Estimating economic and environmental trade-offs of managing nitrogen in Australian sugarcane systems taking agronomic risk into account

Journal of Environmental Management 223 (2018) 264–274

Contents lists available at ScienceDirect

Journal of Environmental Management journal homepage: www.elsevier.com/locate/jenvman

Research article

Estimating economic and environmental trade-offs of managing nitrogen in Australian sugarcane systems taking agronomic risk into account

T

John Kandulua,∗, Peter Thorburnb, Jody Biggsb, Kirsten Verburgc a

CSIRO Land and Water, Waite Campus, Adelaide, Australia CSIRO Agriculture and Food, Brisbane, Australia c CSIRO Agriculture and Food, Canberra, Australia b

A R T I C LE I N FO

A B S T R A C T

Keywords: Coastal water quality Nitrogen use efficiency Marine ecosystems Farm profitability

Use of chemical agricultural inputs such as nitrogen fertilisers (N) in agricultural production can cause diffuse source pollution thereby degrading the health of coastal and marine ecosystems in coastal river catchments. Previous reviewed economic assessments of N management in agricultural production seldom consider broader environmental impacts and uncertain climatic and economic conditions. This paper presents an economic risk framework for assessing economic and environmental trade-offs of N management strategies taking into account variable climatic and economic conditions. The framework is underpinned by a modelling platform that integrates Agricultural Production System sIMulation modelling (APSIM), probability theory, Monte Carlo simulation, and financial risk analysis techniques. We applied the framework to a case study in Tully, a coastal catchment in north-eastern Australia with a well-documented N pollution problem. Our results show that switching from managing N to maximise private net returns to maximising social net returns could reduce expected private net returns by $99 ha−1, but yield additional environmental benefits equal to $191 ha−1. Further, switching from managing N to maximise private returns in years with the highest profit potential (hereafter, good years) to maximising mean social net returns could reduce expected private profits in good years by $277 ha−1, but yield additional environmental benefits equal to $287 ha−1. We contend that it is essential to incorporate farmer risk behaviour and environmental impacts in analyses that inform policies aimed at enhancing adoption of management activities for mitigating deterioration of the health of coastal and marine ecosystems due to diffuse source pollution from agricultural production.

1. Introduction Agricultural production in coastal river catchments has been identified as an important contributor to diffuse source pollution degrading the health of coastal and marine ecosystems (Howarth, 2008; Rabalais et al., 2009). Increasing nitrogen fertiliser application rates (hereafter, N rates) is often associated with higher yields and profits; however, high N rates can result in losses of N to the environment through runoff, deep drainage, volatilisation and denitrification (Canfield et al., 2010; Harmel et al., 2008; Schlesinger, 2009; Thorburn and Wilkinson, 2013). N losses, in particular in the form of dissolved inorganic nitrogen (DIN), can cause problems such as eutrophication, habitat degradation and loss of biodiversity in affected coastal marine ecosystems (Howarth, 2008; Rabalais et al., 2009). In addition, N loss from soils in the form of nitrous oxide (N2O), a potent greenhouse gas, contributes to global warming (Thorburn et al., 2010). Management of N in agricultural production is necessary to mitigate environmental impacts from loss of

∗

Corresponding author. E-mail address: [email protected] (J. Kandulu).

https://doi.org/10.1016/j.jenvman.2018.06.023 Received 31 August 2017; Received in revised form 30 May 2018; Accepted 9 June 2018 0301-4797/ © 2018 Elsevier Ltd. All rights reserved.

N however, consideration of effects of N management on profitability of agricultural enterprises ensures adequate adoption of N management activities (Roebeling et al., 2009). Assessments of alternative N management activities need to take into account trade-offs between competing environmental and economic objectives (van Grieken et al., 2013a). However, economic assessments of N management in agricultural systems, typically assess the impact of applying various N rates on profitability of agricultural enterprises without taking environmental impacts into account (Brennan et al., 2007; Rajsic and Weersink, 2008). Ignoring environmental costs can lead to the application of a higher private economically optimum N rate than the socially optimum N rate that takes environmental costs into account (termed, externalities). Few studies have incorporated environmental costs in assessments of N management activities using measures of central tendency including long-term mean and median cost and benefit values to identify and compare long-term average private and social optimum N rates

Journal of Environmental Management 223 (2018) 264–274

J. Kandulu et al.

Wilkinson, 2013). In addition, N applied to sugarcane is linked with substantive emissions of nitrous oxide (N2O) from soils (Thorburn et al., 2010). The Australian and Queensland governments have, since 2003, implemented policies and targets to reduce exports of pollutants, including N, to the Great Barrier Reef (Kroon et al., 2016; RWQPP, 2009). The sugarcane industry in Queensland has committed to reducing N loss through the adoption of better soil management, use of climate forecasts, legume fallow crops, and N replacement fertiliser management (Schroeder et al., 2010; Thorburn et al., 2011b). The Tully region case study assessment addresses a growing interest by policymakers and agriculturalists to better understand the economic benefits and costs of alternative N management activities.

(Brentrup et al., 2004; Tegtmeier and Duffy, 2004; Van Grinsven et al., 2013; Von Blottnitz et al., 2006). The reviewed studies did not quantify the combined risk from variable climatic and economic factors. However, N application decisions and environmental N losses, are largely influenced by variable climatic and economic conditions (Gandorfer et al., 2011; Monjardino et al., 2013; Rajsic and Weersink, 2008; Sadras, 2002; Sheriff, 2005; Yu et al., 2008). Most agriculturists engaged in production of high-value crops typically apply a higher rate of N than the long-term average economic optimum rate to realise high profits under favourable climatic and economic conditions and to minimise economic losses in years with the lowest profit potential (henceforth, bad years) (Gandorfer et al., 2011; Shillito et al., 2009). Economic assessments that seek to identify long-term average N rates for maximising average private and social returns under expected conditions have limited use and application in contexts where agriculturalists' objectives are to maximise on large profits in years with the highest profit potential, good years, and to minimize the risk of big losses in bad years. This paper presents a framework for assessing the economic and environmental impacts of N management strategies taking into account uncertainty in both climatic and economic conditions. The assessment framework is underpinned by a modelling approach that integrates agricultural production system simulation and financial risk assessment measures of Conditional Value at Risk (CVaR) for the expected return of the lowest and highest possible outcomes with a cumulative probability of five percent (termed, CVaR0.05 and CVaR0.95). CVaR has been applied to assess the risk-mitigating benefit from diversification agricultural enterprises (Kandulu et al., 2012) and the risk mitigating benefit from increasing N rates above the regional optimum (Monjardino et al., 2013). Here we apply the framework to assess economic and environmental impacts of N management in sugarcane production in a sugarcane growing region (Tully) on the Wet Tropical coast of Australia, adjacent to the Barrier Reef World Heritage Area. We use our results to quantify the benefit of adequately incorporating environmental costs and agriculturalists' risk-mitigating behaviour in N management policy decisions for mitigating deterioration of the health of coastal and marine ecosystems due to diffuse source pollution from agricultural production.

3. Methods and data Our methodology involved six distinct steps: 1) developing a conceptual model for calculating net returns and environmental impacts under the three N application strategies, 2) modelling sugarcane yield responses to N application, 3) quantifying uncertainty in parameter values, 4) calculating net returns to sugarcane farmers with and without including environmental costs at six N rates, 5) Comparing the effect on net returns of changing N management strategies under three alternative N management objectives, and 6) systematic uncertainty analysis. 3.1. Developing a conceptual model for calculating net returns Net returns, NR, were calculated using partial budget analysis as the difference between farm revenues and the sum of fertilizer costs and harvesting costs (Fig. 2). To understand the incremental cost vs benefits variable costs of implementing the option, we carried out a gross margin analysis of alternative N management strategies taking into account costs that vary with varying N rate omitting fixed and overhead costs. For example, the fixed component of harvesting costs would not be expected to change with changes in N rates because the call-out fee for harvesters under current contractual arrangements is the same. Thus in a partial or marginal budget analysis, only the difference in the variable component of harvesting costs under different N rates, as influenced by differences in yields under the two N rates, are considered. Farm revenues were calculated as the product of yield and the market price of sugar taking into account: 1) cane payment formula (CPF) – a formula used by Queensland sugar industry to allocate net income from sugar sales between farmers and millers; and 2) cane sugar content (CCS) – calculated as the ratio of the weight of extractable sugar to the weight of one sugarcane at harvest (Di Bella et al., 2014). The costs included were: 1) the cost of N fertiliser based on N rate and unit cost of N fertilisers (assuming the cost of applying fertiliser was constant across all N rates); and 2) the cost of harvesting operations calculated as the product of the unit harvesting cost ($/tonne) charged by contractors to harvest sugarcane and yield per hectare. In addition, environmental costs were quantified and subtracted from farm revenues to calculate social net return based on 1) DIN loss calculations and the unit abatement cost for DIN discharged to coastal ecosystems; and 2) N2O emission calculations, converted to CO2 equivalent (CO2e), and the unit abatement costs for greenhouse gas N2O emitted from soils.

2. Case study context The Tully sugarcane growing region is among the major sugarcane growing regions in Australia's Wet Tropics. Stretching along Australia's north-eastern coastline in Queensland, the Wet Tropics sugar growing region is parallel to the Great Barrier Reef, a World Heritage Site and Australia's most visited tourist attraction containing extensive areas of coral reef, seagrass meadows and fisheries resources (Kroon et al., 2016) (Fig. 1). The broader Wet Tropics region covers an ecologically diverse World Heritage listed area covering 2.2 million hectares and encompassing vast wet tropical rainforests (Kroon et al., 2016). Sugarcane is cultivated as a monoculture in the broader Wet Tropics region with yields varying from year to year between 52 and 125 tonnes ha−1 in response to variable climate with annual rainfalls ranging from 2200 to over 6000 mm. Historical N rates applied by sugarcane growers in the Wet Tropics region range between 140 and 200 kg ha−1 (Thorburn and Wilkinson, 2013). Market prices for sugar and N fertiliser and other farm inputs also vary considerably. Discharges of dissolved inorganic N (DIN) and other pollutants from coastal catchments into the GBR ecosystem is causing a decline in the coral cover and seagrass meadows of the Great Barrier Reef ecosystem (Kroon et al., 2016). N fertiliser applications to sugarcane crops is a major source of DIN exports, and DIN discharges from catchments in the Wet Tropics pose the greatest risk to the health of the GBR (Waterhouse et al., 2012). Annual rates of N applied to sugarcane in the Wet Tropics are estimated at 100 kg ha−1 greater than the amount of N removed from farms in the form of harvested sugarcane (Thorburn and

3.2. Modelling crop yields The Agricultural Production sIMulator (APSIM) was used to simulate annual sugarcane yields under six N rates between 30 kg N ha−1 and 180 kg N ha−1 in increments of 30 kg N ha−1 over a period of 108 years between 1902 and 2010 (Thorburn et al., 2011a). The sugarcane growth simulation model operates on a daily time step and simulates yields driven by variability in climate, N inputs, soil-water balances and nitrogen balances across the 108 simulated years based on historical 265

Journal of Environmental Management 223 (2018) 264–274

J. Kandulu et al.

Fig. 1. Areas of sugarcane production in the Tully sugarcane region and the Great Barrier Reef.

Fig. 2. Organizational structure used to estimate farm revenues and costs and environmental costs under various N application strategies and optimization objectives. 266

Journal of Environmental Management 223 (2018) 264–274

J. Kandulu et al.

Fig. 3. Fitted yield probability distributions (red line) to frequency distributions from APSIM simulated annual sugarcane yields under each of the six N rates between 30 and 180 kg N ha−1 over a period of 108 years between 1902 and 2010. (For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.)

details on APSIM-sugarcane and other applications of this model, readers are referred to Thorburn et al. (2011a). Climate variability is here defined as historical observations of climate variables that influenced agricultural yields in particular temperature and rainfall without isolating the effect of climate change.

records. Several tests against experimental data has validated the capability of the APSIM-Sugarcane model to simulate yield responses to N (Keating et al., 1999; Skocaj et al., 2013; Thorburn et al., 2011a, 2017), N2O emission (Thorburn et al., 2010) and leaching (Thorburn et al., 2011a). The modelling timeframe represents the longest time period for which historical observational climate data records are available for parameterising the APSIM model. The 108 years were based on using the full available meteorological dataset. This was necessary to capture the full range of climatic conditions, in particular, rainfall variability, which impacts strongly on N loss and yield. While it is not excluded that there may be some climate trends over this period these are expected to be small and of secondary importance relative to having a sufficiently large dataset to capture the climate variability itself. For full technical

3.3. Quantifying yield variability We quantified climate-induced variability in yields by fitting probability density functions to frequency distributions of simulated annual yields over 108 years for each of the six N rates between 30 and 180 kg N ha−1 (e.g. Fig. 3). Variability in economic factors including the price of sugar and N fertiliser was also quantified by fitting probability density functions to time-series data on global sugar prices and N 267

Journal of Environmental Management 223 (2018) 264–274

J. Kandulu et al.

Table 1 Notation descriptions and values for NRi calculations (see Equations (1) and (2)). Notation

Description

Unit

Value (range)

Source(s)

CN

Cost of urea fertilizer

$/tonne

103–570

CH D

Harvesting cost paid to contractors The proportion of N lost from fields that is discharged at the end of the catchment Nitrogen abatement cost in the Wet Tropics catchment Price of CO2 emission reduction Market price of sugar

$/tonne

6–7.5 0.07–0.22

(WB, 2015) http://www.indexmundi.com/commodities/?commodity=sugar& months=360¤cy=aud (Mallawaarachchi and Quiggin, 2001; Di Bella et al., 2014) (Thorburn and Wilkinson, 2013)

($/kg) ($/tonne) ($/tonne)

0.91–34 10.23–13.95 213–498

Percentage

10–15

(van Grieken et al., 2013b; Rolfe and Windle, 2016) http://www.cleanenergyregulator.gov.au/ERF/Auctions-results/april-2016 http://www.indexmundi.com/commodities/?commodity=sugar&months= 360¤cy=audWB, 2016 http://www.indexmundi.com/commodities/? commodity=sugar&months=360¤cy=aud (FAO, 2015)

0.03–0.05

(Thorburn et al., 2010)

CA PC PS

CCS

E

Sugar content of sugar cane at harvest calculated as the ratio of the weight of extractable sugar to the weight of one sugarcane The proportion of N applied lost as N2O emissions from soils in CO2 equivalent (CO2e)

The large variation in N abatement cost estimates across reviewed studies reflects differences in the modelled effectiveness of various abatement strategies across catchments. Specifically, biophysical models reviewed provide a wide range of effective N reductions from managing losses at the farm boundary compared with abatement efforts at the end of the catchment. Further, N transmission losses and the extent to which abatement of losses from farms in close proximity to the Great Barrier Reef is more or less cost-effective than prioritising abatement of losses from remote farms (Rolfe and Windle, 2016; Sutton et al., 2011; van Grieken et al., 2013b). In the absence of biophysical modelling specific to our case study, we collated estimates of the cost of abating DIN loss to the Great Barrier Reef from bioeconomic modelling and water quality auction schemes. We used the full range of abatement cost values from multiple studies in the region to enable a comprehensive test on the importance of variable N abatement cost at influencing the robustness of our results and conclusions. Our N abatement cost estimate is consistent with (Compton et al., 2011)who estimated N abatement cost at between $8 and $15 per kg N. E is the proportion of N applied lost as N2O emissions from soils in CO2 equivalent (CO2e) and based on Thorburn et al. (2010). PC is the unit cost of CO2 based on carbon market prices for emission reductions reported by the Australian Government's Clean Energy Regulator through the Emission Reduction Fund program (http://www. cleanenergyregulator.gov.au/ERF/Auctions-results/april-2016). Carbon price is considered a proxy for the unit amount the public is willing to pay to reduce greenhouse gas emissions. Timescale differences in all cost and price parameters were however standardised by adjusting for inflation to ensure variability was not primarily due to inflation. Specifically, data on sugar prices, CO2 emission reduction prices fertiliser costs and harvesting costs are reported in 2016 Australian dollars and adjusted for inflation using CPI data published by the Reserve Bank of Australia (http://www.rba.gov.au/calculator/ annualDecimal.html). We calculated pair-wise Pearson correlation coefficients to quantify correlations between sugar price and sugar cane yields and between the cost of fertilizer and sugar price using historic time-series data. Correlation coefficients between sugar price and sugar cane yields were calculated as −0.29 and the correlation coefficient between the cost of fertilizer and sugar price was calculated as 0.28. Stochastic Monte Carlo simulations were then carried out to draw random parameter values from probability density functions of each of uncertain parameter and used in equations (1)–(3) to calculate 1000 probable net return values under various N rates and generate frequency distributions for NRi taking into account the correlation between sugar prices and sugar cane yields and between fertilizer cost and sugar price. In addition to calculating average net returns, we applied the financial concept of conditional value at risk (CVaR) to calculate the average expected return of

fertiliser (i.e. urea) costs over a period of 20 years from 1995 to 2015 (WB, 2015). Probability density functions of various forms including beta-general, PERT, and triangular distributions were fitted to quantify variability in yields, N losses, and N2O emissions. The probability density function with the best fit was preferred based on chi-squared goodness-of-fit statistical test results. Variability in DIN loss, N2O emissions, the unit cost of abatement of DIN loss and the unit cost of CO2 abatement was also quantified using probability density functions defined using ranges of values from published literature. 3.4. Calculating net returns In what follows, we describe how cost and revenue components for calculating net returns, including farm revenues, fertilizer costs, harvesting costs and environmental costs including DIN abatement and N2O emission costs were calculated under the three N management strategies at various N rates using mathematical equations. Parameter notation descriptions, value ranges used and units and sources used in mathematical equations are summarized in Table 1 and variability in parameter values are quantified using probability distribution functions fitted to the value ranges. Net returns, NR, under each of the six N rates, i, were calculated as:

NRi = (Yi × CPF ) − (CN × Ni ) − (CH × Yi ) − δ (Ni × D × CA) − γ (Ni × E × PC ), δ∈ Δ{0,1}; γ ∈ ϒ {0.1} where Yi is APSIM simulated yield (tonnes of cane per hectare), CPF is the cane payment formula used to allocate net income from sugar sales between farmers and millers calculated as:

CPF = (PS × 9 × 10−3 × (CCS − 4)) + 6.62 × 10−1 Ps is the market price of sugar and CCS is the sugar content in sugarcane at harvest. CN is the cost of N ($tonne−1) and Ni is the amount nitrogen applied. Ps and CN values are based on time-series data on global sugar prices and urea costs between 1995 and 2015 (WB, 2015). CH is the cost of harvesting per tonne per hectare of cane. δ = 0 and ϒ = 0 in calculations of NR without including environmental costs (henceforth, private NR) and δ = 1 and ϒ = 1 in net return calculations that include environmental costs (henceforth, social NR). D is the proportion of N applied that is lost as DIN from fields through runoff and deep drainage and discharged into the GBR (Thorburn et al., 2011a). CA is the unit cost of abatement of DIN discharged at the end of the catchment based on estimates of the unit cost of reducing the amount of N discharged into the Great Barrier Reef through various effective land management activities including nutrient-, fallow-, and tillage management (van Grieken et al., 2013b). 268

Journal of Environmental Management 223 (2018) 264–274

J. Kandulu et al.

additional N realised at N rates higher than 60 kg N ha−1 (Table 2 and Fig. 5a). The N rate at which private net returns are maximized is higher than the N rate with highest social net returns. Specifically, private net returns are maximised at 120 kg N ha−1 being $1224 ha−1 and social net returns are highest at 60 kg N ha−1 being $934 ha−1. Environmental costs consistently increase with increasing N rates due to high expected loss of excess N not utilised by crops to the environment being $96 ha−1 at 30 kg N ha−1 and $575 ha−1 at 180 kg N ha−1. The private CVaR0.95, the weighted average of expected economic returns in good years represented by highest attainable net returns with a cumulative probability of five percent, without including environmental costs, is highest at 150 kg N ha−1 being $2, 748 ha−1 (Table 2 and Fig. 5b). The private CvaR0.05, the weighted average of expected economic returns in bad years represented by lowest net returns with a cumulative probability of five percent, not taking environmental costs into account, is highest at 90 kg N ha−1 being $262 ha−1 (Table 2 and Fig. 5c).

the highest net returns in good years representing highest profit potential years with favourable climatic and economic conditions with a cumulative probability of five percent, CvaR0.95, and bad years with expected extreme losses under unfavourable conditions, CvaR0.05. For full technical details on fitting PDFs to time-series data and calculating correlation coefficients, readers are referred to Kandulu et al. (2012). 3.5. Calculating the impact of switching to the maximum social net return N rate We compared economic and environmental costs and benefits of switching from N rates under each of three private optimization objectives based on grower risk profiles to the maximum social net return N rate. In the absence of empirical information on the prevalence of grower risk profiles, three grower risk profiles were considered: 1) riskaverse growers minimizing expected extreme losses by seeking to maximise private CVaR0.05, 2) risk-neutral grower seeking to maximise average private net returns, and 3) risk-taking grower with the objective of maximising private CvaR0.95. The maximum social net return N rate takes DIN loss and N2O emissions into account.

4.3. Comparing the effect of changing N management strategies on net returns

3.6. Systematic uncertainty analysis In Table 3, we compare economic and environmental costs and benefits under the highest socially N rate (60 kg N ha−1 at $934 ha−1) with three N rates under each of the three private optimization objectives based on grower risk profiles. Specifically, we considered three private strategies: 1) a risk-averse grower applying an N rate that would reduce expected extreme losses and seeks to maximise private CVaR0.05 (90 kg N ha−1 at $262 ha−1); 2) a risk-neutral grower seeking to maximise average private economic returns (120 kg N ha−1 at $1, 224 ha−1); and 3) a risk-seeking grower with the objective of maximising private CvaR0.95 (150 kg N ha−1 at $2, 748 ha−1). Results show that the environmental benefit of switching to the maximum social net return N rate outweighs the economic cost of reduced CvaR0.05, mean returns, and CvaR0.95 under the three private optimization objectives. Switching from an N rate that would maximise private CvaR0.05 (90 kg N ha−1) to one that would maximise mean social economic returns (60 kg N ha−1) would increase expected extreme losses, private CvaR0.05, by $9 ha−1, but reduce environmental costs by $95 ha−1 (Table 3). Switching from an N rate that would maximise average private economic returns (120 kg N ha−1) to one that would maximise mean social economic returns (60 kg N ha−1) would reduce private net returns by $99 ha−1, but also reduce environmental costs by $191 ha−1. Switching from an N rate that would maximise private CvaR0.95 (150 kg N ha−1) to one that would maximise mean social economic returns (60 kg N ha−1) would reduce private CvaR0.95 by $277 ha−1, but reduce environmental costs by $287 ha−1. Overall, a comparison of measures of deterministic central tendency (mean and median) and spread in particular probable low returns in bad years (minimum and CVaR0.05) in good years (maximum and CVaR0.95) and systematic uncertainty analysis which shows consistency in conclusions drawn from expected changes in net returns of shifting between different N application strategies (Tables 2 and 3, and Figs. 4 and 6).

A stochastic BCA model using Monte Carlo simulation was run as a basis for systematic uncertainty analysis. The analysis assessed the contribution of each parameter to variations in net benefit calculations by varying each parameter individually about its range of probable values and calculating probable net benefit value ranges while holding all other parameters at their median values. Use of a stochastic economic model and Monte Carlo simulation to generate probability distributions of net returns enabled assessment of the relative contribution of each uncertain parameter in equations (1) and (2) to variability in estimates of net returns (termed, systematic uncertainty analysis). Specifically, the sensitivity of net return estimates to variability in each uncertain parameter was analysed by measuring the effect of varying each parameter while holding all other parameters at their median values following Kandulu and Connor (2016). 4. Results 4.1. Climate-induced variability in yield, N loss, and N2O emissions Overall, simulated sugarcane yields ranged from 6.4 to 125 tonnes ha−1 per year across each of the six N rates considered (Fig. 3). Median yields ranged between 47 tonnes ha−1 at 30 kg N ha−1 and 63 tonnes ha−1 at 150 and 180 kg N ha−1. There are consistent diminishing yield gains from increased N application with the highest increment in median yield occurring between 30 and 60 kg N ha−1 (11.8 tonnes ha−1). The additional yield gain between 60 and 90 kg N ha−1 is 2.3 tonnes ha−1, from 90 to 120 kg N ha−1 is 0.6 tonnes ha−1, 120 and 150 kg N ha−1 is 0.5 tonnes ha−1 and between 150 and 180 kg N ha−1 0.1 tonnes ha−1. 4.2. Net return estimates Expected net returns are considerably lower when environmental costs are included than when environmental costs are not taken into account (Fig. 4). The difference between including environmental costs and ignoring environmental costs is more appreciable at N rates between 120 and 180 kg N ha−1 than at lower N rates between 30 and 90 kg N ha−1 due to high expected loss of excess N not utilised by crops to the environment. Diminishing marginal net returns from increased N rates can be observed with the highest gain in mean net private and social returns occurring between 30 and 60 kg N ha−1 and smaller gains from

4.4. Systematic uncertainty analysis Net return estimates were substantially sensitive to variability in sugar prices, yield response to N, estimates of cane sugar content unit N abatement costs and unit carbon prices. However, our positive net return results and final conclusions remain robust (Fig. 6). Further, whilst variable N abatement costs led to variable environmental cost estimates, this did not affect final conclusions. Additionally, the optimization of the long-term average maximum social net return N rate was not affected as this was based on the median value. 269

Journal of Environmental Management 223 (2018) 264–274

J. Kandulu et al.

Fig. 4. Frequency distributions of net returns with (red line) and without including environmental costs from 1000 stochastic Monte Carlo simulations that drew random parameter values from probability density functions of each uncertain parameter under each of the six N rates between 30 and 180 kg N ha−1. (For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.)

Uncertain sugar prices contributed most significantly to variability in net returns with estimates of net returns ranging between $278 and $1597 ha−1. Climate-induced variability in yield responses to N also contributed substantially to variability in estimates of net returns with net social return estimates attributable to varying yields ranging between $273 and $1504 ha−1. Estimates of net returns were not

substantially sensitive to unit fertiliser and harvesting costs, DIN loss and N2O emissions. 5. Discussion Using historical climate data, we have estimated economic and

Table 2 Estimates of average net returns, NR, expected returns in good years, CVaR0.95, and expected returns in bad years CVaR0.05, with and without accounting for environmental impacts. (Note that the figures in bold indicate the highest NR ($/ha) under the optimal N rate (kg/ha) for each scenario). N app rate (kg/ha)

Social NR ($ha−1)

Private NR ($ha−1)

Environmental cost ($ha−1)

Social CVaR0.95 ($ha−1)

Private CVaR0.95 ($ha−1)

Social CVaR0.05 ($ha−1)

Private CVaR0.05 ($ha−1)

30 60 90 120 150 180

861 934 916 840 720 595

957 1125 1203 1224 1200 1170

96 192 287 383 479 575

2006 2255 2453 2267 2251 2155

2110 2471 2722 2615 2748 2649

127 54 −70 −224 −357 −512

237 261 262 222 194 149

270

Journal of Environmental Management 223 (2018) 264–274

J. Kandulu et al.

economic net returns and social impacts taking climate-induced yield variability, uncertain economic conditions, and environmental impacts into account. Indirect economic benefits to local, regional and national economies through increases in employment and GDP and livelihood impacts and equity and distribution of costs and benefits between beneficiaries and sectors that incur costs were outside the scope of this analysis. Our results show that switching from private benefit maximising N rate to applying the maximum social net return N rate reduces expected private benefit, but yields substantially greater environmental benefits. Our estimates of net returns were most sensitive to uncertain economic conditions, in particular, sugar prices, and climate variability and its influence on yield response to N application. This is consistent with findings by other studies (Brennan et al., 2007; Oliver and Robertson, 2009; Sadras, 2002). Our results confirm that use of deterministic longterm economic optima alone based on long-term average values in N application decisions would not adequately capture variability and the risk mitigating benefits of supra-optimal N rates (above the long-term economically optimum level). This is consistent with findings by Rajsic and Weersink (2008) who found that use of measures of central tendency such as mean or indeed any single value to represent the economically optimum rate does not account for uncertainty. The economic optimum rate based on the long-term mean value (120 kg ha−1) is significantly less than the optimum rate when risk is taken into account and expected returns in good years are optimized, estimated at approximately 150 kg ha−1. Meaningful interpretation of the optimum N rate based on long-term average net returns, however, is thus challenging in a context of highly variable yield responses to N application due to variable climate, volatile farm input costs including cost of N fertilisers, and variable sugar prices. In a context of incomplete determinism about the precise value of uncertain parameters, the type of uncertainty in our model was measurable uncertainty because it was largely due to inherent temporal variability in values over time due to natural and induced factors. Thus we had considerable confidence that the variability in parameter values was adequately characterized by assigning probability distributions around known parameters value ranges based on observed and validated simulated time-series data. Where time-series of observed or simulated data was not available, a systematic review of the published peer-reviewed literature was carried out to obtain value ranges for some parameters. We incorporate the assumption that agriculturists in highly variable climatic and economic conditions can seek high potential returns in good years and look to avoid low returns in bad years based on empirical evidence and reviewed literature. Our assessment framework provides a more nuanced basis for developing effective policies than previous studies which typically assume that agriculturists apply N rates that maximise long-term average economic returns. Several reviewed studies show that the context within which agriculturists make N application decisions is typically characterised by uncertain climate and economic conditions. Historically, N rates applied by sugarcane growers in the case study region have ranged between 140 and 200 kg ha−1 (Thorburn and Wilkinson, 2013). Yield response based on APSIM results suggest that the yield effect for N rates exceeding 90 kg ha−1 N is very small even

Fig. 5. Mean net returns (a), CVaR0.95 (b) and CVaR0.05 (c) with and without environmental costs calculated from frequency distributions of net returns (Fig. 4) under each of the six N rates between 30 and 180 kg N ha-1.

environmental impacts of nitrogen on sugarcane systems. Using results of this study to inform adaptive responses to future climate will require additional modelling to obtain a detailed understanding of the effects of climate change and its association with future climate variability. A framework for assessing economic and monetised environmental impacts of alternative N management strategies representing alternative N management optimisation objectives based on farmer risk profiles is presented in this paper. Specifically, the framework was applied to assess trade-offs between competing economic and environmental optimization objectives including identifying the highest private

Table 3 Changes in average NR of switching from N rates representing maximum private CVaR0.95 (M5), maximum private NR (MM), and maximum private CVaR0.95 (M95), to maximum social NR (MS). Scenario description

Δ social ($ha−1)

Δ private NR ($ha−1)

Δ Environmental cost ($ha−1)

Δ social CVaR0.95 ($ha−1)

Δ private CVaR0.95 ($ha−1)

Δ social CVaR0.05 ($ha−1)

Δ private CVaR0.05 ($ha−1)

From M5 to MS From MM to MS From M95 to MS

18 94 214

−78 −99 −75

−95 −191 −287

−198 −12 4

−251 −144 −277

124 278 411

9 39 67

271

Journal of Environmental Management 223 (2018) 264–274

J. Kandulu et al.

Fig. 6. A tornado chart showing results on measures of sensitivity of net return estimates to variability in each uncertain parameter holding all other uncertain parameters at their median values for 30 kg N ha−1 with environmental costs.

expected to be lower than the environmental benefit of avoided reef degradation due to reduced N rates (van Grieken et al., 2013a). Our study quantified only two main environmental impacts including GBR degradation by DIN runoff and deep drainage and climate regulation from N2O losses. One limitation of this study is poorly understood underpinning biophysical process models linking N application rates and expected DIN loss to the environment. For example, the difference in timescales between changes in N application rates and responses in terms of yields and DIN losses to receiving aquatic ecosystems in the case study area is poorly understood. In addition, assuming a linear relationship between N application rates and DIN loss may be a gross oversimplification. However, Costanza et al. (2014) argue that there is a degree of uncertainty in most estimated values associated with ecological processes. However, the extent to which uncertainty matters depends on how estimated values will be used. Specifically, less precision may be required if the main aim of an assessment is merely to raise awareness however, more precision may be required to inform the design of optimal incentive payments to growers for improving the stock and quality of ecosystem service. Uncertainties in environmental impact value estimates could be addressed by investing in research towards understanding biophysical dose-response process models. Further, the impact of other forms of N loss including volatilisation and denitrification were not quantified due to lack of underpinning biophysical process and response models however, these impacts have been found to be important in Australia and internationally (Canfield et al., 2010; Davidson et al., 2011; Harmel et al., 2008; Schlesinger, 2009; Sutton et al., 2011; Thorburn and Wilkinson, 2013). Taking these environmental impacts into account would further strengthen the conclusions of our study as well as the case for policy intervention to incentivise efficient N application. Use of enhanced efficiency N fertilisers has been considered as a possible long-run solution for improving economic and environmental outcomes however, studies provide contrasting conclusions on the economics of switching to enhanced efficiency fertilisers (Arrobas et al., 2011; Di Bella et al., 2014; Farmaha and Sims, 2013; Khakbazan et al., 2013; Verburg et al., 2014). However, in most of these studies, environmental impacts of switching to enhanced efficiency N fertilisers and climate-induced yield variability are not quantified. One exception is Zhang et al. (2015) who included environmental impacts but did not model climate-induced yield variability. The only study to our knowledge to model the effects of climate variability on yield is (Sheriff, 2005). Our method can be modified and used to assess tradeoffs between economic and environmental impacts of switching from urea, the

without including environmental impacts. Whilst historical and current application rates are larger than our calculation of the private and social long-term average optima, they are consistent with our finding for the N rate that would maximize the weighted average of expected economic returns in good years estimated at an expected value of 150 kg N ha−1. Further, large variability in N rates could reflect variability in climate and in fertilizer costs nonetheless our finding falls within the range of N rates for optimising expected returns in good years. This reveals that faced with climate uncertainty, sugarcane growers in our case study region typically apply N at rates that will maximise returns in good years representing highest profit potential. Given relative low fertilizer cost as a percentage of expected returns from what is a high value per hectare enterprise, applying high N rates presents a low-risk proposition in the face of high-profit potential (Gandorfer et al., 2011; Shillito et al., 2009). Further, N application decisions in the case study region are to a lesser extent driven by the objective to maximize long-term average returns or to minimize large losses in bad years. The model presented in this paper can provide recommendations in N rate based on historical data on climate and prices. However, farmers typically make decisions at the start of the season with no prior knowledge of future seasonal climatic and economic conditions. Our method can be adapted and applied to calculate expected future net returns using climate and price forecast data. Improvements in seasonal climate forecasting coupled with risk-based assessments of expected returns could enhance efficient N application, increase profitability and reduce environmental costs (Skocaj et al., 2013; Thorburn et al., 2011c). The Australian government is currently developing water quality targets and policies for managing the amount of N discharged to the Great Barrier Reef to restore the health of ecosystems in the reef. Our assessment is relevant and provides results that can be useful for informing policy decisions. For example, our calculations of values of foregone private profits from reducing N rates to meet socially optimal objectives can provide a basis for a financial incentive program by providing a reference point for the amount of money sugarcane growers would be willing to accept to reduce N rates. Specifically, one effective way to achieve a desired N loss reduction target could be through provision of financial incentives in the form of compensation payment schemes offered to growers to recover the cost of implementing lower N rates taking into account the revelation that growers in the case study area apply to maximize returns in good years with high-profit potential. A beneficiary pays cost-sharing arrangement would be economically feasible where the overall cost of incentivising reductions in N rates is 272

Journal of Environmental Management 223 (2018) 264–274

J. Kandulu et al.

most common N fertiliser, to alternative enhanced efficiency N fertilisers taking climatic and economic uncertainty into account. Further, our assessment framework, with some modification, can also be applied to assessments of optimal N rates for various chemical agricultural inputs that can lead to diffuse source pollution in catchments with water resources that serve multiple competing economic and environmental functions including sediments, phosphorous, herbicides, fungicides, and insecticides if relevant underpinning biophysical dose-response process models are available.

Keating, B., Robertson, M., Muchow, R., Huth, N., 1999. Modelling sugarcane production systems: development and performance of the sugarcane module. Field Crop. Res. 61, 253–271. Khakbazan, M., Grant, C.A., Finlay, G., Wu, R., Malhi, S.S., Selles, F., Clayton, G.W., Lupwayi, N.Z., Soon, Y.K., Harker, K.N., 2013. An economic study of controlled release urea and split applications of nitrogen as compared with non-coated urea under conventional and reduced tillage management. Can. J. Plant Sci. 93, 523–534. Kroon, F.J., Thorburn, P., Schaffelke, B., Whitten, S., 2016. Towards protecting the Great Barrier Reef from land-based pollution. Global Change Biol. 22, 1985–2002. Mallawaarachchi, T., Quiggin, J., 2001. Modelling socially optimal land allocations for sugar cane growing in North Queensland: a linked mathematical programming and choice modelling study. Aust. J. Agric. Resour. Econ. 45, 383–409. Monjardino, M., McBeath, T.M., Brennan, L., Llewellyn, R.S., 2013. Are farmers in lowrainfall cropping regions under-fertilising with nitrogen? A risk analysis. Agric. Syst. 116, 37–51. Oliver, Y.M., Robertson, M.J., 2009. Quantifying the benefits of accounting for yield potential in spatially and seasonally responsive nutrient management in a Mediterranean climate. Aust. J. Soil Res. 47, 114–126. Rabalais, N.N., Turner, R.E., Diaz, R.J., Justic, D., 2009. Global change and eutrophication of coastal waters. ICES J. Mar. Sci. 66, 1528–1537. Rajsic, P., Weersink, A., 2008. Do farmers waste fertilizer? A comparison of ex post optimal nitrogen rates and ex ante recommendations by model, site and year. Agric. Syst. 97, 56–67. Roebeling, P.C., van Grieken, M.E., Webster, A.J., Biggs, J., Thorburn, P., 2009. Costeffective water quality improvement in linked terrestrial and marine ecosystems: a spatial environmental-economic modelling approach. Mar. Freshw. Res. 60, 1150–1158. Rolfe, J., Windle, J., 2016. Benchmarking Costs of Improving Agricultural Water Management in GBR Catchments. Report to the National Environmental Science Programme. Reef and Rainforest Research Centre Limited, Cairns 52pp. RWQPP, 2009. Reef Water Quality Protection Plan in. http://www.reefplan.qld.gov.au/ library/pdf/reef-plan-2009.pdf http://www.reefplan.qld.gov.au/library/pdf/reefplan-2009.pdf. Sadras, V., 2002. Interaction between rainfall and nitrogen fertilisation of wheat in environments prone to terminal drought: economic and environmental risk analysis. Field Crop. Res. 77, 201–215. Schlesinger, W.H., 2009. On the fate of anthropogenic nitrogen. P Natl. Acad. Sci. U.S.A. 106, 203–208. Schroeder, B., Hurney, A., Wood, A., Moody, P., 2010. Concepts and value of the nitrogen guidelines contained in the Australian sugar industry's ‘six easy steps’ nutrient management program. In: Proceedings International Society of Sugar Cane Technologists, vol. 27. pp. 11. Sheriff, G., 2005. Efficient Waste? Why farmers over-apply nutrient and the implications for policy design. Rev. Agric. Econ. 27, 542–557. Shillito, R.M., Timlin, D.J., Fleisher, D., Reddy, V.R., Quebedeaux, B., 2009. Yield response of potato to spatially patterned nitrogen application. Agric. Ecosyst. Environ. 129, 107–116. Skocaj, D., Everingham, Y.L., Schroeder, B., 2013. Nitrogen Management Guidelines for Sugarcane Production in Australia: Can These Be Modified for Wet Tropical Conditions Using Seasonal Climate Forecasting? Springer Science Reviews, pp. 1. Sutton, M.A., Howard, C.M., Erisman, J.W., Billen, G., Bleeker, A., Gernnfelt, P., van Grinsven, H., Grizzetti, B., 2011. The European Nitrogen Assessment: Sources, Effects and Policy Perspectives. Cambridge University Press. Tegtmeier, E.M., Duffy, M.D., 2004. External costs of agricultural production in the United States. Int. J. Agric. Sustain. 2, 1–20. Thorburn, P.J., Biggs, J.S., Attard, S.J., Kemei, J., 2011a. Environmental impacts of irrigated sugarcane production: nitrogen lost through runoff and leaching. Agric. Ecosyst. Environ. 144, 1–12. Thorburn, P.J., Biggs, J.S., Collins, K., Probert, M.E., 2010. Using the APSIM model to estimate nitrous oxide emissions from diverse Australian sugarcane production systems. Agric. Ecosyst. Environ. 136, 343–350. Thorburn, P.J., Biggs, J.S., Palmer, J., Meier, E.A., Verburg, K., Skocaj, D.M., 2017. Prioritizing crop management to increase nitrogen use efficiency in Australian sugarcane crops. Front. Plant Sci. 8. Thorburn, P.J., Biggs, J.S., Webster, A.J., Biggs, I.M., 2011b. An improved way to determine nitrogen fertiliser requirements of sugarcane crops to meet global environmental challenges. Plant Soil 339, 51–67. Thorburn, P.J., Jakku, E., Webster, A.J., Everingham, Y.L., 2011c. Agricultural decision support systems facilitating co-learning: a case study on environmental impacts of sugarcane production. Int. J. Agric. Sustain. 9, 322–333. Thorburn, P.J., Wilkinson, S.N., 2013. Conceptual frameworks for estimating the water quality benefits of improved agricultural management practices in large catchments. Agric. Ecosyst. Environ. 180, 192–209. van Grieken, M., Lynam, T., Coggan, A., Whitten, S., Kroon, F., 2013a. Cost effectiveness of design-based water quality improvement regulations in the Great barrier reef catchments. Agric. Ecosyst. Environ. 180, 157–165. van Grieken, M., Poggio, M., Smith, M., Taylor, S.S., Faure, C., Boullier, A., Whittle, S., 2013b. Cost-effectiveness of Management Activities for Water Quality Improvement in Sugarcane Farming. Report to the Reef Rescue Water Quality Research & Development Program. Reef and Rainforest Research Centre Limited, Cairns. Van Grinsven, H.J.M., Holland, M., Jacobsen, B.H., Klimont, Z., Sutton, M.A., Willems, W.J., 2013. Costs and benefits of nitrogen for europe and implications for mitigation. Environ. Sci. Technol. 47, 3571–3579. Verburg, K., Harvey, T.G., Muster, T.H., Brennan McKellar, L.E., Thorburn, P.J., Biggs, J.S., di Bella, L., Wang, W., 2014. Use of Enhanced Efficiency Fertilisers to Increase Fertiliser Nitrogen Use Efficiency in Sugarcane. Sugar Research Australia, Brisbane.

6. Conclusion Ignoring environmental impacts of diffuse source pollution from chemical agricultural inputs can lead to socially sub-optimal outcomes in catchments that service competing economic and environmental functions. Effective management of diffuse source pollution is necessary to mitigate environmental impacts from agricultural systems however, it is important to adequately consider effects on profitability of agricultural enterprises taking into account farmer risk behaviour. We contend that it is essential to incorporate farmer risk behaviour and environmental impacts in analyses that inform policies aimed at enhancing adoption of management activities for mitigating deterioration of the health of coastal and marine ecosystems due to diffuse source pollution from agricultural production. Acknowledgments The authors would like to acknowledge the funding provided by Sugar Research Australia (SRA) and CSIRO. The review provided by Mark Poggio and Stuart Whitten is also gratefully acknowledged. References Arrobas, M., Parada, M.J., Magalhaes, P., Rodrigues, M.A., 2011. Nitrogen-use efficiency and economic efficiency of slow-release N fertilisers applied to irrigated turfs in a Mediterranean environment. Nutrient Cycl. Agroecosyst. 89, 329–339. Brennan, L.E., Robertson, M.J., Dalgliesh, N.P., Brown, S., 2007. Pay-offs to zone management in a variable climate: an example of nitrogen fertiliser on wheat. Aust. J. Agric. Res. 58, 1046–1058. Brentrup, F., Kusters, J., Lammel, J., Barraclough, P., Kuhlmann, H., 2004. Environmental impact assessment of agricultural production systems using the life cycle assessment (LCA) methodology - II. The application to N fertilizer use in winter wheat production systems. Eur. J. Agron. 20, 265–279. Canfield, D.E., Glazer, A.N., Falkowski, P.G., 2010. The evolution and future of Earth's nitrogen cycle. Science 330, 192–196. Compton, J.E., Harrison, J.A., Dennis, R.L., Greaver, T.L., Hill, B.H., Jordan, S.J., Walker, H., Campbell, H.V., 2011. Ecosystem services altered by human changes in the nitrogen cycle: a new perspective for US decision making. Ecol. Lett. 14, 804–815. Costanza, R., de Groot, R., Sutton, P., van der Ploeg, S., Anderson, S.J., Kubiszewski, I., Farber, S., Turner, R.K., 2014. Changes in the global value of ecosystem services. Global Environ. Change 26, 152–158. Davidson, E.A., David, M.B., Galloway, J., Goodale, C.L., Haeuber, R., Harrison, J.A., Howarth, R.W., Jaynes, D.B., Lowrance, R.R., Thomas, N.B., Peel, J.L., 2011. Excess nitrogen in the US environment: trends, risks, and solutions. Issues Ecol. 15, 1–16. Di Bella, L., Stacey, S., Moody, P., Benson, A., Dowie, J., Sluggett, R., 2014. An assessment of controlled release fertilisers in the Australian sugar industry. In: Proceedings of the Australian Society of Sugar Cane Technology, pp. 121–131. FAO, 2015. Food and Agriculture Organisation of the United Nations. http://www.fao. org/es/faodef/fdef03e.HTM. Farmaha, B.S., Sims, A.L., 2013. The influence of polymer-coated urea and urea fertilizer mixtures on spring wheat protein concentrations and economic returns. Agron. J. 105, 1328–1334. Gandorfer, M., Pannell, D., Meyer-Aurich, A., 2011. Analyzing the effects of risk and uncertainty on optimal tillage and nitrogen fertilizer intensity for field crops in Germany. Agric. Syst. 104, 615–622. Harmel, D., Qian, S., Reckhow, K., Casebolt, P., 2008. The MANAGE database: nutrient load and site characteristic updates and runoff concentration data. J. Environ. Qual. 37, 2403–2406. Howarth, R.W., 2008. Coastal nitrogen pollution: a review of sources and trends globally and regionally. Harmful Algae 8, 14–20. Kandulu, J., Connor, J.D., 2016. Improving the effectiveness of aid: an evaluation of prospective Mekong irrigation investments. Int. J. Water Resour. Dev. Kandulu, J.M., Bryan, B.A., King, D., Connor, J.D., 2012. Mitigating economic risk from climate variability in rain-fed agriculture through enterprise mix diversification. Ecol. Econ. 79, 105–112.

273

Journal of Environmental Management 223 (2018) 264–274

J. Kandulu et al.

360¤cy=aud&commodity=urea. World Bank. Yu, Q., Wang, E.L., Smith, C.J., 2008. A modelling investigation into the economic and environmental values of 'perfect' climate forecasts for wheat production under contrasting rainfall conditions. Int. J. Climatol. 28, 255–266. Zhang, X., Mauzerall, D.L., Davidson, E.A., Kanter, D.R., Cai, R.H., 2015. The economic and environmental consequences of implementing nitrogen-efficient technologies and management practices in agriculture. J. Environ. Qual. 44, 312–324.

Von Blottnitz, H., Rabl, A., Boiadjiev, D., Taylor, T., Arnold, S., 2006. Damage costs of nitrogen fertilizer in Europe and their internalization. J. Environ. Plann. Manag. 49, 413–433. Waterhouse, J., Brodie, J., Lewis, S., Mitchell, A., 2012. Quantifying the sources of pollutants in the Great Barrier Reef catchments and the relative risk to reef ecosystems. Mar. Pollut. Bull. 65, 394–406. WB, 2015. http://www.indexmundi.com/commodities/?commodity=sugar&months=

274