A Multivariate Approach to the Analysis of Factorial Fertilization Experiments in Alaskan Arctic Tundra

Abstract

Canonical analysis was used to analyze the results of factorial NPK fertilization trials on size species growing in arctic tundra at Eagel Creek, Alaska, USA (latitude 65°10'N, longitude 145°30'W): the graminoids Eriophorum vaginatum and Carex bigelowii, the deciduous shrubs Betula nana and Vaccinium uliginosum, and the evergreen shrubs V. vitis—idaea and Ledum palustre. In an extension of early univariate analyses, multivariate comparisions were made based on shoot concentrations of N, P, K, Ca, and Mg as a measure of overall nutrient balance. In the unfertilized controls, the graminoids were richer in K than the other growth forms, deciduous shrubs had higher concentrations of N relative to P and Ca than the evergreen shrubs, and the deciduous—evergreen contrasts were greatest early in the growing season. An NKP fertilization triggered disproportionate increases in K concentration in the graminoids and increases N relative to P and Ca in the evergreen shrubs. The effect of fertilization on nutrient balance of the deciduous shrubs in the spring following fertilization was like that of the evergreen shrubs but by late summer had shifted to the graminoid pattern. Nitrogen and phosphorus consistently produced contrasing shifts in the nutrient balance in all three growth forms. In contrast to the two Vaccinium shrubs, the nutrient balance of the graminoid E. vaginatum was strongly affected by any fertilization involving potassium. In general, factorial combinations of N, P, and/or K produced shifts in the nutrient balance qualitatively intermediate to those resulting from the nutrients added singly. Many differences among growth forms apparent here were not discerned in the earlier univariate analyses of these fertilization trials. Particularly when combined with factorial fertilization experiments, canonical analysis can provide considerable insight into the response of nutrient balance to fertilization and help elucidate the ecological significance of these response patterns.