Air Pollution and Climate Drive Annual Growth in Ponderosa Pine Trees in Southern California

: The ponderosa pine ( Pinus ponderosa , Douglas ex C. Lawson) is a climate-sensitive tree species dominant in the mixed conifer stands of the San Bernardino Mountains of California. How-ever, the close proximity to the city of Los Angeles has resulted in extremely high levels of air pollution. Nitrogen (N) deposition, resulting from nitrous oxides emitted from incomplete combustion of fossil fuels, has been recorded in this region since the 1980s. The impact of this N deposition on ponderosa pine growth is complex and often obscured by other stressors including climate, bark beetle attack, and tropospheric ozone pollution. Here I use a 160-year-long (1855–2015) ponderosa pine tree ring chronology to examine the annual response of tree growth to both N deposition and climate in this region. The chronology is generated from 34 tree cores taken near Crestline, CA. A stepwise multiple regression between the tree ring chronology and various climate and air pollution stressors indicates that drought conditions at the end of the rainy season (March) and NO 2 pollution during the water year (pOct-Sep) exhibit primary controls on growth (r 2 -adj = 0.65, p < 0.001). The direct correlation between NO 2 and tree growth suggests that N deposition has a positive impact on ponderosa pine bole growth in this region. However, it is important to note that ozone, a known stressor to ponderosa pine trees, and NO 2 are also highly correlated (r = 0.84, p < 0.05). Chronic exposure to both ozone and nitrogen dioxide may, therefore, have unexpected impacts on tree sensitivity to climate and other stressors in a warming world.


Introduction
The ponderosa pine (Pinus ponderosa, Douglas ex C. Lawson) has long been used as a proxy for climate information in the American Southwest [1][2][3]. Tree ring growth in this seminal species has been found to be sensitive to changes in precipitation [4][5][6], water availability [7], the El Niño Southern Oscillation [8,9], and drought [10]. However, most of these studies have been conducted in the Sierra Nevada, where ponderosa pine trees dominate the interior forest of that region [11]. Fewer chronologies have been used to understand the climate sensitivity of ponderosa pine trees growing in the San Bernardino Mountains of Southern California, where the species is dominant at higher elevations (1524-2438 m) in mixed conifer stands. The region is located in a Mediterranean climate [12] where the wind flow pattern is driven by local topography, island configuration, proximity to the Pacific Ocean, and the semipermanent high pressure of the Northeastern Pacific [13][14][15][16][17][18]. These factors generate a strong sea breeze of westerly onshore flow that is present throughout the year, denominated by the Catalina eddies that grow in MAM and peak during the summer season (JJA) [19]. The climate of the region is distinct from the Sierra Nevada to the North. While mean annual temperature is similar between these two mountain regions (12)(13)(14) • C), the high Sierras (>1500 m) experience much greater rainfall (1778-2032 mm) than the Transverse Ranges (824 mm). Despite this difference, of the 32 ponderosa pine chronologies in the International Tree Ring Data Base (ITRDB, February 2021), only one exists from trees growing in the Transverse Ranges [20].

Tree Ring Chronology
Increment cores were extracted from ponderosa pine trees located near Camp Seely in Crestline, CA (lat: 34.25 • N, lon: 117.29 • W) at an elevation of 1486 m above sea level ( Figure 1). This location was chosen because it is on the west-facing slope of the San Bernardino Mountains and thus maximally exposed to the air pollution blowing eastward from Los Angeles, California. The forest is a mix-conifer forest characterized by Jeffrey (Pinus jeffreyi) and ponderosa pine (Pinus ponderosa) trees, sugar pine (Pinus lambertiana), incense cedar (Calocedrus decurrens), California black oak (Quercus kelloggii), and white fir (Abies concolor) [25]. However, in warmer dry environments like that of the San Bernardino Mountains, ponderosa pine often dominates the forest [37]. Thirty-four cores were extracted from 20 ponderosa pine trees near Camp Seely. All cores were taken at mean breast height (1.3 m above the land surface) in accordance with standard practice [38]. In total, 20 trees were sampled (2 cores per tree where possible) in accordance with standard dendrochronological techniques (Speer, 2010). In the case of 6 trees, only one core per tree was used; however, the expressed population signal (EPS) of the chronology indicated sufficient sampling depth and shared variability among trees was achieved to obtain a stand-level signal [39]. After extraction in the field, the cores were mounted and sanded to expose the transverse section and accentuate ring boundaries. Each core was placed on a Velmex tree ring measuring system (Velmex Inc. Bloomfield, NY, USA) and ring boundaries were cross-dated under a microscope and then measured to 0.001 mm. Once the ring widths for each core (radius) were compiled, the quality control program COFECHA [40] was used to verify the initial crossdating and detect regions of misalignment or low correlation between individual radii. The computer program ARSTAN [41,42] was used to compile individual ring-width series into a standardized chronology. This program removes nonclimatic signals present in tree ring series including geometric and ecological growth trends resulting from intra-tree competition and local dynamics within the canopy. A cubic smoothing spline of 50% frequency response of 67% of the length of each series was used to standardize the series and an autoregressive model to remove autocorrelation effects was employed [43]. Statistics were computed on the chronology including interseries correlation, sensitivity, signal-to-noise ratio (SNR), autocorrelation, and the expressed population signal (EPS), which calculates how well a chronology represents the theoretical population [39]. widths for each core (radius) were compiled, the quality control program COFECHA [40] was used to verify the initial crossdating and detect regions of misalignment or low correlation between individual radii. The computer program ARSTAN [41,42] was used to compile individual ring-width series into a standardized chronology. This program removes non-climatic signals present in tree ring series including geometric and ecological growth trends resulting from intra-tree competition and local dynamics within the canopy. A cubic smoothing spline of 50% frequency response of 67% of the length of each series was used to standardize the series and an autoregressive model to remove autocorrelation effects was employed [43]. Statistics were computed on the chronology including interseries correlation, sensitivity, signal-to-noise ratio (SNR), autocorrelation, and the expressed population signal (EPS), which calculates how well a chronology represents the theoretical population [39]. Map showing the location of the Crestline ponderosa pine chronology used in this study (red star). Green circles represent the locations of existing ITRDB ponderosa pine chronologies, available at https://www.ncdc.noaa.gov/data-access/paleoclimatology-data. Black arrow denotes the dominant wind direction over the study site [44,45] and yellow circles locate the sites of NO2 (southern circle), and ozone (northern circle) air pollution measurements [46].

Climate Data
The site is characterized by a Mediterranean climate with wet winters and dry hot summers. In the San Bernardino mountains, the annual rainfall (824 mm) is less than half of that received in the Sierra Nevada (1778-2032 mm) and particularly deficient in the summer, where June-July-August-September total rainfall averages less than 25 mm. Thus, the trees in the study region experience seasonally limited water availability. The maximum rainfall occurs during the months of December through March while peak temperature occurs in July and August (Figure 2a). Total annual rainfall is 824 ± 1.6 mm and the average annual temperature is 12 ± 0.17 °C. Average maximum monthly temperature and total monthly precipitation data were obtained from the parameter-elevation regressions on independent slopes model (PRISM: https://prism.oregonstate.edu, [47]). The spatial resolution of this data is a 4-km 2 grid-cell centered over the study site (lat: 34.25° N, lon: 117.29° W). PRISM temperature and precipitation data are available from 1895-2015 Map showing the location of the Crestline ponderosa pine chronology used in this study (red star). Green circles represent the locations of existing ITRDB ponderosa pine chronologies, available at https://www.ncdc.noaa.gov/data-access/paleoclimatology-data. Black arrow denotes the dominant wind direction over the study site [44,45] and yellow circles locate the sites of NO 2 (southern circle), and ozone (northern circle) air pollution measurements [46].

Climate Data
The site is characterized by a Mediterranean climate with wet winters and dry hot summers. In the San Bernardino mountains, the annual rainfall (824 mm) is less than half of that received in the Sierra Nevada (1778-2032 mm) and particularly deficient in the summer, where June-July-August-September total rainfall averages less than 25 mm. Thus, the trees in the study region experience seasonally limited water availability. The maximum rainfall occurs during the months of December through March while peak temperature occurs in July and August (Figure 2a). Total annual rainfall is 824 ± 1.6 mm and the average annual temperature is 12 ± 0.17 • C. Average maximum monthly temperature and total monthly precipitation data were obtained from the parameter-elevation regressions on independent slopes model (PRISM: https://prism.oregonstate.edu, [47]). The spatial resolution of this data is a 4-km 2 grid-cell centered over the study site (lat: 34.25 • N, lon: 117.29 • W). PRISM temperature and precipitation data are available from 1895-2015 and were resolved into monthly, seasonal, and clusters of months to compare with the tree ring chronology. Several studies have assessed the reliability of PRISM data to accurately reproduce meteorological station data in southwestern United States [47] and specifically in mountainous regions such as the Transverse Ranges of California [48]. The Standardized Precipitation Evapotranspiration Index (SPEI, [49,50]) was also compared with the tree ring chronology. The SPEI includes both precipitation and evapotranspiration to capture the effect that temperature has on water demand. Data are available from 1901-2015 and calculated monthly at the site location (lat: 34.25 • N, lon: 117.29 • W) [51]. The Palmer Drought Severity Index (PDSI), a measure of the wetness and dryness of a region based on precipitation, moisture supply, runoff and evaporation, was also compared with the tree ring chronology. Data are spatially resolved at 2.5 • × 2.5 • (lat: 34.25 • N, lon: 117.29 • N) and are used from 1933-2015 in this study to overlap with the length of the tree ring chronology [52]. and were resolved into monthly, seasonal, and clusters of months to compare with the tree ring chronology. Several studies have assessed the reliability of PRISM data to accurately reproduce meteorological station data in southwestern United States [47] and specifically in mountainous regions such as the Transverse Ranges of California [48]. The Standardized Precipitation Evapotranspiration Index (SPEI, [49,50]) was also compared with the tree ring chronology. The SPEI includes both precipitation and evapotranspiration to capture the effect that temperature has on water demand. Data are available from 1901-2015 and calculated monthly at the site location (lat: 34.25° N, lon: 117.29° W) [51]. The Palmer Drought Severity Index (PDSI), a measure of the wetness and dryness of a region based on precipitation, moisture supply, runoff and evaporation, was also compared with the tree ring chronology. Data are spatially resolved at 2.5° × 2.5° (lat: 34.25° N, lon: 117.29° N) and are used from 1933-2015 in this study to overlap with the length of the tree ring chronology [52].

Air Pollution Data
In addition to the westward facing position of the study site that directly receives air pollution from Los Angeles, this site was also chosen because of the historical measurements of air pollution (ozone, O3) that have been recorded at Crestline since 1963 [53]. Hourly measurements of ozone have been made in this region since 1979 (lat: 33.99° N, lon: 117.41° W) and are available through the EPA's Air Quality System database [46]. Measurements between 1963 and 1978 are reconstructed from episodic regional measurements from multiple sites, adjusting for differences in elevation, night-time concentration, and peak exposure [54]. Ozone data were resolved into monthly and seasonal composites over the time period and into clusters of months of peak maximum and minimum pollution values. Hourly measurements of nitrogen dioxide are also available through the EPA Air Quality System database [46] and were available over the time period 1980-2020 from the EPA station in Rubidoux, CA (lat: 34.00° N, lon: 117.24° W), although wintertime data

Air Pollution Data
In addition to the westward facing position of the study site that directly receives air pollution from Los Angeles, this site was also chosen because of the historical measurements of air pollution (ozone, O 3 ) that have been recorded at Crestline since 1963 [53]. Hourly measurements of ozone have been made in this region since 1979 (lat: 33.99 • N, lon: 117.41 • W) and are available through the EPA's Air Quality System database [46]. Measurements between 1963 and 1978 are reconstructed from episodic regional measurements from multiple sites, adjusting for differences in elevation, night-time concentration, and peak exposure [54]. Ozone data were resolved into monthly and seasonal composites over the time period and into clusters of months of peak maximum and minimum pollution values. Hourly measurements of nitrogen dioxide are also available through the EPA Air Quality System database [46] and were available over the time period 1980-2020 from the EPA station in Rubidoux, CA (lat: 34.

Climate Response Analysis
The relationship between climate variables and air pollution was examined by modeling the ability of 28 distinct combinations of monthly, seasonal, annual, and clustered grouped variables to predict the ring width chronology. Variables included total monthly precipitation, average monthly temperature, monthly SPEI, PDSI, average daily ozone (O 3 ), and average daily nitrogen dioxide (NO 2 ) from the previous October through the current year's December. Wet season and dry season values were calculated along with the water year and other clustered months to examine relationships between tree growth and these variables. Multiple linear regressions were computed for all 28 time periods and t and p-values were calculated to determine statistical significance. A step-wise regression was employed to examine the relative importance of each variable to the overall tree ring signal. Variables with a variance inflation factor (VIF) greater than 3 were considered colinear and variables with a lower p-value were removed from the regression. A series of 5 separate step-wise regression models were compared.

Results
The Crestline ponderosa pine ring-width series crossdated well and showed good dendroclimatic potential, with the EPS of the site above the threshold limit of 0.85 for the standard chronology ( Table 1). The running EPS shows the signal is robust from 1878 onward ( Figure 3). The Signal to Noise (SNR) ratio is relatively high and the sensitivity is moderate (Table 1). When compared with precipitation, the tree ring chronology shows a statistically significant relationship with rainfall during the late part of the rainy season (February and March) and the month of August ( Figure 4A, Table 2). However, the highest correlations were found between the tree ring chronology and the water year (pOct-Sep, r = 0.39, p < 0.05) and the previous wet season (pOct-Apr, r = 0.38, p < 0.05). Negative correlations were found between temperature and ponderosa pine growth, with the most significant correlation occurring between the tree chronology and temperature during the beginning of the previous rainy season (pOct, r = −0.27, p < 0.05) and the current rainy season (Oct, r = −0.20, p < 0.05) ( Figure 4B, Table 2). A negative correlation was also found to be significant during the month of March of the current rainy season (r = −0.16, p < 0.05). Correlation between the SPEI and the tree chronology revealed a pattern similar to both precipitation and temperature, where statistically significant relationships were found during the start of the previous (pOct) and current rainy seasons (Oct), February, March, and the water year (pOct-Sep) ( Figure 4C, Table 2). The highest correlation, however, was found between SPEI during the wet season (pDec-Mar, r = 0.35, p < 0.05). The chronology showed statistically significant (p < 0.05) correlation with PDSI during all monthly, seasonal, and clustered time periods examined, with the strongest correlations occurring during the month of March (r = 0.46, p < 0.05) and during the rainy season (pOct-Apr, r = 0.44, p < 0.05) ( Figure 4D, Table 2). The highest climate forcing correlations overall were found between the tree chronology and PDSI, although the next strongest correlations were found between the tree chronology and precipitation (Table 2, Supplementary Figure S1).     Monthly average NO2 concentrations were found to be well correlated with tree growth during all months examined, including seasonal clusters (wet season, dry season), and the water year ( Figure 4E). The highest correlation is found between the chronology and May NO2 concentrations (r = 0.71, p < 0.05) and the water year (pOct-Sep, r = 0.66, p  Monthly average NO 2 concentrations were found to be well correlated with tree growth during all months examined, including seasonal clusters (wet season, dry season), and the water year ( Figure 4E). The highest correlation is found between the chronology and May NO 2 concentrations (r = 0.71, p < 0.05) and the water year (pOct-Sep, r = 0.66, p < 0.05). Ozone is also positively correlated with tree growth, particularly when ozone levels peak during the dry season (Jun-Aug, r = 0.67, p < 0.05); however, this is to be expected given that NO 2 and O 3 are strongly correlated (r = 0.84, p < 0.05) ( Figure 4F).
A stepwise multiple linear regression showed that the most important driver of ringwidth increment is NO 2 during the water year (pOct-Sep), which is also observed in the individually computed correlations (Table 3). PDSI during March is also a significant driver of growth and, with NO 2 (pOct-Sep), explains a significant component of tree ring growth variability (r 2 -adj = 0.65, p < 0.001). Combinations of precipitation (pOct-Sep), temperature (pOct), NO 2 (pOct-Sep), and PDSI (March) can be used to represent tree ring variability in several step-wise regression models ( Figure 5). Precipitation is found to be the primary driver of ring-width increment size when only precipitation and temperature are used ( Figure 5). However, when NO 2 is introduced, the model results in a far more significant adjusted r 2 value ( Figure 5). PDSI and precipitation as well as PDSI and temperature were found to covary (VIF > 3) and, thus, are not included together in any stepwise regression model. Similarly, ozone and nitrogen dioxide were found to covary (VIF = 3.4) and are not included together in any model for the same reason. The models that yielded the highest adjusted r 2 values and, thus, best represented the tree ring chronology were PDSI and NO 2 (r 2 -adj = 0.65, p < 0.001) and temperature, precipitation, and NO 2 (r 2 -adj = 0.60, p < 0.05). Using the PDSI (Mar) and NO 2 (pOct-Sep) to reconstruct the tree ring chronology yields the strongest correlation (r 2 = 0.671) ( Figure 6). The individual climate response analyses and the stepwise multiple regressions indicate that drought conditions (PDSI) at the end of the rainy season (March) and NO 2 pollution during the growing year (pOct-Sep) exhibit primary controls on growth for ponderosa pine growing in this region. were PDSI and NO2 (r 2 -adj = 0.65, p < 0.001) and temperature, precipitation, and NO2 (r 2adj = 0.60, p < 0.05). Using the PDSI (Mar) and NO2 (pOct-Sep) to reconstruct the tree ring chronology yields the strongest correlation (r 2 = 0.671) ( Figure 6). The individual climate response analyses and the stepwise multiple regressions indicate that drought conditions (PDSI) at the end of the rainy season (March) and NO2 pollution during the growing year (pOct-Sep) exhibit primary controls on growth for ponderosa pine growing in this region.  The gray bars indicate all variables in the stepwise regression were significant at the 95% level (p < 0.05), the black bar indicates all variables were significant at the 99% level (p < 0.01).

Climate and Tree Increment Growth
Variations in ring width of the ponderosa pine trees from the Crestline region are driven by precipitation and temperature. In general, the climate sensitivity of this chro-

Climate and Tree Increment Growth
Variations in ring width of the ponderosa pine trees from the Crestline region are driven by precipitation and temperature. In general, the climate sensitivity of this chronology to temperature, precipitation, and drought indices is consistent with that found in other studies of Ponderosa pine trees in the southwestern United States [1,[55][56][57]. Rainfall during the growing season has been found to drive growth in ponderosa pine trees growing in the Rocky Mountains [56] and in the Sierra Nevada [58]. McCullough et al. [1] found that wet season precipitation is one of the main drivers of growth across both Pacific interior and coastal forests of ponderosa pine, although trees in the southwest and coast ranges were found to be less sensitive to climate than those located more inland. Their findings are supported by this study, where sensitivity to water year precipitation is statistically significant, but lower than expected (Table 1). Temperature has some control over the tree ring signal, where higher temperatures during the onset of the rainy season (pOct) are associated with lower amounts of basal increment growth ( Figure 4B). Combining temperature and precipitation into a moisture availability index (SPEI) yields a statistically significant relationship with tree growth as well, suggesting that both temperature and precipitation during the wet season determine the growth of the tree bole ( Figure 5). However, the strongest climate determinant of tree growth appears to be moisture availability as represented in the PDSI, which incorporates water demand, moisture availability, temperature, and precipitation ( Table 2). As precipitation decreases and temperature increases, this generates higher water demand, which enhances drought severity and reduces increment growth in ponderosa pine trees in this region. While the stepwise multiple regression reveals that the tree ring chronology can be explained by water year precipitation and rainy season temperature, a much greater portion of the variability is explained when an integrated drought index (PDSI) is used, owing to the interplay between water demand and drought and the ability of this index to retain the memory of previous (9-19 months) climate conditions ( Figure 5) [59]. A more integrated representation of climate conditions yields a higher correlation with tree growth given that tree radial increments are annual.

NO 2 , Ozone, and Tree Growth
Previous studies have discovered a connection between bole growth and increasing N deposition, even in the presence of elevated ozone, which is phytotoxic [60]. In 2020, Fenn et al. [32] measured changes in aboveground biomass in the San Bernardino National Forest over a 10-year period using Forest Inventory and Analysis (FIA) plots. Their findings showed increases in carbon increment across a range of species in response to N deposition. Notably, regions with high levels of both ozone and N deposition still showed a positive increase in carbon increment, needle growth, and aboveground woody biomass [32]. The findings of this study corroborate this assessment, showing that NO 2 concentrations and tree ring width covary significantly. Grulke et al. [61] explain that a nitrogen-driven increase in aboveground biomass may occur at the expense of fine and coarse root growth. This effect is not linear, however, given that moderate exposure to ozone (20-30 ppb) and N deposition (15-25 kg ha −1 yr −1 ) resulted in diminished carbon increment (CI) for ponderosa pine [32]. This may explain why prior studies [62,63] observed diminished growth in response to elevated ozone levels in this region. However, the increased growth in response to elevated NO 2 and ozone in this study may be explained by the extremely high levels of pollution in this region (Figure 4) [64]. The average ozone levels for this study site are much higher (annual average = 50.3 ppb) and, therefore, result in an increase in the allocation of aboveground biomass as observed in the increased ring widths in the trees growing at Crestline. Increased CI growth in response to ozone and N deposition is expected when the ozone levels are sufficiently high (>30 ppb) as confirmed by other studies [32,60].
Nitrogen dioxide emissions throughout the growing season strongly correlate with tree growth and explain the majority (r 2 = 0.50, p < 0.05) of the tree ring signals ( Figure 5).
Because N deposition enhances growth in this region, it may be observed to counter the negative impacts of ozone exposure, by increasing growth in the bole of the tree and enhancing N availability for photosynthetic pigment repair [65]. However, Grulke [26] observed an increase in stomatal conductance and ozone uptake in response to increased N availability, suggesting that trees may experience greater drought stress amid high levels of pollutants, thus enhancing their sensitivity to climate. One way to test this would be to contrast the sensitivity of tree growth to climate before and after the influence of air pollution in the study region, because it is not possible to disentangle the relative contributions of pollution and climate over the common period . However, this requires a longer time series of pollution data over the length of the tree chronology (1855-2015), which is not available. Nevertheless, the variables that best explain the variability in tree ring growth include both moisture availability (PDSI) and nitrogen dioxide, which seem to support this idea ( Figure 5).
One possible hypothesis for the increased growth observed during high nitrogen dioxide years relates to the interplay between precipitation and N deposition, given that arid conditions can reduce N retention and uptake by needles [66,67]. Because the canopy can retain N via dry deposition [68], this stored N may be flushed to the rooting zone during high rainfall events or high rainfall months. One possible explanation, therefore, may be that rainfall events at the end of the rainy season (March-May) flush N to the rooting zone where it is available for uptake. It is also possible that the milder winter conditions of the Transverse Ranges and the mid-elevation of the study site facilitate the biological availability and assimilation of N during spring rainfall events [32].

Colinearity of Ozone and Nitrogen Dioxide
Although a direct link between ozone exposure alone and increased radial increment growth has not been found previously, a correlation between peak ozone emission (July, 79 ppb) and tree growth is observed in this study ( Figure 4F). This is likely related to the co-occurrence of N deposition via nitrous oxides and ozone, as observed in this region (r 2 = 0.71, p < 0.05). As previously mentioned, high levels of ozone and N deposition in combination have been found to stimulate bole growth in ponderosa pine trees in California and this study corroborates this finding [32]. However, because ozone is phytotoxic and has been shown to result in premature needle loss [27], diminished crown capacity, and reduced stomatal conductance [61], using this variable to explain variability in the radial increment of ponderosa pine growth may not be entirely appropriate. Additionally, the covariability of ozone and nitrogen dioxide generated a VIF greater than 3 (VIF = 3.4), which precluded its use in the stepwise multiple regression. Nevertheless, the diffuse impacts of ozone and N deposition can be observed to have significant control over radial growth variability in ponderosa pine trees in a heavily polluted region [69]. Nitrogen dioxide emissions, when combined with moisture availability (PDSI) explain 67.1% (p < 0.01) of the variability in incremental tree growth in the ponderosa pine trees growing at Crestline, CA ( Figure 5). One important question that emerges from this work concerns how ponderosa pine trees in the San Bernardino forest will respond to increasingly high levels of nitrogen, ozone, and increasing temperature in the future, particularly in contrast with those growing in the Sierra Nevada.

Conclusions
Ponderosa pine tree rings at the Crestline site cross-dated well and showed a statistically significant correlation with water year and rainy season precipitation and rainy season temperature. The strongest climatic relationships are found between the tree ring chronology and moisture availability during the water year (PDSI).
Year to year variability in tree growth can be explained by variations in nitrogen dioxide and ozone concentration in this heavily polluted region, particularly during the water year (NO 2 ) and the dry season (ozone). This suggests that trees in this region show increased increment (bole) growth in response to high (>30 ppb) levels of ozone pollution and N deposition, likely associated