Answers are grounded in this paper and cite their sources.
Ants are widely recognized for reducing herbivory, yet how these effects translate into benefits for plant reproductive performance remains unclear. This knowledge gap largely reflects the limited number of studies that have simultaneously quantified ant activity, herbivore abundance, and multiple components of plant reproduction, leaving the mechanisms underlying ant-mediated effects unresolved. To address this gap, we conducted a field experiment in a dry tropical forest ecosystem in Brazil (the Cerrado), where we experimentally manipulated ant presence on the extrafloral nectar-bearing woody plant Senna rugosa (Fabaceae). We monitored ant activity, herbivore abundance, and several reproductive traits—including bud, flower, and seed production, as well as seed weight—throughout a full growing season. Using structural equation modelling, we further evaluated direct associations between ants and plant reproductive traits and indirect pathways mediated through herbivore abundance. Ant exclusion effectively reduced ant abundance, which was accompanied by reductions in bud and flower production and, unexpectedly, by a decrease in herbivore abundance. However, seed number and seed weight did not differ between ant treatments. The structural equation model revealed positive direct associations between ant presence and bud production, which was subsequently positively associated with both flower and caterpillar abundance, whereas the direct association between ants and herbivore abundance was not significant. Ant presence was also positively associated with seed number and seed weight. However, we found no evidence that the effects of ants on plant reproductive traits were mediated by changes in herbivore abundance. Overall, our findings highlight the complexity of ant–plant interactions and suggest that associations between ants and plant reproductive performance cannot be explained solely by changes in herbivore abundance, while the mechanisms underlying these associations remain to be determined.
No takes yet. Share an insight, caveat, or question.
De‐Freitas et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: