Ants on flowers: Protective ants impose a low but variable cost to pollination, moderated by location of extrafloral nectaries and type of flower visitors

Throughout their lifetimes, individuals of most species engage in multiple mutualistic interactions simultaneously. These mutualisms can interfere with each other, leading to ecological costs that can influence the outcome of these interactions. Identifying factors that modulate these costs is pivotal to reaching a deeper understanding of the dynamics of mutualistic interactions. We investigated the ecological costs of protective ants on the pollination of plants with extrafloral nectaries (EFNs) and how plant traits and the type of pollinator can modulate them. We hypothesized that (1) ants negatively affect the patterns of flower visitation and plant reproductive performance and that (2) this effect is relatively higher for plants bearing EFNs close to inflorescences compared to those bearing EFNs only on vegetative tissues. Additionally, we hypothesized that (3) the effect of ants on flower visitation is contingent upon the type of flower visitor. To evaluate these hypotheses, we used a meta-analytical approach based on 27 studies that assessed the effects of ants on patterns of floral visitation and plant reproductive performance. We found that ants impose a low, yet highly variable, cost to pollination. Specifically, ants have a strong negative effect on patterns of flower visitation, especially when the visitors are bees, but this does not translate into reduced plant reproductive performance. The reproductive success of plants with EFNs close to vegetative tissue is enhanced by ant presence. Finally, the negative effect of ants on flower visitors is stronger when EFNs are closer to inflorescences. Synthesis: Our results indicate that, in general, ants have a low impact on the pollination of EFN-bearing plants. This pattern supports theoretical predictions that natural selection acts to lower the costs of protection mutualisms. Our macroscale evaluation further shows that plant traits such as EFN location, as well as the type of flower visitors, modulate this cost. The outcome of ant-pollinator interactions may affect the evolutionary trajectory of traits mediating defense and pollination in EFN-bearing plants.

Methods
Description of methods for data collection: we used keywords (desribed above) to find papers about the effect of ants on pollination on Web of Science and Scopus.

Keywords used to search studies: (“(ant-plant OR myrmecoph*) AND (pollina* OR robbe* OR thie* OR larcen*)”. Based on this query, we used litsearchR R package to find more queries: (1) “(extrafloral nectar*) AND (pollina* OR robbe* OR thie* OR larcen*)”, (2) “(ant-pollinator conflict*) AND (pollina* OR robbe* OR thie* OR larcen*)”, and (3) “(ant-plant mutualis* OR ant-plant interact*) AND (pollina* OR robbe* OR thie* OR larcen*)”.

Articles included in the meta-analysis: 28

Description of methods used for data processing: our searches resulted in approximately 1200 papers. We used litsearchR v. 0.4.1 to remove duplicates (N = 635) and we removed manually more 32 duplicated not detected by litsearchR. To be included in the meta-analysis, the paper must satisfy six eligibility criteria: (1) have treatments with and without ants, (2) there must be a spatial and temporal overlap between ant and flower visitors foraging, (3) experimental or observational study (no theoretical modelling), (4) if the paper evaluated density or abundance of ants, it must report the effect of ant regardless of density or abundance, (5) have no insects secreting honeydew in addition to extrafloral nectaries, (6) plants should bear only extrafloral nectaries as a source of attraction for ants (no housing).