Safe and just Earth system boundaries

This work is an output of the Earth Commission, an independent international scientific assessment initiative hosted by Future Earth (https://earthcommission.org/). The synthesis presented here builds on recent work of the Earth Commission; other scientific literature, such as the PBs; intergovernmental reports, such as those of the IPCC; and World Health Organization guidelines. As the science component of the Global Commons Alliance (https://globalcommonsalliance.org/), the Earth Commission’s theory of change includes providing our results on ESBs to the Science-Based Targets Network, the Systems Change Lab and Earth HQ.

While we acknowledge that any scientific assessment will involve some subjectivity, we have taken several steps to ensure the scientific rigour of our ESBs. (1) Our analysis is founded on a rigorous evidence base (Safe ESBs and Supplementary Methods). (2) Where possible, we determine ESBs at multiple likelihood levels (for climate change, 0.5 °C for low likelihood of passing climate tipping points, 1 °C for moderate likelihood and so on) (Table 1). (3) The nomination process for the Earth Commission and its working groups was an independent process managed by Future Earth (Ethics and inclusion statement). (4) We report the confidence in our ESB assessments (Safe ESBs and Extended Data Table 1).

Safe ESBs

We used two main groups of approaches to setting safe ESBs: a ‘multiple elements’ approach and a ‘spatial aggregation’ approach. We describe these methods here in general terms, with technical details available in Supplementary Methods. These boundaries are aimed at protecting Earth system stability and life-support systems for as many species as possible, but they may not protect all species or all humans today, as further elaborated in our justice analysis.

For climate and biosphere, we assessed critical thresholds for a range of ‘elements’ relevant to each Earth system domain through literature review and modelling.

• For climate, we based our data on those found in a recent assessment of climate tipping elements combined with evidence on biosphere and cryosphere function and palaeoclimate variability (Supplementary Methods).

• For functional integrity, we synthesized the literature on the area needed to secure local NCP, including pollination, pest and disease control, water quality regulation, soil protection, natural hazards mitigation, and physical and psychological experiences (Supplementary Methods).

• For natural ecosystem area, we examined the Earth system NCP of carbon stocks, water flows and habitat for avoiding species extinction (Supplementary Methods).

From these sets of thresholds, we determined boundaries that avoid triggering climate tipping elements or maintain multiple local or Earth system NCPs at different levels of likelihood. To set the climate boundaries, we also used temperature ranges of previous Quaternary interglacials and temperature ranges that maintain biosphere and cryosphere functioning (Supplementary Methods).

For water and nutrients, we identified subglobal boundaries relevant to these systems and then converted them into global boundaries using models or simple aggregation.

• For surface water flows, we used an emerging consensus in the literature to set boundaries on the alterations (increase or decrease) to local-scale surface water flows that protect freshwater ecosystems and fisheries (Supplementary Methods) and applied this to the global land surface area. While the safe alterations can be summed to a global alteration budget, to ensure aquatic ecosystem protection, the safe ESB is best implemented and interpreted according to the subglobal boundary. To derive the safe levels of monthly flow alteration volumes for all land area globally, we analysed water balance model (WBM) runs coupled with the TerraClimate dataset of monthly climate forcings (Supplementary Methods has further information).

• For groundwater, our approach is based on preventing declines in local aquifer levels by setting the maximum safe average annual drawdown equal to the average annual recharge (Supplementary Methods). We estimated the annual groundwater recharge and drawdown for all land surface areas using Gravity Recovery and Climate Experiment satellite data covering the period from 2003 to 2016 coupled with data from the Global Land Data Assimilation National Oceanic and Atmospheric Administration Land Surface Model L4 v.2.1 (Supplementary Methods has more detailed information).

• For nitrogen, we used three regional environmental boundaries: significant disruption to freshwater ecosystems (from total N runoff), groundwater potability (from nitrate leaching) and terrestrial ecosystems (from atmospheric N deposition due to ammonia and nitrogen oxide emissions) across wide areas based on critical concentration limits for each. We mainly relied on a recent study following up previous works that extended the approach of the original PBs. This study used the Integrated Model to Assess the Global Environment (IMAGE) model to derive subglobal boundaries for critical nitrogen losses, surpluses and inputs based on critical concentrations in air and water and then aggregated these into global boundaries (Supplementary Methods has further information).

• For phosphorus, we relied on recent work that used literature-derived critical concentrations for avoiding eutrophication from P runoff to estimate global boundaries for P mined input and surplus based on a global budget calculation, taking into account P recycling, human excreta, soil and sediment retention, and global nutrient rebalancing.

Our approach for the safe aerosol boundaries does not fit neatly into these two categories because we used different methods for the subglobal and global boundaries. Our subglobal safe boundary uses the PB assessment of AODs that avoid tipping of regional monsoon systems. Our global assessment uses recent literature on the consequences of interhemispheric differences in aerosol concentrations on the global monsoon system (Quantifying ESBs and Supplementary Methods have further information).

As a reference for a ‘safe’ Earth climate system state, we used the interglacial Holocene epoch (that is, the state of the Earth system since the last Ice Age some 11,700 years ago. The Holocene’s exceptionally stable global climate system (oscillating <0.5–1 °C from the global pre-industrial 14 °C mean surface temperature) and its configurations of global hydrology, primary production of biomass, biogeochemical cycling and Earth system NCP were the fundamental prerequisites for human development as we know it. We argue that only within a Holocene-like interglacial climate can Earth continue to support human well-being, subject to consumption behaviours and population size. There is no evidence that billions of humans and complex societies can thrive in other known climates, such as a glacial ice age or ‘Hothouse Earth’.

We identified boundaries at multiple levels of likelihoods to reflect underlying scientific uncertainties and variabilities. These uncertainties included epistemic uncertainty in the boundary value for a specific Earth system process or component, such as a tipping element; variability in a boundary value across different places; and uncertainty when aggregating multiple subglobal boundaries into a global boundary. In some cases, these levels are presented with qualitative descriptors of each likelihood level; in other cases, they are presented as a central estimate with an uncertainty range, depending on the available evidence.

Some of our boundary quantifications use assessments of tipping elements since triggering tipping can endanger Earth system stability. Tipping elements commonly undergo changes that are abrupt (that is, faster than the forcing), large and difficult to reverse, although a particular tipping element may not display all three characteristics simultaneously (for example, table 4.10 in ref.). We identified boundaries based on tipping elements that accelerate or lock in change in the same Earth system component or process, such as climate tipping accelerating further climate change or triggering the inevitable loss of an ice sheet, or that trigger a tipping element in another Earth system domain, such as phosphorus concentration reaching a level that triggers eutrophication and disruption of freshwater ecosystems (Table 1).

Safe ESBs: confidence levels

We also assessed the levels of confidence in our safe boundaries (Extended Data Table 1). ‘Confidence’ in this context can be read as ‘degree of certainty in’ or ‘confidence in the validity of’ a specific ESB quantification. We use the same scheme for assessing and communicating confidence as the IPCC, which sets out two components: (1) robustness of the evidence base, judged as limited, medium or robust, considering its type, amount, quality and consistency and (2) degree of scientific agreement across the peer-reviewed literature and among the members of each Earth Commission Working Group, judged as low, medium or high. Based on these two dimensions, five qualifiers can be used to express the level of confidence in a particular ESB quantification: very low, low, medium, high and very high. This self-assessment is an expert judgement based on our understanding of the available literature.

Just (NSH) ESBs

We adopt an Earth system justice lens for both intrinsic and instrumental reasons. We show that some safe ESBs are not strong enough to protect humans and other species today and that we cannot achieve and live within the safe ESBs if inequality is high and resources are unjustly distributed. The evidence from behavioural experiments in public goods provision shows that perceptions of fairness significantly alter the outcomes of such experiments. In particular, individuals in disadvantageous positions insist on fairness even at the risk of large losses by doing so; such experiments suggest that climate change mitigation may not be achieved if rich countries are not perceived as pulling their weight. In common pool resource experiments, rising income inequality leads to a downward spiral of resource overexploitation and scarcity. In such experiments, viewing the problem in terms of fairness can lead to norms that motivate restraining from harvesting. A justice analysis is all the more needed as all science emerges from the value systems that apply in that domain, although these are often not made transparent.

Within the context of our Earth system justice approach, we use three justice criteria or the ‘3Is’: interspecies justice and Earth system stability (I1), intergenerational justice (I2) and intragenerational justice (I3). Our research into interspecies and multispecies justice reveals details regarding the scholarly approaches to these concepts, but there have been no attempts to operationalize these concepts deductively. In our research, we have combined interspecies justice with Earth system stability because Earth system instability undermines non-human species and inductively identified, through domain-specific (for example, climate, biosphere and aerosol loading) approaches, boundaries based on existing scholarship and the logic of that domain. Intergenerational justice refers to the justice between past and present generations (I2a) and between present and future generations (I2b). In general, although not always, our ESBs meet the I2b criteria because they protect future generations but not the present (I2a). Intragenerational justice (I3) combines justice between countries, communities and individuals through an intersectional lens. In balancing between the different justice criteria, we recognize that protecting future generations may impose many trade-offs with the use of resources today and that promoting intragenerational justice will also raise difficult issues regarding how to share resources, risks and responsibilities.

Our concept of harm derives from the justice literature and connects to the terms impact and risk used in the assessment literature. For example, IPCC defines risk as the potential for adverse consequences for human or ecological systems, including to lives; livelihoods; health and well-being; economic, social and cultural assets; infrastructure; services; and ecosystems. These risks are a result of exposure (the presence of people or other assets in regions of Earth system change or hazards, such as populations living near sea level) and of vulnerability (the propensity or disposition to be adversely affected, such as the poor who live in precarious homes or health status). Impact is defined by IPCC as realized risk or consequences. Our harm estimates are mostly based on exposure at different levels of Earth system change.

We recognize four caveats in the justice approach applied in this paper. (1) While staying within the just boundaries as set in this paper is crucial to avoid harm to significant sections of the human population, they are by no means guaranteeing just outcomes. Since just ends can be achieved with unjust means, meeting these boundaries without transformation could significantly harm current generations. (2) While harm to humans is caused in part by increased exposure to biophysical changes, we recognize that harm is also a function of people’s social–economic vulnerability and lack of adaptive capacities. This is beyond the scope of the present paper. (3) Our high levels of aggregation preclude systematic analysis of distributional justice issues in terms of which social subgroups are most harmed under what scenarios. (4) We do not explicitly address possible trade-offs between the three justice criteria. For example, policy instruments for achieving ‘I1’ may well undermine ‘I3’ (for example, limit access to resources for marginal people). Hence, we call for redistribution, liability and compensation.

Each safe ESB has been dealt with slightly differently, with some domains looking at when the system crosses tipping points (for example, climate change), others arguing that tipping points were crossed in the past and trying to recreate boundaries that allow species and systems to function (for example, surface water) and still others taking existing constraints into account in doing so (for example, groundwater). Although the proposals from a safe (and I1) approach fulfil I2b in that they makes space for future generations of humans, they may not guarantee safety for humans today (I2a; for example, climate change; hence, we call for more stringent targets), do not address local human exposure to pollutants (for example, air pollution; hence, we complement with local standards) or may limit access to resources (hence, calling for redistribution, liability, compensation and so on). Finally, while I2a has an explicit temporal dimension, intragenerational justice has an explicit spatial dimension and focuses on whether all people have access to minimum resources and services; how scarce resources are divided or shared between countries, communities and people and the varied justice issues that arise per domain; how environmental risks are spread worldwide and who is most exposed (through, for example, mapping exposure and vulnerability) and how responsibilities are shared between different actors.

To calculate the population exposed to different levels of climate change (Fig. 2), we draw on literature for exposure to sea-level rise at different levels of warming, as well as our own calculations of extreme heat based on output of global models. We acknowledge that these include a limited number of the possible impacts of climate change.

• Projections of sea-level rise need to account for dynamic processes of different complexity and for various spatiotemporal scales. In particular, the immediate response of several sea-level rise contributors (such as ice sheets and inland glaciers) to global warming is only marginal due to their high inertia but can be orders of magnitude higher on centennial timescales. Therefore, to draw a meaningful connection between selected temperature levels and triggered sea-level rise, recent literature has resorted to a twofold approach. The transiently realized sea-level rise throughout the twenty-first century is assessed by pooling Shared Socioeconomic Pathway and Representative Concentration Pathway scenarios by their end-of-century stabilization temperature. Those pools (for example, all scenarios that end up at 2 ± 0.25 °C) are used to drive localized models of sea-level rise, resulting in estimates for sea-level rise at 2100 for different end-of-century warming stabilization levels. Additionally, these twenty-first century projections can be complemented with multi-centennial estimates since long-term sea-level rise is governed by the equilibria of the cryosphere elements and ocean thermal expansion. In the next step, assessing exposure on these different timescales would require population projections, which are available for the twenty-first century but futile for longer timescales. For consistency, we therefore refer to a recent study that quantifies the number of people currently (baseline from that paper: 2010 population of 6.8 billion people) inhabiting land that is subject to inundation by end of this century or on a multi-centennial timescale, without accounting for potential adaptation through migration, coastal defences and so on.

• Wet bulb temperature (TW) exposure was calculated for the historical time period of 1979–2014 and the Shared Socio-Economic Pathway 2-4.5 future scenario for 2015–2100. Wet bulb temperature was calculated following the Davies-Jones method. Global gridded temperature and relative humidity data with a grid spacing of 1.25° × 1.25° at 6-h intervals were downloaded from a bias-corrected global dataset based on 18 models from the Coupled Model Intercomparison Project Phase 6 and the European Centre for Medium-Range Weather Forecasts Reanalysis 5 dataset. We aggregated the data to create a maximum daily TW dataset and then interpolated this to match the 1° × 1° grid spacing of the spatially explicit data for the 2020 population distribution (most recent available, global total 7.7 billion people) from the UN WPP-Adjusted Population Count, v.4.11. We then calculated the wet bulb exposure by summing up the population count for all cells with at least 1 day with a maximum TW > 35 °C. The TW threshold of 35 °C was chosen as it is often considered to be a human physiological limit of tolerance to heat stress. The human body is unable to cool itself beyond TW = 35 °C. An average 1 day per year over this temperature per year is therefore a conservative indicator in assessing human exposure to heat stress, which does not account for annual variability. We then plotted the total number of people exposed to 1 day with a maximum TW > 35 °C in a year against the mean annual global warming associated with that year to construct an exposure–temperature response curve.

• We calculate the number of people displaced from the human climate niche at different levels of warming, following the method of Lenton et al.. The number of people exposed to mean annual temperatures greater than 29 °C was calculated for different global mean temperature increases under four different Shared Socio-Economic Pathways. We used the downscaled spatially explicit output from the Coupled Model Intercomparison Project phase 6 available from the WorldClim v.2.0 database at 0.0833° (approximately 10-km) resolution (available at https://worldclim.org). The exposed population is based on a 2010 population of 6.9 billion with spatial distribution as given by the History Database of the Global Environment 3.2 database. The mean annual temperature threshold of 29 °C was chosen as it is beyond what humans have historically been exposed to.

To calculate current subglobal ESB transgressions (Fig. 3), we use data for the above wet bulb and low-elevation coastal zones as proxies for climate impacts, biosphere functional integrity (Supplementary Methods), surface water and groundwater (Supplementary Methods), exceedance of local safe and just nitrogen surplus and phosphorus concentration (Supplementary Methods) and PM2.5 concentrations. For population, we used the UN WPP-Adjusted Population Count v.4.11.

There are many uncertainties and limitations in this justice analysis. Lack of sufficient data on humans, communities and countries worldwide harmed by biophysical degradation is a key constraint. There is also considerable uncertainty regarding impacts on current generations, future generations, and specific countries and communities. In this paper, we also do not quantify issues of access, explore the implications of access for the safe and just corridor or discuss why it is difficult to meet issues of access without transforming our governance systems.