Ant–hemipteran mutualisms can have positive and negative effects on host plants depending on the level of hemipteran infestation and plant protection conferred by ants against folivory. Differential effects of such mutualisms on plant survival are well documented in undisturbed and ant‐invaded systems, but few have explored how anthropogenic disturbance affects interactions between hemipterans and native ant species and what the consequences may be for recovering ecosystems. Within a fragmented landscape in Costa Rica, restored tropical forests harbor a mutualism between the native ant Wasmannia auropunctata and the scale insect Alecanochiton marquesi on the abundant, early‐successional tree Conostegia xalapensis. I added A. marquesiscales to C. xalapensis seedlings and either allowed or excluded W. auropunctata to investigate if this mutualism leads to increased scale infestation, decreased scale mortality, and decreased folivory. I also examined whether these effects are mediated by the percentage of remnant forest cover in the landscape. I found that seedlings with ants excluded had fewer scale insects and higher herbivory than plants with ants present. I also found evidence that scale mortality due to fungal attack and parasitism was higher on ant‐excluded versus ant‐allowed seedlings but only at sites with high surrounding landscape forest cover. Together, these results suggest that mutualisms between scale insects and native ants can promote scale infestation, reduce folivory on native plant species, and potentially disrupt biological control of scale insects in recovering tropical forests. Further, my experiment underscores the importance of remnant tropical forests as sources of biological control in anthropogenically disturbed landscapes.

Data collected includes abundance data for A. marquesi in both March 2018 when the experiment began and June 2018 at the end of the experiment. Data also included mortality of A. marquesi due to fungi and parasitism, W. auropunctata abundance, and change in percent folivory on leaves of C. xalapensis after the end of the experiment in June 2018. Percent landscape forest cover at buffers out to 650 m are also included in the dataset.