The experiment was a randomized block design with three replicatesof the following treatments: (1) charcoal-filtered air (CF); (2)1 × ambient O3 (1 × O3);(3) 1·5× ambient O3 (1·5 × O3);(4) 1·5 × ambient O3 plus350 p.p.m. CO2 above ambient (target CO2 concentrationwas 700 p.p.m) (1·5 × O3 + CO2);and (5) open-air chamberless plot (OA). Due to the prohibitive costsassociated with supplying twice ambient CO2 to threeadditional open-top chambers, we did not characterize the effectsof elevated CO2 alone on yellow-poplar growth. Gaseswere dispensed 24 h per day from mid-May until mid-Octoberin 1992–96. The polyvinyl-chloride film panels were removedeach season when gas exposures were terminated. In spring 1994,standard OTCs (3 m in diameter, 2·4 mhigh) were extended to a height of 4·6 m. Themean volumetric airflow within chambers was 47 m3 min−1 ± 10% andwas provided by a single blower. In spring 1995, standard OTCs werereplaced with larger open-top chambers (4·6 min diameter, 9·2 m high) similar to those developedby Heagle et al. (1989),but doubled in height. Air was blown into the bottom of each chambervia two 2HP, 91·4-cm-diameter fans (Penn Ventilator Inc.,Philadelphia, PA, USA), each housed in a galvanized sheet-metalbox (123 cm wide, 123 cm high, 123 cm long).Unfiltered ambient air entered each chamber after passing throughparticulate filters and connecting plastic ducts at approximately6·8 m3 s−1.
Ozone was generated from vaporized liquid oxygen with an electricspark discharge O3 generator (OREC Model 03V10-0; OREC,Glendale, AZ, USA) and was dispensed into the 1×O3 and1·5 × O3 OTC's throughTeflon tubing when ambient levels exceeded 0·03 p.p.m.(Rebbeck 1996a; Rebbeck& Loats 1997). Gaseous CO2 was vaporized from liquidCO2 (14 000 kg reservoir, MG Industries,Malvern, PA, USA) and dispensed through Teflon tubing into the OTCsand adjusted manually with needle valves (Part 8513D-2-E-4E-1 A,Brooks Instrument Division, Emerson Electric, Hatfield, PA, USA)to maintain target levels. The O3 concentration was regulatedwith mass-flow control valves (Model FC260, Tylan General, San Diego,CA, USA) through a microcomputer to match a set-point value basedon the most recent ambient O3 reading. Each chamber wasmonitored for O3 and CO2, and air temperature (subsetof chambers) every 2 min. Hourly means were automaticallycalculated and stored on a personal computer. Seasonal average O3 andCO2 concentrations were calculated for each chamber/plot.However, seasonal average concentration often does not correlatewith plant injury because it does not include important exposurefactors such as episodic peaks and the cumulative effects of lowO3 concentrations. To better relate plant responses toO3 treatments, cumulative O3 exposures foreach chamber/plot were estimated (Bortier et al.2000). Two cumulative exposure indices were calculated eachgrowing season: SUM00 (p.p.m. × h)that is equal to the sum of all hourly average O3 concentrations;and SUM06 (p.p.m. × h) isequal to the sum of all hourly concentrations above 0·059 p.p.m.Daily single-point and weekly multi-point calibrations of the O3 monitor(Model 49PS; Thermo Electron Instruments, Hopkinton, MA,USA) were made with a multi-gas calibrator (Model 8500;Monitor Laboratories, San Diego, CA, USA). CO2 levelswere monitored with a Li-Cor Model 6251 infrared gas analyser (Li-Cor Inc., Lincoln, NE,USA) with daily single-point calibrations and zero span checks,and weekly multi-point calibrations using certified CO2 spangases (AGA Industries, Cleveland, OH, USA). Ambient photosyntheticphoton fluence rate (PPFR) over the waveband 400–700 nm(Li-Cor quantum sensor), relative humidity (%RH) and airtemperature [Model XN217 (Hydrometrix RH sensor and Fenwal ElectronicsUUT51J1 thermistor) Campbell Scientific, Logun, UT, USA] weremonitored with a Campbell 21X datalogger.