Soil depth and surrounding vegetation biomass were included as covariates in the analyses. As effect sizes of the covariates were small and effects often restricted to one climate treatment, anova results (with post hoc pairwise Tukey’s HSD tests) are shown for the main factors alongside ancova results for the covariates. In cases where data did not meet assumptions of normality or homogeneity of variances, logarithmic, square root or arcsine (mortality) transformation improved data structure satisfactorily. After 15 years of climate manipulation, above-ground biomass production of the resident vegetation was not affected by WL (GAM, P > 0.7; Fig. 2), but decreased y7 in response to DL and WDL, especially in deep subplots (GAM, P < 0.01). Above-ground biomass production was positively related to soil depth in controls and communities subjected to WL (GAM, P < 0.01 and P  0.3). In the WDS treatment, above-ground biomass production was negatively affected in deep but not in shallow subplots (GAM, P < 0.05). Seedling emergence rate between November 2008 and October 2009 of B. erectus was higher in WL and WDL than in DL communities but did not differ from controls (Fig. 3a and Table 1). Similar results occurred in B. pinnatum seeds, but in contrast to Bromus, seedling emergence did not increase in response to WDL (Fig. 3b). In both species, seedling emergence rates did not differ between seed provenances and were not affected by the amount of surrounding vegetation. In WL and DL communities, seedling emergence of Bromus slightly decreased with increasing soil depth (Table 1; see Fig. S2), although this could also have resulted from very early seedling mortality. Overall, effects of climate manipulations on seedling emergence rates were rather small, but winter warming induced earlier emergence in both species. Between November 2008 and April 2009, seedlings emerged from 62.7% (±0.03 SE) of the Bromus seeds but only from 12.9% (±0.02) of the Brachypodium seeds sown in the controls. In the same period, seedlings emerged from 86.3% (±0.02) of the Bromus seeds and 68.5% (±0.03) of the Brachypodium seeds in communities subjected to WL (WDL 86.6 ± 0.02 and 62.7 ± 0.03 respectively). This shift in the timing of seedling emergence as a result of winter warming led to considerable differences in seedling biomass at the end of the first growing season: in both species and all treatments, seedlings that emerged between November and April (winter cohort) had c. 1.5 times more above-ground biomass than seedlings that emerged between May and June (spring cohort; split-split plot anova of two cohorts in 117 and 113 subplots, respectively; Bromus: F1,199 = 157.6, P < 0.