Agricultural eco-efficiency is promoted as a means of increasing primary production and improving food security (Jansen, 2000). The dominant means for increased eco-efficiency is more intensive use of economic and environmental resources (Gregory et al., 2002). Wilkins (2008) notes that the two ways of achieving intensification include altering the management of individual crop and livestock (and fisheries) enterprises, and altering the land-use system itself. As used here, intensification refers to the integration of livestock and crop production to increase recycling of resources and minimize resource loss; and second, to the use of technologies that directly reduce dependence on nonrenewable resources or improve the efficiency with which they are utilized (e.g., precision agriculture). A third method of achieving agricultural intensification may focus on modifying the regional agricultural system to optimize production for current climatic conditions. In this context, the term agricultural intensification is not limited to production systems characterized by high inputs of capital or labor, or heavy reliance on technologies sourced externally to the system, such as pesticides and chemical fertilizers. Rather, it extends the definition to include increased production resulting from the spatial and temporal concentration of sustainably sourced resources from within the production system.
In all three of the above examples of intensification, the aim is to move the system toward the production efficiency frontier as discussed by Keating et al. (2010) However, increasing amounts of food production has failed to significantly reduce global levels of chronic undernutrition and the absolute number of hungry people over the past few decades (FAO, 2006), due to the lack of a simultaneous consideration of the availability (encompassing both the quantity and quality of food), accessibility, stability of food supplies, and their utilization (FAO, 2000). Collectively these four key dimensions of food security relate to the availability of sufficient quantities of food of appropriate quality, accessed by individuals at all times and utilized through adequate diet, clean water, sanitation, and health care to reach a state of nutritional well-being where all physiological needs are met (Garrett, 1995; FAO, 1996). When viewed in this way, nonfood inputs of an economic, environmental, and social nature are shown to be integral to food security.
Ongoing anthropogenic global warming may have both positive and negative impacts on all four aspects of food security (Schmidhuber and Tubiello, 2007). For example, enhanced concentrations of atmospheric CO2, higher temperatures, and changes in the patterns of precipitation suggest changes in potential yield (Jones and Thornton, 2003; Parry et al., 2004; Liu et al., 2008; Lobell et al., 2008). While some yield benefits may be predicted for the short term, projected increases in climatic variability and the volatility of extreme weather events, and resultant increases in the range of crop pests and diseases, may negate many gains and disrupt the stability of food supply, particularly in developing countries (Rosenzweig and Parry, 1994). Changes in the quality of production (Manderscheid et al., 1995; Kimball et al., 2001) and increased production and energy requirements postharvest (Gleadow et al., 2009) will further challenge the availability of a stable supply of food.
Maintaining the parity of consumer purchasing power offers potential to improve food security. However, the resultant complexity of the interactions between the socioeconomic and environmental drivers on the capacity to buy food is likely to result in geographically varying outcomes under climate change (Liu et al., 2008). As a further aspect of food security, accessing sufficient clean water to enable effective utilization of the nutritional value of food is likely to be further challenged in regions where current access to supplies of good-quality water are already poor and projections of precipitation suggest ongoing reductions (Stige et al., 2006). In addition, projections of an increase in the incidences of food- and water-borne diseases (IPCC, 2007) suggest further reductions in the potential for the nutritional value of food to be fully utilized by those presently most vulnerable. While this brief summary highlights the varying impacts of climate change on food security, it is generally considered that negative effects will dominate already food-insecure livelihoods in the low latitudes (Stern, 2007).
Despite the recognition that numerous social drivers underpin the capacity for food security, the inclusion of an assessment of social outcomes within the eco-efficiency concept has remained on the periphery for over a decade (Schmidheiny and The Business Council for Sustainable Development, 1992; Zoebl, 1996; Jansen, 2000), with little progress made to formalize it within a broader assessment of efficiency. The aim of this paper is therefore to provide a coherent qualitative and quantitative argument for the simultaneous consideration of economic, environmental, and social outcomes resulting from the implementation of technologies and practices. As one of the most significant challenges facing global society today, we use the context of developing and implementing sustainable adaptation response strategies to improve food security under a changing climate, to illustrate the imperative for an explicit triple-bottom-line approach to assessing efficiency. We do this qualitatively and quantitatively by first reviewing the current usage of the term eco-efficiency in the scientific literature domain to assess the extent to which the concept is being applied to the development of climate change adaptation strategies. We also assess the level of consideration given not only to economic and environmental criteria, but also to social aspects such as risk management. Second, we use a crop growth simulation model to examine the economic, environmental, and social outcomes resulting from a change in agricultural management practice. This exercise highlights the potential for conflicting and perverse economic, environmental, and social outcomes to be overlooked when only economic and environmental efficiency is considered.
MATERIALS AND METHODS
Quantitative Literature Analysis
A quantitative literature analysis (also referred to as a simple meta-analysis) was undertaken by identifying published literature relating to eco-efficient agricultural practices and food production activities up to 1 Apr. 2009 using the Internet-based scientific literature search engine Web of Knowledge (http://apps.isiknowledge.com [verified 26 Dec. 2009]). Keywords used in searching the database were divided into variations on the terms eco-efficiency and agricultural production (Table 1 ). Each of the variations relating to the term eco-efficiency was paired with an agricultural term to produce 32 search criteria. From these searches we identified 18 studies relating to eco-efficient agricultural and food production systems that were (i) based on a case study, that is, not a purely theoretical or conceptual consideration; and (ii) focused on primary production or food processing of either a crop, animal, or fish species.
|Eco-efficient terms||Agricultural production terms|
To determine if there was any statistical evidence that case studies with a climate change rationale tended to include consideration of economic, environmental, or social outcomes resulting from the application of a new technology or practice, the 18 research and conference papers found from the review of literature were subjected to a simple meta-analysis. This analysis was undertaken by assessing (i) the extent to which climate change was the objective of the study, and (ii) the extent to which analysis of economic, environmental, or social efficiency was detailed. As there was only one paper from the review that showed climate change to be the primary objective of the study, for the purpose of statistical analysis this was grouped together with papers citing climate change as a secondary or minor objective. A (nonparametric) two-sample Wilcoxon rank-sum test (Lehmann, 1975) was then performed against a one-sided alternative: that is, those papers without a climate change rationale (i.e., scored 0) were likely to include less assessment of economic, environmental, and social outcomes than those papers with a climate change rationale. This is equivalent to the Mann–Whitney test.
Wheat Simulation Experiment
To consider possible tradeoffs between economic, environmental, and social outcomes, the Agricultural Production Systems Simulator (APSIM) (v. 3.6) (Keating et al., 2003) was used to simulate the growth of wheat (Triticum aestivum L.) and three key indicators of environmental performance, namely the amount of N leached below the root zone, surface water runoff, and drainage of water from the base of the soil profile. The wheat module in APSIM has already been validated across a wide range of locations and management practices (Asseng et al., 2000; Wang et al., 2003; Verburg and Bond, 2003; Lilley et al., 2003, 2004), and thus provided confidence that standard APSIM wheat parameter values would adequately simulate growth and environmental performance for this study.
Simulations were undertaken for two diverse locations in the Australian Wheat Belt. The sites were Katanning in Western Australia (33°41′23.9994″ S, 117°33′36″E), which exhibits a winter-dominant rainfall pattern, and Dalby in Queensland (27°10′47.9994″S, 151°15′35.9994″E), which exhibits summer-dominant rainfall. Soil at the Katanning site was a shallow sandy duplex (Isbell 2002) with a wheat plant-available water capacity (PAWC) of 56 mm. Wheat root growth at this site was allowed to reach a maximum depth of 1000 mm in simulations. Two soils were simulated at the Dalby location to explore the influence of soil water holding capacity on model output. One of the soils was a Dalby black Vertisol (Isbell, 2002) with a wheat PAWC of 336 mm, in which the wheat roots were allowed to reach a maximum depth of 1800 mm. The second soil was a self-mulching black earth (Lever Gully soil) (Isbell, 2002) with a wheat PAWC of 185 mm, in which roots were allowed to reach a maximum depth of 1300 mm. Details of physical and chemical characteristics for all three soils were obtained from the APSoil Database (Dalgliesh et al., 2006). This effectively provided three sites for the simulations, which are referred to as Katanning, Dalby 1 (336-mm PAWC), and Dalby 2 (185-mm PAWC).
After parameterization, APSIM was used to simulate the growth of wheat using daily climate data for the period 1900 to 2004 obtained from the Queensland Department of Environment and Resource Management SILO patched point data sets (Jeffrey et al., 2001). The simulations were set to sow wheat on a yearly basis on Day 152 after soil water had been reset to a certain level (as indicated in Table 2 ), with different simulations representing historically dry (fifth percentile), semidry (25th percentile), or average starting soil-moisture conditions for a continuous wheat rotation (with a weedy summer fallow). For all sites soil N was reset to 22 kg ha−1 in the wheat rooting zone just before sowing.
|Statistic||Katanning||Dalby 1||Dalby 2|
|mm water over wheat rooting depth|
The wheat varieties simulated at the Katanning and Dalby sites were ‘Kulin’ and ‘Hartog’, respectively, reflecting those commonly used in the regions. In all cases wheat was planted at 140 plants m−2 and sown to a depth of 30 mm, again reflecting regional practice. Harvesting occurred when the crop reached “harvest-ripe” stage as defined by the APSIM-Wheat model. A grazing event was simulated by removing 75% of the crop residues from the field immediately after harvest.
At each site, and for each of the soil water conditions in Table 2, a matrix of fertilizer strategies was applied. Fertilizer rates (kg N ha−1 as NO3) were set at 0, 25, 50, 75, 100, 150, 200, and 300 and applied as a single application at the time of sowing. Average gross margins for the 105-yr period were estimated using the model output, the production costs listed in Table 3 , and the grain price relationship in Fig. 1 Average gross margins were calculated for two N fertilizer prices (A$1 kg−1 and A$2 kg−1) to explore the impact of differences in N fertilizer cost (or increased application rates used in response to soil degradation [Koning et al., 2008]), on the production incentives for wheat growers and more broadly, the potential implications for national and global food supply.
|Operational cost||Annual expenditure|
|N fertilizer||(i) 1.00|
|Extra field operation||13.32|
Simulated data for the three environmental indicators of N leaching, surface water runoff, and drainage from the bottom of the soil profile (henceforth referred to as drainage), were normalized across the eight N application rates, with 0 being attributed to the worst performance in terms of environmental outcomes and 1 to the best performance. Correlation coefficients were computed to measure the strength of association between the three environmental indicators and gross margin.
Quantitative Literature Analysis
A total of 18 journal and conference papers were identified as utilizing the eco-efficiency concept (Table 4 ). These studies cover either primary production and food processing practices, or renewable energy products and inputs such as fertilizers and pesticides. In just over half (56%) of the studies, climate change is cited as the rationale for determining the eco-efficiency of a technology or practice. Half of the studies in the review measured economic impact, 44% assumed the economic outcome, and only one study omitted to consider the economic impacts resulting from a technology or practice. In contrast, all but one of the studies used empirical data to show environmental benefits. Overall, only 50% of the studies measured both economic and environmental impacts, thereby satisfying the definition of an eco-efficiency assessment.
|Publication||Climate change rationale||Economic||Environmental||Social|
|Basset-Mens et al. (2009)||1||1||2||0|
|Catarino et al. (2007)||1||2||2||0|
|Chen and Sun (2007)||0||1||2||0|
|De Jonge (2004)||1||1||2||0|
|Ingaramo et al. (2009)||0||0||2||0|
|Kim and Dale (2005)||1||2||2||0|
|Kim and Dale (2008a)||2||2||2||0|
|Kim and Dale (2008b)||1||2||2||1|
|Lozano et al. (2009)||0||2||2||0|
|Mouron et al. (2006)||1||2||2||1|
|Nevens et al. (2006)||0||1||2||0|
|Ngoc and Schnitzer (2008)||0||2||2||1|
|Pelletier et al. (2008)||1||1||2||1|
|Reith and Guidry (2003)||0||1||1||0|
|Swanston and Newton (2005)||0||1||2||1|
|Van Passel and Nevens (2007)||1||2||2||2|
In terms of assessing the social impact of a technology or practice, only one of the studies included this in their assessment; 39% assumed either a positive or negative impact, and 54% failed to give it any consideration. Across the range of studies included in this review, only one measured triple-bottom-line outcomes in terms of economic, environmental, and social impacts.
Although there is some visual evidence for an association between a climate change rationale and economic, environmental, or social assessment (Fig. 2 ), this relationship is not statistically significant. When a combined score was constructed by summing the individual scores for economic, environmental, and social content (Fig. 3 ), there seems to be slightly stronger visual evidence for studies with a climate change rationale to also contain an economic, environmental, and social assessment components (i.e., a combined score of 6). Again, this difference is not statistically significant with the Wilcoxon statistic only providing a P value of 12%.
Wheat Simulation Experiment
Wheat yields produced for the eight N fertilizer treatments simulated for the Katanning and Dalby sites are shown in Fig. 4 The solid lines indicate that when fertilizer costs were A$1 kg N−1, maximum gross margin was achieved at Katanning by applying 150 kg N ha−1 and producing a yield of approximately 2500 kg ha−1 However, when the cost of fertilizer doubled to A$2 kg N−1, maximum gross margin was achieved at a reduced N fertilizer application rate of 100 kg N ha−1 and approximately 20% lower yield. A similar, although smaller (10%) reduction in wheat yield also occurred at each of the two Dalby sites when the cost of N fertilizer was doubled.
Figure 5 shows that mean maximum economic efficiency (as measured by gross margin) can be achieved by applying between 100 and 200 kg N ha−1 for wheat grown at Katanning, 150 and 300 kg N ha−1 at the Dalby 1 site, and 75 to 150 kg N ha−1 at the Dalby 2 site. At all three sites, increasing rates of N generally resulted in increased variability in annual gross margin, and, hence, the amount of risk faced by primary producers. While the greatest mean gross margins may be achieved at the Dalby 1 site, these profits are also the most risky of the three sites in terms of annual fluctuations.
Figure 5 also shows the performance of the three environmental indicators (leaching below the root zone, surface water runoff, and drainage of water from the base of the soil profile) for the three highest gross margin estimates at each site. Drainage was the worst (highest normalized score) outcome across the three environmental indicators for the three highest gross margin N application rates simulated at the Katanning site, with leaching being the best (lowest normalized score). While the relative performance of the environmental indicators remained the same across the three highest gross margin N application rates at Katanning, their absolute values changed, with an overall worst environmental outcome resulting at the highest of the three levels of N (200 kg N ha−1).
Similarly, at the Dalby 1 site, the relative performance of the three environmental indicators remained the same across the three highest gross margin N application rates. However, while leaching and drainage reduced with increasing rates of N fertilizer, runoff increased. At the second Dalby site, leaching increased, drainage decreased, and runoff remained almost the same as N rates were increased, resulting in a change in the relative rankings of the environmental indicators.
The correlation coefficients shown in Table 5 indicate a generally negative relationship between the three environmental indicators and gross margin at all three sites with the exception of that between leaching and gross margin at the Katanning site, which increased with the application of higher rates of N fertilizer.
Advancing the assessment of a technology or practice from a single-discipline perspective to both economic and environmental outcomes, as advocated in the eco-efficiency concept, is a positive step toward identifying win-win strategies for sustainable development. The wheat simulations in this study illustrate the potential consequences of a single-discipline perspective on the issue of food security. For example, in the face of an increase in the price of N fertilizer, the rational response anticipated from primary producers using a purely economic criterion is to reduce yields in an attempt to maintain maximum gross margins. Given the global nature of markets for key agricultural inputs such as N, it is reasonable to hypothesize a resulting global-wide reduction in food production, a concomitant drop in global food supply, and increasing volatility in food prices.
In the above example, the private benefits gained by individual primary producers are both rational and efficient from an economic perspective, but fail to recognize the wider potential public costs resulting from a reduction in the availability of food and from lower levels of consumer purchasing power. Including a second disciplinary perspective in the assessment clearly offers increased potential for win-win outcomes and improvements in food security. Indeed, the asymptotic yield curves produced from the wheat simulations in response to increasing applications of N fertilizer suggest maximum yields could be produced at a private cost that is close to maximum gross margin. Such a strategy would simultaneously deliver both private and public benefits.
In addition, the dual consideration of both economic and environmental outcomes also offers some potential for identifying and minimizing unintended perverse incentives and maladaptations. As shown at the Katanning site, increasing levels of N leached from the soil profile are predicted to result from the pursuit of maximum profit margins. Identifying the tradeoffs between economic and environmental efficiency provides opportunities for optimizing outcomes in the face of competing demands. However, the seemingly unintended incentive for primary producers to maximize economic efficiency at the expense of environmental performance also overlooks the associated increases in the variability of profit margins. Indeed, while the greatest mean gross margins may be achieved at the Dalby 1 site, these profits are also the most risky of the three wheat-growing locations in terms of mean annual fluctuations. Increasingly variable incomes are associated with increased levels of risk and poverty for farmers (Ravallion, 1988) and negatively impact sustainable food supply. The broader consideration of efficiency (i.e., reduced levels of risk) provided by considering social outcomes in addition to economic and environmental outcomes offers a further advancement on assessing the efficiency of technologies and practices to address food security.
This study has shown that the eco-efficiency concept is presently being used to assess the potential for technologies and practices to improve food security under a changing climate. In the majority of cases, the assessments are limited to only economic and environmental outcomes. By showing the potential for such studies to result in unintended perverse social and environmental outcomes and incentives for primary producers, we argue that a broader assessment of efficiency would be provided by additional consideration of social outcomes, such as variability in gross margin and the risk profile of primary producers. While this proposition is not new to the eco-efficiency debate (Schmidheiny and The Business Council for Sustainable Development, 1992; Jansen, 2000), the lack of a coherent qualitative and quantitative argument for the simultaneous consideration of economic, environmental, and social outcomes is addressed here for the first time.
A secondary concern regarding the use of the eco-efficiency concept highlighted by this study is the readiness of many authors to assume the nature of the relationship existing between economic and environmental variables. For example, the assumption has been made in a number of the reviewed studies that improved environmental outcomes resulting from the adoption of a practice or technology will lead to positive economic benefits (e.g., Chen and Sun, 2007; de Jonge, 2004; Nevens et al., 2006). The experimental results demonstrate a generally negative relationship between increasing N fertilizer application rates and gross margin, with the exception of wheat growth at Katanning. Differences in this relationship have resulted from the diversity in biophysical conditions and crop management across the sites. Studies that empirically measure environmental benefits and infer concomitant economic benefits, or vice versa, must therefore be considered with some caution.
The four dimensions of food security, (i.e., availability, access, supply stability, and utilization) move beyond the simple quantification of food supply to provide a comprehensive appraisal of the requirements for individuals to reach a state of nutritional well-being in which all physiological needs are met. These dimensions focus simultaneously on economic conditions and drivers (e.g., consumer purchasing power, agricultural labor supply), as well as environmental (e.g., biophysical conditions impacting food quality and the continuity of supply) and social (adequate diet, clean water, sanitation, and health care) considerations. We have demonstrated that it is therefore of little benefit to consider potential advances in any one aspect of food security as reducing the number of people at risk of hunger, if the advances are achieved in isolation of sufficient capacity in all dimensions of food security.
Broadening the assessment of efficiency to include social outcomes may more fully capitalize on technological advances that increase production beyond the present production and efficiency frontiers (Keating et al., 2010; de Wit, 1992), and provide an evidence base to support the delivery of deliberate benefits (Giller et al., 2009). Increased understanding of the social outcomes resulting from a technology change or practice improvement may also contribute to addressing the historically low levels of update and ongoing use observed in the fields of agricultural and development practice (Doss, 2001; Moser and Barrett, 2003). This analysis similarly suggests that the assessment of the efficiency of a technology or practice to improving food security in the face of climate change. It also requires broader consideration of the socioeconomic and biophysical environment necessary to facilitate its adoption (sensu Howden et al., 2007). While it is suggested that socioeconomic developments such as population growth and energy use have relatively greater importance on the number of people at risk of hunger compared to that of climate change (Tubiello and Fischer, 2007), clearly it is important to consider all key factors impacting the ability of an individual to have adequate quantities of food of appropriate quality, accessible at all times, and with the capacity to be utilized sufficiently to reach a state of nutritional well-being in which all physiological needs are met (Garrett, 1995).