ddi12045-sup-0001-TableS1-S5.3-FigureS5.1-AppendixS5.docxWord document65K

Table S1 Actual (2003) and past (1956) patch area and area of grassland habitat in the surrounding landscape (landscape area), proportion of patch area lost between 1956 and 2003, mean annual precipitation (mm), mean annual air temperature (°C) and altitude (m a.s.l.).

Table S2 Plant species richness (total, specialist and non-specialist) at a (1) Quadrat: individual 0.25-m2 quadrats (mean ± standard error of the mean); (2) All quadrats: pooling the 25 quadrats in each patch; (3) Time transect: additional species detected in the time transect; and (4) Patch: Sum of the pooled the quadrats and the additional species from the time transect.

Table S3 Basic statistical description (Mean ± SD) of the differences between stable and regressive patches with mean annual temperature, elevation, mean annual precipitation, geographical gradient, area 2003, area 1956, landscape area 2003 and landscape area 1956.

Table S4 Plant species recorded in this study. Grassland specialists, marked in bold, were selected with the help of regional floras (Bolòs et al., 1993; Rivas–Martínez et al., 2001) and expert advice.

Table S5.1 Results from GLMs for the effects of patch area, patch stability and geographical variation on total, specialist and non-specialist species richness, at patch and quadrat scales.

Table S5.2 Comparison of the importance of actual or past patch area in predicting specialist species richness in models considering all patches together, stable patches and regressive patches. Significance codes: *** P < 0.001; ** 0.01 > P > 0.001; *0.05 P > 0.01.

Table S5.3 Correlation between residuals of the observed and predicted specialist species richness (calculated from the current patch area using the species-area relationship of stable patches) with the proportion of area lost, past patch area and past habitat area in the landscape, in stable and regressive patches.

Figure S5.1 Relationship between the proportion of patch area lost in regressive patches and the residual between observed and predicted specialist species richness, calculated from the current patch area using the species-area relationship of stable patches. The line is the least-squares linear regression. (r2 = 0.57, P = 0.007).

Appendix S5 Results removing from the analyses a set of stable (patches 5, 6, 10 and 11) and regressive (7 and 9) patches to minimize differences in patch size distributions between the two considered groups (stable and regressive).

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.