Ozone exposure-response relationships parametrized for sixteen tree species with varying sensitivity in the United States

Abstract

It is well known that exposure to ambient O₃ can decrease growth in many tree species in the United States (US). Our study reports experimental data from outdoor open-top chamber (OTC) studies that quantify total biomass response changes for seedlings of 16 species native to western and eastern North America, which were exposed to several levels of elevated O₃ for one or more years. The primary objective of this study is to establish a reference set of parameters for these seedling exposure-response relationships using a 3-month (92 day) 12-hr W126 O₃ metric used by US Environmental Protection Agency and other agencies to assess risk to trees from O₃ exposure. We classified the 16 species according to their sensitivity, based on the biomass loss response functions to protect from a 5% biomass loss. The three-month 12-h W126 estimated to result in a 5% biomass loss was 2.5-9.2 ppm-h for sensitive species, 20.8-25.2 ppm-h for intermediate species, and > 28.7 ppm-h for insensitive species. The most sensitive tree species include black cherry, ponderosa pine, quaking aspen, red alder, American sycamore, tulip poplar and winged sumac. These species are ecologically important and widespread across US. The effects of O₃ on whole-plant biomass depended on exposure duration and dynamics and on the number of successive years of exposure. These species-specific exposure-response relationships will allow US agencies and other groups to better estimate biomass losses based on ozone exposures in North America and can be used in risk assessment and scenario analyses.

Introduction

Tropospheric ozone (O₃) is the ambient air pollutant most detrimental to plants (Taylor et al., 1994; Ashmore et al., 2005) and is damaging to forest ecosystems at ambient concentrations in the Northern Hemisphere (Karnosky et al., 2007; Matyssek et al., 2007). Exposure to ambient O₃ levels can decrease stomatal conductance and photosynthetic rate (Novak et al., 2005), as well as decrease stomatal sensitivity to environmental conditions (Paoletti and Grulke 2010), which can lead to reductions in biomass production (Pye 1988, Matyssek et al., 1993, Coleman et al., 1996, US EPA 2006, 2013, 2020) and alter allocation patterns between above and belowground biomass components (Grantz et al., 2006). Over the last 60 years, considerable evidence has demonstrated that ambient concentrations of ozone are sufficient to decrease crop yield and tree growth (Karnosky et al., 2007, US EPA 2006, 2013, 2020). However, the exposure-response relationships between O₃ and decreased tree growth have not been quantified using consistent exposure metrics for many United States (US) tree species.

In the United States, tropospheric O₃ is one of six criteria air pollutants regulated under the Clean Air Act (CAA) through setting primary and secondary National Ambient Air Quality Standards (NAAQSs) to protect human health and welfare, respectively. The current primary and secondary NAAQSs for O₃ are identical and share the form as the fourth-highest daily maximum 8-hour concentration, averaged across three consecutive years and set at a level of 0.07 ppm. However, this O₃ exposure metric is seldom used in vegetation research and other metrics have been shown to better characterize the adverse effects of O₃ on vegetation response (Tingey et al., 1991; Hogsett et al., 1997; Musselman et al., 2006). Experiments on tree seedling growth (Oksanen and Holopainen, 2001) and retrospective analyses of crops (Lee et al., 1988) have shown that the effects of O₃ exposure on plants are cumulative. Other research has demonstrated that plants are comparatively more sensitive to higher concentrations than lower concentrations (Musselman et al., 1983; Hogsett et al., 1985; Nussbaum et al., 1995). Therefore, the metrics of O₃ exposure that produce the best predictions of plant responses are sums of weighted hourly concentrations over the plant growing season, such as W126, SUM06 and AOT40 (Reich 1987; Lee et al., 1988; US EPA, 2020a). In 1996, the United States Environmental Protection Agency (US EPA 1996) considered two specific concentration-weighted indices: (1) the SUM06 index calculated as the sum of all hourly O₃ concentrations ≥ 0.06 ppm, and (2) the sigmoidally-weighted W126 exposure index (Lefohn and Runeckles 1987; Lefohn et al., 1988). The W126 metric uses a continuous sigmoid weighting scheme with inflection point at 0.067 ppm which, to a first-order approximation, is comparable to a 0-1 step function with a jump point at 0.06 ppm for the SUM06 metric. Consequently, both metrics performed similarly for use as a predictor variable to infer the exposure-response relations, based on data from the National Crop Loss Assessment Network (NCLAN) crop studies 1980-1986 (Heck and Cowling 1997; Lee et al., 1988). Recent evidence in the peer-reviewed literature shows a better relationship between W126 and crop loss than between SUM06 and the AOT40 exposure indices and crop loss (McGrath et al., 2015; Mills et al., 2018). Since 2007 (US EPA Staff Paper 2007), US EPA has focused on the daytime 12-hour (8AM to 8PM) 3-month W126 metric for risk assessment and proposed this metric as the secondary standard in 2007 and 2010 (Federal Register proposals 2007, 2010). The use of a 12-h daylight period (8 AM to 8 PM) for a W126 metric is based on evidence that plant stomatal conductance, O₃ concentrations and uptake rates increase to a diurnal peak during the daytime hours (Uddling et al., 2010). This approach was endorsed by the Clean Air Scientific Advisory Committee (CASAC letters from 2007, 2014) which is an independent expert scientific committee set up to advise the US EPA.

Since the NCLAN crop studies, there have been a number of federally funded regression-based ozone exposure studies conducted from 1988-1995 in controlled environments that provide a basis for estimates of exposure-response for seedling tree species. Hogsett et al., (1997) characterized the O₃ exposure-response functions for various tree species based on data from these outdoor open-top chamber (OTC) studies and the three-month 24-h SUM06 metric following its use for NCLAN data (Lee et al., 1988). While more recent evidence favors the use of the 12-h W126 exposure metric for regulatory purposes (US EPA Staff Paper 2007), no peer-reviewed publication has used the 12-h W126 exposure metric for a regression-based meta-analysis of data from existing seedling O₃ exposure studies. Given the different weighting scheme in the summation, and the different daily interval (24 vs. 12 hr), regression estimates using SUM06 and W126 cannot be interconverted.

In general, limitations to understanding exposure-response relationships include: 1) the lack of a common O₃ exposure metric for quantifying plant response to pollutant exposure across studies; 2) the lack of data for a wide range of tree species which may be sensitive to O₃ (Matyssek and Innes 1999); and 3) the interacting effects of other factors that may modify plant response to O₃ (Hogsett et al., 1997). The only way to obtain estimates of the model parameters when using the W126 metric is through reanalyzing the data from experiments themselves. Further, additional tree seedling ozone exposure studies including some less-studied tree species were conducted following Hogsett et al. (1997) that could be used to better determine parameter estimates for exposure-response functions. It is expected that the sensitivity of plant growth to any given level of W126 will differ between tree species because plant response to O₃ varies depending upon species, plant growth strategies, and external environmental factors (light, nutrients, water, temperature) (Tjoelker et al., 1993; Miller et al., 1997). However, how tree species with different anatomical, hydraulic and functional traits respond to O₃ in terms of whole plant biomass is unclear (Grulke and Heath 2020).

The primary objective of this study is to establish a reference set of parameters for seedling exposure-response relationships for 16 North American tree species using the 3-month (92 day) 12-hr W126 metric. The sixteen species are widespread across the U.S. (Figure 1), are ecologically important and include a variety of deciduous and coniferous, and faster and slower growing trees. These parameter values are estimated using experimental data from outdoor OTC studies that quantify whole plant biomass response changes for seedlings exposed to several levels of elevated O₃ for one or more years. Additionally, our analysis of these data includes quantification of the 95% confidence interval of the exposure-response relationships. We compare the different biomass responses between species to illustrate the relative sensitivity of these species to varying levels of ozone exposure using the 12-hr W126 metric. A subset of the species were exposed to ozone for two or more years and/or two or more different O₃ regimes. Using data from these species, we examine whether O₃ effects on plant growth are cumulative over multiple growing seasons and/or dependent on the diurnal and seasonal patterns of O₃ concentrations. These exposure-response relationships will allow US EPA and other groups to better estimate biomass losses based on ozone exposures in North America and can be used in risk assessment and scenario analyses.

Figure 1.

Range distribution maps of the 16 North American tree species within the conterminous U.S. All species extents were derived from USFS's Little (1971, 1977) range maps, reprojected to the Albers Equal Area map projection. The shapefiles used to create the maps were retrieved on 7/6/2021 from the Data Basin organization (https://databasin.org) and the USFS (https://www.fs.fed.us/nrs/atlas/littlefia)

Methods

Study sites and ozone exposure profiles

The seedling O₃ exposure studies for western and eastern tree species were conducted from 1988 to 1995 at the U.S. Environmental Protection Agency research laboratory in Corvallis, Oregon, Michigan Technological University’s Ford Forestry Center in Alberta, Michigan and by researchers from Appalachian State University at Great Smoky Mountains National Park near Gatlinburg, Tennessee (Table 1). Similar experimental protocols were used to expose seedlings to O₃ in 3 m diameter, 2.4 m tall modified open-top chambers (Heagle et al., 1973) as shown in Figure 2. Experiments used a common standard operating procedure developed by the US EPA to ensure federal guidelines for data quality were met (Hogsett et al., 1985). For all studies at all sites, the experimental design was a single-factor nested experiment with a range of O₃ treatment levels including charcoal-filtered air (control), a baseline O₃ profile (1.0x ambient) and several modified O₃ profiles (e.g., 0.5x, 1.5x, 2.0x ambient), with multiple replicate chambers for each treatment. Ozone treatment levels were randomly allocated to the chambers each year and plants were allocated to the chambers using a restricted randomization scheme that minimized the variation in mean plant size (i.e., initial volume calculated as diameter² times height) between chambers.

Table 1.

Table of open-top chamber (OTC) studies of 16 tree species as seedlings by study site.

Site	Species	Oregon Study ID	Study Years	Number of OTCs	Plants per Chamber	Previous Publications
Oregon	Douglas-fir Pseudotsuga menziesii	13	1989-1990	14,111	16 or 24	Hogsett et al (1997)
		35	1991-1992	10	24	Hogsett et al (1997)
	Ponderosa pine Pinus ponderosa	12D	1989	152	16 or 24
		12R	1989-1990	15,111	16 or 24	Hogsett et al (1997)
		26	1990-1991	10	10, 8-233
		34	1991-1992	12,101	16	Hogsett et al (1997)
		45	1992	8	16 or 24
		48	1993	12	24
		57	1994-19954	8	24
	Quaking aspen Populus tremuloides	11	1989	15	16 or 24	Hogsett et al (1997)
		28	1990	10	12
		37	1991	8	24
	Red alder Alnus rubra	14	1989	14	16 or 24	Hogsett et al (1997)
		27	1990	11	16 or 24
		36	1991	10	24	Hogsett et al (1997)
		44	1992	10	7
	Sugar maple Acer saccharum	25	1990	10	10
Michigan	Eastern white pine Pinus strobus		1990	15	24	Hogsett et al (1997)
			1991	15	16	Hogsett et al (1997)
	Quaking aspen Populus tremuloides		1990	15	2 per clone	Karnosky et al. (1996)
			1991	95	5 per clone	Karnosky et al. (1996), Hogsett et al (1997)
			1991	155	16 seedlings	Karnosky et al. (1996), Hogsett et al (1997)
	Sugar maple Acer saccharum		1990	15	8	Hogsett et al (1997)
Great Smoky Mountains National Park, Tennessee	Black cherry Prunus serotina		1989	9	10	Neufeld et al (1995), Hogsett et al (1997)
			1992	15	8	Neufeld et al (1995) , Hogsett et al (1997)
	Chestnut oak Quercus prinus		1991	15	8
	Red maple Acer rubrum		1988	12	7-13	Hogsett et al (1997)
	Sweetgum Liquidambar styraciflua		1989-19906	12	3-8
	American sycamore Platanus occidentalis		1989	9	9
	Table Mountain Pine Pinus pungens		1988-19906	9	11-13
	Tulip popular Liriodendron tulipifera		1990-1991	15	7-8	Hogsett et al (1997)
			1992	15	5-9
	Virginia pine Pinus virginiana		1990-1991	15	6-8	Hogsett et al (1997), Neufeld et al (2000)
			1992	15	3-8	Neufeld et al (2000)
	Winged sumac Rhus copallina		1989	9	4-6
	Yellow buckeye Aesculus flava		1990	15	9

¹Number of open-top chambers were different between harvests 1 and 2.

²Six of the 15 OTC in Study 12D were also included in Study 12R.

³Number of plants per chamber was 10 in 1990 harvest 1 and 8-23 in 1991 harvest 2.

⁴Plants were harvested in 1995 following two years of ozone exposure.

⁵Number of OTCs was 9 for aspen clones and 15 for seedlings in 1991.

⁶Plants were harvested in 1990 following two or three years of ozone exposure.

Figure 2.

Open-top fumigation chambers used in ozone effects research at Twin Creeks near Gatlinburg, Tennessee.

The baseline O₃ profile for studies conducted in Corvallis, Oregon was constructed to reflect the episodic pattern of ambient O₃ concentrations in rural areas for six Midwestern states (Illinois, Indiana, Ohio, Michigan, Wisconsin and Minnesota) (Hogsett et al., 1985). For some seedling studies in Oregon, a second profile was a daily peak profile of equivalent peak concentration and identical diurnal pattern each day (Hogsett et al., 1985); a third profile was a high elevation profile based on 1986 ambient O₃ monitoring data for Big Meadows (1067 m elevation), Shenandoah National Park, Virginia (Figure 3). The hourly O₃ concentrations for the three baseline profiles were weighted using a sigmoid function to generate a series of profiles having the same temporal pattern but different integrated exposure values (Hogsett et al., 1988). While the total exposure values (i.e., the summation of all hourly O₃ concentrations over 24 hours) were equal for the three profiles (episodic, daily peak, and high elevation), the 12-h W126 value for the high elevation high profile (e.g., 48.5 ppm for the 1990 study 11 for quaking aspen (Populus tremuloides) was substantially less than that for the episodic high (97.9 ppm-h) and daily peak high profiles (92.3 ppm-h) (Table 2). Study 34 for ponderosa pine (Pinus ponderosa) included an episodic high profile with the hourly O₃ concentrations either unshifted (EP150) or shifted by 12 hours (EP150 offset) such that peak concentrations occurred during the nighttime. These seven studies that have two or more O₃ profiles are useful to examine the importance of exposure dynamics and test whether plants respond differently to frequent, lower peak O₃ concentrations than to episodic, higher peaks.

Figure 3.

Ozone exposure profiles for open-top chamber studies conducted at US EPA laboratory in Corvallis, Oregon 1989-1995. The requested hourly ozone concentrations over a 30-day period are shown for the A) episodic low, B) episodic high, C) daily peak, D) high elevation low, and E) high elevation high profiles.

Table 2.

Dates of ozone exposure period and harvest ,the 12-h W126 (ppm-h) and 12-h number of hourly O₃ concentrations ≥ 0.10 ppm (N100) exposure metrics for the tree species exposed in open-top chambers at the United States Environmental Protection Agency Pacific Ecological Systems Division laboratory at Corvallis, Oregon. The baseline O₃ profile for studies conducted in Corvallis, Oregon was constructed to reflect the episodic pattern of ambient O₃ concentrations in rural areas for six Midwestern states (Illinois, Indiana, Ohio, Michigan, Wisconsin and Minnesota) (Hogsett et al 1985). For some seedling studies in Oregon, a second profile was a daily peak profile of equivalent peak concentration and identical diurnal pattern each day. A third profile was a high elevation profile. The modified episodic and daily peak profiles were developed by scaling the hourly ozone concentrations using a sigmoid function as described by Hogsett et al (1988). Two or three replicates were used for each ozone treatment level.

Species	Study ID	Harvest Date (Years of Exposure)	Exposure Dates	Exposure Duration (days)	W126 (ppm-h) / N100 values for ozone treatments
					11	2	3	4	5	6
Douglas-fir	132	10/25/1989 (1)	06/07/1989 – 09/27/1989	113	0.0/0.0	13.3/25.1	59.2/240.6	82.8/351.3	103.4/490.9	-
		11/06/1990 (2)	06/05/1990 – 10/03/1990	121	0.0/0.0	13.5/25.3	61.5/252.9	85.8/354.6	109.5/514.6	-
	353	11/18/1991 (1)	06/05/1991 – 09/30/1991	118	0.0/0.0	13.0/24.0	27.7/83.5	59.8 /243.5	82.6/348.1	-
		10/30/1992 (2)	06/02/1992 – 09/21/1992	112	0.1/0.5	11.8/18.9	25.6/78.2	56.9/233.9	79.4/339.7	-
Ponderosa pine	12D4	09/29/1989 (1)	06/07/1989 – 09/27/1989	113	0.0/0.0	7.3/0.0	53.8/4.7	-	100.4/84.1	-
	12R2	10/01/1989 (1)	06/07/1989 – 09/27/1989	113	0.0/0.0	13.3/25.1	59.2/240.6	82.8/351.3	103.4/490.9	-
		10/23/1990 (2)	06/05/1990 – 10/03/1990	121	0.0/0.0	13.5/25.3	61.5/252.9	85.8/354.6	109.5/514.6	-
	265	10/03/1990 (1)	05/16/1990 – 10/03/1990	141	0.0/0.0	15.8/27.4	47.96/210.7	49.06/210.8	130.0/608.2	-
		09/24/1991 (2)	05/01/1991 – 09/16/1991	139	0.5/0.0	16.1/33.3	66.7/287.1	77.7/358.9	128.3/612.9	-
	347	11/04/1991 (1)	06/05/1991 – 09/30/1991	118	0.0/0.0	13.0/24.0	27.7/83.5	59.8/243.5	82.6/348.1	11.6/25.5
		10/12/1992 (2)	06/02/1992 – 09/21/1992	112	0.1/0.5	11.8/18.9	25.6/78.2	56.9/233.9	79.4/339.7	-
	458	09/01/1992 (1)	05/13/1992 – 09/01/1992	112	0.1/0.4	-	-	56.9/232.9	-
	489	09/27/1993 (1)	06/01/1993 – 09/21/1993	113	0.0/0.0	28.0/89.9	51.3/170.6	60.4/252.7	64.4/337.7	64.5/317.8
	5710		06/16/1994 – 10/04/1994	111	0.0/0.0	-	47.9/154.5	56.2/228.8	-	60.1/291.1
		10/10/1995 (2)	06/16/1995 – 10/05/1995	112	0.0/0.0	-	50.0/163.8	57.6/240.0	-	61.1/297.4
Quaking aspen	1111	09/18/1989 (1)	06/06/1989 – 09/18/1989	105	0.0/0.0	6.8/0.0	48.5/4.7	56.0/228.3	97.9/472.0	92.3/82.3
	2812	09/19/1990 (1)	06/05/1990 – 09/19/1990	107	0.0/0.0	12.1/25.3	54.7/228.4	76.6/328.2
	3713	09/11/1991 (1)	06/05/1991 – 09/11/1991	99	0.0/0.0	11.0/18.4	23.4/70.6	-	70.1/296.2	-
Red alder	142	10/01/1989 (1)	06/07/1989 – 09/27/1989	113	0.0/0.0	13.3/25.1	59.2/240.6	82.8/351.3	103.4/490.9	-
	272	10/10/1990 (1)	06/05/1990 – 10/03/1990	121	0.0/0.0	13.5/25.3	61.5/252.9	85.8/354.6	109.5/514.6	-
	363	10/07/1991 (1)	06/05/1991 – 09/30/1991	118	0.0/0.0	13.0/24.0	27.7/83.5	59.8/243.5	82.6/348.1	-
	443	10/05/1992 (1)	06/02/1992 – 09/21/1992	112	0.1/0.5	11.8/18.9	25.6/78.2	56.9/233.9	79.4/339.7	-
Sugar maple	255	10/03/1990 (1)	05/16/1990 – 10/03/1990	141	0.0/0.0	15.8/27.4	47.9/211.1	49.0/211.2	130.0/608.2	-

¹1 = Charcoal-filtered (CF)

²2 = EPXLOW, 3 = EPLOW, 4 = EPMED, 5 = EPHIGH

³2 = EP70 (= EPXLOW), 3 = EP90, 4 = EP120 (= EPLOW), 5 = EP150 (= EPMED)

⁴2 = High Elevation Low (HELLOW), 3 = High Elevation High (HELHI), 5 = Daily Peak (DAILY)

⁵2 = EPXLOW (= EP70), 3 = HIXLOW (= EP175/EP70), 4 = XLOWHI (= EP70/EP175), 5 = EPHIGH (= EP175)

⁶93-day exposure duration, end exposure date 08/16/1990

⁷2 = EP70, 3 = EP90, 4 = EP120, 5 = EP150, 6 = EP150* (= EP150 but offset by 12 hours)

⁸4 = EP120

⁹2 = EP90, 3 = Daily Peak Low (= PEAK135), 4 = EP120, 5 = BROAD120, 6 =Daily Peak High (= PEAK214)

¹⁰3 = PEAK135, 4 = EP120, 6 = PEAK214

¹¹2 = HELLOW, 3 = HELHI, 4 = EPLOW, 5 = EPHIGH, 6 = Daily Peak (or DAILY)

¹²2 = EPXLOW, 3 = EPLOW, 4 = EPMED

¹³2 = EP70, 3 = EP90, 5 = EP150

The baseline O₃ profile for the Michigan studies was based on the 1987 ambient O₃ monitoring data for Washtenaw County, Michigan (Karnosky et al., 1996) and modified using a sigmoid function to generate the 0.5x, 1.5x and 2.0x ambient treatment levels following Hogsett et al. (1988) (Figure 4). For the Tennessee studies, the baseline O₃ profile was based on real-time local ambient air concentrations and modified to generate the 0.5x, 1.0x, 1.5x and 2.0x ambient treatment levels (Neufeld et al., 1995) (Figure 5). The 0.5x ambient treatment level was added in 1989 when the OTC facility in Tennessee was expanded. The 12-h (8 AM to 8 PM) W126 and 12-h number of hourly O₃ concentrations ≥ 0.10 ppm (N100) exposure metrics were initially calculated for the period of exposure for each chamber, study and year (Tables 2-4) and scaled to a 92-day W126 value by the scaling factor of 92 days divided by the duration of exposure in days. Outliers and missing values in the hourly O₃ data files represented less than 1% of the data and the W126 value was adjusted by the scaling factor of the total number of hours in the exposure period divided by the total number of hours of available data. The number of hourly O₃ concentrations ≥ 0.10 ppm ranged from less than 1% of the modified ambient profiles to as high as 45% of several episodic high (Oregon) and 2.0 x (Michigan, Tennessee) profiles (Tables 2-4).

Figure 4.

Ozone exposure profiles for open-top chamber studies conducted at Michigan Technological University’s Ford Forestry Center in 1990. The targeted ozone profiles are based on modified ambient (1.0x) from southern Michigan and are used in both 1990 and 1991 (Karnosky et al., 1996). The actual hourly ozone concentrations for the period June 20 to September 10, 1990 are shown for the A) charcoal-filtered, B) 0.5 x, C) 1.0 x, D) 1.5 x, and E) 2.0 x modified ambient profiles. Sigmoidal weighting was used to develop the 0.5x, 1.5x, and 2.0x profiles.

Figure 5.

Ozone exposure profiles for open-top chamber studies conducted at Great Smoky Mountains National Park, Tennessee in 1990. The targeted ozone profiles are based on modified ambient (1.0x) from a monitoring station in the Park and are used in 1990, 1991 and 1992 (Neufeld et al. 2000). The actual hourly ozone concentrations for the period June 30 to September 12, 1990 are shown for the A) charcoal-filtered, B) 0.5 x, C) 1.0 x, D) 1.5 x, and E) 2.0 x modified ambient profiles. Sigmoidal weighting was used to develop the 0.5x, 1.5x, and 2.0x profiles.

Table 4.

Dates of ozone exposure period and harvest, the 12-h W126 (ppm-h) and 12-h number of hourly O₃ concentrations ≥ 0.10 ppm (N100) exposure metrics for the tree species exposed in open-top chambers at the Great Smoky Mountains National Park, Tennessee.

Species	Harvest Date (Years of Exposure)	Exposure Dates	Exposure Duration (days)	W126 values (ppm-h) / N100 for ozone treatments
				CF	0.5x	1.0x	1.5x	2.0x
American sycamore	08/21/1989 (1)	06/14/1989 – 08/21/1989	69	0.0/0.0	-	1.9/0.5	10.7/14.3	22.1/76.6
Black cherry	10/05/1989 (1)	06/14/1989 – 08/28/1989	76	0.0/0.0	-	1.9/0.0	11.1/10.0	23.0/77.3
	10/13/1992 (1)	05/20/1992 – 10/06/1992	140	0.0/0.0	0.0/0.0	1.4/0.0	15.1/5.0	39.5/102.8
Chestnut Oak	10/22/1991 (1)	05/23/1991 – 10/08/1991	139	0.0/0.0	0.2/1.1	1.2/0.0	16.4/7.6	39.0/87.7
Red Maple	08/24/1988 (1)	07/01/1988 – 08/24/1988	55	0.0/0.0	-	12.012.6	-	59.81300.0
Sweetgum		06/19/1989 – 09/28/1989	102	0.2/0.5	-	2.2/0.5	11.7/14.7	24.1/74.4
	08/09/1990 (2)	06/30/1990 – 08/08/1990	40	0.1/0.0	-	1.0/0.0	6.3/10.3	13.2/31.4
Table Mountain Pine		07/01/1988 – 08/24/1988	55	0.0/0.0	-	12.0/2.6	-	59.8/300.0
		06/15/1989 – 09/28/1989	106	0.2/0.6	-	2.2/0.6	-	25.1/77.3
	08/22/1990 (3)	06/30/1990 – 08/22/1990	54	0.1/0.0	-	1.2/0.0	-	17.5/35.0
Tulip Poplar	09/12/1990 (1)	06/30/1990 – 09/12/1990	75	0.1/0.0	0.12/0.0	1.5/0.0	11.2/12.3	25.3/51.4
	10/08/1991 (2)	05/01/1991 – 10/08/1991	161	0.0/0.0	0.4/1.1	1.5/0.0	18.7/7.9	45.3/101.6
	10/08/1992 (1)	05/20/1992 – 10/08/1992	142	0.0/0.0	0.0/0.0	1.4/0.0	15.2/5.0	39.7/103.0
Virginia Pine	09/27/1990 (1)	06/30/1990 – 09/27/1990	90	0.1/0.0	0.2/0.0	1.8/0.0	13.4/14.7	30.4/61.6
	09/23/1991 (2)	05/06/1991 – 09/23/1991	141	0.0/0.0	0.4/1.1	1.3/0.0	16.9/7.7	40.2/90.1
	10/09/1992 (1)	05/04/1992 – 10/09/1992	159	0.0/0.0	0.1/0.0	2.5/0.0	20.0/17.6	49.1/133.8
Winged Sumac	09/25/1989 (1)	06/19/1989 – 09/25/1989	99	0.2/0.5	-	2.1/0.5	11.4/14.3	23.4/72.2
Yellow Buckeye	07/30/1990 (1)	06/30/1990 – 07/30/1990	31	0.1/0.0	0.4/0.0	2.2/2.5	5.9/12.3	9.8/26.6

¹In 1988, 7-day ozone profiles were based on previous year’s data at the Look Rock monitoring station in the Park (Neufeld et al. 2000). In subsequent years, the ozone profiles were modified ambient treatments. There were three replicates for all treatments in 1988, two in 1989 except for the 2.0x ambient which had three.

²In 1990, the exposure facility expanded from 9 to 15 open-top chambers and a 0.5x treatment was added (Neufeld et al. 2000). There were three replicate chambers for all treatments in 1990-1992.

Harvest biomass measurements

Seedlings were exposed to O₃ in OTC for either a single or multiple growing seasons, and a subset of plants were destructively harvested following O₃ exposure at the end of the final and/or each previous growing season. For each study and harvest, root, needle, and stem biomass for individual trees were summed to calculate total plant biomass, log transformed, and averaged across individuals within a chamber to calculate a chamber mean. Similarly, initial height and diameter of individual trees, when available, were used to calculate initial plant volume as height x diameter², log transformed, and averaged across individuals within a chamber for possible use as a covariate in the regression model.

Statistical analysis

Ozone effects on total plant biomass response are examined using a mixed-effects model analysis to fit a linear or Weibull model (Rawlings and Cure 1985) at the chamber mean level. In a simple linear regression setting, assume we have observations (X₁,Y₁), …, (X_n,Y_n) where Y_i is total plant biomass and X_i is the 12-h W126 metric for chamber i=1,2,…,n. The three-parameter Weibull model that relates total plant biomass (Y) to the 12-h W126 metric (X) has the form

$${\mathrm{Y}}_{\mathrm{i}}=\mathrm{A}{}^{\ast }\text{exp}(\text{-}({\mathrm{X}}_{\mathrm{i}}∕\mathrm{B}{)}^{\mathrm{C}}){}^{\ast }{\mathrm{e}}_{\mathrm{i}}$$ Equation 1

where A, B and C are fixed unknown model parameters and e_i is the multiplicative error term; model parameter B is the O₃ metric value associated with a 63% biomass reduction and C is the shape parameter. Taking the natural logarithm, the Weibull model has the form

$$\text{ln}({\mathrm{Y}}_{\mathrm{i}})=\text{ln}(\mathrm{A})\text{-}({\mathrm{X}}_{\mathrm{i}}∕\mathrm{B}{)}^{\mathrm{C}}+{\mathrm{u}}_{\mathrm{i}}$$ Equation 2

where u_i is a normal independent random variable with mean 0 and variance σ². Note that the linear regression model is a special case of the Weibull model with C=1. The basic three-parameter Weibull model can be extended to include other explanatory variables (i.e., covariates) that correlate with the response variable (Y) but are not affected by the O₃ treatment (Rawlings and Cure, 1985). While initial plant volume was used initially for pre-stratification of plants to the OTCs, several studies required a covariate be included in the Weibull model to account for chamber-to-chamber variation in plant size. We extend the Weibull model to include the log of initial plant volume (Z) as a covariate by the form

$$\text{ln}({\mathrm{Y}}_{\mathrm{i}})=\text{ln}(\mathrm{A}+\mathrm{D}{}^{\ast }({\mathrm{Z}}_{\mathrm{i}}\text{-}\overline{\mathrm{Z}}))\text{-}({\mathrm{X}}_{\mathrm{i}}∕\mathrm{B}{)}^{\mathrm{C}}+{\mathrm{u}}_{\mathrm{i}}.$$ Equation 3

Given we have observations (X_ijk, Y_ijk, Z_ijk) for chamber i=1, 2, …,n_jk, harvest j=1,h_k and study k=1, 2, …,s, there are several possible regression-based approaches to infer the average exposure-response relation for each tree species based on the combined data from multiple studies and harvests. One approach is to fit the Weibull model in Equation 2 (or 3 with covariate) for each study-harvest combination which either has its own set of model parameters (A_jk, B_jk, C_jk) or follows the same line with different intercepts for harvests, i.e., B_jk=B₀ and C_jk=C₀, following Hogsett et al. (1997). As a natural extension of the regression model, the random coefficients model assumes each study-harvest has its own line, but the model parameters come from an assumed distribution. Data were combined across studies and harvests for each tree species and analyzed using the three- or four-parameter Weibull model with or without random coefficients. The three-parameter random coefficients Weibull model has the form

$$\mathrm{E}(\text{ln}({\mathrm{Y}}_{\text{ijk}})\mid {\mathrm{A}}_{\text{jk}},{\mathrm{B}}_{\text{jk}},\mathrm{C})=\text{ln}({\mathrm{A}}_{\text{jk}})\text{-}({\mathrm{X}}_{\text{ijk}}∕{\mathrm{B}}_{\text{jk}}{)}^{\mathrm{C}}$$ Equation 4

where A_jk ~ N(α_j , σ_1j²) and B_jk ~ N(β , σ₂². The shape parameter C is assumed to be fixed and unknown rather than have a stochastic distribution because the parameter space for B and C are jointly constrained. Hierarchical regression analysis is used to test the adequacy of the linear model based on the likelihood ratio test for C=1 at the 0.05 level of significance. Differences in plant size between first- and second-year harvests are modeled by a Weibull model having different Y-intercepts (i.e., parameter A).

For studies having multiple harvests, hierarchical regression analysis is used to test the hypothesis of a greater plant response to O₃ following two years of O₃ exposure relative to a single year of exposure. Harvest differences in O₃ effects are examined by testing the adequacy of a Weibull model with a common B parameter in relation to a Weibull model having separate B parameters for each harvest. For the quaking aspen studies, differences in O₃ effects between clones and seedlings are examined by testing the adequacy of a Weibull model with common B parameter in relation to a Weibull model having separate B parameters for each clone and/or seedling. For some tree species, a different line was fit to some study-harvests using either a Weibull model with random coefficients or a Weibull model with different A and B parameters. The Weibull model with random coefficients could not be fit to some tree species because there were an insufficient number of replicate studies to estimate the model parameter variances. For some other tree species, the fixed-effects Weibull model was selected in favor of the mixed-effects Weibull model based on statistical fit and parsimony. Residual analysis was performed to ensure the data did not violate the model assumptions of normality and random errors.

The Weibull model is used to calculate the predicted relative biomass loss (PRBL) as described by Lee et al. (1990) by the form

$$\text{PRBL(X)}=1-\text{exp}(\text{-}(\mathrm{X}∕\mathrm{B}{)}^{\mathrm{C}}).$$ Equation 5

The PRBL function is used to compare the O₃ effects on plant biomass between tree species and to determine the threshold level associated with a 5% biomass loss for each tree species. A reference level of 5% annual biomass loss was chosen for several reasons. Based on CASAC science advice, US EPA policy assessments have used 5-6% benchmark for a biomass loss that is unacceptable (US EPA 2014). Hogsett et al., (1997) estimated annual biomass losses ranging from 5% to 13% for moderately sensitive tree species at ambient O₃ levels > 20 ppm-h for the three-month 24-h SUM06 metric in the eastern United States. A meta-analysis by Wittig et al., (2009) reported an average annual total biomass loss of 7% following exposure to current ambient O₃ concentrations of 40 ppb on average for 7 or more days across 263 studies. Biogeochemical models predict tropospheric O₃ in combination with nitrogen fertilization will result in annual net primary production reductions of 2.3% to 7.2% for the US 1989-1993 (Felzer et al . 2005). Statistically, a biomass loss of 5% is generally within the experimental design limits of the OTC studies to detect at the 0.05 level of significance when data are combined across studies (e.g., Wittig et al., 2009). Trees unaffected by O₃ have a PRBL of 0% change from control. A positive percent change indicates a decrease in plant biomass in response to a given integrated O₃ exposure value. The 95% confidence interval for PRBL for each tree species is calculated using one of three statistical approaches depending upon curvilinearity and stochastic nature of model parameters. The 95% likelihood confidence interval for PRBL is calculated based on the likelihood ratio test and a chi-squared distribution with 1 degree of freedom when a random coefficients regression model is fit to the combined data following Lee et al., (1990). The 95% confidence interval for PRBL is based on the F-test with 1 and n-p degrees of freedom where n=sample size and p=number of model parameters when a fixed-effects Weibull model with C>1 is used to fit the data. The 95% confidence interval for PRBL is based on the Wald t-test when plant biomass response to O₃ exposure is linear (i.e., C=1) and no random coefficient terms are included in the regression model. The actual coverage of the Wald and F-test confidence intervals for PRBL are close to nominal when curvature effects are low (Lee et al., 1990). All random coefficients model and likelihood interval calculations were performed using the nlmer function in the lme4 package (Bates et al., 2015) and the R programming language (R Core Team 2013). All fixed effects nonlinear regression model calculations were performed using the nls function in the R programming language.

Database

A database of the effects of O₃ on biomass response of trees as seedlings was compiled initially by Hogsett et al., (1997) and supplemented with additional studies conducted by or in collaboration with the Ozone Research Program 1979-2006 at the US EPA research laboratory in Corvallis, Oregon (Table 1). Studies were included if the hourly O₃ monitoring and plant biomass data were available for every open-top chamber in the study, the regression-based design included three or more O₃ treatment levels including a charcoal-filtered (CF) treatment for a control, and the description of experimental design (e.g., ozone treatment and plant assignment to chambers, start and end dates of exposure, harvest date) was sufficient to allow for regression analysis (Table 1). Portions of the data were previously used in peer-reviewed publications and in EPA science reviews (US EPA 1996, 2006, 2013). Several of the studies reported in Hogsett et al., (1997) were excluded because the archived data were incomplete. Several OTC studies for Douglas-fir (Pseudotsuga menziesii), ponderosa pine, tulip poplar (Liriodendron tulipifera) and Virginia pine (Pinus virginiana) exposed plants to elevated O₃ concentrations for multiple growing seasons and were harvested multiple times. For those multi-year studies, biomass was measured at the end of each growing season following exposure. Several Oregon OTC studies for ponderosa pine (Studies 12D, 26, 34, 48, and 57) and quaking aspen (Study 11) exposed plants to two or more different O₃ regimes having the same total O₃ exposure (i.e., SUM00) but different peak O₃ concentrations to examine the role of exposure dynamics on plant biomass response (Table 2). In total, 36 studies for 16 tree species were used in the regression analysis.

Results

Total plant biomass response to O₃ exposure varied by years of exposure, exposure dynamics, and species and was adequately described by either a linear or Weibull function (Table 5). The species-dependent exposure-response models explained between 34% and 98% of the total variation in total plant biomass except for two tolerant tree species, table mountain pine (Pinus pungens) and yellow buckeye (Aesculus flava), which had R² values < 27% (Table 5). Species responses to O₃ exposure ranged from no biomass reductions for yellow buckeye to biomass reductions of 77% for quaking aspen.

Table 5.

Weibull model with random coefficients relating log total plant biomass with the 92-day 12-h W126 ozone exposure metric for 16 tree species based on open-top chamber studies conducted in Oregon, Michigan, and Tennessee 1988-1995. Species are listed in order from most sensitive to least sensitive based on the W126 value associated with 5% biomass loss. The Weibull model has the form, $\mathrm{E}(\text{ln}(\mathrm{Y}))=\text{ln}(\mathrm{A}+\mathrm{D}{}^{\ast }(\mathrm{Z}\text{-}\overline{\mathrm{Z}})+{\text{A2}}^{\ast }\text{Harvest}\_2+{\text{A3}}^{\ast }\text{Study}\_\text{Harvest})\text{-}(\mathrm{X}∕(\mathrm{B}+{\text{B2}}^{\ast }\text{Harvest}\_2){)}^{\mathrm{C}}$ where X=W126, Z is the covariate, ln(Height x Diameter²), Harvest_2 is the 0-1 indicator variable for Harvest 2, Study_Harvest is the 0-1 indicator for a specific study and harvest, (A, B, C, D) are unknown model parameters for the basic Weibull model, and (A2, A3, B2) are the unknown model parameters for testing the differences in intercepts and slopes between harvests.

Species	Fixed Effects							Random Effects				R2
	Basic Weibull Model Parameters				Harvest and Study Differences			Variance Parameters
	A (g)	B (ppm-h)	C	D (g/log(cm3))	A2 (g)	B2 (ppm-h)	A3 (g)	A (g2)	B (ppm-h)2	A2 (g2)	σ2 (ln(g)2)
Black Cherry	42.0 (3.7)	48.8 (8.8)	1.0	17.31 (2.6)				21.6	82.6		0.024	0.85
Tulip Poplar	129.7 (8.8)	69.2 (16.1)	1.0	30.3 (12.7)	129.7 (18.3)		−98.02 (8.7)				0.051	0.94
Ponderosa Pine3	15.8 (1.3)	187.9 (23.3)	1.0	118.6 (81.8)	22.7 (2.7)	−54.53 (26.9)	74.13 (5.1)	9.0		32.4	0.016	0.96
American sycamore	51.6 (3.6)	137.4 (71.9)	1.0								0.016	0.34
Winged Sumac	42.6 (2.5)	34.4 (8.7)	2.1 (1.3)								0.009	0.77
Quaking Aspen4	60.54 (1.9)	66.9 (3.2)	1.5 (0.2)	46.04 (13.1)			19.34 (2.9)				0.052	0.91
Red Alder	53.8 (5.3)	179.8 (22.4)	1.0					101.5	389.5		0.017	0.85
Red Maple	26.4 (1.0)	406.2 (105.8)	1.0								0.005	0.68
Sweetgum	85.3 (4.5)	34.3 (2.6)	9.7 (7.8)								0.018	0.59
Sugar Maple	4.4 (0.2)	559.2 (539.1)	1.0				−1.35 (0.3)				0.027	0.56
Virginia Pine	18.7 (0.6)	644.5 (664.7)	1.0	11.6 (2.5)	3.8 (0.9)		58.216 (2.5)				0.012	0.97
Eastern White Pine	0.3 (0.1)	705.7 (1290.0)	1.0				0.77 (0.1)				0.020	0.94
Chestnut Oak	639.5 (16.4)	811.1 (1339.1)	1.0	337.2 (67.3)							0.005	0.68
Douglas-fir	16.3 (0.5)	1002.1 (433.1)	1.0	9.7 (3.0)	16.5 (1.0)		70.18 (2.9)				0.006	0.98
Table Mountain Pine	304.6 (9.1)	1180.2 (2407.7)	1.0								0.004	0.03
Yellow Buckeye	1785.6 (52.6)			1237.9 (572.8)							0.011	0.26

¹D is the slope of the covariate for 1992 black cherry study.

²A3 is the difference in total plant biomass between the overall mean and 1990 study harvest 1.

³The total plant biomass response to the daily peak high ozone treatment for ponderosa pine study 12D was less than that for the episodic high treatment having similar W126 value (p=0.056). The test of a common plant biomass response to one and two years of ozone exposure (i.e., H₀: B2=0) is rejected at the 0.05 level of significance (p=0.031). A3 is the difference in total plant biomass between the overall mean and study 48 harvest 1.

⁴The total plant biomass response to the daily peak high ozone treatment for aspen study 11 was significantly less than that for the episodic high treatment having similar W126 value (p=0.0007). A is the intercept for aspen clones. The estimated difference in intercepts is −45.8 (SE=1.8, p<0.0001) between seedlings and clones and −25.1 (SE=2.2, p<0.0001) between seedlings and 1991 Michigan clones. D is the slope of the covariate for Oregon studies. The estimated difference in slopes of the covariate for Michigan and Oregon studies is −34.9 (SE=13.3, p=0.009). A3 is the difference in total plant biomass between the overall mean and 1991 Oregon study harvest 1.

⁵A3 is the difference in total plant biomass between the overall mean and Oregon study 25 harvest 1.

⁶A3 is the difference in total plant biomass between the overall mean and 1992 study harvest 1.

⁷A3 is the difference in total plant biomass between the 1990 and 1991 studies harvest 1.

⁸A3 is the difference in total plant biomass between the overall mean and study 35 harvest 2. The difference in biomass between the overall mean and study 35 harvest 1 is 8.8 (SE=0.8, p<0.0001).

Plant response to multiple years of O₃ exposure was better correlated with the current-year W126 metric than the integrated sum of W126 values across years, suggesting that biomass was more strongly influenced by current than previous-year(s) pollutant conditions. Several weighting schemes to integrate the W126 values over two growing seasons were initially considered but there was no improvement in statistical fit using the two-year weighted W126 metrics over the current-year W126 metric. Consequently, the exposure-response functions for individual studies and harvests are reported using the current-year three-month 12-h W126 exposure metric. The use of a common current-year W126 metric allowed for the comparison of exposure-response curves between harvests for the same study as well as between studies for the same species.

Ozone effects on total plant biomass after one or more years of exposure

Reductions in total plant biomass occurred after one or more years of O₃ exposure as indicated by a positive Weibull model parameter B or a negative growth trend (i.e., −1/B) for all tree species except yellow buckeye (Table 5). Plant response to a single year of O₃ exposure was adequately explained by a common line (i.e., model parameter B is fixed, unknown) for all tree species except black cherry (Prunus serotina) and red alder (Alnus rubra) which were better fit using a random coefficients model. The random coefficients models for black cherry and red alder assumed each study had its own line and the model parameter B came from an assumed distribution with significantly positive variance (Table 5). Plant biomass responded linearly to O₃ exposure for 12 tree species, but the negative growth trends were not statistically different from zero at the 0.05 level of significance for chestnut oak (Quercus prinus, p=0.77), eastern white pine (Pinus strobus, p=0.58), sugar maple (Acer saccharum, p=0.30), American sycamore (Platanus occidentalis, p=0.10), table mountain pine (p=0.64), and Virginia pine (p=0.34). In addition, yellow buckeye showed no biomass response to elevated O₃ as described by a constant mean model. The O₃ impacts on plant biomass after one or more years of O₃ exposure were statistically significant at the 0.05 level for black cherry, Douglas-fir, ponderosa pine, quaking aspen, red alder, red maple (Acer rubrum), sweetgum (Liquidambar styraciflua), tulip poplar, and winged sumac (Rhus copallina), suggesting that the current-year biomass response to elevated O₃ was a reliable indicator of sensitivity. Relative plant biomass loss response to O₃ exposure was nonlinear for three species, notably sweetgum (Weibull model parameter C=9.7) which had a plateau-linear response after two years of O₃ exposure and winged sumac (C=2.1) (Figure 6).

Figure 6.

Species-dependent predicted relative biomass loss (PRBL) equations as a function of the three-month 12-h W126 metric for 16 tree species exposed to ozone for one or more growing seasons. The shaded region represents the 95% confidence interval for PRBL when data are combined across studies and/or harvests for each species. Species are ordered from most sensitive to least sensitive.

Cumulative ozone effects on total plant biomass following two years of exposure

Harvest data following a second season of O₃ exposure (denoted Harvest 2) were available to examine the cumulative effects of multiple years of O₃ exposure for Douglas-fir, eastern white pine, ponderosa pine and tulip poplar. Significant reductions in plant growth following one season of O₃ exposure persisted following a second season of O₃ exposure for Douglas-fir, ponderosa pine and tulip poplar. For ponderosa pine, the test of a common plant biomass response to one and two years of ozone exposure (i.e., H₀: B2=0 where B2 is the difference in the model parameter B between harvests 1 and 2) was rejected at the 0.05 level of significance (p=0.031). Reductions in growth of ponderosa pine due to elevated O₃ were greater following two years of exposure than a single year of exposure, indicating that the effects of O₃ were cumulative but less than additive (Figure 7). In contrast, evidence of a cumulative O₃ effect on plant biomass was weak for Douglas-fir and tulip poplar due partially to the low number of replicate study-harvests. Eastern white pine total plant biomass was marginally affected by ozone following one or two years of exposure.

Figure 7.

Comparison of exposure-response equations relating total plant biomass to the three-month 12-h W126 metric for ponderosa pine to examine the immediate and prolonged effects of ozone on plant growth at the end of the growing season after A) one season of ozone exposure and B) two seasons of exposure. Plants exposed to a daily peak profile for one season in 1989 (Study 12D) represent a nearly significant departure from the fitted line (p=0.056).

Effects of exposure dynamics and peak ozone concentrations on plant biomass

The cumulative O₃ effect on plant biomass was adequately described using the Weibull function and the 12-h W126 metric to characterize the exposure-response relationships for all species and O₃ treatments, but a more complex exposure-response relationship was observed for quaking aspen and ponderosa pine exposed to two or more different O₃ profiles (Table 5). Total plant biomass response to the daily peak high ozone treatment for the 1989 quaking aspen study 11 was significantly less than that for the episodic high treatment having similar W126 value (p=0.0007) (Figure 8). The total plant biomass response to the daily peak high ozone treatment for the 1989 ponderosa pine study (Study 12D) was less than that for the episodic high treatment having similar W126 value (p=0.056), while a single line was adequate for ponderosa pine exposed for a single year to two or more different O₃ profiles for the other OTC studies (Figure 7). Plant biomass responses of ponderosa pine exposed to the Midwestern episodic profile offset by 12-h or not offset were adequately fit to a single line for one OTC study (Study 34), indicating the greater role of daytime peak O₃ concentrations than nighttime peak O₃ concentrations in eliciting a plant biomass response.

Figure 8.

A common exposure-response equation relating total plant biomass to the three-month 12-h W126 metric for quaking aspen clones and seedlings examines the effects of ozone on plant growth after one year of exposure. Plants exposed to a daily peak profile for one season in 1989 (Study 11) represent a significant departure from the fitted line (p=0.0007).

Relative O₃ sensitivity of tree species

Data across studies and harvests were combined to calculate the PRBL equation and 12-h W126 value associated with a 5% biomass loss for each species as a measure of O₃-sensitivity (Table 6). We defined three levels of O₃-sensitivity based on the test of a significant O₃ effect and distinct jumps in the W126 value associated with a 5% biomass loss: sensitive - black cherry, ponderosa pine, quaking aspen, red alder, American sycamore, tulip poplar and winged sumac; intermediate –red maple and sweetgum; insensitive – chestnut oak, Douglas-fir, eastern white pine, sugar maple, table mountain pine, Virginia pine, and yellow buckeye (Table 6). Red maple and sweetgum were rated intermediate based on a significant linear response to O₃ (p< 0.05) and a PRBL of 5% at W126 of 20.8 and 25.2 ppm-h, respectively, whereas sugar maple and Virginia pine were rated insensitive based on a non-significant linear response to O₃ (p=0.30 and 0.33, respectively) and a PRBL of 5% at W126 > 28 ppm-h. For consistency, tree species having a W126 >28 ppm-h were classified as insensitive regardless of the statistical test of a significant O₃ effect. There was a distinct break in the W126 values associated with a 5% biomass loss between the sensitive and intermediate groupings. Based on PRBL to protect from a 5% biomass loss, the three-month 12-h W126 estimated range was 2.5-9.2 ppm-h for sensitive species, 20.8-25.2 ppm-h for intermediate species, and ≥ 28.7 ppm-h for insensitive species. Black cherry was by far the most O₃-sensitive tree species sampled based on a PRBL of 18.5% and 33.6 % for W126 values of 10 and 20 ppm-h, respectively (Figure 9). Ponderosa pine and red alder were rated as sensitive based on a statistically significant PRBL of 5% for W126 values less than 10 ppm-h (Table 6). Conifers except for ponderosa pine were among the least O₃-sensitive, particularly table mountain pine and eastern white pine which were rated insensitive based on a PRBL of 5% for W126 values > 28 ppm-h (Figure 9).

Table 6.

Predicted relative biomass loss (PRBL) as a function of the 12-h W126 ozone metric for the sixteen tree species based on the random coefficients predicted total dry weight plant response. Species are listed in order from most sensitive to least sensitive based on the W126 value associated with 5% biomass loss. PRBL = (1-exp(−(W126/B)^C))*100. Ozone rating based on the test of a significant O₃ effect and W126 value associated with 5% biomass loss: 1) sensitive if W126<10 ppm-h; 2) intermediate if 10<W126<28 ppm-h; and 3) insensitive if W126>28 ppm-h and/or a linear slope not significantly different from zero.

	Weibull Parameters		Predicted Relative Biomass Loss and 95% Confidence Interval at W126 values ranging from 10 to 50 ppm-h					W126 value associated with PRBL of 5% (ppm-h)	Ozone sensitivity
Species	B (ppm-h)	C	10 ppm-h (%)	20 ppm-h (%)	30 ppm-h (%)	40 ppm-h (%)	50 ppm-h (%)
Black Cherry	49	1.0	18.5 (3.5,33.5)1	33.6 (9.4, 56.4)	45.9 (13.2,75.0)			2.5	sensitive
Tulip Poplar	69	1.0	13.5 (7.6 19.3)2	25.1 (14.9,35.3)	35.2 (22.0,48.4)			3.5	sensitive
Ponderosa Pine Harvest 1	188	1.0	5.2 (3.9,6.3)1	10.1 (8.3,11.7)	14.8 (12.3,16.9)	19.2 (16.3,21.7)	23.4 (19.9,26.3)	9.6	sensitive
Ponderosa Pine Harvest 2	133	1.0	7.2 (5.5,9.1)1	13.9 (11.1,17.1)	20.1 (16.5,24.3)	25.9 (21.3,30.9)	31.3 (26.1,36.9)	6.0
American sycamore	137	1.0	7.0 (0.0, 15.4)2	13.5 (0.0, 29.1)	19.6 (0.0, 41.3)			7.0	sensitive
Winged Sumac	34	2.1	7.0 (0.3, 25.8)3	27.0 (3.3,44.9)				8.5	sensitive
Quaking Aspen	67	1.5	5.8 (2.2,11.4)3	15.4 (8.2,23.4)	26.3 (17.4,34.6)	37.3 (28.4,44.8)	47.8 (39.8,54.4)	9.0	sensitive
Red Alder	180	1.0	5.4 (3.8, 7.5)1	10.5 (7.5, 13.5)	15.4 (11.2,19.5)	19.9 (15.0,25.5)	24.3 (18.0,30.8)	9.2	sensitive
Red Maple	406	1.0	2.4 (1.0, 3.9)2	4.8 (1.9, 7.7)	7.1 (2.9, 11.3)	9.4 (3.9, 14.9)	11.6 (4.9, 18.3)	20.8	intermediate
Sweetgum	34	9.7	0.0 (0.0, 0.0)3	0.5 (0.0, 4.9)	24.1 (2.0, 46.1)			25.2	intermediate
Sugar Maple	559	1.0	1.8 (0.0, 5.5)1	3.5 (0.0, 10.5)	5.2 (0.0, 15.5)	6.9 (0.0, 20.0)	8.6 (0.0, 24.5)	28.7	insensitive
Virginia Pine	644	1.0	1.5 (0.0, 4.7)2	3.1 (0.0, 9.3)	4.5 (0.0, 14.8)			33.1	insensitive
Eastern White Pine	706	1.0	1.4 (0.0, 6.6)2	2.8 (0.0, 13.1)	4.2 (0.0, 19.4)			36.2	insensitive
Chestnut Oak	811	1.0	1.2 (0.0, 5.6)2	2.4 (0.0, 11.1)				41.6	insensitive
Douglas-fir	1002	1.0	1.0 (0.1, 1.9)2	2.0 (0.3,3.7)	2.9 (0.4, 5.5)	3.9 (0.6, 7.3)	4.9 (0.7, 9.0)	51.4	insensitive
Table Mountain Pine	1180	1.0	0.8 (0.0, 4.9)2	1.7 (0.0, 9.7)	2.5 (0.0, 14.5)			60.5	insensitive
Yellow Buckeye			0.0 (0.0, 0.0)4	0.0 (0.0, 0.0)	0.0 (0.0, 0.0)			> 100	insensitive

¹The 95% confidence interval for PRBL was based on the likelihood ratio test and a chi-squared distribution with 1 degree of freedom when a random coefficients regression model was fit to the combined data.

²The 95% confidence interval for PRBL was based on the Wald t-test when plant biomass response to O₃ exposure was linear (i.e., C=1) and no random coefficient terms were included in the regression model. The lower confidence bound was truncated to 0% as the PRBL parameter space ranged between 0% and 100%.

³The 95% confidence interval for PRBL was based on the F-test with 1 and n-p degrees of freedom where n=sample size and p=number of model parameters when a Weibull model with no random coefficients was used to fit the data.

⁴PRBL was set equal to 0% for all values of W126 based on a constant mean exposure-response function.

Figure 9.

Comparison of the predicted relative biomass loss (PRBL) equations for 16 tree species consisting of four conifers (red lines) and 12 deciduous species (black lines) formed three groups of O₃ sensitivity: sensitive – Black cherry, ponderosa pine, quaking aspen, red alder, American sycamore, tulip poplar, and winged sumac; intermediate –red maple, and sweetgum; insensitive – chestnut oak, Douglas-fir, eastern white pine, sugar maple, table mountain pine, Virginia pine, and yellow buckeye.

Discussion

This study is the first to report whole-plant biomass responses of seedlings of a broad sample of tree species to O₃ exposure in OTCs examined using a Weibull model and the three-month 12-h W126 metric (Tables 5, 6; Figure 9). These parameterized relationships are useful for identifying sensitive tree species in support of risk assessments for trees across the U.S. A wide range of O₃ sensitivities have been reported for many North American tree species based on observational studies of visible foliar symptoms and damage (Campbell et al., 2000) and on controlled studies of the effects of O₃, alone or in combination with other stressors (Hogsett et al., 1997; Chappelka and Samuelson 1998; Isebrands et al., 2000). Response of tree growth to O₃ is affected by many factors operating at various scales, from the leaf (e.g., stomatal conductance, internal anatomical features), to the whole plant (e.g., anti-oxidant production, shift in carbon allocation, plant growth strategy), and to the stand scale (e.g., competition, environmental condition) (Bennett et al., 1992; Laurence et al., 1994; Miller et al. 1997; Plochl et al., 2000; Novak et al., 2008). Further, response may vary over the life span of a tree (Matyssek and Sandermann 2003) as trees get larger (Chappelka and Samuelson 1998). Varying O₃ sensitivities reported for OTC studies of the same tree species reflect differences in experimental protocols, O₃ exposure profile and duration, years of exposure, and genotypes (Karnosky et al., 1996; Dickson et al., 1998). Our results show a wide range of O₃ sensitivities from insensitive (chestnut oak, yellow buckeye, sugar maple, Virginia pine, white pine, Douglas-fir, table mountain pine) to intermediate (red maple, sweetgum) to more sensitive (black cherry, ponderosa pine, tulip poplar, quaking aspen, winged sumac, American sycamore, red alder) for exposures up to 2.0 x ambient and W126 values reaching 86 ppm-h. Our study is the first to report on the sensitivities of chestnut oak and yellow buckeye and fills in data gaps for some less-studied species including red alder, American sycamore and winged sumac (Table 5).

We ranked the species from most to least sensitive according to the W126 associated with a 5% reduction in total biomass as follows: black cherry > tulip poplar > ponderosa pine > American sycamore > winged sumac > quaking aspen > red alder > red maple > sweetgum > sugar maple > Virginia pine > eastern white pine > chestnut oak > Douglas-fir > table mountain pine > yellow buckeye (Table 6). Our rankings are fairly consistent with previous studies of the sensitivities of these species based on foliar injury (Campbell et al., 2000; Bell et al., 2020), growth reductions (Karnosky et al., 1996; Neufeld et al., 1995), GIS-based risk characterization (Hogsett et al., 1997), and computer models (Heck and Furiness 2001). Fast-growing deciduous tree species including quaking aspen, black cherry, tulip poplar and winged sumac were among the most O₃-sensitive tree species in our database. Field-grown tulip poplar and black cherry are considered to be some of the most O₃-sensitive tree species in eastern forests, while red maple, loblolly pine (Pinus taeda), and northern red oak (Quercus rubra) are intermediate, and red spruce (Picea rubens) the most tolerant (Heck and Furiness 2001). Tulip poplar with visible foliar injury ranks higher in O₃ sensitivity than black cherry based on stem growth reductions over a 5- or 10-year period in Great Smoky Mountains National Park (Somers et al., 1998). American sycamore (Platanus occidentalis) ranks highest in O₃-induced foliar stipple injury among 28 eastern tree species based on a continuously stirred tank reactor study (Kline et al., 2008). Western tree species including ponderosa pine and red alder were also ranked as sensitive to O₃, consistent with their known sensitivities based on visible foliar injury (Campbell et al., 2000). Ponderosa pine is considered to be one of the most O₃ sensitive tree species in western North America based on visible foliar injury and basal area loss at O₃ concentrations > 60 ppb (Miller et al., 1983; Miller 1996), while Douglas-fir is considered insensitive based on no reported visible foliar injury due to O₃ in the Pacific Northwest (Campbell et al., 2007). Yellow buckeye, eastern white pine, and Virginia pine were ranked as insensitive, and meet National Park Service criteria for sensitive species but not for bioindicators, based on visible foliar injury (Porter 2003).

Quaking aspen is particularly sensitive to O₃ while hybrid poplars (Populus), tulip poplar, and white pine vary in sensitivities (Isebrands et al., 2000; Rebbeck and Scherzer 2002). Quaking aspen (Karnosky et al., 1992a&b, 1996) and eastern white pine (Houston and Stairs 1973) have a high degree of natural genetic variability in response to O₃ and results from OTC studies reflect that variability (Karnosky 1981; Isebrands et al., 2000). High O₃ concentrations have been shown to cause extensive needle damage to white pine in the southeastern US, including premature needle abscission and decreased basal area increment (McLaughlin 1985; Swank and Vose 1991). However, natural selection may have resulted in genetic shifts in populations of eastern white pine (Bennett et al., 1994), red maple and quaking aspen (Karnosky et al., 1989) in eastern North America towards more O₃-tolerant individuals. Reich and Amundsen (1985) considered eastern white pine to be relatively insensitive to O₃ based on a reduction in net assimilation of 10% at the highest O₃ level, consistent with our ranking of eastern white pine as insensitive to 1.5x and 2.0x ambient treatments.

In western forests experiencing a Mediterranean climate and an annual summer drought, the mechanisms by which Douglas-fir populations have adapted to local environmental conditions and have intermediate drought tolerance may explain its low O₃ sensitivity. Douglas-fir is an isohydric plant that maintains a constant midday leaf water potential by reducing stomatal conductance in response to high vapor pressure deficit (VPD) and drought (Meinzer et al., 2016). Stomatal closure of Douglas-fir due to high VPD and drought may limit the effects of O₃ on growth by reducing the rate of pollutant uptake and consequent damage to plant tissue or impairing physiological processes. Similarly, in eastern forests, tulip poplar and sugar maple are isohydric plants that also close their stomates at moderate water stress which may limit the uptake of O₃ during the summer and in dry years (Smith et al., 2003), and be a reason why some species are not more sensitive to O₃ (Roman et al., 2015; Yi et al. 2017). While stomata in Douglas-fir are highly responsive to vapor pressure deficit (Bond and Kavanagh 1999; McDowell et al., 2002; Unsworth et al., 2004), the stomatal response to O₃ is not well understood (Heath 1994; Grulke and Heath 2020). A unique xylem structure trait of slow-growing ponderosa pine that restricts water transport and contributes to its drought resistance may also contribute to its growth response to O₃ exposure but the mechanistic basis of the interacting effects of O₃ and other factors on plant growth are not well understood (Roskilly et al., 2019).

Except for ponderosa pine, conifers were relatively insensitive compared to broadleaves when exposed to 1.0x and 1.5x ambient and W126 values < 20 ppm-h, consistent with other studies (Reich 1987; Chappelka & Samuelson 1998; Wittig et al., 2009). Conifers are considered less O₃ sensitive than broadleaves at least in the short term (1-5 years) (Skärby et al., 1998) because conifers have lower gas exchange and O₃ uptake and higher antioxidant capacity (Paoletti et al., 2020). However, the correlation of sensitivity with the higher stomatal conductance and O₃ uptake of early successional plants, as suggested by Harkov and Brennan (1982) and Bazzaz (1979), is weak at best, and many other factors including anatomical characteristics for deciduous trees (Bennett et al., 1992) or specific leaf area (Wieser et al., 2013) likely interact with uptake to determine species’ sensitivities. More recent evidence suggests that temperate conifers and broadleaves have different optimal stomatal strategies to manage hydraulic risk during periods of low soil moisture and high evaporative demand (Anderegg et al., 2018; Novick et al., 2019).

Reductions in total plant biomass for the O₃-sensitive tree species typically occurred after one year of O₃ exposure and, for ponderosa pine, were significantly greater after two seasons of O₃ exposure (i.e., cumulative O₃ effect). In contrast, evidence of a cumulative O₃ effect after two years of exposure on biomass was weak for Douglas-fir and tulip poplar due partially to the low number of replicate study-harvests; also, tulip poplar seedlings were so large by year two that they were experiencing some water stress despite watering once or twice daily to alleviate such effects. Eastern white pine total plant biomass was marginally affected by ozone following one or two years of exposure. In the longer-term free air canopy ozone exposure (Aspen FACE), the growth response of older aspen trees over 6 years was consistent with single year OTC experiments on aspen seedlings (US EPA, 2013, 2020a). Other seedling fumigation studies have reported no significant reductions in plant growth to elevated O₃ until the second or third growing season for ponderosa pine (Takemoto et al., 1997; Temple and Miller 1994) and Douglas-fir (Runeckles and Wright 1996). A delayed growth response to O₃ of one or more years has also been reported for sugar maple (Scherzer 1991; Karnosky et al., 2005) and tulip poplar (Rebbeck and Loats 1997). Sugar maple has been described as an intermediate to tolerant species (Rebbeck, 1996), and expresses a significant growth reduction when exposed for three years to a 3.0 x ambient O₃ treatment but not at 1.7 x ambient (Topa et al., 2004). Several mechanisms for cumulative or delayed O₃ effects in trees have been identified, including a disrupted transport or storage of carbohydrates to belowground biomass of Douglas-fir (Runeckles and Wright 1996), impaired bud formation leading to reduced early growth of foliage in the next season in white birch (Betula pendula) (Oksanen 2003), and reduced root carbohydrate content and new root growth of ponderosa pine in the spring following a season of ozone exposure (Andersen et al., 1997).

While a single O₃ exposure index developed from controlled OTC studies may not fully characterize every feature of various O₃ regimes affecting vegetation, the 12-h W126 metric is a reasonable compromise for risk assessment, e.g., to evaluate the secondary NAAQS standard for O₃. Moreover, the 12-h W126 metric may provide consistent protection for vegetation exposed to different temporal patterns of O₃ concentrations. Evidence from our exposure dynamics studies of two O₃ sensitive species, quaking aspen and ponderosa pine, indicated that seedlings responded differentially to O₃ exposure by varying the diurnal and seasonal patterns of O₃ concentrations. Growth response to O₃ was significantly greater for the episodic O₃ profile than the daily peak, high elevation, and the 12-hr shifted episodic profiles having lower daytime O₃ concentrations (Figures 5 and 6). Since the time the exposure studies were performed, air quality policies have been implemented that decreased the number and maximum concentration of episodic peaks in the U.S. (US EPA, 2020b). It has been suggested that one way to measure this is to count the daytime hourly O₃ concentrations of 0.10 ppm and higher (N100), because they may play an important role in eliciting a plant response to O₃ exposure for seedlings as well as for crops (Lefohn and Foley 1992; Lefohn et al., 1997). As seen in Tables 3-5, the N100 count decreases at lower W126 exposure levels. The most sensitive species in our analysis had a biomass loss of 5% at a W126 of 2.5-9.2 ppm-hrs and N100 ranged from 0 to 7 at those exposures. However, there were few experimental exposures between 2.5 and 9.2 ppm-hrs. In that same 2.5-9.2 ppm-hrs range, U.S. O₃ monitoring data from 2016 to 2018 indicate that N100 ranged from 0-10, with an N100 of 0 at most monitors (US EPA, 2020b). While the episodic peaks have been declining across the U.S. in the past few decades, there are still areas with N100 peaks where the sensitive species occur. Therefore, we believe the exposure response relationships reported in this study remain relevant to current O₃ exposures, especially for the sensitive species.

Table 3.

Dates of ozone exposure period and harvest, the 12-h W126 (ppm-h) and 12-h number of hourly O₃ concentrations ≥ 0.10 ppm (N100) exposure metrics for the tree species exposed in open-top chambers at Michigan Technological University’s Ford Forestry Center at Alberta, Michigan. The ozone profiles were based on the modified ambient profile from 1987 ozone data supplied by the Michigan Department of Natural Resources for Washtenaw County, Michigan (Karnosky et al. 1996). The 0.5x, 1.5x, and 2.0x profiles were developed by scaling the hourly ozone concentrations using a sigmoid function as described by Hogsett et al (1988). Three replicates were used for each ozone treatment level.

Species	Harvest Date (Years of Exposure)	Exposure Dates	Exposure Duration (days)	W126 values (ppm-h) / N100 for ozone treatments
				CF	0.5x	1.0x	1.5x	2.0x
Eastern White Pine	09/10/1990 (1)	06/20/1990 – 09/10/1990	83	0.0/0.0	5.7/0.8	6.5/6.6	7.0/6.0	22.5/60.2
	09/14/1991 (1)	06/09/1991 – 09/14/1991	98	0.0/0.0	12.7/28.9	16.6/43.0	27.0/55.7	31.5/85.9
Quaking Aspen Clones	09/10/1990 (1)	06/20/1990 – 09/10/1990	83	0.0/0.0	5.7/0.8	6.5/6.6	7.0/6.0	22.5/60.2
Quaking Aspen Clones	09/14/1991 (1)	06/09/1991 – 09/14/1991	98	0.0/0.0	-	16.6/43.0	-	31.5/85.9
Quaking Aspen Seedling	09/14/1991 (1)	06/09/1991 – 09/14/1991	98	0.0/0.0	12.7/28.9	16.6/43.0	27.0/55.7	31.5/85.9
Sugar Maple	09/10/1990 (1)	06/20/1990 – 09/10/1990	83	0.0/0.0	5.7/0.8	6.5/6.6	7.0/6.0	22.5/60.2

Our empirical findings support the use of a 12-h window (8 AM to 8 PM) for an O₃ exposure metric based on statistical fit and predictive ability for the 16 tree species studied. However, in some circumstances, there may not always be concordance between diurnal patterns of ozone concentrations and those for stomatal conductance (Musselman and Minnick 2000). High elevation sites often have ozone concentrations that remain high at night and even peak at these times (Neufeld et al., 2019). Consequently, a 12-h O₃ exposure metric may not produce optimal predictions at high elevations and for some species with meaningful nocturnal stomatal conductance and flux (Musselman and Minnick 2000, Snyder et al., 2003, Grulke et al., 2004, Barbour et al., 2005, Daley and Phillips 2006). Furthermore, some trees may be more sensitive to ozone if exposed at night, perhaps due to reduced detoxification abilities at night (Matyssek et al., 1995) or more sensitive to afternoon O₃ fluxes than morning fluxes due to diurnal changes in stomatal conductance and antioxidant activity (Heath et al., 2009; Wu et al. 2021). However, the plant’s defensive ability to protect from O₃ damage by altering the stomatal conductance and detoxifying O₃ internally might be limited under higher O₃ concentrations (Musselman et al., 2006).

In eastern United States, annual growth losses of 0%-33% (Hogsett et al., 1997), 0-10% (Chappelka and Samuelson 1998) and 4%-12% (Ollinger et al., 1997) are predicted based on O₃ exposure-response functions for trees as seedlings, which may differ for field conditions or large trees (Chappelka and Samuelson 1998; Matyssek and Sandermann 2003). In the San Bernardino Mountains of California, about 30% of bigcone Douglas-fir (Pseudotsuga macrocarpa) experiencing drought and elevated concentrations of ambient O₃ had growth reductions > 50% between 1951 and 1988 (Peterson et al., 1995). Ozone-induced reductions in growth of Jeffrey pine (Pinus jeffreyi) and ponderosa pine in California are more pronounced in older trees (Peterson et al., 1987; Innes 1993). In Europe, forest yield losses are predicted to range from 10% (Broadmeadow 1998) to as high as 42% for beech (Fagus sylvatica) (Pretzsch et al., 2010). The tree species in our database had a comparable range of predicted biomass reductions of 0%-33% at W126 level of 20 ppm-h, indicating that the O₃ effects on tree growth as seedlings may generalize to mature trees in the field within the margins of biological and genetic variability.

Ambient O₃ exposures over decades may be altering the species composition and O₃ sensitivities of forests in some locations but other stressors (e.g., rising CO₂ and drought) may moderate or exacerbate the direct effects of O₃ on tree growth (Panek et al., 2002). The mixed conifer forest of the San Bernardino Mountains of California is often cited as a classic case of ecosystem impairment due to high ambient O₃ exposure (Fenn et al., 1996). Evidence from the OTC studies suggest the potential risk of O₃ damage to ponderosa pine persists given that the effects of O₃ on tree biomass may accumulate over time. Many O₃-sensitive species reside in the southern Appalachian Mountain region, where pollutant levels over the past 15 years in Great Smoky Mountains National Park have decreased to their lowest levels since measurements began (Neufeld et al. 2019), due to implementation of the NO_x control protocols. As a result, current risks for these species are reduced, although not entirely eliminated, compared to past years when ozone exposures were much higher. Rising atmospheric CO₂ levels, which have increased by 18% since 1988 (NOAA 2022), may further reduce risks because they increase resistance to ozone (Isebrands et al., 2001; King et al., 2005; Paoletti and Grulke, 2005). Evidence from early OTC studies and longitudinal field observations had led researchers to suggest that some populations of eastern white pine, red maple, and quaking aspen were shifting towards more O₃ tolerant individuals due to natural selection (Karnosky et al., 1989; Bennett et al., 1994), but with the recent decrease in ambient ozone exposures, we could see the re-appearance of sensitive individuals given the lack of selective pressures currently.

Conclusion

It is well known that exposure to ambient O₃ can decrease growth in many tree species in the US. The sensitivity to O₃ varies by tree species, is confounded with other stressors (e.g., drought, excessive nitrogen deposition), and can be higher for deciduous than conifer species. The potential risk of O₃ damage to tree growth is expected to be higher for eastern forests comprised of fast-growing, deciduous species than western mixed conifer forests.

The primary objective of this study was to establish a reference set of parameters for seedling exposure-response relationships for 16 North American tree species using the 3-month (92 day) 12-hr W126 metric. The sixteen species are widespread across the U.S. (Figure 1), are ecologically important and include a variety of deciduous and coniferous, and fast and slower growing trees. Our study thus reports exposure-response relationships for an assortment of 16 North American tree species from 36 OTC studies in potential support of risk assessment frameworks used by US EPA, National Park Service and other researchers. This study is the first to report whole-plant biomass responses of seedlings of a broad sample of tree species to O₃ exposure in OTCs examined using a Weibull model and the three-month 12-h W126 metric.

When classified by sensitivity, the three-month 12-h W126 estimated range expected to result in a 5% biomass loss was 2.5-9.2 ppm-h for sensitive species, 20.8-25.2 ppm-h for intermediate species, and > 28.7ppm-h for insensitive species. The most sensitive tree species include a number that are ecologically important and widespread nationally, including black cherry, ponderosa pine, quaking aspen, red alder, American sycamore, tulip poplar and winged sumac.

These species-specific exposure-response relationships will allow US Agencies and other groups to calculate the exposure-response relationships for many O₃ exposure metrics of interest and to better estimate biomass losses based on ozone exposures in North America, to be used in risk assessment and scenario analyses. We have therefore made our database available to the public at the US EPA ScienceHub (https://sciencehub.epa.gov/sciencehub/distribution/7295/download).