Relative importance of factors controlling the success of Oxalis acetosella: an example of linear modelling in ecological research

An ecological hypothesis concerning the relative importance of factors governing the success of Oxalis acetosella on mesic upland forest sites in southern Finland was formulated and tested statistically by means of a multivariate linear model. The data consisted of a stratified random sample of 40 plots with biological measurements and associated observations on environmental variables. The covariance matrix between the incorporated variables was computed and the model parameters were estimated using the method of maximum likelihood provided by the computer programme LISREL V. The results supported the hypothesis based on previous observations and experiments: Oxalis acetosella benefits from a high nutrient level of the soil, but is decisively dependent on the shelter provided by the tree stand. When both spruce stands and well-lit pine stands on mesic and relatively rich forest sites were considered simultaneously, only a weak correlation was found between light availability and site fertility. Accordingly, the presumed value of Oxalis acetosella as an indicator of site fertility appears to be questionable. The adequacy of linear modelling in an ecological context is discussed. The kind of models applied in this study have only a limited application range in non-experimental ecological research. However, linear modelling may contribute to solving particular ecological problems in cases where short environmental gradients are considered so that nonlinearity is not a dominant feature.