Antarctic ozone hole modifies iodine geochemistry on the Antarctic Plateau

Polar stratospheric ozone has decreased since the 1970s due to anthropogenic emissions of chlorofluorocarbons and halons, resulting in the formation of an ozone hole over Antarctica. The effects of the ozone hole and the associated increase in incoming UV radiation on terrestrial and marine ecosystems are well established; however, the impact on geochemical cycles of ice photoactive elements, such as iodine, remains mostly unexplored. Here, we present the first iodine record from the inner Antarctic Plateau (Dome C) that covers approximately the last 212 years (1800-2012 CE). Our results show that the iodine concentration in ice remained constant during the pre-ozone hole period (1800-1974 CE) but has declined twofold since the onset of the ozone hole era (~1975 CE), closely tracking the total ozone evolution over Antarctica. Based on ice core observations, laboratory measurements and chemistry-climate model simulations, we propose that the iodine decrease since ~1975 is caused by enhanced iodine re-emission from snowpack due to the ozone hole-driven increase in UV radiation reaching the Antarctic Plateau. These findings suggest the potential for ice core iodine records from the inner Antarctic Plateau to be as an archive for past stratospheric ozone trends.

at Dome C. In details, the ice core depth used for this study was converted into meters of water 44 equivalent using the available average density profile for Dome C 2 to fit the depth profile 45 calculated by Gautier et al.. Then, to match our core collected in 2012 with the 5 ice core 46 chronology calculated by Gautier et al. 2016, we used the annual snow accumulation data 47 Germany) was used to analyze major ions including acetate, methanesulphonic acid (MSA), 72 chloride, ammonium and nitrate 5 . The identification was performed using an IC 73 (ThermoScientific™ Dionex™ ICS-5000, Waltham, US) equipped with an anionic exchange 74 column (Dionex Ion Pac AS 19 2 × 250 mm) and a guard column (Dionex Ion Pac AG19 2 × 50 75 mm). Sodium hydroxide (NaOH), used as mobile phase, was produced by an eluent generator 76 (Dionex ICS 5000EG, Thermo Scientific). The NaOH gradient with a 0.25 mL min -1 flow rate 77 was: 0-6 min at 15 mM; 6-15 min gradient from 15 to 45 mM; 15-23 min, column cleaning with 78 45 mM; 23-33 min; equilibration at 15 mM. The injection volume was 100 µL. A suppressor 79 (ASRS 500, 2 mm, Thermo Scientific) removed NaOH before entering the MS source. The IC 80 was coupled to a single quadrupole mass spectrometer (MSQ Plus™, Thermo Scientific™) with 81 a negative electrospray source (ESI) that operated with a temperature of 550 °C and a needle 82 voltage of 3 kV. The mass spectrometer parameters are reported in Barbaro

S3.2 -Correlation test 119
We have used the Pearson correlation (R) of the datasets before and after the identified change point as 120 an additional confirmation of the occurrence of the tipping point in 1974. After splitting the identified 121 dataset into two segments, for each one we have calculated R and the associated p-value by converting 122 R-value to a t-statistic.

123
S3.3 -Dome C iodine ice core concentration and stratospheric ozone. 126 We computed the correlation between annual ice core measured [I] with modelled sunlit total ozone 127 columns above Dome C (TOC DC ). The correlation obtained for the ozone hole period 1975 -2011 is: r = 128 0.398, p-value = 0.015 while for the pre-ozone hole period 1950 -1974 the correlation obtained is r = 129 The negative association between annual ice core [I] and AF 300 is confirmed by the Pearson coefficient 133 that becomes negative and more significant during the ozone hole period, while is not significant for the Thanks to the availability of the data from two snow trenches in an area close to the drilling site (see S2

138
for further details), we validated the ice core chemical analysis with these samples. Merging the dataset 139 was necessary to complete the sodium record that presents a gap in the period 1989-1997 due to 140 analytical failure. The good agreement among the three records allowed us to merge the three datasets 141 for both Na and I. We used the same matching procedure for Na and I based on the chronology described   Figure S1). The first one 154 has a low temporal resolution due to the relatively scarce annual accumulation (80 mm we yr -1 ) since the 155 aim was to investigate the multi-millennia variability of iodine species and only a few samples covered 156 the Holocene 7 . On the contrary, the Law Dome ice core (740 mm we yr -1 ) covered the period from 1929 157 to 1988 at a sub-annual resolution and does not present a decrease in iodine concentration from the late 158 70s to the early 80s 8 . We argue that this distinct behaviour is related to the mean location height, the

198
The surface concentrations of CO 2 , CH 4 , H 2 and N 2 O were specified following previous works 19,20 .

199
The CAM-Chem configuration used here extends from the surface to approximately 40 km (3.5  Antarctic ice-records as described above in section S4 and shown in Figure S1.

238
The 300.5 nm bin, which has a mean bandwidth of 3.625 nm, has been used for adressing the trend in the

290
In addition to all the processes described in the main text, we further discuss here the possible 291 effect of sample storage and transport on iodine concentration. It has been reported that firn samples, 292 being characterized by a lower density, can lead to significant iodine losses during storage 29 . However, 293 this was reported only for layers having a density close to 0.83 kg L -1 (i.e. close to the firn to ice 294 transition). In this record, the density values in the first 3.5 meters (i.e. when the iodine decrease was 295 detected) slightly increased from 0.32 to 0.38 kg L -1 and this cannot explain the observed two-fold  iodine day-to-night variability and seasonal variability, respectively 8,30 . A key parameter that can 308 enhance and activate iodine volatilization from snow\firn\ice samples is the exposure to light 7 but all the 309 samples from both the shallow core and the snow pits were stored under dark conditions until they were 310 processed and analysed. We underline that correct sample storage is fundamental to minimize iodine 311 losses from ice core and snow samples. To assure reliable sample representativeness, the cold chain 312 (temperature equal or below -20°C) as well as dark conditions must be guaranteed for the entire 313 transportation process.
Iodine can also be lost during the analysis from melted ice samples. To avoid this risk, the 315 samples were stored at temperature below -20°C and melted only immediately before the analysis. Any