Public health nutrition challenges have become more complex in recent years because diets are not only expected to support health and well-being, but there is an increasing expectation that they are also environmentally sustainable^{(Reference Garnett1–Reference Gheewala, Jungbluth and Notarnicola4)}. Part of the responsibility rests with the systems of food production, processing and distribution. However, population shifts to diets that are lower in environmental impacts could also contribute to improving sustainability, based on the notion of sustainable lifestyles expressed in Sustainable Development Goal 12⁽⁵⁾. The largest body of evidence relating to sustainable diets concerns greenhouse gas emissions^{(Reference Ridoutt, Hendrie and Noakes2)}. An alarming finding is that many lower greenhouse gas emission dietary patterns are linked to poor nutritional and health indicators^{(Reference Payne, Scarborough and Cobiac6,Reference Ridoutt, Baird and Hendrie7)} , highlighting the need for lower environmental impact diets to also consider nutritional adequacy and support longstanding public health nutrition objectives^{(Reference Ridoutt, Hendrie and Noakes8)}. Evidence in relation to the impacts of diets on water scarcity is also beginning to emerge^{(Reference Hess, Andersson and Mena9–Reference Ridoutt, Baird and Anastasiou13)}. Water scarcity reflects the availability of water relative to the natural rate of replenishment. As water scarcity increases, the availability of water for human uses and for the environment diminishes. Water scarcity is a major international environmental concern⁽⁵⁾. The food system is critically relevant to resolving water scarcity since agriculture alone accounts for around 70 % of global freshwater withdrawals⁽¹⁴⁾.

In Australia, a large (>9000) sample of self-reported adult daily diets were recently assessed for diet quality and water scarcity footprint that assesses contribution to water scarcity^{(Reference Ridoutt, Baird and Anastasiou13)}. Using a quadrant analysis approach, a subgroup of diets was identified with both higher diet quality and lower water scarcity footprint. This subgroup had an average water scarcity footprint 43 % lower than the current average diet and 64 % lower than the subgroup of diets with both lower diet quality and higher water scarcity footprint. These findings demonstrate that large reductions in dietary water scarcity footprint are possible. However, the question remains as to whether such reductions are adequate, or necessary, to be considered sustainable. The pursuit of unnecessarily aggressive reductions could limit dietary diversity.

To guard against major and potentially irreversible earth system change, a variety of planetary boundaries^{(Reference Rockström, Steffen and Noone15,Reference Rockström, Steffen and Noone16)} or absolute environmental limits^{(Reference Bjørn, Chandrakumar and Boulay17,Reference Sandin, Peters and Svanström18)} have been proposed. These boundaries represent thresholds for natural resource use and emissions to the environment that should not be exceeded. This approach to sustainability assessment has emerged in recognition that with the global population increasing and standards of living generally rising, marginal improvements in eco-efficiency may not be enough to avert serious environmental change^{(Reference Bjørn, Diamond and Owsianiak19)}. For example, a major study of EU consumption, supported by international trade, recently concluded that environmental impacts exceed a fair share of the so-called ‘safe operating space’ within which humanity’s footprint is within the planetary boundaries^{(Reference Sala, Benini and Beylot20)}. Critically, food consumption was identified as one of the main drivers of environmental impact.

Our study involved evaluation of the water scarcity footprint of Australian adult diets in relation to proposed planetary boundaries for global freshwater use, that is freshwater withdrawals for agriculture and industry. Our objective was to assess the absolute sustainability of water use supporting Australian dietary habits and the adequacy of current dietary guidelines⁽²¹⁾ if they are to support sustainable water use in addition to health and well-being. To our knowledge, this is the first study to assess the water scarcity footprint of individual self-reported diets within a planetary boundary framework.

Methods

Dietary intake data

Dietary intake data, covering 9341 adults (19 years and above), were obtained from the National Nutrition and Physical Activity component of the Australian Health Survey⁽²²⁾. This survey, conducted by the Australian Bureau of Statistics over the period 2011–2013, using a 24-h recall process and a complex sampling method⁽²³⁾, remains the most detailed and nationally representative source of dietary intake information in Australia. For each individual, the data describe quantities of foods and beverages consumed on the day prior to a face-to-face interview with a trained assessor.

As described elsewhere^{(Reference Ridoutt, Baird and Anastasiou13)}, mixed dishes were disaggregated into their basic components and cooked food portions were converted to equivalent raw quantities. In addition, adjustments for under-reporting were made using estimates of the under-reported food energy from the Australian Bureau of Statistics⁽²³⁾. For each individual daily diet, total energy intake was determined using data obtained from the Australian Food Composition Database⁽²⁴⁾, along with the number of serves of each of the food groups described in the Australian Dietary Guidelines⁽²¹⁾. A diet quality score (out of 100) was also quantified, using an index that describes degree of compliance with the guidelines^{(Reference Golley and Hendrie25)}. A higher score reflects higher compliance with the Guidelines.

Water scarcity footprint modelling

The evaluation of water use across a food system is complex as water scarcity can vary greatly from one geographic region to another. Water use from regions of scarcity and abundance cannot be simply aggregated as this is not environmentally meaningful^{(Reference Pfister, Boulay and Berger26)}. Instead, a water scarcity footprint needs to be quantified, as described in ISO14046:2014⁽²⁷⁾, taking into account the spatial distribution of water use and the local water scarcity conditions. In Australia, the water scarcity footprint of the major agricultural commodities has been assessed^{(Reference Ridoutt, Hadjikakou and Nolan28)}, as well as processed food products of local and imported origin^{(Reference Ridoutt, Baird and Anastasiou13)}.

That said, water scarcity is a human construct. Water scarcity footprint results obtained using different water scarcity models are typically highly correlated^{(Reference Ridoutt, Baird and Anastasiou13,Reference Ridoutt and Hodges29)} ; however, they can differ in magnitude. Therefore, in this study, an ensemble method was used, as is common when working with climate data from a variety of models^{(Reference Flato, Marotzke, Abiodun, Stocker, Qin and Plattner30)}. To characterise the water scarcity footprint of foods consumed in Australia, a multi-model ensemble was calculated as the arithmetic mean of results obtained from three different water scarcity models reported previously^{(Reference Ridoutt, Baird and Anastasiou13)}. Data for almost 150 separate food items are presented in the Supplementary Material. Water scarcity footprint results were scaled relative to water use at the global average level of water scarcity (i.e. litres equivalent, L-eq) to enable direct comparison with the planetary boundary for water use.

Planetary boundary analysis

The authors of the planetary boundary concept initially proposed a boundary for global freshwater consumption of 4000 km³/year, with a zone of uncertainty extending to 6000 m³/year^{(Reference Rockström, Steffen and Noone15)}. By allocating 70 % of this available water use to the food system⁽¹⁴⁾ and sharing it equally among the 7·8 billion global citizens, the maximum water use to support an individual daily diet is in the range of 983–1475 L (Table 1). Subsequent analysis, based on more complex modelling, has revised downwards the planetary boundary to 2800 km³/year, with a zone of uncertainty of 1100–4500 km³/year^{(Reference Gerten, Hoff and Rockström31)}. Anticipation of higher future global populations also constrains the water use available to support an individual daily diet (Table 1).

Table 1 Downscaling the planetary boundary for freshwater consumption to define a boundary for water use to support an individual daily diet that is sustainable

As dietary water scarcity footprint results presented in this study are expressed relative to water use at the global average water scarcity, they can be directly compared against the planetary boundary for water use downscaled to the level of an individual diet. We assessed the average (i.e. mean) Australian adult daily diet. In addition, a quadrant analysis was undertaken for the 19- to 50-year age group used in the Australian Dietary Guidelines⁽²¹⁾ to define a higher diet quality/lower water scarcity footprint subgroup and a lower diet quality/higher water scarcity footprint subgroup. For this age group, the water scarcity footprint of a recommended diet based on the Australian Dietary Guidelines⁽²¹⁾ was also quantified.

Results

Using the multi-model ensemble approach, the water scarcity footprint of the average Australian adult daily diet was 432·6 L-eq (95 % CI 432·5, 432·8, n 9341). Average energy intake was 10 458 kJ. As has been reported elsewhere^{(Reference Ridoutt, Baird and Anastasiou13)}, the largest contribution to the water scarcity footprint was from discretionary foods (26·1 %). These foods, sometimes also referred to as indulgence foods, are energy-dense and nutrient-poor foods high in saturated fat and/or added sugars, salt or alcohol. The Australian Dietary Guidelines⁽²¹⁾ recommend that these foods are consumed only occasionally and in small quantities, although most Australians consume these foods excessively. Concerning the core food groups defined in the Australian Dietary Guidelines⁽²¹⁾, fruits made the largest contribution to the water scarcity footprint of the average Australian adult diet (20·0 %), followed by dairy foods and alternatives (14·4 %). The Australian Dietary Guidelines⁽²¹⁾ group dairy foods like milk, cheese and yogurt, together with non-dairy alternatives such as soya, cereal and nut beverages. Fresh meats (beef, lamb, poultry and pork) and alternatives (fish, eggs, tofu, legumes/beans) contributed 12·1 % of the water scarcity footprint; cereal/grain foods and vegetables contributed 11·8 and 7·7 %, respectively.

The water available to sustain a daily diet depends on the estimated planetary boundary for water use, as well as the share that is apportioned to the food system and the global population. The smallest estimate for a current population of 7·8 billion is 695 litres/d (Table 1). On this basis, the average Australian daily diet is well within the planetary boundary for water use (Fig. 1). Considering the large 19- to 50-year age group (n 5157), diets that were both higher in diet quality and lower in water scarcity footprint required only 245 L-eq/d (95 % CI 244·7, 245·0), below even the lowest zone of uncertainty for the planetary boundary (Table 1). Only diets that were both lower in diet quality and higher in water scarcity footprint reached the boundary (699 L-eq/d; 95 % CI 698·9, 700·9).

Fig. 1 Water scarcity footprint (WSF) of Australian adult dietary patterns compared with planetary boundaries for dietary water use for a current global population of 7·8 billion and a future global population of 9 billion

Compared with the current average diet, a recommended diet⁽²¹⁾ requires substantially reducing the number of servings of discretionary foods and increasing the number of servings from all the five core food groups (Table 2). If the current diet was scaled accordingly, the water scarcity footprint would increase to 521 L-eq/d. However, with lower water scarcity footprint food choices (as exhibited by the higher diet quality and lower water scarcity footprint subgroup), the recommended diet can be achieved with less water use (379 L-eq/d; Table 2). Either way, the recommended diet was also found to be within the planetary boundary for a current population of 7·8 billion and a future population of 9 billion (Fig. 1).

Table 2 Food intake and water scarcity footprint (WSF) for the current adult daily diet (19–50 years), the current diet scaled to the recommended servings in the Australian Dietary Guidelines⁽²¹⁾, and a recommended diet with improved WSF intensity based on the higher diet quality and lower WSF subgroup. Food groups are as defined by the Australian Dietary Guidelines. The recommended number of servings is based on the average for men and women. WSF are expressed relative to water use at the global average water stress