Soil, site and management components of variation in species composition of agricultural grasslands in western Norway

A partitioning of the total variance in species composition of grasslands associated with increasing fertilizer inputs (unfertilized pastures, artificially fertilized hay meadows and intensively cultivated grassland) in western Norway was undertaken. The partitioning was carried out with (partial) constrained ordinations (canonical correspondence analysis) and associated Monte Carlo permutation tests. Explained variation was high (0·651), with soil chemistry, management and site explaining 0·271, 0·228 and 0·052 of the variation, respectively, and the interaction between soil and management explaining 0·100 of the variation. However, much of the measured soil chemistry was considered to be an effect of management. The soil chemistry variable explaining most variation was extractable P content, associated with high soil extractable P contents in the intensively cultivated grassland due to high fertilizer applications. However, soil extractable P content did not explain differences in species composition when grasslands with smaller differences in fertilizer inputs (meadows vs. pastures) were compared. Total soil C and N contents and C:N ratios explained significant variation between all grassland types. Lowest levels of these variables were measured in the grassland with the highest fertilizer inputs, suggesting that total soil N content cannot be used as a predictor of fertility. Significantly higher soil N contents were measured in grasslands of low productivity, probably due to a low decomposition rate of stress‐tolerant plants and the consequent low availability of soil N and accumulation of soil organic matter. Ellenberg N values reflected soil chemical differences in this study with high Ellenberg N values reflecting high Ca, Mg and P contents and pH values and low total C and N contents and C:N ratios.