Effect of tree stand and site variables on alpha diversity of ground vegetation in the forests of Northern Karelia

Generalised linear models (GLMs) were used to estimate the alpha diversity of the ground vegetation of boreal managed forests. Alpha diversity was des cribed using Simpson's reciprocal, which was employed as a response variabl e in modelling, the explanatory variables being the tree and site character istics The data were from Northern Karelia, Eastern Finland. Several combin ations of error and link functions were used to examine the patterns in the data, the best results being obtained with a model containing the normal e rror and identity link function, and the second power of the explanatory va riable and one containing the gamma error and reciprocal link function. Sit e type, the presence of pubescent birch and the basal area of the stand wer e the significant variables in these models. It was concluded that GLMs are suitable tools for estimating biodiversity w ithin the scope of the present data. In addition, it was found that the hig hest ground vegetation diversity exists within fertile stands with a mixtur e of birch and a low stand basal area. Models of this kind would be useful in forest management planning in order to study the effect of management on vegetational diversity using the values of different tree stand and site v ariables before and after treatment. (C) 2000 Academic Press.