Ant-snatching assassin bugs carry a ‘backpack’ of ant corpses as an antipredator strategy. From photographs, we quantified the relative size and number of ants in these backpacks. We found a trade-off between size and number of carried ants, suggesting that trophic allometry has implications beyond energy acquisition, potentially affecting camouflage.

We searched for photographs of Acanthaspis and Inara individuals in online public image repositories (Deviant Art, Flickr, iNaturalist; search term: “assassin bug”) and selected those (i) in high resolution (i.e., enough to precisely visualize the extremities of the bug and ants) and (ii) that showed the assassin bugs in a side view (lateral photograph). In each selected photo, we quantified our two operational variables: PPSR and the number of prey. As PPSR is dimensionless, we measured the size of the predator and the prey in pixels using the ImageJ software since in the photographs there was no scale to know their real size. The predator size was measured by its body length (from the tip of the rostrum to the tip of the abdomen). We assumed that all ants on a given backpack have the same body size. We also used the head length (from the mid-point of the anterior clypeal margin to the mid-point of the posterior margin) as a proxy for prey size. In all selected photographs, at least one head was in a front view and, if there was more than one ant in this position, we calculated the mean head length.

We used a mathematical approximation to estimate the number of ants present in the predator’s backpack as it was not possible to count them precisely from the photographs. First, we calculated the volume of a single ant based on Tschinkel (2013), which indicates that the gaster volume of Solenopsis represents approximately 57% of its total body volume. We used this proportion for all ants, because this fine allometric characterization has rarely been estimated for other genera. Then, considering the ant's gaster as a spheroid (i.e., an ellipsoid with two equal semi-diameters) (Tschinkel, 2013), we measured its length (GL) and width (GW), and used the formula of spheroid volume to quantify it. We also considered the predator’s backpack as a spheroid. To calculate the total volume of the backpack, we measured its length (BL) and height (BH), and used the same formula of spheroid volume. After calculating the two volumes (of the backpack and of an ant), we used the maximally random jammed (MRJ) parameter of 0.637 (Donev et al., 2004) to estimate the percentage of the backpack volume occupied by ants. Then, we divided the resulting occupied volume by the volume of an individual ant to estimate the total number of ants.

To test the relationship between the number of carried ant carcasses and the relative size of predators and their prey, we fitted a Generalized Linear Model with Poisson error distribution. PPSR was included as the predictor variable and the number of ants in the backpack (rounded to next integer) as the response variable. Statistical analyses were performed in the R environment (R Core Team, 2021).

The only program required is R.