Skip to main content

Data from: Impact of ground-level ozone on tropical forests

Cite this dataset

Cheesman, Alexander et al. (2024). Data from: Impact of ground-level ozone on tropical forests [Dataset]. Dryad.


Elevated ground-level ozone (O3), a result of human activity, is known to reduce plant productivity but its impact on tropical forests remains unclear. Here for the first-time we measured the O3 susceptibility in a range of ten tropical tree species and use these data to determine how changing O3 exposure has impacted tropical forest productivity and the global carbon cycle.

README: Data from: Impact of ground-level ozone on tropical forests

Represents 32 data files (9.32MB) and an R script. Data files are a combination of .csv and netCDF and include both experimental data and model results.

Description of the data and file structure

Experimental Data

Consists of 13 data files representing a summary of experimental conditions and data produced by a series of experiments using Open Top Chambers to determine the O3 susceptibility of  ten tropical tree species.  

Experimental conditions: Ten data files of O3 concentrations and meteorological conditions for the OTC experiments conducted on ten species of tropical tree seedlings listed in Extended Data Table 1 of the manuscript. Each file has the same structure and parameters (see below)

            "Ba_merged_data.csv" = Brachichyton acerifolius (Malvaceae)
            "Cb_merged_data.csv" =Carallia brachiata (Rhizophoraceae)
            "Ci_merged_data.csv"=Calophyllum inophyllum (Calophyllaceae)
            "Cr_merged_data.csv"=Chionanthus ramiflorus (Oleaceae)
            "Dd_merged_data.csv"=Darlingia darlingiana (Proteaceae)
            "Fp_merged_data.csv"=Flindersia pimenteliana (Rutaceae)
            "Hn_merged_data.csv" =Homalanthus novo-guineensis (Euphorbiaceae)
            "Ie_merged_data.csv" =Inga edulis (Fabaceae)
            "Sg_merged_data.csv" =Syzygium gustavioides (Myrtaceae)
            "Tc_merged_data.csv" =Theobroma cacao (Malvaceae)

With each of the ten files each consisting of 17 columns with the same format, set up to act as an input for DO3SE model.

              Day_of_year= Julian day of year
              Hour_of_day = h interval of the day
              Temperature = Air temperature (°C)
              VPD = Vapour pressure deficit (kPa)
              PAR = Photosynthetically active radiation (µmol /m2 /s)
              Pa = atmospheric pressure (kPa)
              Wind = wind speed = 0.75 (m /s)
              ppt = precipitation (mm) set to 0 as plants irrigated
              Cham_1 = O3 concentrations in chamber 1 (ppb)
              Cham_2 = O3 concentrations in chamber 2 (ppb)
              Cham_3 = O3 concentrations in chamber 3 (ppb)
              Cham_4 = O3 concentrations in chamber 4 (ppb)
              Cham_5 = O3 concentrations in chamber 5 (ppb)
              Cham_6 = O3 concentrations in chamber 6 (ppb)
              Cham_7 = O3 concentrations in chamber 7 (ppb)
              Cham_8 = O3 concentrations in chamber 8 (ppb)

Experimental Biomass:  A singe file containing the final harvest biomass and leaf-level functional traits of plants from ten tropical species grown in OTC's under a range of O3 concentrations. Values of NA indicates "value not available"

Experimental_Biomass.csv” a single file consisting of 19 columns including.
              Code = Species code (See Table 1)
              Plant_ID = Plant identifier within experiment
              Cham_ID = Chamber identifier (i.e. 1 to 9)
              Leaves_dw = Final harvest dry weight of leaf biomass (g)
              Petiole_dw = Final harvest dry weight of petiole biomass (g)
              Stems_dw = Final harvest dry weight of stem biomass (g)
              Roots_dw = Final harvest dry weight of root biomass (g)
              AGB = Total Above ground biomass (g)
              Biomass = Total Biomass (g)
              LA_Av = Leaf area average of target leaves (cm2)
              LMA = Whole leaf mass per unit area of target leaves (g /m2)
              LMA_lamina = lamina only leaf mass per unit area of target leaves (g /m2)
              Flagged = Individual flagged and removed from biomass analysis due to issues with data (e.g. herbivory, irrigation failure)
              TAC = Total antioxidant capacity of target leaves (mg Ascorbic acid equivalents /g)
              TPC = Total phenolic content of target leaves (mg Gallic acid equivalents /g)
              delta_13C = leaf mass δ15N (‰)
              delta_15N = leaf mass δ13C (‰)
              TC = Total carbon content (%)
              TN = Total nitrogen content (%)

DO3SE model parameters. A single file consisting of input parameters required for species-specific parameterization of DO3SE model to calculate O3 fluxes.

DOS3E_model_parameters.csv” a single csv file containing 15 columns
               Code = Species code (see Table1)
               m = Species-specific sensitivity to An (dimensionless)
               g0 = minimum stomatal conductance (mmol H2O m−2 s)
               Vcmax = maximum rate of carboxylation (µmol m−2 s-1)
               Jmax = maximum rate of elctron transport at 25°C (µmol m−2 s−1)
               Gmax_O3 = maximum stomatal conductance to O3 (mmol m−2 PLA s−1)
               fmin = fmin (fraction)
               Tmin = Miniumum temperature (°C)
               Tmax = Maximum temperature (°C)
               Topt = Optimum temperature (°C)
               VPDmax = VPD for max stomatal conductance (kPa)
               VPDmin = VPD for min stomatal conductance (kPa)
               PAR = light response fraction (unitless)
               Lm = effective leaf width (m)

Dose response functions – A single file consisting of the linear decline in relative biomass against O3 exposure metrics.

Dose_response_functions.csv” = file consists of 6 columns including
              Code= Species Code (See Table 1)
              POD0_J_DRF = Relative biomass decline per unit POD0 parametrized using Jarvis (% /POD0)
              POD1_J_DRF = Relative biomass decline per unit POD1 parametrized using Jarvis (% /POD1)
              POD0_P_DRF = Relative biomass decline per unit POD1 parametrized using Photosynthesis model (% /POD0)
              POD1_P_DRF = Relative biomass decline per unit POD1 parametrized using Photosynthesis model (% /POD1)
              AOT40_DRF = relative Biomass decline per unit AOT40 (% /AOT40)

Model Data

A total of 19 data files representing both gridded(netCDF) data at a spatial resolution of 1.25◦ latitude by 1.875◦ longitude, and .csv summaries of models specifically data files represent:

Forest cover types: Fractional cover of different existing and potential forest. Representing a regridding of the available spatial data sets at the resolution of JULES modelling framework.”: variable = area, units=m2. Total area of JULES modelling grid cell extents.”: variable = CF, units=frac. Fractional cover of tropical forest PFT (i.e. BET-Tr) in the year 2015 as per Harper et al., (Harper et al., 2023)”: variable = SF, units=frac. Fractional cover of existing secondary tropical forest as per Vancutsem et al., (2021)”: variable = PF, units =frac. Fractional cover of potential secondary forest cover as per (Griscom et al., 2017)

 Ozone concentrations: representing global gridded outputs from UKESM1 part of a CMIP6 historical simulations.” variable = O3, units =ppb. Predicted average annual O3 concentrations from 1900 to 1909.” variable = O3, units =ppb. Predicted average annual O3 concentration from 2005 to 2014.
O3_change” variable = O3, units =ppb. Change in average annual mean O3 concentrations when comparing pre-industrial (1900-1909) and present day (2005-2014) averages.

Calibration of JULES: representing model outputs for relative NPP decline for forested grid cells for the year 2009.

“JULES_susceptibility.csv” = file containing 3 columns.
             Rel_NPP = relative decline in annual NPP (%)
             POD1 = calculated annual POD1 (mmol/ m2 )
             Susceptibility = level of O3 susceptibility modelled (i.e. low, moderate or high)

Impacts of O3 on NPP: representing comparison of JULES model simulations of Net Primary productivity (NPP) in regions of current forest cover comparing simulations with fixed O3 concentrations (i.e. 1900 to 1909) and those with transient O3 (i.e. 1900 to 2014 – taken as 1 h metrology from UKESM1) and assuming one of three O3 susceptibilities.” variable =NPP, units =kg-C /y. Predicted decline in NPP when assuming low O3 susceptibility.” variable =NPP, units =kg-C /y. Predicted decline in NPP when assuming moderate O3 susceptibility.” variable =NPP, units =kg-C /y. Predicted decline in NPP when assuming high O3 susceptibility.” variable =NPP, units =% of control. Predicted decline in NPP when assuming low O3 susceptibility.” variable =NPP, units =% of control. Predicted decline in NPP when assuming moderate O3 susceptibility.” variable =NPP, units =% of control. Predicted decline in NPP when assuming high O3 susceptibility.

Accumulated impact of O3 on tropical carbon stocks: representing the yearly summary of the comparison between simulated terrestrial carbon stocks (i.e. soils and vegetation) using fixed pre-industrial O3 exposure and those with transient O3 when assuming one of three O3 susceptibilities

Delta_carbon.csv” = file consists of 7 columns.
          Year= year of simulation (i.e 1900-2014)
          Veg_high = Difference in vegetation assuming high O3 susceptibility (Pg-Carbon)
          Veg_mod = Difference in vegetation assuming moderate O3 susceptibility (Pg-Carbon)
          Veg_low = Difference in vegetation assuming low O3 susceptibility (Pg-Carbon)
          Soils_high = Difference in soils assuming high O3 susceptibility (Pg-Carbon)
          Soils_mod = Difference in soils assuming moderate O3 susceptibility (Pg-Carbon)
          Soils_low = Difference in soils assuming low O3 susceptibility (Pg-Carbon)

JULES outputs across tropical forests- three data files constituting the predicted NPP decline for each modelled grid cell across tropical forest types assuming either a low, moderate or high O3 susceptibility.  

Forest_types_NPP_Low.csv” = file contains 4 columns including.
         index= Julian day of year
         NPP_decline = calculated NPP decline (%)
         weights = Area weighting (fraction of forest cover x area of grid cell . m2)
        Type= Type of tropical forest (i.e. Current Forest, Secondary Forest, Potential Forest)

 “Forest_types_NPP_Mod.csv” = file contains 4 columns including.
         index= Julian day of year
         NPP_decline = calculated NPP decline (%)
         weights = Area weighting (fraction of forest cover x area of grid cell . m2)
        Type= Type of tropical forest (i.e. Current Forest, Secondary Forest, Potential Forest)

Forest_types_NPP_High.csv” = file contains 4 columns including.
         index= Julian day of year
         NPP_decline = calculated NPP decline (%)
         weights = Area weighting (fraction of forest cover x area of grid cell . m2)
         Type= Type of tropical forest (i.e. Current Forest, Secondary Forest, Potential Forest)

Ozone concentrations across forest types: Representing a summary of O3 concentrations seen across different forest types in the present and in the year 2050. Represents a summary of CMIP6 model predictions under different Shared Socioeconomic Pathways.

SSP_2050.csv” = file contains 6 columns
          Type = forest cover type (i.e. Tropical Forest, Secondary forest, Potential forest)
          pd.forest. = present day average O3
          X2050.forest.= average [O3] in 2050 (ppb)
          X2050.forest.std = standard deviation in [O3] in 2050 (ppb)
 = Difference in O3 between 2014 and 2050
          Scenario = Shared socioeconomic pathway O3 modelled by UKESM1 CMIP6 under (i.e. SSP1-2.6, SSP2-4.5, SSP3-7.0, SSP5-8.5)

Sharing/Access information

Data inputs reported here were was derived from the following sources:

  • Griscom BW, Adams J, Ellis PW, Houghton RA, Lomax G, Miteva DA, et al. Natural climate solutions. Proceedings of the National Academy of Sciences of the United States of America 2017; 114: 11645-11650.
  • Sellar AA, Jones CG, Mulcahy JP, Tang Y, Yool A, Wiltshire A, et al. UKESM1: Description and Evaluation of the U.K. Earth System Model. Journal of Advances in Modeling Earth Systems 2019; 11: 4513-4558
  • Vancutsem C, Achard F, Pekel J-F, Vieilledent G, Carboni S, Simonetti D, et al. Long-term (1990–2019) monitoring of forest cover changes in the humid tropics. Science Advances 2021; 7: eabe1603.
  • Harper KL, Lamarche C, Hartley A, Peylin P, Ottlé C, Bastrikov V, et al. A 29-year time series of annual 300 m resolution plant-functional-type maps for climate models. Earth Syst. Sci. Data 2023; 15: 1465-1499.


An R script is included that allows for the creation of all figure panels used in the manuscript.


This deposit represents i) data on experimental conditions and results from work investigating the impacts of ozone (O­3) on saplings of ten tropical tree species, ii) how this empirical data was used to parameterize a global land surface model as well as iii) global model outputs for the impacts of O3 on tropical forests and the global carbon cycle and iv) formatted data and R-code needed to replicate all figures within the associated manuscript.

Experimental facility

All empirical work was conducted at the joint University of Exeter (UoE) and James Cook University (JCU) TropOz research facility ( located at the James Cook University’s Environmental Research Complex ( on the Nguma-bada campus in far-north Queensland, Australia. The facility consists of nine independently controlled and monitored Open Top Chambers (OTC) which allows for the examination of plants grown under ambient CO2 and nine different O3 concentrations using a gradient experimental design (Kreyling et al., 2018). The chambers (internal volume 22.2 m3) were ventilated with charcoal filtered air augmented with O3 generated on site and supplied to each chamber between 8:00 and 17:00 to attain a range of nine different O3 exposures, with typical mean chamber concentrations for each experiment ranging from 25 to 112 ppb during the hours of O3 exposure. A single UV-absorption O3 analyser (Model 205, 2B technology, Boulder CO, USA) was used to continuously monitor [O3] within sample air being sequentially sourced from each chamber. A typical sampling sequence, accounting for solenoid switching and deadspace turnover, resulted in 4 readings (of 10 sec instrument averages) per chamber every ~15 min. This chamber-data was averaged and then interpolated via linear approximation to produce continuous dataset at a five min resolution for each chamber. Meteorological measurements including air temperature (T), relative air humidity (RH), shortwave radiation and photosynthetically active radiation (PAR) were recorded using a single meteorological monitoring station (Campbell Scientific, Logan UT, USA) established in the central OTC with data averaged over every five or ten minutes. These high resolution data were averaged to provide DO3SE model inputs (see calculating O3 flux below).

Plant material

Ten tropical tree species (Table 1) from nine different families were selected to represent a broad a range in successional-stage and leaf morphological traits (i.e. leaf mass per unit area ranging from 58.1 to 142.5 g m−2). Species response to O3 were examined under a rolling experimental program where different species were introduced to the chambers at different times and considered as independent of each other. All seedlings were sourced locally and planted when ~20 cm tall into either 20 or 60 L pots filled with locally sourced ‘garden mix’ topsoil mixed 3:1 with Quincan (a local form of scoria) to improve drainage. Potted saplings were acclimatized to full sun before 3 or 4 replicate plants were transferred into each OTC, with typical O3 fumigation for 9h per day (from 08:00 to 17:00) and plants grown for between 61 and 191 (average =150) days. Plants were irrigated daily using individual drip irrigation and fertilized as required using a controlled release fertilizer (Osmocote Native Formula, ScottsMicacle-Gro, Marysville, OH, USA).

At the end of the experiment, plants were harvested for total biomass and separated into leaves, stems, and roots to calculate biomass partitioning. Dry biomass was determined after oven-drying at 70 °C until constant weight was achieved.

 Table 1: Details of ten tropical tree species grown in open top chambers under various O3 concentrations as part of this data deposit.

Code Species Family Start of experiment End of experiment
Ba  Brachichyton acerifolius  Malvaceae 10/09/2021 10/11/2021
Cb  Carallia brachiata  Rhizophoraceae 25/05/2022 17/11/2022
Ci  Calophyllum inophyllum  Clusiaceae 08/04/2021 24/09/2021
Cr  Chionanthus ramiflorus  Oleaceae 18/05/2022 18/11/2022
Dd  Darlingia darlingiana  Proteaceae 28/03/2022 05/10/2022
Fp  Flindersia pimenteliana  Rutaceae 30/07/2022 30/11/2022
Hn  Homalanthus novo-guineensis  Euphorbiaceae 14/07/2020 10/12/2020
Ie  Inga edulis  Fabaceae 28/04/2021 19/08/2021
Sg Syzygium gustavioides Myrtaceae 01/5/2022 05/10/2022
Tc  Theobroma cacao  Malvaceae 13/07/2020 13/01/2021

Leaf sampling and biochemical characterization

Leaf functional traits: Between 6 and 12 fully expanded mature leaves (leaf number dependent upon leaf size) were collected from every plant at the end of the O3 exposure period. These target leaves were scanned to determine leaf area and shape, weighed to determine fresh mass, and half were dried (70°C, 72h) for determination of whole leaf-LMA (whole leaf mass per unit area g m-2). In the remaining fresh leaves, lamina tissue (without midrib or major veins) was excised using a scalpel. Lamina tissue was rescanned, wrapped in tin foil and snap-frozen in liquid N2 - to await lyophilization, and weighing to determine lamina-LMA (lamina only LMA), and biochemical analysis.

Freeze-dried leaf lamina samples were ground into fine powder (Rocklabs Bench Top Ring Mill, Scott, Dunedin New Zealand) and stored in airtight vials before analysis for biochemical characteristics. The total carbon (TC) and nitrogen (TN) as well as δ13C and δ15N were determined using a Costech Elemental Analyser fitted with a zero-blank auto-sampler coupled via a ConFloIV to a ThermoFinnigan DeltaVPLUS using Continuous-Flow Isotope Ratio Mass Spectrometry (EA-IRMS) at James Cook University’s Advanced Analytical Centre. The δ13C results are reported as per mil (‰) deviations from the VPDB reference, while δ15N is compared to atmospheric N.

Total antioxidant capacity (TAC) was measured by the ferric reducing antioxidant power (FRAP) assay (Benzie and Strain, 1996) with some modifications. Leaf samples (~30 mg) were extracted in cold 50% acetone (Ritmejerytė et al., 2019). For FRAP assay, fresh FRAP reagent was prepared by mixing 300 mM acetate buffer (pH 3.6), 10 mM TPTZ (2,4,6-tripyridyl-s-triazine) solution (in 40 mM HCl) and 20 mM FeCl3.6H2O solution and kept it warm at 37°C in a water bath until used. The volume ratio of the acetate buffer, TPTZ and FeCl3.6H2O solutions was 10:1:1 respectively. Ascorbic acid was used to make a standard calibration curve in the range of 0–250 µg mL−1. Leaf extracts (10 µL) were placed in 96-well plate to react with the FRAP solution (190 µL) for 30 min in the dark. The absorbance of the reaction mixture was then measured at 593 nm with a microplate reader (FLUOstar OPTIMA, BMG LABTECH Pty. Ltd). Results were expressed as ascorbic acid equivalents (mg AAE g−1 dry weight).

Total phenolic content (TPC) was measured by the Folin–Ciocalteau method (Cork and Krockenberger, 1991; Singleton and Rossi, 1965). Briefly, a 20-μl aliquot of the leaf extract and 380 µL distilled H2O was mixed with 25 µL Folin–Ciocalteu reagent. After 3 min, 75 µL 20% Na2CO3 (w/v) was added to the reaction mixture and incubated in the assay tubes at room temperature for 20 minutes. The absorbance of the mixture was then measured at 765 nm with a microplate reader (FLUOstar OPTIMA, BMG LABTECH Pty. Ltd). Gallic acid was used as a standard and total phenolic content was expressed as Gallic acid equivalents (mg GAE g–1).

Calculating O3 flux

To determine the accumulated Phytotoxic Ozone Dose (POD1 , mmol m−2, above a threshold 1 nmol O3 m−2 projected leaf area (PLA) s−1) we used the Deposition of O3 for Stomatal Exchange (DO3SE) model vn 3.1 ( (Büker et al., 2012) employing two methods of modelling dynamic stomatal conductance (gs); specifically a combined photosynthesis-stomatal conductance model assuming optimal stomatal behaviour (Ball et al., 1987; Medlyn et al., 2011) and an empirical-multiplicative gs model.

The combined photosynthesis-stomatal conductance model was parameterized for each species using estimates of photosynthetic characteristics (e.g. Vcmax and Jmax) and a stomatal response variable g1 (Ball et al., 1987; Medlyn et al., 2011)  (Extended Data Table 1) determined using a portable photosynthesis analyser (LI-6400XT, LiCOR Biosciences, Lincoln NE, USA). Leaf-level gas-exchange data was collected on the newest fully developed leaf of plants grown under the lowest [O3] (control plants, typically n=4) and consisted of both ACi curves and survey measurements collected every three minutes for ~24 h per leaf using an inlet-buffer volume and with the LI-6400xt tracking ambient PAR and temperatures. Average night-time conductance (i.e. 18:00 to 06:00) was taken as g0 while day-time (06:00 to 18:00) gas exchange data were used to estimate leaf-level g1 using ‘FitBB’  function of the plantecophys package (Duursma, 2015). Fitting of ACi curves to determine Vcmax and Jmax were carried out using ‘plantecophys::fitaci’ using updated default temperature response parameters (

In addition, we carried out a parameterization of DO3SE using an empirical-multiplicative gs model (Emberson et al., 2000; Jarvis, 1976).  In the weeks prior to harvest, measurements of gs were made on the youngest fully expanded mature leaf of all trees across the range in O3 exposure using a SC-1 Leaf Porometer (Decagon Devices, Pullman, WA, USA). Point measurements of gs on both abaxial and adaxial leaf surfaces were collected over a range of time and weather conditions and coupled to the closest meteorological data as recorded by the experimental system. These values were used to parameterize DO3SE model (See data file “DOS3E_inputs.csv” ), using the method described by Hayes et al., (2020) allowing for the calculation of POD1 as per CLRTAP (2017).

Ozone dose-response functions

Species level O3 dose-response functions were calculated using the linear decline in chamber-average relative biomass of each species against POD1 determined using both a combined photosynthesis-stomatal conductance and empirical-multiplicative model of gs. Relative biomass of each chamber was derived using the y-intercept of a regression between average chamber total biomass and O3 exposure (i.e., the hypothetically maximum biomass at POD1 = 0). It is important to note that this approach does not account for potential hormetic responses (Agathokleous et al., 2019) however the slope coefficients of the linear dose response function are still commonly used to calculate O3 susceptibility (Pleijel et al., 2022). Although the two methods used for calculating POD1 generally agreed we selected the observed distribution in O3 susceptibility using the combined photosynthesis-stomatal conductance model to calibrate the JULES given the same model is employed within the DGVM itself.

Data quality control methods

Meteorological data collected inside the OTC’s during the experimental runs was done so with factory calibrated sensors and was compared against a local meteorological station maintained at JCU’s Environmental Research Complex to ensure sensible data. O3 concentration monitoring was checked against a zero-air standard (activated charcoal filtered air) every sampling round (i.e. ~22 min). Final biomass partitioning was determined using annually calibrated scales with data input and scale accuracy checked by reweighing suspected outliers after additional drying time.  

Modelling impacts of O3 exposure on tropical forests

To examine the potential implications of O3 on tropical forests, we used the Joint UK Land Environment Simulator (JULES) vn 5.6. (Best et al., 2011; Clark et al., 2011) a land surface model used to study soil-vegetation-atmosphere interactions at a spatial resolution of 1.25° latitude by 1.875° longitude. This modelling framework incorporates continuous and spatially explicit environmental information (e.g. meteorological conditions, and [O3]) and includes the representation of vegetation responses to atmospheric composition, e.g. CO2 (Huntingford et al., 2013), aerosols (Mercado et al., 2009; Rap et al., 2018), and O3 (Leung et al., 2022), as well as interaction with other abiotic factors such as temperature (Huntingford et al., 2017), drought (Harper et al., 2021), and changes in nutrient cycling (Huntingford et al., 2022).

The O3 damage scheme employed in JULES is the same as that implemented by Sitch et al. (2007) (Sitch et al., 2007) and Oliver et al., (2018) (see Eqs. 1, 2, and 3) with updates to plant functional types (PFT) and their physiology as per Harper et al, (2016, 2018) (Harper et al., 2016; Harper et al., 2018), and of photosynthetic and stomatal functional traits following Oliver et al., (2022). The schema works by modifying net photosynthesis (Anet) and stomatal conductance (gs) by an O3 damage factor (F) at every time step. With F defined (Eq. 1) by a sensitivity parameter (α) and the flux of O3 above a threshold (y) so that A decreases linearly (Eq. 2) as O3 flux increases above the threshold and the rate of decrease depends on the sensitivity parameter. The decrease in A affects the Net Primary productivity (NPP), and the model assumes that a) O3 damage is instantaneous at the point of uptake and b) results in a coordinated reduction in gs (Eq.3). 

Eq.1        F = 1 – α × (Flux-O3 > y)

Eq.2        Amod =Anet ×

Eq.3        gmod = gs × F

JULES routinely scales flux of O3 to the canopy by calculating at each canopy layer and for shaded and sunlit leaves separately, whereas DO3SE (used to determine O3 susceptibilities) calculates for the uppermost sunlit leaves only. Therefore, to compare the dose response function in JULES during calibration to those observed using DO3SE we used the O3 flux to the top canopy layer sunlit leaves only, as per Oliver et al., (2018).

Calibrating JULES to observed O3 susceptibility of tropical trees

Given the broad range in O3 susceptibility observed across tropical trees species tested we calibrated the tropical broadleaf evergreen (BET-Tr) PFT to replicate one of three O3 susceptibilities based upon the range we observed (i.e low =  −0. 38 % POD1−1, moderate = −0.50 % POD1−1, and high = −0. 95 % POD1−1) by iterative adjustment of the O3 response factor of photosynthesis (α) after comparison of annual NPP (years 2009 to 2011) with observed dose-response functions (DRF; % biomass decline POD1−1) (Table 2). For additional modelled tropical PFT’s we chose to set the α of C3 grasses to that observed in C3 trees given the lack of comparable data, and for C4 grasses that observed in Saccharum spontaneum cv. Mandalay (Cheesman et al., 2023) (i.e. α = 0.04 at a threshold of 2 nmol m−2 s−1).

Table 2: Calibration of JULES O3 response factor (α) for low, moderate and high O3 susceptibility for the tropical broadleaf evergreen tree PFT (i.e. BET-Tr). The term α was calibrated to replicate three observed susceptibilities in biomass decline (i.e. low, moderate and high) by iterative adjustment.

O3 susecptibility JULES O3 sensitivity paramater (α)

DRF-slope target 

(% POD1 -1)

DRF -slope observed 

(% POD1 -1)

Low 0.03 0.376 0.381
Moderate 0.05 0.499 0.508
High 0.12 0.946 0.949

Simulation details

We ran pan-tropical simulations for the period 1st Jan 1900 to 31st Dec 2014 (1900 to 2014) assuming a homogeneous O3 susceptibility of tropical trees to one of the three levels identified (i.e. low, moderate, high) using dynamic vegetation (Cox, 2001; Harper et al., 2018), allowing PFT fractional cover and Leaf Area Index (LAI) to change with varying environmental conditions over the last century. Standard inputs across simulations included variation in [CO2] and land-use land-cover change as used in the Global Carbon Budget 2020 (Friedlingstein et al., 2020), with meteorology and forcing data from CRU-JRA v 2.3 (University of East Anglia Climatic Research Unit; Harris I.C., 2022). To identify the impacts of anthropogenic-derived O3, we compared simulations with ‘fixed’ [O3] representing the average seen between 1900 and 1909 to those with a ‘transient’ [O3] (i.e. 1900 to 2014). In both cases [O3] were taken as an hourly input from UKESM1 (Sellar et al., 2019) part of a CMIP6 historical  simulation (Eyring et al., 2016), all simulations included an initial 1000-year model spin up (1900 to 1920 climatology cyclically repeated) with fixed [O3].

Impacts of changing O3 over the 20th Century.

To examine the influence of changing [O3] on current pan-tropical NPP, we examined the difference in modelled 10-year mean (2005 to 2014) NPP of the BET-Tr PFT, in both fixed and transient [O3] simulations. All BET-Tr performance data were weighted by current observed forest extent using the ESA CCI landcover fractions (Extended Data Fig. 6) from 2015 translated into JULES PFTs as per Harper et al., (2023) and grid-cell area. With broad geographic areas assigned using Natural Earth

In considering the cumulative impacts of changing [O3] upon tropical terrestrial carbon pools we examined the time series of total terrestrial pool size between static and transient [O3] simulations; including impacts of both NPP decline, shifts in dynamic vegetation, and changes in soil biogeochemistry. Rates of change were considered over the entire simulation (i.e., 115 years, 1900 to 2014) as well as since the year 2000.

Impacts of O3 on nature-based solutions

To examine the potential impacts of O3 susceptibility on nature-based solutions to climate change we examined the impact of recent (i.e. 2014) air quality in regions of secondary forest regrowth (Vancutsem et al., 2021) and potential forest restoration (Griscom et al., 2017) assuming the same range in O3 susceptibility across forest types. Original-resolution binary data were re-grided to the resolution of JULES model environment using the reproject function in google earth engine (see data deposit). This allowed the calculation of each modelled grid cell covered by a) existing forest, b) secondary forest regrowth and c) area of potential forest restoration.


Agathokleous E, Belz RG, Calatayud V, De Marco A, Hoshika Y, Kitao M, et al. Predicting the effect of ozone on vegetation via linear non-threshold (LNT), threshold and hormetic dose-response models Predicting the effect of ozone on vegetation via linear non-threshold (LNT), threshold and hormetic dose-response models. Science of the Total Environment 2019; 649: 61-74.

Ball J, Berry J, Woodrow I. A model predicting stomatal conductance and its contribution to the control of photosynthesis under different environmental conditions. In: J. B, editor. In Progress in Photosynthesis Research. Martinus Nijhoff Publishers, Dordrecht, The Netherlands., 1987, pp. 221-224.

Benzie IF, Strain JJ. The ferric reducing ability of plasma (FRAP) as a measure of “antioxidant power”: the FRAP assay. Analytical biochemistry 1996; 239: 70-76.

Best MJ, Pryor M, Clark DB, Rooney GG, Essery RLH, Ménard CB, et al. The Joint UK Land Environment Simulator (JULES), model description – Part 1: Energy and water fluxes. Geosci. Model Dev. 2011; 4: 677-699.

Büker P, Morrissey T, Briolat A, Falk R, Simpson D, Tuovinen JP, et al. DO3SE modelling of soil moisture to determine ozone flux to forest trees. Atmos. Chem. Phys. 2012; 12: 5537-5562.

Cheesman AW, Brown F, Farha MN, Rosan TM, Folberth GA, Hayes F, et al. Impacts of ground-level ozone on sugarcane production. Science of The Total Environment 2023; 904: 166817.

Clark DB, Mercado LM, Sitch S, Jones CD, Gedney N, Best MJ, et al. The Joint UK Land Environment Simulator (JULES), model description – Part 2: Carbon fluxes and vegetation dynamics. Geosci. Model Dev. 2011; 4: 701-722.

CLRTAP, Mills G, Harmens H, Hayes F, Pleijel H, Büker P, et al. Mapping critical levels for vegetation. Revised Chapter 3 of the Manual on Methodologies and Criteria for Modelling and Mapping Critical Loads and Levels and Air Pollution Effects, Risks and Trends. Geneva, Switzerland: UNECE, 2017.

Cork SJ, Krockenberger AK. Methods and pitfalls of extracting condensed tannins and other phenolics from plants: insights from investigations on Eucalyptus leaves. Journal of chemical ecology 1991; 17: 123-134.

Cox P. Description of the TRIFFID dynamic global vegetation model. Hadley Centre Technical Note 2001; 24.

Duursma RA. Plantecophys - An R Package for Analysing and Modelling Leaf Gas Exchange Data. Plos One 2015; 10: 13.

Emberson LD, Ashmore MR, Cambridge HM, Simpson D, Tuovinen JP. Modelling stomatal ozone flux across Europe. Environmental Pollution 2000; 109: 403-413.

Eyring V, Bony S, Meehl GA, Senior CA, Stevens B, Stouffer RJ, et al. Overview of the Coupled Model Intercomparison Project Phase 6 (CMIP6) experimental design and organization. Geoscientific Model Development 2016; 9: 1937-1958.

Friedlingstein P, O'Sullivan M, Jones MW, Andrew RM, Hauck J, Olsen A, et al. Global Carbon Budget 2020. Earth Syst. Sci. Data 2020; 12: 3269-3340.

Griscom BW, Adams J, Ellis PW, Houghton RA, Lomax G, Miteva DA, et al. Natural climate solutions. Proceedings of the National Academy of Sciences of the United States of America 2017; 114: 11645-11650.

Harper AB, Cox PM, Friedlingstein P, Wiltshire AJ, Jones CD, Sitch S, et al. Improved representation of plant functional types and physiology in the Joint UK Land Environment Simulator (JULES v4.2) using plant trait information. Geosci. Model Dev. 2016; 9: 2415-2440.

Harper AB, Williams KE, McGuire PC, Rojas MCD, Hemming D, Verhoef A, et al. Improvement of modeling plant responses to low soil moisture in JULESvn4.9 and evaluation against flux tower measurements. Geoscientific Model Development 2021; 14: 3269-3294.

Harper AB, Wiltshire AJ, Cox PM, Friedlingstein P, Jones CD, Mercado LM, et al. Vegetation distribution and terrestrial carbon cycle in a carbon cycle configuration of JULES4.6 with new plant functional types. Geosci. Model Dev. 2018; 11: 2857-2873.

Harper KL, Lamarche C, Hartley A, Peylin P, Ottlé C, Bastrikov V, et al. A 29-year time series of annual 300 m resolution plant-functional-type maps for climate models. Earth Syst. Sci. Data 2023; 15: 1465-1499.

Hayes F, Harmens H, Sharps K, Radbourne A. Ozone dose-response relationships for tropical crops reveal potential threat to legume and wheat production, but not to millets. Scientific African 2020; 9: e00482.

Huntingford C, Atkin OK, Martinez-de la Torre A, Mercado LM, Heskel MA, Harper AB, et al. Implications of improved representations of plant respiration in a changing climate. Nature Communications 2017; 8: 11.

Huntingford C, Burke EJ, Jones CD, Jeffers ES, Wiltshire AJ. Nitrogen cycle impacts on CO2 fertilisation and climate forcing of land carbon stores. Environmental Research Letters 2022; 17: 12.

Huntingford C, Zelazowski P, Galbraith D, Mercado LM, Sitch S, Fisher R, et al. Simulated resilience of tropical rainforests to CO2-induced climate change. Nature Geoscience 2013; 6: 268-273.

Jarvis PG. Interpretation of variations in leaf-water potential and stomatal conductance found in canopies in field. Philosophical Transactions of the Royal Society of London Series B-Biological Sciences 1976; 273: 593-610.

Kreyling J, Schweiger AH, Bahn M, Ineson P, Migliavacca M, Morel-Journel T, et al. To replicate, or not to replicate – that is the question: how to tackle nonlinear responses in ecological experiments. Ecology Letters 2018; 21: 1629-1638.

Leung F, Sitch S, Tai APK, Wiltshire AJ, Gornall JL, Folberth GA, et al. CO2 fertilization of crops offsets yield losses due to future surface ozone damage and climate change. Environmental Research Letters 2022; 17.

Medlyn BE, Duursma RA, Eamus D, Ellsworth DS, Prentice IC, Barton CVM, et al. Reconciling the optimal and empirical approaches to modelling stomatal conductance. Global Change Biology 2011; 17: 2134-2144.

Mercado LM, Bellouin N, Sitch S, Boucher O, Huntingford C, Wild M, et al. Impact of changes in diffuse radiation on the global land carbon sink. Nature 2009; 458: 1014-1017.

Oliver RJ, Mercado LM, Clark DB, Huntingford C, Taylor CM, Vidale PL, et al. Improved representation of plant physiology in the JULES-vn5.6 land surface model: photosynthesis, stomatal conductance and thermal acclimation. Geosci. Model Dev. 2022; 15: 5567-5592.

Oliver RJ, Mercado LM, Sitch S, Simpson D, Medlyn BE, Lin YS, et al. Large but decreasing effect of ozone on the European carbon sink. Biogeosciences 2018; 15: 4245-4269.

Pleijel H, Danielsson H, Broberg MC. Benefits of the Phytotoxic Ozone Dose (POD) index in dose-response functions for wheat yield loss. Atmospheric Environment 2022; 268: 7.

Rap A, Scott CE, Reddington CL, Mercado L, Ellis RJ, Garraway S, et al. Enhanced global primary production by biogenic aerosol via diffuse radiation fertilization. Nature Geoscience 2018; 11: 640-644.

Ritmejerytė E, Boughton BA, Bayly MJ, Miller RE. Divergent responses of above-and below-ground chemical defence to nitrogen and phosphorus supply in waratahs (Telopea speciosissima). Functional Plant Biology 2019; 46: 1134-1145.

Sellar AA, Jones CG, Mulcahy JP, Tang Y, Yool A, Wiltshire A, et al. UKESM1: Description and Evaluation of the U.K. Earth System Model. Journal of Advances in Modeling Earth Systems 2019; 11: 4513-4558.

Singleton VL, Rossi JA. Colorimetry of total phenolics with phosphomolybdic-phosphotungstic acid reagents. American journal of Enology and Viticulture 1965; 16: 144-158.

Sitch S, Cox PM, Collins WJ, Huntingford C. Indirect radiative forcing of climate change through ozone effects on the land-carbon sink. Nature 2007; 448: 791-794.

University of East Anglia Climatic Research Unit; Harris I.C. CRU JRA v2.3: A forcings dataset of gridded land surface blend of Climatic Research Unit (CRU) and Japanese reanalysis (JRA) data; Jan.1901 - Dec.2021. NERC EDS Centre for Environmental Data Analysis, 2022.

Vancutsem C, Achard F, Pekel J-F, Vieilledent G, Carboni S, Simonetti D, et al. Long-term (1990–2019) monitoring of forest cover changes in the humid tropics. Science Advances 2021; 7: eabe1603.


Natural Environment Research Council, Award: NE/R001812/1