Quantifying and categorising national extinction-risk footprints
Previous studies have used number of species threats6,7, countryside species-area relationship1,3,17, and potentially disappeared fraction of species4 to quantify biodiversity loss. We introduce the non-normalised Species Threat Abatement and Restoration (nSTAR) metric as the quantifiable representation of biodiversity loss in our analysis, a unit-less, species-centred metric which relies on detailed information curated in the IUCN Red List of Threatened Species11. On its own, this metric can be used to support production-based accounting of the extinction risk of species and identify the most significant threats at a specific location to inform direct interventions26. However, once manipulated into a structure that allows it to be appended to a multi-region input–output (MRIO) table, an environmentally-extended MRIO can be created. This unlocks the power of consumption-based accounting of this extinction risk, connecting the direct environmental impact with the consumption which ultimately induces it.IUCN Red List of Threatened SpeciesThe IUCN Red List version 2020–211 provided information on extinction risk for over 122,000 species and details of the threats acting on those species, including the threat classification, scope, timing, and severity. The species scope was limited to comprehensively assessed terrestrial species, ensuring that only species which have been assessed across all countries were included, and thus eliminating any geographical bias introduced by incomplete assessments27. Species with an extinction risk category of Near Threatened (NT), Vulnerable (VU), Endangered (EN), or Critically Endangered (CR) were included. Three species were excluded to avoid double counting where two different extinction risk categories were provided for the same species, leaving 5295 amphibian, mammal, and bird species in scope.The information contained in the IUCN Red List regarding the threats facing each species is crucial, since many of these threats are attributable to economic activity28,29. Specialist assessors are required to assign one or more of 118 different threat classes to each species’ record, with additional documentation of the severity, scope and timing of each threat recommended, based on the impact of that threat on the species’ population30. To connect this threat information to economic sectors, a key requirement for input–output analysis, background information on threat classes was sourced from the IUCN Threats Classification Scheme version 3.229. Each threat was assessed for connection to each of the 6357 economic sectors classified in the UN Statistics Division Central Product Classification Standard31, based on the likelihood that activity associated with each sector directly contributes to the threat being assessed. As an example, the economic sectors associated with rice cultivation were allocated to the threats grouped under IUCN Threat Class 2.1—Annual & perennial non-timber crops. A total of 55 out of 118 threats were allocated to at least one economic sector, with higher-level threat classes excluded from this allocation if information was available for the associated lower-level threat classes to avoid double counting. Species threats driven by activity that cannot be attributed to an economic sector, such as invasive species, were not allocated to any sectors and as a result, the extinction-risk footprint does not necessarily represent the full magnitude of extinction risk for each species. While not all threats were allocated to an economic sector, all economic sectors were allocated to at least one threat. Further details on the connection of economic sectors to threats are available in Supplementary Note S5, which includes a link to the detailed 6357 × 118 binary concordance matrix used to execute these sector-threat allocations.The IUCN Red List also requires inclusion of a range map and habitat classification, which were combined with remote sensed land cover and elevation data to generate a high-resolution area of habitat (AOH) map for each in-scope species32,33. These maps, reapplied from Strassburg et al.34, were used to calculate the percentage of each species’ AOH present in each country.Quantifying biodiversity loss: the nSTAR metricThis detailed information from the IUCN Red List was used to calculate the nSTAR metric, which quantifies each threat’s impact, rather than just its presence, on each species. Adapted from the newly developed Species Threat Abatement and Restoration metric (STAR)26 by removing the normalisation step, the nSTAR metric, which has no units, was calculated for each species in two stages.First, a numeric representation of each species’ extinction risk category (Wi) was determined, following the equal steps methodology introduced by Butchart et al.35. Extinction risk categories of Data Deficient (DD) and Least Concern (LC) were assigned Wi = 0, Near Threatened (NT) was assigned Wi = 1, Vulnerable (VU) was assigned Wi = 2, Endangered (EN) was assigned Wi = 3, and Critically Endangered (CR) was assigned Wi = 4.Next, a Threat Impact score (TSij) for each threat (j) acting on a species (i) was determined based on the scope and severity information recorded for that threat, according to the values set out in Table 1, which are adapted from those proposed by Garnett et al.36. Reapplying the methodology of the STAR metric, where no value was recorded for the scope or severity of a threat, the median possible value for these were used, and only threats noted as Ongoing or Future were included. Further details on these methodological choices and sensitivity analyses to support them are available in Mair et al.26.Table 1 Numeric representation of threat information.Full size tableThe numeric nSTAR value for each species-threat combination (ij) was calculated by multiplying the value representing the species’ extinction risk category (Wi) by the Threat Impact score (TSij) for that threat:$${text{nSTAR}}_{ij} = W_{i} *TS_{ij}$$
(1)
The total nSTAR for species (i) can be calculated by multiplying the extinction risk category value (Wi) for that species by the sum of all Threat Impact scores for the species:$${text{nSTAR}}_{i} = W_{i} *(TS_{i1} + TS_{i2} + TS_{i3} + cdots + TS_{ij} )$$
(2)
Once calculated according to Eq. (1), the nSTARij value for each species-threat combination was allocated to economic sectors using the 6357 × 118 sector-threat concordance (available in Supplementary Note S5), which was normalised based on the economic size of each sector. Finally these nSTAR values, derived for each species-sector combination, were allocated to each country based on the country’s share of the AOH for that species, calculated from the intersection of the species’ AOH map with each country’s borders34.The nSTAR metric introduced here differs from the STAR metric from which it is adapted in that the normalisation step executed at this point in the STAR methodology is omitted. This ensures that the nSTAR metric is both additive and independent across all three dimensions of species, country, and economic sector, a necessary condition for use in input–output analysis. The STAR metric normalises the total value calculated in Eq. (2) to ensure that the total STAR value for any species is equal to Wi * 100, resulting in all species with the same extinction risk category being allocated the same STAR value regardless of the number of threats acting on them26. This normalisation facilitates the aggregation of the STAR metric by species taxonomy however it is problematic when aggregating the STAR metric by threat, since the STAR value attributed to each species-threat combination will be dependent not only on the characteristics of that threat, but also on the number and characteristics of other threats acting on the species. This dependence on more than one variable in the calculation of the STAR value for each species-threat combination means that it is not suitable for aggregation by threat and, by extension, economic sectors once the threat to sector allocation has been carried out.In order to provide a metric which can be aggregated and disaggregated across species, sector, and country hierarchies the nSTAR methodology excludes this normalisation step. Consistent with the STAR methodology, the nSTAR metric is calculated using numeric values only and therefore has no unit of measure26.Input–output analysisOnce calculated, the nSTAR metric was partnered with the global supply-chain data available in the 2013 Eora MRIO, chosen for its extensive coverage of 190 regions (189 countries and one ‘rest of world’ region) and between 26 and 1022 economic sectors in each country, depending on the level of detail in each country’s publicly available National Accounts12.A satellite block, or Q matrix, was created using the nSTAR values for 5295 species across 6357 economic sectors for 190 regions. This satellite block was then aggregated to match the sectoral structure of the Eora MRIO, a total of 14,839 country-sector combinations. A process flow diagram to illustrate the stages of data manipulation required to convert the IUCN Red List data to a satellite block ready for use with the Eora MRIO is included in Supplementary Fig. S5.The Eora MRIO provided the intermediate transaction matrix T, the final demand matrix Y, and the value-added matrix V. Consumption based footprints were calculated by connecting the nSTAR value captured in the satellite block Q to the final demand matrix Y following Leontief’s methodology9,10. Central to this methodology is the Leontief Inverse L, a concise mathematical representation of the interdependencies across all economic sectors, which is expressed as:$${mathbf{L}} = left( {{mathbf{I}}{-}{mathbf{A}}} right)^{{ – {1}}}$$
(3)
where I is an identity matrix with dimensions equal to the those of the intermediate transaction matrix T, and A is the direct requirements matrix, derived from the T matrix in a number of stages. First the total output vector x is calculated, then diagonalised, and the inverse calculated to derive ({widehat{mathbf{X}}}^{-1}), which returns the direct requirements matrix A when multiplied by T.Next the satellite block was converted into an intensity matrix q by multiplying Q by ({widehat{mathbf{X}}}^{-1}) to calculate the nSTAR value attributable to each dollar of total output produced by each sector. Once the q, L and Y matrices are available, the consumption extinction-risk footprint for a sector k (fk) can be calculated using Eq. (4):$${mathbf{f}}_{k} = {mathbf{q}}*{mathbf{L}}*{mathbf{Y}}_{k}$$
(4)
where Yk represents the final demand for that sector. Consumption extinction-risk footprint values were generated for each species-sector-country combination, a total of more than 78 million datapoints.Further matrix manipulation was used to calculate the country-level imported, exported, and domestic extinction-risk footprints. For each country the final demand matrix, Y, was separated into two matrices, Ydom, representing demand from that country for the economic sectors in that country, and Yoth, representing demand from all other countries for the economic sectors in that country. Next, the intensity matrix, q, was separated into two matrices, qdom, representing the nSTAR intensity for each of the species within that country’s borders, and qoth, representing the nSTAR intensity for all remaining species. The three sub-footprints for each country were calculated using Eqs. (5), (6) & (7). A simplified illustration of this methodology is included in Supplementary Fig. S3.$${mathbf{f}}_{dom} = {mathbf{q}}_{dom} *{mathbf{L}}*{mathbf{Y}}_{dom}$$
(5)
$${mathbf{f}}_{exp} = {mathbf{q}}_{dom} *{mathbf{L}}*{mathbf{Y}}_{oth}$$
(6)
$${mathbf{f}}_{imp} = {mathbf{q}}_{oth} *{mathbf{L}}*{mathbf{Y}}_{dom}$$
(7)
Imported, exported, and domestic extinction-risk footprints were calculated for 188 countries.LimitationsWhile very powerful in unravelling the intricacies of the global economy, there are limitations to the effectiveness of input–output analysis. Since it relies on National Accounts data, only activity which can be directly connected into reported economic activity is captured. This means that any activities that are not transacted within the boundaries of the formal economy, such as subsistence hunting and illegal logging, will be excluded unless they have been incorporated into the relevant country’s National Accounts data. The exclusion of threats due to their timing or non-economic classification (such as geological events, disease, and invasive species) resulted in a zero nSTAR value for 519 species, leaving 4776 species with a material nSTAR value. In addition, any limitations in the sector categorisations, their spatial and technological homogeneity, or assumptions included in the allocation of economic activity to sectors within the National Accounts data in each country will be propagated through to the footprint calculations. These limitations are common to consumption-based analyses5,6,7,17,25 and we do not further address them here.Further limitations exist with the use of the scope and severity data for each threat captured in the IUCN Red List, since this does not take into account interaction between threats, or between the severity and scope of an individual threat36. As a result, the impact from a single threat acting on a species may be overstated, and higher nSTAR values attributed to that species than would otherwise be warranted. In addition, any variations in the location, scope, or severity of threats acting across a species’ distribution range are not captured and thus the impact of different economic sectors may be over or under-represented26.There is a temporal displacement between the economic activity and the extinction risk used in this analysis. The extinction risk category assigned to each species is due to the cumulative sum of current and historical impacts acting on it, while the value of economic interactions used to trace this extinction risk through the global economy is based on one year of activity. This is typical of related approaches1,6, and may not introduce much uncertainty given that current economic activity is higher than at any time in history37. Nevertheless, there is no doubt that some current extinction risk is due to past economic activity and development of methods to incorporate this temporal dimension would be a valuable research avenue. More
