Global and domestic drivers of sustainability indicators

The inclusion of sustainability in national food-based dietary guidelines is increasing as policymakers prioritise sustainability. While only four countries (Brazil, Germany, Qatar, Sweden) included sustainability in their national dietary guidelines in 2016^{(Reference Gonzalez Fischer and Garnett19)}, this had increased to 37 countries by 2022^{(Reference James-Martin, Baird and Hendrie20)}. Twenty-nine of these countries, primarily from Europe and Latin America, included environmental impacts in their consideration of sustainability. In public-facing documents, environmental messaging focused on reducing food waste, while in background documentation, there was a greater consideration of on-farm environmental impacts (greenhouse gas emissions, water, land use, nitrogen and phosphorus application, chemical pollution, biodiversity⁽²¹⁾). A much smaller emphasis was placed on the use of antibiotics and hormones in food production and minimising plastics in food packaging. The most recent version of the Australian Dietary Guidelines⁽²²⁾ includes an appendix that directly addresses the environmental sustainability of food. A review of the 2013 document, due to be released in 2026, will go further and consider sustainability, as defined by accessibility, affordability, equity and environmental impact, in a set of updated guidelines⁽²³⁾.

The evolving landscape of sustainability policies and initiatives, both globally and domestically, is also leading to a major shift in how environmental impact is reported in agribusiness. The need to demonstrate sustainability credentials is, for some stakeholders, driven by economic factors, such as market access or risk management. Businesses also align themselves with global initiatives such as the UN Sustainable Development Goals^{(Reference Nations24)}, the Global Reporting Initiative⁽²⁵⁾ and the Science-Based Target Initiative⁽²⁶⁾, which require the measurement of indicators to ensure a documented commitment to responsible practices.

The global and domestic economic pressures to quantify the environmental impacts of food production and diets are complex and are increasing the availability of information on environmental performance of businesses, including on farm. For instance, global risk management and disclosure frameworks, such as the Taskforce for Nature Related Financial Disclosures (TNFD) and climate-related disclosures incorporated in the standards of the International Sustainability Standards Board, exert significant influence on large businesses and financial institutions and drive the need for environmental, social and governance reporting, but not just at the corporate scale. In several countries, including Australia, climate reporting is becoming mandatory, encompassing Scope 3 (supply chain) greenhouse gas emissions for large businesses and financial institutions, which extends the need for reporting through the supply chain to verify the sustainability of agricultural production and processing of food items. For example, retailers need to demonstrate their emissions profile, including Scope 3 emissions, and consequently pass this requirement onto their suppliers through their procurement processes. Notably, nature-based reporting under the TNFD currently operates on a voluntary basis, but many countries adopt similar reporting frameworks. In the long term, evidence-based base-lining and continuous improvement of practices will lead to more sustainable agricultural production systems and diets.

Developing indicators and metrics for sustainable diets

Sustainability indicators gained prominence following the 1992 Rio Summit⁽²⁷⁾, which underscored the need for measurements to track progress towards sustainable development. Various international groups have since provided direction for assessing the environmental impact of food and diets. For example, the Food and Agricultural Organisation of the United Nations (FAO) and the World Health Organisation (WHO)⁽²¹⁾ have established guiding principles for sustainable healthy diets that include (a) maintaining greenhouse gas emissions, water and land use, nitrogen and phosphorus application and chemical pollution within set targets and (b) preserving biodiversity, including that of crops, livestock, forest derived foods and aquatic genetic resources and avoiding overfishing and overhunting.

The high-level principles provide a framework for developing relevant indicators. These have been further refined through the consideration of the planetary boundaries framework that defines the safe operating space for human societies^{(Reference Steffen, Richardson and Rockström28)}. Over time, a diverse range of indicators has emerged^{(Reference Sparling, White and Boakye29)}, and five broad environmental issues (climate change, land scarcity, pollution of land and water resources, water scarcity and biodiversity loss) have been established as priorities (Table 1). The interrogation of global datasets has provided indicators, typically greenhouse gas emissions, land use, energy use, acidification potential, eutrophication potential and water use. These have then been collectively or selectively used for dietary recommendations in many countries^{(Reference Poore and Nemecek4,Reference van Dooren, Aiking and Vellinga30,Reference Clark and Tilman32–Reference Harwatt, Benton and Bengtsson34,Reference Kesse-Guyot, Chaltiel and Wang44,Reference Aldaya, Ibanez and Dominguez-Lacueva45)} . While these issues and indicators broadly address planetary boundaries, some aspects, such as biosphere integrity to address genetic diversity and the role of the biosphere in system function, are usually omitted due to insufficient data^{(Reference Clark and Tilman32)}, complex metrics or because risks to biodiversity are assumed to be covered through the application of other indicators. Similarly, novel entities or indicators of toxicity such as chemical pollution or contamination are rarely addressed^{(Reference Ran, Cederberg and Jonell31)}. Other indicators are adapted; for example, land use refers primarily to forested areas within the planetary boundaries framework^{(Reference Steffen, Richardson and Rockström28)} but relates to the use of crop land in the Nordic Nutrition Recommendations for sustainable food production and consumption^{(Reference Harwatt, Benton and Bengtsson34)}. Land use is also sometimes used as an indicator of competition for productive land and other times as a surrogate for biodiversity. Soil health and degradation fall outside defined planetary boundaries but are also threats to sustainability and have been proposed as a tenth planetary boundary^{(Reference Ran, Cederberg and Jonell31)}.

Table 1. Environmental issues, indicators and metrics most commonly used for assessing the environmental impact of diets. Information partly derived from van Dooren et al. ^{(Reference van Dooren, Aiking and Vellinga30)} and Ran et al. ^{(Reference Ran, Cederberg and Jonell31)}

* Interpretation of indicator type differs from sourced publications based on local environmental considerations.

The indicators historically used to inform dietary recommendations are usually global pressure indicators (Table 1), that is, indicators of environmental load. These are different from impact indicators that aim to capture the consequences of the environmental load^{(Reference Ran, Cederberg and Jonell31,Reference Smeets and Weterings46)} . Therefore, while geographic and system differences in expression of these pressure indicators are recognised^{(Reference Poore and Nemecek4,Reference Ran, Cederberg and Jonell31)} , recommendations on changes in foods often lack region and system specificity and fail to reflect different environmental priorities. Addressing this gap requires the use of impact indicators that are prioritised for, and can be applied at, a regional or farming system level. Some such indicators have been developed from the pressure indicators (Table 1), but the list is incomplete. An alternative set of indicators has recently been proposed with more emphasis on impacts^{(Reference Ran, Cederberg and Jonell31)}. The basic indicators include carbon footprint, blue water consumption, biodiversity impact from land use, ecotoxicity or pesticide use and exploitation of wild fish stocks. These provide opportunities for assessments at a regional or farming system scale and to capture environmental trade-offs. For more comprehensive assessments, land use, soil quality and green water consumption are proposed, with impacts on sea floor, energy use and nitrogen and phosphorus output only used for high ambition assessment.

While efforts have been made to standardise indicators and metrics for corporate sustainability reporting⁽⁴⁷⁾, their applicability varies due to contextual factors such as country, agricultural commodity, production system or region, and little guidance exists on appropriate choices. Ideally, indicators should reflect these diverse contexts; if not, the conditions limiting their applicability must be stated^{(Reference Moldan, Dahl, Hák, Moldan and Dahl48)}.

The alignment of indicators for sustainable development across countries was demonstrated in 2002 in a country comparison of United Nations Conference on Sustainable Development categories and themes^{(Reference Hass, Brunvoll and Hoie49)}. However, while global datasets provide a consistent basis for estimating impacts and enabling country and system comparisons, they often fail to capture the nuances of food production at national and sub-national scales. This is clearly illustrated in the background reports for the 2024 State of Food and Agriculture Report, which reported that the use of global-level datasets often overestimated, and sometimes underestimated, the impacts of food production on the environment in all the case study countries investigated (Australia, Brazil, Colombia, Ethiopia, India, UK)^{(Reference Vittis, Mosnier and Smith50)}. This was particularly true for land use change and greenhouse gas emissions. For Australia, the global Historic Land Dynamics Assessment+ (HILDA+) did not accurately capture extensive land use patterns, mapping large areas of conservation and protected lands (e.g. the entire Simpson Desert) to modified pastures and grazing^{(Reference Navarro Garcia, Sperling and Islam51)}. There were also inconsistencies in land use change between the HILDA+ and local datasets. Similarly, the use of FAO Tier 1 methods overestimated greenhouse gas emissions from the Intergovernmental Panel on Climate Change Agriculture sector by 46 %^{(Reference Poore and Nemecek4,Reference Ran, Cederberg and Jonell31,Reference Smeets and Weterings46)} .

Overall, sustainability indicators are crucial for informing dietary recommendations and assessing the environmental impact of food production. However, their effectiveness depends on their ability to capture regional and system-specific nuances, as well as their alignment with broader sustainability goals. The ongoing refinement of indicators and metrics will be essential to ensure that sustainability assessments accurately reflect the complexities of food systems worldwide.

Australia – a case for relevant indicators and metrics

Locally relevant indicators are required to drive on-farm changes that make food production systems, and therefore diets, more sustainable. To illustrate this point, we use examples from three important and contrasting food production systems and regions within Australia: the pastoral production systems of the rangelands, the mixed farming zone of southern Australia, and the Murray–Darling Basin. These systems and regions are diverse in their management, climate and physical characteristics and therefore face different environmental challenges and opportunities.

The rangelands cover around 80 % of Australia⁽⁵²⁾ (Fig. 1) and span a diverse group of relatively undisturbed ecosystems including tropical savannas, woodlands, shrublands and grasslands. Soil and rainfall constraints mean that most of this land is not suitable for cropping or horticulture, and livestock remain the only large-scale option for food production. Cattle, sheep and goats in rangeland systems generally graze unimproved native pastures with minimal inputs, so productivity is low compared to more intensive systems. The low inputs mean such systems are close to circular, and well-managed extensive grazing can provide a range of ecosystem services^{(Reference Manzano, de Aragão Pereira and Windisch55,Reference Manzano, Rowntree and Thompson56)} . Elsewhere, the mixed farming zone (crops and livestock) comprises 35–41 million hectares (ha) of improved pasture with 21–31 million ha used for crops in any 1 year^{(Reference Pethick, Bryden and Mann57)} (Fig. 1). Much of this land is used for both crops and livestock at different times within a year or in a crop/livestock rotation between years, with most cropping farms also including livestock^{(Reference Bell, McCormick and Hackney58)}. Although much smaller than the rangelands, the mixed farming zone is a key driver of food production, producing most of Australia’s crops and 40 % of the livestock^{(Reference Bell, McCormick and Hackney58)}. Crops (predominantly wheat, oats, barley, canola and lupins, with some chickpeas and lentils) are produced almost exclusively in rain-fed systems utilising green water only. Livestock use mostly non-arable land not suited for alternative food production or dual-purpose arable land used for both cropping and grazing in areas of low water stress^{(Reference Wiedemann, Yan and Murphy59)}. Finally, the Murray–Darling Basin is the largest river basin in the country, covering around 1 million km² of eastern Australia^{(Reference Holland, Luck and Max Finlayson60)} (Fig. 1), and home to a unique collection of flora and fauna. Two-thirds of Australia’s irrigation occurs in this region, which produces around 40 % of the country’s agricultural produce, including 100 % of the nation’s rice; more than 50 % of the grapes, citrus, stone fruit and pome fruit; and 25 % of dairy farms⁽⁶¹⁾. There is also overlap with both the rangelands and mixed farming zones. While some environmental issues such as biodiversity loss and carbon footprint are common across these production systems and regions, others vary despite overlaps in geography and commodity production (Table 2). In addition, the most used global pressure indicators (Table 1) are not always adequate to inform local dietary recommendations to support sustainable farming practices and, if applied indiscriminately, may increase local environmental impact.

Figure 1. Map of key agricultural land uses⁽⁵³⁾ overlaid with the location of the Murray–Darling Basin⁽⁵⁴⁾ and rangelands⁽⁵²⁾. The rangelands are predominantly natural pastures grazed by livestock and areas used for conservation or other protected resources. The mixed crop-livestock zone includes areas with cropping and sown pastures.

Table 2. Examples of how environmental indicators differ between production systems

Livestock contribute 85 % of greenhouse gas emissions (66 Mt CO₂e) from the Australian agriculture sector via enteric methane and manure⁽⁶²⁾, thus reducing greenhouse gas emissions is a priority for both the rangeland and mixed farming systems where the majority of ruminant livestock are produced. However, focusing only on emissions (greenhouse gas emissions and energy use indicators, Table 1) does not account for opportunities to store carbon in grazing landscapes. Research that takes into account both emissions and sinks indicates that net emissions from the Australian sheep and cattle industries decreased nearly 70 % from 2011 and 2020^{(Reference Bowen Butchart, Christie-Whitehead and Roberts63)}. Within some grazing systems, carbon sequestration in soils or vegetation partially or fully offsets greenhouse gas emissions^{(Reference Thomas, Sanderman and Eady64–Reference Mayberry, Bartlett and Moss67)}, although net emissions and emission intensity are likely to remain higher than for cropping systems^{(Reference Thomas, Sanderman and Eady64–Reference Mayberry, Bartlett and Moss67)}. Greenhouse gas emissions are commonly presented as CO₂e using Global Warming Potential (GWP) 100, which is a measure of how much a greenhouse gas (e.g. methane or nitrous oxide) warms the earth compared to CO₂ over 100 years. However, this approach does not take into account the atmospheric lifetime of different greenhouse gases, and the use of GWP* or radiative forcing-based climate footprints may offer different conclusions^{(Reference Ridoutt68)}.

Biodiversity loss is important for all three systems and regions, but pressure indicators such as the area of land used to produce a unit of food (Table 1) do not provide a balanced comparison. For example, while low-input and extensive grazing systems like those found in rangeland areas use much larger areas of land to produce a unit of food compared to intensive systems, they generally do so with smaller impacts on the environment because livestock graze native vegetation, which has not been cleared to the same extent as cropping and intensive grazing systems. In these rangeland systems, biodiversity loss is driven not by land use per se but by overgrazing and land degradation^{(Reference Waters, Orgill and Melville69)}. Therefore, specific impact indicators that account for land quality (e.g. habitat loss, land degradation) or the abundance of plant and animal species (e.g. species loss) may be more appropriate^{(Reference Harrison, Palma and Buendia70)}.

Similarly, land scarcity indicators (Table 1) tend to focus on the absolute area of land used for food production rather than the area of arable land. This approach fails to acknowledge the ability of livestock systems to produce food from non-arable land and crop byproducts and the efficiencies associated with dual-purpose mixed systems. In a global meta-analysis, Clark and Tillman^{(Reference Clark and Tilman32)} indicated that ruminant meat required 0·018–0·03 m² land/kJ, whereas the cropland footprint calculated by Ridoutt and Navarro Garcia^{(Reference Ridoutt, Anastasiou and Baird71)} for Australian beef indicates <0·001 m²/kJ. Presumably, the difference relates to the inclusion of land that is not suited to other food production in the first reference. Similarly, net protein contribution (NPC) can be used to indicate the efficiency of protein production systems, with grassfed beef production having an NPC of 1597, compared to an NPC of 1·96 for grain-finished beef ^{(Reference Thomas, Beletse and Dominik40)}. An NPC greater than 1 indicates a positive contribution of protein to the food system, with higher values signifying higher contributions and greater efficiency.

Finally, indicators relating to nutrient pollution (e.g. acidification and eutrophication potential, Table 1) are most relevant to intensive production systems, especially those draining into major and/or ecologically significant river systems. Thus, these indicators are extremely important for the Murray–Darling Basin, but of little relevance to the extensive rangeland systems.

The Australian agriculture sustainability framework

In response to the mismatch between international sustainability indicators and Australian production conditions, many agricultural industries have developed (or are developing) reporting frameworks to define, measure and report on the sustainability performance of that sector (cf. Australian Beef Sustainability Framework, Australian Dairy Sustainability Framework, Australian Grain Industry Sustainability Framework, Sheep Sustainability Framework and others). These frameworks help the industries they pertain to shape their sustainability narrative based on credible and documented information^{(Reference McRobert, Gregg and Fox17)} and are, by design, more representative of Australian conditions.

A whole of agriculture national framework, the Australian Agricultural Sustainability Framework (AASF), has been designed to closely align with these industry frameworks, as well as many international sustainability initiatives, and has been created to provide coordination and guidance to the whole agricultural value chain^{(Reference McRobert, Gregg and Fox17)}. The AASF is built around the three themes of sustainability: economic, environmental and social, and the current iteration (version 4·2, released October 2023) contains 17 principles (desired outcomes, e.g. net anthropogenic greenhouse gas emissions are limited to minimise climate change) and 43 criteria (conditions to be met to comply with the principle, e.g. greenhouse gas emissions are reduced throughout lifecycle)^{(Reference McRobert, Gregg and Fox17)}.

A key challenge for reporting against a sustainability framework is the identification of suitable indicators and the acquisition of data supporting these indicators. For the AASF, it is recognised that a selection of indicators, relevant for specific contexts, is needed to enable the use of the framework for a wide range of purposes. These may be to produce a whole of Australian agriculture sustainability report through to guiding an agricultural company to understand their sustainability status. For many of the potential uses of the AASF, data to support reporting is collected at the individual producer/processor level. To access this data requires the ability to efficiently and effectively share data in a trusted and secure way across the industry. To enable this data sharing, the designers of the AASF have focused on the design of a data sharing ecosystem^{(Reference Lemon and Kostanski72)}, the core of which is a Register of Indicators. The purpose of this register is to provide guidance to data collectors (producers and processors) on what data to collect, direction to those seeking data on what data can be asked for, and confidence to data and digital service providers on what data needs to be supported within their service offerings.

To instil trust in the Register of Indicators, the register is governed by an independent panel of trusted individuals who are advised by groups of experts with respect to specific indicators. The register’s management processes are transparent and designed to ensure that the register contains only indicators that meet specific assessment criteria and have been identified in an inclusive process that represents relevant stakeholders.

Associated with the Register of Indicators, the AASF Data Ecosystem also includes a catalogue of preferred methods for measuring individual indicators along with catalogue of published datasets containing measures of individual indicators. As stated above, the purpose of the AASF Data Ecosystem is to enable sharing of sustainability data across the agriculture sector in support of a variety of use cases. To achieve this, the Register of Indicators needs to list enough indicators to be useful while not having so many that the register becomes unmanageable.

Tracing provenance and environmental impact

Relying on a limited set of indicators and metrics derived from global databases to categorise individual foods or entire diets is therefore an oversimplified approach to assign environmental impact. This paper has provided examples where indicators and metrics need to be derived for a specific location or production system if regional environmental priorities are to be addressed. National improvement then comes through an aggregation of regional improvements and global improvement through aggregation of national improvements.

Counterintuitively, classification of foods and products based on global databases could even have a negative impact on regional environments. From a producer’s point of view, if their product has received a poor global environmental impact assessment with no system discrimination (and vice versa), there is very little incentive to make environmental improvements; the drivers of change are more likely to be economic than environmental.

Information on provenance linked with sustainability credentials can complement dietary recommendations based on global datasets. This can be done through the determination of geographic origin through isotopic fingerprints^{(Reference Pauli and Rodriguez Curras73,74)} . More likely, full traceability systems using digital technologies such as serialised labelling (barcodes, QR codes), digital, secure record-keeping (Blockchain) and/or artificial intelligence will be required to associate a specific food item with place-based on-farm measurement and monitoring of production practices used at all stages of the supply chain^{(Reference Wallace and Manning75–Reference Voisin, Godrich and Blake77)}.

The provision of transparent and consistent information can help to improve consumer awareness and confidence, with subtle changes in marketing and context leading to green nudging and changed buying behaviour^{(Reference Ferrari, Cavaliere and De Marchi78)}. While the main incentive for farmers to implement and demonstrate technology and management to reduce environmental impacts of their produce is currently driven through the demand generated by the sustainable procurement standards from the retail sector^{(Reference Ricketts, Palmer and Navarro-Garcia79)}, going forward, consumer choice could become a stronger driver than what it currently is.