Introduction

"The pressures we exert on the planet have become so great that scientists are considering whether the Earth has entered an entirely new geological epoch: the Anthropocene, or the age of humans. It means that we are the first people to live in an age defined by human choice, in which the dominant risk to our survival is ourselves.” Achim Steiner, UNDP Administrator (pg. 7) [1].

Current food production methods and consumption patterns are unsustainable in supporting human and planetary health [2••,3•,4]. It is estimated that 21–37% of the total net greenhouse gas emissions (GHGe) [3•, 5••] are a result of the inputs and actions required to sustain the current global food system. The 2019 Intergovernmental Panel on Climate Change (IPCC) [5••] quantified that, on average, 16–27% of global anthropogenic emissions were a result of actions and inputs from the food system up to the farm gate (i.e. food production), while the remaining 5–10% were post-farm gate (i.e. food processing, consumption, and waste). Therefore, factors across the whole food chain are important to consider. Fifty percent of the world’s habitable land is used for agriculture [6], and 60% of terrestrial biodiversity loss is related to food systems [7]. Furthermore, an estimated two-thirds of freshwater withdrawals are for irrigation [8]. Changes to the food system provide critical opportunities for solutions to help mitigate climate change, as recognised by global bodies [2••, 9]. International goals that aim to achieve such solutions include the United Nations’ Sustainable Development Goal number 12 (Ensure sustainable production and consumption) [3•, 4] and the 2016 Paris Agreement on Climate Change emission targets [10], to which Australia and New Zealand (NZ) are signatories. The most recent 2021 IPCC report signifies a code red warning that immediate change is required to mitigate climate warming that is likely to reach 1.5 °C above pre-industrial levels between 2030 and 2052 if it continues to increase at the current rate [11••].

Undoubtedly, both food production and consumption practices need to shift to feed the predicted nearly 10 billion people by 2050, to achieve human health within finite planetary boundaries [2••]. The EAT Lancet report published in 2019 has drawn attention to more sustainable dietary practices required to meet this challenge and coined the term “The Great Food Transformation” [2••]. Within countries, recommendations about the food and dietary patterns associated with health are provided by evidence-based dietary guidelines, with pictorial food guides often used to translate dietary guideline statements into practical advice for consumers [12]. Despite attempts to incorporate environmental sustainability, the 2013 Australian Dietary Guidelines (ADG) have been criticised for making only brief reference to sustainability considerations in an appendix [13]. In comparison, the recently updated NZ Eating and Activity Guidelines for NZ adults have considered environmental sustainability [14]. With the ADG currently under review, it is timely to consider the environmental impacts associated with food consumption in Australia.

One approach to contribute to meaningful debate on practical and holistic policy changes, including the development of dietary guidelines, is to consider both the health and environmental impacts associated with consumption of foods and dietary patterns [15, 16•]. The aim of this rapid review was to assess the environmental impacts associated with food consumption (both actual and apparent) in Australia and NZ. Due to the shared food standards between Australia and NZ (Food Standards Australia and New Zealand), and the close geographical proximity of the two countries that facilitates food trade, the environmental impacts of food consumption in both countries were assessed.

Methods

The rapid review protocol was registered with PROSPERO (CRD42020221623). Exposures of interest were actual and apparent consumption of individual foods, food groups, or dietary patterns in Australia and NZ. Outcomes of interest included but were not limited to, land use, water use, biodiversity loss, and GHGe (Appendix 1). Search terms were developed to reflect the Populations, Exposures, and Outcomes of interest. Keyword searches were executed across four databases (Environmental Science Index, Web of Science, Scopus, and Medline). A grey literature search using Google with country filters was also conducted using a reduced number of keywords.

Database searches were limited to the English language and primary studies, reviews, and reports that presented data from Australia and/or NZ were eligible for inclusion. Publications that included data from other countries were eligible for inclusion only if it was possible to extract relevant data specific to Australia and/or NZ. A publication limit was set to the last 10 years (01/01/10 to 02/12/20) to include publications since the evidence review [17] that informed the 2013 ADG. The reference lists of included studies and reports were also used to identify relevant literature.

Results of the database searches were uploaded to Covidence and duplicates removed. Two authors (SF, LG) independently screened titles and abstracts of identified articles, while the full text of those deemed relevant was screened by SF and EB, and discrepancies were resolved by a fourth reviewer (KC). Data from the included articles was extracted by two reviewers (SF and EB) using a standardised proforma with discrepancies being resolved in conjunction with a fourth reviewer (KC).

Findings

A total of 21,428 unique articles were identified through database searches (Fig. 1) resulting in 308 full-text articles being assessed for eligibility, of which 283 were excluded, leaving 25 articles for final inclusion. An additional two reports identified from the grey literature search were included. Due to the heterogeneity of the data extracted only the following environmental indicators are reported: GHGe, land use, ecological footprint, cropland footprint, and water use/footprint/scarcity which meant that seven studies that did not report these outcomes were excluded. Environmental indicators not included were phosphorus footprint, nitrogen footprint, river environments, and energy use. For a glossary of terminology, refer to Appendix 2.

Fig. 1
figure 1

PRISMA flow diagram of studies and reports included in the rapid review

Full size image

Table 1 provides a summary of the 20 included articles. Most articles provided Australian data (n = 18) while two provided NZ data. Over half of the articles examined individual foods or food groups (n = 15). Of the articles that examined dietary patterns (n = 13), over half also provided data on the foods or food groups that were included within those dietary patterns (n = 8). There was a high degree of heterogeneity in the environmental indicators that were used within the cited studies with GHGe being most commonly reported (n = 12), followed by water use (n = 4), water footprint (n = 3), carbon footprint (n = 3), land use (n = 2), and, lastly, ecological footprint (n = 2). The following indicators were additionally included in some of the articles: non-CO2 GHGe (n = 1), water scarcity footprint (n = 1), and cropland footprint (n = 1).

Table 1 Summary of included articles (n = 20)
Full size table

The findings for the following environmental indicators are described below: GHGe; water use/footprints, and water scarcity footprints; and land use, ecological footprints, and cropland footprints.

GHGe

The industrialisation of farming and food production systems and rising volumes of food waste in landfill has led to the food system becoming one of the leading greenhouse gas-emitting industries responsible for one-quarter of total global emissions [3•, 8]. To measure and compare the quantity of total emissions from the food industry, many researchers utilise kilograms of carbon dioxide equivalents (kg CO2e), a metric that accounts for all greenhouse gases and their relative potential for warming the earth’s surface over the next hundred years [3•].

Australia’s total GHGe for 2020 was an estimated 510.1 Mt CO2e [38], and food system-related GHGe currently represent 14.2% of the country’s total annual emissions [30]. Weekly food consumption based on household expenditure is estimated to be responsible for an average of 80 kg CO2 week−1 household−1 [31]. This estimate excludes other greenhouse gases and additionally does not consider the emissions related to the act of food preparation in the home or food wasted. Food waste is estimated to contribute 6% of Australia’s total food-related GHGe [30].

Using estimates from recent national data on Australia’s food consumption patterns, the average Australian’s diet is responsible for a total of 19.7 kg CO2e person−1 day−1; however, the standard deviation varies considerably due to the highly varied dietary intakes of Australians [28]. The average estimated GHGe increased by 5.2 kg CO2e person−1 day−1 between 1995 and 2011 [27]. These differences over time can be explained by (1) an increase in average energy intake per person between dietary surveys (9400 kJ vs 10,224 kJ per day) as total energy intake has been found to be positively correlated with total dietary GHGe (r = 0.54 (p < 0.001)) [28] and (2) the use of an updated environmentally extended input–output (EEIO) model in the later study. Despite similar average energy intakes to Australians, the average NZ diet (mean 9103 kJ) has lower emissions, with a typical diet consumed by an adult male estimated to emit 10.1 kg CO2eq person−1 day−1 [36] and an average adult diet emitting 6.6 kg CO2eq person−1 day−1 [22•].

When the average Australian diet was compared to the ADG-recommended diet in two studies [18, 20••], the diets had similar GHGe (Table 1). Conversely, when Hendrie et al. [28] compared the GHGe of two other current dietary patterns to the recommended eating pattern for adults aged 19–50 years, the GHGe was approximately 5 kg CO2e day−1 higher for those adults who consumed a lower quality, higher GHGe diet (LQHE diet), compared to the recommended diet. In contrast, those adults who consumed a higher quality, lower GHGe (HQLE) diet had emissions 6.5 kg CO2e day−1 lower than the ADG-recommended diet. The HQLE diet comprised less than half the number of serves of discretionary foods compared to an average Australian diet and included an average of 0.2 serves less of milk and dairy foods.

The quantity of GHGe produced by a country is influenced by the composition of the population’s diet. In the ADG, foods are categorised as core foods (e.g. fruit, vegetables, cereals, lean meat, eggs, poultry, dairy) or non-core “discretionary” foods (e.g. sugar-sweetened beverages, alcohol, confectionary, processed meats). Table 2 provides an overview of the GHGe associated with consumption of core food groups and discretionary foods. Four studies provided data on the percent contribution to total food-related GHGe, three on GHGe per kg of food consumed (kg CO2e kg−1), and two on GHGe per day (kg CO2e day−1). There were some differences in categorisation of foods which makes comparisons challenging.

Table 2 Average GHGe from consumption of core food groups and discretionary foods in Australia and New Zealand
Full size table

Core foods are estimated to contribute 67–73% to total food-related GHGe in Australia [25, 27, 28], whereas discretionary foods are estimated to contribute between 27 and 33% of the total food-related GHGe in Australia and NZ [22•, 25, 27, 28]. Processed meats were the highest contributors (11–15%) [25, 28]. The high contribution from discretionary foods highlights not only the impact of the production of these foods on the environment but also the higher than recommended consumption of these low nutritional quality foods in Australia and NZ.

Of the core foods, 26–34% of contributions come from the meat and alternatives group in Australia. Fruit (3.5%) and vegetables (6.5%) were the two lowest contributors, but intake of fruit and vegetables currently falls far below recommended intakes nationally [28]. In contrast, in NZ, the highest contributions to food-related GHGe were from meat, seafood, and egg consumption (35%) followed by highly processed foods such as bakery items and ice cream (24%) [22•]. A similar pattern is observed for GHGe per kg of food consumed and GHGe per day with red meat, in particular beef being the main contributor to the GHGe of core foods. There is a difference in emissions for beef between Australia and NZ (NZ 21 kg CO2e kg−1 and Queensland beef 30.8 kg CO2e kg−1) [22•, 37]. The seafood group had the largest range in average GHGe per kg. Comparing emissions per kg of food may not be a particularly useful metric as overall contribution from that food will depend on the quantity consumed per serve, as well as the frequency of consumption. For example, discretionary foods may have a lower CO2e kg−1 value than red meat but may be consumed multiple times a day by some consumers.

Water Use/Footprints and Water Scarcity Footprints

In Australia, the amount of water required for food production is high, with 60% of the water available for human use being used for irrigated agriculture [39]. Of the eight million megalitres (ML) of water used for Australia’s agricultural production in 2018–2019, 1 million of this was used for fruit and nut crops, 882,000 ML for sugar cane, 388, 933 ML for vegetable crops for human consumption, and 75,600 ML for rice [40]. With most of Australia’s agricultural production exported (71%), total water used does not accurately reflect that used for food consumed in Australia [41].

Seven studies in Australia have looked at the impact of total dietary intake and/or food products on environmental indicators related to water, but differences in measures between studies make comparisons difficult. One approach is to measure total use, as reported above and the other similar measure is the water footprint (m3). There is a large variation in these measures across different food products. Reutter et al. [30] analysed the food system contribution to total water use in Australia using data from 2000. Approximately 60% of water used was for food production, with 13% embodied in the food as consumed, on average, by the Australian consumer [30]. Almost 5% is embodied in the food that is wasted by the consumer. In another study, researchers used national intake data to determine the contribution of consumption to water footprint. The average weekly water footprint of Australian households related to their food consumption was estimated at 35 m3 [31]. The three highest contributors to this water footprint, based on household expenditure data, were bakery products, flour and cereals (39%), meat (20%), and meals outside the home and fast food (16%) [31]. This is likely an underestimate of the water footprint for household food consumption as the researchers did not account for water used when households prepared or consumed the food at home.

In comparison, Candy et al. [18] found that the highest contributors to the overall 758 GL of water per year used in Greater Melbourne for food production were dairy (53.1%), beef and lamb (26.3%), followed by vegetables (8.2%), nuts (7.7%), and sugar (3.6%). All other food product groups contributed less than 2% each. Hadjikakou et al. [25] estimated the percent contribution to total blue water (i.e. irrigation) of different types of discretionary foods that were purchased by households per week. The three highest contributors were processed meat products (9.2%), alcohol (7.3%) and condiments, confectionary, food additives, and pre-prepared meals (7.3%). Researchers in two other studies used food expenditure or purchasing data. The relatively high water footprint in Australia (total: 2085 Mm3 year−1; per capita: 1082 m3 year−1) related to wheat consumption can help explain why the bakery products, flour, and cereals of food group category are the highest contributor [29]. Another example cited in our review is the water footprint of fresh mango, that has been estimated as 87 L kg−1 with 53% of the footprint associated with distribution and consumption waste [32].

Currently, Ridoutt et al. [33•] are the only researchers to determine the water scarcity footprint (WSF) of Australian adult dietary patterns. Ridoutt and colleagues [33•] addressed specific environmental concerns related to water use by estimating the associated WSF. The WSF for Australian adult daily diets averaged 362 L-eq person−1 day−1; however, this estimate is highly variable (SD = 218 L-eq person−1 day−1) due to wide differences in dietary patterns [33•]. Discretionary foods contributed 24.6% to the overall WSF of Australian adult diets, followed by fruits (18.9%), dairy and dairy alternatives (16.1%), bread and cereals (12.6%), and fresh meat and alternatives (11.6%) [33•]. When these findings are compared to those of Reynolds et al. [31], there are some differences in contribution between foods, such as fruit and nuts only contributing 2% to the total water footprint, but contributing 18.9% to the WSF.

Land Use, Ecological Footprints, and Cropland Footprints

The percent of land used for agriculture globally is 77% for livestock, meat, and dairy and 23% for crops (excluding feed) [6]. In Melbourne, land use related to agriculture was higher than the global average with beef and lamb consumption being responsible for 90% of land use [18]. The next two highest contributors were dairy (2.6%) and pig and chicken meat (2.2%) [18]. Consequently, when dietary patterns exclude these agricultural products, land use is lower. Candy et al. [20••] modelled two dietary patterns, a healthy mixed (HM) diet and a healthy plant-based (HPB) diet, compared to the current diet (TPWO). The HPB diet had the lowest land use, while the HM diet required 25% more land use than the current diet.

In terms of overall food consumption, research from the Australian Capital Territory (ACT) found that the impacts from apparent food consumption, estimated from food expenditure data, accounted for 50% of the territory’s total Ecological Footprint (2.12 million hectares) in 2017–2018, which is over nine times the size of the ACT [21]. One demand component of the Ecological Footprint is the cropland footprint. Ridoutt et al. [35], in accordance with best practice to address specific environmental concerns related to cropland use, estimated three types of cropland footprints: cropland scarcity footprint (CSF), cropland biodiversity footprint (CBF), and cropland malnutrition footprint (CMF).

The average CSF in Australia exceeds the global target (7.1 m2 year-e person−1 day−1 compared to 6.1 m2 year-e person−1 day−1) [35]. However, given the diversity of diets, in terms of overall energy intake and the types of foods consumed, many adults are below the target. If all Australian adults consumed the food choices evident in the high dietary quality-low cropland footprint (HDQ-LCF) dietary pattern (see Table 1), the CSF would decrease by around 20% and CBF (CBF) by 21%. On the other hand, for Australian adults whose diet conforms to the food choices and portion sizes recommended by the current ADG (i.e. increased core food intake and reduced discretionary foods), the CSF would increase by 0.3 m2 year-e per day, representing an overall increase of 4% [35]. Therefore, the estimated CSF still exceeds the global target.

In Australia, discretionary foods contribute the highest percentage share to the Ecological Footprint (35%) [25] and to cropland footprints (36%) [35], compared to other food groups. Ridoutt et al. [35] found that the second largest contributor to cropland footprints was the meat and alternatives group (23.9–27.4%), with poultry (9.5–11.7%), and beef and lamb (7.3–8.8%) being the main contributors within this food group. The third highest contributor was the grain (cereals) food group, with approximately 12% contribution to total cropland footprint.

Within the discretionary food category, the percent contribution to the Ecological Footprint was highest for processed meats and fattier/salty sausages (17.5%), followed by alcohol (7.0%), condiments, confectionary, food additives, and prepared meals (3.0%) and cakes, biscuits, puddings, and related products (2.9%) [25]. A similar pattern was seen for discretionary food product categories and contribution to CSF and CBF. The highest contributors to CSF were processed meat products (12.5%), alcoholic beverages (6.0%) and biscuits, cakes, waffles (3.3%) and muesli bars, confectionary, and chocolate (2.5%). For CBF, the contributions were similar, namely, processed meat products (10.1%), alcohol (5.5%), muesli bars, confectionary and chocolate (4.7%) and biscuits, cakes, and waffles (4.3%) [35].

Discussion

There was a range of approaches used to assess the environmental impacts of food and/or diets in the studies we reviewed. We have observed that the researchers tended to choose a single environmental indicator to assess the environmental impact of food and/or diets such as GHGe, cropland footprint, or water footprint. Aldaya et al. [42••] reported similar findings in their recent review which highlighted the most frequently used indicators and approaches for assessing sustainable healthy diets.

Interestingly, we were still able to see clear trends. For example, discretionary foods are consistently one of two highest contributors to environmental impacts across multiple metrics: GHGe, cropland footprints, ecological footprint, and water scarcity footprint [22•, 25, 27, 28, 33•, 35]. The meat and alternatives group also had a high environmental impact across multiple metrics although the water scarcity footprint was lower for this group compared to dairy products, cereals, grains, and fruit and vegetables [18, 20••, 22•, 28, 31, 33•, 35]. Fruits and vegetables generally had a low environmental impact in other metrics. However, as fruit and vegetable intake currently falls below recommendations in both Australia and NZ [22•, 28], this is unlikely to be the case when actual intake meets recommendations.

In Australia, most consumers sourced their food domestically in 2020 with only 11% of food consumption by value from imported food [41], a reduction from 15.4% in 2015–2016. The majority of these foods are processed products (9.6%) such as beverages and frozen vegetables. These imported foods play an important role in meeting consumer preferences for taste and variety and may not always have a higher environmental impact compared to domestically produced alternatives. Farmery et al. [23] found that imported seafood does not necessarily have a higher carbon footprint than domestically produced seafood. Imported foods can also provide supply when domestic production is impacted by drought, such as in the case of rice in 2019–20.

Poore and Nemecek [8] reported that moving from current diets to a diet that excludes animal products would reduce the food-related GHGe globally by 6.6 billion metric tonnes of CO2e. In the studies reported above which modelled different dietary patterns, this benefit was not clearly observed. Wilson et al. [36] estimated that the GHGe of a vegan diet in NZ was approximately 0.6 kg CO2e person−1 day−1 higher compared to a dietary pattern which minimised GHGe whilst achieving nutrient levels. In comparison, Candy et al. [20••] found that a plant-based diet in Australia had lower GHGe compared to other modelled diets. As the recommended diets for the two countries differed until recently (i.e. NZ recommended three serves of vegetables compared with Australia’s five serves), it could explain the difference in results. Additionally, NZ has low carbon emissions per kg of beef and lamb produced compared to the rest of the world; around 25% of the global average which could impact on the variation [43]. The dietary pattern modelling indicates there are alternative ways to reduce GHGe than just eliminating animal products such as meat, eggs, and dairy. For example in NZ, a 4% reduction was observed in GHGe when the eating pattern was shifted to meet the minimum recommendations of the NZ Dietary Guidelines [22•]. Therefore, rather than eliminating particular foods from a diet, a more realistic goal would be to aim to meet the dietary recommendations and choose mostly sustainable options.

The choice of metric used (or combination thereof) is important to consider as it may under-represent the environmental impacts of foods. Only four groups of researchers used metrics which accounted for multiple environmental impacts [21, 25, 32, 35]. One such metric is the Ecological Footprint. This tool is described as a useful resource-accounting tool that measures how fast individuals, groups of people, or activities consume energy and resources (including plant-based food products, livestock and fish products) compared to how fast nature can absorb our waste (carbon emissions) and generate new resources to replenish those that have been used [44]. The Ecological Footprint tracks the use of productive surface areas such as cropland, grazing land, and fishing grounds and is measured in global hectares (gHa). In contrast, the cropland footprint does not include grazing land for livestock and therefore may not accurately represent the environmental impacts from meat and dairy. Furthermore, not all water footprint calculations consider environmental relevance especially for Australia where variation in local water stress is extreme between regions [45]. The water scarcity footprint is a much more useful metric in Australia and could be a useful tool to identify appropriate regions for crop growing in the future.

Furthermore, some of the metrics used in the cited studies do not allow for differences in agricultural practices and the use of renewable energy. Lifecycle assessments (LCA) are commonly used for analysing the environmental impacts of agricultural products and consider metrics such as resource use, pollutant emissions, and land use. However, van der Werf, Knudsen, and Cederberg [46•] recently highlighted that the LCA method often does not account for alternative production methods and can misrepresent less intensive farming systems such as organic agriculture. Regenerative animal farming methods can have positive environmental impacts such as the integration of livestock into agricultural crops for manure, reduction or elimination of tillage and cover crop leading to improved soil health, increased sequestration of carbon, and increased biodiversity [47].

Another potential problem with using an LCA is the availability of specific country data as there are multiple factors that contribute to the difference between GHGe between countries and within locations in a country. The authors of the two NZ studies stated that there is a lack of available food-related GHGe and waste data and LCA analysis in NZ which resulted in both Drew et al. [22•] and Wilson et al. [36] utilising data estimates from the UK. This may have misrepresented the emissions from NZ’s food system particularly as Australia has the largest share of NZ’s food imports (and vice versa).

This rapid review synthesises the latest evidence on associations between dietary exposures and environmental outcomes in Australia and NZ. This evidence is highly relevant in the context of dietary guideline development. To inform the next iteration of the ADG, the findings of this review could be used as a starting point to summarise the environmental impact associated with current food consumption behaviours, and against which to make comparisons with alternative dietary patterns. Comprehensive modelling is required to investigate the nutritional adequacy of food substitution effects of more sustainable food choices [48]. For example, reductions in the recommended number of serves of meat could be balanced by increases in the recommended number of serves of legumes or nuts. Undoubtedly, to promote both health and environmental sustainability, dietary guidelines should continue to recommend reduced consumption of discretionary foods.

Conclusions

The results of this rapid review demonstrate that there is context-specific evidence available that describes the environmental impacts associated with food consumption in Australia and NZ. Most of the articles included in this review provided evidence related to the consumption of individual foods, food groups, or dietary patterns on GHGe. Whilst there are commonalities between different environmental indicators such as the impact of discretionary food consumption on CSF, WSF, and GHGe, there is wide variation in these indicators for other foods such as fruit. Modelling of current food consumption data to those dietary patterns recommended as being optimal for health does not necessarily result in an improvement in all environmental indicators. Thus, it is essential that environmental sustainability is considered in the revision of Australia’s Dietary Guidelines, as has been recently done for NZ, in order to ensure that the foods and dietary patterns that are recommended are associated with both positive health and environmental outcomes.