How an organism’s sensory system functions is central to how it navigates its environment. The insect olfactory system is a prominent model for investigating how ecological factors impact sensory reception and processing. Notably, work in Lepidoptera led to the discovery of vastly expanded structures, termed macroglomerular complexes (MGCs), within the primary olfactory processing centre. MGCs typically process pheromonal cues, are usually larger in males, and provide classic examples of how variation in the size of neural structures reflects the importance of sensory cues. Though prevalent across moths, MGCs were lost during the origin of butterflies, consistent with evidence that courtship initiation in butterflies is primarily reliant on visual cues, rather than long distance chemical signals. However, an MGC was recently described in a species of ithomiine butterfly, suggesting that this once lost neural adaptation has re-emerged in this tribe. Here, we show that MGC-like morphologies are widely distributed across ithomiines, but vary in both their structure and prevalence of sexual dimorphism. Based on this interspecific variation we suggest that the ithomiine MGC is involved in processing both plant and pheromonal cues, which have similarities in their chemical constitution, and co-evolved with an increased importance of plant derived chemical compounds.

i) Animals

Specimens were collected in the Estación Científica Yasuní, in the Parque Nacional Yasuní, Orellana Province, Ecuador, during November-December 2011, and September-October 2012, under permit 0033-FAU-MAE-DPO-PNY and exported under permits 001-FAU-MAE-DPO-PNY and 006-EXP-CIEN-FAU-DPO-PNY. Permits were obtained from Parque Nacional Yasuní, Ministerio Del Ambiente, La Dirección Provincial de Orellana. Species representing 12 genera, excluding Godyris, were selected on the basis of phylogenetic distribution and available sample size, and represent 8 of the 10 ithomiini subtribes (Table S1; Figure 1). These species are: Melinaea spp., Mechanitis polymnia, Forbestra olivencia, Methona grandior/curvifascia, Ithomia amarilla, Hypothyris anastasia, Napeogenes larina, Oleria gunila, Hyposcada illinissa, Callithomia lenea, Pseudoscada florula, and Hypoleria Sarepta. To increase sample size for Melinaea, which is in an important phylogenetic position, we include individuals from M. menophilus, mnasias and marsaeus (see Supplementary Material for further discussion). Dissection and fixation of specimens were performed at the Estación Científica Yasuní. Brains were exposed by removing a section of head carapace under HEPES-buffered saline (HBS; 150 mM NaCL; 5mM KCL; 5 mM CaCl2; 25 mM sucrose; 10mM HEPES; ph 7.4), before being fixed with zinc formaldehyde solution (ZnFA; 0.25% [18.4 mM] ZnCl₂; 0.788% [135mM] NaCl; 1.2% [35mM] sucrose; 1% formaldehyde) for 16-20 hours. Extraneous head tissue was then removed, and brains were washed three times in HBS. Samples were transferred to 80% methanol/20% DMSO for at least two hours, then stored in 100% methanol. Samples were kept at room temperature until return to the UK, then transferred to -20°C.

ii) Immunohistochemistry

Samples were rehydrated using serial Tris buffer-methanol solutions (90%, 70%, 50%, 30% and 0%), each for 10 minutes. Brains were then incubated for two hours in NGS-PBS_d (5% Normal Goat Serum, 1% DMSO, 94% 0.1M PBS), before being exposed to anti-SYNORF1 (Antibody 3C11; Developmental Studies Hybridoma Bank, University of Iowa, Iowa City, IA, RRID: AB_2315424; Buchner, 1996) in solution with PBS_d-NGS, at a ratio of [1:30], for 3.5 days at 4°C. Non-bound antibody was removed by washing with PBS_d (1% DMSO, 99% 0.1M PBS) three times. Goat anti-mouse secondary antibody, Cy2-conjugated (Jackson ImmunoResearch; Cat No. 115–225‐146, RRID: AB_2307343, West Grove, PA), was then applied at [1:100] in PBS_d-NGS for 2.5 days at 4°C. Samples were imbued with glycerol through graded exposure in 0.1M Tris buffer (1%, 2%, 4% each for two hours, and 8%, 15%, 30%, 60%, 70%, 80%, each for 1 hour), and full dehydrated by washing with 100% ethanol (3 x 30 minutes). Methyl Salicylate was then underlaid, and the brains allowed to sink. This was repeated twice before transfer to storage vials of methyl salicylate.

iii) Confocal imaging and Image Segmentation

Samples were mounted in methyl salicylate held between two cover slips either side of a hole bored through an aluminium slide. Mechanitis and Ithomina were imaged on a Leica SP5 microscope using a 10x 0.4NA objective. All other species were scanned on an Olympus IX3-SSU using a 10x 0.4NA objective. As we do not compare raw volumes across species, the use of different microscopes does not affect our subsequent analyses. For each individual, a single stack was taken encompassing the whole of one antennal lobe, chosen at random, with a z-step of 1µm between each optical section and a x-y resolution of 1024x1024 pixels. Consistent light detection was ensured by adjusting the laser intensity and gain with depth. To correct for the artefactual shortening of the z-dimension of images due to the air objective lenses, a correction factor of 1.52 was applied to the z-dimension of the image stacks (Heinze and Reppert, 2012; Montgomery and Ott, 2015). Image segmentation was performed in Amira 5.4.1 (ThermoFisher Scientific; RID: SCR_007353). Volumes of evaluated areas were exported using the measure statistics tool. Volumes for surface models were plotted using the nat R package (Bates et al. 2020).

Putative MGCs were identified on the basis of internal fibrous structure and location at the base of the antennal nerve, as described in Montgomery and Ott (2015). To formally examine the presence of MGs, each individual glomerulus was segmented in two focal male ALs for each species. In Methona 4 males were analysed in this way to confirm the apparent lack of both MGs and MGCs (see results). Males were chosen for this initial assessment as the MGs clustered around the base of the antennal nerve are larger in Godyris males, and no enlarged glomeruli outside this cluster observed in females (Montgomery and Ott, 2015), as is also commonly observed across moths (Rospars and Hildebrand, 1992; Huetteroth and Schachtner, 2005; El Jundi et al., 2009b; Løfaldli et al., 2010; Yan et al. 2019). Three tests were used to evaluate whether MGs were present in focal individuals, based on methodologies used in previous publications (Montgomery and Ott, 2015; Keubler et al, 2010; Kelber et al. 2009). As the method used by Kelber et al (2009) was found to be the most conservative (see Supplementary File) we focus on these results in the main text. Under this method, a glomerulus is considered to be a MG if its volume is greater than the 90th percentile of glomeruli volumes plus k times the difference between the 10^th and 90^th percentiles (Kuebler et al. 2010). We discriminated between MGs and normal glomeruli using a threshold k value of 1.5, which defines a moderate outlier (Sachs, 1988). MGs were considered to be present if they passed the discrimination threshold in both fully segmented males. We note that our definition of a MG is dependent upon the volumetric distribution of the other glomeruli, meaning that expansion of non-MGC glomeruli may obscure the presence of glomeruli expanded to a similar degree as MGs in other genera. We therefore provide the glomeruli size distributions for all species (Figures S2-14).

For subsequent evaluations of sexual dimorphism we scanned multiple males and females for each species in the same way as described above. In total 178 individuals were measured for this dataset with an average of 15 individuals/species. Samples size for each species are provided in Table S1, and range from 2-12 males and 4-11 females. In all individuals of both sexes we then segmented: i) all glomeruli comprising the putative MGC, including both MGs and closely associated ‘satellite’ glomeruli. Satellite glomeruli were included where they form a distinct unit of glomeruli with the MGs, which are together raised with respect to the anterior surface of the remaining glomeruli, and have the fibrous internal structure characteristic of MGCs (Montgomery and Ott, 2015); ii) the combined volume of all glomeruli; and iii) total antennal lobe volume including glomeruli and the internal antennal lobe hub. Data for Godyris zavalata were taken from Montgomery and Ott (2015) but images were checked for consistency with newly obtained data.