Figure 1.
Site-pair comparison and interaction specific site-pair combinations.
A) Site-pair comparison. Site 1 and 2 each have unique plants and pollinators. The central square represents the interaction-matrix between shared species. Here, six interactions are present in both sites (interaction consistency, filled squares) and six interactions are only observed in one of the two sites (interaction turnover, open squares). Unique species to either site 1 or 2 were discarded and only the central matrix was used for analysing the turnover of pairwise interactions. B) Interaction specific site-pair combinations. This (hypothetical) interaction is observed at sites 1 and 6 (filled squares) while the species pair is also present at site 3, however without interacting (open square). One or both species are absent from the remaining sites (in grey) and they are excluded from the analysis for this particular interaction. Three site-pair combinations are possible in this case; 1↔3 and 6↔3: interaction turnover and 1↔6: no interaction turnover (interaction consistency).
Figure 2.
Beta diversity of species and interactions and geographical distance.
Total interaction beta diversity βWN (black), plant beta diversity βPlants (green), beta diversity of interactions between shared species βOS (red), and pollinator beta diversity βPollinators (blue) as a function of geographical distance between sites. All measures relate positively to geographical distance. Only for βPollinators the correlation was non-significant (βWN: P = 0.002, R2 = 0.40, α = 0.02; βOS: P = 0.047, R2 = 0.19, α = 0.03; βPlants: P = 0.004, R2 = 0.37, α = 0.04; βPollinators: P = 0.086, R2 = 0.15, α = 0.01). Shaded areas delimit corresponding 95% confidence intervals.
Figure 3.
Interaction frequency and interaction turnover.
Average interaction frequency is negatively related to the probability of interaction turnover. The more frequent interactions show lower probabilities of turnover between sites. Superimposed points result in darker marks.
Figure 4.
Flower abundance and the probability of losing an interaction.
Whether pairwise interactions are realized or not is highly dependent on local flower abundance of the plant species. The figure shows the increasing probability of losing an interaction when going from a site with higher local flower abundance of the plant species to a site with lower local flower abundance. The larger the difference in local flower abundance, the stronger the effect. Superimposed points result in darker marks.
Table 1.
Model terms and their estimates.