Figure 2

Three examples of non-parametric relationships (visualized here with Generalized Additive Models) between ranked contributions of species scores to biodiversity values. Species are ranked from common (low rank) to rare (high rank), low occurrence in poems (low rank) to high occurrence in poems (high rank), and from negative trends (low rank) to positive trends (high rank). Most bivariate relationships investigated were weak and not statistically significant. However, rare species scored low for poetry (P = 0.0026). Rare seed-eating species also tended to score lower for weed seed predation. The hump in the population trends vs seed predation figure shows that some species scoring highest for pest regulation tended to be ranked intermediately for trends (i.e. relatively stable populations or increasing populations). All bivariate relationships between measured biodiversity values are presented in the Extended Results Fig. 2. Relationships including weed seed predation are restricted to a subset including only seed-eating species.