Parallel and divergent morphological adaptations underlying the evolution of jumping ability in ants

Material We selected one worker specimen preserved in ethanol (70–99%) from each genus for which jumping behavior has been previously documented, to carry out detailed computed tomography (CT) scanning and 3D morphometry: Harpegnathos saltator Jerdon, 1851 (unique specimen identifier: CASENT0764679), Gigantiops destructor Fabricius, 1804 (CASENT0709414), Odontomachus rixosus Smith, F., 1957 (CASENT9741319) and Myrmecia nigrocincta Smith, F., 1858 (CASENT0741302). As a control, we imaged a small set of non-jumping ant species: Euponera sikorae Forel, 1891 (CASENT0709898), Formica rufa Linnaeus, 1761 (CASENT0741323), Odontomachus kuroiwae Matsumura, 1912 (CASENT0741313) and Nothomyrmecia macrops Clark, 1934 (CASENT0795539). An effort was made to select comparison species as closely related as possible to the jumping species considering availability of preserved specimens. All specimens were stained in iodine solution 3–7 days prior to the scanning date. The length of the leg segments was measured three times within the same specimen (in the right and/or left legs). For the specimens for which no information on the body mass was available, the whole body was scanned, and body volume was used as proxy for body mass via assuming a uniform density of 1040 kg/m3: Euponera sikorae (CASENT0741359); Formica rufa (ANTSCAN, CASENT0709272); Gigantiops destructor (CASENT0744574); Nothomyrmecia macrops (CASENT0741364). Micro-CT scanning and 3D-reconstruction Micro-CT scans were generated with a Zeiss Xradia 510 Versa 3D X-ray microscope operated with the Zeiss Scout-and-Scan Control System software (version 14.0.14829.38124) at the Okinawa Institute of Science and Technology Graduate University, Japan. Scans were conducted with a 40 kV (75 μA) / 3 W beam strength under a 4x magnification. Voxel size and exposure time depended on specimen size (Suppl. Table S1). As the mesosoma of ants exceeds the field-of-view of the camera at high magnification, vertical stitching of serial scans was used. 3D reconstructions of the resulting scan projection data were done with the Zeiss Scout-and-Scan Control System Reconstructor (version 14.0.14829.38124) and saved in txm file format. Postprocessing of txm raw data was done with Amira 2019.2 (Visage Imaging GmbH, Berlin, Germany) to segment individual structures into discrete tissue volumes. The segmented voxels were then exported with the plugin script “multiExport” (Engelkes et al., 2018) in Amira 2019.2 as 2D TIFF image stacks. VG-Studio 3.4 (Volume Graphics GmbH, Heidelberg, Germany) was used to create volume renders from the TIFF image series. Muscle architecture was reconstructed with Amira 2019.2 XTracing extension, following the workflow presented in Katzke, Puchenkov, Stark, & Economo (2022). The accuracy of the tracing algorithm in Katzke et al. (2022) was 92% for fiber length estimation and 100% for the pennation angle estimation. Muscle identity and nomenclature follows Aibekova et al. 2022. Muscles most relevant in the movement of the legs were segmented (Ipcm2, Iscm4, I-, II-, IIIscm1, II-, IIIscm2, I-, II-, IIIscm3, Ipcm8, II-, IIIscm6, Ipcm4, II-, IIIpcm3_4, I-, II-, IIIctm1, I-, II-, IIIctm2, I-, II-, IIIctm3); in addition, large muscles including the indirect muscle of the head (Idvm5), the levator (IA1), and one of the rotators (IA2) of abdomen were segmented for control. We want to mention one caveat related to preparation technique. It is possible there may be different degrees of muscle shrinkage due to the preservation in high ethanol concentrations (70–99%) This could in principle cause spurious differences across species, as the level of contraction may vary among species, however we have no evidence this was an issue in this case, and it is unlikely to affect the broad differences identified in this study.