Planetary boundaries

Planetary boundaries are proposed limits to human pressures on nine processes that regulate the stability and resilience of the Earth system. Together, they define a “safe operating space for humanity,” intended to preserve conditions broadly resembling those of the Holocene, the approximately 11,700-year interval during which agriculture and complex societies developed. Introduced in 2009 by an international research team led by Johan Rockström, the framework treats environmental change as a problem of interacting planetary processes rather than separate environmental pressures. Crossing a boundary indicates increasing risk, not necessarily an immediate tipping point or inevitable collapse. (stockholmresilience.org)

The nine processes are climate change, biosphere integrity, land-system change, freshwater change, biogeochemical flows, ocean acidification, stratospheric ozone depletion, atmospheric aerosol loading, and the introduction of novel entities such as synthetic chemicals. Their boundaries use different indicators and operate across different spatial scales. The framework connects research on biodiversity, water, atmospheric chemistry, and ecological functioning with questions of sustainable development. Its scientific development includes continuing revisions to indicators, boundary values, and methods for assessing interactions. (doi.org)

The 2023 comprehensive assessment concluded that six boundaries had been transgressed; the September 2026 Planetary Health Check reported seven, following the assessment of ocean acidification as transgressed in 2025. These assessments distinguish global boundary status from local environmental conditions and do not themselves determine how responsibilities or resource entitlements should be distributed. Applications therefore involve both scientific measurement and explicit choices about scale, allocation, and equity. (pubmed.ncbi.nlm.nih.gov)

Earth-system foundations and the safe operating space

The framework takes the Holocene, the interglacial epoch beginning approximately 11,700 years ago, as its reference state. Its relative environmental stability provided the setting in which agriculture and complex societies developed. This reference does not imply an unchanging planet: rainfall, vegetation and regional climates varied throughout the epoch. Rather, the objective is to preserve the Earth-system conditions within which contemporary societies developed, not to freeze every landscape or environmental measurement at a historical value. Johan Rockström and colleagues’ 2009 formulation therefore used the range of Holocene variability as a reference for identifying a desirable planetary state. (stockholmresilience.org)

Earth-system resilience means the capacity to absorb disturbances while retaining the functioning characteristic of a Holocene-like state. In Will Steffen and colleagues’ 2015 formulation, resilience concerns persistence under both sustained human pressures and shocks. It depends on physical and biological regulation, rather than climatic stability alone. The biosphere, for example, influences material and energy flows, while feedbacks can either counteract disturbances or amplify them. A process can weaken planetary resilience through gradual degradation even when no single abrupt global threshold has been demonstrated for that process. (doi.org)

A control variable is a measurable indicator selected to represent the state of, or human pressure on, an Earth-system process. A boundary value specifies the proposed limit for that indicator; the safe operating space is the domain on the lower-risk side of the boundaries. Variables need not all measure the same kind of quantity. For example, the 2023 assessment used atmospheric carbon dioxide concentration and anthropogenic effective radiative forcing as complementary climate indicators, expressed respectively in parts per million and watts per square metre. These describe atmospheric composition and an energy perturbation, not interchangeable measures of environmental damage. (stockholmresilience.org)

A precautionary boundary is not identical to a biophysical threshold. A threshold is a transition in system functioning, potentially abrupt or difficult to reverse; the boundary is placed before the estimated danger level. In the 2015 framework, it lies at the lower-risk end of a zone of uncertainty, reflecting incomplete knowledge and intrinsic Earth-system variability. Crossing the boundary therefore indicates increasing risk, not proof that an irreversible transition has already occurred. The separation also provides time to respond before a threshold is reached. Processes without demonstrated global thresholds can still warrant boundaries because accumulated regional changes erode planetary resilience. (doi.org)

These limits address Earth-system functioning, not the date when a mineral or fuel reserve will be exhausted. They are also distinct from local pollution standards: a planetary boundary concerns large-scale stability, whereas local limits address particular exposures or ecological conditions. Nor does the framework calculate a fixed global carrying capacity or prescribe one population size. Its original formulation instead identifies biophysical constraints within which different pathways of human development remain possible. (stockholmresilience.org)

Origins and successive revisions

In 2009, Johan Rockström and colleagues introduced the framework in two complementary publications: “A safe operating space for humanity” in Nature and the fuller “Planetary Boundaries: Exploring the Safe Operating Space for Humanity” in Ecology and Society. They identified nine Earth-system processes: climate change, biodiversity loss, interference with nitrogen and phosphorus cycles, stratospheric ozone depletion, ocean acidification, global freshwater use, land-system change, atmospheric aerosol loading, and chemical pollution. Nitrogen and phosphorus were treated together within one boundary. The authors proposed numerical boundaries for seven processes, leaving aerosol loading and chemical pollution unquantified. Initial values included an atmospheric carbon dioxide concentration of 350 parts per million and global consumptive freshwater use of 4,000 cubic kilometres per year. These were explicitly preliminary proposals rather than a completed set of established planetary thresholds. (nature.com)

The first major revision, led by Will Steffen and published in Science in 2015, retained nine boundaries but revised their indicators and strengthened their treatment of geographical variation. Several acquired a two-tier structure connecting global limits with regional conditions; for example, land-system change shifted from the proportion of land converted to cropland toward remaining forest cover, differentiated among tropical, temperate, and boreal forests. The revision also identified climate change and biosphere integrity as core boundaries: substantial and persistent transgression of either could, in the authors’ assessment, independently drive the Earth system into a different state. Other boundaries could undermine these core processes without necessarily producing such a transition on their own. This hierarchy distinguished their Earth-system roles rather than ranking their immediate importance to human welfare. (doi.org)

Replacing “biodiversity loss” with “biosphere integrity” broadened the biological boundary beyond species disappearance to include the functioning of the biosphere. A specialist study by Georgina Mace and colleagues, published in Global Environmental Change in 2014, had examined weaknesses in extinction rate and species richness as planetary indicators. It proposed considering the genetic library of life, functional diversity, and biome condition and extent. The 2015 framework consequently distinguished genetic diversity from functional integrity, retaining extinction rate as a proxy for the former and proposing the Biodiversity Intactness Index as an interim indicator for the latter. In the 2023 comprehensive revision led by Katherine Richardson, human appropriation of net primary production replaced that interim functional indicator, connecting the assessment more directly to the energy available to sustain ecological processes. (openresearch-repository.anu.edu.au)

The 2015 renaming of “chemical pollution” as “novel entities” likewise expanded the boundary’s scope. It encompassed new substances, new forms of existing substances, and modified life forms, including engineered materials and naturally occurring elements mobilized by human activities. A specialist assessment led by Linn Persson, published in Environmental Science & Technology in 2022, examined synthetic chemicals, including plastics, within this broader category. Its conclusion that the boundary had been exceeded rested on evidence that production and releases were increasing faster than global capacities for safety assessment and environmental monitoring—not on the identification of a single worldwide toxic-concentration threshold. This assessment supplied an operational basis for a boundary that had remained unquantified in the earlier framework. (doi.org)

Freshwater underwent a further conceptual expansion from “freshwater use” to “freshwater change.” Lan Wang-Erlandsson and colleagues’ 2022 study, “A planetary boundary for green water,” argued for including soil moisture and its role in terrestrial ecological and climatic regulation, rather than focusing predominantly on consumption of river, lake, and groundwater resources. The 2023 comprehensive revision incorporated both blue-water and green-water components, represented respectively by streamflow and root-zone soil moisture. Their control variables measured the proportion of global ice-free land experiencing departures from preindustrial variability, capturing unusually wet as well as unusually dry conditions. The revised boundary therefore addressed disruption of the terrestrial water cycle, not simply the volume of water withdrawn or consumed. (nature.com)

Climate change and ocean acidification

The climate-change boundary uses atmospheric carbon dioxide concentration and radiative forcing as complementary control variables. In 2009, Johan Rockström and colleagues proposed a concentration boundary of 350 parts per million (ppm), compared with a preindustrial concentration of about 280 ppm, and a maximum positive forcing of $+1\ \mathrm{W\,m^{-2}}$ relative to preindustrial conditions. The concentration indicator tracks a major, long-lived driver of climate change; the forcing indicator captures the combined influence of carbon dioxide and other agents affecting Earth’s energy budget. These were precautionary boundary values, not concentrations or forcing levels at which an abrupt global transition was demonstrated to occur. (stockholmresilience.org)

Radiative forcing measures a change in net downward radiative energy flux caused by an external driver, before the radiative response associated with surface-temperature change. Positive forcing tends to warm the climate; negative forcing tends to cool it. Unlike carbon dioxide concentration alone, net anthropogenic forcing includes warming contributions from other greenhouse gases and cooling contributions from aerosols. Neither control variable is interchangeable with a temperature target such as $1.5^\circ\mathrm{C}$ or $2^\circ\mathrm{C}$: concentration and forcing describe drivers, whereas temperature describes the climate response. The relationship depends on feedbacks, ocean heat uptake, and the history of forcing, so a concentration boundary does not specify a unique contemporaneous temperature. (ipcc.ch)

The ocean absorbs approximately one-quarter of human carbon dioxide emissions, slowing atmospheric accumulation while changing seawater chemistry. Dissolved carbon dioxide participates in chemical equilibria that produce bicarbonate and hydrogen ions:

Additional hydrogen ions react with carbonate ions:

The result is lower pH and reduced carbonate-ion availability. “Acidification” denotes this shift toward greater acidity, not necessarily a transition to pH below 7; surface seawater remains alkaline. (stockholmresilience.org)

The ocean-acidification control variable is the global mean surface-water aragonite saturation state, $\Omega_{\mathrm{arag}}$. Aragonite is a calcium carbonate mineral used by reef-building corals, pteropods, and other marine organisms. Its saturation state is expressed as

where $K_{\mathrm{sp}}^{*}$ is the apparent solubility product under the relevant seawater conditions. Values below 1 indicate undersaturation, favoring dissolution of exposed aragonite. Biological effects are not confined to that threshold: declining saturation can increase the energetic cost of forming and maintaining shells and skeletons even while seawater remains supersaturated. (data.noaa.gov)

The proposed ocean-acidification boundary retains mean surface aragonite saturation at at least 80% of its preindustrial value. This is a relative reference level, not an absolute saturation state of 0.8, and it places a precautionary limit above widespread undersaturation. A global mean also does not imply uniform exposure: cold waters and upwelling regions can experience less favorable carbonate conditions than the average. Although acidification and warming share carbon dioxide as a driver, they represent different mechanisms—carbonate chemistry and radiative energy imbalance, respectively. An intervention that reduced temperature without reducing atmospheric carbon dioxide would therefore not, by itself, remove the chemical pressure underlying acidification. (stockholmresilience.org)

Biosphere integrity and land-system change

Biosphere integrity comprises two dimensions: the diversity of life and its contribution to Earth-system functioning. The 2015 revision led by Will Steffen distinguished genetic diversity, which supplies the evolutionary capacity for adaptation, from functional integrity, through which organisms regulate energy flows and material cycles. A species count cannot describe organisms’ abundances, ecological roles, or interactions: species may remain present while declining populations cease to perform important functions. The revision therefore paired an extinction-rate indicator with the Biodiversity Intactness Index, an interim measure of native-species abundance relative to minimally disturbed conditions. (doi.org)

The genetic-diversity control variable measures extinctions per million species-years, abbreviated E/MSY. One E/MSY corresponds to one extinction annually among a million species. The 2023 assessment led by Katherine Richardson retained a boundary below 10 E/MSY and estimated the contemporary rate at more than 100 E/MSY. This is a proxy for erosion of evolutionary diversity, not a direct measurement of genetic variation within populations. (doi.org)

For functional integrity, the 2023 assessment introduced human appropriation of net primary production, or HANPP. Net primary production is the carbon fixed through photosynthesis minus plants’ respiratory expenditure; it supplies energy and organic material to ecosystems. HANPP combines biomass harvested or destroyed by people with productivity changes caused by land use. Expressing appropriation against estimated preindustrial Holocene production avoids treating human-induced productivity increases as an unrestricted ecological surplus. The assessment proposed a boundary below 10% of this reference production, with increasing risk between 10% and 20%, and estimated appropriation at approximately 30%. HANPP is a terrestrial functional proxy, not a comprehensive measure of marine and terrestrial ecological condition. (doi.org)

The land-system boundary instead tracks remaining forest area because forest conversion changes biophysical climate regulation. In 2015, the proposed minimum remaining cover was 85% of potential forest area for tropical forests, 50% for temperate forests, and 85% for boreal forests; the globally aggregated boundary was approximately 75%. These percentages refer to each biome’s reference forest extent, not to the proportion of all land that should carry trees. Tropical forests strongly influence evapotranspiration and moisture recycling, whereas boreal forests strongly affect surface reflectivity and energy exchange, particularly through their interaction with snow cover. Temperate forests were assigned a less stringent area boundary because their biophysical effects on global climate were assessed as weaker. (doi.org)

Forest extent and ecological condition nevertheless measure different changes. Habitat fragmentation divides remaining habitat into smaller, isolated patches and increases exposure to edge conditions; substantial degradation can therefore occur within land still classified as forest. A 2015 synthesis by Nick Haddad and colleagues, covering fragmentation experiments on five continents over 35 years, found biodiversity reductions of 13–75%, alongside decreased biomass and altered nutrient cycling. Effects were greatest in small, isolated fragments and increased over time. Thus, retaining a given forest area does not establish that its populations, connectivity, or ecosystem functions remain intact. (doi.org)

Freshwater change: blue water and green water

Blue water comprises water in rivers, lakes, and groundwater, including flows and stores available to aquatic ecosystems or for human abstraction. Green water comprises water associated with the unsaturated soil zone, vegetation, and evapotranspiration: precipitation retained in soil can be taken up by roots and returned to the atmosphere through transpiration, or evaporate directly. These are interconnected components of the water cycle, not separate supplies. Root-zone soil moisture links rainfall to plant growth, while evapotranspiration transfers both moisture and energy between land and atmosphere. In 2022, Lan Wang-Erlandsson and colleagues proposed explicitly incorporating green water into the freshwater boundary, extending its scope beyond accessible surface water and groundwater. (nature.com)

Earlier assessments emphasized global consumptive blue-water use, with a proposed limit of $4{,}000\ \mathrm{km}^{3}\,\mathrm{yr}^{-1}$. Consumption differs from withdrawal because some withdrawn water returns to rivers or aquifers rather than being evaporated or otherwise removed from immediately available supplies. This aggregate approach could miss changes caused by altered rainfall, vegetation, and soil-water dynamics, as well as the geographical and seasonal concentration of pressures. The 2023 assessment therefore replaced the principal global consumption indicator with two measures of freshwater change: departures in streamflow, representing blue water, and root-zone soil moisture, representing green water. Consequently, an assessment could find substantial hydrological disruption even when estimated global consumption remained below the earlier volumetric limit. (doi.org)

The revised method compares monthly conditions in each $0.5^\circ$ grid cell with a modelled preindustrial reference period, nominally 1661–1860. Values below the local fifth percentile or above the ninety-fifth percentile count as dry or wet deviations. Their areas are aggregated across global ice-free land, and monthly percentages are averaged annually. The proposed boundaries correspond to the upper end of preindustrial variability in this global affected-area measure: 10.2% for streamflow and 11.1% for soil moisture. Miina Porkka and colleagues’ 2024 analysis, which supplied the approach used in the 2023 assessment, estimated corresponding affected areas of 18.2% and 15.8% for 1996–2005, not for the publication year. These are precautionary reference-based limits, rather than demonstrated global tipping points; both anomalous drying and wetting contribute to exceedance. (nature.com)

Irrigation illustrates why the two components must be assessed together: diverting river water onto fields can increase soil moisture and evapotranspiration while reducing downstream streamflow. In the Amu Darya and Syr Darya drainage basins, the 2024 analysis identified increasingly frequent dry streamflow deviations alongside wet soil-moisture deviations from the late 1960s, consistent with irrigation expansion. Dams redistribute flows through storage and controlled releases, changing seasonal flood pulses even where annual water totals change little; altered flood pulses can affect floodplain habitats. Deforestation and other land-cover changes modify evaporation, soil moisture, precipitation, and runoff, so their hydrological effects are not captured by abstraction totals alone. (nature.com)

Groundwater abstraction adds delayed effects that may extend beyond the immediate pumping location. Pumping initially draws on aquifer storage, but continued withdrawal can intercept groundwater that would otherwise discharge into rivers, or induce river water to infiltrate into the aquifer. The resulting streamflow depletion depends on hydraulic connections and can persist after pumping stops. Thus, groundwater and surface-water pressures cannot be treated independently: a withdrawal recorded underground may subsequently reduce water availability for downstream ecosystems and users. Streamflow serves as a useful blue-water indicator, but does not directly measure every change in groundwater storage. (pubs.usgs.gov)

Nitrogen and phosphorus flows

Industrial production and agriculture alter nutrient cycles by increasing the supply of biologically available nitrogen and phosphorus and redistributing them among soils, waters, and the atmosphere. Industrial nitrogen fixation converts atmospheric $N_2$ into reactive compounds used in fertilizers, while cultivation of nitrogen-fixing crops increases biological fixation. Phosphorus extraction instead mobilizes phosphate from geological deposits, supplementing its natural release through rock weathering. Fertilizer application, livestock manure, and wastewater then transfer these nutrients into agricultural and aquatic systems. The boundary concerns the ecological consequences of these accelerated flows, rather than the exhaustion of phosphate reserves. (stockholmresilience.org)

In the 2015 revision led by Will Steffen, the nitrogen control variable was industrial and intentional biological fixation, with a proposed boundary of $62\ \mathrm{Tg\,N\,yr^{-1}}$ and an uncertainty zone extending to $82\ \mathrm{Tg\,N\,yr^{-1}}$. One teragram equals one million metric tonnes; these quantities measure elemental nitrogen, not the total mass of fertilizer products. The assessment estimated fixation at approximately $150\ \mathrm{Tg\,N\,yr^{-1}}$. This is an upstream pressure indicator: it measures nitrogen newly made available through deliberate human activity, not nitrogen concentrations in rivers or the quantity ultimately reaching the sea. (doi.org)

Phosphorus has two complementary control variables. The global variable measures delivery from freshwater systems to the ocean, with a boundary of $11\ \mathrm{Tg\,P\,yr^{-1}}$, intended to limit the risk of widespread marine oxygen depletion. The regional variable measures phosphorus from fertilizers applied to erodible agricultural soils, with a globally aggregated boundary of $6.2\ \mathrm{Tg\,P\,yr^{-1}}$ and an uncertainty zone extending to $11.2\ \mathrm{Tg\,P\,yr^{-1}}$. The latter addresses freshwater eutrophication before phosphorus reaches coastal waters. Soil application and ocean delivery are therefore different stages of nutrient transport, not interchangeable measures. The 2023 assessment led by Katherine Richardson retained these boundaries and reported approximately $22\ \mathrm{Tg\,P\,yr^{-1}}$ reaching the ocean. (doi.org)

Excess loading promotes eutrophication: nutrient enrichment stimulates algae and aquatic plants, increasing organic-matter production. Respiration and microbial decomposition subsequently consume dissolved oxygen, potentially producing hypoxia or anoxia and eliminating organisms unable to tolerate oxygen-poor conditions. The 2009 framework identified oxygen depletion in the Baltic Sea as an example of regional disruption associated with altered nutrient flows. Delivery depends on transport pathways as well as inputs: surface runoff, agricultural drainage, and groundwater carry nutrients into receiving waters, while erosion transports nutrient-bearing soil particles. Consequently, fertilizer application does not translate directly into an equivalent annual aquatic load. (epa.gov)

Global totals also conceal geographically concentrated surpluses. Where fertilizer and manure inputs exceed plant uptake, nutrients can accumulate or escape into surrounding environments; the resulting pressure depends on their location within a drainage basin and their connection to receiving waters. An EPA-associated watershed study published in 2022 found that critical source areas occupying only 2–8% of watershed land accounted for 75% of observed nutrient flux. Such concentration explains why a global aggregate cannot establish whether an individual lake, river, or coastal ecosystem is protected: comparable total inputs can produce different losses depending on soils, rainfall, drainage, and nutrient placement. (epa.gov)

Stratospheric ozone depletion and atmospheric aerosol loading

Stratospheric ozone depletion results principally from chlorine and bromine released from long-lived compounds such as chlorofluorocarbons and halons. These substances survive transport into the stratosphere, where ultraviolet radiation breaks them down; the released halogens participate in catalytic reaction cycles that destroy ozone while regenerating the reactive halogen species. Over Antarctica, reactions on polar stratospheric cloud particles activate chlorine during winter, enabling rapid ozone destruction when sunlight returns in spring. Because stratospheric ozone absorbs ultraviolet-B radiation, depletion increases the amount reaching the surface, although actual exposure also depends on clouds, aerosols, altitude, and solar elevation. (csl.noaa.gov)

The boundary uses total-column ozone: the amount of ozone integrated vertically through the atmosphere, rather than its concentration at a single altitude. Measurements are expressed in Dobson units (DU). Johan Rockström and colleagues’ 2009 formulation proposed limiting the reduction to less than 5% of a preindustrial reference of 290 DU, corresponding to approximately 275 DU. This globally framed indicator must be distinguished from the much larger seasonal depletion over Antarctica. It measures the persistence of the protective ozone column, not emissions of ozone-depleting chemicals themselves. (stockholmresilience.org)

Controls under the 1987 Montreal Protocol have reduced atmospheric abundances of ozone-depleting substances, and the 2023 planetary-boundaries assessment classified stratospheric ozone depletion as within its boundary, reporting slight recovery. That classification does not mean that the Antarctic ozone hole has disappeared. The 2022 international ozone assessment projected a return to 1980 total-column values around 2040 for the near-global average, around 2045 for the Arctic, and around 2066 for Antarctica, conditional on continued compliance and the assessment’s assumed emissions pathways. These recovery dates refer to 1980 conditions, not directly to the framework’s preindustrial reference. (pmc.ncbi.nlm.nih.gov)

Atmospheric aerosols are suspended solid particles or liquid droplets, including mineral dust, sea salt, sulfate particles, and combustion-derived soot. They alter the Earth’s energy budget through scattering and absorption: sulfate-rich particles generally reflect sunlight, whereas black carbon absorbs it and heats the surrounding air. Aerosols also provide nuclei for cloud-droplet formation, changing droplet number, cloud reflectivity, and precipitation processes. Their effects therefore depend on composition, size, height, and location, not simply particle abundance. Aerosol optical depth (AOD) is a dimensionless measure of the attenuation of light by particles through an atmospheric column; satellite instruments and ground-based sun photometers measure its spatial distribution. (science.nasa.gov)

The 2023 assessment introduced the annual mean interhemispheric difference in AOD as a global control variable, proposing a boundary of 0.1 against an estimated contemporary difference of approximately $0.076 \pm 0.006$. Unequal aerosol loading between hemispheres changes their relative heating and can shift tropical rainfall and monsoon circulation. Regional assessment remains necessary: the same study retained a provisional South Asian AOD boundary of 0.25, with an uncertainty range of 0.25–0.5, linked to risks of reduced monsoon rainfall. A hemispheric difference does not describe every regional burden, so remaining below its global boundary cannot establish that individual regions have acceptable aerosol conditions. The 2026 Planetary Health Check likewise reported global improvement while warning that unsafe regional aerosol levels persist. (science.org)

Novel entities and chemical pollution

Novel entities are substances, materials, or modified organisms whose human introduction creates geologically novel exposures with the potential to disrupt Earth-system processes. The category includes synthetic pesticides and industrial chemicals, plastics made from polymers, and engineered nanomaterials. Its broader formulations also include naturally occurring substances, such as heavy metals, mobilized and redistributed by human activities. Novelty therefore concerns not only molecular invention but also new forms, quantities, and environmental distributions. The boundary addresses potential disruption of the biosphere and other Earth-system processes, rather than treating every instance of local contamination as evidence of a planetary threshold being crossed. (developmentaid.org)

In January 2022, Linn Persson and colleagues assessed the boundary as transgressed because chemical production and releases were increasing faster than societies’ capacity to evaluate safety and monitor environmental consequences. Their assessment reported a roughly fiftyfold increase in global chemical production since 1950 and a 79% increase in plastics production between 2000 and 2015. These production trends indicated growing pressure, not equivalent increases in environmental damage: substances differ greatly in their hazards, uses, and release pathways. The authors evaluated candidate control variables against three criteria—feasibility, relevance, and comprehensiveness—and proposed complementary measures rather than a single universal concentration limit. (doi.org)

The underlying information gap extends beyond missing toxicity tests. A 2020 analysis led by Zhanyun Wang examined 22 chemical inventories covering 19 countries and regions and identified more than 350,000 chemicals and chemical mixtures registered for production and use. This was an inventory total, not a count of individually verified pollutants in the environment. More than 50,000 entries had confidential identities, while up to 70,000 were ambiguously described. Such limitations complicate comparisons between inventories, identification of substances in environmental samples, and linkage of measured exposures to production or hazard information. (pubmed.ncbi.nlm.nih.gov)

Persistence, mobility, and mixture exposure further obstruct aggregation into one control variable. Persistent substances can accumulate when continuing inputs exceed degradation or removal, while mobile substances disperse beyond their release locations. Consequently, stopping emissions need not immediately eliminate environmental exposure. Meanwhile, testing chemicals separately does not fully characterize the combined effects of mixtures or effects under different environmental conditions. The 2023 comprehensive boundary assessment adopted the percentage of synthetic chemicals released without adequate safety testing as a control variable, assigning a boundary of zero. This was a precautionary assessment criterion concerning unassessed releases, not an experimentally demonstrated concentration at which the entire Earth system changes state. (developmentaid.org)

A July 2026 methodological proposal led by Annika Jahnke called for complementary pressure, state, and impact indicators. Pressure indicators concern extraction, production, and releases; state indicators describe distribution, accumulation, and exposure; impact indicators address resulting biological or physical effects. Production data offer opportunities for early intervention but lie further from demonstrated ecological consequences, whereas impact measurements are closer to damage but may provide later warning. Proposed monitoring approaches include chemical screening combined with mixture-effect modelling, effect-based testing, and remote sensing of plastics. The authors distinguished protection from all unknowns from operational precaution that accepts explicitly bounded uncertainty, challenging the practicality of a zero-uncharacterized-release criterion. Their proposal develops an assessment architecture; it does not establish a single validated global threshold for the combined effects of all novel entities. (pubmed.ncbi.nlm.nih.gov)

Interactions, feedbacks, and compound pressures

The nine boundaries describe interacting processes, not nine independent allowances for environmental disturbance. Transgression of one can change the risk associated with another: degradation of the biosphere, for example, can weaken carbon storage and thereby amplify climate change, while warming places additional stress on ecosystems. The 2023 assessment led by Katherine Richardson explicitly identified such changes in risk gradients, including mutually reinforcing effects between climate change and biosphere integrity. Consequently, keeping an individual control variable below its proposed boundary does not isolate that process from disturbances elsewhere in the Earth system. The safe operating space depends partly on the capacity of other processes to absorb disturbance and maintain stabilizing functions. (environmentandsociety.org)

Forest loss illustrates the coupling of land, freshwater, and climate. Removing trees reduces carbon held in vegetation and changes evapotranspiration, including the return of soil water to the atmosphere through leaves. Reduced evaporation can diminish surface cooling and the supply of atmospheric moisture available for rainfall farther downwind. Changes in precipitation and root-zone soil moisture then affect vegetation growth and carbon uptake, connecting the water cycle back to the carbon cycle. Lan Wang-Erlandsson and colleagues’ 2022 green-water assessment emphasized these ecohydrological and hydroclimatic interactions. An increase in local river flow after clearing, caused by reduced plant water use, can therefore coexist with impaired moisture recycling; greater runoff is not necessarily evidence of improved Earth-system functioning. (nature.com)

Warming, drought, fire, and ecosystem degradation can act as compound pressures rather than separate disturbances. Soil-moisture deficits constrain plant growth and carbon uptake, while drying vegetation increases susceptibility to fire; burning releases stored carbon and damages the vegetation responsible for subsequent uptake. These processes can form an amplifying feedback when reduced carbon storage contributes to further warming. In its September 2026 account of the Planetary Health Check, the Potsdam Institute for Climate Impact Research highlighted repeated heat, drought, fire, and degradation as pressures capable of weakening terrestrial carbon sinks and turning some regional ecosystems into carbon sources. This is a regional and process-specific finding, not evidence that all terrestrial ecosystems have ceased absorbing carbon. (nature.com)

Direct atmospheric measurements provide evidence of these interactions in Amazonia. In a 2021 study, Luciana Gatti and colleagues analysed 590 aircraft vertical profiles of carbon dioxide and carbon monoxide collected at four sites during 2010–2018. Carbon emissions were greater in eastern than western Amazonia, and southeastern Amazonia was a net carbon source even after fire emissions were separated from the regional balance. The more heavily deforested eastern region had also experienced stronger dry-season warming and moisture stress. The study linked these conditions to increased ecosystem stress, fire occurrence, and carbon emissions; its geographically differentiated results do not establish a single basin-wide transition date or threshold. (nature.com)

Future responses remain conditional on model assumptions. The 2023 boundary assessment used idealized experiments combining atmospheric carbon dioxide concentrations of 350, 450, and 550 ppm with different remaining forest areas. Land-system transgression produced losses of terrestrial carbon stocks of approximately 100–200 billion tonnes of carbon. These experiments explored consequences of sustained forcing rather than forecasting a particular emissions pathway; atmospheric carbon dioxide was prescribed, so carbon released by vegetation did not feed back into its atmospheric concentration. The 2026 Planetary Health Check separately raised the question of whether models overestimate land-sink resilience. Observed regional weakening thus supports concern about coupled pressures without determining the magnitude or timing of future global sink changes. (environmentandsociety.org)

Assessment methods and reported boundary status

Assessments combine instrumental observations, historical and palaeoclimate reconstructions, process-based models, and synthesis of specialist literature. These sources serve different purposes: measurements estimate present conditions; reconstructions establish reference states; models estimate quantities that cannot be observed globally and test responses to disturbance; literature synthesis informs boundary placement. Reference periods differ between indicators rather than constituting one uniform “Holocene average.” For example, the 2023 freshwater assessment used simulated streamflow and root-zone soil moisture during 1661–1860 to establish preindustrial variability. Monthly departures were identified locally and aggregated into the percentage of global land experiencing unusually wet or dry conditions; boundaries were set at the 95th percentile of the preindustrial global-area distribution. (environmentandsociety.org)

Katherine Richardson and colleagues’ peer-reviewed study, published in Science Advances on 13 September 2023, reported six transgressions: climate change, biosphere integrity, land-system change, freshwater change, biogeochemical flows, and novel entities. The Planetary Health Checks published by the Potsdam Institute for Climate Impact Research on 24 September 2025 and 21 September 2026 reported seven, adding ocean acidification. These are institutional scientific reports, distinct in publication type from the 2023 journal article. Global aerosol loading and stratospheric ozone depletion remained within their boundaries in these assessments; this classification does not exclude regional exceedances or environmental harm. (pubmed.ncbi.nlm.nih.gov)

The table preserves the 2023 assessment snapshot, rather than mixing successive indicator definitions. Values are those assembled for that publication, not synchronized observations made during 2023. The uncertainty column identifies important qualifications; numerical risk-zone endpoints are not statistical confidence intervals. (doi.org)

Table values and qualifications follow the 2023 study’s Table 1 and accompanying analysis. Nitrogen and phosphorus constitute one boundary, as do the two components of biosphere integrity and freshwater change; indicator counts therefore exceed boundary counts. (doi.org)

Publication year must also be distinguished from the year a transgression occurred. The 2025 ocean-acidification assessment reported surface aragonite saturation of 2.84 against a revised boundary of 2.86, retaining the criterion of 80% of preindustrial saturation but updating its numerical baseline. The change from the 2023 classification therefore reflects revised reconstruction and reference conditions as well as environmental change—not simply a directly observed crossing during calendar year 2025. Ocean-carbon observations are spatially incomplete, so gridded reconstructions help estimate global carbonate chemistry between sampled locations. (pik-potsdam.de)

Radial diagrams transform heterogeneous indicators onto normalized scales. In the 2023 figure, the origin represents mean Holocene conditions and the boundary generally occupies a common radius, with exceptions for freshwater components; wedge lengths use logarithmic scaling. Green denotes the safe operating space, yellow through red increasing risk, and purple high risk. Blurred wedge ends indicate poorly constrained values or risk-zone limits. Normalization enables a compact comparison of boundary status, but wedge area is not a common unit of ecological damage: a larger sector cannot be read as proportionately greater species loss, economic cost, or probability of an Earth-system transition. (environmentandsociety.org)

Applications, downscaling, and related frameworks

National applications translate global boundaries into benchmarks against which environmental pressures can be compared. Two approaches are commonly distinguished: top-down allocation, which distributes a global limit among countries or other actors, and bottom-up assessment, which identifies regional ecological limits and relates them to planetary processes. Daniel O’Neill and colleagues’ 2018 study combined downscaled boundaries with environmental footprints and social indicators for roughly 150 countries. It allocated shares according to population and compared seven biophysical indicators with eleven social outcomes. These seven indicators were not seven directly measured planetary boundaries: they included ecological and material footprints alongside indicators derived from selected boundary processes. Such applications therefore require explicit choices about allocation and indicator substitution, rather than simply reproducing the global assessment at smaller scales. (eprints.whiterose.ac.uk)

The accounting perspective determines where responsibility for pressure is recorded. Territorial accounts measure pressures occurring within a jurisdiction; consumption-based accounts attribute pressures throughout global value chains to the final demand that drives production. Imported food, for example, can embody agricultural water use and land occupation outside the consuming country or city. A reduction in domestic production therefore need not imply a corresponding reduction in its consumption footprint. O’Neill and colleagues used consumption-based indicators accounting for international trade. However, annual national totals can conceal ecological differences: freshwater appropriation must also be examined at the river-basin scale and over shorter periods, because water availability and competing demands vary geographically and seasonally. (eprints.whiterose.ac.uk)

For cities and businesses, translation connects footprint inventories with organizational targets, operational responsibilities and monitoring. Xuemei Bai and colleagues’ 2024 review, Translating Earth system boundaries for cities and businesses, identified sharing approaches, formulated ten translation principles and proposed a protocol with explicit control steps. Its emphasis was on making scientific and normative judgements transparent, including assumptions and uncertainties. Targets for geographically heterogeneous processes also require local ecological information. The Earth Commission’s 2023 assessment, for example, proposed maintaining at least 20–25% diverse seminatural habitat within each square kilometre of human-modified land to support local ecosystem services. Habitat elsewhere cannot necessarily replace nearby pollination, soil protection or water regulation. (research-portal.uu.nl)

Doughnut economics, introduced by Kate Raworth in 2012, combines an ecological ceiling derived from the nine planetary boundaries with a social foundation covering essential human needs. Its developed formulation includes twelve social priorities, such as food, health, education, housing and political voice. The space between deprivation below the foundation and ecological overshoot beyond the ceiling is represented as a doughnut-shaped “safe and just” space; it is not an alternative measurement of the nine Earth-system processes. Amsterdam’s 2020 City Doughnut illustrates urban application through four lenses: local social conditions, local ecological conditions, worldwide ecological impacts and worldwide social impacts. This structure connects residents’ well-being with effects extending beyond the municipal boundary. (doughnuteconomics.org)

The Earth Commission introduced a distinct set of safe and just Earth-system boundaries in 2023, in a study led by Johan Rockström and Joyeeta Gupta. “Safe” boundaries address Earth-system stability and resilience; “just” boundaries additionally seek to minimize exposure to significant human harm, including loss of livelihoods, displacement and food or water insecurity. The integrated boundary takes the stricter of the two. For climate, the study proposed a safe boundary of 1.5 °C above preindustrial temperature but a stricter just boundary of 1.0 °C. Its assessment covered five domains—climate, biosphere, freshwater, nutrients and aerosols—and quantified eight global boundaries, seven of which it assessed as exceeded. That count is not interchangeable with the nine planetary boundaries: the domains and indicators differ, and the assessment did not establish corresponding boundaries for ocean acidification, stratospheric ozone depletion or novel entities. (nature.com)

Scientific limitations and questions of allocation

The evidential basis of the nine boundaries is uneven: evidence that a pressure disrupts Earth-system functioning is not necessarily evidence for a single global threshold. Johan Rockström and colleagues explicitly distinguished these cases in 2009. Climate change and stratospheric ozone depletion involve large-scale systemic thresholds, whereas land conversion, freshwater alteration, and biodiversity loss can accumulate through geographically distributed changes without a demonstrated planet-wide tipping point. For spatially heterogeneous processes, regional thresholds and their connections therefore underpin the proposed global limit. A global aggregate cannot by itself establish whether a particular drainage basin or ecosystem remains within locally tolerable conditions. The original authors also acknowledged that placing a boundary a precautionary distance from danger involves judgments about acceptable risk, not simply measurement of a natural constant. (stockholmresilience.org)

Changes in definitions complicate comparisons between successive assessments. Katherine Richardson and colleagues’ 2023 update replaced the earlier global freshwater-consumption indicator with measures of departures in streamflow and root-zone soil moisture, and introduced human appropriation of net primary production as the functional biosphere-integrity indicator. These changes assess different aspects of environmental disturbance; they are not merely updated observations of unchanged variables. A consistent historical comparison requires recalculating earlier conditions using the same indicator, reference period, and boundary definition. Likewise, the six transgressions reported in 2023 are classifications, not six equivalent quantities of damage: a count omits the magnitude of each departure, its uncertainty, and interactions among processes. The assessment’s normalized diagram does not provide a common physical unit in which losses across boundaries can be added. (doi.org)

Biophysical estimates do not determine how environmental pressures or reduction obligations should be distributed. In their 2023 perspective on Earth-system justice, Joyeeta Gupta, Diana Liverman, and colleagues distinguished maintaining planetary stability from preventing harm and securing access to food, water, energy, and infrastructure. A limit protecting aggregate Earth-system functioning is not inherently an equitable distribution of benefits or risks. Historical responsibility, development needs, intergenerational obligations, and participation in decision-making introduce normative questions that cannot be settled by refining measurement uncertainty alone. Their approach also recognizes that protecting people against significant harm may require a more stringent limit than one concerned only with Earth-system stability. (nature.com)

Allocation methods consequently produce different national or organizational allowances from the same global estimate. Equal-per-capita allocation divides available space by population; grandfathering preserves existing shares of pressure; ability-to-pay approaches assign greater adjustment obligations to wealthier populations. None follows uniquely from Earth-system science. An illustrative dairy-sector study compared three allocation principles and two methods of scaling individual shares to sectors, producing six allocation combinations. It showed how an assessment could change with the ethical and accounting choices even when the measured environmental pressure remained unchanged. Historical-responsibility approaches face an additional evidential limitation: reconstructing cumulative carbon dioxide emissions is more feasible than reconstructing comparable historical contributions to biosphere degradation. Thus allocation uncertainty includes both disagreements about fairness and incomplete historical inventories. (link.springer.com)

Unresolved methodological questions concern the duration of overshoot, cumulative exposure, interaction strength, and reversibility. The 2023 assessment identified only partial process-based understanding of interactions among boundary domains. (doi.org) A 2026 perspective, Planetary Boundary for Novel Entities: Time for a Reboot, illustrates the operational problem: production or releases alone do not capture accumulated stocks, delayed effects, or exposure to mixtures. It proposed complementary pressure, state, and impact indicators, including production volumes, environmental concentrations, and mixture-effect measurements. Persistent substances can remain in sediments or other reservoirs after releases decline, so reduced pressure need not imply rapid recovery. Translating such information into accountable targets still requires specifying which actors control releases, which locations and life-cycle stages are covered, what monitoring demonstrates progress, and how remaining uncertainty is treated. These proposed indicators support assessment and governance; they do not establish a demonstrated universal chemical threshold. (pubs.acs.org)

References

1. Planetary Boundaries: Exploring the Safe Operating Space for Humanitystockholmresilience.org
2. Steffen_PlanetaryBoundaries_2015healthy.ucla.edu
3. A safe operating space for humanitynature.com
4. Planetary boundaries: Guiding human development on a changing planetboku.ac.at
5. Approaches to defining a planetary boundary for biodiversityopenresearch-repository.anu.edu.au
6. Outside the Safe Operating Space of the Planetary Boundary for Novel Entitiesexplore.openaire.eu
7. A planetary boundary for green waternature.com
8. Chapter 7: The Earth’s Energy Budget, Climate Feedbacks, and Climate Sensitivityipcc.ch
9. Ocean acidification - Stockholm Resilience Centrestockholmresilience.org
10. Ocean acidificationnoaa.gov
11. Dataset Overview | National Centers for Environmental Information (NCEI)data.noaa.gov
12. California Current Acidification Network (C-CAN)pmel.noaa.gov
13. Notable shifts beyond pre-industrial streamflow and soil moisture conditions transgress the planetary boundary for freshwater changenature.com
14. Streamflow Depletion by Wells—Understanding and Managing the Effects of Groundwater Pumping on Streamflowpubs.usgs.gov
15. Streamflow depletion by wells--Understanding and managing the effects of groundwater pumping on streamflowusgs.gov
16. Sources and Solutionsepa.gov
17. Nutrientsepa.gov
18. Limited progress in nutrient pollution in the U.S. caused by spatially persistent nutrient sourcesassessments.epa.gov
19. Scientific Assessment of Ozone Depletion 2022: Twenty Questions and Answers About the Ozone Layercsl.noaa.gov
20. Earth beyond six of nine planetary boundariespmc.ncbi.nlm.nih.gov
21. Scientific Assessment of Ozone Depletion 2022: Executive Summarycsl.noaa.gov
22. Aerosols: Tiny Particles, Big Impactscience.nasa.gov
23. Planetary Health Check 2026 finds mounting pressures across Earth’s life support systemspik-potsdam.de
24. Outside the Safe Operating Space of the Planetary Boundary for Novel Entitiesdevelopmentaid.org
25. Outside the Safe Operating Space of the Planetary Boundary for Novel Entitiespubmed.ncbi.nlm.nih.gov
26. Toward a Global Understanding of Chemical Pollution: A First Comprehensive Analysis of National and Regional Chemical Inventoriespubmed.ncbi.nlm.nih.gov
27. Earth beyond six of nine planetary boundariesjussemper.org
28. Planetary Boundary for Novel Entities: Time for a Rebootpubmed.ncbi.nlm.nih.gov
29. Amazonia as a carbon source linked to deforestation and climate changenature.com
30. Amazonia as a carbon source linked to deforestation and climate changepubmed.ncbi.nlm.nih.gov
31. Planetary Health Check 2026 finds mounting pressures across Earth’s life support systemspik-potsdam.de
32. Earth beyond six of nine planetary boundaries — PubMedpubmed.ncbi.nlm.nih.gov
33. Planetary Health Check 2025 — Executive Summarypik-potsdam.de
34. Planetary Health Check 2026: A Scientific Assessment of the State of the Planetpublications.pik-potsdam.de
35. SCOPE Data Highlights Ocean Acidification in Planetary Health Check 2025oceancarbon-scope.org
36. A good life for all within planetary boundarieseprints.whiterose.ac.uk
37. Translating Earth system boundaries for cities and businessesresearch-portal.uu.nl
38. Safe and just Earth system boundariesnature.com
39. What is the Doughnut?doughnuteconomics.org
40. The Amsterdam City Doughnut: A Tool for Transformative Actiondoughnuteconomics.org
41. Earth system justice needed to identify and live within Earth system boundariesnature.com
42. Sharing the safe operating space: Exploring ethical allocation principles to operationalize the planetary boundaries and assess absolute sustainability at individual and industrial sector levelslink.springer.com