Preparation of ant colonies
We collected seven colonies of Diacamma sp. by using bamboo traps (a piece of shoot about 30 cm in length and 5 cm in diameter) in Sueyoshi Park, Okinawa Island, Japan (26°23′N, 127°72′E), from September to early December 2017. These bamboo tubes make good nests and are regularly chosen by Diacamma sp. colonies (Fukumoto 1983). After rain, Diacamma sp. colonies often relocate to another suitable nest. So, by placing bamboo traps on the ground, colonies would relocate into them after each rain, which allowed us to collect whole colonies. We brought the colonies into the laboratory and marked all workers and gamergates (functional queens) individually with a combination of three dots from an oil paint marker (Mitsubishi Pencil Co., Ltd., Tokyo, Japan), one dot each on the dorsal regions of the anterior thorax, metathorax, and petiole (six different colors used in total). After marking, each colony was transferred to a plastic container (44 cm × 30 cm × 7.5 cm high) with a bamboo tube as a nest. They were left for 2–3 days in a climate-controlled room (25 ± 2 °C; 14L:10D) to acclimate to the bamboo nest. After we measured the size of each ant colony, it was transported back to Sueyoshi Park. Each bamboo nest with an ant colony was buried in the ground in an area of the park where other Diacamma sp. colonies were present. The bamboo tubes were covered completely with soil except for the entrance of the tube.
Behavioral observations
Two to three days after the bamboo nests were placed in the field, we observed the behavior of the marked workers outside of the nests during the daytime. We tracked individual marked workers that came out from the nest entrance. We recorded their behavioral responses when encountering non-nestmates, and the distance from the nest where the encounter occurred. An encounter was defined as a marked worker touching a non-nestmate with its antennae. The behavioral responses were designated as attacks or non-attacks to evaluate the aggressiveness of marked workers. We recorded a behavioral response as an attack when a marked worker bit or chased a non-nestmate and as a non-attack when a marked worker just touched or escaped the non-nestmate. We observed encounter events for more than 10 marked workers per colony, and 1 to 5 times per marked worker. In total, we recorded 220 behavioral responses of 101 workers from seven colonies.
During these observations, to investigate the foraging ranges of ant workers, we placed pegs (~ 10 cm in length, 0.5 cm in diameter) into the tracks of each marked worker on the ground until the ant went back to the nest. We then recorded the linear distance from the nest opening to the farthest peg. We tracked over 10 marked workers per colony, and 111 individuals of the seven colonies in total.
Behavioral response toward presented non-nestmates
To complement these observations, we experimentally forced encounter events in the field. We collected Diacamma sp. workers from the other colonies nesting within a 3-m radius from the marked colony by using tweezers that had been washed with acetone. The collected workers were tied with a polyester thread (~ 30 cm in length, 0.12 mm in diameter) around their petioles under hypothermic anesthesia (hereafter an “attached worker”). After recovering from the anesthesia, the attached workers were then used for the experiment.
We presented attached workers to marked workers outside of the marked worker’s nest every 2–10 min; we brought down an attached worker softly ahead of a walking marked worker. We then recorded behavioral responses of marked workers when touching the attached workers and distance from the nest, as described above. Although attached workers were used for the test repeatedly, if an attached worker was attacked by a marked worker during an encounter, it was replaced with another individual from the same alien colony for the following tests. Ten marked workers were haphazardly selected per colony and were presented with attached workers 10 times per individual. This test was conducted for six marked colonies. A colony that had 70 workers was not used (see Results), because it moved out of our study site. In total, we recorded 600 behavioral responses from 60 individuals.
Statistical analyses
To test whether ant aggressiveness was affected by distance from the nest and colony size, behavioral response data were analyzed by a generalized linear mixed model (GLMM) with a binomial distribution and logit link function, using behavioral responses (i.e., attack or non-attack) as the response variable, distance from the nest, colony size, and the distance × colony size interaction as the fixed effects, and individual as the random effect. The significance of the effects of distance from the nest and colony size was tested using a chi-square likelihood ratio test. Likewise, in the forced encounters, the influence of distance from the nest and colony size on behavioral responses were analyzed by using a GLMM and chi-square likelihood ratio test. The statistical analyses were performed using R ver. 3.4.2 (R Developmental Core Team 2017).