Future ground-level concentrations of phytotoxic ozone are projected to grow in the Northern Hemisphere, at a rate depending on emission scenarios. We explored the likely changes in net ecosystem production (NEP) due to the increasing concentration of tropospheric ozone by applying a Generalized Additive Mixed Model based on measurements of ozone concentration ([O3]) and stomatal ozone flux (FsO3), at a mountainous Norway spruce forest in the Czech Republic, Central Europe. A dataset covering the growing period (May-August 2009) was examined in this case study. A predictive model based on FsO3 was found to be marginally more accurate than a model using [O3] alone for prediction of the course of NEP when compared to NEP measured by the eddy covariance technique. Both higher [O3] and FsO3 were found to reduce NEP. NEP simulated at low, pre-industrial FsO3 (0.5 nmol m-2 s-1) was higher by 24.8% as compared to NEP assessed at current rates of FsO3 (8.32 nmol m-2 s-1). However, NEP simulated at high FsO3 (17 nmol m-2 s-1), likely in the future, was reduced by 14.1% as compared to NEP values at current FsO3. The interaction between environmental factors and stomatal conductance is discussed in this paper.
Tropospheric ozone (O3), a common phytotoxic secondary air pollutant, reduces growth and carbon sequestration potential of terrestrial vegetation (
Recent studies at different hierarchical levels have shown that O3 reduces photosynthetic carbon assimilation and changes carbon allocation in trees and forest ecosystems (
To quantify the detrimental dose of O3 on vegetation, several indices have been established: AOT40 (accumulated exposure over a threshold of 40 ppbv) and PODy (accumulated ozone flux above a given flux threshold
Contrary to these findings, growth decreased by 6.6% in Swiss forests during the period 1991-2011, particularly in European beech and Norway spruce (
In the present study, we attempted to quantify the effect of O3 on Net Ecosystem Production (NEP), defined as the difference between ecosystem-level photosynthetic uptake of CO2 and ecosystem respiratory loss of CO2. Our attention was focussed on a mountainous spruce forest in the Czech Republic. NEP values measured by eddy covariance were compared to those estimated by the Generalized Additive Mixed Model (GAMM). We tested the efficacy of [O3]- and stomatal ozone flux (FsO3)-based GAMM models to predict diurnal and seasonal changes in NEP. To determine O3 effects, NEP was calculated in a simulation model at low, pre-industrial [O3] (12 ppbv) and FsO3 (0.5 nmol m-2 s-1) as well as at high [O3] (80 ppbv) and FsO3 (17 nmol m-2 s-1) expected at the end of the century (
The forest stand is located at the Bílý Kríž experimental research site within the Beskydy Mountains in the north-east of the Czech Republic (49° 30′ N, 18° 32′ E; 875-908 m a.s.l.). The experimental research site forms part of several international research networks and infrastructures: CzeCOS (Czech Carbon Observation System), ICOS (Integrated Carbon Observation System), and AnaEE (Analysis and Experimentation on Ecosystems). The forest stand (99%
The area has a moderately cool (mean air temperature 6.8 °C) and humid (mean relative air humidity 84%) climate with high annual precipitation (mean annual precipitation 1318 mm; years 1998-2009). Due to an even distribution of precipitation over the year, values of soil volumetric water content remain high during the growing season ranging between 20 and 30% irrespective of sky conditions. The region is characterized by low concentrations of nitrogen oxides (below 10 ppbv) and high [O3], exceeding 80 ppbv during summer months (
The measurements were conducted from May 28 to September 30, 2009, during the daytime (06:00-18:00 GMT+1) to cover the periods characterized by well-developed turbulent mixing. A meteorological mast (36 m tall) situated within the studied stand was equipped with a set of meteorological sensors and an eddy covariance system. Air temperature (
The barometers PTB110® (Vaisala, Vantaa, Finland) and SPA 511 B5UB® (CRESSTO, Roznov pod Radhostem, Czech Republic) were used to measure air pressure. Global radiation (GR) was measured by a pyranometer CM6B® (Kipp & Zonen, Delft, Netherlands). The 2D ultrasonic anemometer 50.5 (Met One Instruments, Grants Pass, OR, USA) was used for the measurement of horizontal wind speed. All standard meteorological measurements were made on a vertical profile at heights of 2, 7, 10, 11.5, 12.25, 13, 13.5, 14, 15, 17, 21, and 28 m above the soil surface. The signals from all sensors were recorded every 30 s and stored as half-hourly averages using a data logger (Delta-T®, Burwell, Cambridgeshire, UK). In addition, precipitation was recorded by a precipitation gauge 386C (Met One Instruments, Grants Pass, OR, USA).
An eddy covariance system was used to measure the CO2 and water vapour fluxes between the forest stand and the atmosphere. The system consisted of a Gill R3® ultrasonic anemometer (Gill Instruments, Hampshire, UK) and an enclosed-path infrared gas analyser LI-7200® (LI-COR Biosciences, Lincoln, NE, USA) placed 18 m above the soil surface. The post-processing of high frequency data (20 Hz) was performed by EddyPro® software (LI-COR Biosciences, USA) according to recent recommendations (
[O3] was measured at 5, 15, and 25 m above the soil surface using slow-response O3 analysers O341M® (Environment S.A., Poissy, France). The signals from all O3 analysers were recorded as half-hourly averages.
Stomatal ozone flux (FsO3) to the Norway spruce forest was calculated according to Cieslik (
where
where
Stomatal resistance component
where
The model was applied using environmental variables measured at the experimental station, although soil moisture deficit was estimated as a function of precipitation and daily mean surface temperature according to the principles of water budget. For more details and the site-specific model parameterisation see
The Generalized Additive Mixed Model (GAMM) was implemented in the “mgcv” package of the R ver. 3.4.0 program (
All predictors were tested and fitted by a linear model, in the form of splines as smooth functions and as 1st degree linear interactions. Predictors were centred and, in the case of GR and RH, a square root transformation was used. Day of observation and autocorrelation pattern were tested as random factors. The best model was selected based on the AIC criteria and
The mean incoming daily sum of GR was 15.8 MJ m-2, mean
The results of GAMM model for NEP prediction including [O3] and FsO3 are shown in
Statistically significant (
NEP was predicted using [O3] (
Multi-factorial analysis (Fig. S2 in Supplementary material) revealed that high FsO3 (17 nmol m-2 s-1) has the most significant effect on NEP reduction in summer, while NEP decrease is driven more by GR than FsO3 in spring and autumn. Moreover, VPD-induced stomatal closure was a main factor limiting NEP and stomatal O3 flux in Norway spruce trees throughout the whole growing season. Therefore, a strong relationship between NEP reduction and stomatal conductance, not only the simple [O3] outside the leaves, can be suggested.
Taking to the account the whole dataset, the GAMM based calculated NEP was higher by up to 12.4% at low, preindustrial ozone concentration (12 ppbv) as compared to the measured NEP at the present concentration of 42.9 ± 11.2 ppbv (mean ± SD -
In addition, we present estimated changes in NEP under the conditions of low/high [O3] and FsO3 during individual months of the growing season (May-September -
Projections of O3-induced damage are uncertain due to numerous scenarios dealing with different future [O3] or FsO3. Ozone in the Northern Hemisphere is projected to increase further by 20-25% between 2015 and 2050, and by 40-60% by 2100, if current ozone-precursors (CO2, volatile organic compounds, NOx) emission trends continue (
Numerous studies have shown that elevated [O3] generally results in reduced photosynthesis, chlorophyll content and whole-plant growth, decreased stomatal conductance, an altered antioxidant system, accelerated senescence and changes the plant metabolism, although the extent of the effects varies by species, length of exposure, [O3] and/or co-occurrence of other stress factors (reviewed in
We predicted higher NEP (by 24.6% on average) at low FsO3 of 0.5 nmol m-2 s-1 as compared to NEP at current FsO3 of 8.32 ± 3.66 nmol m-2 s-1 with the highest absolute reduction occurring in June and July (
Norway spruce, one of the most common and important timber trees in Europe, was found to be an O3 sensitive tree species, showing substantial reductions in photosynthetic carbon uptake under elevated [O3] (
It should be noted, however, that there is a contrasting evidence of long-term O3 effects detected after several growing seasons. Since several research groups reported no negative effect of elevated [O3] on carbon assimilation in forest ecosystems (
We have found that stomatal ozone flux model tracks diurnal changes in NEP more precisely that ozone concentration model does. NEP simulated at low, pre-industrial FsO3 (0.5 nmol m-2 s-1) was higher by 24.8% as compared to NEP assessed at current FsO3 (8.32 nmol m-2 s-1). Further increase in FsO3 (up to 17 nmol m-2 s-1) may lead to subsequent reduction in NEP by 14.1% in average as compared to current NEP values, but this reduction may change during the growing season. These results are in agreements with observations in other European coniferous forests. Relatively high site-specific variability in [O3] effect of photosynthetic carbon uptake is likely caused by species-specific sensitivity of stomata to environmental drivers. Also a co-occurrence of other stress factors, particularly extreme temperatures and drought, leading to the changes in stomatal conductance and stomatal ozone flux, may influence a final response of trees to [O3].
This work was supported by the Ministry of Education, Youth and Sports of the Czech Republic within the project “SustES - Adaptation strategies for sustainable ecosystem services and food security under adverse environmental conditions” (CZ.02. 1.01/0.0/0.0/16_019/0000797).
Diurnal courses of global radiation (A), air temperature (B), vapour pressure deficit (VPD, C), and ozone concentration (D) during the representative days.
Relative occurrence of stomatal ozone flux (FsO3) during the sunlight hours modelled for the period May 28-September 30, 2009.
Modelled relationships between net ecosystem production (NEP) and ozone concentration (A) and between NEP and stomatal ozone flux (B) for four representative days (full line). Models were applied with fixed predictors (see
Predicted diurnal courses of net ecosystem production (NEP) using Generalized Additive Mixed Model with ozone concentration (O3, A,B,C,D) and stomatal ozone flux (FsO3, E,F,G,H). Three levels of O3 and FsO3 were used to predict NEP: Blue line - low O3 (10 ppbv) and FsO3 (0.5 nmol m-2 s-1), red line - high O3 (80 ppbv) and FsO3 (17 nmol m-2 s-1), and black line - actual, measured O3 and FsO3. Green line represents NEP measured by an eddy covariance system. Shaded area represents prediction bounds at 95% confidence level.
Meteorological variables of the period May 28-September 30, 2009. [O3]: ozone concentration; (FsO3): stomatal ozone flux; (VPD): vapour pressure deficit; (GR): global radiation; (RH): relative air humidity; (Tair): air temperature; (Tr): transpiration; (NEP): net ecosystem production: (SD): standard deviation.
Variable | Unit | Mean | SD | Range |
---|---|---|---|---|
NEP | µmol m-2 s-1 | 12.1 | 9.2 | -7.9 - 31.7 |
[O3] | ppbv | 42.9 | 11.2 | 12.0 - 79.5 |
FsO3 | nmol m-2 s-1 | 8.3 | 3.6 | 0.072 - 17.6 |
GR | W m-2 | 342.3 | 247.4 | 0.86 - 974 |
Tair | °C | 17.7 | 6.1 | 1.3 - 33.8 |
RH | % | 75.2 | 17.3 | 37.6 - 99.9 |
Tr | mm h-1 | 0.0878 | 0.0754 | 0 - 0.37 |
VPD | kPa | 0.624 | 0.585 | 0.0006 - 2.89 |
Outputs of Generalized Additive Mixed Model (GAMM) predicting mixed effects of [O3] and other selected micrometeorological variables (predictors) on net ecosystem production (NEP). Linear predictors and linear interaction statistics are shown as
Predictor/Interaction | Goodness of fit | Statistics | |||
---|---|---|---|---|---|
Parametric estimate | Edf |
|
|||
[O3] | -0.042609 | - | 4.655 | - | 3.55e-06*** |
VPD | 1.686854 | - | 2.226 | - | 0.026154* |
[O3] × GR | -0.002573 | - | -3.390 | - | 0.000717*** |
GR × RH | -0.100322 | - | -6.310 | - | 3.73e-10*** |
GR × |
0.013316 | - | 6.919 | - | 6.84e-12*** |
GR × Tr | -0.476992 | - | -3.499 | - | 0.000481*** |
VPD × GR | -0.280066 | - | -8.314 | - | <2e-16*** |
s(GR) | - | 2.987 | - | 6835.670 | <2e-16*** |
s(RH) | - | 2.878 | - | 22.424 | 6.5e-13*** |
s( |
- | 2.4577 | - | 6.775 | 0.000807*** |
s(Tr) | - | 2.384 | - | 5.168 | 0.003775** |
Outputs of Generalized Additive Mixed Model (GAMM) predicting mixed effects of stomatal ozone flux (FsO3) and other selected micrometeorological variables (predictors) on net ecosystem production (NEP). Linear predictors and linear interaction statistics are shown as
Predictor/Interaction | Goodness of fit | Statistics | |||
---|---|---|---|---|---|
Parametricestimate | Edf |
|
|||
VPD | 2.301725 | - | 2.850 | - | 0.00443** |
GR × FsO3 | -0.016042 | - | -4.577 | - | 5.14e-06*** |
GR × RH | -0.142519 | - | -7.082 | - | 2.23e-12*** |
GR × |
0.014084 | - | 7.378 | - | 2.73e-13*** |
VPD × GR | -0.388303 | - | -9.721 | - | <2e-16*** |
s(GR) | - | 2.973 | - | 6314.858 | <2e-16*** |
s(RH) | - | 2.865 | - | 18.079 | 5.99e-08*** |
s( |
- | 2.824 | - | 11.175 | 8.78e-07*** |
s(FsO3) | - | 2.933 | - | 22.547 | 7.77e-13*** |
s(Tr) | - | 2.701 | - | 7.112 | 0.000131*** |
Mean values of ambient (Amb) ozone concentration ([O3]; ppbv) and stomatal ozone flux (FsO3; nmol m-2 s-1), and Net ecosystem production (NEP; µmol m-2 s-1) predicted by a Generalized Additive Mixed Model for the conditions of actual (measured), high, and low [O3] and FsO3 during the individual months of the growing season and the whole dataset covering May 28-September 30, 2009. Δ represents percentage difference of predicted NEP to mean NEP measured by an eddy covariance system. (a) values of Δ represent percentage decrease in NEP induced by high O3 (80 ppbv) and/or FsO3 (17 nmol m-2 s-1); (b): values of Δ represent percentage increase in NEP at low O3 (12 ppbv) and/or FsO3 (0.5 nmol m-2 s-1).
Months | Conditions | Amb [O3] | Low [O3] | High [O3] | Amb FsO3 | Low FsO3 | High FsO3 |
---|---|---|---|---|---|---|---|
May | [O3]/FsO3 | 36.5 | 12 | 80 | 5.5 | 0.5 | 17 |
NEP | 7.9 | 9.6 | 7.2 | 7.9 | 10.8 | 7.2 | |
Δ | - | 21.5 b | 8.9 a | - | 36.7 b | 8.9 a | |
June | [O3]/FsO3 | 42 | 12 | 80 | 9.1 | 0.5 | 17 |
NEP | 10.2 | 12.5 | 9.7 | 10.2 | 14.1 | 7.7 | |
Δ | - | 22.5 b | 4.9 a | - | 38.2 b | 24.5 a | |
July | [O3]/FsO3 | 46.2 | 12 | 80 | 8.98 | 0.5 | 17 |
NEP | 15.3 | 17.2 | 13.8 | 15.3 | 18.7 | 13.5 | |
Δ | - | 12.4 b | 9.8 a | - | 22.2 b | 11.8 a | |
Aug | [O3]/FsO3 | 45 | 12 | 80 | 8.5 | 0.5 | 17 |
NEP | 12.6 | 13.9 | 11 | 12.6 | 15.4 | 10.9 | |
Δ | - | 10.3 b | 12.7 a | - | 22.2 b | 13.5 a | |
Sept | [O3]/FsO3 | 39.1 | 12 | 80 | 7.2 | 0.5 | 17 |
NEP | 10.4 | 10.8 | 8.3 | 10.4 | 12.3 | 8.5 | |
Δ | - | 3.8 | 20.2 | - | 18.3 | 18.3 | |
Whole dataset | [O3]/FsO3 | 42.9 | 12 | 80 | 8.3 | 0.5 | 17 |
NEP | 12.1 | 13.6 | 10.7 | 12.1 | 15.1 | 10.4 | |
Δ | - | 12.4 | 11.6 | - | 24.8 | 14.1 |
Fig. S1 - Diurnal courses of modelled stomatal ozone flux (FsO3) in Norway spruce forest during the investigated period of May 28 - September 30, 2009.
Fig. S2 - Principal component analysis revealing dependencies of environmental variables among each other.