively.Other crops removed large amounts of other nutrients, particularly K from the arable land.These included cassava, Irish potato and pigeon peas, taking away 15.3, 11.0 and 9.8 kg/ha respectively.The possible implication of the observed trends in nutrient removal through crop harvests is that continuous growing of these crops that extract large quantities of nutrients from the soil may lead to continuous deterioration of the soil if equal or larger quantities of nutrients are not replaced into the soil through the various input sources.The ultimate impact is continued soil degradation and lowered productivity of the land and limited future sustainability.
Regarding crop residue removals (Output 2), the larger proportion of the total residue appears to be that taken away from the arable land by burning and for animal feed.The two types of residue removals in Baura and Bolisa village accounted for 73% and nearly 71% of the residue produced in the farm respectively, and contributing to considerable removals as presented in crop residue removals (Output 2), accounting for 5.7 kg N/ha, 0.4 kg P/ha and 4.3 kg K/ha being removed from the agricultural system through crop residue removals.It has been established from this study that among the crops grown, maize, sorghum, finger millet and pearl millet straws removed the largest amounts of all N, P and K, partly because of the high nutrient concentration in their residues and in part because of the larger land area under these crops.Reducing the quantities of residues removed from the cropland would thus significantly help maintain the status of soil fertility in the existing farming system.
Figure 4 presented the quantities of eroded soil material by estimating the quantities of soil loss under each crop in respective villages.Respective quantities of eroded soil material were obtained multiplying the crop factors in the USLE.At village level, the calculated average soil losses from croplands owned by the respondent households were 62.4, 16.5 and 11.2 tonnes/ha/year in Mafai, Baura and Bolisa respectively; the highest soil loss being under sunflower, maize, sorghum, finger millet and pigeon pea crop covers.The figures for eroded soil material are slightly smaller than those reported for some parts of Dodoma region.In Mpwapwa for example, Christiansson (1981), reported yearly values as high as 105 tonnes/ha/yr (that is, 70 m 3 /ha/year) under mixed native food crops in flat cultivation.In maize alone values of soil loss were 96.2 tonnes/ha/yr (that is, 64 m 3 /ha/yr), while in sorghum and pearl millet it was 85.1 tonnes/ha (that is, 56.7 m 3 /ha/yr) and 99 tonnes/ha/yr (that is, 66.6 m 3 /ha/yr) respectively.Higher figures are also possible under similar rainfall conditions.In Kenya for example, figures recorded are as high as 350 tonnes/ha/year (Bekele and Thomas, 1992).In all the three villages erosion removed smaller quantities of both available P and exchangeable K.This may be explained by the fact that often the available P and exchangeable K in the soil are in smaller proportions compared to quantities of total N. Mafai village contributed a larger proportion, over 80% of the total nitrogen loss in eroded soil material (Table 10).This is attributed to the steeper slopes of the terrain in this village that resulted in larger values of both S and P factors in the USLE (Table 1) with a consequent effect of compounding the calculated soil loss compared to Baura and Bolisa.Another explanation to this observation is that Mafai soils are nearly four times richer in topsoil N than both Baura and Bolisa (Table 2).It is evident from Figure 4 and Table 10, and studies elsewhere (Christiansson, 1981) that land-use systems under cereal crops such as maize, finger millet and sorghum, and root and tuber crops (such as sweet and round potato) grown on steep slopes could be highly erosive, leading to large quantities of valuable nutrients being transported away in eroded materials, particularly if no sound soil conservation measures are practised.

Predominantly nutrient mining agricultural system
In all the three villages there were negative nitrogen budgets to the magnitudes of -252.4,-9.58 and -13.53 kg/ha/year in Mafai, Baura and Bolisa respectively (Table 11).Implicitly this gives reflections on the prevalence of a general nutrient mining from the cropping system in the study area, which leads to a net negative nutrient budget, particularly for N and marginal positive budgets for P and K.The removals of nutrients is from eroded soil material and nutrient removals in harvested products have been found to be the major pathway for N losses, accounting for 95.33 and 13.97 kg/ha respectively.As for P and K budget, largest removals appeared to be in form of harvested products.Thus careful management of these output pathways is likely to enhance soil fertility and Reference for crop cover factor (C) are found in Smaling et al. (1993), Roose (1986) and Lewis (1987).What is definitely worrying from the findings of this study is the large deficit of the most growth limiting nitrogen.The point of concern here is how long the nutrient resource base is going to sustain production given the present soil management system and extraction rates.Both P and K are in marginal positive nutrient budgets.Such marginal positive trends do not rule out the possibility of these nutrients becoming limiting in the near future, or in some parts of the landscape, especially if nutrient uptakes in harvested products, residue removals or erosion loss are not replenished by equal or larger quantities.
The use of inorganic fertilisers (Input 1) remains the major source of nutrient supplement into the soil (at least in the short-run) to offset the continuous depletion that has resulted into the observed negative nutrient budgets.No fertiliser recommendations exist for the study area, and for Dodoma region as a whole.However, based on the calculated nutrient budget, 5 bags (1 bag = 50 kg) of UREA (46%N) or 9 bags of N-P-K (20-10-10) compound fertiliser would be required to offset the calculated negative N-budget.Another approach to replenish fertility in the soil would be increased planting of leguminous crops and tree species which fix nitrogen and add organic matter to the soil and reduce soil erosion.Leguminous trees incorporated in the farm are reported to fix up to 50 kg N/ha/yr and raise the productivity of semiarid soils (Mortimore, 1989).This can be adopted in areas with similar climatic conditions such as Kondoa.Trees incorporated in crop fields not only improve the fertility status of the soil but also reduces soil erosion through reduction of runoff (Young, 1991;Liniger, 1992).

Implication of the nutrient budget on agricultural sustainability
The agricultural system of the study area is characterised by declining crop productivity, mainly associated with negative balances for major nutrients like nitrogen.The amount of nutrients leaving the soil in harvested crops, by erosion and by residue removals, is grossly exceeding the nutrient input to the system by natural processes such as biological nitrogen fixation and atmospheric deposition, or artificially by organic manure and inorganic fertilisers.
Increasing and sustaining agricultural production in the inherently erodible and low fertility soils of the study area is a major challenge.This has resulted from among others factors, soil erosion and increased soil degradation in farmland, continuous harvesting without proportionate nutrient replenishment and residue removals.
One of the reasons for low agricultural production is the gradual process of soil nutrient depletion in many of the sub-Saharan land-use systems (CWFS, 1985;Smaling, 1993).Among the processes responsible for this nutrient depletion are soil erosion, crop harvesting, residue removals and leaching, phenomena that have received less public attention compared to aspects like crop failures due to erratic rainfall and climatic and nonclimatic factors.The nutrient status of the soil dictates the productivity of land and dwindling soil fertility has become one of the major constraints to crop production, in the Kondoa area as well other parts of Tanzania and Africa in general.This has resulted into reduced agricultural sustainability (Lynam and Herdt, 1989;Spenser and Swift, 1992;Rockström, 1997).Thus, in order to ensure agricultural sustainability there is need to enhance the management of all the nutrient input pathways while minimising the nutrient output pathways.

Conclusions
It has been established in the foregoing discussion that the agricultural system in the study area faces extremely large nutrient deficit, particularly for N, as expressed by the enormous negative nutrient budgets.The major concern among the nutrient output factors from the agricultural land is soil erosion.This nutrient loss pathway accounted for more than 80% of the N losses in eroded material.Thus efforts to control erosion are necessary to reduce the amount of soil removed from the arable land, and as such help to halt the negative nutrient budget trends in the farming system.Increased effort on on-farm soil conservation is an important undertaking.Among the on-farm conservation strategies could be the increased incorporation of crop residues into the soil (Input 2).For ensured sustainability of the agricultural system protection of the land against soil erosion has to take a long-term perspective.In the short run, one of the important factors in the efforts to raise agricultural productivity is increased availability and application of inorganic fertilisers, especially under conditions of heavy population pressures and dwindling reserves of uncultivated arable land.Given the growing land pressure in the study area and elsewhere in the country and rising fertiliser prices, a subsidy on fertilisers is critical especially to the resource-poor farmers to ensure food security and soil productivity.However, further research has to be conducted on crop responses to fertilisers in this area so as to come out with more feasible recommendations bearing in mind the climatic and socioeconomic conditions that prevail in the area.
Manure application (Input 2) provides nutrient and additional benefits such as increased water storage and nutrient retention in the soil.However, since not all households have got access to farmyard manure following expulsion of livestock from the area in the late 1970s, increased use of this input is unlikely to be achieved in the short-run.Present average manure application rate has been observed to be far below the recommended rates for the sorghum, pearl millet, and maize based farming system of study area.Such low rates pose a particular challenge given shortages of inorganic fertilisers.Incorporation of crop residues into the soil would, apart from contributing significantly to the maintenance of high nutrient levels in the soil, is also necessary to provide surface cover to the ground that is required so as to reduce surface runoff and control soil loss by erosion.Other conservation efforts such as including agroforestry trees in croplands and making contour ridges to control soil erosion need to be encouraged.Finally, a well-planned long term conservation strategy becomes an absolute necessity if agricultural productivity is to be made sustainable under such fragile semiarid environments.

Figure 1 .
Figure 1.Map of the Irangi Hills.The mapped area forms the Kondoa Eroded Area (KEA), an area recognised for its severe soil erosion with extensive gullies.The insert shows the location of the KEA within Tanzania.Source: Kangalawe (2001).

Figure 2 .
Figure 2. Area cultivated and average N, P and K application (kg/ha) from inorganic fertilisers.

Table 1 .
Land characteristics used in the calculation of soil loss by erosion in the survey villages.Refer to Table12for crop cover factor (C).total crop residue produced in the farm 55.4% is incorporated into the soil, 25.1% used as feed to livestock or other purposes and 19.5% is burnt.In Baura the corresponding values were 27, 25 and 48% respectively.Those for Bolisa village were 29.4,44.4 and 26.2%.The residue yield per hectare was calculated based on harvest indices of the different crops. *

Table 2 .
Some characteristics from composite soils samples (0-20 cm depth) in the study area.

Table 3 .
N, P and K nutrients (in kg)application from fertilisers by the surveyed households.

Table 4 .
Area cultivated and average N, P and K application (kg/ha) from inorganic fertilisers.

Table 5 .
Nutrient addition (kg/ha) from farmyard manure in the survey village.

Table 7 .
Nutrient content in harvested products and in residues (in percent).

Table 8 .
Total crop residue yield (kg/ha) used in the nutrient input and output calculations.
na = not grown by the respondents during the period under study.
Table8that among the crops involved in the nutrient budget analysis, sunflower had Total crop residue yield (kg/ha) used in the nutrient input and output calculations.Some crops were not grown by all the respondents during the period under study (e.g.cassava, cowpeas, groundnut, Irish potato, pearl millet, sugarcane), which raise their mean values.

Table 9 .
Nutrient addition (kg/ha) into the soil through incorporation of crop residues.
1 na = not grown by the respondents during the period under study.

Table 10 .
Nutrient removals (kg/ha) in harvested products in the three villages.

Table 11 .
Nutrient loss (kg/ha) from agricultural land through removals of crop residues.

Table 12 .
Crop cover factors and the calculated soil loss (tonnes/ha) by erosion under different crop covers.

Table 13 .
Nutrient loss (kg/ha) in eroded soil material under different cropping systems in the survey villages.

Table 14 .
Summary on nutrient input, output (in kg/ha/yr) and nutrient budget 1 from the different sources.
1Nutrient budget = Total input -Total output in kg/ha/yr.