Document: BAH-3-91-5

Use of response surface methodology and ANOVA in detecting air-pollution effects on net photosynthesis under varying temperature and light conditions.

MOMEN, B.* ¹, P.D.ANDERSON ² and J.A.HELMS ³

University of Maryland, College Park, MD 20742 USA ¹
USDA Forest Service, Rhinelander, WI 54501 USA ²
University of California, Berkeley, CA 94720 USA ³

Abstract:
Studies of air pollution effects on plant ecophysiology under changing microclimate are complex requiring efficient statistical design and proper analysis to assure feasibility and correct results. We quantified effects of rain acidity and ozone on net photosynthesis (A_n) of ponderosa pine seedlings under varying temperature and light conditions. Seedlings were exposed to combinations of rain with pH 5.1 and 3, and ambient and twice ambient ozone regimes for 14 months. Subsequently, A_n of current and one-yr-old foliage was measured under nine combinations of three temperatures (18, 25, 32 ^oC) and three light regimes (250, 500, 1000 mol photons m^-2 s^-1). The design was a CRD replicated twice with regard to rain-by-ozone combinations (main plots). Foliage classes were sub plots within main plots, Temperature levels were sub plots within foliage classes, and light levels were sub plots within temperature levels. This allowed the reduction of experimental units to a feasible number of eight, but resulted in repeated measurements for some factors requiring identification and assignment of proper error terms to avoid pseudoreplication. Although repeated-measures ANOVA (multiway) or split-plot ANOVA (univariate) are suitable, identification and assignment of proper errors are not intuitive when multiple repeated measures exist. We used a response surface methodology consisting of least square regression, canonical, and ridge analyses to construct quadratic, polynomial response surface models of A_n response to temperature and light levels per rain-by-ozone combination. temp, temp², light, A_n mean, A_n max, Temp. at A_n max, and Light at A_n max were calculated or estimated. The effects of rain acidity and ozone on these responses were then quantified using a two-way ANOVA, which did not require identification and assignment of any error term beyond the default mean square of error. Rain-by-ozone or ozone effect was not significant on any response. Rain with pH 3 decreased A_n mean and A_n max significantly in current-yr foliage only.

Keywords: ResponseSurface Methodology, Repeated Measures ANOVA, Air-pollution Effects, Ponderosa pine

This abstract is being presented at: 3:45 PM in session:
Oral Session #46: Modeling Populations and Statistical Ecology.