Vertical differentiation in root placement is one of the potential mechanisms of plant niche differentiation. It can be due to the remarkable plasticity of roots in response to nutrients and neighbours, but most data on it come from pot or garden experiments. The roles of vertical differentiation and of plasticity in it in the field are thus not well known. We examined species‐specific root vertical distribution in a montane grassland using quantitative real‐time PCR. We asked whether individual species differ in their rooting depths, whether such differences are associated with above‐ground functional traits (such as height or specific leaf area), and whether they respond to the presence of a competitor. This response was assessed by comparison of species‐specific vertical profiles between control plots and plots where the dominant species, Festuca rubra, had been removed. Vertical profiles of individual species varied considerably, from species with most root biomass concentrated in the uppermost (<2 cm) soil layer, through species with uniform vertical distribution, to a species with roots predominantly below 8 cm (Nardus stricta). Species at the fast end of the plant economy spectrum were more likely to place their roots in the uppermost layers. Grassland species, thus, exploit different parts of the below‐ground resources in spite of their short stature, minor differences in height above‐ground and shallow soil. While below‐ground and above‐ground biomasses of most species were higher in the removal plots, species rooting patterns did not change in response to the removal. The interspecific differences in vertical profiles were thus due to species' innate differences, not to plastic responses to the presence of the dominant species. Synthesis. The findings imply that vertical root differentiation in the field is strong and can contribute to niche differentiation. However, the role of root plasticity in natural systems may be considerably weaker than in artificial systems with few species and strong nutrient gradients. This absence of the plastic response in the field is likely to be due to a fairly homogeneous distribution of nutrients in the soil and to the predominantly symmetric nature of below‐ground competition.

Summary Several dynamic models have shown that dynamics of legumes and grasses can result in periodic behaviour. These oscillations arise due to delays in nitrogen flows coupled with differences in ability to compete for light. However, long‐term time series on legume dynamics that could be used to test predictions of these models are almost non‐existent. We examine legume oscillations in a semi‐natural mountain grassland using a long‐term (Tilde; 30 years) data series on aboveground biomass of individual species and on nitrogen and phosphorus content over time. Using autocorrelation analysis, we show that there is a strong periodicity (with period of 8–9 years) of legume and grass biomass and nitrogen content in the grass biomass. These three variables are in fairly stable phase shifts relative to each other, with a grass peak followed by a peak in C : N ratio in grasses which is followed by a legume peak. Phosphorus content in either legume or grass biomass does not show synchronous cycling with legume or grass biomass or nitrogen content in grass. Fitting a dynamic linear model to the data showed that legumes affect nitrogen content in grasses, and grass biomass both affects and is affected by nitrogen content. In contrast, there is no negative effect of grasses on legumes, indicating some other process must be responsible for the legume decline. Manuring, which was occasionally applied to the plots, also does not seem to affect the cycling. Second‐order term for legumes showed some evidence of self‐inhibitory effects in legumes, but phosphorus content in legumes shows no support for phosphorus limitation. The most likely explanation of the legume decline should be sought elsewhere (pathogens, soil biota etc.). Synthesis. Long‐term data support the existing the claim that legume dynamics are the key driver of nitrogen dynamics in nutrient‐poor semi‐natural grasslands. Grasses benefit from the nutrient enrichment due to legume cycling, but are a passive element and do not play a role in legume limitation. Apart from the role of nutrient cycling, these legume‐driven nutrient dynamics also constitutes processes by which long‐term richness of meadows is maintained.