Second, we tested if measures of individual EFs differed among years, seasons, species richness levels, and plot identities (Fig. 1). Considering all EFs, year explained on average 4.7% of the variation of EFs, and season explained on average 4.1%. Additionally, species richness explained 8.6%, while plot identity explained 21.3% of the variation of individual EFs. These differences explained about one-fourth of the variation of the individual EFs. One-third of the variation in individual EFs was unexplained: 34.1% for EFs, which were measured in several seasons, and 48.3% for EFs measured in just one season. All tested interaction terms explained only a small part of the total variation of the EFs (Fig. 1).
Different variables explained the variation of EFs in different classes of EFs. For example, for invasion resistance and plant productivity, SR explained a large proportion of the variation (on average, 23.1% and 21.0%; Fig. 1, green). For consumer-related functions, year explained a large proportion (on average 11.9%; Fig. 1, red). For plant nutrients, plot identity explained, on average, 38% of the variation (Fig. 1, blue). This means, that classes of EFs were differently affected by biological and environmental conditions (classes of EF: F7,245=4.2, p=<0.01; Driver F12,245=38.35, p<0.01; classes of EF: Driver: F60,245=5.4, p<0.01, where ”driver” represents the year, season, SR, plotID and their interactions).