Role of natural halogen chemistry on the evolution of global stratospheric ozone depletion

Abstract

The stratospheric ozone layer acts as Earth’s primary shield against harmful ultraviolet radiation. Anthropogenic emissions of chlorinated and brominated compounds caused substantial ozone depletion during the 20^th century, including the formation of the Antarctic ozone hole. While the role played by these anthropogenic compounds has been well established, the contribution of naturally emitted halogens, particularly bromine and iodine, remains poorly quantified. Here, using chemistry-climate model simulations spanning 1910 to 2100, we show that natural halogen chemistry makes a major contribution to lower-stratospheric ozone loss. Between 1990 and 2020, when anthropogenic halogen loading reached its peak, natural halogens accounted for up to half of the total halogen-induced ozone loss in the extrapolar lower-stratosphere and substantially enhanced ozone destruction along the Antarctic periphery by ~35%. Their relative influence is projected to increase by the end of this century, exceeding 80% of total halogen-induced ozone loss in the extrapolar lower-stratosphere and delaying the recovery of the natural Antarctic seasonal ozone cycle by nearly five decades. These findings reveal that natural halogens play a critical but underrepresented role in shaping the past and future changes in the ozone layer and should be accurately represented in ozone projections.

Halogens from natural sources account for a substantial fraction of stratospheric ozone depletion, contributing up to half of recent lower-stratospheric ozone loss and delaying Antarctic ozone recovery by several decades.

Introduction

The stratospheric ozone layer is essential for protecting life on Earth from harmful ultraviolet (UV) radiation¹. The discovery of severe ozone depletion in the 1980s, particularly the onset of the Antarctic ozone hole², led to extensive research into the chemical and meteorological processes driving stratospheric ozone loss. Anthropogenically produced long-lived halogens (LLH) such as chlorofluorocarbons (CFCs), halons, and hydrochlorofluorocarbons (HCFCs) were identified as the primary drivers of human-induced ozone depletion³. These compounds are chemically stable in the troposphere, allowing them to reach the stratosphere where, due to the increased UV radiation, they photochemically release reactive halogen (chlorine and bromine) radicals that catalyze ozone destruction^4,5. This human-driven destruction is hereafter referred to as ozone depletion. The Montreal Protocol and its subsequent amendments have successfully limited anthropogenic emissions of these ozone-depleting substances (ODSs), with confirmed evidence of effectiveness reflected in the early signs of ozone recovery, not only over Antarctica but across the world’s ozone layer^6–8.

In addition to the dominant anthropogenic LLH contribution, the Earth’s stratosphere has always contained halogens arising from natural sources, including chlorine (Cl), bromine (Br), and iodine (I). This natural halogen background originates primarily from oceanic and terrestrial ecosystems, biomass burning, and salt-marsh environments, which emit LLH such as methyl chloride (CH₃Cl) and methyl bromide (CH₃Br)^9–11. These natural sources are currently estimated to account for ~16% of the total chlorine and 29% of the total bromine reaching the stratosphere¹. Another source of stratospheric natural halogens are short-lived halogens (SLH)¹², including organic very short-lived (VSL) halocarbons and inorganic halogen species (collectively called Cl_y, Br_y and I_y, respectively) with lifetimes under six months, emitted predominantly from marine biogenic and abiotic processes^13,14. These natural SLH precursors are rapidly converted into reactive halogens close to the tropopause¹², with brominated SLH contributing an estimated 26% of the total stratospheric bromine injection, nearly equal to the natural CH₃Br fraction¹. More recently, natural SLH iodine has also been shown to contribute to the net stratospheric halogen budget^15–17. Despite their low concentrations compared to peak anthropogenic long-lived halogen sources, the higher reactivity of natural Br and I suggests that these compounds play a substantial role in ozone chemistry^18,19. This influence is particularly pronounced in the tropical lower stratosphere, where efficient convective uplift injects precursors upward^20,21. It is also evident over Antarctica, where crossed-halogen reactions between anthropogenic Cl and natural Br and I increase the depth and size of the ozone hole^17,22,23.

While the role of anthropogenic LLH on stratospheric ozone depletion is well established, and the relative contribution of LLH compared to that of other greenhouse gases (GHGs) such as CO₂, CH₄ and N₂O to the long-term evolution of the ozone layer has been widely studied^24–27, the time-varying background contribution of natural halogens to the total halogen-driven ozone loss remains unquantified. This gap is particularly relevant for a comprehensive understanding and reliable projection of the future recovery of the ozone layer. Using a state-of-the-art Earth System Model with interactive chemistry, we present a two-century assessment of the relative contribution of natural and anthropogenic halogens to the global stratospheric ozone loss across pre-industrial, anthropogenically perturbed and recovery phases. We show that the relative importance of natural versus anthropogenic halogens has shifted markedly over time: natural halogens dominated stratospheric ozone loss until around 1960, and still accounted for about 25% of the whole stratosphere ozone destruction by 1980 (nearly 50% in the lower stratosphere), in part by enhancing the efficiency of anthropogenic LLH-driven loss through cross-halogen reactions. Notably, natural bromine and iodine exert a strong influence on the polar regions, particularly on the seasonal cycle and spatiotemporal distribution of ozone over the Antarctic. Critically, natural halogens amplify anthropogenic impacts on the Antarctic ozone hole by accelerating its onset and delaying the projected recovery of its seasonal cycle by several decades. These findings demonstrate that natural halogen chemistry has played, and will continue to play, a key role in shaping the evolution of global stratospheric ozone.

Results

Global stratosphere

To quantify the role of natural halogens in the evolution of global stratospheric ozone depletion, we employ the Community Earth System Model version 2 with coupled short-lived halogen chemistry (CESM2-SLH)²⁸ and compare two long-term sets of simulations, one including natural plus anthropogenic halogen sources (henceforth AllHAL), and another one excluding the natural fraction (henceforth NoNatHAL), i.e., anthropogenic fraction only. The contrast between the two sets of simulations, each of them based on a three-member ensemble initialized from distinct, physically consistent coupled atmosphere-ocean states, allows us to isolate the natural contribution to the total halogen-mediated stratospheric ozone loss relative to the overall ozone budget for different time periods and to characterize internal variability and uncertainties of each sensitivity (see “Methods”). The resulting AllHAL ensemble reproduces the observed evolution of total column ozone and falls within the spread of other state-of-the-art chemistry-climate models prescribed with equivalent forcings (see “Methods” and Supplementary Fig. S1), providing confidence in the suitability of the simulations for the attribution analysis presented below.

Figure 1A shows the global-mean O₃ loss rate through halogen-mediated reactions averaged across the whole stratosphere. During the early 20^th century, the mean stratospheric O₃ loss rate is exclusively due to natural sources (Fig. 1A). However, from the 1940s onward, halogen-driven ozone depletion increases sharply due to rising anthropogenic concentrations (Supplementary Figs. S2–S4), surpassing the background natural contribution by the mid-1960s (Supplementary Fig. S5). Human-induced global whole-stratosphere ozone depletion peaks during the 20^th−21^st centuries turnover, primarily driven by anthropogenic LLH chlorine (Supplementary Figs. S2 and S4A, and Table S1). Following the implementation of the Montreal Protocol, human-induced halogen ozone depletion declines after 2000, although due to their much longer photochemical lifetimes, anthropogenic halogens remain the dominant contributors to global whole stratospheric ozone depletion throughout the 21^st century.

Fig. 1. Normalized halogen-mediated O₃ loss rate.

A Whole stratospheric annual mean depletion rates, averaged across the ensemble, normalized to the O₃ volume mixing ratio, averaged globally from the tropopause to 3 hPa. Colors represent the contribution of different halogen families to O₃ loss (see legend). Pie-charts represent the ensemble-mean percentage contribution of the normalized depletion rates of each halogen family relative to the total halogen-mediated losses averaged across 1910–1940, 1990–2020 and 2080–2099; the corresponding full range across ensemble members is reported in Supplementary Table S3. B 1990–2020 zonal-mean profile of the natural contribution relative to the total halogen-mediated O₃ depletion (%) and (C) its average vertical profile between 60 °S and 60 °N, decomposed into families of reactions (colors in legend). Horizontal lines indicate the ~30 hPa upper limit of the lower stratosphere region and the ~100 hPa upper and ~300 hPa lower bounds of the tropopause within the extrapolar latitudes (60° S and 60° N). D Same as A but averaged within the extrapolar lower stratosphere (i.e., box in panel B), from the tropopause up to 30 hPa, 60° S–60° N. Hatched areas in A, C and D differentiate the portion contributed by halogens from anthropogenic (ant) sources from that of natural sources (empty areas).

Though weakly, our simulations reveal that anthropogenic halogens began depleting stratospheric ozone as early as the 1920s (left pie-chart on Fig. 1A), and significantly after 1940, about three to five decades before the role of LLH on stratospheric ozone loss was first reported^4,29. This early impact is consistent with the initial emergence of carbon tetrachloride (CCl₄) emissions in the 1910s³⁰, preceding the substantial increase in CFCs concentrations from the 1940s onward¹ (Supplementary Fig. S4A). In the 1990–2020 period, when anthropogenic halogen-induced ozone depletion reached its peak, natural halogens continued to have a non-negligible influence on global stratospheric O₃ loss. They accounted for ~15% of the total global halogen-mediated ozone loss, primarily through Cl- (12%) and Br-driven (3%) reactions (see middle pie-chart in Fig. 1A). In this period, the contribution of natural iodine integrated over the whole stratosphere is minimal, dwarfed by the dominant anthropogenic sources of Br and Cl.

Lower stratosphere

While natural halogens contribute notably to ozone depletion across the whole stratosphere (Fig. 1A), their impact is markedly more pronounced in the lower stratosphere, i.e., from the tropopause up to ~30 hPa, with respect to the middle/upper stratosphere (from ~30 hPa to ~3 hPa; Supplementary Table S1). This shift in halogen-induced stratospheric O₃ loss between the lower and middle/upper stratosphere (Fig. 1B and C) responds to the substantial difference in photochemical lifetimes between natural SLHs and anthropogenic LLHs¹. Indeed, during the peak period of halogen-induced ozone depletion (1990–2020), our simulations indicate that, despite their much lower concentrations compared to anthropogenic LLHs (see Supplementary Fig. S6), natural halogens still exerted a dominant influence on ozone loss across much of the lower stratosphere, with the strongest effect occurring at extrapolar latitudes near the tropopause (Fig. 1B–D). This localized behavior at low stratospheric altitudes is particularly important, as the influence of ozone variability on the Earth’s net radiative balance maximizes in this region^31,32.

In contrast to extrapolar and Arctic latitudes, an opposite stratospheric signal emerges over Antarctica, where natural halogens suppress normalized ozone depletion in the core of the polar lower stratosphere while amplifying ozone loss along its periphery (Fig. 1B). These negative anomalies can be counterintuitive and should not be misinterpreted as natural halogens preventing ozone destruction (i.e., absolute ozone abundance in AllHAL remains always lower than in NoNatHAL, see Supplementary Fig. S7). Instead, the blue Antarctic region in Fig. 1B reflects a local suppression in the efficiency of normalized ozone depletion, driven by a redistribution of chemical pathways. Specifically, natural bromine and iodine enhance Cl-Br cross-reactions, which dominate in the polar periphery, while reducing the effectiveness of pure Cl reactions in the polar core. Given the widespread magnitude and spatial variability of these effects, we next examine natural halogen effects across different latitudinal bands: extrapolar (60° S−60° N) and Antarctic (90° S−60° S) regions. This analysis considers differences in the release of reactive halogens from natural and anthropogenic sources, their contrasting vertical distribution and the increased chemical reactivity of iodine- and bromine-driven ozone-loss chemical cycles (including cross-reactions) relative to the predominantly chlorine-based anthropogenic ozone depletion.

Extrapolar latitudes (60° S−60° N)

The influence of natural halogens in the extrapolar region (Fig. 1B, boxed; see “Methods”) is strongest near the tropopause and decreases with increasing altitude (Fig. 1C). Notably, natural halogens exerted a profound influence on lower-stratospheric ozone loss during the early 20^th century and continue to account for nearly half of the total halogen-mediated loss even during the period of maximum anthropogenic ozone depletion (averaging 49.3% between 1990 and 2020; middle pie-chart in Fig. 1D). Among these, iodine-driven reactions contribute to the largest natural individual share (15.7%), whereas natural bromine accounts for up to 26.1% when combining cross-reactions with Cl (13.5%) and reactions involving only natural Br (12.6%). Importantly, only for Cl does the anthropogenic contribution (37.4%) exceed the natural component (7.5%) (Fig. 1D), consistent with the fact that stratospheric inorganic Cl is proportionally more abundant from anthropogenic sources than from natural ones (Supplementary Fig. S6 A and B). At 100 hPa, natural sources account for ~three-quarters (73.4%) of halogen-mediated ozone loss, dropping to about one-third (33.7%) at 30 hPa (Supplementary Fig. S8) for the 1990–2020 period. Above this height, ozone depletion intensifies sharply, driven predominantly by the large abundance of inorganic chlorine due to the effective photodecomposition of anthropogenic LLH at these altitudes (Supplementary see Fig. S8). Indeed, anthropogenic halogens contribute over 80% of the total halogen-induced loss from ~10 hPa upward (Supplementary Fig. S8). Overall, even as their relative importance decreases with increasing altitude, natural halogens remain a dominant driver of ozone loss across the extrapolar lower stratosphere, reflecting a substantially higher ozone-depleting efficiency compared to anthropogenic halogens.

Our simulations reveal that the enhanced ozone-loss efficiency of natural halogens compared to LLHs arises mainly from the specific chemical properties of the species involved, particularly their shorter photochemical lifetimes, rather than from their source origin (i.e., natural vs. anthropogenic). Given that most of the halogen sources included in the CESM2-SLH model are in the form of non-reactive organic compounds (see “Methods”), all those halogenated source-gases remain unreactive until they photochemically decompose in the lower stratosphere to release their reactive Cl, Br and I atoms, which occurs at different heights. Due to the much larger lifetime of LLH, Cl photodecomposition occurs predominantly in the upper stratosphere (with exception of CH₃Cl that is partially photolyzed in the lower stratosphere), while most iodine halocarbons are photo decomposed before reaching the tropopause (Supplementary Table S1). For the case of Br, both CH₃Br and organic SLH bromine photodecompose in the lower stratosphere, the latter reacting faster than the former, though with lower abundance. These lifetime and concentration differences between LLH and SLH are reflected in the reactive halogen release contributed from different groups of source-gases (Fig. 2 and Supplementary Fig. S9). During 1990-2020, the release of most of the reactive Cl content occurs predominantly in the upper stratosphere, driven by anthropogenic LLH species (Fig. 2A). In contrast, the major release of reactive Br and I occur in the lower stratosphere, where natural short-lived substances dominate. Near the tropopause, Br release originates mainly from natural organic SLH, while natural CH₃Br becomes increasingly relevant with altitudes up to ~30 hPa (Fig. 2B). For I, the dominant source of organic release is CH₃I (Fig. 2C), which occurs predominantly below the tropopause, and therefore most stratospheric iodine follows product gas injection^15,17. These patterns underscore that halogen activation is primarily controlled by species-specific photodecomposition rates and vertical transport, with the relative influence of natural Cl and Br sources in the extrapolar lower stratosphere expected to increase in the future as anthropogenic LLH emissions decline (Supplementary Fig. S9) and convective transport intensifies under global warming³³.

Fig. 2. Release of reactive halogens in extrapolar latitudes during 1990–2020.

Vertical profiles of in-situ release rates of the corresponding reactive (A) Cl, (B) Br and (C) I atoms arising from the photochemical degradation of organic source gases (see legend). The in-situ halogen sources considered in this work are long-lived (LLH) and organic SLH (i.e., VSL halocarbons) (see “Methods”). Hatched areas differentiate the contribution from anthropogenic (ant) halogens from that from natural sources. Here, we note that all LLH are considered anthropogenic, with the exception of CH₃Cl (A) and CH₃Br (B) which also possess a natural contribution (see “Methods” and Supplementary Fig. S2). For the case of organic SLH, Br (B) and I (C) emissions are assumed to be only of natural origin, while for Cl (A) a distinction between natural organic SLH (SLHnat) and anthropogenic SLH (SLHant) species is considered. Horizontal lines indicate the ~30 hPa upper limit of the lower stratosphere region and the ~100 hPa upper and ~300 hPa lower bounds of the tropopause within the extrapolar range (60° S and 60° N).

Although the absolute release of inorganic Cl far exceeds that of Br and I, the latter exhibit greater ozone-destruction efficiency^17,32. This enhanced efficiency arises from higher photolytic rates and more effective catalytic cycles involving Br and I^15,19,34. To account for these effects, we computed the Cl-equivalent ozone loss potential in the extrapolar lower stratosphere (Fig. 3), normalizing halogen-specific ozone loss rates per unit of inorganic halogen abundance and expressing Br and I normalized values relative to Cl (see “Methods”). This metric integrates: (i) the higher Br and I fraction in natural organic SLH emissions (Supplementary Fig. S2); (ii) the rapid photodecomposition of short-lived halocarbons in the lower stratosphere compared to LLHs (Fig. 2); (iii) the enhanced catalytic recycling of Br and I species; and (iv) the dominant Cl abundance throughout the stratosphere (Supplementary Fig. S6). Conceptually, this approach follows the concept of Effective Equivalent Stratospheric Chlorine (EESC) used in WMO assessments³⁵ but is based on local ozone-loss chemistry rather than surface mixing ratios and tabulated lifetimes.

Fig. 3. Cl-equivalent ozone loss potential in the extrapolar lower stratosphere.

A Absolute volume mixing ratio of inorganic halogens (Cl_y, Br_y and I_y). B Relative contribution of different halogen families to the total halogen-mediated O₃ depletion rate normalized to the O₃ volume mixing ratio (see legend). C Cl-equivalent volume mixing ratios of inorganic halogens, i.e., with Br_y and I_y adjusted to reflect their relative ozone depletion efficiency equivalent to Cl_y (Br_y^Cl-eq and I_y^Cl-eq) (see “Methods”). Average from the tropopause up to 30 hPa, 60 °S–60 °N and across the three ensemble members from 1960 to 2099. Hatched areas differentiate the portion contributed by halogens from anthropogenic sources (ant) from that of natural sources. All quantities represent ensemble-mean values from 1960 onwards (single member before).

Results show that, during 1990–2020, inorganic iodine (I_y) and bromine (Br_y) are ~362 and 63 times more efficient, respectively, than inorganic chlorine (Cl_y) at depleting ozone in the extrapolar lower stratosphere (Supplementary Fig. S10). This agrees with previous estimates of the enhanced efficiency of bromine compared to chlorine¹⁹. Consequently, although natural sources contribute only ~15% of the total inorganic halogen burden in this region (Fig. 3A), their high photochemical efficiency accounts for ~40% of total halogen-mediated ozone loss (Fig. 3B). These findings disentangle how halogen chemical speciation enhances the contribution of natural sources to ozone loss, amplifying their role in the lower stratosphere beyond what would be expected from their comparatively small abundance relative to anthropogenic halogens.

By the end of the 21^st century, as stratospheric halogen loading is projected to return almost to the pre-human perturbed regime, natural sources tend to re-emerge as the dominant component of the total halogen-driven ozone-loss in the lower stratosphere, accounting for 82.8% by 2080–2099 (Fig. 1D). This transition away from the Cl-driven peak of human-induced ozone layer depletion is marked by a growing contribution from I- and Br-mediated reactions, rising from 15.7% and 12.6% in 1990–2020 to 39.9% and 27.1% by the end of the century, respectively. In contrast, Cl-Br cross-reactions decline markedly from 13.5% to 7.6%, while natural Cl-mediated reactions barely change from 7.5% to 8.1%. This uneven evolution points to a synergistic interaction in which natural bromine enhances Cl-Br cross-reactions more efficiently when anthropogenic Cl is available^22,23. As anthropogenic Cl abundance declines in the future, the weakening of this natural-anthropogenic synergy underscores how interactions between natural and anthropogenic halogens shape the overall impact of halogens on stratospheric ozone balance.

Antarctic

Natural halogens play a key role in reshaping the spatio-temporal distribution of Antarctic ozone depletion pathways during the period of minimum ozone levels (1990-2020). Figure 4A shows the change in annual mean halogen-mediated ozone loss rate normalized to local ozone concentrations between the AllHAL and NoNatHAL simulations. This comparison reveals that natural halogens reduce ozone depletion efficiency in the core (90°−80° S) of the Antarctic lower stratosphere while enhancing ozone loss toward the polar cap periphery (80°−60° S) (Figs. 1B and 4A). A breakdown of the normalized natural halogen contribution by reaction families reveals that Cl-Br cross-reactions dominate ozone loss at the polar periphery, while individual Cl-driven depletion (largely linked to anthropogenic ODSs) is suppressed at the polar core in the presence of natural halogens (Supplementary Fig. S11). This pattern further underscores the aforementioned natural-anthropogenic synergy: natural Br interacts with anthropogenic Cl to amplify ozone depletion along the polar periphery, where pure-Cl cycles are less effective, while simultaneously suppressing the efficiency of the dominant pure-Cl channel within the core of the polar vortex.

Fig. 4. Antarctic 1990–2020 annual mean halogen-mediated ozone loss rate normalized to the O₃ volume mixing ratio and radiative effect.

A Polar projection averaged in the lower-stratosphere (tropopause – 30 hPa). B Vertical profile of the zonal mean averaged between 80° and 60° S (delimited by solid lines in A), decomposed into families of reactions mediated by Cl, Br, I and Cl-Br crossed-reactions (see legend). Hatched area differentiates the portion contributed by halogens from anthropogenic (ant) sources from that of natural sources. Horizontal dashed lines indicate the upper and the lower bounds of the tropopause within the 80°–60° S range. C Polar projection and (D) zonal mean of the contribution of natural halogens to the excess of downwelling solar flux at the surface due to the ozone loss in 1990–2020 relative to 1910–1940 (∆SW_↓^O3). All quantities represent ensemble-mean values. The 95% confidence interval, accounting for both interannual variability and ensemble spread (see “Methods”), is used to identify regions where the O₃ loss rate (A) and ∆SW_↓^O3 (C) are not statistically significant (gray shading) and to quantify the uncertainty in the zonal-mean ∆SW_↓^O3 (D).

Figure 4B distinguishes the contribution from natural and anthropogenic sources for each individual halogen family to the total halogen-mediated depletion in the periphery of the Antarctic lower stratosphere (80°−60° S). Natural sources account, on an annual average, for 35% of the total halogen-mediated ozone loss in the lower stratosphere. Within this contribution, Cl-Br cross-reactions dominate (21.2%), followed by Cl-mediated (6%), I-mediated (5%), and Br-mediated pathways (2.8%). This effect peaks just above the tropopause (at ~140 hPa), where natural halogens account for about 55% of the net halogen-mediated ozone loss. This effect then declines to ~30% at 30 hPa, coinciding with the local minimum in normalized halogen-mediated loss. Above this level, the halogen-mediated depletion increases rapidly and is dominated by anthropogenic ODSs, which account for more than 80% of the total loss from ~10 hPa upward (Supplementary Fig. S12). This demonstrates that natural halogens have a pronounced impact on ozone depletion both at the geographical (Fig. 4A) and vertical (Fig. 4B) periphery of the Antarctic lower stratosphere, where their interactions with anthropogenic species, particularly through Cl-Br cross-reactions, substantially amplify lower-stratospheric ozone loss^17,22,23.

To further assess the implications of this ozone depletion on Antarctic ecosystems, we evaluate the excess of downwelling solar flux at the surface attributable to the natural-halogen induced ozone loss during 1990–2020. This excess is defined as the difference between the mean values for the maximum depletion period (1990–2020) and the pre-ODS period (1910–1940), and we quantify how much of the enhanced downwelling radiation is driven by natural halogens (see “Methods”). The analysis reveals that natural halogens contribute an excess of 0.10 ± 0.04 W m⁻² in shortwave radiation reaching the polar cap surface (90°−60° S) on an annual mean, accounting for up to 10% of the total halogen-induced increase in this period of maximum ozone depletion (Fig. 4C). The zonal-mean distribution underscores the stronger influence of natural halogens along the Antarctic periphery, exceeding their impact within the polar core (Fig. 4D). At the South Pole, the annual mean radiative excess is 0.10 ± 0.06 W m⁻², peaking to about 0.13 ± 0.05 W m⁻² around 75 °S (80°−70° S mean), and then decreasing sharply toward lower latitudes. Absolute changes in downwelling shortwave radiation are influenced not only by the release rate of reactive halogens from natural sources and their impact on ozone-loss efficiency, but also by the amount of solar radiation reaching the top of the atmosphere, both of which vary with latitude and season. This reflects the multiple dependence of the Earth’s radiative effect from natural SLH on diverse chemical and physical factors³⁶.

Natural halogens also play an important role in modulating the perturbation induced by anthropogenic halogens on the evolution of the Antarctic stratospheric ozone during the 20^th and 21^st century. Figure 5 shows changes in the natural seasonal cycle of Antarctic ozone from 1910 until 2100 for the AllHAL and NoNatHAL simulations. Most notably, the onset of the ozone hole represents a major perturbation of the natural solar-driven ozone cycle (expressed as the anomaly relative to the preceding Austral winter month of July) at high latitudes²³. Under largely unperturbed natural conditions (1910–1960), total ozone within the Antarctic polar cap (90°−60° S) typically exhibits higher values during the austral spring (September-November) than in the preceding polar winter, which is attributed to photochemical production and stratospheric transport^3,37. With the outbreak of anthropogenic ODSs emissions after 1940 and the subsequent formation of the ozone hole, this natural regime over the Antarctic polar cap was reversed, reaching a pronounced minimum during September-October as a result of catalytic chemical depletion mediated by chlorine and bromine from human sources^4,5. Our multi-model ensemble results indicate that this anthropogenic perturbation is strongly dependent on the predicted background of natural halogens represented in the model: natural sources significantly amplify this perturbation by advancing the reversal of the seasonal natural regime by 7–10 years and, more importantly, by substantially delaying the predicted recovery of the natural Antarctic ozone seasonality by 47–52 years (Fig. 5). Therefore, our findings show that natural halogens significantly re-shape both the onset and persistence of the Antarctic ozone hole’s seasonality following the increase and decline in dominant anthropogenic ODSs.

Fig. 5. Disruption and recovery of the natural seasonal cycle of mean Antarctic stratospheric O₃ column (90°−60° S).

Monthly stratospheric O₃ differences relative to July, shown as ensemble means from 1960 onwards (single member before), for the simulations (A) including all halogen sources (AllHAL) and (B) excluding natural halogen sources (NoNatHAL). Dashed horizontal lines highlight the difference values in October, with black markers indicating the zero-crossing points to assess the timing of disruption and recovery of the natural seasonal cycle. Error bars represent the range of zero-crossing points derived from the minimum and maximum O₃ values across the ensemble.

Discussion

Several studies have highlighted the critical importance of the Montreal Protocol and its amendments, demonstrating major benefits for life on Earth by estimating “World Avoided” environmental achievements compared to a business-as-usual scenario^7,38–40. These assessments assumed a continued rise in anthropogenic halogen emission and compared the projected evolution of the ozone layer under that trajectory with more realistic scenarios in which anthropogenic LLH emissions are controlled. While unreported industrial emissions in East Asia have been shown to threaten full ozone recovery^41–44, current projections largely assume a negligible role of halogens naturally emitted from the oceans and biosphere. In other words, a detailed attribution of the role of the natural halogen background on the ozone layer evolution remains missing. This gap is critical for three main reasons: (i) ozone layer recovery estimates must account for the pre-industrial (natural) halogen baseline as a reference; (ii) natural bromine and iodine, due to their short lifetimes and high reactivity, are disproportionately more efficient at destroying ozone than chlorine in the lowermost stratosphere; and (iii) this region is highly sensitive to climate-driven changes in temperature and composition near the tropopause, which induce substantial perturbations in the radiative balance.

Our results show that even during the period of maximum anthropogenic influence (1990-2020), natural halogens contributed ~15% of the global whole stratosphere halogen-mediated ozone loss, primarily through Cl- (12%) and Br-cycles (3%), and accounted for at least half of the halogen-driven depletion in the lower stratosphere (Fig. 1). This disproportionate high efficiency stems from the shorter lifetimes (Fig. 2) and much higher reactivity (Fig. 3) of natural Br and I compared to anthropogenic chlorinated ODSs, which require longer residence times and higher altitudes for photolytic release of reactive Cl atoms. This effect is particularly distinctive over Antarctica, where the natural halogen background alters the spatial pattern of halogen-driven ozone depletion (dominated by anthropogenic sources) by suppressing the efficiency of Cl-driven loss within the polar core and enhancing the Cl-Br cross-reactions along the surrounding periphery. This influence becomes most evident when examining Antarctic ozone seasonality: natural halogens contribute to substantially delay the projected recovery of the pre-industrial seasonal ozone cycle (following the decline in anthropogenic ODSs emissions) by about five decades.

Similar evaluations in the Arctic (60°–90°N) indicate that the annual mean contribution of natural halogens to lower-stratospheric ozone loss is broadly comparable to that in extrapolar regions. However, we note that most fully-coupled chemistry-climate models, including the CESM2-SLH setup used here, exhibit temperature biases in the Arctic lower stratosphere and are unable to reproduce Northern Hemisphere sudden stratospheric warmings unless constrained by prescribed meteorological fields, which exert a strong control on springtime ozone chemistry^45–48. Consequently, regional estimates for the Arctic derived from our fully-coupled model configuration should be interpreted with caution. Despite these limitations, we find that natural halogens account for roughly half of the Arctic halogen-driven ozone depletion in recent decades (1990–2020), increasing to nearly 80% by the end of the century (Supplementary Fig. S13).

An important consideration for future ozone projections is the role of climate- and composition-driven feedbacks linking oceanic emissions of natural halogens and anthropogenic forcing. Recent work by Saiz-Lopez et al.³⁶ highlights that natural SLH sources and their climate impacts are strongly modulated by feedbacks with anthropogenic emissions, mediated through changes in atmospheric composition and oxidative capacity, as well as through climate-chemistry interactions. For example, anthropogenic increases in tropospheric ozone can amplify natural inorganic iodine emissions via enhanced ozone deposition to the ocean surface⁴⁹, while anthropogenic aerosols further enhance inorganic SLH recycling through heterogeneous reactions⁵⁰. In parallel, climate-driven changes in ocean biogeochemistry (through changes in ocean stratification, productivity, and biological activity) can substantially increase marine emissions of SLH halocarbons under warming scenarios^49,51. Together, these feedbacks demonstrate that natural halogens actively influence key chemical and physical processes in the atmosphere and are expected to continue doing so under future climate change^24–27,36. In the present study, while recycling processes of inorganic halogens are represented in the model setup²⁸, oceanic emissions of SLH halocarbons are prescribed as constant in time^21,36, and their response to climate variability is therefore not explicitly captured. Consequently, our projections should be interpreted as a conservative lower-bound estimate of the contribution of natural halogens relative to anthropogenic halogens regardless of the scenario considered. This conservative context is further supported by the use of a high-GHG scenario (see “Methods”), which yields relatively high baseline stratospheric ozone by the end of the 21^st century. Under other scenarios with lower projections of stratospheric ozone recovery, the impact of natural halogens on normalized ozone loss would likely be larger.

Previous studies have shown that SLH exert a disproportionally strong radiative effect through ozone loss compared to long-lived halocarbons³², while Saiz-Lopez et al.³⁶ further quantified the evolving role of natural SLH in Earth’s net radiative balance from pre-industrial conditions to the present and into the future. In addition, recent work links enhanced reactive chlorine activation in the lower stratosphere to heterogeneous chemistry on wildfire-driven aerosols⁵², a process expected to intensify under global warming. Accurately projecting ozone layer recovery and its associated climate feedbacks therefore, requires explicit consideration of the natural halogen baseline, including the strong vertical dependence of Cl-, Br- and I-driven ozone-loss efficiency. Overall, our results underscore the important, yet often overlooked, role of natural halogens in the past, present and future evolution of the stratospheric ozone layer.

Methods

Model configuration and simulations setup

To evaluate the stratospheric ozone response to natural versus anthropogenic halogen sources, we performed a set of simulations using the Community Earth System Model version 2 with on-line short-lived halogen chemistry (CESM2-SLH)²⁸. The model was run in fully coupled configuration, employing the Whole Atmosphere Community Climate Model v6 (WACCM6) as the atmospheric component. This setup features high vertical resolution in the upper atmosphere, with 70 vertical levels extending from the surface to 6 × 10⁻⁶ hPa (∼140 km), and horizontal resolution of ~2°^28,53.

Two simulations were conducted to isolate the effects of natural halogen emissions. The first included all known natural halogen emissions along with the model’s default input anthropogenic species (AllHAL), while the second excluded natural halogen sources (NoNatHAL). The impact of natural sources was then quantified as the difference between the two simulations (AllHAL − NoNatHAL). Natural emissions comprise oceanic sources of organic SHL (i.e., VSL halocarbons: CHBr₃, CH₂Br₂, CH₂BrCl, CHBr₂Cl, CHBrCl₂, CH₃I, CH₂I₂, CH₂IBr and CH₂ICl)⁵⁴, and abiotic emissions of inorganic iodine species (I₂ and HOI)⁵⁵ (Supplementary Fig. S3). Along with these natural sources, the default CESM2-SLH chemical scheme²⁸ also accounts for additional bromine and chlorine release through sea-salt aerosol dehalogenation^36,56.

Additionally, the default lower boundary condition (LBC) dataset includes surface mixing ratios for CH₃Cl and CH₃Br (Supplementary Fig. S4), the only long-lived halocarbons with both natural and anthropogenic sources. For these species, we partitioned their time series into natural and anthropogenic components directly from the LBC dataset. The pre-1930s part exhibits a stationary annual cycle with no long-term trend, which we take as representative of the natural background (Supplementary Figs. S2 and S4). Any increase above this pre-1930 baseline is attributed to anthropogenic emissions and is included as the only input of CH₃Cl and CH₃Br in the NoNatHAL configuration. In addition to CH₃Cl and CH₃Br, anthropogenic emissions of LLH and ODSs species followed the A1 scenario from WMO⁵⁷, including the following halogenated species: CH₃CCl₃, CCl₄, CFC-11, CFC-12, CFC-113, HCFC-22, CFC-114, CFC-115, HCFC-141b, HCFC-142b, H-1301, H-1211 and H-2402. Anthropogenic SLH chlorine compounds (i.e., VSL-Cl halocarbons) were prescribed using time-dependent surface LBCs for CHCl₃ and C₂H₄Cl₂, and surface emissions for CH₂Cl₂ and C₂Cl₄ based on Hossaini et al.⁵⁸ and Claxton et al.⁵⁹

To account for internal variability, particularly in the timing and magnitude of ozone hole emergence and recovery, three ensemble members were run for both the AllHAL and the NoNatHAL scenarios. One member covered the full 1900-2099 period, while the other two focused on the 1960-2099 period; the ensemble spread arises from different initial conditions in the coupled atmosphere–ocean system. For the future projection (2014 onward), all simulations followed the Shared Socioeconomic Pathway SSP3-7.0 scenario⁶⁰. Using a single GHG pathway across experiments avoids introducing additional uncertainty from divergent future GHG trajectories and facilitates the attribution of halogen-driven ozone changes. Moreover, SSP3-7.0 represents a high-GHG pathway that yields relatively high baseline stratospheric ozone by the end of the 21^st century in chemistry-climate models⁴⁵, providing a conservative context for quantifying halogen-driven ozone loss.

The resulting AllHAL ensemble accurately captures the long‑term variability of the total ozone column (TOC). Supplementary Fig. S1 shows the TOC anomaly relative to 1980, comparing the three‑member AllHAL ensemble with observations from the Bodeker Scientific Filled Total Column Ozone Version 3.4 (BSFilledTCO‑v3.4)⁶¹ satellite-based database and with state‑of‑the‑art chemistry-climate models from the Coupled Model Intercomparison Project phase 6 (CMIP6)⁶² considered in previous multimodel evaluations of stratospheric ozone⁴⁵. The AllHAL ensemble reproduces the observed timing and magnitude of the late‑20^th‑century ozone decline and the early‑21^st-century recovery, both for austral spring (October) Antarctic conditions and for the global annual mean. Throughout the observational period (1979–2016), the ensemble mean remains close to the observational record and lies well within the spread of CMIP6 models. Furthermore, the ensemble spread (represented as ±2 standard deviations) encompasses much of the variability exhibited by the other CMIP6 models, indicating that the three-member ensemble considered in this work provides a reasonable estimate of the uncertainty among other state-of-the-art chemistry-climate models. Notably, the AllHAL ensemble also lies within the range of future TOC trends simulated by CMIP6 models under the same SSP3‑7.0 scenario. Taken together, this evaluation confirms that our model configuration captures the key dynamical and chemical processes governing long‑term ozone variability and supports the adequacy of the three‑member ensemble strategy used here, providing a robust basis for the attribution analysis presented in this study.

Ozone loss rates normalized to the O₃ budget

Direct comparison of absolute stratospheric ozone concentration and depletion rates between the AllHAL and NoNatHAL simulations is challenging due to their differing baseline ozone levels. In particular, the absence of natural halogens in NoNatHAL results in systematically higher stratospheric ozone abundance, which, if assessed only in absolute terms, can obscure the chemical impact of natural halogens on ozone. To address this, we employed a normalized metric that expresses ozone loss rates, in units of parts per billion by volume per unit time (ppbv_O3 month⁻¹), relative to the local ozone abundance in each simulation (ppbv_O3), yielding normalized ozone loss rates in units of month⁻¹. This approach accounts for background ozone differences and isolates the chemical effect of natural halogen, ensuring unbiased interpretation of halogen-induced ozone depletion. While non-normalized loss rates produced qualitatively similar patterns, normalization provides a more robust basis for quantitative comparison.

Ozone loss rates assessed in this work are grouped by families of reactions mediated by Cl, Br, I and the Cl-Br cross reactions (Supplementary Table S2), enabling the attribution of the total halogen-mediated ozone loss to individual halogen families.

Breakdown of natural and anthropogenic halogen contributions

To isolate the natural contribution from the total halogen-mediated ozone loss rate, we subtracted the normalized ozone loss in the NoNatHAL simulation from that in AllHAL, representing this difference with empty shading in all figures across the main text and Supplementary. Conversely, results for NoNatHAL (i.e., anthropogenic halogens only) are represented with hatched shading. The same approach was applied to the total inorganic chlorine (Cl_y = Cl + ClO + 2Cl₂ + 2Cl₂O₂ + OClO + HOCl + ClONO₂ + HCl + BrCl + ClNO₂ + ICl), the total inorganic Br (Br_y = Br + BrO + 2Br₂ + BrCl + BrNO₂ + BrONO₂ + HBr + HOBr + IBr) and total inorganic I (I_y = I + 2I₂ + IO + OIO + HI + HOI + INO + INO₂ + IONO₂ + IBr + ICl + 2I₂O₂ + 2I₂O₃ + 2I₂O₄): AllHAL – NoNatHAL represents the natural fraction, while NoNatHAL reflects the anthropogenic component of the total inorganic halogen content. Accordingly, throughout all figures, hatched areas denote anthropogenic contributions and empty shading indicate that from natural sources.

Two complementary procedures were used to evaluate halogen source gas abundance and the release of reactive halogens across the atmosphere. First, source gas abundance was calculated by summing the halogen content in each unreactive organic species (scaled by its halogen atomicity) within its family (see Supplementary Table S1). For CH₃Cl and CH₃Br, which have mixed natural and anthropogenic sources, mixing ratios were partitioned as described in “Model configuration and simulations setup” (see Supplementary Fig. S4). Second, in-situ release of reactive Cl, Br and I at different altitudes (reflecting their distinct lifetimes), was assessed by summing all photochemical release rates from photolysis and reactions with OH, O¹D and Cl (accounting for the total number of halogen atoms released in each process) and then averaging by atmospheric region and time. This approach highlights the contrasting chemical efficiency of long-lived versus short-lived species, as illustrated along the extrapolar stratospheric column (1990–2020; Fig. 2) and the lower stratosphere timeline (Supplementary Fig. S9).

Cl-equivalent ozone loss potential

To assess the relative ozone-depleting efficiency of individual halogen families, we implemented a time-dependent scaling approach. Specifically, the ozone-depleting efficiency of Br_y and I_y in the extrapolar lower-stratosphere was quantified relative to that of Cl_y. First, we computed the ratio of the normalized ozone loss rate (month⁻¹) associated with each halogen family (Supplementary Table S2) and divided it by the abundance of the corresponding inorganic halogen (Cl_y, Br_y or I_y), yielding an efficiency per unit halogen in units of month⁻¹ pptv_X⁻¹ (X = Cl_y, Br_y, I_y). Following previous studies^23,36, the Cl-Br cross reactions of ozone loss were attributed to bromine, given its dominant role in these catalytic cycles. These calculations were performed for the regional mean in the extrapolar lower stratosphere (i.e., 60° S–60° N; from the tropopause to 30 hPa). Finally, we expressed the resulting efficiencies for Br_y and I_y relatives to that of Cl_y, producing time-varying scaling factors (Supplementary Fig. S10) that convert Br_y and I_y abundances into Cl-equivalent units, respectively denoted as Br_y^Cl-eq and I_y^Cl-eq (represented in Fig. 3C). These metrics highlight the disproportionately high ozone depletion potential of iodine and bromine compared to chlorine, despite their significantly lower atmospheric concentrations (Fig. 3A).

Excess of downwelling solar flux at surface

To quantify the excess of shortwave solar radiation reaching the Earth’s surface due to ozone depletion driven by natural halogens (∆SW_↓^O3), we assessed the downwelling solar flux at surface (FSDS). This magnitude is computed by the Rapid Radiative Transfer Model for Global Circulation Models (RRTMG) package, which is implemented by default in CESM2^63,64. RRTMG provides spectrally resolved radiative transfer calculations, accounting for the individual contributions of different atmospheric compounds (e.g., ozone, water vapor, aerosols, etc.) to both incoming and outgoing radiation throughout the atmosphere^65,66. To isolate the component of the radiative effect attributed to ozone changes, we applied the single subtraction approach used in previous work³⁶. Specifically, the excess radiation attributable to natural halogens (∆SW_↓^O3) was obtained as the difference between the two sets of simulations (AllHAL – NoNatHAL) and expressed as the difference between the mean for 1990–2020 and the mean for the unperturbed 1910–1940 period.

Confidence intervals for assessing statistical significance

Uncertainty in values averaged over analysis periods and across the ensemble was assessed using the 95% confidence interval (CI) of the mean. For each analysis period, annual mean values from all ensemble members were pooled into a single sample to evaluate the uncertainty associated with both the interannual variability and the ensemble spread (i.e., internal climate variability). The standard deviation ($\sigma$) was computed across this pooled sample, and the standard error of the mean was estimated as $\sigma /\sqrt{N}$, where $N$ is the total number of annual values in the pooled set. The 95% confidence interval was then defined as:

$${CI}_{95\%}=t_{0.025,N-1}\frac{\sigma }{\sqrt{N}},$$ 1

where $t_{0.025,N-1}$ is the two-tailed Student’s t critical value with $N-1$ degrees of freedom. Particularly in spatial patterns (i.e., Fig. 4A and C), mean values whose absolute magnitude was smaller than their corresponding Cl were considered not statistically significant at the 95% level.

Author contributions

A.S.-L. designed research. J.V., R.P.F., C.A.C. and A.S.-L. initiated the study in collaboration with D.K. and D.M.; J.V. and R.P.F. set up the model, performed the simulations and evaluated the output with the help of C.A.C.; R.P.F and A.R. assessed halogen release rates. All authors discussed the findings and commented on the manuscript. J.V. produced the figures and wrote the manuscript with contributions from all authors.

Peer review

Peer review information

Nature Communications thanks Ross Salawitch and the other, anonymous, reviewer(s) for their contribution to the peer review of this work. A peer review file is available.

Funding

This work was supported by the Spanish Ministry of Science, Innovation and Universities, the Agency and the European Regional Development Fund (Project PID2023-152856OB-I00 IOSTRAT funded by MCIU/AEI/10.13039/501100011033/ERDF, EU) and the European Research Council Executive Agency (Project ERC-2016-COG 726349 CLIMAHAL). R.P.F. would like to thank financial support from the Argentine Secretary of Science (REMATE IF-2023-85161983-APN).

Data availability

The Community Earth System Model (CESM) is developed and maintained by the National Center for Atmospheric Research (NCAR) under the support of the NSF. The released CESM2-SLH tag used in this work (cesm2.2-asdbranch_slh_tag) can be cloned from GitHub (https://github.com/RafaPedroFernandez/CESM/tree/cesm2.2-asdbranch_slh_tag), which was originally forked from the latest CESM2 branch available in the NCAR repository (https://github.com/ESCOMP/CESM/tree/cesm2.2-asdbranch). The total ozone column observational (BSFilledTCO‑v3.4) and CMIP6 data used in the validation of the model setup (Supplementary Figs. S3 and S4) are respectively available from https://www.bodekerscientific.com/data/total-column-ozone and https://metagrid.esgf-west.org/search/cmip6. All data to reproduce all figures and analysis in this study are available in a Zenodo database (https://doi.org/10.5281/zenodo.21234495)⁶⁷.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at https://doi.org/10.1038/s41467-026-76600-6.