A consumption‐based analysis of extinction risk in Australia

Australia has an important role to play in protecting biodiversity, yet has a poor track record in preventing species extinctions. While invasive species and fire regimes are significant contributors to extinction risk in Australia, many species are threatened by habitat loss and other activities that support demand for products and services. Here, we connect the direct impact on in‐scope species to the final consumption ultimately supported by those activities, quantifying a territorial and consumption extinction‐risk footprint for each economic sector, state, territory, and local government area. We identify consumption of goods and services provided by the construction sector as the most significant consumption driver of extinction risk and show that 30% of Australia's extinction‐risk footprint is exported. Our findings highlight the importance of local government as the last line of defense against species extinctions and suggest that programs that influence consumption behavior may be required to reduce extinction risk in Australia.


INTRODUCTION
As one of the world's 17 megadiverse countries (Mittermeier et al., 1997), and with high levels of species endemism (Steffen et al., 2009), Australia has an important role to play in the conservation of biodiversity. It is well placed to do so financially, enjoying the 13th highest GDP in the world (World Bank, 2021), but its track record since European settlement has been weak (Ritchie et al., 2013;Woinarski et al., 2015). The broad scope of biodiversity, which includes ecosystem, species, and genetic diversity (Secretariat of the Convention on Biological Diversity, 2000), makes it difficult to succinctly quantify losses or gains; however, all three aspects of biodiversity are of con-fragmentation, such as agriculture, transportation, urban development, and mining, can be attributed to direct human activity (Kearney et al., 2019). This direct activity is often linked to economic transactions as natural resources are used as inputs to complex supply chains that ultimately feed, accommodate, move, and equip Australia and her trading partners. Local interventions such as protected areas and species recovery plans are important for moderating the impact of this direct activity; however, they are not sufficient (Galli et al., 2014). A more comprehensive approach to biodiversity conservation can be facilitated by understanding the consumption activities that ultimately create the demand for these direct activities.
Consumption-based analyses, using the input-output methodology first proposed by Leontief (1936), have identified the consumption drivers of biodiversity loss on a global scale. Using various proxies for the state of biodiversity such as land use, habitat loss, species threats, and species extinction risk, these analyses have highlighted the significant role of food production and consumption (Irwin et al., 2022;Kitzes et al., 2017;Marques et al., 2019;Wilting et al., 2017) and trade in agricultural commodities (Green et al., 2019;Lenzen et al., 2012) in driving global biodiversity loss. At a local scale, Bull et al. (2022) assessed the impact of a university's consumption patterns on global biodiversity, finding that the supply chains that support its research activities have the greatest impact.
To understand the consumption activities that contribute to extinction risk in one country, Australia, we reapply the methodology developed by Irwin et al. (2022) that relies on the comprehensive species data available in the International Union for Conservation of Nature (IUCN) Red List of Threatened Species (IUCN, 2021). We use these species data to quantify a metric for assessing species' extinction risk, to map the distribution of this metric across Australia, and to connect this extinction risk to consumption using input-output analysis, thereby calculating a territorial and consumption extinction-risk footprint for each in-scope species, sector, and region. These extinction-risk footprints reveal connections between the direct activity driving threats to Australian species and the consumption that creates the demand for that activity.
The consumption-based account of extinction risk that this analysis provides could be useful for quantifying supply chain impacts on biodiversity loss, which is increasingly referenced in reporting frameworks and legislation. The Taskforce for Nature-related Financial Disclosures (TNFD), for example, is expected to include the requirement that reporting entities consider both their direct impact on nature at the location of their operations and the indirect impact associated with their supply chain activities (TNFD, 2022). Similarly, changes to the European Union's Sustainability Reporting Standards, which come into force in 2024, expand the scope of environmental factors reported against to incorporate biodiversity, and require any reporting entity to report on its own operations and its supply chains across all countries (European Parliament, 2022).

Data
Comprehensive species data were obtained from the IUCN Red List of Threatened Species version 2021-2 (IUCN, 2021), which contains information on each species' distribution, habitat preferences, extinction risk category, and the threats acting on it, including details on the scope, severity, and timing of each threat. We rely on the IUCN Red List rather than the national threatened species list, collated under the Commonwealth Environmental Protection and Biodiversity Conservation Act (EPBCA) (Department of Agriculture, Water, and the Environment, n.d.), or any state-or territory-specific threatened species lists, to ensure a consistent classification of extinction risk across the country and to utilize the detailed threat, habitat, and distribution information available in the IUCN Red List. The species scope was limited to comprehensively assessed species to avoid any bias driven by incomplete assessments. Only species assessed as being threatened with extinction, indicated by an extinction risk category of Near Threatened (NT), Vulnerable (VU), Endangered (EN), or Critically Endangered (CR), were included, a total of 276 species. This scope was further reduced to those species with a terrestrial area of habitat (AOH), leaving 239 mammal, bird, and amphibian species in scope (30 CR, 54 EN, 73 VU, and 82 NT).

Area of habitat maps
Range data polygons for each species were downloaded from the IUCN Red List (IUCN, 2020a) and information on elevation and habitat preferences (where present) was noted. Australian Land Cover Classifications (Lymburner et al., 2015) were grouped to match the IUCN Habitat Classifications (IUCN, 2012) as closely as possible (Supporting Information S1). Contour data for Australia were obtained from Geoscience Australia (2004) to provide elevation information, and shapefiles that captured the intersection of range data, elevation, and habitat preferences were generated using Geographic Information System software, providing an AOH map for each species. The digital boundaries of each local government area (LGA) were sourced from the Australian Bureau of Statistics (2020), and the intersection of the species AOH maps and LGA boundaries was used to calculate the percentage of each species' AOH found within each LGA. Mapping species information by political boundaries rather than using land tenure types (Kearney et al., 2022) or natural boundaries such as catchments (Evans et al., 2011) and biogeographic regions (Cresswell & Murphy, 2016) supports both the connection into economic activity needed to implement an input-output analysis and the identification of policy and management interventions at all three levels of government.
LGA boundaries are wholly contained within state and territory boundaries, allowing the aggregation of data to a subnational and national level, as needed.

Quantifying extinction risk
To obtain a numeric representation of species extinction risk, we used a modified version of the Species Threat Abatement and Restoration metric (Mair et al., 2021), known as the nSTAR metric. This unit-less metric was calculated using the comprehensive threat information for each species maintained in the IUCN Red List (IUCN, 2021) that is converted to a numeric value following the methodology outlined in detail in Irwin et al. (2022) and briefly summarized here. First, the extinction risk category for each species was assigned a numeric value, with CR = 4, EN = 3, VU = 2, NT = 1, and all other categories = 0. Next, a Threat Impact score was calculated for each threat acting on each species, based on the scope and severity information recorded in the IUCN Red List and referencing the values in Supporting Information S2. Threats occurring in the past were excluded, and the median possible value for scope and severity was used where there was no information on these attributes recorded (Mair et al., 2021). The nSTAR value was calculated for each threat acting on each species by multiplying the Threat Impact score by the numeric representation of that species' extinction risk category.
Once the nSTAR value was calculated for each speciesthreat combination, it was allocated to economic sectors using a sector to threat concordance generated by assessing the information on each threat available in the IUCN Threat Classification Scheme (IUCN, 2020b), connecting it to the relevant economic sectors used in Australia's National Accounts (Australian Bureau of Statistics, 2012), and normalizing the concordance generated based on the size of each economic sector (Supporting Information S3). Finally, these species-sector nSTAR values were allocated to each LGA based on the percentage of each species' AOH falling within the boundaries of the LGA. This generated a 239 × 1284 matrix for each of Australia's 545 LGAs, ready for further manipulation into a satellite block for use in input-output analysis.
The allocation of the nSTAR value to economic sectors, necessary to undertake input-output analysis, excluded those threats that cannot be directly connected into economic activity, such as invasive species, disease, and fire regimes, from the calculations. As a result, the final nSTAR value does not necessarily represent the total extinction risk of each species from all threats. When aggregated at a regional level, this nSTAR value provides a territorial extinction-risk footprint for the region in question and quantifies the direct, or productionbased, impact on the species within that region. Inputoutput analysis is then used to calculate a consumption extinction-risk footprint, quantifying the indirect, or consumption-based, impact on species.

Input-output analysis
Input-output analysis allows the connection of a direct environmental impact, in this case the extinction risk quantified by the nSTAR value, to the final consumption of products and services that may be geographically far removed from that environmental impact (Leontief, 1970). The methodology uses National Accounts data, which document the many interactions between actors within an economy, to trace the flow of an environmental impact from its source to the location of final demand. We used the Australian National Accounts Input-Output tables for 2017, curated within the Australian IELab (Lenzen et al., 2017), to carry out this analysis. This provided us with monetary transaction data contained in three matrices, including the intermediate transaction matrix (T), which provides information on all transactions between sectors within the economy. The second matrix, the value-added matrix (V), provides information on other inputs to production required by each sector, such as compensation of employees, taxes, and imports. The third matrix, the final demand matrix (Y), provides information on the value of transactions at the point of final consumption, where the final consumer spends their money on each sector. The 239 × 1284 matrices generated for each LGA were manipulated using sector and regional aggregators to create a satellite block (or Q matrix) that matched the sectoral structure of the intermediate transactions matrix (T), providing the final input needed to calculate the consumption extinction-risk footprint.
To derive the consumption extinction-risk footprint, first the total output vector (x) of the input-output table was calculated by adding the summations of each of the T and Y matrices, then diagonalized to generatê, which was inverted to givê− 1 . The direct requirements matrix A was then calculated by multiplying T and̂− 1 , and with an identity matrix (I) with dimensions equal to T, the Leontief Inverse (L) was calculated using Equation (1): The Leontief Inverse captures the full supply chain interdependencies within the economy, and the nSTAR value attributable to each $1 of total output from each sector was captured in the direct intensity matrix (q), calculated by multiplying the satellite block Q bŷ− 1 . The consumption extinction-risk footprint value for each sector (k) was calculated from this direct intensity matrix (q), the Leontief Inverse (L), and the final demand for the sector (Y k ) using Equation (2): Further details on input-output methodology, including its limitations, are included in Supporting Information S4.

RESULTS
Two approaches were used to account for species extinction risk in Australia. The first, a production-based accounting approach, provides the territorial extinctionrisk footprint and quantifies the direct impact of economically induced threats acting on the species located within each region. The second, a consumption-based accounting approach, provides the consumption extinction-risk footprint and quantifies the impact of final consumption of the products and services provided by a given sector or location on all species. At a total species or global level, the values of the territorial and consumption extinction-risk footprints will be equal.

Territorial extinction-risk footprints by LGA
All of Australia's LGAs are home to at least one threatened or Near Threatened (NT) species, with Clarence Valley in New South Wales hosting the most at 64. Figure 1a maps the number of threatened or NT species occurring in each LGA, with the darker color representing a higher number of species. The territorial extinction-risk footprint, which considers the extinction risk category of each species within the LGA, along with the number, scope, and severity of threats acting on it, is mapped in Figure 1b. This more elaborate measure shows that Australia's territorial extinction-risk footprint is concen-trated in Far North Queensland, with Mareeba, Douglas, and Tablelands LGAs together custodians of 12.7% of the national territorial extinction-risk footprint. In contrast, 17 LGAs within the Greater Adelaide area represent only 0.03% of Australia's territorial extinction-risk footprint. Detailed results by state, territory, and LGA are available in Supporting Information S5.

Consumption extinction-risk footprints-Sector results
Consumption extinction-risk footprints were calculated according to the sector of final demand, the point at which households, governments, or exporting entities purchase a product or service. The impact of any intermediate economic transactions that take place before the point of final demand, for example, when wheat is purchased by a flour mill so it can supply a bread manufacturer, will be recorded against the sector of final demand, in this case food and beverage products. Figure S1 provides a visualization of the change in sector contribution as the extinction-risk footprint flows from the sector that generates the direct, or production-based, impact to the sector that generates the indirect, or consumption-based, impact.
Consumption of products and services provided by the construction sector is a significant driver of economically induced extinction risk in Australia, representing 22% of the consumption extinction-risk footprint, followed by trade, repairs, and hospitality at 14% and food and beverage products at 11%. Figure 2 illustrates the contributions made by consumption of each sector to Australia's consumption extinction-risk footprint. These results differ from the global extinction-risk footprint distribution where food and beverage consumption has the greatest contribution, at 20% of total, and the construction sector generates 16% of total (Irwin et al., 2022). Further details on the consumption extinction-risk footprint, including results by Class, are available in Supporting Information S7.
Demand originating in other countries generated 30% of the extinction risk experienced by Australian species, with exports constituting 90% of the consumption extinctionrisk footprint associated with demand for products and services provided by the mining sector, and 100% of the consumption extinction-risk footprint associated with final demand for broadacre crops.

Consumption extinction-risk footprint-Geographic results
Each region is a custodian of the species within its borders, and its territorial extinction-risk footprint represents F I G U R E 1 The distribution of threatened or Near Threatened species and territorial extinction-risk footprint across Australia. Local government areas are color coded according to (a) the number of threatened or Near Threatened species found within their boundaries and (b) the size of the territorial extinction-risk footprint, a measure of extinction risk that takes into account a species' extinction risk category and the scope and severity of each threat acting on it. The darker the color, the higher the relevant value. Note that there are no units for extinction-risk footprint.

F I G U R E 2
Australia's extinction-risk footprint by consumption sector. Each sector's footprint quantifies the contribution that consumption of the goods and services produced by that sector has on the extinction risk of all in-scope species. Numbers in orange indicate the percentage of each sector's consumption extinction-risk footprint that is driven by exports. the impact of consumption from all locations on the threatened and NT species within the region. Each region is also a consumer of products and services that induce direct threats on species across all regions, and this impact is quantified in its consumption extinction-risk footprint. Figure 3 illustrates each state and territory's contribution to Australia's territorial and consumption extinction-risk footprint, and, for comparison, each state and territory's share of Australia's population and land area. Victoria, South Australia, and the Australian Capital Territory are net importers of extinction-risk footprint from other regions, with their consumption footprint greater than their territorial footprint. In contrast, Queensland, Tasmania, and the Northern Territory are net exporters of extinction-risk footprint. Once population sizes are considered, however, the per-capita consumption footprints in the Northern Territory (426 per million people) and Tasmania (421 per million people) are well above the national average of 245 per million people. As previously noted, 30% of the consumption extinction-risk footprint is exported to other countries, embodied in products and services as they are exported.
Exploring the differences between territorial and consumption extinction-risk footprints at a LGA level reveals that the consumption extinction-risk footprint is concentrated in metropolitan areas, in line with the concentration of population. Figure 4a maps the consumption extinctionrisk footprint by LGA, with the color gradation becoming darker as the footprint value increases. The ratio of consumption footprint to territorial footprint illustrates the shift of extinction-risk footprint from the location of direct impact to the location of consumption and is mapped Australia-wide in Figure 4b, and for selected LGAs in Figure 5. Residents of those LGAs with a ratio greater than 1 (color coded blue in Figure 5) contribute to the extinction risk of Australian species through their consumption, even though their everyday activities may have a minimal direct impact on species. The extinction risk impacts that have accumulated through the supply chains that deliver the products and services consumed by those residents are F I G U R E 3 Each region's contribution to Australia's total area, population, consumption footprint, and territorial footprint. The territorial footprint represents the impact of all locations' consumption on the species within each state or territory's borders, while the consumption footprint represents the impact of each region's consumption on all species in Australia. borne by species that may be far removed from the final consumption location. Detailed results by state, territory, and LGA are available in Supporting Information S5.

Extinction-risk footprint-Species results
At a species level, the territorial extinction-risk and consumption extinction-risk footprints are equal when aggregated across all sectors and locations, although as noted in Methods, these footprints do not necessarily represent the total extinction risk of each species from all threats. The highest extinction-risk footprint was found for the Orange-bellied Parrot (Neophema chrysogaster), a CR bird (BirdLife International, 2018). The extinction risk attributable to economic activity for this species represents 49% of the total, and 33% of this was generated by demand for the construction sector. This demand drives extinction risk through residential and commercial development, which is expected to cause very rapid declines in the population (BirdLife International, 2018).
Residential and commercial development, along with transportation and service corridors, was also a significant threat acting on one of Australia's most charismatic species, the Koala (Phascolarctos cinereus) (Woinarski & Burbidge, 2020). The extinction risk attributable to economic activity for the Koala represented 19% of the total, and of this, 58% was generated by demand for the outputs of the construction sector, 40% generated by demand for the Queensland construction sector alone. Most of the consumption impacting this species originated in Australia, with only 12% of the extinction-risk footprint exported.
Demand for the food and beverage, crops, and beef sectors contributed 35% of the economically induced extinction-risk footprint for the Australian Painted-snipe (Rostratula australis), which represented 69% of the total for this EN bird dependent on wetlands (BirdLife International, 2016). Agroindustry farming and abstraction of surface water are the most severe threats acting on this species, activities that are undertaken to meet the demand for food and crop products. Exports contributed 37% of the economically induced extinction-risk footprint for this species, with 11% attributable to the export of broadacre crops.

DISCUSSION
The construction industry is ranked third in terms of industry contribution to Australia's economy, contributing 9.7% of GDP in 2019-2020 (Australian Bureau of Statistics, 2021). Our analysis shows that this economic activity comes at a cost to Australia's species, with 22% of the national consumption extinction-risk footprint generated by demand for this sector's products and services. This footprint is driven in part by the prevalence of residential and commercial development as a threat, which affects F I G U R E 4 The distribution of consumption extinction-risk footprint across Australia. Local government areas are color coded according to (a) the size of the consumption extinction-risk footprint and (b) the ratio of consumption extinction-risk footprint to territorial extinction-risk footprint, indicating the shift of extinction-risk footprint from the location of direct impact to the location of consumption. The darker the color, the higher the value.

F I G U R E 5
The distribution of consumption extinction-risk footprint for selected local government areas (LGAs).
LGAs are color coded according to the ratio of consumption extinction-risk footprint to territorial extinction-risk footprint for: (a) Sydney and surrounds; (b) Melbourne and surrounds; (c) Southeast Queensland and northern New South Wales; (d) Adelaide and surrounds; (e) Perth and surrounds; and (f) Tasmania. The darker the color, the higher the ratio, indicating a shift of extinction-risk footprint from the location of direct impact to the location of consumption.
23% of the 239 species in scope and has a direct connection to construction activity. Demand for construction also drives extinction risk via the accumulation of small impacts within the supply chains that support the sector. All construction materials are extracted from nature at some point-trees are felled to produce timber, ores are mined and converted to metal products-and other resources are consumed to facilitate these conversions, bringing with them additional extinction-risk impacts. Our findings differ from previous global studies, which found that food consumption is the highest consumption driver of biodiversity loss (Irwin et al., 2022;Kitzes et al., 2017;Marques et al., 2019;Wilting et al., 2017), indicating the importance of undertaking country-level consumption-based analyses to identify effective policy and advocacy responses to address biodiversity loss. In Australia, the significance of the construction sector as a contributor to extinction risk supports the recommendation from Samuel (2020) that the federal EPBCA should be revised to consider the cumulative impacts associated with development, in addition to the current large-scale, project-by-project approach. Our findings indicate that an industry-specific assessment could provide meaningful insights into these cumulative impacts and inform legislative changes at both the federal and state levels.
Local governments also have an important role to play in safeguarding threatened species since it is at this level that most development and construction projects are approved, effectively establishing local governments as the last line of defense for species. Those LGAs with high territorial extinction-risk footprints (color coded blue in Figure 1b) are custodians of high numbers of species that are close to extinction or under threat from multiple activities, so any development decisions are likely to have a direct impact on species. More rigorous criteria could be applied when assessing development applications in these locations, and rezoning undeveloped land based on species distribution maps could protect critical habitat from further development. In LGAs with high consumption extinction-risk footprints (color coded blue in Figures 4 and 5), it may be more effective to focus on individual consumption behavior as a means of reducing species extinction risk. The flow of territorial extinction-risk footprint to consumption extinction-risk footprint is most pronounced in metropolitan LGAs, consistent with their higher population densities, indicating that a change in behavior at an individual level within those LGAs could have a meaningful impact on reducing the extinction-risk experienced across Australia. Behavioural domains targeted at reducing biodiversity loss include those focused on lifestyle, stewardship, social, consumption, advocacy, and donation behaviors (Selinske et al., 2020). Within the domain of consumption behaviors, certification schemes have been shown to be effective in influencing consumption choices, and research into labeling schemes has demonstrated that many consumers are willing to accept a price premium in return for purchasing wildlife-friendly products (Herring et al., 2022;Selinske et al., 2020).
The introduction of a certification scheme related to species extinction risk could be an effective mechanism to reduce the impact of consumption on Australia's species. Building on current schemes such as those managed by the Forest and Marine Stewarship Councils, such a certification scheme could be extended to all construction materials and food products. Given that the construction; food and beverage; and trade, repairs, and hospitality sectors currently contribute 47% of the consumption extinctionrisk footprint in Australia, such end-consumer-focused initiatives, partnered with local government regulations regarding development, could reduce species-threatening activities across Australia.
Australia has an important role to play in protecting biodiversity. Conservation measures directed at the location of biodiversity loss are necessary, but they may not be sufficient if the underlying demand for activities that threaten species is not also addressed. An understanding of the location and sectors of consumption that drive biodiversity loss, such as that provided by the methodology presented here, can identify further interventions to mitigate that loss. Combining these mitigation actions with those focused on threats that are not driven by economic activity, such as invasive species and fire regimes, will provide a more comprehensive arsenal with which to reduce species extinctions and support the conservation of Australia's unique biodiversity.

A U T H O R C O N T R I B U T I O N S
Amanda Irwin and Arne Geschke conceived the study. Amanda Irwin carried out data manipulation and analysis and wrote the first draft. All authors edited and approved the manuscript.

A C K N O W L E D G M E N T S
The authors wish to thank Sebastian Juraszek of the University of Sydney for his management of the computer infrastructure that supported this work. This research is supported by an Australian Government Research Training Program (RTP) Scholarship.

C O N F L I C T O F I N T E R E S T S TAT E M E N T
The authors declare no conflicts of interest.

D ATA AVA I L A B I L I T Y S TAT E M E N T
The IUCN Red List of Threatened species is available at https://apiv3.iucnredlist.org/. The sector to threat concordance used in this research is available at https://hdl. handle.net/2123/26254. O R C I D Amanda Irwin https://orcid.org/0000-0001-8374-9284