Methods
Pogonomyrmex body measurements
P. barbatus and P. badius were used for measurements as a large quantity of individuals were available from previous field and lab sampling. Samples used for measurement were either stored at -80° C (most P. barbatus ) or dried (P. badius ); for the former, individuals were from years of field and lab sampling, some of the same colonies as were part of previous studies (Smith et al. 2012, 2015, 2018), and similarly for the latter (Smith and Tschinkel 2006, Smith 2007). For both species the objective was to maximize variation in body size for each caste rather than maintain sufficient sampling to examine colony-level effects. P. barbatus lab colonies produced microgynes and micromales as well as minim/nanitic-like workers, sometimes after years of lab rearing. These micro individuals were included to help disentangle the effects of size and caste. Because size variation was the primary variable being maximized in the sampling design, some colonies are only represented by single individuals while others with many. A total of 31 colonies of P. barbatus were included spread across 25 gynes, 19 males, and 38 minor workers (82 total). A total of 11 P. badius colonies were sampled with 17 gynes, 23 males, 21 major workers and 23 minor workers.
All ants measured were dried and separated into body and appendage segments using a stereo microscope, and glued to card stock (on their right side) to help standardize focal plane and angle among samples. All measurements were done using a a stereomicroscope (SZX7, Olympus, USA) with an attached camera (Retiga 2000R, Q-Imagine, Canada) and measured after calibration using iSolutions-Lite software (iMT Technology, USA); magnification of images differed for body parts and castes, but ranged from 25 to 56X.
Body parts measured were chosen for their repeatability and frequency of use in ant studies in order to facilitate cross-study comparisons. Pictures and descriptions of these standard measures are available at https://www.antwiki.org/wiki/Morphological_Measurements (edited by S. Shattuck). A total of 16 measurements were taken for each ant. Length: Head length (full-face view, edge of clypeus to head vertex), mesasoma/Weber’s length (profile view, head attachment to petiole attachment), gaster length (dorsal view, post-petiole attachment to end of first gastral tergite along post-petiole to sting axis), the length of the tibia and femur of each leg (fore, mid, hind) was measured; Width: head width 1 (full-face view, head width across the eyes), head width 2 (full-face view, width at clypeus), head width 3 (full-face view, width at mid-point between vertex of head and eyes), gaster width (dorsal view, widest point of first gaster segment perpendicular to length axis); Height: thorax height at three points (profile view, measured at the anterior top of each leg, fore, mid, and hind). Ant length (sum of head, mesasoma, and gaster length) was used as a proxy of ant size when comparing to multivariate statistics (below).
All analyses were done using R 4.1.2 (R Core Team 2021) and the RStudio interface (RStudio Team 2022). Plots were constructed using ggplot2 (Wickham 2016) and where relevant combined using ggpubr (Kassambara 2020).
Non-metric multidimensional scaling (NMDS) was used to compare body size and shape across the measured characters in both P. barbatus andP. badius . Bray-Curtis distances were used and only two dimensions were extracted for this analysis – the metaMDS function from the package vegan 2.5 (Oksanen et al. 2020) was used, and then points were extracted and plotted. Differences among castes was assessed statistically using PERMANOVA in size-shape space using the function ‘adonis’ from vegan (Oksanen et al. 2020) with pairwise comparisons made using adonis.pair from the EcolUtils package (Salazar 2022). Furthermore, differences among caste in shape was tested using ANOVA on the second scaling axis (orthogonal to size).