Table 4 Statistical results for the analysis of VOC compound group emissions from experimental plant communities grown in pots in the greenhouse (Fig. 2). The table shows effects of species richness (1 to 3 plant species), species composition (four levels, Trifolium pratense monoculture (1sp), two species mixture of T. pratense and Geranium pratense, two species mixture of T. pratense and Dactylis glomerata and the three species mixture containing T. pratense, G. pratense and D. glomerata) and species identity (with the effect of D. glomerata or G. pratense presence shown in separate columns) on the VOC emission from experimental plant communities. Statistical analyses were performed with absolute amounts (ng g^− 1 h^− 1) and relative amounts (% of the total VOC emission) of VOCs emitted. The first column in both panels (A and B) denotes the statistical results for the effect of caterpillar herbivory. As the effect of herbivory was the same in these models (where it was always fitted first), the statistical result is only presented once. Interactions between species richness, species composition and herbivory (Spodoptera littoralis caterpillar herbivory in one individual of the focal plant species T. pratense in each community) are reported in the column “interactions”. Species richness, species composition and species identity (presence or absence of a species) were tested in separate analysis of variance models following transformation of data to meet assumption of normality (see text for details). Statistically significant results are depicted in bold with asterisks indicating level of significance (***P < 0.001, **P < 0.01 and *P ≤ 0.05) and degrees of freedom are provided at the top of the table as F (df1, df2), n = 5. See Additional file 3: Table S5 for full models

From: Plant volatile emission depends on the species composition of the neighboring plant community