Introduction

Continuous cropping without adequate restorative practices poses a serious threat to sustainability of agro-ecosystems. The compound growth rate in yield of major crops in India is either declining or negative over the period of 1980–1981 to 2011–2012 (MoA 2013). The increase in food production must have to come from increased productivity, since horizontal expansion of cultivable area is not possible at this juncture of exploding population as that in India (Chhonkar 2003). However, at the same time, soil nutrient depletion and other forms of degradation threaten the increase in productivity (Surendran and Murugappan 2007a; Surendran et al. 2016). Previously, much research was focused on increasing the agricultural production but a gradual shift was made toward a long-term perspective considering both current and future production as well as the environmental impacts.

Soil fertility decline generally does not get the same public attention as droughts, pest infestation, etc., since it is a gradual process and not associated with catastrophes and mass starvation; it is largely imperceptible. The preservation and maintenance of fertility necessitate the investigation of nutrient element regime of the soil, which represent the life media for microbial activities as well as crop. This necessitates a regular monitoring of changes in soil fertility that occurs in the soil. For understanding the role of different process, a budgetary approach offers good tool through analyzing the turnover of nutrients in the soil-plant system at different spatial scales (World Bank 2007; IISD 2009; OECD 2010; Leip et al. 2011; Fethi and Leip 2015).

Soil nutrient balance—a review

Soil fertility decline can be accessed via expert knowledge systems, the monitoring of soil chemical properties over time (chronosequences) or at different sites (biosequences), and the calculation of nutrient balances, with the last one being the most used and cost-efficient technique (Cobo et al. 2010). Suggestions to use nutrient balances in nutrient management dates back more than 150 years but have only become accepted in farmer’s practice in the last decades (Van Noordwijk 1999; Surendran and Murugappan 2010; Cobo et al. 2010). Stoorvogel and Smaling (1990) and Smaling (1998) calculated the nutrient balance for 35 Sub-Saharan African countries and reported the seriousness of nutrient depletion on future food production. Nutrient balances have been monitored by number of authors in European Union and they found that N, P, and K budgets are in the range of deficit to surplus (OECD 2010). However, it has only been in the last decade, as concerns for soil fertility decline have increased and the limitations of standard chemical fertilizers have been recognized; thus, the nutrient budgeting and balance analyses have come to the fore (De jager et al. 2001; Bekunda and Manzi 2003; Sheldrick et al. 2003; Surendran and Murugappan 2007a; Phong et al. 2011).

Understanding and modeling nutrient cycles in food and related agro-industrial systems is a crucial task. Although nutrient management has been addressed at the plot and farm scales for many years now in the agricultural sciences, there is a need to upscale these approaches to capture the additional drivers of nutrient cycles that may occur at the local, i.e., district scale. In evaluating nutrient budgets, it is important to be aware of the use of different system boundaries at regional scale (Leip et al. 2011; Phong et al. 2011). Larger-scale estimates of nutrient balance and budgeting have become increasingly influential in policy discussion related to soil fertility management and sustainable agriculture (IFPRI 1995; FAO 2004; OECD 2010).

The contamination of freshwater systems by excess nutrients results in many negative effects on the ecosystem, including anoxia, fish deaths, and the development of toxic algal blooms. Moreover, these can have adverse impacts on livestock, wildlife, and human health (Duong et al. 2012; Sutton et al. 2013). These impacts include negative effects on biodiversity, eutrophication, nitrate accumulation in waters, acidification, nitrous oxide emissions, and risks to human health due to exposure to ozone and particulate matter (Smil 2011). The agricultural sector is an important source of the N that ends up in groundwater and surface water and the atmosphere (Erisman et al. 2013; Fowler et al. 2013). Similarly, mining of nutrients will result in deficiency of nutrients and negatively affects plant growth, resulted in reduction in food grain production, and affect the sustainability of the entire ecosystem.

Some of the nutrient flows, viz., leaching, denitrification, and erosion losses, are hard to quantify on a routine basis. These flows are often estimated on the basis of (pedo) transfer functions, an approach followed for Sub-Saharan Africa (SSA) by Smaling et al. (1993) and Stoorvogel et al. (1993) and adopted and also critically reviewed by others (Smaling 1998; FAO 2003; Schlecht and Hiernaux 2004; Faerge and Magid 2004; De Ridder et al. 2004; Lam et al. 2005; Dang 2005; Khai et al. 2007; Wijnhoud 2007; Lesschen et al. 2007; Phong et al. 2011). Bassanino et al. (2011) calculated nutrient budgets according to the IRENA European methodology in Italy and found the highest nutrient surpluses (103, 39, and 95 kg ha−1 for N, P, and K, respectively) in the most intensely managed area. For EU27, N surplus is 55 kg N ha−1 year−1 in a soil budget and 65 kg N2O N ha−1 year−1 and 67 kg N ha−1 year−1 in land and farm budgets, respectively (Leip et al. 2011). A comparison of nitrogen (N) budgets of agro-ecosystems is made for the EU27 countries by four models, i.e., INTEGRATOR, IDEAg, MITERRA, and IMAGE, with different complexity and data requirements. The models estimated a comparable total N input in European agriculture, i.e., 23.3–25.7 M t N year−1, but N uptake varies more, i.e., from 11.3 to 15.4 M t N year−1 leading to total N surpluses varying from 10.4 to 13.2 M t N year−1 (De vries et al. 2011).

Even though there are lot of such studies across the globe, information on nutrient balance research in India is scarce. Much of the published data in India is on the national/regional level and deals only with the fertilizer inputs and random estimates of crop nutrient uptake. Researchers in the past have reported indiscriminate mining of N, P, and K from soil reserves in all the agro-climatic zones across India (Yadav et al. 2001; Katyal 2001; Kumar et al. 2001; Krishna Prasad et al. 2004; Surendran and Murugappan 2007a, 2007b) without considering the soil losses through erosion and leaching. The first phase of our study on Coimbatore District showed negative balances for N and K and positive for P indicating the mining of nutrients (Surendran and Murugappan, 2007b), and hence, the authors thought of developing a strategy for sustainable agricultural production. Keeping these facts in view, the present study was carried out to calculate nutrient budget of Erode and Coimbatore Districts of Tamil Nadu state of India, by using the decision support model “NUTrient MONitoring (NUTMON)-Toolbox” (Vlaming et al. 2001), and come out with a policy Decision Support System (DSS) and strategies for improving the agricultural productivity in the Western Zone of Tamil Nadu.

Materials and methods

Study area

The study area is comprised of two contiguous districts in the western agro-climatic zone of Tamil Nadu in the southern part of India, and site characteristics are explained in Table 1 and Fig. 1. Usually, dry climate prevails in most part of the district except western part, which has a semi-dry climate. Somayanur, Pichanur, Peelamedu, Irugur, Palathurai, Periyanaickenpalayam, Noyyal, Chavadiparai, Dasarapatti, Palladam, and Anaimalai series cover a major part of the soil series in the study area (Soil Atlas 1998). Soil map of both the districts are presented in Fig. 2 (NBSSLUP 1997). The groundwater level fluctuates annually between 3 and 15 m in wet areas and 15 and 35 m in dry areas. The main source of irrigation in the study area is by canals, tanks, and wells. Cropping pattern varies widely with the soil types and irrigation facilities.

Table 1 Description of the study area
Full size table
Fig. 1
figure 1

Geographical location of the study area

Full size image
Fig. 2
figure 2

Soil map of Coimbatore and Erode Districts

Full size image

Model description

The present study aims at application of the NUTMON concept/software tool which is developed by integrating different knowledge systems (farmers and scientific) and interdisciplinary links (soil science, agronomy, and economics) at various spatial scales (field, farm, and catchment/district level) on the assessment of nutrient inflows and outflows.

Modules

NUTMON-Toolbox module was used to calculate the nutrient flows between the units and nutrient balances. This module includes five inflows, viz., mineral fertilizers (IN1), manure (IN2), deposition (IN3), biological N fixation (IN4), and sedimentation (IN5), and five outflows, viz., harvested product (OUT1), crop residues (OUT2), leaching (OUT3), gaseous losses (OUT4), and erosion (OUT5).

Calculating inputs and outputs

Nutrient flows are quantified in three different ways, viz., by using primary data, estimates, and assumptions (Smaling and Fresco 1993; Surendran and Murugappan 2007b, 2010). Flows directly related to farm management were quantified from the primary data. Flows quantified in this way are the use of chemical fertilizer (IN1), organic inputs (IN2), farm products (OUT1), and other organic products (OUT2) and redistribution of household waste, crop residues, and farmyard manure (FYM). The resulting data fall in the category of primary data. These flows are quantified using the following equation:

$$ \mathrm{Flows}={{\displaystyle \sum}}_x\mathit{\mathsf{w}}\mathit{\mathsf{d}}\ \mathit{\mathsf{Pro}}{\mathit{\mathsf{d}}}_{x.t} \times \mathit{\mathsf{f}}\mathit{\mathsf{r}}\ \mathit{\mathsf{Pro}}{\mathit{\mathsf{d}}}_{\mathit{\mathsf{x}}} $$
(1)

where wd Prod x,t  = amount of product x in month t kg and fr Prod x  = nutrient content in product x in kg/kg

Information on nutrient use applied through chemical fertilizers (IN1) per ton of nitrogen, potassium, and phosphorus (NPK) was obtained from the FAI database (FAI 2010). Manure production (IN2) in each district was calculated by multiplying the per capita manure with livestock population (Murugappan et al. 1999). The quantity of manure produced by individual animal was calculated based on average body weight and by using the equation developed by Merck Vet Manual (1998). Removal of harvested produce (OUT1) entails loss of nutrients and the quantity being determined by the average yield of the particular crop and its nutrient content. The average yield for all the crops cultivated in this district has been taken from the Season and Crop Reports (2010) published by the Government of Tamil Nadu, and by using the average nutrient content of each crop (Tandon 1997), OUT1 (nutrients exported out of the farm in crop produces) was calculated. Nutrient export in crop residues (OUT2) was calculated in a similar way by assuming that only 20 % of the generated residue is being returned directly into field as a source of nutrients and the remaining 80 % is being fed to the animals or burned as fuel (Tandon 1992).

Transfer function or models used

Atmospheric deposition (IN3), biological N fixation (BNF, IN4), leaching (OUT3), and gaseous losses (OUT4) were quantified fully on the basis of off-site knowledge using transfer functions, and the resulting data are estimates. Due to lack of point data on wet and dry depositions (IN3) at district level, the built-in transfer functions in NUTMON-Toolbox were used to calculate IN3 as done by Smaling et al. (1993), where nutrient input was considered as a function of square root of average rainfall in mm year−1. Inflow through atmospheric deposition (IN3) in month t kg−1 is calculated using the in-built regression equation of NUTMON-Toolbox, which is given in Eq. (2).

Nutrient from atmospheric deposition

$$ \left(\mathrm{Area}/10,000\right) \times \left(\mathrm{SQRT}\left(\mathrm{PrecAnnual}\right)\right) \times \left({\mathrm{PrecMonth}}_t \times \mathrm{Annual}\right) \times \mathrm{R}\mathrm{e}\mathrm{g}.\ \mathrm{Coef} $$
(2)

where Area = area hectares; PrecAnnual = precipitation mm/year; PrecMonth t  = precipitation mm/month; Reg. Coef = regression coefficients for N, P, and K.

Non-symbiotic N fixation (IN4b) is calculated using a function relating N fixation with mean annual precipitation. A small rainfall-dependent contribution from non-symbiotic fixers was accounted as per the procedure of Stoorvogel and Smaling (1990). N input through biological fixation (IN4) is given as

$$ =\mathrm{IN}4\kern0.5em \mathrm{N}\mathrm{o}\mathrm{n}\hbox{-} \mathrm{Symb}\ t\ p\kern0.5em + \mathrm{IN}4\ \mathrm{Symb}\ t\ p $$
(3)

Non-symbiotic N fixation by crops in primary production unit (PPU) p in month t kg is given in Eq. (4).

$$ \mathrm{IN}4\ \mathrm{N}\mathrm{o}\mathrm{n}\hbox{-} \mathrm{Symb}\ t\ p = \left(\mathrm{Area}/10,000\right)\kern0.5em \times \kern0.5em \left(1/12\right) \times \kern0.5em 2\kern0.5em + \left(\mathrm{PrecAnnual} - 1350\right)\kern0.5em \times \kern0.5em 0.005\Big) $$
(4)

The relative contribution of symbiotic and associative N fixation to that of free living organisms to the global total was taken as 70:30 as assessed by Paul (1988). The amount of N fixed (IN4) was calculated in the present study based on this assessment. N contribution from groundwater (IN5) is considered negligible in tropical conditions (Carolien kroeze et al. 2003), and therefore, only P and K inputs from sedimentation were accounted in IN5 based on the results of Abedin et al. (1991) and Handa (1998), who, respectively, calculated 1.5 and 10 kg K year−1 as inputs from sediments. Leaching of N and K (OUT3) is assumed to be uniform for all soil-bound sub-systems, whereas leaching of P is assumed to be zero. The percentages of leaching for both nutrients are calculated as a function of the clay percentage of the soil and the mean annual precipitation using transfer functions based on built-in model (Smaling et al. 1993).

For N,

$$ \left(\mathrm{Mineralized}\ {N}_p/12\right)\kern0.5em + \mathrm{IN}1\ \mathrm{MinFert}\ {N}_{p\ t} + \mathrm{IN}1\ \mathrm{MinOrg}\ {N}_{p\ t}\Big)\times \left(2.1\times {10}^{-2}\times \mathrm{PrecAnnual}-3.9\right) $$
(5)

For K,

$$ \mathrm{frLeach}\ {K}_p\times \left(\;\left(\mathrm{Exch}{K}_p \times 1/12\right)\kern0.5em +\kern0.5em \mathrm{IN}1\ \mathrm{MinFert}\ {K}_{p\ t} + \mathrm{IN}1\ \mathrm{MinOrg}\ {K}_{p\ t}\right) $$
(6)

where IN1 MinFert p,t is the inflow from fertilizers on PPU in month t, IN1 MinOrg p,t is the inflow from organic manures on PPU in month t, frLeach p is the fraction of potassium leached from PPU p, and ExchK p is the exchangeable K in soil PPU p.

The mineralization rate to calculate OUT3 with respect to soil N was assessed in a column study. In this representative, surface soil samples were collected from the study area and packed in the column of PVC pipes having a diameter of 10 cm and height of 45 cm. The soils in the column were maintained at field capacity level. Prior to incubation, the soil was washed with distilled water and the moisture content was maintained at field capacity throughout the experimentation. The soil samples were extracted for NH4 and NO3–N using 2 M KCl as per the procedure of Bremner and Keeney (1985). The leachate was collected everyday and analyzed for NH4 and NO3–N immediately, until a static condition was reached. Based on the released quantity of NH4 and NO3–N, the mineralization rate of nitrogen was calculated. Total soil N was derived from collecting representative samples covering entire district and analyzed for its total N content. N mineralized (N min) in 0–20-cm soil layer is calculated using Eq. (9).

$$ {N}_{\min } = 20 \times {N}_{\mathrm{tot}} \times M $$
(7)

The percentage of gaseous loss (OUT4) of N is assumed to be the same for each primary production compartment and is calculated as a function of the clay percentage of the soil and the mean annual precipitation using a transfer function (Smaling et al. 1993). Gaseous losses (OUT4) are calculated by multiplying the loss percentage by fertilizer N, mineralized soil N, and given in Eq. (10).

$$ \left(\mathrm{Soil}\ \mathrm{N} + \mathrm{Fertilizer}\ \mathrm{N}\;\right)\kern0.37em \times -9.4\kern0.5em + 0.13 \times {\mathrm{frclay}}_p \times 100\kern0.5em + 0.01 \times \mathrm{PrecAnnual} $$
(8)

Erosion (OUT5) can occur in any of the primary production compartments. Soil loss (kg ha−1 year−1) is estimated using the universal soil loss equation (Wischmeier and Smith 1978). Soil loss is converted to nutrient loss (kg ha−1 year−1), using the total N, P, and K contents (%) of the soil and an enrichment factor.

$$ \Big(\mathrm{Soil}\ {\mathrm{loss}}_f \times 1000 \times {\mathrm{frSoil}}_p \times \mathrm{EnrichFact} \times \mathrm{SoilFormFact} \times \left({\mathrm{Area}}_p/{\mathrm{Area}}_f\right) \times {\mathrm{Cusle}}_p $$
(9)

where Soil loss is the soil loss from FSU, frSoil is the nutrient content in soil on PPU, EnrichFact is the enrichment factor, and Cusle p is the USLE crop cover factor for PPU p.

It is not easy to derive R factor from commonly collected meteorological data. On the basis of literature data, this was set at 0.25 % for the study area. The K factor also varies with types of soil, and previous studies showed that K value ranged from 0.197 to 0.217 with 0.202 as an average. It has been found as 0.11 for Chambal ravines (Pratap et al. 1978). But for major Indian soil types, the value of K was 0.12 (Biswas and Mukherjee 1982). Slope (S) and L were determined as per the procedure of Mitchell and Brubenzer (1980). The degree of land cover also varies, and it was difficult to quantify in terms for the district. The value of C factor estimated for maize crop ranged from 0.266 to 2.528 based on the growing practices (Agnihotri et al. 1987). So, from the previous studies, an average C factor was estimated for the entire district as 1.4. Land management factor P was derived from Wenner (1981). The estimated average inherent soil fertility was used to translate soil loss data into N, P, and K losses and by multiplying by an enrichment factor to arrive at OUT5. Enrichment occurs because of the fact that the finest soil particles are the first to be dislodged during erosion and eroded soil material tends to contain more nutrients than the soil. In the present study, the enrichment factor is set at 1.5 for N, P, and K by assuming a ratio of 1:1.5 for the nutrient content of the original soil to that of the eroded soil.

Strategy development using a DSS

Nutrient depletion is the result of a net imbalance, between incoming and outgoing nutrients in farm inputs and outputs. Because many aspects of farm management influence these processes, there is a need for a “basket of technology options,” addressing the various causes of depletion. However, fertilizer being the major input, a strategy was worked out for a selected farm under the study area using a DSS named DSSIFER software (Decision Support System for Integrated Fertilizer Recommendation).

DSSIFER is a computer software, which gives crop-, site-, and season-specific fertilizer prescriptions for a specified yield target/percentage yield sufficiency based on the equations generated through soil test crop correlation studies (STCRs) which were conducted earlier in Tamil Nadu state (Surendran and Murugappan 2010).

Fertilizer prescription of a specified yield target

$$ U=\alpha\;S + \beta\;F $$
(10)

Fertilizer prescription for percentage yield

$$ Y = A\ \left(1\ \hbox{--}\ {e}^{-\mathrm{c}1\mathrm{b}-\mathrm{c}\mathrm{x}}\right) $$
(11)

Strategies were worked out for achieving a neutral nutrient balance in the farm using DSSIFER-based fertilizer prescription rate (both organic and inorganic) for the soil fertility status of the individual crops, by taking into account the season and specific yield target.

Description of the selected farm

The irrigated farm selected for developing a strategy is located at Singampettai village in Bhavani block of Erode District of Tamil Nadu. The cultivable area of the farm is of 5.87 ha. The farm is irrigated through the well located within the farm. The soil of the farm taxonomically belongs to Ammapettai soil series (Soil Atlas 1998). The farm comprises of four farm section units (FSUs) and is divided based on the homogenous soil properties, slope, and crops grown in the farm. These FSUs consist of nine crops, viz., sugarcane, turmeric, sugarcane, brinjal, fallow, sunflower, tapioca, ratoon cane, and tapioca (Fig. 1). Nutrients for the farm were mainly through chemical fertilizers and organic manures that are met from external sources besides on-farm-generated manures. The farmer besides using on-farm manure also purchases manure off-farm and imports it into the farm. This was included as IN2a and IN2b. Besides, a part of crop residue was also directly recycled into the farm by incorporation/burning. Outflows from the farm were crop uptake (OUT1), removal in crop residue (OUT2), leaching (OUT3), gaseous loss (OUT4), and erosion loss (OUT5). Nutrient balance was worked out using NUTMON for both farmer’s practice of fertilizer application and DSSIFER-based fertilizer application.

Results and discussion

Quantification of inputs

The consumption of fertilizer in the study area has registered a spectacular growth, i.e., 30 times (0.6 to 19 MT) during the last three decades, owing to the adoption of green revolution technological packages (FAI 2010). The consumption of NPK fertilizers (IN1) in Coimbatore District were 31,986, 10,793, and 22,715 t, respectively, and in Erode district were 36,732, 11,406, and 13,422 t, respectively (Surendran and Murugappan 2007b; FAI 2010). The population of livestock and manure produced by individual animal and the nutrient potential in animal manure are furnished in Table 2. Animal manure enters the system after collection from livestock units in the farm itself (on-farm manure) or imported from nearby farms (off-farm manure). Available literature indicates that under Indian conditions, only 40 % of the total manure is used in agriculture and rest being used either as cooking fuel or wasted (Tandon 1997; Redding 1999). Therefore, out of the total potential nutrient generated from manures, the quantity that enters into the farm as nutrients (IN2) is given in Table 2.

Table 2 Nutrient potential from manure in Western Zone of Tamil Nadu
Full size table

The N, P, and K depositions (IN3) were derived from the built-in transfer functions in NUTMON-Toolbox. These values for Coimbatore and Erode Districts were 3.70 N, 0.60 P, and 2.42 K and 3.59 N, 0.59 P, and 2.36 K kg ha−1 year−1, respectively. Land depositions of NH3/NH4 from the atmosphere provide about 10–20 kg N ha−1 year−1 (Derwent et al. 1988). Similar results of nutrient deposition were also reported by Sapek and Sapek (1993). They have estimated that 17 kg of N and 0.5 kg of P ha−1 year−1 were deposited in land from atmosphere. Abedin et al. (1991) and Handa (1988) have found that the annual inputs of K through atmospheric deposition exceeded 10 kg ha−1. These results are in agreement with the observations made on nutrient deposition in the present study. Since the variations in climate within the study area are not considerable, the quantified deposition data was extrapolated at the district level. Non-symbiotic N fixation calculated for Coimbatore and Erode Districts were 572 and 630 t year−1, respectively. Considering that the relative contribution of symbiotic and associative N fixation and non-symbiotic N fixation by free living organisms to the global total to be in the ratio of 70:30, the amount of N fixed through symbiotic fixation was arrived for Coimbatore and Erode Districts (IN4) as 1470 and 1335 t year−1.

Quantification of outputs

Nutrients exported through harvested produce (OUT1) and residues (OUT2) in Erode and Coimbatore Districts have been calculated and presented in Tables 3, 4, 5, and 6. Total losses of N through leaching (OUT3) for Coimbatore and Erode Districts were found to be 8962 and 8095 t year−1, respectively (28.41 and 28.30 kg N ha−1 year−1, respectively). This calculated leaching loss of N from the system is similar to the value estimated by Bjornberg et al. (1996) (17–72 kg N ha−1 year−1). As suggested by Dobermann et al. (1996), the leaching loss of P was assumed to be negligible, as most of the soils in the study area tend to retain/fix P. In fine-textured soils, K leaching generally does not exceed 2 kg ha−1 year−1 (Tisdale et al. 1985). However, leaching of K on acid sandy soils in southern Nigeria accounted to 16 kg ha−1 year−1 of soil-derived K and 10 kg ha−1 year−1 of surface-applied K at an application rate of 60 kg ha−1 year−1 (Omoti et al. 1983). Average K concentrations in soil water extracted by means of ceramic suction cups at 1-m depth were 0.6 mg K L−1 corresponding to a K leaching loss of 1.5 kg ha−1 year−1 (Askegaard and Eriksen 2000). The calculated K losses due to leaching in Coimbatore and Erode Districts were 2054 and 1860 t year−1.

Table 3 Nutrient export through harvested produces of Coimbatore District
Full size table
Table 4 Nutrient export through crop residues of Coimbatore District
Full size table
Table 5 Nutrient export in harvested produces of Erode District
Full size table
Table 6 Nutrient export in crop residues of Erode District
Full size table

N losses (OUT4) calculated using the built-in multiple regression equation in NUTMON-Toolbox for Coimbatore and Erode Districts were 555.17 and 639 t year−1. The estimated soil losses using the universal soil loss equation for these districts were 355 and 383 kg ha–1 year−1. The estimated soil loss clearly matched with the soil loss calculated on red and black soils of the study area by Santhanabosu and Sivanappan (1989), who reported losses to the tune of 0.236 to 585 t ha−1 year−1.

Quantification of the nutrient balance

Quantification of the nutrient balance of Coimbatore District in western zone of Tamil Nadu revealed that the sum of the input factors (IN1 to IN5) minus output factors (OUT1 to OUT5) produced a negative balance for N (–3160 t year−1) and K (–3073 t year−1) and a positive balance for P (+6423 t year−1) (Surendran and Murugappan 2007b). Similar trend of results was noticed in Erode District; the balances for N and K were negative (−1033 and −17,811 t year−1, respectively), whereas the P balance was positive (+4580 t year−1; Table 7). The per hectare N and K balances were also negative for both the districts (−3.3 N and −58.6 N and −10.1 K and −9.8 K kg ha−1 year−1+, respectively), whereas P registered a positive balance in cases (+14.5 and +20.5 kg ha−1 year−1, respectively; Fig. 3). The aggregated nutrient balances for the two districts were also negative for N (−4193 t year−1) and K (−20,884 t year−1) and positive for P (+11,003 t year−1; Table 7). The positive balance of P is the result of the accumulation of P over years due to P fertilizer application, and also, the losses were low since the soils in the study area tend to fix P (Kumaraswamy 2001). The positive balance of P will result in an increased risk of nutrient emissions to the environment causing nutrient toxicity. The enhancement of P in soil reserves may lead to the contamination of surface and groundwater causing accelerated eutrophication and poses risks of toxicity to aquatic life. Similarly, for the negative balances of N and K, the reason being the sum of emissions was much higher than the emissions. The negative balances of N and K imply that a net depletion of these nutrients from the soil reserves occurs. N is mobile in the soil system and is also lost from the system by leaching, volatilization of NH3 in soils whose pH is more than neutral, and denitrification in soils where submergence is a practice. All the three processes operate in the study area. While comparing the districts, it was found that the high negative K balance in Erode District is mainly due to existing cropping system, in which K-depleting crops like turmeric, tapioca, banana, maize, and paddy are grown. In the case of K, removal of harvested product (OUT1 and OUT2) proved to be the strongest negative contributor, followed by leaching which occurs in the study area since the soil characteristics are conducive for leaching. Yet, the wider negative balance obtained may be due to sub-optimal use of inputs in the study area (Murugappan et al. 1999). Continued nutrient mining process that goes at the expense of soil nutrient from the mineral and organic matter reserves limits the crop yield and renders the land chemically degraded (Murugappan 2000). We can undoubtedly infer that the current practice of cropping system and nutrient management are exhaustive in terms of N and K withdrawals and cause greater drain of these nutrients from soil reserves. This process unchecked might lead to an irreversible loss of soil fertility and eventually jeopardize the production in the years to come and leaving the soils unfertile for the posterity. Declining soil fertility also prevents income generation of the rural community and triggers the migration of the rural population into urban centers in search of income and food at the expense of social security. A nutrient audit model described in this study can effectively play a role in assessing the problems and helps developing strategies and practices that can be used to make useful policy interventions.

Table 7 Soil nutrient balance for Western Zone of Tamil Nadu
Full size table
Fig. 3
figure 3

Soil nutrient balance for Erode and Coimbatore Districts of Tamil Nadu under NUTMON alone and with NUTMON and DSSIFER combinations

Full size image

Strategy development using a DSS

Nutrient balance at crop and farm levels by NUTMON

Nutrient balances at crop level covering all the FSUs in the farm, which were generated using NUTMON-Toolbox, are presented in Table 8. All the crop activities showed a negative balance of N except brinjal, where the N balance (7.4 kg ha−1) was positive. A mixed pattern of positive and negative balances was observed with P. A positive full balance of 16.1 kg P ha−1 was noticed with tapioca followed by turmeric (14.1 kg ha−1). P balance was highly negative (−63.8 kg ha−1) with sugarcane. K balance also showed a mixed trend of results. The highest negative K full balance was observed with sugarcane (–204.2 kg ha−1). However, the positive K balance was noticed in sunflower (24.7 kg ha−1). In nutshell, none of the crop activities exhibited a positive balance for all the three nutrients. For the farm as a whole, the nutrient balance was expressed as the sum of inputs minus the sum of outputs covering all FSUs, SPUs, and RUs. There has been a slight variation in the nutrient balance of the farm than the individual crops. NUTMON-Toolbox-generated nutrient balance for the experimental farm as a whole showed that the full balances were positive for P and negative for N and K, similar to the trend observed at district level data mentioned earlier (Fig. 3).

Table 8 NUTMON-Toolbox-generated nutrient balance at crop level for the irrigated large farm in Singampettai
Full size table

Nutrient balance for the farm using NUTMON combined with DSSIFER

The fertilizer recommendation derived from DSSIFER tool for the crops grown at Singampettai irrigated farm were 815.8, 252.5, and 586.4 kg of N, P, and K, respectively (Table 9). The fertilizer prescription for most of the crops revealed that the presently followed state recommendation is sub-optimal and an upward revision is the need of the hour to make the crop production profitable and sustainable. NUTMON-Toolbox-generated nutrient balance was positive for all the three nutrients when the fertilizer program is DSSIFER based. Among the N, P, and K balances, N balances were highly positive at both partial and full balance modes (88.0 and 68.3 kg ha−1, respectively).

Table 9 NUTMON-Toolbox-generated nutrient balance when fertilizer program is from DSSIFER for the experimental farms in Singampettai
Full size table

Nutrient balance at district level by using NUTMON combined with DSSIFER

Nutrient balances were computed using the NUTMON-Toolbox by replacing the farmer’s practice of fertilization with DSSIFER-generated crop and site-specific fertilizer recommendations and the farmer’s crop yield with the yield targets fixed in DSSIFER. The results revealed that the nutrient balances were positive at both farm and district levels. The mean N, P, and K balances were 19.6, 38.7, and 2.1 kg ha−1 year−1 for Erode District and 63.9, 71.2, and 37.7 kg ha−1 year−1 at district level (Fig. 3). Thus, fertilizer program generated by DSSIFER is not only balanced but also ensures sustainability of the agro-production systems.

Other management options for mitigation of nutrient mining

The major negative contributor is the outflow through harvested crop produce, which cannot be curtailed since the main aim of the farmers and policy makers is to enhance the productivity to feed the enormous population. Solutions to nutrient depletion and accumulation need to focus on economically feasible and socially acceptable technologies. There is a wide range of management interventions (nutrient-saving technologies, viz., increasing the use efficiency and preventing/minimizing the losses) to influence soil nutrient balances.

Some of the strategies are discussed below, which are relevant to the study area.

  1. 1.

    Split applications of fertilizers can be made to match the nutrient requirement of the crop with that of the nutrient availability in soil, thereby increasing the efficiency of applied fertilizers.

  2. 2.

    The farmers in the study area have to be trained for efficient recycling of farm wastes, proper manure collection, and storage methods so as to achieve a positive balance.

  3. 3.

    Farmers should be trained in such a way to know about the whole system of their farm, nutrient inflows and outflows creating awareness about the activities which deplete their soil fertility, and also training on efficient management techniques to mitigate them (Murugappan 2000).

  4. 4.

    Crop rotation should involve shallow-rooted crop and deep-rooted crop for efficient transfer of nutrient flow from sub-soil to surface soil (Hartemink 1994).

  5. 5.

    Introduction of green manures and legumes in the system is one of the technological options to replenish the soil nitrogen level without any external inputs.

  6. 6.

    Even though farmers are mainly concerned with the current season, awareness has to be created about the effects of soil fertility decline. Feedback mechanism has to be developed or included in the training program so that everyone will know about declining soil fertility, which in turn reduce the yield in a long run.

  7. 7.

    Nutrient-depleting and nutrient-accumulating crops should be in the cropping system, thereby making the system able for recuperating the fertility level.

  8. 8.

    Adoption of precision agriculture, site-specific nutrient management, and drip fertigation will help to improve/sustain the soil fertility level (Jayakumar et al. 2014; 2015).

  9. 9.

    Practice of integrated nutrient management comprising integrated usage of chemical fertilizers and other source of organic manures such as biofertilizers will result in sustainable crop yields without any detrimental effect on agro-ecological balance (Surendran and Vani 2013; Surendran et al. 2016).

Conclusions

The calculated nutrient balances for Coimbatore and Erode Districts in a semi-arid region of South India, using NUTMON methodology, were negative for nitrogen (N −3.3 and −10.1 kg ha−1) and potassium (K −58.6 and −9.8 kg ha−1) and positive for phosphorus (P +14.5 and 20.5 kg ha−1). Soil nutrient pool has to offset the negative balance of N and K; there will be an expected mining of nutrient from the soil reserve in the study area. DSSIFER-based fertilizer recommendation turns the negative balances of N and K in to a positive one at crop, farm, and district levels. These DSSs, viz., NUTMON and DSSIFER, serve as a tool to identify the depletion of nutrients and help to suggest the management options using a systematic approach. The management options to mitigate nutrient mining with an integrated system approach are also discussed. However, it is concluded that one single technology does not solve these nutrient-related problems and solutions have to be sought in a suite of technologies through integrated nutrient management (INM)/best management practices (BMPs).